-------
$
"o
_
u
C
Q.
1
I
C/3
LLJ
02
13
cc
25
24
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4 -
3 -
2 -
1
0
AREAL LOADING PER
FILTERING CYCLE (FFPP)
«fc 0.05 LB/FT2 \
$ 0.15 LB/FT2 [REVERSE AIR
£ 0.25 LB/FT2 )
O SHAKE
UTILITY FABRIC FILTERS
O MONTICELLO* (SHAKE)
O ARAPAHOE NO. 3*
4 CAMEO NO. 2*
A CHEROKEE NO. 3*
D ECOLAIRE*
* HARRINGTON (SHAKE)*
B KRAMER*
V MARTIN DRAKE*
*. NIXON*
A NUCLA (SHAKE)**
* TUBE SHEET AP
** FLANGE - FLANGE
1.0 1.7 2.0
FILTERING AIR TO-CLOTH. acfm/ft2
3.0
4517-360
Figure 3. Average operating AP versus A/C for the FFPP and nine baghouses installed at
coal fired plants.
154
-------
is a summary of the average AP vs. air/cloth ratio for the reverse-gas com-
partments of the FFPP at three values of areal loading (from Figure 1), and
five full scale reverse-gas units. Also, similar data are plotted for three
shaker units and from the FFPP deflate/shake tests. Notice that it was pos-
sible to measure the actual tube sheet Ap for some units, while at others
only flange-flange data were available, which includes an unknown pressure
drop across the ductwork. The data are generally similar for the reverse-gas
units, all of which are located in Colorado and installed on boilers burning
low-sulfur coal.
Good agreement is seen between data taken at the FFPP and the full-
scale units. Notice the linear dependence of AP upon A/C below A/C = 3 acfm/
ft2, indicating the possibility of operating at higher A/C values than the
traditional 2 acfm/ft2 for savings in capital costs. The increased slope
above A/C = 3 is not well understood, but appears to result from differences
in permeability of dust cakes formed at different face velocities.
As expected the units cleaned by shake/deflate tend to operate at sig-
nificantly lower pressure drops than the reverse-gas units, except at the
Monticello station, which burns Texas lignite. This station has experienced
continuous problems of high pressure drop and dust bleed through, and the
interaction of its unusual ash with fabrics of different types is currently
being studied.
DUST CAKE ANALYSIS
Samples were cut from the bags at each test site after the compart-
ments were cleaned and shut down. Rather thick residual dust cakes were
found in each instance. Figure 4 is a cross sectional photograph illustrat-
ing a typical example from a reverse-gas unit. The cake is rugged, with
large nodular deposits, deep fissures, and a thinner deposit along the "fold
line" where the bag flexed during reverse-gas cleaning. The actual weights
of the dust cakes are summarized in Table 1. The weight of dust per unit
area ranged from 0.4 to 0.6 lb/ft2 for all of the installations investigated
(40-60 Ib/bag). Considering that this is the residual cake left after clean-
ing and that the calculated accumulation of dust during one hour of filtering
is expected to be about 0.05 lb/ft2, it is concluded that only about 10 per
cent of the dust cake is removed by reverse-gas cleaning—probably the
freshest part of the cake. Recall again that the data presented in Figure 4
and Table 1 are for bags that have been "cleaned" by reverse-gas. Thus, the
residual dust cake is the one analyzed.
Studies were made of the bag swatches to determine how permeability of
the dust cake, or alternately the drag (AP vs. A/C ratio) depended upon its
physical properties. The swatches were placed in a large filter holder and
their AP vs A/C characteristics measured. Dust was then removed from the
swatches for further analysis. The distribution of particle sizes in the
range of 4-80 ym was measured using a Coulter counter, and the results re-
lated to the measured drag.
n
Figure 5 is a summary of all the permeability data from 1 ft samples
taken from the top, middle, and bottom of bags at the five full-scale units
155
-------
ui
-
FABRIC
NODULES
FISSURES
FOLD AREA
i 2345673 =
IC'-s
'OOt-s
1234 6789
12245S7S3 123455789 123456789 1 2 i
i-40996
234 (5789 1234 6789 123
4617-358
Figure 4. Cross sectional view of fiber glass bag and dust cake from a baghouse cleaned by reverse gas.
-------
Table 1. Summary of the weights of dust cakes from several baghouse installations.
PLANT
BAG TYPE
MARTIN DRAKE
NIXON
CHEROKEE
(UNIT 3)
CAMEO
(UNIT 2)
ARAPAHOE
(UNIT 3)
FFPP
SHAKE (1/82)
FFPP
REV. AIR (1/81)
ECOLAI RE
PI LOT PLANT
MENARDI SOUTHERN
601-T (9.5 oz/yd)
MENARDI-SOUTHERN
601-T (9.5 oz/yd)
FIBERGLASS WITH ACID
RESISTANT COAT
(9.5 oz/yd)
MENARDI-SOUTHERN
601-T (9.5 oz/yd)
MENARDI-SOUTHERN
601-T (9.5 oz/yd)
ALBANY INTERNATIONAL
Q53-S3016 TRICOAT
TYPE 11T (NO RINGS)
(9.2-9.6 oz/yd)
ALBANY INTERNATIONAL
Q53-S3016 TRICOAT
TYPE 12T (RINGS)
(9.2-9.6 oz/yd)
FABRIC FILTER CO.
ACID-FLEX
STYLE 504-1AF
SIZE OF BAGS
(LxD)
30.5'x 12"
31.75' x 12'
33.8'x 12"
22.1' x 8"
22.1' x 8"
34'x 11.5"
34'x 11.5"
24' x 8"
AVERAGE RESIDUAL WEIGHT
OF DUST CAKE
(After Cleaning)
LB/FT2
0.5
0.4
0.4
0.5
0.6
0.2
0.2 (1/81)
0.6 (1/82)
0.15(11/81)
1.0 (2/82)
4617-267
157
-------
Li.
co
q
o
o
o
LU
_l
Q
Q
Q.
O
• //// TOP
MIDDLE
BOTTOM
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
MEDIAN PARTICLE DIAMETER AT TOP, MIDDLE AND BOTTOM
2.0
MEDIAN PARTICLE DIAMETER AT MIDDLE
4617-266A
Figure 5. Relative size of particles and drag of dust cakes found at the top, middle,
and bottom of bags from several baghouses.
158
-------
and the FFPP. The data are normalized by dividing both the values of drag
and median diameter measured at the top and bottom of the bags by the values
obtained at the middle of the bags and dividing by the mass of dust per unit
area. This method of normalization is intended to eliminate variations im-
posed by differences in the inlet particle size distribution at the differ-
ent power plant installations and variations in thickness (weight) of the
values.
From Figure 5 it can be seen that there is little correlation between
the drag and the median diameters of the particles comprising the cakes. The
Coulter analysis indicates that the particles near the top of the bags, how-
ever, were somewhat smaller than those from the middle and bottom. These
data suggest that the drag is not determined by the thickness of the cake or
particle size, but by the macroscopic structure; i.e., the cracks, voids, and
fold lines.
Measurements of the ash chemistry, structure, and resistivity are be-
ing made in order to gain a better understanding of the factors that deter-
mine the drag, or pressure drop. Although the measured concentrations are
small, the chemical analyses do show progressively higher concentrations of
sulfate in (1) the hopper samples, (2) smooth cake samples, and (3) nodular
cake samples, respectively. This is thought to confirm the conclusion that
an equilibrium situation is reached wherein a relatively small amount of ash
is deposited and removed during each filtering and cleaning cycle, while the
heavy, residual cake remains relatively undisturbed. No cementitious struc-
ture has been found, however, in our microscopic analysis. Thus, there is
no indication that the relatively higher sulfate concentration plays a sig-
nificant role in the formation or tenacity of the dust cake nodules.
CURRENT FFPP TESTS
At the FFPP, research has been initiated to investigate the effective-
ness of sonic energy enhancement on reverse gas cleaning. Two types of com-
mercially available sonic energy sources (horns) are currently being evalu-
ated. Cleaning sequences which are being tested range from simultaneous use
of reverse gas and sonic energy to sonic energy alone. Early in 1983 test-
ing will begin using a "generic" sonic energy source capable of operation at
a continuum of fundamental frequencies and energy outputs. This will allow
effective sonic energy density within a baghouse compartment (number of horns
and location). The use of sonic energy to augment reverse air cleaning is
dramatically illustrated in Figure 6. Here the sonic horn was activated for
10 seconds in the middle of a 30 second reverse gas period. The reverse gas
air-to-cloth ratio was 2.5 ft/min. The pressure drop was reduced by more
than one-half when the horn was used.
CONCLUSIONS
In the FFPP and the full-scale baghouses, heavy residual dust cakes
are left on the bags after reverse-gas cleaning. Only about 10 per cent of
the cake is removed during each cleaning cycle, yet the drag is reduced by
approximately a factor of two. Accordingly, it is concluded that the
159
-------
01
o
c
E
o
o
I
u
c
CL"
-1
LU
O
PILOT BAGHOUSE ON LINE
AFTER POWER PLANT OUTAGE
10
THE HORN WAS ACTIVATED HERE AND DURING
ALL SUBSEQUENT CLEANING CYCLES
CONTROL COMPARTMENT
COMPARTMENT WITH HORN
I I I I I I I
I I I
10
15
20
25
MAY, 1982
30
4517-390
Figure 6. Pressure drop/time history illustrating the effectiveness of sonic bag cleaning. Both compartments were
operated at an A/C ratio of 1.2 ACFM/FT? and cleaned every 3 hours by reverse gas enhanced with
sonic energy.
-------
pressure drop is determined as much by the structure of the cake as by the
amount of dust present. For the sites evaluated, there was no particular
correlation between the pressure drop and particle size or chemistry.
AP vs. air/cloth ratio appears to be a nearly linear function, with a
sharp increase in slope near 3 acfm/ft2. All of the reverse-gas units eval-
uated performed similarly, but at a higher pressure drop than has been re-
ported for most shaker units. If a method could be found to remove the cake
more effectively, it is expected that the operating pressure drop of reverse-
gas baghouses could be lowered by a large fraction without a significant
loss in collection efficiency.
The dependence of pressure drop on areal dust loading (time between
cleaning) is complex. For higher values of air/cloth (>2.2 acfm/ft2) the
average pressure drop in the FFPP increased slightly as the dust loading on
the bags (dwell time) increased. For lower values of air/cloth, the average
pressure drop decreased as the dwell time increased.
After 10,000 hours of reverse-gas operation no bags have failed. The
collection efficiency has consistently exceeded 99.9% after the first few
days of operation. Several weeks are required for the system to stabilize
for each test condition.
Over the range tested (RA/C of 2.0 to 6.0 acfm/ft2), the effectiveness
of bag cleaning was not a strong function of reverse-gas volume. Since a
value of 6.0 acfm/ft is very high for practical operation, reverse-gas
cleaning appears to be a marginal method of removing dust cakes from fabric
filters.
Preliminary indications are that a sonic aided reverse-gas cleaning is
very effective in removing residual dust cakes. Future plans include de-
tailed studies of that cleaning mode with the goal of determining the opti-
mum distribution, intensity, frequency, and timing of the sonic energy.
Extended tests of shake/deflate cleaning are also planned.
The emphasis in all these tests will be to achieve a reduction in
pressure drop and baghouse size without sacrificing reliability or effi-
ciency.
ACKNOWLEDGEMENTS
We would like to express our appreciation to the staff and management
of the Public Service Company of Colorado and the Colorado Springs Depart-
ment of Public Utilities for their cooperation and assistance in evaluating
the fullscale baghouses. The data reported here were all taken by staff
members of Southern Research Institute. The FFPP is operated by Kaiser En-
gineers under the direction of Lou Rettenmaier of EPRI. We would like to
thank Richard Hooper of EPRI for his continuing support in acquiring and in-
terpreting the data.
161
-------
The work described in this paper was not funded by the U.S. Environ-
mental Protection Agency and therefore the contents do not necessarily re-
flect the views of the Agency and no official endorsement should be inferred.
162
-------
REFERENCES
1. Smith, W. B., R. C. Carr, K. M. Gushing, and G. B. Gilbert. Air load
and startup tests of a 10 MW fabric filter pilot plant. Presented at
the Fabric Filter Forum, Phoenix, Arizona, January 1981.
2. Smith, W. B., R. C. Carr, and K. M. Gushing. Characterization of a 10
MW fabric filter pilot plant. Presented at the Third Symposium on
Transfer and Utilization of Particulate Control Technology. Sponsored
by the U.S. EPA, Orlando, Florida, March 1981.
3. Carr, R. C., W. B. Smith, and K. M. Gushing. Characterization of a 10
MW fabric filter pilot plant. Paper 81-9.2. Presented at the Annual
Meeting of the APCA, Philadelphia, Pennsylvania, June 1981.
4. Carr, R. C., W. B. Smith, and K. M. Gushing. Test results from operat-
ing fabric filters: full scale and Arapahoe 10MW pilot plant. Present-
ed at the EPRI Conference on Fabric Filter Technology for Coal-Fired
Power Plants, Denver, Colorado, July,, 1981.
5. Smith, W. B., K. M. Cushing, and R. C. Carr. Measurement procedures and
supporting research for fabric filters. Presented at the EPRI Confer-
ence on Fabric Filter Technology for Coal-Fired Power Plants, Denver,
Colorado, July, 1981.
6. Ladd, K. L., R. L. Chambers, 0. C. Plunk, and S. L. Kunka. Fabric fil-
ter system study second annual report. EPA-600/57-81-037, July, 1981.
7. Ensor, D. S., S. Cowen, A. Shendrikar, G. Markowski, G. Woffinden,
R. Pearson, and R. Sheck. Kramer station fabric filter evaluation.
EPRI Report No. CS-1669, January, 1981.
8. Ensor, D. S., R. G. Hooper, and R. W. Sheck. Determination of the
fractional efficiency, opacity characteristics, engineering, and econom-
ic aspects of a fabric filter operating on a utility boiler, EPRI
Report No. FP-297, November, 1976.
153
-------
THE DESIGN, INSTALLATION, AMD
INITIAL OPERATION OF THE
W. H. SAMMIS PLANT UNIT 3 FABRIC FILTER
by: Dennis R. Ross, P.E.
Generating Plant Staff Engineer
Ohio Edison Company
Akron, Ohio UU308
James R. Howard,
Plant Engineer
W. H. Sammis Plant
Ohio Edison Company
Stratton, Ohio U3961
R. Mark Golightley,
Plant Engineer
W. H. Sammis Plant
Ohio Edison Company
Stratton, Ohio
ABSTRACT
This paper will discuss the design philosophy and preliminary
operating experience of the first unit of four units to be retro-
fitted with fabric filters at the W. H. Sammis Plant. Each unit
is l80 MW net. The unit was operational late August 1982.
The paper will identify design criteria initially specified,
design enhancements during the course of the project, construction
highlights, and preliminary operating experience. Results from the
first months of operation will include pressure drop and opacity.
In addition, start-up, shutdown, and maintenance procedures will
be discussed.
164
-------
INTRODUCTION
The W. H. Sammis Plant, located on the Ohio River north of
Steubenville, Ohio, is the second largest of the Ohio Edison Company
system coal-fired plants. Currently, the units burn coal mined
locally as well as out-of-state coal. The first four units have a
Net Demonstrated Capacity (NDC) of 180 MW each, with gas flows of
approximately 755,000 ACFM each at 305°F. Unit 5 is 300 MW NDC with
a gas flow of approximately 1,300,000 ACFM at 2TO°F. Units 6 & 7
are 600 MW NDC each, with gas flows of approximately 2,600,000 ACFM
each at 270°F. Flue gas from Units 1 & 2 are combined into Chimney
No. 1 and Units 3 & ^ are combined into Chimney No. 2. Each of these
chimneys are 500' high. Units 5 & 6 share the No. 3 Chimney (850* high)
while Unit 7 has its own 1,000* high chimney. All seven units have
electrostatic precipitators which were originally installed with the
boilers. Units 1-^ went into service during the period 1959-62, Unit
5-1967 and Units 6 & 7-1969 and 1971, respectively. The Company is
currently engaged in a program to retrofit high efficiency particulate
collection equipment on all seven units, as the existing precipitators
can not achieve the level of collection required to satisfy the emission
limits.
For the purpose of this discussion, primarily the equipment,
fabric filters, which have been purchased for the Units 1-U will be
presented. Units 5-7 are being equipped with electrostatic pre-
cipitators.
BACKGROUND
All of the units were designed and equipped with the then current
state-of-the art emission control features of electrostatic precipi-
tators and tall stacks. The precipitators, for a variety of reasons,
have never performed to the highest level expected of them. With
development of particulate emission limitations as required by the
Clean Air Act of 1970, it became apparent that further steps would
be needed to meet the newly imposed regulations.
In 1972, Gilbert/Commonwealth, Ohio Edison's Architect/Engineer,
was commissioned to perform detailed studies on all of the Ohio Edison
Company plants to determine a compliance strategy to satisfy the limi-
tations imposed by the new regulations. These studies were updated
routinely, primarily because of changes in the proposed emission regu-
lations and fuel availability. Most recently, in mid-1977, the U.S.
EPA finalized S02 standards for Ohio.
165
-------
The federal standards finalized in 1977 are site-specific and even
stack-specific. Currently, the standards for the W. H. Sammis Plant are
.1 Ibs/MMBTU for particulate and 2.91 Ibs/MMBTU for SC>2 from each chimney.
Utilizing alternate equations to determine allowable S02 emissions per-
mitted some variability to the emissions from individual chimneys.
Needless to say, some interesting combinations were investigated by
which compliance could be met. The ultimate goal was to establish a
strategy to meet both particulate and S02 emission regulations through
the selection of the proper combination of coals.
A predominant factor in the development and final selection of
control strategies was the evaluation of site-specific requirements
which also affected the economics. Probably the most important con-
sideration was the space available for equipment installation.
The plant is located between railroad tracks and 500' high hills
on the west and the Ohio River on the east. Between the plant and
the Ohio River is Ohio Route 7, a four-lane highway. Available space
for equipment between the plant and the highway was minimal at best.
Locating the control equipment on the plant roof and in the
south yard were considered, but these locations, upon some detailed
evaluation, were determined to be not feasible. Other ideas were
explored, including the location of the equipment on a deck-like
structure built over the highway. The more this concept was explored,
the more feasible it appeared. Preliminary contact with the highway
department was encouraging, and the concept was further developed.
The control strategy ultimately selected for particulate and
S02 emission compliance was the use of low sulfur (less than 1$)
for Units ±-k and coal not to exceed 2.8% sulfur for Units 5-7 and
the installation of new high efficiency particulate collection equip-
ment on all seven units. This strategy was the least costly to
satisfy the regulations. In accordance with this control strategy,
it was determined that the particulate collectors had to have an ef-
ficiency of approximately 99.1*1$ in order to meet the 0.1 Ibs/MMBTU
particulate emission regulation.
In January 198l, a Consent Order was signed with the U.S.
EPA which established the compliance dates for meeting particulate
emission limits, as shown in Table No. 1.
165
-------
TABLE 1. W. H. SAMMIS PLANT COMPLIANCE SCHEDULE
Unit No. Compliance Date
1
2
3
h
5
6
7
January 31, 1983
August" 31, 1982
June 30, 1982
November 1, 1982
April 30, 1981+
August 31, 198U
December 31, 1982
The Consent Order incorporated new particulate collection
equipment installation on the seven boilers at the W. H. Sammis
Plant and five boilers at three other plants and upgrading par-
ticulate collection equipment on eleven boilers at three plants.
Non-compliance with particulate emission limitations at the W. H.
Sammis Plant could cost Ohio Edison Company $7,500 per day per
boiler. Interim deadlines established by the Consent Order which
addressed primarily construction activities could also cost the
Company $5,000 per boiler per day fines, (l)
The total effect of the Consent Order requires that Ohio Edison
Company expend approximately $550 million to reduce particulate emis-
sions. At the W. H. Sammis Plant alone, approximately $^50 million
will be expended.
CONSTRUCTION AND DESIGN HIGHLIGHTS
As stated previously, the site constraints posed a major problem
from an engineering and construction standpoint. The deck concept
became the only viable method of supporting equipment. The deck spans
a four-lane highway between the plant and the Ohio River. It is 915
feet long and lUO feet wide and allows 20 feet of clearance above the
highway. The foundation has 192 caisson piers approximately 6 feet
in diameter that go to bedrock. Approximately 12,000 tons of structural
steel, breeching, fabric filters and electrostatic precipitators, are
supported by the deck. The deck construction cost approximately $27
million.^' The deck is shown on Figure No. 1.
Access to the deck is limited, so construction is staged such that
work activities progress from the plant toward the Ohio River and from
the middle of the deck toward the ends of the deck. Material and sub-
assemblies from a nearby marshalling yard were moved by barges to the
deck as needed. Figure No. 2 shows construction activities on the deck.
167
-------
Draft losses associated with the new fabric filters and the ex-
tensive ductwork necessitated new ID fans for the Units 1-U "boilers.
Modified radial tip fans with two-speed PAM motors are being used.
The PAM motors were selected to reduce low load power consumption,
to allow easier fan start-up and to reduce the risk of boiler implosions.
Our design criteria for the fabric filters and the ductwork from
the air heater outlet to the chimney is +_30 IWG. An economic evaluation
was conducted relative to strengthening the boiler to meet the new de-
sign criteria for implosion protection. In lieu of boiler strengthening,
it was decided to install additional draft controls. A Bailey Controls
Network 90 implosion protection package was selected for the Units 1-U
boilers.
Throughout the course of the project, many types of models were
utilized. Initially, our consultants developed a table-top scale model
of the immediate plant vicinity. This was used internally to develop
and present alternative equipment locations influenced by the site
constraints. This model was subsequently used in discussions with
various regulatory agencies in securing approvals for the deck and
barge unloading facility construction, as well as negotiations
relative to the Consent Order.
Construction of the deck had to be accomplished while maintaining
highway traffic flow at all times. A model was constructed which in-
corporated scaled-equipment to aid in planning the deck construction.
Retrofit projects produce unique problems from a standpoint of
engineering and construction. Site constraints, especially those
imposed at the W. H. Sammis Plant, at times, make these problems even
more burdensome. Existing facilities must be moved to accommodate new
equipment and facilities must be altered and strengthened to accommodate
the new design. Our consultants originally constructed a 3/8" scale
model of the back portion of the plant from Units U through 7 to verify
design layouts. All equipment, piping, steel and ducting was shown in
detail. As the project progressed, this model became more of a design
construction and planning tool than a design verification model. Inter-
ferences were quickly recognized and changes were made during the design
stage.
With the addition of new ID fans and additional lengths of
breeching, boiler control and implosion protection were of great
concern. Our consultants mathematically modeled the gas side of the
boiler and its controls. This model defined problem areas and analyzed
the effectiveness of different control schemes.
As a standard design requirement, we require the precipitator
vendors to flow model the precipitator and duct system to optimize
flow distribution and pressure drop. At the onset of the project,
this was not required of the fabric filter vendor. The industry
had been led to believe that a fabric filter was a flow-balancing,
163
-------
pressure-equalizing device. However, during the project, it was
decided to conduct a flow model study of the fabric filter. A more
detailed discussion of the fabric filter model will be addressed later.
FABRIC FILTER DESIGN
Coal being burned in Units 1-h is low sulfur eastern coal with
design specifications of: 0.6 to 0.8$ sulfur; 10-20% ash; and 10,000-
12,000 BTU/lb. Design gas flow is 75^,000 ACFM each at 305°F. In
1977, early in the planning for the current particulate control program,
it was decided to solicit bids for electrostatic precipitators and
fabric filters. In June 1978, American Air Filter was issued a
letter of intent to supply four fabric filters, one each for Units l-"k,
Table No. 2 lists pertinent data of the fabric filter.
TABLE 2. FABRIC FILTER DESIGN INFORMATION
Fabric Filter Supplier American Air Filter
Number of Compartments 12
Type of Cleaning Reverse air
Air-to-Cloth Ratio-Design
Gross 1.93:1
Net(l) 2.1:1
Net-Net(2) 2.31:1 2
Cloth area per Compartment 32,632 ft.
Filter Bags
Per Compartment 312
Total 3,1^
Filter Bag Configuration 3w6w3x26
(l) One compartment off-line for cleaning
(2) One compartment off-line for cleaning, and one
down for maintenance
The fabric filter reverse air and outlet dampers are poppet-type
and the inlet damper is a flat-plate butterfly damper. Due to site
constraints and the desire to utilize existing chimneys, the inlet and
outlet of the fabric filter are on the same side. Figure No. 3 illu-
strates the confined duct arrangement. The reverse air system has two
100$ sized fans for each unit. The reverse air circuit is kept hot by
circulating clean flue gas. When a compartment requires cleaning the
reverse-air bypass dampers close, and the individual compartment reverse-
air damper opens. A manually operated compartment ventilation system is
also provided. Automatic control of the fabric filter is accomplished
by an Allen-Bradley programmable controller. Pressure drop information
is shown in Figure No. h.
169
-------
The fabric filter is bypassed via two Andco flap-door type
dampers. The fabric filter will automatically be bypassed on high
flue gas temperature and high differential pressure across the fabric
filter. The fabric filter is manually bypassed on start-up and can be
operator initiated for emergency or abnormal situations.
The filter bags are fiberglass with Teflon B coating. Additional
information is shown in Table No. 3. In Unit No. 3, thirty-six acid
resistant bags have been installed for test purposes. These bags are
style QT8-S3016 with Filter Resources' acid resistant proprietary
finish.
TABLE 3. FILTER BAG DATA
Bag Manufacturer Filter Resources
Bag Style Q79-S3016
Bag Finish 10% Teflon B
Weight 9.5 oz/yd2
Weave 3x1 Twill
Count 5U x 30
Permeability 35-6o(l)
Yarn Designation
Fill 75 1/2
Warp 75 1/0
Bag Dimensions
Diameter 11 3A"
Length 3Uf 7"
Anticollapse Rings
Number 8 per bag
Type Cadmium plated carbon steel
(l) CFM/ft.2 @ 0.5 IWG
FABRIC FILTER FLOW MODEL STUDY
When the fabric filter was initially procured, a flow model study
was not a requirement. After seeing results of the flow model con-
ducted for the EPRI pilot plant located at the Arapahoe Plant, it was
deemed appropriate to flow model the W. H. Sammis Plant fabric filter.
American Air Filter was approached with this request, and they agreed
that a flow model would be beneficial.
The model work was conducted by NELS Incorporated of St. Catherines,
Ontario, Canada. The model was 1:12.57 scale and made of plexiglass.
It incorporated all the ductwork from the air heater outlet to the
fabric filter, the fabric filter, and ductwork from the fabric filter
170
-------
to the ID fain inlet. Structural members that affected gas flow,
dampers, and all thimbles were included in the model. A second phase
of the model addressed the ID fan outlet duct to and including eight
diameters of the chimney.
The primary objectives of the modeling program were: minimize
pressure drop; achieve acceptable flow distribution to the ID fans;
determine acceptable flow and particulate distribution to the com-
partments; determine acceptable flow and particulate distribution
through the tubesheets; minimize hopper reentrainment; and evaluate
the system under bypass conditions. Unlike precipitator modeling which
has acceptance criteria established by the industry, fabric filter
flow modeling had no established industry guidelines. A target
balance for flow within +10% and particulate distribution within +20%
of theoretically equal flow was established. Through the tubesheet,
a target balance having an RMS deviation less than 25% was established.
Vaning was added to minimize the pressure losses and to improve
the flow and dust distributions. Flow distribution values were
generally within or close to the target values. Dust distribution
was generally outside the target values. It was recognized that
dust distribution could not effectively be controlled without going
to massive internal control structures and increased pressure losses.
The system pressure loss is essentially constant for various operating
configurations.
Perhaps the two most important findings of the program pertained
to hopper reentrainment and system bypass. Flow patterns in the
hoppers indicated severe reentrainment. The dust flows entering
the hoppers were initially concentrated at the hopper inlet and
continued to recirculate in the hopper without dispersing. This
condition was rectified by installing ladder vanes and a false floor
in the inlet duct and flow deflectors in the hopper.
The original outlet duct configuration with poppet-type bypass
dampers had an unacceptably high pressure drop of IT.7 IWG for a full
load bypass. The entire outlet duct and bypass duct was modified.
The subsequent pressure drop was near that expected under normal
operating conditions. (3, U;
PREOPERATIONAL TESTING AND TRAINING
As the fabric filter installation for Unit 3 neared completion,
approximately U months before start-up, a team comprised of plant
and vendor personnel began checking out the various systems. Wiring
was verified, and limit switches were set on all dampers which provided
inputs to the programmable controller. The cleaning cycle was then
operated in both the pressure and time initiated mode using the program-
mable controller. All other equipment was checked-out in the same
171
-------
manner. This attention vas useful in that problems encountered with
the reverse air fans, limit switches and control panels were detected
and corrected before start-up.
In conjunction with the checkout activities, training was conducted
by various vendors and Gilbert/Commonwealth at the plant site. This
training was directed toward both operating and maintenance personnel.
Throughout the *t month period prior to start-up, trips were
taken by various engineering and plant supervisory personnel to
operating installations. This first-hand observation and discussion
opportunity is extremely beneficial. Learning from the experience
of others allows valuable insight into potential problem areas and
gives guidance for operating and maintenance personnel. This also
establishes a line of communication to solve future problems that
may be encountered.
Unit 3 came off line on June 22, 1982 for a scheduled turbine
overhaul and boiler inspection. During this outage, final tie-in
of the new ductwork was accomplished. Initial firing of the boiler
began on August 28, 1982. Finally, on September 1, 1982, precoating
of the fabric filter was ready to begin.
It was our intent to precoat the fabric filter on line rather
than precoat with cold flyash. This was a change from the original
thinking during the project. With American Air Filter's approval,
the following criteria was established: flue gas temperature to be
305°F or greater; gas flow to be approximately U00,000 ACFM; fluidizing
air and hopper heaters at least 2U hours prior to start-up; the unit
had to be on total coal fire; all 12 compartments in service; and no
cleaning until a 3 IWG pressure drop was experienced across a compart-
ment. At 1:^+0 a.m., on September 1, 1982, the bypass dampers were
closed and flyash laden flue gas was introduced into the fabric filter.
At start-up, the opacity monitor was not operable. The indi-
vidual compartment differential pressure gauges were operable, but
needed to be recalibrated. The primary indication that the fabric
filter was operating was the flange-to-flange differential pressure
recorder. Precoating continued until ^:00 a.m., when the first cleaning
cycle was initiated.
Reverse air damper problems were experienced on several of the
compartments. The damper shaft was hanging up on the guide bearings.
With 9 of 12 compartments operating, the pressure drop was approxi-
mately 3.9 IWG at lUO MW. At 11:26 a.m., the fabric filter was cleaned
down from a differential pressure of ^.3 IWG to 2.6 IWG. We have
elected to maintain a differential pressure of 2.6 to 3.6 IWG. The
design on the Bailey Network 90 implosion protection system needs to
be reviewed and settings readjusted. In this interim period, we are
overcleaning the fabric filter to minimize system pressure losses.
.172
-------
PRELIMINARY OPERATION
Since September 1, 1982, there have been approximately 15 boiler
trips. These trips have been caused mainly by the boiler implosion
protection system signals and equipment. To date, the fabric filter has
not caused a unit trip.
On September 6, 1982, the fabric filter was inspected internally.
At this time, 16 bags were found to have dislodged from the thimbles.
Twelve of the 16 bags were discovered in Compartment B-l which was the
first compartment bagged. Some holes in the tubesheet and at the thimbles
were also discovered. All bags were then retensioned.
The compartments were again inspected on September 2 it, 1982. It was
observed that some bags needed retensioning, but on a random basis.
Several holes in the tubesheets were observed. Leakage between the inlet
plenum and the interior of Compartment 'B-l was observed. This has been
repaired.
Since September 1, 1982, thirteen bags have been removed. Seven of
the bags had failed. Six of the bags were removed because there appeared
to be problems with the bag. Of these, one was an acid resistant coated
bag. These failures have been concentrated mainly in the cuff and first
ring areas. The bags are being analyzed. Currently, we are attributing
these failures to installation problems, normal infant mortality of the
bags,' and problems caused by holes in the tubesheet and back wall.
The opacity monitors were operable on September 13, 1982. Opacity
has generally been running between 9-13$. The opacity is less than the
20$ required opacity, but it is higher than anticipated. This has been
attributed to instrument calibration, some instrument debugging problems,
and possible overcleaning of the bags.
MAINTENANCE CONSIDERATIONS
The plant personnel are committed to a comprehensive preventative
maintenance program. This is based on discussions with American Air
Filter and various utilities with operating fabric filters. Checklists
have been developed to aid in this program. Specific equipment and areas
are highlighted, and space is provided to note observations. This data
will be used to provide historical information to troubleshoot problem
areas.
Because the filter bags are so critical to the operation of the fabric
filter, a filter bag mapping program has been instituted. A form has been
developed which shows a compartment with all of the thimbles. An alpha-
numeric matrix is utilized to locate the filter bags. When a compartment
is inspected, the form is used to note observations such as retensioning
173
-------
requirements, "bag blousing and bags dislodged from their thimbles. A
second sheet is provided to note bag failures. On this sheet, a filter
bag profile is shown. Recorded observations would be bag type (teflon
or acid resistant coating), location of failure, description of failure
and corrective action taken.
Expansion targets have been installed at the outside corners of
the fabric filter. These are located exterior to the fabric filter so
they can be easily observed. It was decided to install these targets
for early detection of expansion problems or problems with the column
sliding plates.
FUTURE PLANS
While it is too early to fully assess the operation of the fabric
filter, we are optimistic. Experience gained from this unit will be
factored into the other three units. Several tasks have yet to be ac-
complished.
As previously discussed, the cleaning cycle will be optimized after
adjustments to the Bailey Network 90 implosion protection system are
completed. From industry experience, the cleaning cycle optimization
could take several months.
The American Air Filter test program of the Filter Resources'
QT8-S3016 acid resistant bags is scheduled to cover three years. Ohio
Edison is contemplating its own filter bag test program on Unit 1. The
primary purpose is to determine candidates for future filter bag replace-
ment.
Reprogramming of the fabric filter programmable controller is required
for two reasons. Currently, the cleaning cycle is programmed to stop at
each compartment. If a compartment is isolated, the cleaning cycle will
still stop at the compartment for its alloted time even though it is
not physically cleaned. The program will be changed to skip an isolated
compartment. Also, during the cleaning cycle, the outlet and reverse air
dampers are moving simultaneously for a given period of time. The program
will be changed to correct this.
At American Air Filter's request, a temperature survey was conducted
on a compartment, B-6. They are concerned about the fabric filter tempera-
ture at low load operation due to low air-to-cloth ratio, low flue gas
temperature, temperature stratification and potential acid dewpoint
problems. The area of concern is primarily at or near the tubesheet
elevation. We have already added extra insulation on the lower 8' of
the compartment exterior wall. Flue stops have also been added in the
insulation to prevent convection currents. Data from the test is currently
being analyzed.
The work described in this paper was not funded by the U.S. Environ-
174
-------
mental Protection Agency, and therefore the contents do not necessarily
reflect the view of the Agency and no official endorsement should be
inferred.
175
-------
REFERENCES
1. Fines Overhead, Utility Races to Meet Air Rules; Engineering
News Record, August 5, 1982, pp. 26-27.
2. Ibid.
3. Ross, D. R., and Bowen, C.F.P; 1979. The Modeling of a Baghouse,
Proceedings of the Vth International Fabric Alternatives Forum;
Phoenix, Arizona, December 5-6. pp. (3-l) - (3-31).
U. Bowen, C.F.P; 1981. Gas Flow Model Study. Proceedings of the
5th International Fabric Alternatives Forum. Phoenix, Arizona,
January 15-16. pp. (l-l) - (l-M).
176
-------
Figure 1. View or deck taken from the southbound lane of Route 7-
Unit 1 fabric filter is in the "background.
Figure 2. View from deck looking towards Unit k fabric filter.
Structural steel and ductwork in middle of the picture
is for Unit 5 precipitator.
177
-------
Figure 5. Typical duct arrangement on Units 1-1*. Fabric filter inlet duct
enters from the left. Fabric filter outlet duct is at top of
picture. Duct entering the chimney is the ID fan outlet duct.
> -
5 -
•
I
.
r
6 3
/\t^fi*\ t. /ceo
Figure 1*. Guaranteed and anticipated fabric filter pressure drop vith
two corrpartments out of service.
178
-------
RESULTS FROM THE FABRIC FILTER
EVALUATION PROGRAM AT COYOTE UNIT #1
H. James Peters, Arthur A. Reisinger and W. Theron Grubb
Wheelabrator-Frye Inc.
Air Pollution Control Division
600 Grant Street
Pittsburgh, Pennsylvania 15219
and
Merrill Lewis
Montana Dakota Utilities
Beulah, North Dakota
ABSTRACT
The fabric filter at Coyote Unit #1 has been in service since May 1981
collecting lignite flyash and sodium-based spray dryer product as the second
stage of the dry scrubbing system installed by the Wheelabrator-Frye
Inc./Rockwell International Joint Venture.
A program is being conducted to evaluate the performance of acrylic,
polyester, Nomex, and fiberglass filtration fabrics in full compartment
trials. The unit uses predominantly acrylic fabric and operates at filter
ratios of typically 3:1 using combination deflation air and mechanical
shaking as the fabric cleaning method.
This paper highlights the results of the fabric evaluation program to
date including comparisons of the pressure drop, throughput, replacement
history, filtration parameters, and changes in physical properties of the
various fabrics over the first year of operation. Discussion of the overall
performance of the unit and economic and technical considerations for fabric
selection is also covered.
179
-------
BACKGROUND
In late 1976, prior to final equipment selection for the Coyote project,
an extensive pilot program was undertaken to demonstrate the use of nahcolite
and a fabric filter as an SO? and particulate collection process. The
program was conducted at the Leland Olds Plant of Basin Electric Power
Cooperative by Wheelabrator-Frye Inc. and Superior Oil under the sponsorship
of the Coyote Project Utilities and their engineer, Bechtel. The resultant
four months of tests proved to be very positive with respect to 502 rernova^
and sorbent utilization as well as particulate collection. S02 removal in
the range of 70-90 percent was achieved at stoichiometric ratios of 1.0 to
1.5(1). This compares favorably with reports (2,3) describing other testing
of dry injection systems using nahcolite and trona.
Early in 1977 it became obvious that the lack of commercial availability
of nahcolite would preclude a commercial installation of the process. As an
alternative to the dry injection system utilizing nahcolite, testing was
conducted by Wheelabrator-Frye and Rockwell International using a spray dryer
upstream of the pilot baghouse as a means of injecting an alkaline material
into the flue gas stream for S02 absorption. This two stage SOp and
particulate removal system (4) proved to be technically and economically
favorable and was selected for the Coyote Station.
PLANT DESCRIPTION
The 410 megawatt Coyote Station which has been on stream since the
second quarter of 1981, is located in the city of Beulah in Mercer County,
North Dakota and is fired with locally mined lignite fuel. The boiler is a
balanced draft cyclone-fired unit with twelve cyclones and steam driven
forced draft fans. The system is designed for lignite fuel having an average
of 7,046 Btu per pound, 7% ash and 0.78% sulfur. During operation flue gas
exits two air heaters and flows through four 46-foot diameter spray
dryers into a 38-compartment fabric filter. Two axial flow induced draft
fans discharge the filtered flue gas to a single stack. The flue gas
temperature at the inlet to the fabric filter is in the range of 210°-
220°F. Gas temperatures beyond fabric capabilities are alarmed and gas
flow is diverted through the fabric filter bypass system.
PILOT TESTING
Testing for the Coyote Station was first conducted in 1976 and examined
the use of nahcolite as a dry absorbent in a pilot fabric filter
incorporating twelve 11-1/2" diameter by 30 ft. bags with combination
deflation air - mechanical shake cleaning. During the nacholite testing and
in subsequent testing of the spray dryer dry scrubbing process, fabric filter
data was gathered on various fiberglass, acrylic, and polyester fabrics and
correlated with boiler and spray dryer operating conditions.
180
-------
The resulting correlations allow estimates of filter fabric pressure
drop during full scale operation through a simple model.
AP
K
L
T
AP = SE V + K L T V2
fabric pressure drop at time T, in w.g.
effective residual drag coefficient, in w.g.-min/ft
filter velocity, ft/min
dustload drag coefficient, in w.g.-ft-min/grain
particulate loading, grains/ftj
elapsed time since cleaning, minutes
The duration of exposure to flue gas during pilot testing was limited to
at most a few months for each fabric, therefore no assessment of long term
performance or bag life could be made. In the course of pilot testing no bag
failures occured which were caused by flue gas conditions. All of the fabric
types tested were deemed acceptable for the service.
COMPARATIVE PRHBURK DROP
COYOTB STATION FABRIC PII.TBR PBRPOIIMANCB
ACTUAL PERFORMANCE
: 6.1 01. ACRYLIC
• OVERALL AVERAGE
300 «0
BOILER LOAD-MW
1 2
FILTER VELOCITY FPM
2 3
FILTER VELOCITY FPM
Figures 1 and 2. Pilot plant and actual operating pressure drop.
Figure 1 shows the predicted fabric pressure drop as a function of filter
velocity for the lightweight acrylic, polyester, and fiberglass fabrics based
on continuous cleaning cycles for the fabric and expected grain loading.
131
-------
Figure 2 shows the predicted fabric pressure drop for the lightweight
acrylic only with representative actual operating points from full scale
operation. The approximate relationship between boiler load and filter
velocity is also indicated. A comparison of the pressure drop versus the
filter velocity from pilot unit correlations and full-scale operation shows
the effectiveness of scale-up from pilot plants.
PROGRAM OBJECTIVES
Although the pilot work with various fabrics had been successful, due
consideration was given to the process as a new application, and a long-term
program to determine the best fabric for this system in terms of operating
and replacement costs was initiated. A recommendation was made to the prime
contractor and the Coyote partners to allow installation of several full-
compartment trials of various candidate fabrics. This program would allow
evalution of the fabrics in this new application and would enhance the
selection of future replacement bags. The program began with the selection
of fabric types and the number of full compartments of each which were to be
installed.
In conjunction with the fabric selection, a monitoring program was
established to review the operation of the entire collector for comparison
with the pilot unit and to perform specific testing of the various fabrics.
The primary objectives of this comprehensive program were to determine the
fabric dimensional stability, filtration performance, and baglife.
FABRIC ALTERNATIVES
The process of selecting the proper fabric for the Coyote Station
baghouse involved considerations of fabric durability, filtration performance
(efficiency, throughput, drag, and cleanability), and economics. Since the
design operating temperature was 180°F, synthetic fabrics were considered
as alternatives to fiberglass fabrics, which were being utilized in all coal-
fired boiler baghouses in the United States at that time. The synthetic
fabrics had a lower cost, potentially better durability in the deflate/shake
cleaning system, and lower operating pressure drop than fiberglass fabrics as
indicated in pilot tests. Fabrics considered as alternatives were those made
of acrylic (Dralon T), aramid (Nomex), and polyester fibers.
Polyester fabric was the most economical choice and performed well in
the short pilot plant tests durations at up to 200°F. Laboratory
exposure to 180°F/40fc RH with and without SOp and with and without
alkaline dust exposure resulted in no appreciable strength loss of the
polyester fabrics. However, there was still a concern that with the high
moisture content of the flue gas and the high alkalinity of the particulate,
if the baghouse were to operate for extended periods of time in the 220-
250°F range, the polyester fiber might be subject to hydrolysis resulting
in severe strength loss and premature bag failure.
182
-------
Oral on T homopolymer acrylic fabric had been successfully used on coal-
fired boiler baghouses in Australia. Acrylic fiber is formed by addition
polymerization and thus is not susceptible to hydrolysis as are polyester and
Nomex, which are formed by condensation polymerization. The design operating
temperature was safely below the maximum temperature rating of 284°F for
Dralon T acrylic. The only concern regarding the use of acrylic fabric was
dimensional stability. Extensibility problems had been previously
encountered with acrylic fabrics in other filtration applications. These
stability problems were discussed with fabric and bag suppliers who generally
agreed that a stable acrylic fabric could be produced by proper heat-setting
and finishing techniques.
Nomex fabric was considered, in spite of its high relative cost, because
of its potential for increased cleanability and lower pressure drop compared
to fiberglass, based on experience in the Southwestern Public Service Co.
Harrington Station Unit #2 baghouse. In addition, Nomex fabric would be able
to withstand the normal flue gas temperatures at the spray dryer inlet.
Because of the alkalinity of the excess soda ash reagent, there was little
concern about acid degradation of the Nomex fabric.
Fiberglass fabric was included based on successful operation in other
coal-fired boiler baghouses and for full evaluation as a replacement fabric
in the unlikely event that spray dryer operation could not provide the
necessary flue gas cooling for the synthetic fabrics. Since deflate/shake
cleaning was to be used, synthetic fibers were expected to provide superior
bag life to fiberglass which has inherently poor flex abrasion resistance.
Fi ber
Type
TABLE 1. NOMINAL PROPERTIES OF INITIAL FABRICS INSTALLED.
Yarn
Type
Weight Finish
(oz/yd2) Finish Type Code
No.
Compartments
Acrylic Spun 6.1
Acrylic Spun 10.0
Nomex Spun 5.5
Polyester Filament/Spun 6.1
Polyester Woolen Spun 12.0
Glass Filament/Text. 10.0
Glass Filament/Text. 13.5
Glass Filament/Text. 13.5
Glass Filament/Text. 13.5
H20 Repel!ant E
H20 Repel 1ant E
Acid Resistant F
Heat-Set G
Acid Resistant F
Teflon B (10%) A
Acid Resistant B
Acid Resistant C
Acid Resistant D
19
12
1
1
1
1
1
1
1
183
-------
Based on all of the above considerations, acrylic was chosen as the
primary fabric for use at Coyote. This choice was based on the expectancy of
lower fabric pressure loss, increased bag life, and reduced cost compared to
fiberglass fabric, without the potential risk of hydrolysis which might be
encountered with polyester fabric. Special fabric finishing and rigorous
quality control testing were utilized to minimize any potential stretching of
the acrylic bags. Two different acrylic fabrics were selected: the 6
oz./yd^ fabric which had been used for all pilot plant and laboratory
evaluations and a 10 oz./yd2 fabric similar to that used successfully for
boiler applications in Australia and in many industrial applications. A list
of the nine initial fabrics selected and their nominal properties is given in
Table 1.
FABRIC FILTER OPERATION
The Coyote fabric filter is cleaned by a combination of deflation air and
mechanical shaking. During cleaning, an individual compartment is isolated
from gas flow and controlled reverse flow is used to collapse the bags to a
partially open configuration. The bags are then cleaned by shaking action
for 10-15 seconds, time is allowed for dust to settle, bags are gently
reinflated and the compartment is placed back into filtration service.
In normal operation, overall fabric pressure drop (including inlet and
outlet manifold and housing losses) is maintained by pressure drop
initiation of the cleaning cycle at 6" w.g. At low loads, where the 6"
setpoint would not be reached within several hours, cleaning cycles are
initiated on 3-6 hour intervals. This maintains an average dust loading on
the fabric and allows the fabric filter to easily accommodate rapid boiler
load increases. Continuous cycles (each compartment cleans after one hour
filtering time) are used at full boiler load to maintain approximately 4"
fabric and 6" overall pressure loss at 3 ft/min filter velocity.
The fabric filter contributes significantly to the sulfur dioxide
removal as the second stage of the flue gas desulfurization process.
Residual sodium carbonate leaving the spray dryer continues to react with
SOg during filtration. The fabric filter typically accounts for as much as
20 percent of the overall S02 removal.
The fabric filter performance as a particulate collection device is as
anticipated. Particulate emissions have been measured and are well below the
required new source performance standards.
In order to properly evaluate the overall performance of the fabric
filter and the relative performance of the various fabrics, data
on individual compartment pressure drop and relative flow was collected at
periodic intervals.
Under steady operating conditions the pressure loss across the fabric
was measured for each of the compartments immediately before and after the
compartment was isolated for cleaning. Together with the individual
184
-------
compartment AP data, sufficient information on boiler and spray dryer
operation was taken to estimate the total gas flow and participate loading
to the fabric filter.
Twelve compartments, including the nine alternative fabrics plus three
additional acrylic compartments were outfitted with pressure taps 'to measure
the pressure loss across the outlet valve opening. This measurement allows
the flow rate through the compartment to be determined. To ensure meaningful
comparisons the data was coordinated with the fabric filter cleaning cycle
and taken both immediately before and after cleaning. The outlet valve AP
data enables relative flow comparisons between any of the test fabrics. In
combination with total gas flow data, a weighted average of the data allows
the filter velocity for each of the fabrics to be established.
This individual compartment flow can also be compared with flow rates
calculated directly from the outlet valve AP value based on generalized flow
correlations developed during scale model flow studies of the outlet duct and
valve air flow configuration. Good agreement was found between the model
study correlation and flow rates derived from system operation and weighted
average data.
With individual compartment AP and filter velocity data available as
well as system operating data, effective residual drag and dust load drag
coefficients can be determined for each individual fabric as had been done
during pilot testing.
RESULTS OF FABRIC TESTING
During scheduled boiler outages, test compartments were inspected and
sample bags removed for measurement of fabric strength and other fabric
properties. In addition, control bags were measured to track dimensional
stability, and several bags in each test compartment were tested for filter
drag (5) and residual dust weight.
The results of Mullen Burst testing are presented in Figure 3. Results
from measurements of M.I.T. flex vs. length of service for the fiberglass
fabrics are given in Figure 4 and for the synthetic fabrics in Figure 5.
Note that except for the polyester fabrics, the strength as measured by both
tests stabilized at levels indicating good potential baglife. Of particular
interest is the M.I.T. flex tests performed on the synthetic fabrics. This
type of testing is typical for fiberglass filtration fabrics but not usually
performed on synthetic bags. The M.I.T. flex values were a clear indicator
of the progressive hydrolysis of the polyester which eventually led to
failure and replacement. The absolute value of the M.I.T. flex test rather
than the percentage loss appears to be a better indicator of potential bag
failure.
Fabric in-situ drag measurements and residual dust weights were for
the most part in agreement with pressure drop and relative flow measurements
185
-------
taken during operation, although the state of the bags at the time actually
tested (elapsed time since cleaning, length of time out of service, ambient
conditions, etc.) was not necessarily the same for each fabric style.
The lightweight acrylic and Nomex fabrics consistently indicated the
highest relative flow (throughput), low drag values, and low residual dust
weights. The 10 oz. acrylic and all of the fiberglass fabrics operated at
higher drag, residual dust, and pressure drop with lower throughput.
Of interest is the low residual dust loadings found for the fabrics.
Residual dust weight for the lightweight acrylic was as low as 7 Ibs. This
can be contrasted with as much as 50-100 Ibs. for fiberglass fabrics in
reverse air collectors filtering boiler flyash. The shaking action on the
lightweight fabric maintains a low residual dust, and at the same time
particulate collection efficiency is maintained at a high level. Residual
dust weights for the 10 oz. acrylic and fiberglass fabrics in the Coyote
fabric filter were roughly 50% greater than that of the lighter acrylic.
1
i
1W
MULLEN BURST TESTING
200
LENGTH Of SERVICE (DAYII
100 200
LENGTH Of SERVICE (DAYS)
Figure 3. Results of Mullen Burst tests for synthetic and fiberglass fabric.
Overall performance of the lightweight acrylic fabric has been superior
to the other test fabrics. Although a substantial decrease in Mullen burst
strength of 50-60% was observed during the initial six months of operation,
the fabric physical properties have stabilized at a level which should result
in more than adequate bag service life. The few bag failures which did occur
were due to random physical damage during installation and minor construction
modifications.
186
-------
The Nomex fabric has, also performed very well to date. Fabric strength
has stabilized at a level which indicates good potential baglife. A full
compartment of Nomex has been in service for 27 months at the Southwestern
Public Service Harrington #2 baghouse. The Nomex fabric has exhibited
excellent cleanability, low pressure drop, and low filter drag as determined
during single bag testing. The relative flow has been slightly lower than
the lightweight acrylic based on compartment outlet valveAP measurements.
The 10 oz/yd^ acrylic fabric has exhibited consistently higher
pressure drop than the lightweight acrylic fabric. Single bag filter drag
was higher than for the other fabrics and compartmental relative flow was
lower than the 6 oz/yd2 acrylic. The bulky yarn construction and
resistance to shake cleaning of the heavier weight fabric apparently result
in poor cleanability.
Fiberglass fabrics exhibit excellent strength retention after twelve
months operation, as would be expected at the low operating temperature
(relative to the 550°F capability of glass fabrics) and moderately alkaline
conditions to which they have been exposed.
MIT FLEX TESTING
LENGTH Of SERVICE IDAYSI
LENGTH OF SERVICE IOAYSI
Figure 4. Results of M.I.T. flex tests for fiberglass fabrics.
Polyester fabrics experienced drastic loss of strength and high rate of
failure due to chemical degradation of the polyester fiber. Exposure to 240-
250°F operating temperatures in the presence of the typically high flue gas
moisture content and moderate alkalinity resulted in hydrolysis of the
fiber. If it had been known that actual operating temperatures would be 230-
250°F rather than the 180°F design, polyester fabrics would likely have
been omitted as candidates for evaluation.
187
-------
A
MIT FLEX TESTING
L
ACRYLIC FABRICS
O !"• -
MIT FLEX TESTING
LENGTH OF SERVICE JOAVSI
I I
IN a*
LENGTH OF SERVICE (DAVS*
I
no
POLYESTER ««t
LENGTH OF SERVICE (DAYS)
LENGTH OF SERVICE IDAVSI
Figure 5. Results of M.I.T. flex test for synthetic fabrics.
183
-------
The overall performance of the Coyote filtration fabrics has been
excellent. Excluding the polyester fabrics only 35 failures have occurred in
the first eighteen months for a failure rate of less than 0.5%. These
failures (detailed in Table 2) are all due to miscellaneous physical damage
which occurred during installation or as a result of minor construction
modifications required inside the compartments after bag installation.
TABLE 2. BAG FAILURE SUMMARY.
Start-up through September 8, 1982 (excluding polyester bags*):
Primary bag failures - 33
Secondary failures (damage by adjacent bag failure) - 2_
Total over sixteen months - 35
Primary failures by fabric type:
Acrylic
6 oz.
10 oz.
Fiberglass
10 oz. - Finish A
13.5 oz. Finish B
13.5 oz. Finish C
13.5 oz. Finish D
Nomex
5.5 oz.
Total
Number failed
16
11
1
0
0
4
I
33
Percent failed
0.41
0.45
0.49
0
0
1.96
0.49
0.45
Causes of bag failure:
Specific physical damage:
Abrasion against internals 18
Damage to bottom during installation 7
Top cuff damaged by bag clamp 2
Cause undetermined 6
Secondary failures: 2
Total 35
*Note: Both types of polyester bags were severely damaged due to hydrolysis of
the fiber. Prior to total replacement 25 12 oz. bags and 29 6 oz. bags failed.
189
-------
ACTUAL VS DESIGN
The fabric filter was designed for a nominal air-to-cloth ratio of 3-to-
1 with two compartments cleaning and two compartments off-line for
maintenance. Tests performed during the past eighteen months indicate that
air-to-cloth ratios in a range of 2.5 to 3.8-to-l have been achieved. During
this time the fabric filter has maintained pressure losses of 4-8 inches
(including manifold and housing losses) with stable pressure drop at each
flow rate. During periods of reduced boiler load, the fabric filter pressure
loss operates at the previous values for that specific load. Although
fabric air-to-cloth ratios have been higher than anticipated the fabric
filter has consistently operated without difficulty. Filtration performance
(pressure drop, cleanability, and efficiency) has remained relatively stable
in spite of a wide variety of boiler and FGD system operating conditions.
Pressure excursions, due to boiler load swings, uneven gas distribution from
spray dryers, fabric filter control or equipment malfunctions, etc., have
only been temporary, and when system operation returned to normal, so did
fabric pressure drop.
The higher temperature operation (230-250°F operation for much of the
first twelve months compared to 180°F design) resulted in discoloration of
the acrylic fabric as would be expected, but had no serious effect on fabric
strength, expected service life or dimensional stability. The higher
temperature did accelerate the hydrolysis failure of the polyester fabrics
leading to their replacement with acrylic and at this time polyester is not
considered a viable alternative for this application.
ECONOMICS
The use of acrylic fabric in lieu of fiberglass offers significant
economic advantages due to:
o Lower fabric cost.
o Lower pressure drop at equivalent throughput.
o Less possibility of installation damages.
o Potentially longer bag life.
Bags fabricated of the six ounce acrylic fabric are priced 20% less than
a nominal ten ounce fiberglass bag with 10% teflon finish yielding a clear
economic advantage for the acrylic in capital costs. In addition, for the
Coyote operation, assuming one inch of pressure loss is equivalent to
$50,000 per year operating costs, lower pressure drop operation reflects
significant savings. Longer fabric life may also be realized for the acrylic
fabric.
Nomex can be considered a viable alternative fabric. However, the
economics indicate that an additional year of baglife would be required of
190
-------
the Nomex assuming a three-year life for the acrylic, and also a one inch
lower pressure loss. This remains to be substantiated through
continued operation.
OUTLOOK
The fabric filter evaluation program at Coyote will continue through the
life of the initial set of bags. Modifications to the cleaning action for
the fiberglass bags are being considered in order to improve their relative
performance. The basic objective of identifying an optimum fabric selection
for eventual rebagging of Coyote has been satisfied by the superior
performance of the lightweight acrylic.
After eighteen months of fabric filter operation at Coyote with superior
performance in terms of low pressure drop at high filter velocities, fabric
replacement experience and service life expectancy the combination deflation -
mechanical shaker type fabric filter using synthetic bags is demonstrated to
be an exceptional selection for the dry scrubbing application. In a side-by-
side comparison the general serviceability of fiberglass bags with shaker
cleaning is also demonstrated. Fabric filter operating temperatures can be
maintained below the upper limits for the synthetic fabric even during
periods of off-normal operation by the spray dryers. The ability to use
acrylic instead of fiberglass offers significant economic advantages and
potentially longer fabric life.
The work described in this paper was not funded by the U.S. Environmental
Protection Agency and therefore the contents do not necessarily reflect the
views of the Agency and no official endorsement should be inferred.
REFERENCES
1. Estcourt, V. F., Grutle, R.O.M., Gehri, D. C., and Peters, H. J., "Tests
of a Two-Stage Combined Dry Scrubber/S02 Absorber using Sodium or
Calcium", American Power Conference, Chicago, Illinois, April 1978.
2. EPRI Report FP-207, "Evaluation of Dry Alkalies for Removing Sulfur
Dioxide from Boiler Flue Gases", October, 1976.
3. Muzio, L. J., Sonnichsen, T. W. et al., "Demonstration of S02 Removal
on a Coal-Fired Boiler by Injection of Dry Sodium Compounds", EPA/EPRI
Symposium on Flue Gas Desulfurization, Hollywood, Florida, May 1982.
4. U.S. Patent No. 4,197,298, "Sequential Removal of Sulfur Oxides from Hot
Gases" issued to Wheelabrator-Frye Inc. and Rockwell International.
5. Grubb, W. T. and Banks, R. R., "Field Evaluation of the Drag of
Individual Filter Bags", 4th Symposium on the Transfer and Utilization
of Particulate Control Technology, Houston, Texas, October, 1982.
191
-------
BAGHOUSE PERFORMANCE AND ASH CHARACTERIZATION AT THE
ARAPAHOE POWER STATION
by: Robert S. Dahlin
Southern Research Institute
Birmingham, Alabama 35255
D. Richard Sears
U.S. Department of Energy
Grand Forks, North Dakota 58202
George P. Green
Public Service Company of Colorado
Denver, Colorado 80202
ABSTRACT
This paper presents the results of a field test conducted in March, 1981,
on the Unit 3 baghouse of the Arapahoe Station of the Public Service Company
of Colorado. The unit was burning a subbituminous coal from Routt County,
Colorado, and was retrofitted with a baghouse in 1979. Baghouse performance
was found to be excellent with an overall mass efficiency of 99.98 percent.
The estimated cumulative collection efficiency of all particles smaller than
two microns was 99.92 percent.
The average electrical resistivity of the fly ash was found to be 6 x 10
ohm*cm at 266°F. This was consistent with the low level of SOs found in the
flue gas (^ 0.3 ppm). The measured resistivity agreed reasonably well with
that predicted by Bickelhaupt's technique using the analysis of the ash ob-
tained by atomic absorption spectrometry. These results suggest that a high
SCA and possibly flue gas conditioning would be required to obtain the same
collection efficiency from an ESP.
A microanalytical characterization of ash elemental and mineral composi-
tion has been performed. Using SEM, detailed major element analyses for ^ 240
individual ash particles have yielded concentration frequency distributions,
composition as function of particle size, and inter-element concentration
correlations. The results have been correlated with coal and ash mineralogy.
Suggestions are made concerning possible application of this data to practical
problems of fly ash control and disposal.
192
-------
INTRODUCTION
The Grand Forks Energy Technology Center of the U.S. Department of Energy,
through its contractor Southern Research Institute, is conducting a series of
field tests to study the characteristics and control of fly ash from the com-
bustion of low-rank Western coals. Emphasis is being placed on two types of
coal: Western subbituminous coal and Texas lignite. The overall objective of
the project is to provide data that will be useful in evaluating particulate
control technologies for these coals.
To date, two field tests have been performed. The first field test, the
subject of this paper, was conducted in March 1981, at Unit 3 of the Arapahoe
Power Station of the Public Service Company of Colorado. The unit, retrofitted
with a baghouse that was placed in service in May 1979, was burning a low-
sulfur subbituminous coal from Routt County, Colorado. The second field test,
which was jointly sponsored by DOE and EPRI, was completed in May 1982 at the
San Miguel Station of the San Miguel Electric Cooperative located between
San Antonio and Corpus Christi, Texas. Results of the San Miguel test will be
reported elsewhere.
OBJECTIVES AND SCOPE
The primary objectives of the Arapahoe test were to evaluate the per-
formance of the baghouse and characterize the fly ash. This information will
be used, along with information from other power plants, in evaluating various
particulate control technologies for low-rank Western fuels. In this case,
the fuel is a low-sulfur Western subbituminous coal. Baghouse performance was
evaluated in terms of mass efficiency, fractional efficiency, and pressure
drop. The fly ash was characterized in terms of its particle size distribution,
chemical composition, and electrical resistivity.
The scope of the test program encompassed the following elements:
(1) Determination of inlet and outlet mass loadings and mass
efficiency by EPA Method 17,
(2) Measurement of inlet and outlet particle size distributions
and fractional efficiency using calibrated cascade
impactors,
(3) Documentation of boiler and baghouse operating conditions
during testing,
(4) Sampling and analysis of coal by ASTM methods,
(5) Sampling and analysis of fly ash using glass fiber thimbles
and a five stage cyclone assembly for sampling with analysis
of major elements by Atomic Absorption Spectrometry (AA) and
minor elements by Spark Source Mass Spectrometry (SSMS),
193
-------
(6) Analysis of major flue gas components by Orsat apparatus with
moisture determined by adsorption on Drierite dessicant,
(7) Determination of ash resistivity using an in situ point-
plane probe,
(8) Analysis of particulate and gaseous organics by gas
chromatography (GC) and high performance liquid chromatog-
raphy (HPLC) of extracts of fly ash and XAD-2 resin exposed
to flue gas in a SASS train,
(9) Measurement of SOa and SO3 (H2SOt») levels in the flue gas by
the Cheney-Homolya method,
(10) Collection of two metric tons of coal for use in the pilot-
scale Particulate Test Combustor at Grand Forks,
(11) Mineral characterization of the coal by x-ray diffraction
(XRD) of as-received coal and of oxygen plasma low tempera-
ture ashed (LTA) coal, as well as by chemical fractionation,
(12) Major element analysis of coal and size-fraction-
ated ash by neutron activation analysis (NAA), and
(13) Detailed microanalytical characterization of size fraction-
ated ash using scanning electron microscopy/electron micro-
probe (SEM).
Space limitations preclude detailed presentation of the entire effort here.
Therefore, several phases of this study will appear elsewhere. See, for
example, references 13 and 14 for a more complete exposition of the micro-
analytical results.
SITE DESCRIPTION
BOILER
The Unit No. 3 steam generator at Arapahoe is a vertically down-fired, dry-
bottom boiler supplied by Babcock and Wilcox. The unit was placed in service
in 1951, and now operates as an intermediate peaking unit and used for cycling
duty. Nominal generating capacity is 46 megawatts. The design specifications
and nominal operating conditions of the boiler are summarized in Table 1.
FUEL
At the time of the testing, the Unit No. 3 boiler was burning a low-
sulfur subbituminous coal from Routt County, Colorado. Proximate and ultimate
analyses of the coal performed after completion of the testing revealed that
the coal was of somewhat higher quality than anticipated, being borderline
between a subbituminous A and a high volatile C bituminous coal. Proximate
and ultimate analyses of daily composite coal samples are given in Table 2.
194
-------
TABLE 1. BOILER DESIGN AND NOMINAL OPERATING PARAMETERS
Manufacturer
Generating Capacity
Year On-Line
Duty Cycle
Firing Method
Slagging Method
Steam Header Temperature
Steam Header Pressure
Fuel
Heating Value
Moisture
Ash
Volatile Matter
Fixed Carbon
Sulfur
Excess Combustion Air
Fuel Feed Rate
Combustion Air Flowrate
Babcock and Wilcox
46 Megawatts
1951
Cycling*
p-c vertical (down-fired)
Dry bottom
900°F (482°C)
870 psig (6100 kPa)
Western subbituminous coal
10,600 BTU/lb (5890 kcal/kg)
8%
10%
40%
42%
0.5%
15%
46,800 Ib/hr (21,200 kg/hr)
450,000 Ib/hr (204,000 kg/hr)
*Generally base-loaded Monday through Friday; boiler designed
for quick start-up and unit taken off line during low-load
periods.
TABLE 2. PROXIMATE AND ULTIMATE COAL ANALYSES
(AS RECEIVED)
Date
Sampled
3/3/81 3/4/81 3/5/81 Average
Proximate
% Ash
% Volatile
% Fixed Carbon
BTU/lb
% Sulfur
4.62
9.31
43.63
42.44
10,738
0.49
13.38
10.46
34.82
41.33
10,454
0.48
6.19
9.94
41.74
42.13
10,786
0.48
8.06
9.90
40.06
41.97
10,659
0.48
195
-------
TABLE 2. PROXIMATE AND ULTIMATE COAL ANALYSES
(AS RECEIVED) (Cont' d)
Date
Sampled
3/3/81 3/4/81 3/5/81 Average
Ultimate
% Moisture
% Carbon
% Hydrogen
% Nitrogen
% Chlorine
% Sulfur
% Ash ,
% Oxygen
4.62
61.02
5.15
1.47
0.07
0.50
9.32
17.85
13.38
59.75
4.19
1.39
0.07
0.49
10.46
10.27
6.19
60.62
4.95
1.38
0.04
0.48
9.93
16.41
8.06
60.46
4.76
1.41
0.06
0.49
9.90
14.84
//Analyses performed by Commercial Testing and Engineering,
Inc. using standard ASTM methods.
tBy difference.
Perhaps the most noteworthy feature is the low sulfur content (^ 0.48%). The
high moisture content on March 4 was believed to be due to snowfall during the
preceding night. Otherwise, the composition of the coal appears to have
remained fairly constant during testing.
BAGHOUSE
Unit No. 3 was retrofitted with a Joy-Western baghouse which was placed
in service in May 1979. The baghouse is a reverse-air unit with Teflon-coated
fiber glass bags. The design air-to-cloth ratio (A/C) was 2.0 acfm/ft2 (0.010
m/s) with all compartments on line and 2.16 acfm/ft2 (0.011 m/s) with one
compartment out for cleaning, at a design gas flow of 315,000 acfm (149 m3/s).
Actual gas flows measured during the testing were significantly lower, with an
average of 238,000 acfm (112 n»3/s), resulting in A/C values of 1.51 acfm/ft2
(0.0077 m/s), and 1.63 acfm/ft2 (0.0083 m/s) with all compartments on line and
one compartment out, respectively. The discrepancy between the design and
measured gas flows is a direct result of the conservative baghouse design.
The baghouse is divided into 14 compartments with 236 bags in each
compartment. The bags are 8 in (20 cm) in diameter and 22 ft (6.7 m) in
height. A schematic drawing of the baghouse is shown in Figure 1. The
drawing also shows the locations of the inlet and outlet test ports. Table
3 gives a summary of the design parameters of the baghouse.
196
-------
STACK
OUTLET
TEST PORTS (8)
FLUE GAS FROM
NO. 3 BOILER
BAGHOUSE
/
WxMAV
\^yr
INLET TEST PORTS (4)
ID FANS (2)
Figure 1. Schematic Side View of Baghouse
TABLE 3. BAGHOUSE DESIGN PARAMETERS
Supplier
Year On-Line
Design Gas Flowrate
Design A/C Ratio
(With One Compartment Off-Line)
Cleaning Method
Design Pressure Drop
Fabric
No. of Compartments
No. of Bags Per Compartment
Bag Diameter
Bag Length
Efficiency Guarantee
Joy-Wes tern
1979
315,000 acfm @ 290°F (149 m3/s @ 143°C)
2.0 acfm/ft2 (0.010 m/s)
2.16 acfm/ft2 (0.011 m/s)
Reverse Air
6" H20 (11.2 mm Hg)
Teflon Coated Fiberglass
14
236
8 in (20 cm)
22 ft (6.7 m)
99.8 Percent
197
-------
During the initial operation of the baghouse, minor discrepancies were
noted such as thimble leaks and valve seatings that required adjustments.
However, these repairs were expeditiously made and initial performance tests
conducted.
After approximately two months of operation, the flange to flange pres-
sure drop exceeded design to the point that reductions in unit load were
required. In an effort to keep the baghouse on line, hand cleaning and beat-
ing of the bags became necessary on a regular basis to remove cake build—up
on the inside of the bags. Several attempts were made by plant personnel to
correct the pressure drop problem, such as replacing the original woven fiber-
glass teflon coated bag with a teflon B coated bag, and burning different
types of coal. Attempts were also made to optimize cleaning cycles and model
flow distribution patterns.
All efforts at optimizing the cleaning cycle, including adjusting
reverse air periods, pressure-initiated cleaning cycles and continuous clean-
ing cycles, and installing the teflon B bags, did not improve the problem
associated with the high flange to flange pressure drop. Ultimately, the
pressure drop problems were resolved by the installation of sonic cleaning
devices (1).
PROCEDURES AND RESULTS
MASS LOADING AND MASS EFFICIENCY
Inlet and outlet mass loadings were determined by isokinetically collect-
ing suspended fly ash on in situ filters using the equipment and procedures
specified in EPA Method 17 (2). Although Method 5 is usually required for
compliance testing, Method 17 was deemed to be more appropriate for mass
efficiency determinations. Method 17 gives a better representation of the
true particulate concentration at actual flue gas conditions because it avoids
the possibility of new particle formation by condensation. Each sampling run
involved a traverse of the entire duct area.
The measured mass loadings and efficiencies are given in Table A. Based
on a total of six runs, the mean inlet and outlet loadings were 3.20 gr/dscf
(7330 mg/m3) and 0.00055 gr/dscf (1.26 mg/m3). The standard deviations on the
inlet and outlet loadings were 15.7% and 23.6% of the mean values, respective-
ly. The mean mass efficiency was 99.983%, corresponding to a mean penetration
of 0.017%. The standard deviation on the penetration was 19.0% of the mean
value. Thus, at the 90% confidence level, the penetration is in the range
of 0.012 to 0.023%, or the efficiency is in the range of 99.977 to 99.988%.
This corresponds fairly well to the range of measured efficiency values.
The outlet mass loadings were comparable to those reported by Ensor
et al (3) for the Kramer Station fabric filter, which was operating at an A/C
value of 1.67 acfm/ft2 (0.0085 m/s). The mass efficiencies were well above the
design value, certainly due in part to the conservative design of the baghouse.
The average emission rate was calculated to be approximately 0.001 lb/10 BTU
(0.43 yg/J). The stack opacity recorded by the plant visiometer was essential-
198
-------
ly zero during all testing.
during all testing.
Unit No. 3 was running at essentially 100% laod
TABLE 4. MEASURED MASS LOADINGS AND EFFICIENCIES
t
Run
No.
1
2
3
4
5
6
Flue gas
Flowrate
acfm (m /s)
229,000 (108)
231,000 (109)
241,000 (114)
241,000 (114)
242,000 (114)
241,000 (114)
Temp
. Mass Loading
Inlet
°F (°C) grains/dscf
262
271
273
268
264
258
(128)
(133)
(134)
(131)
(129)
(126)
3.
3.
3.
2.
3.
2.
13
35
84
45
59
86
(7170)
(7670)
(8790)
(5610)
(8220)
(6550)
Outlet
(mg/m3)
0.00066
0.00044
0.00058
0.00036
0.00071
0.00055
Baghouse
Efficiency
(1.
(1.
(1.
(0.
(1.
(1.
5)
0)
3)
82)
6)
3)
99
99
99
99
99
99
7,
.979
.987
.985
.985
.980
.981
tObtained by EPA Method 17.
PARTICLE SIZE DISTRIBUTION AND FRACTIONAL EFFICIENCY
Inlet and outlet particle size distributions were measured using modified
Brink impactors at the inlet and University of Washington Mark III impactors
at the outlet. Eight real impactor runs and two blank runs with filters
ahead of the impactors were performed in separate sampling ports at the inlet.
At the outlet, three complete traverses of duct area were performed, with no
blank runs. The blank runs at the inlet did not indicate any appreciable
substrate interference problems with the flue gas. Glass fiber substrates
were used at the inlet and greased metal foils were used at the outlet.
The raw impactor data were reduced using an updated version of the
Cascade Impactor Data Reduction System (CIDRS) originally described by
Johnson, et al (4). The resulting inlet and outlet size distributions are
given in Table 5 on the basis of cumulative percentages of particulate mass
contained in particles smaller than various diameters. Based upon these
results, it appears that submicron particles account for about 0.84% of the
particulate mass at the Inlet. At the outlet, about 7.7% of the particles by
mass are smaller than 1 ym. The mass median diameter (mmd) of the inlet dis-
tribution is beyond the upper particle size limit of the modified Brink
impactors. However, an extrapolation of the data using an osculating poly-
nomial (5) suggests that the inlet mmd is about 20 ym. The mmd of the outlet
distribution is about 4.5 ym.
Inlet and outlet differential mass distributions (AM/AlogD versus D) were
also computed using the same data reduction system. Fractional penetration
values were computed as the ratio of outlet to inlet values of AM/AlogD at
various particle diameters. Figure 2 shows the fractional efficiency curve
constructed from these calculations. The calculated fractional efficiency
199
-------
TABLE 5. INLET AND OUTLET PARTICLE SIZE DISTRIBUTIONS
Particle Diameter
(van)
0.4
0.6
0.8
1
2
4
6
8
Cumulative Percentage of Particulate Mass
Contained in Particles Smaller Than
Indicated Diameter (%)
Inlet Outlet
0.36
0.48
0.64
0.84
3.0
12
18
23
2.1
3.8
5.7
7.7
19
43
64
79
•fAverage of eight inlet runs with Brink impactors and three outlet
runs with University of Washington Mark III impactors.
99.99
o
\L
u.
u
99.9
TT
0.5 1.0 2.0 5.0 10
DIAMETER (urn) 4.51-4
Figure 2. Fractional Efficiency Curve.
200
-------
decreased steadily from 99.988% at 10 ym to 99.87% at 0.5 ym. The impactor
data are not reliable below about 0.5 ym. Despite the apparent sensitivity to
particle size, the baghous'e still attained an estimated cumulative collection
efficiency of 99.92% for all particles smaller than 2 ym. In absolute terms,
this is well in excess of the fine particle collection efficiencies typically
obtained with most conventional control equipment.
FLY ASH ANALYSIS
Fly ash from an inlet mass train thimble was ignited and then digested by
standard acid attack for analysis by atomic absorption spectrometr The
results are given in Table 6. The results are reported as the we_ ..it percent-
ages of the elements as their oxides, but this is not intended to imply that
they actually occur in this form. Based on the relatively low alkali metal
content of this fly ash, the ash would be expected to exhibit an inherently
high electrical resistivity in the absence of any conditioning effects from
flue gas species.
TABLE 6. CHEMICAL ANALYSIS OF FLY ASH OBTAINED BY ..ATOMIC ABSORPTION
SPECTROMETRY ON A DIGESTED SAMPLE"'
(weight percentage as oxide)
Li20 0.03
Na20 0.46
K20 1.3
MgO 1.4
CaO 5.1
Fe203 3.6
A1203 25.1
Si02 57.4
Ti02 1.3
P205 1.0
S03, 0.68
LOI 1.8
Total 99.17
#The sample was taken from the thimble of a mass train which isokineti-
cally sampled the flue gas at the inlet of the baghouse.
tLoss on ignition.
201
-------
FLUE GAS COMPOSITION
The major flue gas components were determined by the standard Orsat
method with water vapor determined gravimetrically by adsorption on Drierite
dessicant. The average composition of the flue gas was as follows: 73% Na,
4.8% Oa, 13.5% COa, and 8.7% HaO. The oxygen level was somewhat higher than
expected from the amount of excess air reportedly being used in the boiler,
but this difference could be attributable to air inleakage ahead of the
sampling ports.
Sulfur oxides were determined by the method of Cheney and Homolya (6).
Sulfuric acid, which was originally present as either SOa or HaSOi*, was
collected by selective condensation in a temperature-controlled tube packed
with a Pyrex glass plug. Sulfur dioxide was captured by absorption in a
solution of 3% HzOa. The recovered condensate and the spent peroxide solution
were analyzed for sulfate content by titration. Two of the eight analyses
were performed by ion chromatography as a check on the titration. The two
analytical techniques gave results that agreed to within an average percent
difference of 15.8%.
The results of eight sampling runs yielded a mean SO3 (HaSOO concentra-
tion in the flue gas of 0.29 ppmv with a standard deviation of 0.039 ppmv.
The mean SOz concentration was 391 ppmv with a standard deviation of 11.5 ppmv.
At the concentration levels found in these analyses, the SO3 (HzSOi*) would be
expected -to have very little effect on the electrical resistivity of the ash.
ELECTRICAL RESISTIVITY OF ASH
Although the collector in this case is a baghouse, the electrical resis-
tivity of this ash may be of interest to utilities considering various types
of collectors for a new plant burning a similar coal. Electrostatic effects
play an important role in current theories of filtration, so it is also
reasonable to expect that the resistivity of the ash may influence its filtra-
tion characteristics. Indirect evidence of this effect has recently been
obtained in an EPRI-sponsored study of fabric filtration (7).
An in situ point-plane probe was used to measure ash resistivity by the
sparkover method. This procedure has been described in detail elsewhere (8).
A series of nine measurements were made at the inlet sampling location. The
mean value of sparkover resistivity was 6 x 1011 ohm*cm with a standard
deviation of 4 x 1011 ohm»cm. The considerable scatter in the data is partly
due to temporal and spatial fluctuations of flue gas temperature and SO3 and
moisture levels.
The measured resistivity was compared to that predicted by the technique
of Bickelhaupt (9) using the ash composition given in Table 6 and assuming no
conditioning effect of the SOa. At the mean flue gas temperature during the
in situ resistivity measurements (266°F/130°C) the predicted resistivity was
5 x 1011 ohm*cm at a flue gas moisture level of 8.7%. Thus, the predicted
value of resistivity agreed with the mean in situ measurement to within 20%
despite the considerable scatter in the in situ data.
202
-------
PRESSURE DROP
The pressure drop across the baghouse was monitored throughout the test
program. However, these pressure drop data were probably not representative
of typical baghouse operating conditions at Arapahoe since the bags were
cleaned during a plant outage that occurred about two weeks prior to the test
program. Although the measured pressure drops were well below the design
value of 6 in HaO (11.2 mm Hg) , excessive pressure drop has been a problem in
the past as previously noted. During the test, the pressure drop typically
varied from about 3.4 in HaO (6.4 mm Hg) near the end of the cleaning cycle
up to about 5.0 in H20 (9.3 mm Hg ) near the start of the cleaning cycle.
The average pressure drop was estimated to be about 4.3 in HaO (8.0 mm Hg).
DISCUSSION OF FIELD TEST RESULTS
The low-sulfur Routt County coal burned at Arapahoe produces a low-alkali,
inherently resistive ash. In the absence of flue gas conditioning, a
relatively large precipitator would be required to attain collection effi-
ciencies comparable to those achieved by the Arapahoe baghouse. A comparable
collection efficiency was reported by Lodge Cottrell (10) for the Dave Johnston
Unit 1 cold-side precipitator without flue gas conditioning. The measured
collection efficiency was 99.96% at a specific collection area (SCA) of
661 ft2/1000 acfm (130 m2/m3/s). The sulfur content of the coal was in the
range of 0.4 to 0.7%. With flue gas conditioning, the Pleasant Prairie Unit 1
precipitator achieved a measured efficiency of 99.96% (11). The SCA was 440
ft2/1000 acfm (86.6 m2/m3/s) and the fuel was a low-sulfur (0.37%) Powder
River Basin coal. The Pleasant Prairie precipitator was equipped with a
sulfur burner that supplied roughly 5 ppmv of SO3 to the flue gas ahead of
the ESP. These results suggest that comparable efficiencies may be attainable
in relatively large precipitators without conditioning and in somewhat smaller
precipitators with conditioning. Efficiencies of 99.98% for the baghouse and
99.96% for the ESP correspond to penetrations varying by a factor of two.
At the Arapahoe Station, however, baghouse penetrations also varied by a
factor of two (0.012 - 0.023%) at the 90% confidence level. Consequently
the baghouse and ESP efficiencies quoted above must be regarded as comparable
within experimental uncertainty.
Excessive pressure drop, although not documented during this test, has
been a problem at Arapahoe as it has at a number of other baghouse installa-
tions. The causes of this problem are generally not well understood. This
is clearly an area in which additional research is required. The effects of
electrostatic phenomena and the interactions of the filter cake with various
flue gas species need to be investigated.
LABORATORY CHARACTERIZATION OF ARAPAHOE ASH
Baghouse inlet ash samples comprising five size-fractionated cuts were
obtained using the 5-stage series multicyclone sampling/sizing system
developed by SoRI (12).
203
-------
Laboratory investigation of the multicyclone cuts included:
• Major element composition
• Particle size distribution within the cuts
• Scanning electron microscope investigation (including
electron microprobe) study of particle size, morphology,
and individual particle composition
• Statistical analysis of SEM data to develop inter-element
correlations and composition-size correlations
• Auger spectrometry study of individual particles including
application of secondary ion mass spectrometry (SIMS) and
electron spectroscopy for chemical analysis (ESCA)
Major element analyses of the first four multicyclone stages were performed
by x-ray fluorescence. No significant departures from bulk composition were
observed (cf. Table 6 for the latter).
A quite different approach is provided by scanning electron microscopy/
electron microprobe (SEM).* SEM analysis can be applied to individual ash
particles. If sufficient particles are analyzed, the data can be subjected to
on-line statistical analysis. Also, morphological information obtained by
SEM may have relevance to ash collectability. E.g. Arapahoe ash in the >6.5 y
range is composed predominantly of typical spherical ash particles solidified
from the molten state. Submicron Arapahoe ash, however, contains frequent
hollow pleurospheres, particles with surface irregularities, and numerous
non-spherical particles. The density and the surface interactions of such
particles will differ from those of the larger, smooth spheres. This in turn
may be important in filter cake formation and cohesion.
SEM determination of particle size and shape and SEM-electron microprobe,,
analysis of major element composition were done on approximately 50 particles
from each of the five multicyclone stages. The data obtained from all five
stages were combined to form a population of ^ 240 particles and the data were
analyzed using the SEM's on-line computational system.
Because of space limitations, we can present here only a few examples of
the detailed composition information and correlation analyses. Additional
*GFETC employs a JEOL JSM 35 scanning electron microscope. X-rays are detec-
ted with a Kevex lithium-drifted silicon detector and the elemental analysis
is performed by means of a Tracer Northern NS-880 x-ray analyzer. (Reference
to specific brand names and models is done to facilitate understanding and
neither constitutes nor implies endorsement by the Department of Energy.)
To eliminate human bias in selection of particles to be measured and analyzed,
photographs of SEM fields are overlaid with a grid. Random-number-generated
grid coordinates are used to select particles for analysis.
204
-------
information may be found in Refs 14 and 15 and in a forthcoming detailed
report of the SoRI-GFETC field test results.
Figure 3 displays frequency distributions for two constituents important
in moderating ash resistivity: NazO and 80s. Frequency distributions of this
type can help one understand the range of particle resistivities presented to
an ESP.
Figure 4 displays the same elemental concentrations as functions of
particle size. At least in the case of SO 3, there is clearly an increased
concentration in the 0-5 ym size range. Because this range includes sizes
for which ESP efficiencies often are poorest, studies of this type may be
useful in diagnosing successes and failures of various conditioning strate-
gies.
An example of a well-defined inter-element correlation is that shown by
AlaOs versus SiOa, Figure 5. This is the only example of a functional rela-
tionship which is preserved to a significant degree in data for each individual
multicyclone stage. In this plot, as in all of the intervariable correlation
plots, each datum represents a specific, individual, identifiable, serial-
numbered ash particle. By tracking it back through the statistical system,
each particle's concentration of other elements may be recovered. A major
concentration of data in Figure 5 falls in a region bounded by 50-60 pct.SiOz
and 25-35 pct.AlaOa. This lies within the range of compositions of dehydrated
illite (16).
Using x-ray diffraction applied to low temperature, oxygen plasma ashed
coal, species identified include a-quartz, calcite, and a clay, kaolinite.
Possible minor constituents which could not be positively confirmed included
dolomite and anorthite,
The thermal decomposition of kaolinite (and illite) proceeds through a
sequence of stages. Above 1095°F mullite and cristobalite are formed (16, 17,
18). Clearly, mullite is found in some coal ash. Hulett and Weinberger have
isolated spherical mullite skeletons in Eastern coal ash particles by removing
the glass phases using a 1 pct.H2Fz etch (17). Stinespring and Stewart have
discussed these transformations (18) and their relation to fly ash surface
chemistry. X-ray diffraction of Arapahoe multicyclone ash confirms the
presence of mullite and a-quartz, but not cristobalite. No other species are
observed.
The SEM/microprobe analysis of individual particles does not distinguish
phases — microprobe intensities contain contributions from surface atoms as
well as atoms within the interior; the combined signal cannot be "unfolded"
to reveal the relative contributions from several phases containing a common
element. Figure 5 can best be understood as representing a population of ash
particles containing dehydrated aluminosilicates (especially kaolinites or
illites) which have been converted thermally at least in part to mixtures of
mullite-like phases and quartz. The high SiOa, descending leg of Figure 5,
would correspond to ash particles high in SiOa to the exclusion of AlzOs-rich
phases. In addition, all particles contain phases with measurable concentra-
tions of some of the other eight "major" elements.
205
-------
SODIUM OXIDE
50-
5 BO-
. 20-
o
£ 10-
40-1
10-
SULFUR TRIOXIDE
0 I 23456789 10
Na90 Concentration Midpoint
if 0-&1
S. 0123456789 10
SOj Concentration Midpoint
Figure 3. Examples of Frequency Distributions of Elemental Concentrations
5-
4-
3-
o
CM
5 2H
I
No^O vs SIZE
S0 vs SIZE
8-
6-
T
2-
£'*'•?'.''•*•';!»'. i-'L.f-—~r
0 20 40 60 80 100
Diameter, \un
—i—• 1 1
20 40 60 80 100
Diameter, pm
Figure 4. Examples of Concentration-Particle Size Correlations
4O-I
30-
?20H
<
OH
AI203 vs Si02
••
lOH
6-
4-
2-
0
S03 vs MqO
20 40 60 80
%SiO2
IOO
O 2 4 6 8 IO 12 14
% MqO
Figure 5. Examples of Inter-element Concentration Correlations
206
-------
Ultimately it may be possible to employ detailed coal mineral analysis,
in advance of pilot plant test burns, to predict the chemical species in the
fly ash. This would be useful in planning ash disposal strategy, and possibly
in predicting collector performance. Meanwhile, it is becoming possible to
employ detailed ash characterization studies to predict aspects of ash pro-
perties beyond bulk resistivity.
For example: by using SEM's data it is possible to produce not only
binary, but also ternary and higher order inter-element correlations and also
to calculate various concentration-dependent properties of individual parti-
cles. In another paper in this symposium (19) this approach is used to
calculate frequency distributions of resistivity and resistivity-particle size
correlations.
In summary: Arapahoe ash characterization studies performed at GFETC
have resulted in:
• Particle morphology information probably relevant to fabric
filter performance.
• Detailed composition data for individual particles.
• Frequency distributions of individual elements, data maps
of composition versus particle size, and inter-element
concentration correlations.
• Identification of mineral species in coal and ash.
• Suggestions that these studies may be used to understand
and predict ash behavior in particle collection devices
such as ESPs.
ACKNOWLEDGMENTS
The authors gratefully acknowledge the contributions of the staff of
Southern Research Institute who participated in the field test at Arapahoe.
The field test personnel were supervised by Mr. G. H. Marchant, Jr.
Mr. W. R. Dickson was responsible for the gas sampling and analysis and ash
analysis. Other members of the test crew included: Messieurs W. S. Hall,
T. F. Hammond, E. C: Landham, C. V. Lindsey, J. S. O'Neal, Z. A. Peich,
D. Sanders, and T. A. White.
We also wish to acknowledge the contributions of GFETC staff who per-
formed extensive microanalyses, microscopy and data analyses. Principal
contributors were S. A. Benson, D. P. McColler, S. J. Miller, and D. K. Rindt.
This work was funded by the U.S. Department of Energy, Grand Forks Energy
Technology Center, Grand Forks, North Dakota, under Contract No. DE-AC18-
80FC10225.
207
-------
DISCLOSURE
The work described in this paper was not funded by the U.S. Environmental
Protection Agency and therefore the contents do not necessarily reflect the
views of the Agency and no .official endorsement should be inferred.
REFERENCES
1. Menard, A. R. and Richards, R. M. The Use of Sonic Horns as an Assist to
Reverse Air Cleaning of a Fabric Filter Dust Collector. This Symposium,
Session A-5.
2. Code of Federal Regulations. Title 40, Part 60, Appendix A. July 1,
1978. pp. 220-237.
3. Ensor, D, S., Cowen, S., Shendrikar, A., Markowski, G., Woffinden, G.,
Pearson, R. and Scheck, R. Kramer Station Fabric Filter Evaluation.
EPRI-CS-1669. Electric Power Research Institute, Palo Alto, California,
1981.
4. Johnson, J. W., Clinard, G. I., Felix, L. G. and McCain, J. D. A Computer-
Based Cascade Impactor Data Reduction System. EPA-600/7-78-042. U.S.
Environmental Protection Agency, Research Triangle Park, NC, 1978. 592 pp.
5. Johnson, J. W., Pyle, B. E. and Smith, W. B. Extending Precision in a
Computer-Based Cascade Impactor Data Reduction System. In: Proceedings of
the Second Symposium on Advances in Particle Sampling and Measurement.
EPA-600/9-80-004. U.S. Environmental Protection Agency, Research Triangle
Park, NC, 1980. pp. 146-166.
6. Cheney, J. L. and Homolya, J. B. Sampling Parameters for Sulfate Measure-
ment and Characterization. Environ. Sci. Technol. 13:584, 1979.
7. Felix, L. G., Merritt, R. L. and Smith, W. B. Particulate Emission and
Operating Characterization of a Fabric Filter Pilot Plant. SoRI-EAS-82-
549. Southern Research Institute, Birmingham, Alabama, 1982.
8. Nichols, G. B. Techniques for Measuring Fly Ash Resistivity. EPA-650/2-
74-079. U.S. Environmental Protection Agency, Research Triangle Park, NC,
1974.
9. Bickelhaupt, R. E. A Technique for Predicting Fly Ash Resistivity. EPA-
600/7-79-204. U.S. Environmental Protection Agency, Research Triangle
Park, NC, 1979. 105 pp.
10. Smock, R. W. Cold-Side Precipitators Performing Well on Western Coal
Power Plants. Electric Light and Power. 59(5):21, 1981.
208
-------
REFERENCES (Cont'd)
11, Eskra, B. J. and McKinney, B. G. One Year's Operating Experience with S03
Conditioning on a Large Coal-Fired Unit's Electrostatic Precipitator.
Paper presented at the 75th Annual Meeting of the Air Pollution Control
Association, New Orleans, Louisiana. June 20-25, 1982.
12. Smith, W. B. and Wilson, R. R., Jr. Development and Laboratory Evaluation
of a Five-Stage Cyclone System. EPA-600/7-78-008. U.S. Environmental
Protection Agency, Research Triangle Park, NC, 1978.
13. Hart, F. C. and DeLaney, B. T. The Impact of RCRA (PL 94-580) on Utility
Solid Wastes. FP-878. Electric Power Research Institute, Palo Alto,
California, 1978.
14. Sears, D. R. Particulate Control for Low Rank Coals. In; DOE/GFETC/
QTR-81/3-4. U.S. Department of Energy, Grand Forks, North Dakota, 1982.
15. Sears, D. R. Hydrocarbon and Tracte Element Emissions from Combustion.
In; DOE/GFETC/QTR-81/3-4. U.S. Department of Energy, Grand Forks, North
Dakota, 1982.
16. Deer, W. A., Howie, R. A. and Zussman, J. Rock Forming Minerals.
Vol. 3 Sheet Silicates. Longman, London, 1976.
17. Hulett, L. D. and Weinberger, A. J. Some Etching Studies of the Micro-
structure and Composition of Large Aluminosilicate Particles in Fly Ash
from Coal Burning Power Plants. Environ. Sci. Techno1. 14:965, 1980.
18. Stinespring, C. D. and Stewart, G. W. The Surface Chemistry of Alumino-
silicate Particles—Application to Combustion Stream Chemistry. METC/RI-
79/7. Morgantown Energy Technology Center Report. Morgantown, WV, 1979.
19. Sears, D. R. , Benson, S. A., McCollor, D. P. and Miller, S. J. Fly Ash
from Texas Lignite and Western Subbituminous Coal: A Comparative Character-
ization. This Symposium, Session C-4.
209
-------
AN EVALUATION OF FULL-SCALE FABRIC FILTERS ON UTILITY BOILERS
by: John W. Richardson, John D. McKenna, John C. Mycock
ETS, Inc.
Roanoke, Virginia 24018
ABSTRACT
The objective of this EPA sponsored program was to determine the
particulate emission concentrations of a coal-fired electric utility of
greater than 100 MW output. Tests were also conducted to determine
gaseous constituent concentrations of the flue gas and particle size
via a cascade impactor.
Testing was conducted at Southwestern Public Service's Harrington
Station, Unit 3, between July 6, 1981, and July 11, 1981. A total of
three (3) outlet tests and one (1) inlet test were performed. Due to the
absence of inlet test ports, inlet testing was done by bypassing the bag-
house and testing at the outlet ports of the stack.
Emissions for Unit 3 are controlled by two (2) baghouse systems, an
east and a west, each with its own operating control system and bypass
dampers for start-ups, emergency operation and shutdown. Each system
incorporates shake/deflate cleaning, and consists of 32 compartments with
204 bags per compartment for a total of 6,528 bags.
Average outlet concentration resulted in a lower than expected value,
of 0.007 Ibs./lO Btu.* The loading of the inlet testing was 2.0 Ibs./lO
Btu giving a 99.65% collection efficiency for the baghouse. This emission
rate is,significantly lower than the existing Federal standard of 0.03
Ibs./lO Btu. Particle sizing tests indicated that the mass geometric
mean diameter for outlet tests 1 through 3 ranged from 7.5 to 13 /urn with
an extrapolated inlet mass diameter of 60 /urn, and a baghouse collection
efficiency of 99.86%.
This paper has been reviewed in accordance with the U. S. Environ-
mental Protection Agency's peer and administrative review policies and
approved for presentation and publication.
~C*0 Readers more familiar with the metric system may use the conversion
factors at the end of this paper.
210
-------
I. INTRODUCTION
This test series of Southwestern Public Service, Harrington Station
Unit 3, was carried out as part of EFA Contract Number 68-02-3649, titled:
An Evaluation Of Full-Scale Fabric Filters On Utility Boilers.
The purpose of this project is to evaluate and characterize the per-
formance of three (3) full-scale fabric filter units controlling 100 MW
or larger coal-fired power plants. Efforts are being made to select sites
that are representative of different boiler and coal types, with control
systems characteristic of modern fabric filters.
Three outlet particulate emission tests were conducted at SPS's
Harrington Unit 3 during the week of July 6-11, 1981,to determine parti-
culate emissions in Ibs./lO BTU.
Baghouse and control room data, along with basic baghouse maintenance
and operational data, were also obtained.
Test protocol followed that specified in the Federal Register. Dia-
grams of the stack and sampling ports'may be found in Figures 1 and 2.
A total of four (4) outlet particulate tests and one inlet particu-
late test were performed. Run 1 was voided because an incorrect stack
moisture content was assumed which resulted in the wrong size sampling
nozzle being used. Runs 2, 3, and 5 (outlet tests) are considered valid
tests. Run 4 was the inlet particulate test which was conducted by by-
passing the baghouse and testing at the stack outlet ports. A summary of
these test data may be found in Table 1.
II. DESCRIPTION OF PROCESS AND CONTROL DEVICES
SPS's Harrington Station Unit 3 consists of a tangentially fired,
Combustion Engineering steam generator, capable of producing 2,700,000
pounds of steam per hour at 2500 psig, 1005 F superheat, and 1005 F
reheat.
Pulverized Western coal averaging 8,475 Btu/lb., 0.3% sulfur and
5.5% ash is burned.
Particulate emissions are controlled by Wheelabrator-Frye, Inc. bag-
houses designed to operate at a flue gas flow of 1,650,000 ACFM at 313°F,
with a minimum design efficiency of 98.6%.
There are two (2) baghouse systems at Harrington Unit 3, an east and
a west system; each with its own operating control system and bypass dam-
pers for start-up, emergency operation and shutdown. A schematic of the
boiler and baghouse configuration may be found in Figure 3.
211
-------
300'
6" Mam in 1 SanpHim Torts
Continuous Monitor ing. Port j
4 @90U
220' 8"
14' 7" Inlet: Brccchins
Inlet Breeching
Figure 1
Harrington Unit 3 Stack Configuration
212
-------
Tr.nvorsc Pt.
1
2
3
Distance in Inches
56.3
81.8
119.2
Figure 2
Traverse Points Harrington Unit 3
213
-------
DATA SUMMARY TABLE
Southwestern Public Service, Harrington, Unit 3
OUTLET EMISSION DATA
July 6-11, 1981
Run
&
Date
Run 1
7/8/81
Run 2
7/9/81
Run 3
7/9/81
Run 5
7/10/81
Particulate
Emissions
Outlet
lbs/10 BTl!
VOID
0.009
0.009
0.003
Particulate
Emissions
Outlet
Ibs/hr
35
34.6
11.7
Particulate
Emissions
Outlet
fcr/ACF
0.0024
0.0022
0.0008
MW
Production
7. Opacity
341
5.5
350
5.5
349
5.6
Flow
Outlet
ACFM
DSCFM
1,710,000
944,000
1,740,000
912,000
1,710,000
938,000
Outlet
Temp.
°F
323
323
323
Or sat
% CO 2
Inlet
Outlet
In:
a
Out:
11.6
In:
Out:
11.6
In:
Out:
11.6
Orsat
% o2
Inlet
Outlet
In:
Out:
6.5
Tn:
Out:
6.5
In:
Out:
6.5
Orsat
% CO
Inlet
Outlet
In:
Out:
0.0
In:
Out:
0.0
In:
Out:
0.0
Baghouse
Differential
Pressure
in. H00
2
East BH:
6.98
West BH:
6.26
East BH:
8.1
West BH:
6.7
East BH:
7.1
West BH:
6.6
Stack
Gas
Moisture, 7
In:
Out:
In:
Out:
10.6
In:
Out:
14.3
In:
Out:
10.2
INLET EMISSION DATA
Run
&
Date
Run 4
7/9/81
a Not t
Particulate
E-nissions
Inlet
lbs/10 ETTU
2.0
ested
Particulate
Emissions
Inlet
Ibs/hr
7960
Particulate
Emissions
Inlet
gr/ACF
0.517
Mr/
Production
7. Opacity
349
98
Flow
Inlet
ACFM
DSCFM
1,780,00
947,00
Inlet
Temp.
-------
Stack
West Boghouse
^ C / •s= East Baghouse
North
o
Ilarrinj;Lon No. 3 Flue Gas Flow
Figure 3
215
-------
This shake/deflate cleaning system consists of 32 compartments with
204 bags per compartment for a total of 6,528 bags; each bag is 11.5
inches in diameter and 30 feet, six inches long. Bag spacing is 14.0
inches center-to-center, with a bag reach of two. Design air-to-cloth
ratio is 2.81:1 gross, 2.90:1 with one compartment down and 3.0:1 with
two compartments down. Total cloth area is 6.0 X 10 ft.^ The baghouse
is outfitted with 50% 14 ounce fiberglass acid resistant coating Criswell
4491625 bags and 50% Criswell 442 570C-2 10 ounce Teflon coated fiberglass
bags. All bags are of caps and eye bolts design. Three and one-half
inches of fiberglass insulation covers the baghouse exterior.
TEST PORT LOCATIONS
Sampling was conducted at four (4) ports spaced equi-distance around
a 20.8 ft. diameter stack. Three traverse points were assigned to each of
the four (4) ports , resulting in a twelve (12) point traverse particulate
test. A diagram of the stack traverse points may be found in Figure 2.
TESTING METHODOLOGY
Particulate emission tests were conducted according to U.S. EPA Re-
ference Method 5 procedures in conjunction with Methods 1, 2, 3 and 4.
Each test included a 12-point traverse with a ten (10) minute sampling
duration for each point.
Assembly and use of the imp actor train followed stater-of-the— art
protocol and general Method 5 sampling train procedures.' Special pre-
caution was taken to avoid rough handling of loaded impactors, overloading
of the impactor and in the performance of hot leak tests.
Test 4 was an inlet particulate test conducted by bypassing the
house and directing the particulate directly to the outlet stack. Test 4
incorporated EPA Method 17 using a Norton AN-899 5.0 urn retention Alundum
thimble .
SAMPLING EQUIPMENT AND PROCEDURES
Particulate
The particulate sampling equipment used is referred to as the "EPA
Particulate Sampling Train", designed and developed by the EPA. ' The
"train" used in this testing was assembled by.ETS, Inc. in accordance with
construction details published by the EPA. A schematic of the samp-
ling train is shown in Figure 4.
The EPA sampling train consists of a stainless steel sampling nozzle,
a Method 5 filter holder containing an 87 millimeter Schleicher and
Scherell #1 HV High Purity Glass Filter, a series of four (4) Greenburg-
Smith impingers, a check valve, a leakless vacuum pump, a dry gas meter,
216
-------
Temperature Sensor
1.8 to 2.5 cm
(0.75 to 1 In.")
i - ^ I Probe
Impinger Train Optional, May Be Replaced
By An Equivalent Condenser
1.8 cm (0.75 In.) f
Pitot Tube
Probe
Temperature
— Sensor
Heated
Area
Therr.oneter
Filter Holder
Thermoizeter
N)
Stack Wall
reverse-ivpe
Pitot Tube
I
Check Valve
Vacuum Line
Pitot
Manometer
Ir.pingers By-Pass
Ice Bath
Orifice
Main Vacuum Gauge
Valve
Thermometers
Dry Gas
Meter
Air-Tight
Pump
Method 5 Sampling Train
Figure 4
-------
and a calibrated orifice. The impingers and connecting tubes are made
of Pyrex glass and are connected with glass ball and socket joints. The
probe is made of type 316 stainless steel.
Using the Type "S" pitot tube, a velocity traverse is performed along
all traverse axis during each particulate run. The velocity pressure at
each sampling point is measured by means of an inclined manometer. A
thermocouple is also attached to the pitot tube to measure gas stream
temperature. Thermal potential is measured with an Omega portable poten-
tiometer.
Prior to, and at the conclusion of, each run the complete sampling
train including probe and nozzle is leak tested by plugging the nozzle
with a rubber stopper and then applying a vacuum of 15 inches of mercury
to the system. The maximum allowable leakage rate of 0.02 cfm was not
exceeded in any of the leak tests.
During testing, isokinetic sampling conditions are achieved by main-
taining the velocity of the stack gas entering the sampling nozzle equi-
valent to the stack gas velocity at each sampling point. The required
velocity of the stack gas entering the sampling nozzle is obtained by
adjusting the pressure drop across the calibrated orifice. The pressure
differential across the calibrated orifice is measured by an inclined/
vertical manometer with an inclined range of 0-1.0 inches of water and a
vertical range from 1.0 to 10.0 inches of water.
At the completion of each test, the sampling nozzle,inside of the
probe, inside of the thimble holder and front-half of the glass fiber
filter holder were washed with acetone. The washings were collected in
separate storage containers. The Alundum thimble and the glass fiber
filter were removed from their respective fiber holders and stored in
separate storage containers. The contents of the first three impingers
were measured and discarded. The silica gel from the fourth impinger was
transferred to a separate storage container. All storage containers were
returned to the laboratory for analyses.
Tests to determine carbon dioxide, oxygen and carbon monoxide were
conducted via an Orsat analysis according to EPA Method 3. A calibrated
cylinder was run as a blank.
III. DISCUSSION OF RESULTS
During this test series, there were no deviations from normal opera-
ting conditions for Unit 3 that could be determined from the control room
or baghouse control room data or conferences with the boiler operators.
Control room data monitored during the test period compared closely with
previous data.
Strip charts monitoring ash discharge, 862* 02 and NO concentrations,
percent opacity, baghouse outlet temperature and pressure drop were col-
lected and the data were recorded. Flue gas volumes for the three (3)
218
-------
outlet and one inlet particulate tests were within an expected and con-
sistent range.
Coal and ash samples were taken for analysis. Approximately five
pound integrated coal samples were taken during each test, of which appro-
ximately one-half pound was analyzed for ash, sulfur, moisture and Btu
content (proximate fuel analysis). Coal parameter concentrations remained
within the expected and design range.
Outlet particulate emissions at Harrington Unit 3 averaged 0.007
Ibs./lO Btu. Emission rates were also expressed in Ibs./hr. and aver-
aged 27.39 Ibs./hr. for the three (3) outlet tests. The single inlet
emission test performed by bypassing the baghouse resulted in an emission
rate of 2.0 lbs/10 Btu or 7958.5 Ibs./hr. Emission rates for outlet
runs 2 and 3 were very close at 0.009 lbs./106 Btu (35.9 Ibs./hr.), and
0.009 lbs./106 Btu (34.6 Ibs./hr.), respectively, while outlet run 5
resulted in an unexpectedly low emission rate of 0.003 Ibs./lO Btu or
11.7 Ibs./hr. The inlet run performed at the outlet stack ports also
resulted in a lower emission rate than expected when compared to previous
test results conducted at SPS Harrington Unit 3.
Cascade impactor sampling was performed by Research Triangle Insti-
tute, July 8 and 9, 1981. A total of three (3) outlet and one (1) inlet
impactor tests were performed. Figure 5 summarizes the cumulative mass
size distribution results for the three (3) outlet impactor tests.
All outlet impactor tests including a blank had some stages with
weight losses in spite of extended sampling times. The weight losses on
the substrates were believed to be due to chemical reactions of the flue
gas with the grease to form volatile materials, although the Apiezon-H
grease has performed satisfactorily at other test sites. All of the out-
let size distribution data were adjusted using the blank test substrate
weights. The mass geometric mean diameter for outlet tests 1 through 3
ranged from 7.5 to 13 Mm with an extrapolated inlet mass diameter of 60 /urn,
and a baghouse collection efficiency of 99.86%.
Southwestern Public Service has expressed concern that the results
obtained during this series of tests do not represent typical operation
of the Harrington Unit 3. The following paragraph is taken from a letter
sent to EPA.
"SPS questions the validity of the data because there is substantial
disagreement with previous testing by SPS and GCA on Harrington Unit 2
and by SPS on Unit 3. Additionally, results of the inlet grain loading
test'(Run 4) by ETS are lower than the theoretical values by 50%.
Finally, it seems unreasonable that the lowest particulate loadings
measured came the day after the bypasses had been opened the night before."
ETS agreed to include this paragraph in the paper to make the SPS
reservations known, however, we feel that the emission levels measured are
representative of the baghouse performance at the time of the test. Inlet
219
-------
u
99.9&1
99.9-
99.8-
99-
98-
95-
90-
70-
60-
50-
40-
30-
20-
10-
5-
2-
1-
.5-
.2-
.1-
.05-
.01-
.1
Run No. 2 on July 8, 1981;
3 hrs.
A Run No. 3 on July 9, 1981;
8 hrs.
x Run No. 5 on July 10, 1981;
13 hrs 51 min.
TIIIII i ITT i 1111 i iiii
1 10
AERODYNAMIC DIAMETER (pm)
122
Figure 5 Cumulative mass size distribution for
three outlet impactor samples.
220
-------
testing may be suspect since the inlet sample was obtained at the stack
outlet sampling port's after bypassing the baghouse. However, as can be
seen in Table 2, the results of the outlet tests compare with emission
tests conducted at other coal-fired utilities. This conclusion is also
supported by facts such as the 15 hour sampling time required to obtain a
representative impactor sample, a visual opacity reading between 0 and 5
percent during all the outlet tests, and the presence of an RTI quality
assurance specialist during the entire testing program.
Bag analyses were performed at ETS* laboratory in Roanoke, Virginia.
A battery of tests were performed including permeability, tensile strength,
MIT flex, and Mullen Burst. A total of four (4) bags (2 used and 2 new)
were tested.
Overall, the bag analyses conducted by ETS, Inc. indicate a greater
percentage loss of strength in terms of MIT flex, Mullen Burst and tensile
strength than previous tests performed by another firm. Permeability as
received is essentially the same when comparing ETS's and the other firms'
results. The cleaning procedures differ between the two laboratories but,
in general, indicate approximately the same permeability improvement after
cleaning. A more detailed comparison of the bag analyses may be found in
Table 3.
It should be noted that ETS used a .04 in. jaw and four (4) pound
weight and the other firm used a .03 in. jaw and four (4)'pound weight
when performing the MIT flex test. Another possible contributing factor
to the differences is the other firms bag testing was done in April 1981,
and the ETS bag analyses were performed in September 1981; thus, the bags
were on stream longer before the ETS tests were conducted.
Laboratory analytical procedures incorporated were those mandated by
the Federal Register Part II, Environmental Protection Agency, Revision to
Reference Methods 1-8.
Second Test Location
Particulate emission tests were also conducted at Pennsylvania Power
and Light's Brunner Island Station Unit 1 August 10-14, 1981 and August 30-
September 4, 1982.
Unit 1 consists of a 345 MW Westinghouse turbine generator unit sup-
plied by a continuous operating Combustion Engineering type CL pulverized
coal-fired boiler; with a maximum rating of 2,100,000 Ibs. of steam/hr. at
1005°F and 2750 PSIG at the superheater outlet and reheat rating of
1,947,000 Ibs. steam/hr. at 1005°F and 600 PSIG.
Particulate control is handled by a Carborundum baghouse with a
designed gas flow of 1,200,000 ACFM at 330°F.
The average particulate emission rate for five (5) outlet tests
221
-------
TABLE 2
MASS PARTICULATE EMISSIONS OF VARIOUS COAL-FIRED POWER PLANTS
SITE EMISSIONS.gr/ACF
1. SPS - Harrington Unit 3 0.002
2. Colorado Ute, Nucla Station 0.003
3. Northern States Power
Riverside Station 0.003 - 0.006
4. Nebraska Public Power
Kramer Station 0.02
5. Pennsylvania Power & Light
Sunbury Station 0.005 - 0.006
6. Pennsylvania Power & Light
Brunner Island 0.0085
7. Northern States Power
Elk River 0.004
8. SPS - Harrington Unit 2 0.017
9. Colorado Ute
Bullock Station 0.012
222
-------
TABLE 3 - FABRIC TESTING RESULTS COMPARISON
Bag Typo
Acid
Resistant (625)
Used
Tet '.on - 3
L'sed
Acid
Resistant (625)
New
Teflon - 3
New
Fabric Tests
Flex (Fill)a
Top - 48
Middle - 55
Bottom - 68
Top - 58
Middle - 57
Bottom - 71
Top - 59
Middle - 50
Bottom - 49
Top - 84
Middle - 88
Bottom - 89
Fill - 2,500
Warp - 12,500
Fill - 3,013
Warp - 10,400
Fill - 500
Warp - 15,000
Fill - 629
Warp - 9,946
Mullen Burst a Tensile a
13 - 297, loss
32% loss
2 - 20% loss
247. loss
700 PSI
619 PSI
525 PSI
343 PSI
Fill - 0-207,
Warp - 1-177.
Fill - 437, ;
Warp - 317.
Fill - 0-167.'
Warp - 0-7%
Fill - 18%
Warp - 277.
Fill - 300
Warp - 600
Fill - 252
Warp - 465
Fill - 130
Warp - 250
Fill - 87
Warp - 228
Permeabi 1 ii:y a
As received - 1.48-2.75
After wash - 48
As received - 1.2-2.1
After vacuum - 60.4
As received - 2.31-2.75
After wash - 55
As received - 1 .8-2.0
^After wash - 46.4
Testing Co.
Other Lab
ETS, Inc.
Other Lab
ETS, Inc.
Other Lab
ETS, Inc.
Other Lab
ETS, Inc.
ro
ro
OJ
MIT Flex
Mullen Burst
Tensile
Permeability
Measurement Parameters
cycles
Ibs/sq. inch
Ibs/inch
CFM/sq. ft.
Other Lab - .03 in. jaws @ 4 Ibs.
ETS, Ince - ,QA in. jaws @ 4 Ibs.
-------
u
performed in 1981 was 0.0364 Ibs./lO Btu. Inlet emissions averaged
13.08 Ibs./lO Btu.
A total of six (6) cascade impactor tests were performed, two (2)
inlet, two (2) outlet, and two (2) blank runs. Mean extrapolated particle
size was 54.3 microns at the inlet and 8.72 microns at the outlet. These
impactor tests indicated that the baghouse collected more than 99.9% of
the particulate matter in the flue gas.
This fabric filter characterization program continues and efforts con-
tinue to secure a third site for testing.
METRIC EQUIVALENTS
Readers more familiar with metric units may use the following equiva-
lents to convert to that system:
Nonmetric
Btu
in. -
in.
°F
ft
Ib
micron
oz
Times
1055
2.54
6.45
5/9(°F-32)
0.30
0.09
28.32
0.06
0.45
1.00
28.35
Yields Metric
J
cm-
cm
m
liter
g
kg
Jim
g
224
-------
REFERENCES
1. Harris, D. Bruce. "Procedures For Cascade Impactor Calibration and
Operation in Process Streams". EPA 600/2-77-004, (NTIS PB263623),
January 1977.
2. "Code of Federal Regulations". Method 17, Title 40, Part 60,
Appendix A.
3. Martin, Robert M. "Construction Details of Isokinetic Source
Sampling Equipment". EPA report APTD-0581 (NTIS PB203060), April
1971.
225
-------
Status of SPS Investigation of
Harrington Station Unit 2 Fabric Filter System
Richard Chambers
Southwestern Public Service Company
P. 0. Box 1261
Amarillo, Texas 79170
Dale Harmon
Industrial Environmental Research Laboratory
Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
This paper describes activities during the fourth year of the SPS/EPA
fabric filtration study at Harrington Station. The scope of the overall
study and its purposes were outlined in earlier papers.
During the fourth year of this project, sufficient information was
available to make certain comparisons between Harrington Station's high
air-to-cloth ratio (3.4 ft/min) shake/deflate fabric filter system and low
air-to-cloth reverse air units. Specifically, comparison of bag life
economics and pressure drop performance is addressed. Sizing of reverse
air and shake/deflate baghouses is discussed in light of current operating
data from Harrington Station and other installations.
Operation and maintenance costs are shown for the Harrington Unit 2
fabric filter system, and suggestions for improved shaker design are given.
Plans to investigate sonic augmentation of bag cleaning are described.
This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved
for presentation and publication.
INTRODUCTION
Electric utility experience with fabric filtration for particulate
control of coal-fired boiler emissions has increased dramatically over the
last 5 years. Southwestern Public Service (SPS) established itself as a
pioneer in utility fabric filtration with the construction of two shake/
deflate baghouses in 1978 and 1980.
Since 1977, SPS, under contract to the Environmental Protection
Agency, has been collecting data at Harrington Station, Unit 2; therefore,
a great deal of information is available on the Unit 2 system. Results of
this study have become increasingly important as other utilities, seeking
solutions to emission control problems, look for design and performance
specifications based on actual operating experience.
226
-------
This paper presents information SPS has accumulated that will be of
value to utilities evaluating the applicability of shake/deflate cleaning
for a specific installation. Comparisons are made between shake/deflate
and reverse air fabric filters in the areas of pressure drop, sizing, bag
life and fabric replacement economics, and operation and maintenance costs.
The EPA study at Harrington Station is now into its fifth and final
year. The results of this program are providing much needed knowledge of
the effectiveness of a shake/deflate fabric filter system used in associa-
tion with low sulfur western coal. A final report will be published at the
conclusion of the study.
BAG LIFE
There is very limited data available to the utility industry regarding
the bag life of various fabrics in reverse air and shake/deflate collectors
because few collectors have been on line long enough to establish fabric
life, especially on large units. According to a 1981 EPRI report (1),
"Preliminary results from reporting utilities indicate bag lives in excess
of two years, with several units exceeding three years." As more exper-
ience is gained, many of these collectors will undoubtedly achieve bag life
in the 3 to 6 year range, but as yet 'no reliable fabric life predictions
are available for reverse air units.
Harrington Unit 2, however, has been on line since 1978 and bag life
data is available from this shake/deflate unit. Full compartment tests
were done on a variety of fabrics to determine both their pressure drop
performance and wear characteristics (2).
2
Of the fabrics tested at Harrington, the plied, 10-oz/yd Teflon-
coated materials have shown the best bag life and performance. These tests
demonstrate that bag life in the 3.0 to 3.5 year range is obtainable in a
shake/deflate collector employing suitable fabrics.
Recent testing on the Unit 2 filter has shown the possibility of
extending bag life beyond the 3 to 3.5 year range. Fabric wear occurs
mostly during the cleaning cycle where the fabric is subject to flexing and
abrasion. Therefore, reducing the shaking time and increasing the filtra-
tion time between cleaning cycles should yield gains in bag life. Prelimi-
nary testing results have shown that reducing the shake times at Harrington
from 20 to 5 seconds does not adversely affect pressure drop, implying
satisfactory cleaning with just a 5-second shake. In addition, preliminary
results of attempts to increase filtering time before cleaning look promi-
sing. With these changes, it is anticipated that bag life on Harrington 2
can be extended to 4 to 4.5 years.
The Harrington Unit 3 fabric filter operates at a design air-to-cloth
ratio of 3.0 ft/min, compared to 3.4 ft/min for Unit 2. Although it is too
early to determine what bag life the unit will eventually experience, after
only 2 years of operation (start-up summer 1980), the fabric appears to be
in better shape than the Unit 2 fabric after 2 years service. Most likely
this is because the Unit 3 filter is cleaned less frequently due to its
227
-------
lower pressure drop. On this basis it is expected that the lower air-to-
cloth ratio fabric filter on Unit 3 will exceed 4 years bag life on its
first set of bags.
FABRIC REPLACEMENT COSTS FOR R/A AND S/D FABRIC FILTERS
In comparing fabric replacement costs for reverse air (R/A) and
shake/deflate (S/D) collectors the following factors must be kept in mind:
1. The rather substantial price difference between ringed R/A and
unringed shaker bags.
2. The larger number of R/A bags compared to a higher air-to-cloth
shaker.
3. The higher wear rate of shaker bags vs. reverse air bags due to
cleaning forces.
To quantify the interrelationship between these factors the levelized
fabric replacement costs vs. bag life are considered for a 1100 MW unit
equipped as follows:
2.0 ft/min with 2 of 32 compartments
2.7 ft/min with 2 of 32 compartments
1. R/A baghouse, a/c
out-of-service.
2. S/D baghouse, a/c
out-of-service.
3. S/D baghouse, a/c = 3.0 ft/min with 2 of 32 compartments
out-of-service.
4. S/D baghouse, a/c = 3.4 ft/min with 2 of 32 compartments
out-of-service.
For this analysis:
Bag installation cost
Bag dimensions
R/A bag cost
S/D bag cost
Inflation
Cost of capital
$8.30/bag
33.5 x 1 ft
$61.60/bag (7 rings)
$48.50/bag (no rings)
5%.
11%
The breakdown of replacement costs is shown in Table I.
TABLE I. FABRIC REPLACEMENT COSTS *
A/C
(ft/min)
2.0
2.7
3.0
3.4
Cleaning
R/A
S/D
S/D
S/D
Fabric
Cost ($)
1,345,000
784,700
706,200
623,100
Instal-
lation
Cost ($)
181,300
134,300
120,900
106,600
Total
Replacement
Cost ($)
1,526,300
919,000
827,100
729,700
* at 1100 MW
The results of this analysis are shown in Figure 1.
228
-------
The most striking, result is the very rapid initial decline of all the
curves. In the 1 to 5 year range of bag life, dramatic changes in replace-
ment cost occur from year to year. These curves strongly point out the
magnitude of the savings to be gained by optimizing fabric life.
The highest curve in Figure 1 is that for the reverse air unit. This
happens because the reverse air filter has the larger number of bags (hav-
ing the lowest air-to-cloth ratio) and the highest priced fabric. Although
it is difficult to compare directly, reverse air cleaning is generally
expected to be less damaging to the fabric and gives better bag life. The
purpose of these curves is to indicate how much longer bag life needs to be
to obtain the same fabric replacement cost as the shake/deflate units. The
smallest of the shake/deflate units is the one with an air-to-cloth ratio
of 3.4 ft/min. The pressure drop and resulting fan cost at this
air-to-cloth ratio is estimated to be equivalent to the reverse air
collector at a 2.0 ft/min air-to-cloth ratio (Figures 2 and 3). Since this
collector has the same air-to-cloth ratio as Harrington, it could be
assumed that bag life would be at least in the 3 to 3.5 year range
(comparable to Harrington). As shown„ in the lower portion of Figure 1, a
3.0-year bag life on this unit is roughly equivalent to 5.5 years life on
the reverse air system.
Bag life should increase on the shake/deflate units as the air-to-
cloth ratio decreases since less cleaning will be required to maintain a
given pressure drop. Comparing a 4-year bag life on the 3.0 ft/min air-to-
cloth ratio shake/deflate collector with a reverse air unit gives a
6.3-year bag life to achieve equivalent fabric replacement costs.
The analysis discussed above makes a number of assumptions that may
not actually be realized in practice. The intent, however, is not a
definitive analysis, but to illustrate the interplay of the factors
involved for a fairly reasonable set of assumptions. The one point that
can be drawn from the analysis is that the shake/deflate design can be
competitive with reverse air designs in terms of fabric replacement costs
under certain conditions.
PRESSURE DROP AT HARRINGTON STATION
Figure 4 shows the pressure drop history of the Harrington Unt 2
baghouse (beginning after the first rebagging of the unit.) The pressure
drops shown are flange-to-flange pressure drops, calculated at full load
(a/c=3.4 ft/min) from operational data taken for each month. (Note: Pres-
sure drops in Figure 2 are lower since they are tube-sheet pressure drops.)
As Figure 4 shows, much variation in pressure drop occurs over an extended
period of time due to variations in operating and filtration conditions.
The full-load flange-to-flange pressure drop has averaged 7.87 in.
w.g. over the past 38 months of operation. At a reduced load (318 MW) and
an air-to-cloth ratio of 3.0 ft/min, the pressure drop averaged 6.85 in.
w.g. for this same period.
229
-------
Point
X
O
A
D
ft./mln
2.0
2.7
3.0
3.4
Type
R/A
S/D
S/D
S/D
(5.6)
6
Figure 1,
Boglife.yr
Levelized fabric replacement cost (1100 mw unit)
230
-------
CT
*
C
•
0.
<3
UJ
UJ
X
co
UJ
03
UJ
e>
cr
UJ
25
24
23
22
2!
20
19
I 8
17
16
15
14
13
12
I I
10
9
8
7
6
5
4
3
2
1 I—
0
ARAPAHOE PILOT UNIT
AREAL LOADING PER
FILTERING CYCLE (FFPP)
0.25 LB/FT2
O.I5 LB/FT2 > REVERSE AIR
0.25 LB/FT
O SHAKE
UTILITY FABRIC FILTERS
O ARAPAHOE No. 3
A CAMEO No. 2
A CHEROKEE No. 3
D ECOLAIRE
• HARRINGTON (SHAKE)
D KRAMER
V MARTIN DRAKE
NIXON
HARRINGTON PILOT R/A
— ADAPTED(with permission) FROM EPRI.
HARRINGTON PILOT S/D
1.0
I.7 2.0
2.5
3.O 3.3
FILTERING AIR-TO-CLOTH , acfm/ft.
Figure 2. Comparison of EPRI/Harrington data.
231
-------
6 -r
5 '
4 •
X
o
o
o_
O
O
* 3
e
o
Q.
Z
2- •
s/o
1 • •
I
1.5
2.0 2.5 3.O
Design A/C Ratio
3.5
4.0
Figure 3. Fan horsepower costs for 1100 raw unit.
232
-------
to
OJ
GJ
West Boghouse
Bughouse
JASONDJFMAMJJASONDJFMAMJJASONDJFMAMJJA
1979 1980 1981 1982
Figure 4. Monthly average full-load pressure drop.
-------
Curve fitting the pressure drops versus air-to-cloth ratio over a long
period of time has led to the following correlation for pressure drop:
AP = 1.51V1*35 (1)
where V is the air-to-cloth ratio in ft/min and A P is the pressure drop in
in. w.g. The equation represents both of Harrington Station's fabric
filters (Units 2 and 3) very well in predicting the average pressure drop
at some given air-to-cloth ratio over a long period of time.
The pilot baghouse (3, 4) at Harrington Station has been able to
essentially duplicate the performance of the prototype units at Harrington
when operated in the shake/deflate mode. The pressure drop correlation
developed for the pilot is almost identical to that for the prototype.
AP = 1.60V1'35 (2)
PRESSURE DROPS FROM OTHER UNITS
Care should be exercised when selecting collectors for comparisons of
pressure drop due to the low capacity factor of some units. A baghouse on
a small boiler that is used for peaking purposes only cannot be expected to
give the pressure drops that a larger base loaded unit would give. Figure
5 shows the increase in pressure drop at Harrington Station at constant MW
load with respect to time over a 38-hour period. This plot illustrates the
effect that extended periods of high load can have on pressure drop.
Pressure drop data for the following comparison were taken from four
sources:
1. Harrington Station.
2. EPRI data on utility baghouses.
3. EPRI's fabric filter pilot plant.
4. Harrington Station's pilot unit.
Figure 2 shows the tube sheet pressure drops and air-to^cloth ratios
of these units. All of the shake/deflate data seem to correlate fairly
well as shown in the lower right-hand corner of the graph. The pilot unit
data from Harrington and EPRI agree well with each other and with operating
experience from many of the reverse air prototype units below an air-to-
cloth ratio of 2.0 ft/min. Above 2.0 ft/min, the Harrington pilot gives
higher resulting pressure drops than the EPRI pilot, the Ecolare unit, and
the Kramer.
ANNUAL I.D. FAN HORSEPOWER COSTS
To compare the I.D. fan horsepower costs associated with the pressure
drops shown in Figure 2, costs for a 1100 MW unit are again considered
based on the Harrington pilot work shown in Figure 2. Casing losses for
234
-------
7.5-
-------
reverse air were approximated at 2.0 in. w.g. to obtain the total flange-
to-flange pressure loss. In practice the casing losses vary between about
1.5 and 3.0 in. w.g. Since data was available for the kilowatt-hours
associated with pressure drop across Harrington Unit 2, these numbers were
used along with Harrington's busbar electrical cost of 19.17 mills/kWh.
Strictly speaking, this means that Harrington's fan and fluid drive effi-
ciencies are being assumed; in addition, a capacity factor of 0.70 was
assumed. While these assumptions and the Harrington pilot projections are
certainly only approximations of the real cases, they should serve the
purpose here of a "ballpark" comparison of I.D. fan horsepower costs for
shake/deflate and reverse air units at various air-to-cloth ratios. The
resulting yearly costs of fan horsepower versus air-to-cloth ratio are
shown in Figure 3.
The 2.7 ft/min air-to-cloth shake/deflate collector shows equivalent
fan horsepower costs to reverse air at an air-to-cloth ratio of 1.6. As
expected, the 3.0 ft/min air-to-cloth ratio shake/deflate unit has higher
pressure drop and is equivalent to reverse air in fan costs at an air-to-
cloth ratio of approximately 1.75 ft/min. The 3.4 ft/min air-to-cloth
ratio shake/deflate case has even higher pressure losses and is equivalent
to reverse air in fan costs at an air-to-cloth ratio of approximately 2.0
ft/min.
OPERATION AND MAINTENANCE COSTS
Operation and maintenance costs for Harrington Station's Unit 2 fabric.
filter system for a 3-year period are shown in Table II (in 1982 dollars)
in terms of dollars per installed kilowatt per year.
TABLE II. OPERATION AND MAINTENANCE COSTS
FOR HARRINGTON STATION UNIT 2
Year
81-82
80-81
79-80
Materials
0.055
0.248
0.570
Maint. Labor
0.167
0.179
0.170
Oper. Labor
0.192
0.215
Avg. 0.291 0.172 0.204
The material costs shown include bag purchases in 1979. The average
O&M cost over this 3-year period is $0.667/kW/yr.
No data was available for R/A collectors to compare with these
numbers. R/A O&M costs should vary from those of a S/D collector mainly
for the following reasons:
236
-------
1. Difference in bag replacement costs.
2. R/A fan horsepower pressure, drop and maintenance, compared to the
much smaller deflation fan on S/D collectors.
3. No shaker maiiltenance on R/A baghouses.
IMPROVEMENTS IN SHAKER DESIGN
Operating experience over the last 5 years at Harrington Station has
pointed to several areas of improvement in shake/deflate design. These
improvements, described below, are aimed at minimizing system cost and
annual operation and maintenance costs.
1. Amplitude of shake: The distance the top of the bag is moved
during a shaking cycle can profoundly affect bag life since the fabric may
actually fold over and abrade against itself during cleaning. Maintaining
shake amplitude at around 1 in. at reasonable shake frequencies could
lengthen baglife considerably.
2. Deflation control should be established on flow rather than
pressure drop. The deflation air-to-cloth ratio employed at Harrington is
approximately 0.2 ft/min and produces .-a pressure drop in the 0.1-0.3 in.
w.g. range. Control with this small pressure drop is difficult at best.
3. Multiple shakers should be used on larger compartments in an
effort to keep the inertia of the shaking system to a minimum and reduce
wear on bearings and supports.
4. Two-speed or variable-speed shaker motors might be of use in
minimizing fabric wear. Under conditions where less cleaning is desirable
the shaker motors could be put on low speed to lessen fabric damage.
5. Shaker tube bearings that are mounted in the compartment wall
must be carefully designed for adequate wall stiffness to avoid any
movement that could damage the bearings. In addition the shaker tube
bearings need to be protected from overheating.
SONIC AUGMENTED CLEANING
A number of utilities are employing sonic horns to augment fabric
cleaning. Therefore, EPA and SPS decided that a study of sonic augmenta-
tion was needed to determine its effectiveness compared to shaking and to
look at the effects it has on fabric life. After consulting with the
Fuller Co., six sonic horns were purchased and plans were made to suspend
them evenly spaced beneath the two walkways in one compartment in the Unit
2 baghouse at Harrington Station.
During testing the horns will be activated during the deflation cycle
for approximately 10 seconds. Both flow and pressure drop will be measured
on the compartment and then compared with adjacent compartments. Testing
will be done with and without shakers to determine the relative magnitude
of the cleaning forces being generated by the sonic horns.
237
-------
Fabric samples will be taken from the compartment at certain intervals
and tested to determine if there is any difference in fabric wear between
the two cleaning techniques.
CONCLUSIONS
Southwestern Public Service has gained a great deal of knowledge from
the EPA/SPS fabric filter study that has been of immediate benefit in the
proper operation of Southwestern's fabric filter systems. The research
performed under this program has allowed SPS to identify the performance
and wear characteristics of the commonly used fabrics and led to identify-
ing a successful fabric for the unit in terms of both bag life and pressure
drop. Fabric studies have revealed the mechanism of fabric wear under
shake cleaning conditions that has been invaluable in understanding bag
wear.
As useful as this program has been, fabric filtration in the utility
industry still has a need for more fundamental research. Ash filtration
properties are as yet very poorly understood and are beyond prediction,
making accurate sizing of fabric filters difficult. Little research has
been done in the area of designing fabrics specifically for utility
filtration which would be a very fruitful area of research. Additional
work is needed in the fluid mechanical design of baghouses to minimize
casing losses and ensure good ash distribution.
As electric utilities convert to coal, the challenge of balancing
environmental compliance with rising electric rates increases the
significance of research in the field of fabric filtration. Studies such
as the one conducted at Harrington Station can best yield results that
reflect utility standards and regulatory interests.
REFERENCES
1. Operating History and Current Status of. Fabric Filters in the Utility
Industry, EPRI Report 1401. July 1981.
2. Chambers, R., Ladd, K., and Kunka, S. SPS Experience with Fabric
Filtration. Paper presented at Fifth International Fabric Filter
Forum, Phoenix, Arizona, January 1981.
3. Ladd, K., Hooks, W., Kunka, S.,and Harmon, D. SPS Pilot Baghouse
Operation. In: Third Symposium on the Transfer and Utilization of
Particulate Control Technology. Volume I. Control of Emissions from
Boilers, EPA-600/9-82-005a, July 1982. (NTIS PB83-149583), pp 55-64,
July 1982.
4. Chambers, R., Kunka, S., and Harmon, D. Update of SPS Pilot Baghouse
Operation. Paper presented at Fourth Symposium on the Transfer and
Utilization of Particulate Control Technology, Houston, Texas, October
1982.
238
-------
UPDATE OF SPS PILOT BAGHOUSE OPERATION
Richard Chambers and Sherry Kunka
Southwestern Public Service Company
P. 0. Box 1261
Amarillo, Texas 79170
Dale Harmon
Industrial Environmental Research Laboratory
Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
An option of the 1977 Environmental Protection Agency (EPA) contract
with Southwestern Public Service Company (SPS) to assess the performance of
a large prototype fabric filter system provided for the installation of a
pilot baghouse. The pilot unit was placed in service in October 1979.
This paper describes the test program being conducted in the pilot unit by
SPS for the EPA. These test programs to date have primarily involved the
evaluation of the filtration characteristics of a number of fabrics in both
the shake/deflate and reverse air cleaning modes.
This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved
for presentation and publication.
INTRODUCTION AND BACKGROUND
In 1977, the Environmental Protection Agency (EPA) executed a contract
with Southwestern Public Service Company (SPS) to assess the performance of
a large fabric filter system used on a new utility boiler burning low
sulfur Western coal. One option of this contract provided for the instal-
lation of a pilot filter system on-stream with Harrington Station Unit 2.
The EPA elected to exercise this option in 1979, and SPS agreed to operate
and maintain a pilot unit at Harrington Station.
The pilot unit is a Wheelabrator-Frye, Inc. Model 366, Series 11.5 RS
Dustube Dust Collector. It has two six-bag compartments and initially was
fitted with a Criswell Teflon-coated fabric; bags were 11.5 in. in diameter
and 366 in. long. Cloth area per compartment is 549 ft . The unit is
operated by a 480-V control panel and an instrumentation panel.
The pilot unit is installed at the southeast corner of the east bag-
house on Harrington's Unit 2. Inlet flue gas to the pilot is pulled from
the east inlet duct of the main baghouse and mixed with hot preheater flue
gas through a mixing valve to maintain constant temperature. (See Figure
1.)
239
-------
NJ
<->
AIR
GAS DUCT
TO AIR
PREHEATEF
(700°F)
IT
<
1
£
3-
b.
PREHEATEf
HOT INLET
x-BLOCK DAMPER
[s<1 n ^ r
HOT INLET
h PURGE. J
<*
• I
\
*•
COLD INLET
BLOCK DAMPER^
TEMPERATURE
/CONTROL DAMPER
r~~j
A_J -
iFANl
7 V OUTLET
^ ABLOCK VALVP
DUCT FROM AIR
PREHEATER TO
BAGHOUSE
(350°F)
AIR
FLOW "~
!
-J
' V<}—1
COMPARTMENT COMPARTMENT
ON LINE\ (CLEANING
"— ^ ^
/BY PASS THROTTL-
ING DAMPER
" " ' "1
BY PASS DAMPER r —
p
FLOW
1
S
r\
CONTROL
-------
In October of 1979, the pilot baghouse was placed in service. Several
initial problems bad to be resolved before a satisfactory operating level
could be attained (1). By July of 1980, tbe pilot unit was ready for
testing of particulate and gaseous emissions. Tbis first series of tests
confirmed that flue gas composition and particulate loading for the pilot
unit were comparable to those of the main baghouse.
The next series of tests, which are discussed in this paper, were
designed to reflect the interests of both SPS and EPA. These areas of
testing include scale-up studies with shake/deflate cleaning, reverse air
fabric studies, and special interest studies, such as ringed bag testing
and design equations. The pilot baghouse is now being prepared for an
electrostatic enhancement program to be conducted at Harrington Station.
SCALE-UP WITH SHAKE/DEFLATE CLEANING
OVERVIEW
The first objective of the pilot baghouse study following start-up and
check-out was to prove that suitable scale-up to the prototype unit exist-
ed. Data were collected over a 5-month period with the pilot unit opera-
ting in various modes considered to be representative of the prototype bag-
house.
The fabric chosen for this initial program was W. W. Criswell's 442-
57DC-2 (10-oz. Teflon-coated). Ideally, for comparison, the same fabric
used in the prototype unit would be used in the pilot unit; however, this
mode of operation was not possible due to the large number of fabrics being
tested in the prototype (see Table I). The Criswell fabric was selected
for the pilot unit because it represented the majority of fabrics in the
prototype unit and had a proven performance record.
In the first phase of the pilot scale-up work, the unit was operated
around the clock for 1 month at the design air-to-cloth ratio of 3.4
ft/min. Phase two called for alternating operation between high and low
loads. Therefore, in June the pilot unit was run at air-to-cloth ratios of
2.8 and 3.4 ft/min, and in July at air-to-cloth ratios of 2.2 and 3.4
ft/min.
Because none of the above test methods attempted to cycle load with
the boiler, an effort was made in the third phase of the scale-up study to
simulate the daily load cycle. During August and September the pilot unit
was run at an air-to-cloth ratio of 3.4 ft/min from 8:00 a.m. to 3:30 p.m.
The air-to-cloth ratio was then reduced to 1.94 ft/min until the next
morning.
OPERATING PARAMETERS
Cleaning cycle parameters and operating conditions for the scale-up
study are:
241
-------
TABLE I. PROTOTYPE FABRIC SUMMARY
West Boghouse
East Baghouse
1
*
3
*
5
*
7 1 Nomex All-
Spun;2 Nomex
Comb.jSCriso-
f lex 446;
4Crisoflex
449;BaL «
9
*
(Warp- In)
II
*
13
*
2
*
4
*
Menardi -
Southern
Teflon
Test Bags
8
*
10
*
12
*
14
*
IS
* *
17
* •»
19 Original bags
equipped with
special
shaker
mechanism
21
Criswell 442
Teflon B
Test Bags
23
Fabric Filters
502 Tri-Treat
Test Bags
25
Criswell 449
Teflon B
Test Bags
27
* *
•*• «•
* *
Fabric Rlters
All-Filament
Teflon
34 Acid Flex;
34 Tri-Treat
Balance :
Original Bags
Globe -
Albany
Nomex
* *
•*• *
16
18
20
22
24
26
28
* Criswell 442 Teflon B; (10.5 oz.) (Rebagging complete).
** Criswell 449 Tri-Treat;(14oz.) (Rebogging complete).
242
-------
Inlet Temperature 400°F
Grain Loading 0.75-1.2 gr/acfm
Filtering Cycle 30 min.
First Null Period 20 sec.
Deflate 40 sec.
Shake 20 sec.
Settle "no-flow" 20 sec.
Cleaning Cycle 30 min.
DATA REDUCTION
Pressure drop data was collected throughout each filtration cycle by
strip chart recorders. A typical filtration cycle is shown in Figure 2.
According to filtration theory the pressure drop, AP, across a single
compartment fabric filter is given by
AP = Ap „+ K_cV2t (1)
e 2
where
Ap = effective Ap after cleaning
e
K~ = the specific resistance
c = grain loading
V = air-to-cloth ratio
t = filtration time.
The effective pressure drop after cleaning, A p , was determined by
extrapolating the linear portion of the filtration curve shown in Figure 2
to zero time.
The specific resistance can be calculated by rearranging the equation
above and setting Ap equal to the terminal A p. Grain loading, however, was
not measured on a regular basis during the testing program. Therefore only
values of K c can be calculated from the data without some assumption about
the absolute value and constancy of the grain loading. K c was calculated
from
K0c = (Apt - Ap )/tV2 (2)
2 e
243
-------
o
k.
Q
Terminal Pressure Drop,APt
Slopes K2cV
Effective Pressure Drop ,APe
Filtration Time
Figure 2. Typical filtration cycle.
40
35
3O
25
20
IS
10
I
Range of R/A
Values
2.O Z.2 2.5 2.8 3.4
Air-To-Cloth Ratio, f t/min
Figure 3.
results.
244
-------
For comparison with the prototype, the terminal and effective pressure
drops were averaged to give an average cell-plate pressure drop. The pilot
cell-plate pressure drop data cannot be simply compared to the prototype
flange-to-flange pressure drops: one includes "casing" losses, and the
other does not. Casing losses, of course, account for entrance effects in-
to hoppers, pressure drops across dampers, duct losses, and other losses
not associated with the fabric.
Casing losses across the Harrington fabric filters are given approxi-
mately by:
A PC = 0.156V
(3)
To put the pilot and prototype data on the same basis and allow direct
comparison of the numbers, 0.156V was added to the average cell-plate
pressure drop obtained from the pilot unit.
RESULTS
Table II shows the average flange-to-flange pressure drops from the
pilot and prototype units for the 5 months of the study. The numbers agree
remarkably well overall. The range of the flange-to-flange pressure drop
values measured at each operating condition is shown in the last column of
Table II. The pressure drops for the pilot and prototype units fall within
a fairly narrow range for each operating condition except for June and
August. In June the pressure drop on the east side of the prototype was
unusually high, and in August the east side pressure drop was unusually
high. No reason for these differences have been identified but it is
suspected that operational problems are responsible rather than filtration
phenomena.
TABLE II. SUMMARY OF SCALE-UP STUDY
WITH SHAKE/DEFLATE CLEANING
MONTH
May
June
July
August
September
Average
A/C,
f t/min
3.40
2.80
3.40
2.20
3.40
1.94
3.40
1.94
2.50
3.40
a Flange-to-flange
Prototype
East
7.21
6.23
8.08
4.61
8.00
3.88
7.86
4.11
5.70
8.48
pressure drop
F/Fa
West
7.50
5.73
7.27
4.84
8.27
3.74
7.97
3.99
5.76
8.85
Pilot
East"
7.66
6.23
7.59
4.29
8.71
3.58
9.18
4.34
5.87
8.80
F/Fa
West"
7.58
6.02
7.12
4.57
8.23
3.96
8.26
3.73
5.22
8.81
Range
0.45
0.50
0.96
0.55
0.71
0.38
1.32
0.61
0.65
0.33
0.65
In. w.g.
245
-------
One surprising result of the study was that cycling the pilot did not
give any significant improvement in agreement of data. Apparently it is
not necessary in pilot studies of this type to follow the load cycle since
good results can be obtained by simply operating at the conditions of
interest.
REVERSE AIR CLEANING
OVERVIEW
The next phase of the pilot program concerned reverse air cleaning
studies on a number of different fabric types. A summary of fabrics
studied in the pilot unit is shown in Table III.
TABLE III. PILOT FABRIC SUMMARY
Manufacturer*
Fabric
Style
Finish
Weighjj
oz/yd
Weave
Ken 617F
FF 502-1
GA 877
FF 502-1
MS 601
FF 504-1
FF 504-1
MS MS601T (7 Rings)
MS MS601T (5 Rings)
MS 601
MS 601
MS 601
MS 509-NC (7 Rings)
CS-428
AF
Q78
TC
TC
AF
TC
TC
TC
TC
TC
TC
TC
10
14
14
14
10
10
10
10
10
10
10
10
9
3x1 T
3x1 T
3x1 T
3x1 T
3x1 T
3x1 T
3x1 T
3x1 T
3x1 T
3x2 T
Crowfoot
2x2 T
3x1 T
MS - Menardi-Southern
FF - Fabric Filters
Twill
Napped
Ken - Kennecott
GA - Globe Albany
TC - Teflon Coating
AF - Acid Flex (1-625)
24f>
-------
The reverse air study looked at the pressure drop performance of
various fabrics having different weights and coatings as well as numbers of
rings. In addition, several fabrics were examined with weave types that
are not commonly encountered in utility filtration systems. The vast
majority of fabric used today in baghouses has a 3x1 twill weave. To get
an idea of how other weaves would react to utility filtration conditions, a
3x2 twill, a 2x2 twill, and a crowfoot weave fabric were examined. In
addition, a napped fabric was evaluated to examine the effectiveness of
napping as compared to texturizing in producing a filtration surface.
OPERATING CONDITIONS
The operating and cleaning cycle parameters employed during the
reverse air testing are:
Operating
Inlet temperature
Grain loading
Filtering cycle
Cleaning
First settle
R/A settle
Second settle
R/A air-to-cloth
400°F
'0.75-1.2 gr/acfm
60 rain.
30 sec.
45 sec.
30 sec.
1.3-1.5
DATA REDUCTION
The reverse air data were reduced in the same way as the shake/deflate
study, with one exception. To estimate the flange-to-flange pressure drop
of a multicompartment prototype, 2.0 in. w.g. was added to the average
cell-plate pressure drop instead of 0.156V * Typically, the casing losses
in a large reverse air collector are in the 1.5 to 3.0 in. w.g. range and
2.0 in. w.g. assumes fairly good design.
RESULTS OF REVERSE AIR STUDY ON STANDARD FABRICS
Results of reverse air on standard fabrics are shown in Table IV. All
of the fabrics tested resulted in fairly high pressure drops. The measured
cell-plate pressure drops ranged from 5.26 to 6.52 in. w.g. The flange-to-
flange pressure drops for a multicompartment prototype were estimated from
this data. These pressure drops ranged from 7.26 to 8.52 in. w.g. at an
air- to-cloth ratio of 2.0 ft/min.
247
-------
TABLE IV. RESULTS OF REVERSE AIR STUDY ON STANDARD FABRICS
Fabric
Style
Ken-617F
FF502-1AF
GA 877-Q78
FF502-1TC
MS 601-T
FF504-1AF
FF504-1TC
No. of
Rings
6
6
6
6
6
6
6
Air-to-
Clotha
2.30
(2.00)
2.00
2.00
2.00
2.00
2.00
2.00
Vb
32.29
(Calculated
31.21
27.36
33.89
27.19
29.24
32.22
AP
After
Clean-
ing
6.31
Values)
5.98
4.81
5.84
4.92
5.10
5.81
AP
Before
Clean-
ing0
7.78
7.05
5.70
7.00
5.80
6.05
6.92
Avg.C
AP
7.05
(5.84)
6.52
5.26
6.42
5.36
5.58
6.37
F/Fd
AP
9.05
(7.84)
8.52
7.26
8.42
7.36
7.58
8.37
ft/min
b in. w.g. min. ft. gr.
Ib cu. ft.
m
in. w.g.
Estimated flange-to-flange
Unlike the remainder of the fabrics, the first test fabric (Ken 617F)
was run at an air-to-cloth ratio of 2.3 ft/min. The results of this test
indicated that the higher air-to-cloth ratio was not necessary to compen-
sate for a short testing period. After the Kennecott test the air-to-cloth
was lowered to 2.0 ft/min. To allow comparison with the rest of the fab-
rics, the Kennecott results were projected at an air-to-cloth ratio of 2.0
ft/min.
The data on the MS601-T and GA877-Q78 fabrics (see Table IV) must be
viewed cautiously since neither came to equilibrium during the testing
period. In both cases the pressure drop monotonically increased throughout
the testing period. This phenomenon of slowly approaching operating
equilibrium has also been observed for the MS601-T fabric in full-compart-
ment testing on the Harrington prototype baghouses.
FF502-1AF and FF504-1TC fabrics had the same pressure drops. This
result is again in agreement with full-compartment testing done on the
shake/deflate Harrington filter. In these full compartment studies no
difference has been noted between 10-oz. Teflon-coated fabrics and 14-oz.
acid-resistant coated fabrics in terms of pressure drop.
248
-------
The FF502-ITC fabric is a 14-oz. fabric with a 10 percent Teflon
coating. This combination represents the "high-priced spread" in that the
more expensive 14-oz. fabric and Teflon coating are. combined. The fabric
showed pressure drops in the same range as fabrics FF502-1AF and 'FF504-1TC.
Testing of 14-oz. Teflon-coated materials in the prototype indicated
reduced flow ( 10 percent), but the pressure drop was not substantially
different than the commonly used FF502-1AF and FF504-1TC fabrics.
2
The FF504-1AF fabric is a lO-oz./yd fabric with an aci-'-resistant
coating. This fabric is essentially the "low-priced spread," c jining the
least expensive coating and fabric. The fabric exhibited one of the lower
pressure drops observed in the pilot for the standard fabrics, 7.58 in.
w.g. In the prototype this fabric has shown no lower pressure drop than
the other fabrics. Baglife on this fabric is now approximately 20 months
in two compartments in Unit 2. Observed wear on the fabric seems to indi-
cate that inferior bag life will be obtained under shake/deflate condi-
tions. The exact economic trade-off of bag cost versus bag life will be
studied for this fabric to see if it offers an economic advantage despite
its shorter bag life.
RING STUDY
The effect of various numbers of rings in a bag was studied using
samples having five, six, and seven rings (equally spaced). The fabric
chosen for this study was a lO-oz./yd Teflon-coated material. Originally
only MS601-T was to be used in the study, but its failure to come to
equilibrium made it necessary to include data previously obtained on
another lO-oz./yd Teflon-coated material (FF504-1TC). The results
obtained on these fabrics are shown in Table V.
TABLE V. RESULTS OF RING STUDY
Fabric
Style
MS601-T
FF504-1TC
MS601-T
MS601-T
No. of
Rings
5
6
6
7
va
36.70
32.22
27.19
31.74
AP
After
Clean-
b
ing
5.76
5.81
4.92
5.65
AP
Before
Clean-
ing
6.94
6.92
5.80
6.68
Avt.
AP6
6.35
6.37
5.36
6.16
F/Fh o
A Pb'c
8.35
8.37
7.36
8.16
»1
in. w.g. miii. ft. gr.
Ib cu. ft.
m
b .
in. w.g.
Q
Estimated flange-to-flange
249
-------
The pressure drops for the three fabrics that obtained equilibrium
during the testing show very little difference. From this data it would
appear that the number of rings in a bag do not have profound effects on
pressure drop. Outside the range of five to seven rings in a 30-ft bag,
however, some effect on pressure drop may occur.
NON-STANDARD FABRICS
One area of fabric filter research left virtually untouched in the
utility industry is the effect of fabric weaves and construction on perfor-
mance of filtration materials. To date only texturized 3x1 twill weaves
have been employed to any extent on glass fabrics. Certain experiences
with 3x1 and 3x2 twill Nomex fabrics at Harrington pointed to weave as
being a potentially major factor in fabric performance.
Several fabric weaves and constructions that are uncommon to utility
application were tested in the pilot unit. Table VI lists the results for
these non-standard fabrics.
TABLE VI. RESULTS OF TESTING ON NON-STANDARD FABRICS
Fabric
Style
MS 3x2 T
MS-CF
MS 2x2 T
MS509N
No. of
Rings Weave
6 3x2 T
6 Crowfoot
6 2x2 T
7 Napped
32.51
26.93
16.07
32.69
AP
After
Clean-
5.05
5.02
3.09
5.73
AP
Before
Clean-
ing
6.71
5.94
3.57
6.75
Avgfi
APB
5.61
5.48
3.33
6.24
P-
7.61
7.48
5.33
8.24
in. w.g. min. ft. gr.
Ib cu. ft.
m
in. w.g.
Estimated flange-to-flange
The 3x2 twill and the crowfoot fabrics have significantly lower pres-
sure drops than most of the fabrics in common use. No operational diffi-
culties were encountered during the testing of either of these fabrics, but
the crowfoot weave exhibited a rather strange phenomenon. Initially the
pressure drop rose after start-up and leveled off at close to 5.5 in. w.g.
For the last 3 days of the test period, however, the pressure drop de-
250
-------
creased suddenly to,approximately 4.3 in. w.g. To be sure of interpreting
this data conservatively, the last few days of operation were ignored in
the data analysis. Further analysis of the crowfoot fabric would certainly
be wise.
The 2x2 twill fabric exhibited pressure drops below 2 in. w.g. for the
first 4 days and increased slowly throughout the testing period. Only the
last few days of testing were used to evaluate the 2x2 twill fabric. Since
it did not come to equilibrium by leveling off in pressure drop, the data
are not satisfactory. Again, more testing should have been done, but time
was not available.
The seven-ringed napped fabric with a pressure drop of 8.24 in. w.g.
failed to show any advantage over the 3x1 twill seven-ringed texturized
fabric at a pressure drop of 8.16 in. w.g.
ADDITIONAL STUDIES
K c RESULTS
The K.c values obtained during the scale-up period are shown versus
air-to-cloth ratio in Figure 3.
K c decreases as the air-to-cloth ratio is lowered. At air-to-cloth
ratios lower than 2.7 ft/min, however, no significant changes in K c are
observed. If grain loading can be assumed constant, this implies tnat K
also decreases with air-to-cloth ratio and does not change significantly
below an air-to-cloth ratio of 2.7 ft/min. The minor dip at an air-to-
cloth ratio of 2.5 ft/min may be an anomaly. If not, it is due to settling
out of heavier particles as the velocity gets lower and lower.
The range of K c values measured for reverse air cleaning on the
various test bags is also shown in Figure 3. As shown, the K values for
reverse air cleaning are substantially larger than the K« values obtained
for shake/deflate cleaning at the same air-to-cloth ratio (2.0 ft/min).
Cleaning then is shown to influence K as well as ash properties. This
effect could reflect the length of time required to"repair" the ash cake
after cleaning. Reverse air cleaning causes fissures in the ash cake along
with some areas where cake has sloughed off. After cleaning, these areas
are quick to rebuild with the result that any additional cake is generally
distributed and influences the pressure drop increase. After shake/deflate
cleaning, large areas of the cloth are stripped bare to the surface. The
time required to repair these larger areas strongly influences the rate at
which pressure drop can increase across the filter cake.
DESIGN EQUATIONS
The pilot data obtained in the scale-up and reverse air studies can be
used to develop pressure drop relationships for use in design. The equa-
tions are:
251
-------
AP = 1.60V1'34 Shake/deflate (A)
AP = 3.29V1'34 Reverse air (5)
Figure 4 shows a plot of pressure drop versus air-to-cloth ratio
predicted from these equations.
Strictly speaking, these correlations apply only to one ash type,
certain fabrics, and a given set of cleaning conditions. The data used in
developing these correlations, however, represent the cloth used in over 90
percent of the baghouses in operation on utility boilers. In addition, the
cleaning conditions employed are representative of those used by many
installations. These equations then should be useful in modeling filtra-
tion of ashes similar to the Wyoming ash produced at Harrington Station.
Because the operating conditions during testing were deliberately
chosen to simulate operations at high capacity factors, these correlations
should not be used for peaking units and other low load applications.
CONCLUSIONS
The fabric filter pilot unit at Harrington Station has successfully
demonstrated the ability of a pilot unit to model large scale fabric filter
systems. The pressure drop model developed from the pilot unit describes
both the Harrington fabric filters (Unit 2 and Unit 3) with good accuracy
and should be a useful tool in design of units with similar fly ashes.
Although no direct comparison from pilot to prototype was available
for the reverse air studies, there is no apparent reason that the reverse
air equation should not be accurate for the ash type studied.
The experience gained from Harrington Station's pilot and prototype
units indicates that the typical design goal of 6.0 in. w.g. flange-to-
flange pressure drop can be obtained with a properly designed shake/deflate
baghouse at an air-to-cloth ratio of approximately 2.7 ft/min. The reverse
air study indicates the maximum air-to-cloth ratio for a reverse air bag-
house on western ash is in the 1.5-1.7 ft/min range.
The reverse air fabric study points to the need for more research in
the area of fabric weaves and construction. Shown below in Table VII are
all the fabrics tested arranged according to increasing pressure drop.
252
-------
15
14
13
12
IO-
_§.£
U-i 9
1 o
2a 8
i
fi»3 7
5
4
3
2
I •
0-
Reverse Air
Shake / Deflate
1.5
2.0 2.5
Air to Cloth Ratio, ft/min
3.0
3.5
Figure 4. Plot of pressure drop vs. air-to-cloth.
253
-------
TABLE VII. FABRICS BY PRESSURE DROP
Fabric
Weigh£
oz/yd Coating Weave
F/Fd
AP Comments
MS601-T
GA 877-Q78
MS601-T
MS601-T
FF504-1AF
MS601-T
Ken-617-F
MS601-T (7 Rings)
MS509N (7 Rings)
MS601-T (5 Rings)
FF504-1TC
FF502-1TC
FF502-1AF
10
14
10
10
10
10
10
10
10
10
10
14
14
Teflon
078
Teflon
Teflon
Acid Flex
Teflon
CS-428
Teflon
Teflon
Teflon
Teflon
Teflon
Acid Flex
2x2
3x1
3x1
Twill
Twill
Twill
Crowfoot
3x1
3x2
3x1
3x1
3x1
3x1
3x1
3x1
3x1
Twill
Twill
Twill
Twill
Twill
Twill
Twill
Twill
Twill
5
7
7
7
7
7
7
8
8
8
8
8
8
.33
.26
.36
.48
.58
.61
.84
.16
.24
.35
.37
.42
.52
Equilibrium not
Equilibrium not
Equilibrium not
Calculated
obtained
obtained
obtained
Flange-to-flange, in. w.g.
The highest pressure drops recorded during tests were for the three
most commonly used fabrics. Two of the non-standard weaves, the 3x2 twill
and the crowfoot, did show significantly lower pressure compared to the
commonly used types. Although the 2x2 twill fabric did not come to equi-
librium, its behavior during the test program wpuld indicate that it may
also be a candidate for enhanced performance. The three lowest pressure
drop fabrics are those that did not come to equilibrium during the testing
period, and no clear conclusion can be drawn from the data about their
performance.
These results demonstrate that better performance may be obtained if
further research is aimed at developing fabrics especially suited for util-
ity filtration systems. Considering the costs of premature bag replacement
and the cost of high pressure drop, further studies could lead to greater
cost-effectiveness.
REFERENCES
1. Ladd, K. L., Hooks, W. , Kunka, S. L., and Harmon, D. SPS Pilot Bag-
house Operation. In Third Symposium on the Transfer and Utilization
of Particulate Control Technology: Volume I. Control of Emissions
From Coal-Fired Boilers, EPA-600/9-82-005a (NTIS PB83-149583), pp
55-64, July 1982.
254
-------
THE USE OF SONIC AIR HORNS AS AN ASSIST TO REVERSE
AIR CLEANING OF A FABRIC FILTER DUST COLLECTOR
by: Alan R. Menard
R. Mark Richards
Public Service Company of Colorado
Arapahoe Station
Denver, Colorado 80201
ABSTRACT
A detailed summary of the installation, test proqram, and performance of
sonic air horns used as an assist to reverse air cleaning of an operating
FFDC is presented. Four sonic air horns were installed and tested on com-
partment number 11 of Public Service Company of Colorado's Arapahoe Unit #3
FFDC. The overall objective of the test program was to reduce the pressure
drop across the tube sheet of the test compartment. Sonic cleaning reduced
the compartment tube sheet differential pressure by approximately 27 percent;
this was sufficient to warrant the purchase and installation of two (2) sonic
air horns in each of the 14 baghouse compartments.
255
-------
INTRODUCTION
The need to decrease the pressure drop across the filtering elements in
a Fabric Filter Dust Collector (FFDC) is inherent to the reduction of its
operating costs. Public Service Company of Colorado's Arapahce Unit #3 FFDC
has experienced flange to flange pressure drops on the order of 10 inches VWC
at full load as opposed to the design pressure drop of 6 inches VWC. Much of
this discrepancy in pressure drop can be attributed to the formation of
nodules (Reference 1) or cake on the inner surface of the filtering elements
which is not removed during the reverse air cleaning cycle (Figure III-l).
The formation of nodules or cake reduces the effective filtering area of the
baghouse, which in turn, increases operating air-to-cloth ratio.
A solution was sought by Arapahoe Station personnel that would decrease
the overall pressure drop of Arapahoe Unit #3 FFDC. Because the nodule/cake
buildup on the filtering elements was felt to be directly related to the
pressure drop problem, the use of sonic air horns used in conjunction with
the reverse air cleaning cycle was thought to be a viable solution to the
problem of fly ash removal. A test program was devised that would determine
the effect of using sonic air horns during the reverse air cleaning cycle as
a means of reducing the pressure drop across the filtering elements.
Four Fuller air horns were purchased and installed in compartment #11 of
Arapahoe Unit #3 FFDC. Compartment #4, having the most similar flow
characteristics to compartment #11, was used to collect base data for compar-
ison purposes during the test period. The use of the sonic air horns made an
appreciable improvement in the performance of the test compartment; signifi-
cant enough to warrant the purchase and installation of horns in all 14 bag-
house compartments.
This report summarizes the installation, test program and performance
of sonic air horns used as an assist to reverse air cleaning of a FFDC.
256
-------
TECHNICAL DISCUSSION
BAGHOUSE DESIGN AND HISTORY
Public Service Company of Colorado's Arapahoe Unit #3 FFDC was designed
for 315,000 acfm flow with a 6-inch VWC pressure drop flange to flange, and
a 2.16 operating air-to-cloth ratio. The boiler/turbine combination has a
48 MW capability and operates on low sulphur, sub-bituminous western coal.
The baghouse is a reverse air cleaning type, consisting of two rows of seven
modular compartments with inlet, outlet, and reverse air ducts located
between the two rows of compartments (see Figure II-l). Each compartment
contains a total of 236 bags which are arranged as shown in Figure II-2.
The bags are 8 inch diameter, 22 feet long, woven fiberglass with teflon B
coating. The baghouse was started'up on May 18, 1979, and reached initial
full load operation on May 21, 1979. For complete design specifications on
Arapahoe Unit #3 baghouse, see Table 1.
After initial operation of the baghouse, several mechanical problem
areas such as valve seating, air cylinder operation, and thimble leaks were
recognized and repaired. With these problems solved, performance testing of
the baghouse began. Originally, the baghouse was outfitted with 8-1/4 inch
diameter Acme Mills woven fiberglass bags with teflon coating. After
approximately two months of operation, the baghouse experienced flange to
flange pressure drops inexcessof the design pressure drop of 6 inches VWC
at full load. Acceptable pressure drops could only be obtained by hand clap-
ping and/or beating the bags to remove the cake buildup on the inside sur-
face of the filtering elements. However, as the cake began to build up on
the bags, the pressure drop would again exceed design levels. The baghouse
manufacturer felt that the pressure drop problem could be solved by replacing
the O.E.M. bags with Menardi Southern bags with teflon B coating.
Compartments 1 and 7 were outfitted with the Menardi Southern bags and
a test program was initiated. The performance of the two compartments out-
fitted with the new bags was significantly better than the other compartments
(Reference 2). It was decided to install the Menardi Southern bags in all
14 baghouse compartments. Eight-inch diameter bags were chosen as opposed to
8-1/4 inch diameter bags because of long lead times associated with obtaining
the 8-1/4 inch diameter bags. This three percent reduction in filtering area
was felt to be outweighed by the performance increase with the new bags.
During this time period, data acguisition and testing was initiated to
optimize the operation of the baghouse. Different types of coal were burned.
Although some coals increased operating flange to flange pressure drop, no
coal was found that decreased pressure drop. Optimization of the cleaning
cycle was also attempted. Multiple settling/reverse air periods, continuous
257
-------
Table 1
ARAPAHOE UNIT 3 FFDC DESIGN SPECIFICATIONS
DESIGN CONDITIONS
Process .......................................... Power Generation
Suspended Material ............................... Fly Ash
Fuel ............................................. Pulverized Coal
Gas:
Source ................................... ..... Coal-Fired Boiler
Gas Flow (max. acfm) .......................... 315,000
Temperature (°F) .............................. 290
Outlet Loading (fr/acf) ....................... 0.007
Pressure Drop (in. VWC) ....................... 6.0
Design Pressure (in. VWC) ..................... ±20
Total Filter Area (sq ft) ..................... 156,995
Effective Filter Area (sq ft)± ................ 145,781
Total Filter Ratio ........................ ---- 2:1
Active Filter Ratio± ......................... 2.16:1
Reverse Air Flow:
Total (max acfm) ............................ 16,700
Static Pressure (in VWC) .................... 10.6
Reverse Air Fan (qty) ....................... 2 (1 on standby)
FFDC DATA
Type . ............................................ Modular
No. of Units ...... .... ........................... 2
No. Compartments/Units ........................... 7
Baqs:
Total Quantity ................................ 3304
Per Compartment ............................... 236
Material ...................................... Menardi Southern, Teflon B
coating
Diameter (in.) ................................ 8
Length (ft) ...... ............................. 22
Power Requirements ................. .. ............ 480V, 3 phase, 60Hz
Air Flow Regulation
Hopper Inlet (Isolation) ...................... Butterfly ° 14
Reverse Air/Outlet Valve ...................... Double Disc0 14
Reverse Air Control (Nul 1 ) Damper ............. Butterfly ° 1
Outlet Duct (Isolation) ____ . .................. Butterfly ° 14
By-Pass Damper ................. . .............. Poppet ° 2
±With one compartment out for cleaning and one compartment out for
maintenance
0 Pneumatic operated
258
-------
Figure II-l Arapahoe Unit 3 Baghouse Compartment Locations
2
/3
10
8
t
NOKTH
\
XKJUET
259
-------
Figure II-2 Typical Compartment Filter Element Location
A 8 C
D C F
H 1
/
2
3
-f
j^
..}
1C
7
8
3
SO
//
/a
/3
H
tS
Ho
17
16
n
-">,-\
>..-'
•? /
^-/
f-— f~^
23
?--i
OOO
ooo
OOO
ooo
ooo
ooo
ooo
ooo
ooo
ooo
0 0 O
ooo
ooo
ooo
ooo
ooo
O O O
C) O O
ooo
0 O O
O O O
C 0 O
C1 O 0
coo
OOOG
0000
C O 0 O
oooo
oooo
oooo
oooo
oooo
oooo
oooo
oooo
oooo
oooo
oooo
oooo
oooo
O O 0 0
.oooo
O O O O
o o o 6
0 0 0 0
oooo
oooo
. !.„.,-. , i
ooo
ooo
000
ooo
ooo
ooo
ooo
000
ooo
ooo
0 O O
O O 0
O 0 0
0 0 O
ooo
ooo
0 O 0
ooo
o o o
ooo
ooo
OO ^)
on ) •
00 )
i
!
-- DOOR.
260
-------
cleaning as well as differential pressure initiation of the cleaning cycle
were also tested. None of these experiments made any significant improve-
ment in flange to flange pressure drop.
After approximately three months of operating with the new Menardi
Southern bags, the pressure drop from the inlet flange to the outlet flange
again exceeded the design specification of 6 inches VWC. To maintain a
reasonable pressure drop, the bags still had to be periodically hand beaten
to remove the cake/nodule buildup on the inside surface of the filtering
elements. As a result of this, Arapahoe Station personnel, with assistance
from the staff at EPRI's Emission Control Test Facility at Arapahoe Station
sought a solution to the pressure drop problem. Because all attempts at
optimizing the cleaning cycle and the switching to a different filtering
element made no appreciable improvement on the overall flange to flange
pressure drop, it was felt that the nodule/cake buildup on the filtering
elements was directly related to the pressure drop problem. The use of sonic
air horns as an assist to the reverse air cleaning cycle was felt to be a
viable solution to the problem of cake buildup not being removed from the
filtering elements with reverse air cleaning only. A test program was devel-
oped that would determine the effect of sonic cleaning on the nodule/cake
buildup problem, and ultimately would determine if sonic cleaning would re-
duce the operating flange to flange pressure drop.
HORN INSTALLATION
Four sonic air horns were obtained on a trial basis for installation at
Arapahoe Station from Fuller Company. Because of time limitations, a test
program was devised that was somewhat limited in scope but would still deter-
mine the overall effectiveness of sonic cleaning. Arapahoe Station person-
nel, in conjunction with the staff at EPRI's Emission Control Test Facility
at Arapahoe Station, assembled a list of critical parameters that would be
monitored during the test period. During these initial discussions, decisions
were made on horn placement, boiler loading, horn operation and special
equipment needed to control horn operation.
Four Fuller air horns, model 310-64-1-0099-01 were installed in compart-
ment 11 of Arapahoe Unit #3 FFDC. It was thought that the compartment should
be divided into four equal quadrants, and a horn placed at the center of each
quadrant. The location of the "Doghouse" containing the outlet and compart-
ment isolation valves made this objective impossible and the horns were
placed as shown in Figure II-3. Figure II-4 shows the installation of a
typical horn through the roof of the compartment.
Each air horn requires 70-90 psig air at 50 scfm. The control air
system that operates the existing air cylinders on top of the baghouse, at
80 psig, was adequate to supply the horns. However, it was felt that an air
receiver tank was necessary to supply the four horns with a sufficient volume
of air. A 30 cubic foot air receiver tank was purchased and installed on top
of the baghouse. A one (1) inch diameter manifold was run from the receiver
tank to the horns with one-half U) inch copper tubing branching from the
manifold to each horn.
261
-------
Figure II - 3 Test Compartment Horn Locations
Horns
"Doghouse" •
Outlet
Duct
Reverse
Air Duct
Side View
c
Horns
3 0
Top View
262
North
-------
Figure 11-4 Compartment Horn Installation
263
-------
The question of when to operate the horns and for what duration during
the cleaning cycle was discussed. Having limited experience with sonic air
horns, three options were considered:
(1) Operate the horns during the initial settling period (no reverse
air);
(2) Operate horns during reverse air cleaning;
(3) Operate horns during the final settling period.
Because of time limitations on the Fuller air horn trial period, it was
not feasible to optimize horn operation. It was decided that option two
would achieve the best results. The horns would be operated as an assist to
the reverse air fan for a period of ten seconds. This proved to be the best
method of horn operation and will be discussed later.
A means of controlling horn operation was devised. A normally closed
solenoid valve was installed in the line exiting the air receiver tank that
would open when horn operation was desired. At the baghouse control panel,
reverse air damper movement (initiation of compartment clean) would initiate
a primary timer. After 40 seconds, the first timer would initiate a second
timer which operated the solenoid valve for a period of ten seconds. This
system served to operate the horns at the approximate time the reverse air
flow reached the compartment.
TEST PROGRAM
The objective of the test program was to determine the effect of sonic
cleaning on flow and pressure drop in the test compartment. An increase in
flow and a decrease in pressure drop in the test compartment was desired.
Four Fuller air horns were installed on compartment #11 of Arapahoe Unit #3
FFDC and it was designated the test compartment. Compartment #4, having the
most similar flow characteristics to compartment #11, was used to accumulate
base data for comparison purposes. The following is a list of parameters
that were monitored during the test period:
(1) Pressure drop across the thimble sheet of the test compartments;
(2) Exit flow of the test compartments;
(3) Flange to flange pressure drop;
(4) Inlet, outlet, reverse air, and compartments 4 and 11 exit gas
temperature;
(5) Compartment ash pulling times;
(6) Bag weights - before and after cleaning.
For comparison purposes of data taken on different days, it was decided
264
-------
that a constant boiler load was essential. Variations in load would cause
fluctuations in flow, pressure drop, etc., which would have made data inter-
pretation and correlation more difficult. The unit was base loaded at 46 MW
(full load) one (1) day per week during the test period. Although the horns
continued to operate during periods of varying load, the majority of relev-
ant data was taken during the periods of base loading. The instrumentation
and recorders were installed and were ready for operation by February 22,
1982.
Before the test period began, bags were weighed in each of the test
compartments before and after cleaning. Ash pulling times were io record-
ed. The recorders were placed in service to accumulate base dat^. before any
horns were operated.
It was decided to test the horns in successive combinations of 1, 2, or
4 horns to determine the minimum amount of horns required for successful sonic
cleaning. On February 23, 1982, one horn was operated. On February 25,
1982, 2 horns and on March 29, 1982, 4 horns were tested. Ash pulling times
were recorded each Monday.
During the test period, the baghouse was on a continuous clean cycle.
There was a ten minute delay between-each compartment clean, and it would
take approximately 2.75 hours to clean all 14 compartments. Reverse air flow
was held constant during the test program. For a complete log of the test
program, see Appendix A.
255
-------
TEST RESULTS AND DISCUSSION
Introduction
The use of sonic air horns as an assist to reverse air cleaning proved
to be a viable solution to the problems of nodule/cake formation on the
filtering elements and increased pressure drop in an operating FFDC. During
the test period, bag weights, filter element cake thickness and tube sheet
differential pressure were all reduced in the test compartment. In addition,
flow through compartment 11 increased during the period of horn operation.
Results indicated that two (2) horns per compartment were sufficient to
adequately clean the compartment and maintain the reduction in pressure drop.
As a result of this test program, Public Service Company of Colorado has
purchased and installed two (2) sonic air horns in each of the fourteen (14)
baghouse compartments at Arapahoe Station's Unit #3 FFDC. This section
summarizes the performance of sonic air horns used as an assist to reverse
air cleaning of a FFDC.
Discussion of Results
The overall objective of the test program was to reduce the pressure
drop across the thimble sheet in the test compartment. The following para-
meters were monitored during the test program:
(1) Bag weights and samples;
(2) Compartment ash pulling times;
(3) Thimble sheet differential pressure;
(4) Test compartment exit flows.
A summary of the effect of sonic cleaning on each of the above listed
parameters is presented below.
Bag Weights and Samples
Probably the most significant effect of sonic cleaning was on the fil-
tering elements (bags) themselves. Before the sonic horns were operated, bags
wereweighed in both compartments 4 and 11. Bag weights ranged from 34 to 55
pounds with the average weight being approximately 46 pounds. At several
points during the test program, the test compartments were isolated and bags
were weighed. Compartment 4's filtering elements remained the same in
weight while the bag weights in compartment 11 were dramatically reduced.
266
-------
kr*w . VT, , • %,»TMB^MNfTT1'1? • " <§• '• •' v • iiy^v •». ^^
-vBilff ^>
Figure III-l Nodule/Cake Formation on a Typical Filtering Element - Without
Sonic Cleaning
• .
,
. • -*:.
Figure III-2 Nodule/Cake Formation on a Typical Filterinn Element - With
Sonic Cleanina
267
-------
Weights in the test compartment ranged from 12-1/2 to 25 pounds with the
average weight being approximately 18 pounds. A new filtering element weighs
approximately 9 pounds. The sonic air horns reduced the amount of fly ash"
cake buildup on the filtering elements an average of 76 percent. This
reduction in weight has an added benefit. The filtering elements are ten-
sioned to 50 pounds in the compartments. As bag weights approach 50 pounds,
it is difficult to maintain proper bag tensioning which affects the way the
bags collapse during reverse air cleaning. With the reduction in bag weights
due to sonic cleaning, more uniform filtering element tensioning is possible.
For a complete tabulation of bag weights taken during the test program see
Table 2.
Samples of the filtering elements were taken in each of the test com-
partments before and after cleaning. Samples were cut from the top, middle
and bottom of selected bags without disturbing cake buildup. The difference
in nodule formation with and without sonic cleaning can be seen in Figures
III-l and III-2.
Compartment Ash Pulling Times
Another means of determining the effectiveness of sonic cleaning was to
monitor the amount of time required to remove the fly ash from the compart-
ment ash hoppers. An increase in the amount of time required to remove the
ash from the hopper on the compartment with sonic horns would indicate that
the cake buildup on the filtering elements was being removed. Ash pulling
times before the test program began were recorded and are shown in Figure
III-3. Figure III-4 shows compartment ash pulling times during one and two
horn operation. Assuming that the curves in Figure III-4 would be similar
in shape to the curves in Figure III-3 without sonic cleaning, ash pulling
times in compartment 11 increased by approximately 90 percent. It was ex-
pected that after the majority of cake buildup in the test compartment had
been removed by sonic cleaning, the ash pulling times for compartment 11
would decrease from the ash pulling times recorded when the horns were first
operated. Figure III-5 shows the compartment ash pulling times after two
weeks of continuous two hourn operation. The ash pulling time for compart-
ment 11 should be slightly higher than normal because of the increase in flow
through the compartment due to sonic cleaning (discussed below). However,
the increased amount of fly ash collected in compartment 11 over the 2.75
hour time period between cleanings is difficult to measure and does not show
up readily in the data.
Thimble Sheet Differential Pressure
The pressure drops across the tube sheet of compartment 4 and 11 were
recorded on a two pen strip chart recorder. Data was recorded at full load
before any horns were operated. Tube sheet differential pressure measured at
that time was 6.5 inches VWC immediately after the compartment cleaned and
would decay to 8.5-9 inches VWC immediately before cleaning (see Figure III-
6).
On February 23, 1982, with the unit stabilized at 45 MW, one horn in the
northeast corner of the compartment was operated. Pressure drop across the
tube sheet in compartment 11 was reduced to 3.2 inches VWC immediately after
268
-------
Bag #
13G
20D
IOC
200
15C
10G
19G
16H
11D
140
18G
8H
5C
50
5G
5H
12C
120
12G
12H
20C
200
20G
20H
7C
200
13G
20D
20G
80
8H
12G
15C
Compartment
4
4
4
11
11
11
11
11
11
4
4
4
11
11
11
11
11
11
11
11
11
11
11
11
11
11
11
4
4
11
11
11
11
Weight
6 Ibs.
4 Ibs.
34 Ibs.
46i Ibs.
49 Ibs.
52 Ibs.
47 Ibs.
49 Ibs.
55 Ibs.
45X Ibs.
45 Ibs.
37 Ibs.
12| Ibs.
14 Ibs.
15 Ibs.
16* Ibs.
15 Ibs.
14 Ibs.
144 Ibs.
15 Ibs.
19 Ibs.
154 Ibs.
15| Ibs.
18 Ibs.
18i Ibs.
18 Ibs.
19 Ibs.
36 Ibs.
34 J Ibs.
20 Ibs.
24 Ibs.
19 Ibs.
22 Ibs.
Table 2
BAG WEIGHTS
Comments
0 horns, before cleaning
0 horns, before cleaning
0 horns, before cleaning
0 horns, before cleaning
0 horns, before cleaning
0 horns, before cleaning
0 horns, after cleaning
0 horns, after cleaning
0 horns, after cleaning
0 horns, after cleaning
0 horns, after cleaning
0 horns, after cleaning
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, before cleaning 3/17/82
2 horns, before cleaning 3/17/82
2 horns, before cleaning 3/17/82
0 horns, before cleaning 3/17/82
0 horns, before cleaning 3/17/82
2 horns, after cleaning 3/18/82
2 horns, after cleaning 3/18/82
2 horns, after cleaning 3/18/82
2 horns, after cleaning 3/18/82
269
-------
15D
17G
22D
22G
3C
3H
8C
8D
11G
11H
17C
20G
3C
3H
11D
11G
21C
21H
11
11
11
11
11
11
11
11
11
11
11
11
4
4
4
4
4
4
22i Ibs.
25 Ibs.
24 Ibs.
24i Ibs.
18 Ibs.
22 Ibs.
21i Ibs.
18 Ibs.
171 Ibs.
21 Ibs.
20 Ibs.
24i Ibs.
40 Ibs.
35i Ibs.
35i Ibs.
38 Ibs.
34 Ibs.
35 Ibs.
2 horns, after
2 horns, after
2 horns, after
2 horns, after
4 horns, after
4 horns, after
4 horns, after
4 horns, after
4 horns, after
4 horns, after
4 horns, after
4 horns, after
0 horns, after
0 horns, after
0 horns, after
0 horns, after
0 horns, after
0 horns, after
cleaning 3/18/82
cleaning 3/18/82
cleaning 3/18/82
cleaning 3/18/82
cleaning
cleaning
cleaning
cleaning
cleaning
cleaning
cleaning
cleaning
cleaning
cleaning
cleaning
cleaning
cleaning
cleaning
270
-------
Figure III - 3 Compartment Ash Pulling Times - Pre Test Data
to
m
-------
Figure III-4 Compartment Ash Pulling Times During One and Two Horn Operation
12-
10-
8-
to
~j
to
M
0)
1
CO
6-
•s
k-
Note: Differences in ash pulling times for one and two horn operation are
due to boiler load fluctuations and time between ash pulling cycles.
The intent of the Figure is to show time differences between
compartment 11 and the rest of the curve.
2 horns operating
horn operating
,
_,.
-------
Figure III - £ Compartment Ash Pulling Times After Two Weeks of Continuous Two Horn Operation
to
18-
16-
CO
0
ft
CQ
k-
T~
2
k
~T
5
7
10 11 12 13
Compartment
-------
FIGURE Itt-6. COMPARTMENT TUBE SHEET
PRESSURE DROP AT FULL LOAD
NO HORNS OPERATING
85 10-
-------
cleaning (see Figure III-7). It was assumed that the pressure drop across
the tube sheet would increase at the same rate as in the compartment without
sonic horns. Pressure drop increased at a more rapid rate until it was equal
to the pressure drop in compartment 4 immediately before compartment 11
cleaned. This phenomenon can be explained because accompanying the decrease
in pressure drop there was a substantial increase in flow through the test
compartment. This indicates that a greater amount of fly ash was being col-
lected in the compartment with sonic horns than in the other compartments,
hence, the rapid increase in pressure drop.
On February 25, 1982, two horns, one in the northeast and one in the
southwest corner of the test compartment were operated. Pressure drop across
the tube sheet was reduced to 3.0 inches VWC immediately after the compart-
ment cleaned. Again the increased rate of pressure drop decay was observed.
During this time period, unit 3 was loaded as P.S.Co. system require-
ments mandated. This served two purposes; first, to allow the baghouse test
compartment to experience a more "real world" environment, and secondly to
allow the test compartment time to seek its own steady state operating point.
Pressure drop across the tube sheet increased to 4 inches VWC at full load
immediately after cleaning after about one week of continuous two horn
operation. This pressure drop remained the same for the remainder of the two
horn test period (approximately 1 month). It was felt that steady state
operation had been reached.
Figure II1-8 shows tube sheet pressure drop versus time for two horn
operation after steady state operation had been reached. The dashed line in
the Figure represents a corrected pressure drop curve which assumes that the
pressure drop decay would be similar to the curve of the compartment with no
horns operating if sonic horns were installed in all fourteen compartments.
During baghouse operation without sonic horns, flange to flange press-
ure drop measured between 9 and 9.5 inches VWC or approximately i to 1 inch
VWC higher than the tube sheet differential pressure immediately before the
compartment cleaned. It is assumed that the flange to flange pressure drop
curve would be similar in shape with the use of sonic horns in all fourteen
compartments to the pressure drop curve without sonic horns. This would
equate to a flange to flange pressure drop of 6.5 to 7 inches VWC if two
sonic horns were installed in all fourteen compartments. This represents
approximately a 27 percent decrease in flange to flange pressure drop if the
horns were installed in all fourteen compartments.
We believe that this is a conservative estimate of performance improve-
ment because of the significant increase in flow through the test compart-
ment. A portion of the pressure drop through the compartment could be due
to the fact that the test compartment flow was higher than the flow in the
remaining 13 compartments (discussed below). When sonic horns are installed
on all fourteen compartments, the flow will be more evenly distributed
throughout the baghouse and the pressure drop across the tube sheets should
decrease.
It is interesting to note that the horns were accidentally operated at
275
-------
Figure III - 7 Compartment Tube Sheet Pressure Drop at Pull Load -
One Horn Operating
0)
1
CM
o
m
,8
I
8-
CD
-p
o>
CO
I
10 .
8.
6-
2 .
Compartment 11
Compartment
Clean •—v^
Compartment
1
Time, Hours
276
-------
FIGURE
8. COMPARTMENT TUBE SHEET
PRESSURE DROP AT FULL LOAD
TWO HORN OPERATION
o:
UJ
0
CO
LLJ
X
O
en
UJ
o:
o_
Theoretical curve assumes the pressure drop curve
would be similar in shape to the pressure drop curve
with no horns operating if the sonic horns were
installed, on all fourteen compartments.
IO
6
CL
O
cr
Q
Ld
or
ID
r-COMPARTMENT 4
COMPARTMENT l!
234
TIME. HOURS
277
-------
the wrong time during the two horn test period. Because of a mistaken ad-
justment in the timing sequence that operated the horns, the horns were oper-
ated during the initial settling period of the compartment clean (no reverse
air). Pressure drop across the tube sheet in compartment 11 returned to its
original no horn levels after two days of operating in this manner. With the
timing sequence problem solved, the horns were again operated during the re-
verse air period. Tube sheet pressure drop returned to 4 inches VWC immedi-
ately after cleaning.
On March 29, 1982, all four horns in the test compartment were operated
with the unit at full load. Pressure drop across the tube sheet of the
compartment with sonic horns was 5 inches VWC immediately after cleaning as
opposed to 7 inches VWC in the test compartment without sonic horns immedi-
ately after cleaning. The significance of this data is somewhat questionable
because at the time the boiler was operating on a different coal than that
which was used in the beginning of the test program. This new coal has been
known to significantly increase the operating flange to flange pressure drop
of the baghouse and on that day, flange to flange pressure drop was recorded
at 9.5 to 10 inches VWC. It can be seen from Figure III-9 that the sonic
horns did reduce the pressure drop across the tube sheet in the test com-
partment when compared to the pressure drop across the tube sheet of the
compartment without sonic horns.
Test Compartment Exit Flows
Exit flows were measured in the outlet ducts of compartment 4 and 11 us-
ing a S-type pitot tube. Data was recorded at full load before any horns
were operated and flow through both test compartments was from 23,600 acfm
immediately after cleaning to 15,850 acfm immediately before each compart-
ment cleaned (see Figure 111-10). A summary of test compartment exit flows
is presented in Table 3 and is also shown in Figures III-ll, 111-12, and
111-13. It can be seen that flow through the test compartment with sonic
cleaning did increase significantly and conversely, flow through the test
compartment without sonic horns decreased. It is assumed that flow through
the remaining compartments in the baghouse also decreased due to the 15.3
percent increase in flow through compartment 11. The greatest proportion of
flow followed the path of least resistance; through the compartment with
sonic cleaning. This would explain the rapid increase in tube sheet press-
ure drop noted earlier. With the installation of sonic horns in all four-
teen (14) compartments, flow will be more evenly distributed and will return
to the original no horn levels.
Because pressure drop across the tube sheet achieved steady state oper-
ation at an acceptable level with the use of two (2) sonic horns, it was
decided that two (2) horns per compartment would be the minimum amount of
horns required for successful sonic cleaning. Four (4) sonic horns per
compartment would have the advantage of more rapid removal of the initial
cake buildup., Once the initial cake buildup is removed from the filtering
elements, four horns show no appreciable improvement in compartment perfor-
mance.
278
-------
Table 3 - Test Compartments Exit Flows - Immediately Before and After Cleaning
0 Horns
1 Horns
2 Horns
4 Horns
Immediately After Cleaning
Compartment 11
23,600 acfm
29,900 acfm
27,200 acfm
26,500 acfm
. Compartment 4
23,600 acfm
22,400 acfm
20,800 acfm
20,100 acfm
Immediately Before Cleaning
Compartment 11
15,800 acfm
17,500 acfm
15,800 acfm
15,800 acfm
. Compartment 4
15,800 acfm
17,500 acfm
13,500 acfm
15,800 acfm
279
-------
Figure III- 9 Compartment Tube Sheet Pressure Drop at Full Load
Four Horns Operating
10-
I
•8 B
03
!
oT
I 6
m
I
•3
•H
(D
§
(D
o>
.Q
2-
Time, Hours
280
-------
Figure III- 10 Compartments l± and 11 Exit Flow vs Time - No Horns Operating
30,ooo-
o
H
I
20,000-
10,000-
Compartment 11
T~
3
Time, Hours
T
k
~r
5
Figure 111-11 Test Compartments Exit Flows vs Time - One Horn Operating
30,000-
20,000-
10,000-
o
o
-Compartment 11
T 1—
2 3
Time, Hours
T
5
281
-------
Figure 111-12 Test Compartments Exit Flow vs Time - Two Horn Steady
State Operation
30,ooon
8
20,000-
10,000-
o
Compartment 11
Compartment
TIME, HOURS
Figure 111-13 Test Compartments Exit Flows vs Time - Four Horns Operating
30,000n
20,000-
8
10,000-
o
•Compartment 11
T 1
2 3
TIME, HOURS
T~
k
282
-------
CONCLUSION
Because of boiler instabilities and resultant load limitations, along
with the high costs associated with operating Induced Draft (ID) fans at high
pressure drops, the need to reduce the operating flange to flange pressure
drop of a FFDC is imperative. Assuming the 27 percent decrease in flange to
flange pressure drop noted earlier, a savings of approximately $52,400 per
year could be realized (Reference 3). It is estimated that the total cost of
purchasing and installing two horns in each of the fourteen baghouse compart-
ments would be $53,000. With the I.D. fan cost savings alone, this invest-
ment would pay for itself in one year. Added savings would be realized in
the elimination of the costs associated with the labor required for periodic-
ally hand clapping the filtering elements to remove the cake buildup.
The use of sonic air horns as an assist to reverse air cleaning is an
inexpensive and effective means of reducing the pressure drop across the
tube sheet of an operating baghouse. As a result of the information obtained
from this test program, Public Service Company of Colorado has purchased and
installed two sonic air horns in each of the fourteen compartments on
Arapahoe Unit #3 FFDC. Preliminary test results indicate a significant re-
duction in flange to flange pressure drop.
A test program has been initiated that will not only determine the net
reduction in flange to flange pressure drop of Arapahoe Unit #3 FFDC, but
will determine if higher operating air-to-cloth ratios are obtainable with
the use of sonic cleaning. If higher air-to-cloth ratios are achievable
while still maintaining acceptable pressure drops, significant capital sav-
ings could be realized on new baghouse installations through the reduction
of the number of compartments and/or total filtering area required for the
successful operation of a FFDC.
The work described in this paper was not funded
by the U.S. Environmental Protection Agency and
therefore the contents do not necessarily reflect
the views of the Agency and no official endorsement
should be inferred.
283
-------
REFERENCES
1. "Nodular Deposits in Fabric Filters", Robert F. Lembach and Gaylord W.
Penny, Journal of the Air Pollution Control Association, August, 1979.
2. Letter, Ronald F. Ross, Joy Industrial Equipment Company, to Gordon
Schott, P. S. Co., April 8, 1980.
3. "Evaluating Baghouse Systems for Energy Efficiency", Plant Energy
Management, February, 1982.
284
-------
APPENDIX A
TEST LOG
Date
2/22 Load 46 MW. Final setting up of horns. Obtained base data.
2/23 Load 45 MW. Operated 1 horn in NE corner of compartment.
Dropped 3 inches inAP.
2/24 Load 45 MW. Continued 1 horn test. Pulled ash and timed.
2/25 Load 46 MW. Operated 1 horn NE corner. At 2:30 p.m. 2 horns
operated NE and SW corners. Dropped 1 inch inAP.
2/26 Load 46 MW. Continued 2 horn test. Pulled ash and timed.
3/1 Load variable. Compartment isolated took bag weights. Com-
partment back in service. Continue with 2 horn test (NE and
SW corners) for next two weeks.
3/2 Load variable. Continued 2 horn test.
3/3 Load variable. Continued 2 horn test.
3/4 Load variable. Continued 2 horn test.
3/5 Load variable. Continued 2 horn test.
3/8 Load 46 MW. Continued 2 horn test. Pulled ash and timed.
3/9 Load variable. Continued 2 horn test.
3/10 Load variable. Continued 2 horn test.
3/11 Load variable. Inspected compartment 4 for bag leaks. Found
nothing wrong. Stood in compartment 4 while the horns operated
in compartment 11 to see if there was any sound attenuation to
explain the decrease in thimble sheet AP in |4. Felt no excess
vibration while horns were operating.
3/15 Load 46 MW. Two horn test continues. Pulled ash and timed.
Switched out wrong Agastat so horns were operating before reverse
air. Horns had no effect on tube sheetAP. Agastats fixed.
3/16 Load variable. Continued 2 horn test.
3/17 Load variable. Continued 2 horn test. Took bag samples in 4
and 11. Weighed bags in both compartments. All data taken before
cleaning.
285
-------
3/18 Load variable. 2 horns operating. Cut samples and weighed
bags in compartment 11 after cleaning. Took coal sample.
3/19 Load variable. 2 horns operating.
3/22 Load 46 MW. 2 horns operating. Pulled ash and timed.
2/23 Load variable. Continued 2 horn test.
3/24 Load variable. Continued 2 horn test.
3/25 Load variable. Continued 2 horn test.
3/26 Load variable. Continued 2 horn test.
3/29 Load 46 MW. Continued 2 horn test. At approximately 1:00 p,
operating 4 horns.
3/30 Load variable. Continued 4 horn test.
3/31 Load variable. Continued 4 horn test.
4/1 Load variable. Continued 4 horn test.
4/2 Load variable. Continued 4 horn test.
4/5 Load 46 MW. Continued 4 horn test.
4/6 Load variable. Continued 4 horn test.
4/7 Horns shut off at 7:00 a.m. Weighed bags in compartment 11.
No leaks found in compartment 11.
4/8 Load variable. No horns.
4/9 Load variable. No horns.
4/10 Load 46 MW. No horns. Pulled ash and timed.
286
-------
'ELECTROSTATIC' STIMULATION OF REVERSE-AIR^CLEANED FABRIC FILTERS"
by: D. A. Furlong and G, P. Greiner
ETS, Inc.
Suite C-103, 3140 Chaparral Dr., SW
Roanoke, VA 24018
D. W. Van Osdell
Research Triangle Institute
Research Triangle Park, NC 27709
L. S. Hovis
U. S. Environmental Protection Agency
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
ABSTRACT
The concept of electrostatic stimulation of fabric filtration (ESFF)
has been investigated on a slipstream of a pulverized-coal-fired boiler
using reverse-air-cleaned, woven-fiberglass fiber bags. Operation was
demonstrated using ESFF at a gas-to-cloth ratio (G/C) of 6 ft/min. An
un-electrified control house was simultaneously operated at a G/C of
3 ft/min. Under these conditions, the ESFF house maintained a pressure
drop equal to or less than the control baghouse. In addition to reducing
the filter cake pressure drop, ESFF was observed to apparently have long
term benefits in preventing irremovable dust buildup in the fabric.
This paper has been reviewed in accordance with the U. S. Environ-
mental Protection Agency's peer and administrative review policies and
approved for presentation and publication.
Readers more familiar with metric units may multiply ft/min by 0.305 for
the equivalent m/min.
287
-------
INTRODUCTION
This paper describes development activities from May 1981 to March
1982 using a pilot scale, slipstream fabric filter employing electro-
static stimulation of fabric filtration (ESFF).
The ESFF concept is based on laboratory scale tests at the Textile
Research Institute (TRI) as reported by Lamb and Costanza. The TRI
concept of ESFF consists of applying an electric field parallel to the
surface of the fabric through use of electrodes located at the upstream
surface of the filter fabric. Figure 1 is a schematic of a TRI electrode
"harness" located on the outside of a pulse cleaned bag and electrically
connected so that alternate electrode wires are at high electrical poten-
tial, producing an electric field of 2 to 4 kV/cm.
A team composed of Research Triangle Institute (RTI), ETS, Inc., and
TRI, under contract to the U.S. EPA, is evaluating this concept at pilot
scale. The pilot scale program was started in October 1979. The first
year's primary objective was to verify the TRI laboratory results under
actual flue gas conditions. EPA, as manager, and RTI, as prime contractor,
have directed the technical effort and performed initial electrical hard-
ware development and construction. ETS, Inc. has designed, built, installed,
and is operating the pilot unit. ETS and RTI are further developing the
ESFF hardware. The pilot plant is installed on a slipstream of a pulverized-
coal boiler at the E. I. DuPont de Nemours & Company plant, Waynesboro,
Virginia.
The pilot unit was designed to operate in either the pulse-jet (outside
collection) or reverse air (inside collection) cleaning mode. The baghouses
were first operated as pulse-jet collectors.- The principal conclusions of
the pulse-jet work, as reported previously, ' were:
1) At any given face velocity, the pilot ESFF baghouse had a
reduced residual pressure drop and a reduced rate of pressure
drop increase when compared to the pilot conventional bag-
house .
2) The pilot ESFF baghouse could be operated in a stable
fashion at face velocities up to about twice the stable
operating conditions for the pilot conventional baghouse.
3) A reduction in total annualized cost of about 30 percent
was estimated for an industrial boiler application utilizing
an ESFF baghouse at twice the conventional face velocity.
4) Particulate emissions from the two baghouses were not signi-
ficantly different.
5) The electrode configuration developed in the course of the
study, a combination pulse-jet cage/electrode, located on
the downstream side of the* filter fabric, was feasible and
had potential for commercial use.
288
-------
In May 1981, the pilot plant was converted to reverse-air cleaning.
This paper summarizes the development activities and reports preliminary
results and conclusions based on the work from May 1981 to March 1982.
THE PILOT SYSTEM
The pilot system provides the capability of simultaneous monitoring
of the ESFF effect through use of two identical baghouses receiving their
inputs from a common slipstream. One baghouse is supplemented with electri-
cal enhancement equipment; the other is not.
Figure 2 is an isometric drawing of the pilot unit, and Figure 3 is
the system schematic. Each baghouse can accept up to 9 reverse-air cleaned
bags each 20.3 cm (8 in.) in diameter and 224 cm (8 ft) long, or up to
13 pulse cleaned bags, 11.7 cm (4-5/8 in.) in diameter and 224 cm (8 ft)
long.
The provision of parallel control and experimental units is an impor-
tant feature of the pilot plant in that it permits operation in either of
two modes, parallel or separate. In parallel operation, both baghouses have
the same inlets, compensating for the cumulative effect of varying inlet
conditions. Any differences between the two units, either intentional or
unintentional, can be evaluated by parallel operation. Future work is
planned where separate tests will be conducted in each baghouse after
sufficient experience is in hand to negate the further need of a "control"
baghouse.
The pilot unit is installed on a slipstream from an industrial boiler
baghouse. Four pulverized-coal boilers are used with a normal load of
about 160,000 kg/hr (350,000 Ib/hr) of steam. The coal fed to the boilers
changes frequently, with sulfur content varying from less than 1 percent to
about 2 percent, and ash content ranging from 5 to 15 percent. The boilers
are sometimes co-fired with No. 2 or No. 6 fuel oil. Inlet dust loadings
at the pilot plant have varied from 0.6 to 5 g/scm (0.25 to 2 gr/scf);
0.7 g/scm (0.3 gr/scf) is typical. The mass mean diameter of dust particles
averaged 6.3 /um.
ESFF TEST HARDWARE
The electrical power requirements of the ESFF system are very low.
The power consumption measured by TRI and confirmed in the pilot system is
about 1 W/m^ (0.1 W/ft ) of cloth. This ESFF power requirement is equiva-
lent to approximately 0.1 in. of water pressure drop at a G/C of 6 ft/min.
The electrical system used in the pilot unit consisted of a variable
DC power supply (0 to 20 kV at 5 mA) for each of five power supply networks,
Current and primary voltage were measured separately in each network.
289
-------
Current limiting and meter protection circuits were used.
The electrode designs utilized to date have been based on the combi-
nation cage/electrode developed during the pulse-jet work. Again, these
cages were placed on the downstream (clean) side of the fabric, so as to
not restrict cleaning of the bag and to avoid dirt buildup on the cage.
These cages (patterned after the pulse-jet cage/electrode, constructed of
welded 1/8-in. (0.318 cm) rods, and referred to as "rigid cages") exhibited
two problems: first, fabric wear was observed at the insulator locations;
and second, the wire spacing of 2.5 cm coupled with insulator limitations
restricted field strength to about 4 kV/cm. New cages were fabricated
using 34 1/8-in. diameter electrodes (1.68 cm spacing). These electrodes
were held in place by woven fiberglass straps having sewn-in pockets.
These cages are referred to as flexible cages, since the electrode/strap
assembly is quite flexible prior to installation on the bags.
Fiberglass bags were used for the entire reverse-air test series. The
bags were 20.3 cm (8 in.) in diameter and 224 cm (8 ft) long; no anti-
collapse rings were used. These bags were made of J. P. Stevens Style
No. 648, 17 oz (482 g) texturized fiberglass fabric with a 10 percent
Teflon B finish. Each bag was tensioned to 40 Ib (18 kg) using a conven-
tional tensioning spring.
RESULTS AND OBSERVATIONS
Electrostatic stimulation of fabric filtration has two primary e.ffects
on filter behavior when compared to conventional filtration at the same G/C:
(1) the average pressure drop is reduced through a combination of lower
residual pressure drop and a reduced pressure drop rise during a filtering
cycle; and (2) the ESFF baghouse can operate at a higher stable G/C.
To define and compare the effects on pressure drop rise during a
filtering cycle, a figure of merit called PDR (pressure drop ratio) is
used. PDR compares the beginning and ending Ap across one cleaning cycle
using electrostatic enhancement with the comparable AP change without
electrostatic enhancement.
(AP - A P ) ESFF
PDR is defined as: r
(Apf - Ap^ ) control
whereAP is the pressure drop across the bags,
f refers to the final state (just prior to cleaning), and
r refers to the residual state (just after cleaning).
290
-------
For an idealized filter cycle, in which the bag pressure drop increases
linearly with time and for constant dust loading from AP to AP , it can
be shown that: r
(K ) ESFF
PDR = *•
(K ) control
where K_ is the specific cake resistance.
PDR, then, is the ratio of the flow resistance of a dust cake collected
by an ESFF system to that collected by a conventional fabric filter.
Figure 4 presents typical pressure drop recordings for both the con-
ventional and ESFF pilot baghouses in reverse-air operation. The rate of
increase of pressure drop in the ESFF baghouse when compared to the con-
ventional pilot baghouse is evident. PDR can be seen to vary depending on
the length of the filter cycle, and thus,.is most useful for comparison of
two baghouses operating with the same cleaning cycle. As the two baghouses
were operating at the same face velocity when these data were obtained, the
ESFF baghouse also has a reduced residual pressure drop. Thus, the ESFF
effect is to produce a lower average pressure drop when operated at the
same filtering velocity.
The reverse-air baghouse operation was started up with a break-in
period of about 1 week at a G/C of 2 ft/min, with no electrical field in
the ESFF baghouse. During this period, the two baghouses exhibited essen-
tially identical pressure drop characteristics, indicating that the two
baghouses could be analyzed as identical baghouses in parallel. The electri-
cal field was then put on the bags, and was left on almost continuously for
the next several months.
Figures 5 and 6 present a chronological summary of the results during
May and June 1981, using reverse-air bag cleaning. Field strength and
gas-to-cloth ratio were independent variables. The PDR and average pressure
drop were measured for each filtering cycle and averaged for the day. The
gas-to-cloth ratio was progressively increased from 2 to 6 ft/min for the
ESFF baghouse; however, the control house could not operate stably above a
G/C of 4.5 ft/min. Note also that on May 12 and 13, when the electric field
was removed, the average AP of the ESFF house rapidly approached that of
the control house. Note particularly, that from June 10 to June 30, a 21-
day period of continuous 24-hour operation, the ESFF house operated at a
G/C of 6 ft/min, with an average AP between 6.4 and 7.1 in. (16.3 and 18.0
cm) of water, while the control house operated at a G/C of only 3 ft/min
and yet had an average AP of 7.4 to 8.3 in. (18.8 to 21.1 cm) of water.
Stated another way, ESFF allowed a doubling of the velocity without a major
change in pressure drop.
Figure 6 demonstrates the ability of an ESFF baghouse to operate at a
higher stable G/C. Note that from June 9 to 13, the control baghouse demon-
strated inability to operate stably at a G/C of 4.5 ft/min; subsequently,
291
-------
it was reduced to a G/C of 3 ft/min. Simultaneously, the ESFF baghouse
operated at a G/C of 6 ft/min from June 10 to 30.
Based on these encouraging results, a series of tests were undertaken
to evaluate the effects of field strength and gas-to-cloth ratio on pressure
drop characteristics.
Figure 7 summarizes the evaluation of field strengths of 3 and 4 kV/cm
on average Ap over a G/C ratio range of 1 to 6 ft/min. This test series
used the rigid cages; hence, fields above 4 kV/cm were not possible. These
data show graphically the benefit derived from the ESFF effect. The con-
ventional baghouse had a higher averageAp where it was capable of stable
operation. Increasing the electric field from 3 to 4 kV/cm does not appear
to offer an advantage.
By the end of this test series, the disadvantages mentioned previously
for the rigid cage had become evident, and it was necessary to install new
bags in the ESFF pilot baghouse. Figure 8 is a chronological summary of
testing with new bags and the new flexible cages at field strengths up to
6 kV/cm. After a break-in period at a G/C of 2 ft/min on July 26 through
28, the G/C was increased to 4 ft/min and an electric field of 4 kV/cm
resulted in an immediate decrease in the average Ap compared to the control
house. Figure 9 summarizes the results of this initial approximately 1
month test period with the flexible cages. These results indicate two
things. First, little or no benefit was derived from field strengths above
3 kV/cm. Second, note the results at 0 kV/cm. As the ESFF bags were rela-
tively new while the conventional bags had been in use since the beginning
of the test program, some difference in theAp response of the two bag sets
at zero field was expected. However, the magnitude of the difference was
surprising, and suggested that something other than the electric field
might be causing at least a portion of theAp advantage being shown by the
ESFF baghouse. An extensive investigation of the situation was undertaken,
and several factors which served to exaggerate the advantage of the ESFF
baghouse were identified. A concise explanation is still in the process of
being defined; however, three contributing features stand out so far:
(1) As is well known, reverse-air baghouses are extremely sensitive
to cleaning parameters. Although efforts had been made to favor
the control house by cleaning it last, this, in fact, may have
favored the ESFF house, possibly by giving it a pre-coat after
cleaning. In addition, a slight leak below the tubesheet in the
control baghouse, balanced by a similar sized leak above the tube-
sheet in the ESFF baghouse, raised the effective gas-to-cloth
ratio in the conventional baghouse. These factors were quantified
by experimentation, and their mathematical inclusion in the data
analysis indicated that the ESFF baghouse at zero field still had
a lower average AP than did the conventional baghouse. The same
corrections reduce the size of the advantage which the data had
indicated for the ESFF baghouse at the various field strengths.
(2) In May, as shown in Figure 5, the field was removed after 2 weeks
292
-------
of operation,'then the APS at zero field strength were essen-
tially equal. In September, after 2 months of operation, the
ESFF bags are better even with no field. This certainly sug-
gests the possibility of a long term, cumulative effect of the
electric field by reducing the residual A? by avoiding the
buildup of un-removable dust accumulations deep in the fabric
weave.
(3) During the investigation into the zero field behavior, the ESFF
bags were operated for approximately 2 months without the electric
field. It was noted that, after this zero field operation, the
low residual Ap results which were obtained during the early
operation were no longer observed. The hypothesis is that the
long term lack of an electric field allows the fabric to build an
interstitial dust layer which cannot be removed by normal means,
leading to the increased residualAPS. A return to new fabric
will try to verify this observation.
CONCLUSIONS
Although the evaluation of electrostatic augmentation of reverse-air-
cleaned bags is not as yet complete, the observations to date of primary
significance are:
(1) Starting with new fabrics, ESFF operation for an extended period
(21 days) at a G/C of twice (6 vs. 3 ft/min) that of the control
house was demonstrated at comparable or lower pressure drops
(Figure 6).
(2) Interruption of the electric field for periods of 1 to 2 days
did not eliminate most of the benefits of ESFF. Removal of the
field for several months did diminish the benefits of ESFF upon
reapplication of the field.
(3) ESFF appears to maintain the fabric in a "near new" condition
relative to entrapped particles within the fabric. It is, of
course, these entrapped particles that result in the residual
Ap after cleaning. This observation is consistent with earlier
work, indicating that under ESFF conditions more particulate is
captured near the upstream side of the fabric, and penetration
into the fabric is reduced. Thus ESFF prevents, or at least
significantly retards, the buildup in residual
FUTURE WORK
Both the rigid and the flexible cages are recognized to have practical
and economic limitations for full-scale filter use; hence, electrode design
293
-------
is continuing. The most promising concept at this time for reverse-air-
cleaned applications appears to be a "woven" electrode, although this has
not as yet been tested in the pilot system. Woven electrodes consist of
conductive yars woven into the warp of the fabric at the desired electrode
spacing. Yarns consisting of stainless steel filaments, graphite filaments,
or a graphitized high temperature nylon are now being investigated. These
conductive yarns, and the techniques for weaving them, have been previously
used to produce anti-static fabrics for both filtration applications and
for carpets. Woven electrodes in both glass and Teflon fabrics are being
investigated.
ACKNOWLEDGEMENTS
This work was supported by EPA Contract 68-02-3186 from Industrial
Environmental Research Laboratory, Research Triangle Park, NC. The assis-
tance provided by E. I. duPont de Nemours and Company, Inc. is gratefully
acknowledged. JChe pilot unit was located at their Waynesboro, Virginia,
plant. Teflon^fabric was donated to the test program by DuPont.
294
-------
REFERENCES
1. Lamb, G. E. R., and P. A. Costanza. A Low-Energy Electrified Filter
System, Filtration and Separation 17:319, 1980.
2. Van Osdell, D. W., G. P. Greiner, G. Lamb, and L. S. Hovis, "Electro-
static Augmentation of Fabric Filtration", In proceedings of the Third
Symposium on the Transfer and Utilization of Particulate Control Techno-
logy, Volume I. Control of Emissions from Coal-Fired Boilers,
EPA-600/9-82-005a, July 1982.
3. Greiner, G. P., D. A. Furlong, D. W. Van Osdell, and L. S. Hovis.
"Electrostatic Stimulation of Fabric Filtration", JAPCA, 31:1125-1130,
1981 (presented at the 74th Annual Meeting of the APCA, Philadelphia,
Pennsylvania, June 1981).
295
-------
TO POWER SUPPLY
Hh
1.5 cm
HIGH VOLTAGE HARNESS
SEPARATE, INSULATED
BAG SUPPORT "CAGE"
:IBERGLASS YARN
-ELECTRODES
STAINLESS STEEL
WIRE-0.58 mm
HARNESS INSTALLED ON BAG
Figure 1
TRI HARNESS
Conventional Baghoute
Figure 2
PILOT PLANT
296
-------
OUTLET (TOP)
TRAILER
( .. 0 J
v x „ s KP J-
V ~ , r *
TEST PORTS
Figure 3
SYSTEM SCHEMATIC
1:00 12:00 1:00 10:00
9:00 8:00 p.m.
Figure 4
REVERSE AIR PILOT BAGHOUSE PRESSURE DROP TRACE
297
-------
cc
Q
Q.
o>
2
1.0
0.8
0.6
0.4
0.2
0
8
ll 6
itS 4
'*Z
o o
.2
ui
0
8--
raX
0°
it
Conventional
\
11 16
Date: May 1981
21
26
j
31
Figure 5
ESFF CHRONOLOGICAL SUMMARY, MAY 1981
298
-------
cc
Q
Q.
OJ
1.0
0.8
0.6
0.4
0.2
0
8
6
: 4
2
0
12
£U>
_ CM p
|d 8
II6
2
0
8
o
to
UJ
ESFF
\
Conventional
11 16 21
Date: June 1981
26
_J
31
Figure 6
ESFF CHRONOLOGICAL SUMMARY, JUNE 1981
299
-------
10
8
a.
in
Csj
d
~ 6
c
d 5
2
o
£
1 4
a.
0)
4 3
ESFF
3 kV/cm
I
234
Gas-to-Cloth Ratio, ft/min (= 0.5 cm/s)
Figure 7
DELTA P AVG. VS G/C RATIO
RIGID CAGE
APRIL 24 TO JULY 15
300
-------
1.0
0.8
cO.6
Q
a.
0.4
0.2
I 8
£
«S 6
•8°
4
2
0
I
Hi
\
!l8
£»
I2
*a«
s°*
.s
<'~
0
,2< 8
SE
ceo
x:10. 6
£4
II 2
25
Conventional
v
I
July 1981
31
10 15
August 1981
Date
20
25
30
Figure 8
ESFF CHRONOLOGICAL SUMMARY, JULY 25 - AUGUST 31, 1981
301
-------
8
£
O
CM
d 4
o
£
Q.
O
O)
I
Conventional
Baghouse
ESFF Baghouse
Strength
0 kV/cm
6 kV/cm
2 and 4 kV/cm
3 kV/cm
JL
_L
234
Gas-to-Cloth Ratio, ft/min (= 0.5 cm/s)
Figure 9
DELTA P AVG. VS G/C RATIO
FLEXIBLE CAGE
JULY 27 TO SEPT. 21
302
-------
ELECTRICAL STIMULATION OF FABRIC FILTRATION;
ENHANCEMENT BY PARTICLE PRECHARGING
by: George E. R. Lamb, Richard I. Jones and William B. Lee
Textile Research Institute
Princeton, New Jersey 08540
ABSTRACT
The reductions in pressure drop that accompany the establishment of a
strong electric field near a filter fabric appear to be due to three
separate mechanisms. One is the formation of a more porous dust cake due
to dust capture in the low packing density regions of the fabric. A
second mechanism is attraction of particles to the bag wall which causes
the bag to act like a precipitator. The thickness of dust cake is then
greater near the entrance than at the end of the bag, and this results in
a lower pressure drop. The third mechanism involves attraction of parti-
cles to the bag electrodes. The dust is then deposited in bands with
relatively thin deposits in between. Measurements and visual inspections
of the dust deposits indicate that the second and third effects are en-
hanced when the aerosol is charged. A particle charger of new design
appears to be particularly suitable for this purpose, and is found to
cause major changes in filtration performance.
This paper has been reviewed in accordance with
the U.S. Environmental Protection Agency's peer
and administrative review policies and approved
for presentation and publication.
303
-------
INTRODUCTION
A number of investigations have been directed to the development of sys-
tems for electrically stimulated fabric filtration (ESFF). Some have taken
the approach of establishing an electric field near the filter (1-4). Others
have sought to improve performance by adding charges to the aerosol particles
(5-7). In a few. cases, both effects have been combined (8,9).
When electric fields are applied in the vicinity of the filter, perfor-
mance is improved by virtue of the increased capture efficiency of single
fibers in the presence of a field (10). This causes the dust cake to form
closer to the surface of the fabric. The cake is then more easily removed
when the filter is cleaned, and less dust is retained (10). If, in addition,
the upstream surface of the filter has a nap, the dust cake will form in the
nap and be more porous. (The nap is the fuzzy region on the surface of some
fabrics where the fiber volume fraction is smaller than in the interior.)
Both these effects will allow the filter to run at lower pressure drop than
without the electric field. Penetration is usually reduced by ESFF. This,
intuitively, is also an expected result of the better single fiber efficiency,
but an indirect mechanism may involve reduced seepage because of the upstream
shift of the dust deposits.
The results in Table 1 illustrate these effects. They were obtained
with a series of bags made of Teflon felts with increasing degrees of nap.
It can be seen that,as the nap becomes more pronounced, not only overall pres-
sure drop levels are reduced, but also pressure drop ratios (PDR). If APf
is the pressure drop just before a filter is cleaned, and AP. the pressure
drop just after, then the PDR is defined as the ratio of (AP--AP.) when the
electric field is applied to the corresponding value with no field. The
table also shows that the field reduces penetration.
TABLE 1. EFFECT OF SURFACE LOW-DENSITY LAYER ON PRESSURE DROP REDUCTION
Av.
field
(kV/cm)
0
2
4
. .,
Control
AP.,APf
(mm H20)
7,23
6,16
7,15
PDR
0.63
0.50
Eff
98.
98.
98.
•
03
80
88
fi*
(mm
7,
7,
5,
Low nap
H26)
22
15
12
Eff.
PDR (%)
98.04
0.53 98.93
0.47 99.13
APi>
(mm
5,
3,
3,
High nap
H26)
17
9
4
Eff
PDR (%)
96.
0.50 98.
0.08 98.
•
83
51
76
Charging the aerosol upstream of the bag has also been found to result
in lower pressure drops and lower penetrations, but few workers have taken
account of the loss of dust which occurs either in the precharger or in the
duct from precharger to baghouse. In such cases, it is difficult to assess
the effects of charging on bag performance. One study (7), in which the dust
collected by the bag was weighed, showed no difference in specific cake resis-
tance with precharging. The lower pressure drop was due to reduced dust
loading.
304
-------
EXPERIMENTAL
In order to study the effects of precharging without interference from
reduced dust loading, the charger design shown in Figure 1 was adopted.
AE
r
i
(-
h
1-
i
hi
ioSOL
I
T
-s1
PJ
Yr-
— »•-
ESFF
POWER
SUPPLY
PRECHARC
POWEF
SUPPL
iN
Figure 1. Location of "lightning rod" precharger in reverse air baghouse.
The charger, a metal rod provided with wire bristles at the upper end,was
positioned by means of an insulated support, so that its tip was just below
the entrance to the bag in a one-bag laboratory baghouse. With this ar-
rangement it was expected that all dust would enter the bag because of the
high air velocity at the bag entrance. The bags used were provided with
electrodes on the upstream surfaces.
®
Table 2 lists results obtained with a Teflon felt bag provided with
so-called "printed" electrodes, which in this case consisted of narrow
bands of a conducting elastomer composition. The use of "printed" elec-
trodes is a technique undergoing trial. If successful, it should permit a
length of fabric to be fitted with an electrode pattern more easily and
cheaply than would be the case with wire electrodes. The results in Table
2 may be taken as an indication that such electrodes can provide electrical
stimulation as effectively as wires. The question to be examined in the
future is one of durability.
The results in Table 2 show that,when aerosol precharging is combined
with an electrified bag, reductions in PDR of almost an order of magnitude
can be obtained. It is clear that, when bag electrodes and precharger are
raised to potentials of opposite signs, the pressure drop reductions are
greater than when the signs are the same. It can also be seen that; with
this bag, activating the precharger and not the bag electrodes causes only a
minor reduction in pressure drop.
305
-------
TABLE 2. PRESSURE DROP RATIOS WITH ESFF AND PRECHARGED AEROSOL
Teflon®
Bag field
(kV/cm)
0
0
+2
+2
-2
-2
-3
-3
+2
-2
>
bag with printed electrodes. Inlet concentration 4.8
Face velocity 6 ft/min (3 cm/s).
Precharger
voltage (kV)
0
-9
0
-9
0
-9
0
-9
+9
+9
AP±, AP
16,110
16,107
15,83
14,28
15,60
15,41
15,52
15,35
16,39
17,31
g/m3.
PDR
0.97
0.72
0.15
0.48
0.28
0.39
0.21
0.24
0.15
Figure 2 shows results obtained with woven glass bags having sewn-in
150 ym bare copper wire electrodes. It can be seen that, as before, without
a potential on the precharger, the reductions in the pressure drop are mod-
est. With the precharger activated, PDR values fall considerably lower.
Precharger
Voltage
a:
o
a.
1.0
0.8
0.6
0.4
O.E
2.3 4
AVERAGE FIELD (kV/cm)
3 cm/s
1.5 cm/s
Figure 2. Dependence of PDR on bag field after extended run.
no precharger.
Top curves
306
-------
In order to study the mechanism by which precharging reduces pressure
drop, the bag support was modified to allow the bag to be removed easily
from the baghouse. The sewn seam of the bag was replaced with an overlap
held closed by clamps. With this arrangement, the bag(after being removed
from the baghouse) could be carefully opened and the dust cake could be
examined. Visual inspection immediately showed that nonuniforin distribu-
tion of the dust deposit occurred both in the longitudinal and in the tang-
ential direction. Deposition was heavier at the lower end of the bag and
on the electrodes, as shown in Figure 3.
Figure 3. Dust deposition pattern with 4 kV/cm field and -15 kV precharger
potential.
Mass distribution with respect to bag height was measured by placing a
mask at a series of positions over the opened bag and aspirating the dust
from the area within the mask. The dust was collected on a filter and
weighed. This was done for various combinations of bag fields and pre-
307
-------
changer potentials. The results are plotted in Figure 4: in all cases
more dust collects at the bottom than at the top of the bag. When this
occurs with no voltage applied, the cause is assumed to be settling by
gravity; but with an applied field, the tendency for early deposition is
increased. Values of PDR are listed alongside the curves. It should be
noted that these were values obtained from the pressure drop reached
after 15 minutes, starting from an almost clean bag. The bag was
cleaned by vacuuming each time so as to allow measurement of the dust
deposit areal densities plotted in Figure 4. The PDRs in Figure 4 must
therefore not be regarded as typical of values that would be obtained after
prolonged conditioning. They nevertheless bear a certain relationship
to the curves. It can be seen that curves with greater slopes tend to cor-
<8
16
(O
Q
fe~: «°
El
in g
5i 8
20 40 60 80 - 100 120
DISTANCE FROM BOTTOM OF BAG (cm)
Figure 4. Distribution of dust mass with respect to distance from the
bottom of the bag. B and P indicate average field on bag in
kV/cm and voltage on precharger in kV, respectively.
respond to lower PDR. Deviations from this trend must be attributed to
nonuniformity in distribution in a tangential direction, such as shown in
Figure 3, which is a photograph of the opened bag after being run 15 minutes
with an average +4 kV/cm field between wires and a -15 kV potential on the
precharger (44,-15). It can be seen that: (1) Most of the dust collects on
the positive wires. (2)' The amount collected, as plotted in Figure 4, Is
greatest at the bottom and diminishes with distance up the bag. (3) Some
308
-------
collection occurs on the ground wires. This may be dust that, having depos-
ited on the positive wires, is reentrained after gaining a positive charge.
(4) Spaces between wires are left essentially clean.
Photographs taken of the bag surface and dust deposits are shown in
Figures 5, 6, and 7. In all these photographs, pieces of adhesive tape had
been stuck to the clean bag over portions of some of the electrodes. The
patterns of dust deposition on the tape lead to some interesting conclusions
about the motion of dust particles during filtration, as discussed below.
Figure 5 shows the bag surface after vacuuming (a) and with a dust cake laid
at no applied voltage (b). No dust has collected on the pieces of adhesive
tape because there is no air flow normal to the tape. Figure 6 shows dust
deposited with a 4 kV/cm field between bag electrodes and +15 kV (a) and -15
kV (b) on the precharger. In both cases the dust forms visible bands on the
wires, but the bands are sharper when bag voltage and precharger voltage are
of opposite sign. An interesting feature of this figure is that dust was
deposited on the adhesive tape in an amount apparently equal to that else-
where on the band. This indicates that particle velocities due to the field
near the wires are much greater than the gas velocity. The bag is thus acting
.like a precipitator. Figure 7 shows the appearance of the dust deposits
when only a bag potential was applied (a). Here there are still visible
bands, but they are more diffuse and dust is again deposited on both the
electrodes and the adhesive tapes. When only the precharger was activated
(b and c) , the dust deposit was uniform, and there was some deposition on
the tapes, presumably of dust driven to the wall by space charge effects.
After each of the 15-minute runs specified in Figure 4, the bag was
weighed and the mass of dust determined. Table 3 shows that no matter what
the combination of bag and charger potential, the mass collected in 15 mi-
nutes was always the same. Thus the changes in pressure drop were not to
any extent due to reduced total dust loading. This appears to disagree with
.the data in Figure 4. In that figure the total mass of dust should be pro-
portional to the area under each curve, and the lower curves appear to enclose
smaller areas. The explanation must be that the sharp upswings to the left
of the lower curves were plotted with too few points to show accurately the
amount in those regions. Figure 3 shows that heavy deposition occurs with-
in a few centimeters of the bottom of the bag; the telephone (12 cm high)
to the right of the photograph gives an idea of scale.
TABLE 3. MASS OF DUST COLLECTED ON BAG IN 15 MINUTES
Bag field Precharger Mass of dust
(kV/cm) voltage collected (g)
in two trials
0 0 30, 31
+4 0 33, 31
+4 -15 27, 29
+4 +15 37, -
0 -15 35, 27
0 +15 32, 31
309
-------
!
CLEAN I •
Figure 5. Appearance of the inner surface of the bag:
a) after vacuum cleaning through a wire screen; and
b) with an average of 7.5 mg/cm2 of fly ash
deposited with no electric field or precharging.
No fly ash has deposited on the tape in either case.
In Figures 5, 6, and 7, horizontal bands are due to
anticollapse rings inside the bag.
310
-------
v.. .;'.& fc ''
^•'.•TO* ' Vv'-.-'vi ' ''-V '
--*• ''
Figure 6. Appearance of the inner surface of the bag:
a) after deposition with 4 kV/cm between bag
electrodes and +15 kV on the precharger; and
b) after deposition with 4 kV/cm between bag
electrodes and -15 kV on the precharger.
Points to note: (1) Dust collects on tape,
(2) Bands are narrower with negative precharging.
311
-------
*:.;•.', y,
vv-'.-.
Figure 7. Appearance of inner surface of bag:
a) after deposition with 4 kV/cm between bag electrodes and no pre-
charging. Dust deposits in bands on both sets of electrodes and on
tapes.
b) with positive precharging only. Faint bands are visible on grounded
electrodes, including area covered by tape.
c) with negative precharging only. No bands are visible.
312
-------
Any technique capable of reducing pressure drop offers several
possible benefits: one can choose to operate at a lower energy consumption
level, allow longer intervals between cleanings, or, by raising face velo-
cities, use fewer bags. A recent study (11) has shown that the third of
these options gives the largest savings in operating costs, and it may be
assumed that in any future commercial development of ESFF, face velocities
will be higher than those in present practice. However, limits to face
velocity are also imposed by falling efficiency. The possible increases
in face velocities attainable in the combined bag electrode/precharger
mode were explored by making measurements of efficiency at a series of
velocities. Figure 8 shows efficiencies obtained with no field 0,0),
with +4 kV/cm average field between electrodes (+4,0), and with ,4 kV/cm
and -15 kV on the precharger (+4,-15). There is a clear enhancement of
efficiency at all velocities. The bag used was the same one in Figures 3
to 7, made of textured woven glass. Long-term tests in a pilot plant will
be needed to determine the maximum velocity at which stable operation is
possible. It is reasonable to assume, however, that with a precharger/
electrified bag combination, considerable increases in face velocity will
be possible.
B. P
o 99
z
UJ
u.
LL.
UJ
98
0,0
3 6
FACE VELOCITY (cm/s)
Figure 8. Dependence of efficiency on face velocity for woven glass bag
with and without ESFF. B and P indicate the same quantities as
in Figure 4.
313
-------
CONCLUSIONS
Use of a precharger of new design in conjunction with electric fields
generated by electrodes in the bag walls has led to reduced pressure drop
and lower dust penetration. The reduction in pressure drop is due at least
partly to lessened uniformity of the dust cake, which is heavier at the
bottom of the bag and on the electrodes. The improved performance was not
due to reduced dust loading, since no changes in total mass collected on the
bag were observed. With this type of precharger/electrif ied bag combination,
it should be possible to operate a baghouse at face velocities several times
higher than those of current practice.
314
-------
REFERENCES
1. linoya, K. and Makino, K. Application of electric field effects to
dust collection filters. Aerosol Science 5: 357, 1974.
2. Rivers, R. D. Non-ionizing electrostatic air filters. ASHRAE
Journal 37: February 1962.
3. Frederick, E. R. Some effects of electrostatic charges in fabric
filtration. In Proceedings: Symposium on Use of Fabric Filters
for the Control of Submicron Particulates, EPA-650/2-74-043
(NTIS PB 237629), May 1974.
4. Bergman, W., Hebard, H., Taylor, R., and Lum, B. Electrostatic filters
generated by electric fields. Paper presented at the Second World
Filtration Congress, London, England, September 1979.
5. Lundgren, D.A. and Whitby, K.T. Effect of particle electrostatic
charge on filtration by fibrous filters. I & E C Process Design and
Development 4:345, 1965.
6. Helfritch, D.J. Performance of an electrostatically aided fabric
filter. CEP 54: August 1977.
7. Hovis, L.S., Abbott, J.H., Donovan, R.P., and Pareja, C.A.
Electrically charged flyash experiments in a laboratory shaker bag-
house. In Third Symposium on the Transfer and Utilization of Particu-
late Control Technology, Vol. I, EPA-600/9-82-005a (NTIS PB83-149583),
July 1982.
8. Penney, G.W. Electrostatic effects on fabric filtration; Vol. I,
EPA-600/7-78-142a (NTIS PB288576), September 1978.
9. Lamb, G.E.R. and Costanza, P.A. Electrical stimulation of fabric
filtration - Effects of electrode current. Textile Res. J. 51:389,
1981.
10. Lamb, G.E.R., Costanza, P.A., and Turner, J. H. Role of filter
structure and electrostatics in dust-cake formation. Textile Res. J.
50: 661, 1980.
11. Van Osdell, D. W., Ranade, M. B., Greiner, G. P. and Furlong, D. F.
Electrostatic Augmentation of Fabric Filtration: Pulse-Jet Pilot
Unit Experience, EPA-600/7-82-062 (NTIS PB83-168625), November 1982.
315
-------
ESFF AS A FIELD EFFECT
By: L. S. Hovis and G. H. Ramsey
Industrial Environmental Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
and
R. P. Donovan
Research Triangle Institute
P. 0. Box 12194
Research Triangle Park, North Carolina 27709
ABSTRACT
Evidence to suggest that the mechanism on which ESFF (electrostatic
stimulation of fabric filtration) depends is an electric-field-only mechanism
(as 'opposed to a Coulomb mechanism that depends on both electric field and
electric charge) includes:
1. A room-temperature high-humidity factorial experiment in which
both external electric field and fly-ash electrical charge were
independent variables.
2. Selected experiments carried out at low relative humidity and room
temperature.
3. Enhanced filtration measurements made with a 60-Hz ac electric
field applied to the bag electrodes.
4. Published precharging data of others in which enhancement vanishes
at high relative humidity.
All of the data collected in EPA/IERL's Research Triangle Park laboratory
(items 1-3, above) refer to experiments in which pulverized coal fly ash was
the dust source. Dusts from other sources were used in the experiments of
item 4.
This paper has been reviewed in accordance with the U. S. Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication.
316
-------
ESFF (electrostatic stimulation of fabric filtration), an acronym coined
by reseachers at Textile Research Institute(l), refers to a technique for im-
proving fabric filtration performance by applying an external, nonionizing
electric field parallel to the fabric surface during the filtration process.
The improved performance manifests itself as a reduced drag and reduced
penetration (although not all researchers report a reduced penetration--
Reference 2, for example).
Typical explanations of the observed enhancement invoke Coulomb capture
forces between the electrically charged dust particles and the externally
applied surface field, expecially on the low-fiber-density upstream surface
of the fabric.(1) Particles collected on these low-fiber-density surfaces do
not contribute as much to fabric drag as those collected in the more tightly
packed interstices of the bulk fabric. Qualitatively, this model of ESFF
portrays the action of the electric field as enhanced upstream capture of
electrically charged dust at low-fiber-density 'surface sites.
In such a model, particle charge is expected to be a variable of first
order significance since the Coulomb force experienced by a particle approach-
ing the fabric depends on both particle charge and the electric field in the
region through which it passes. Other electrical forces exist, however, such
as polarization forces; and other non-Coulomb-force-field-dependent mechanisms
could be significant. This paper reviews data that support ESFF effects based
on mechanisms other than Coulomb capture.
HIGH HUMIDITY FACTORIAL
This experimental series used both fly-ash electrical charge and an ex-
ternal electric field at the fabric surface as independent variables in a
factorial experiment measuring the performance of a pulse-jet baghouse. De-
pendent variables of the experiment included pressure drop across the bags
(AP), outlet fly-ash concentration (C ), and an effective cake resistance
(K2'). The last quantity was calculated from increase in AP during a filtra-
tion cycle and the corresponding increase in fly-ash mass on the bags, de-
termined by carefully demounting and weighing.
In the context of this factorial experiment, high relative humidity
means 52 to 59 percent. The specific value itself is not important except
as an indication that the electrical resistivity of the fly ash/fabric filtra-
tion media would be low enough to dissipate the electrical charge deposited
along with the collected fly ash without significant charge accumulation
in/on the dust cake. Since the maximum current carried to the bags by the
fly ash under any of the test conditions was 150 to 200 nA, having the bag
electrode current in the 150- to 200-jjA range, three orders of magnitude
higher, ensures that the field at the bag surface depends primarily on the
voltage applied to the electrodes. Thus, keeping the relative humidity
greater than 50 percent kept the electical resistivity of this particular
test fly ash/fabric combination low enough to avoid the complications
brought about by an electric field component attributable to deposited
charge.
317
-------
The details of the experimental procedure and apparatus appear else-
where. (3) The key conclusions are summarized in Figures 1 through 4, the
first two of which plot normalized K2' and &P as a function of precharger
current (Figure 1) and field voltage (Figure 2). Figure 1 shows that K2'
does not change with increasing fly-ash electrical charge, as measured by the
corona precharger current. (The average charge/mass of the fly ash varied
over the range ±0.2 [tC/g to -2.0 pC/g as the power supply current varied over
the range plotted.) The dashed line in Figure 1 plots the normalized AP_
values corresponding to these precharger power settings. APn is the final
pressure drop (AP^,) minus the residual pressure drop (APR) in one filtration
cycle. The reduction in APn with increasing precharger current is the electri-
cal enhancement effect attributed to precharging.(4-6) Since K2* (which depends
on the ratio APn/M_, where fL is the increase in fly-ash mass on the bag that
caused APR to increase to AP^) does not change, Figure 1 implies that the APD
decrease has been compensated by a similar decrease in M^. Figure 3, which
compares the normalized APn curve with the corresponding normalized M^ curve,
shows that that is exactly what has happened and that the electrical enhance-
ment described by the reduced values of AP are achieved by reducing the M_^
on the bags. The precharger acts as a collecting prefilter, a conclusion
reported at the previous symposium for both pulse-jet(7) and shaker baghouse(8)
experiments.
The field voltage dependencies (Figures 2 and 4) tell a different story.
Figure 2 shows that the normalized K2' and the normalized AP_ decrease together
as a function of external field voltage. Figure 4 confirms that M._ is inde-
pendent of field voltage meaning that the mass of fly ash reaching the bags
is independent of bag electrode voltage. These plots show, therefore, that
the permeability of the deposit has increased, resulting in a decrease in
cake airflow resistance, because of the field.
The mechanism by which this increase in permeability takes place cannot
be further specified other than to say that Coulomb forces do not dominate
the mechanism. This conclusion follows from the observation that charging
the fly ash to a high charge-to-mass ratio (Q/M) does not alter the magnitude
of the permeability enhancement; that is, the reduction in K2* brought about
by the external field is the same for the lowest Q/M fly ash as for the highest.
A mechanism that depended on Coulomb forces primarily should produce increasing
response with increasing Q/M at constant field. That the fly-ash deposit does
not exhibit this property leads to the conclusion that ESFF, in this experimental
setup at least, depends upon electrical interactions other than Coulomb attraction.
While the capability of the precharger to add charge to the fly ash
that reaches the bags is indisputable (the bag electrodes can be used to
measure the increased current contributed by the precharger), the distribution
of this charge by particle size is not known. Conceivably, all the charge
could be added to a size fraction of the fly ash that proves insignificant in
ESFF. Two additional experiments are planned to further clarify the conclusions:
1. Charge spectrometry by the Millikan apparatus.
2. Charge neutralization by operating the precharger in the ac mode.
313
-------
The first supplementing experiment will measure the distribution of
charge among the various sizes of fly-ash particles; the second will attempt
to neutralize the natural charge now present on the fly ash so as to achieve
a data point of even lower Q/M than when the precharger is OFF.
LOW HUMIDITY OBSERVATIONS
When the high humidity factorial experiment is not carried out at "high
humidity," by definition significant electrical charge can accumulate on the
bags because the charges carried by the fly ash do not leak off as readily
as when operating at high humidity. The cage voltage(9) is one measure of
accumulating bag charge. At operating relative humidity of 35 percent or less,
the cage voltage (cage voltage is the voltage on the electrically isolated
support cage of a pulse-jet bag, as read by a high impedance voltmeter such
as an electrometer) generally builds up to values exceeding ±1000 V dc, sug-
gesting that the net electric field at the fabric surface now has a component
attributable to accumulating charge in addition to the applied field that
dominated the high humidity case.
Furthermore, because these charges will drift in the externally applied
field, they distribute themselves so as to neutralize the external field over
much of the fabric surface similar to the action described by Walkenhorst in
his metallic fiber filter.(10) Figure 5 sketches an extension of the Walken-
horst model to a fabric filter as envisioned here.
Laboratory observations confirm that the ESFF effect diminishes at low
relative humidity and that charge accumulation increases. However, because
experiments correlating cage voltage with K2' and other performance measures
are incomplete, this evidence is only speculative at present. Should the
interaction depicted in Figure 5 prove valid, then a low frequency field
reversal as discussed by Walkenhorst(lO) should restore the ESFF interaction.
AC ESFF
A third general experimental approach is to examine ESFF as a function
of field frequency. The reason such observations are significant is that
Coulomb forces depend on field direction, while other field dependent mecha-
nisms such as dielectrophoresis do not. Dielectrophoresis, the net attractive
force experienced by a polarized neutral body in an electric field gradient,
depends on the direction of the field gradient alone. The net force is always
in the direction of increasing field strength; thus, reversing the polarity
of the field does not change the direction of the net polarization force in
the typical ESFF experimental configuration. The net Coulomb force under ac
operation, however, is zero for time periods long compared to 1/f, where f is
the ac frequency.
To date, the only frequency investigated has been 60 Hz. At 60 Hz, the
ESFF mechanism appears intact and operative, although limitations in available
power supplies have restricted the available field strength. A fly-ash fabric
combination that yielded a pressure drop ratio (PDR) of 0.6 with 8000 V dc
across electrodes 1.8 cm apart produces a PDR of 0.7 with 5000 V rms (60 Hz).
The PDR is the ratio of AP with an electric field to AP without a field.
Face velocity during both measurements was approximately 5 cm/s.
319
-------
While the observation of an ac ESFF effect suggests that polarization
is a significant mechanism, Coulomb forces cannot be ruled out until much
higher frequency data are available.
FIBROUS FILTER ELECTRICAL ENHANCEMENT
Other researchers have reported electrically enhanced fibrous filtra-
tion through the use of prechargers(ll), electric fields across the media
(12), or both (13). Calculations of the single-fiber collection efficiency
of an electrically charged particle flowing past a charged fiber(l4) predict
that the addition of this Coulomb force to the classical capture mechanisms
influences single-fiber efficiency primarily at Stokes numbers below about 2
(Figure 6). For typical filter parameters, this conclusion implies that
Coulomb forces are important only for particles less than about 5 pm diameter.
These Coulomb-force-based predictions agree reasonably well with pub-
lished data, using either a precharger (Figure 7) or an external field across
the filter (Figure 8). In these two plots, overall filter efficiency is the
ordinate and particle diameter, the abscissa.
Figures 7 and 8 represent an electrical enhancement of fibrous filtra-
tion attributable to a Coulomb force. What will be argued next is that these
enhancements are not ESFF-type enhancements as described in References 1, 2,
and 3 and, in fact, exhibit behavior different from the reported ESFF en-
hancement. Furthermore, the behavior exhibited by the electrically enhanced
fibrous filters is consistent with Coulomb force enhancement, while that ex-
hibited in the ESFF experiments is not.
No significance is attached here to the fact that the primary enhance-
ment criterion differs between the two types of filters. For the fibrous
filter, collection efficiency is the primary measure of performance; for fab-
ric filters, it is pressure drop, drag, K2', or some normalized ratio describ-
ing filter energy consumption. What is crucial is that the enhancement of
each depends differently on certain independent variables.
Take the data in Figure 8 first. Bergman et al.(12) report that the
enhancement depicted in Figure 8 largely disappears if the dust source is
passed through a charge neutralizer prior to the filter. This behavior is
consistent with a Coulomb force which depends on both particle charge and the
electric field at the collector. With fabric filters, as described in the
high humidity factorial experiment, increasing the particle charge produced
no discernible effect on K2'. While increasing the particle charge is not
identical to the Bergman et al.(12) experiment, a Coulomb-force-dominated
variable should respond to any change in particle charge. That it did not do
so in the fabric filter experiment led to the conclusion that Coulomb forces
do not dominate the ESFF enhancement observed in the high humidity factorial
experiment.
The charge neutralization experiment planned as continuing work is
similar to the experiment reported by Bergman, who concluded that "natural"
particle charge was essential for the electrical enhancement he observed
and as a Coulomb mechanism would require.
320
-------
The precharger-induced enhancement represented in Figure 7 is reported
by Yu and Teague(ll) to disappear when carried out at high humidity or with
conductive dusts. This observation agrees with a model picturing charge
accumulation on the filter as the source of the electric field at the filter
surface that is required for the Coulomb capture force to be present. When
these charges are allowed to leak off, the field disappears, and the Coulomb
force vanishes as does the enhancement. These observations are consistent
with the hypothesis that this fibrous filtration enhancement is Coulomb-force
dominated.
The ESFF data presented earlier exhibit just the opposite behavior. It
is at high humidity—no surface charge accumulation—that the ESFF enhancement
is most pronounced. Charge accumulation seems to be associated with diminished
ESFF, although this conclusion is very tentative at present.
What the preceding paragraphs have attempted to do is to review the proper-
ties of a type of electrically enhanced filtration thought to be dominated by
Coulomb forces (i.e., fibrous filtration) and then show that certain dependen-
cies of these filters differ from those observed with ESFF and that, further-
more, those differences can be understood by attributing the electrical enhance-
ment of fibrous filtration to Coulomb forces and the electrical enhancement of
fabric filtration of fly ash to some unspecified but nonCoulomb mechanism.
CONCLUSIONS
The evidence cited supports the existence of a non-Coulomb mechanism of
ESFF. While the evidence is neither conclusive nor exclusive (in that some
experiments remain to be done before the conclusion becomes solidly established
and nothing presented rules out the existence of conditions under which Coulomb
forces could be significant in ESFF), the data make a non-Coulomb mechanism or
mechanisms the most likely explanation of the ESFF phenomenon observed in
the EPA/IERL-RTP in-house laboratories. The data do not distinguish between
an electrical polarization mechanism and other non-Coulomb mechanisms. They
do suggest that the magnitude of the electrical charge on the incoming fly ash
is not an important variable, a surprising conclusion in view of the recognized
dominance of Coulomb forces in most electrified systems.
REFERENCES
1. Lamb, G. E. R. and Costanza, P. A. A low-energy electrified filter
system. Filtration and Separation. July/Aug.:319-322, 1980.
2. Greiner, G. P., Furlong, D. A., VanOsdell, D. W., and Hovis, L. S.
Electrostatic stimulation of fabric filtration. J. Air Pollut. Control
Assoc. 31:1125-1130, 1981.
3. Donovan, R. P., Hovis, L. S., Ramsey, G. H., and Ensor, D. S. Electric-
field-enhanced fabric filtration of electrically charged flyash.
Aerosol Science and Technology. l_:385-399, 1982.
4. Ariman, T., and Helfritch, D. J. Pressure drop in electrostatic fabric
filtration. In: Second Symposium on the Transfer and Utilization of
Particulate Control Technology, Vol. III. EPA-600/9-80-039c (NTIS
No. PB-81-144800) (Sept.), 1980. pp. 222-236.
321
-------
5. Chudleigh, P. W. , and Bainbridge, N. W. Electrostatic effects in fabric
filters during build-up of the dust cake. Filtration and Separation.
17:309-3ir, 1980.
6. linoya, K., and Mori, Y. Experimental advances in fabric filtration
technology in Japan—effects of a corona precharger and relative humid-
ity on filter performance. In: Second Symposium on the Transfer and
Utilization of Particulate Control Technology, Vol. III. EPA-600/9-80-039c
(NTIS No. PB-81-144800) (Sept.), 1980. pp. 237-250.
7. Donovan, R. P., Hovis, L. S., Ramsey, G. H., and Abbott, J. H. Pulse-
jet filtration with electrically charged flyash. In: Third Symposium
on the Transfer and Utilization of Particulate Control Technology, Vol. I.
EPA-600/9-82-005a (NTIS No. PB-83-149583) (July), 1982. pp. 11-22.
8. Hovis, L. S., Abbott, J. H., Donovan, R. P., and Pareja, C. A. Elec-
trically charged flyash experiments in a laboratory shaker baghouse.
In: Third Symposium on the Transfer and Utilization of Particulate
Control Technology, Vol. I. EPA-600/9-82-005a (NTIS No. PB-83-149583)
(July), 1982. pp. 23-34.
9. Donovan, R. P., Ogan, R. L., and Turner, J. H. Electrostatic effects in
pulse-jet fabric filtration of room temperature flyash. In: M. P.
Freeman and J. A. Fitzpatrick (eds.), Theory, Practice and Process
Principles for Physical Separations. Engineering Foundation, New York,
New York, 1981. pp. 445-466.
10. Walkenhorst, W. Deliberations and studies on the filtration of dust-
laden gases with special allowance for electric forces. Staub-Reinholt
Luft. 29(12):1-13, 1969.
11. Yu, H. S;, and Teague, R. K. Performance of electrostatic fiberbed. In:
Extended Abstracts of the 1st Annual Conference of the American Associa-
tion for Aerosol Research, Santa Monica, California. February 17-19,
1982. pp. 19-3-19-4.
12. Bergman, W., Hebard, H. D., Taylor, R. D., and Lum, B. Y. Electrostatic
filters generated by electric fields. In: Proceedings of The Second
World Filtration Congress, 1979.
13. Ho, C. P., the Bahnson Co., Winston-Salem, N.C.
14. Loeffler, F. The influence of electrostatic forces and of the proba-
bility of adhesion for particle collection in fibrous filters. In:
Proceedings: Symposium on New Concepts for Fine Particle Control.
EPA-600/7-78-170 (NTIS No. PB-292-095) (August), 1978. pp. 206-236.
322
-------
1.0
0.5
o
_o
0 13.5 ft/min (6.9 cm/s)
• 9.8 ft/min (5.0 cm/s)
2 Precharger
EAR
DPrecharger
EAR
DO
20 50
Precharger Current
100
Figure 1. Normalized K.'2 and APD as a function of precharger current.
(Each I, value is summed over three values of field voltage.)
1.0
0.5
Field
EAR
EAR
o Field (no data
points)
DO
° 13.5 ft/min (6.9 cm/s)
• 9.8 ft/min (5.O cm/s)
Field Voltage (kV)
Figure 2. Normalized K£ and AF^, as a function of field voltage.
(Each D value is summed over three values of precharger current.)
323
-------
1.0
0.5
D Precharger
0 13.5 ft/min (6.9 cm/s)
• 9.8 ft/min (5.0 cm/s)
20 50
Precharger Current (/iA)
100
Figure 3. Reduction in dust mass (MD) deposited on bags
because of particle precharging.
1.0
0.5
D Field
0 13.5 ft/min (6.9 cm/s)
• 9.8 ft/min (5.0 cm/s)
LM,
DO
DAR
D Field
L'AR,
4 8
Field Voltage (kV)
(no data
"points)
Figure 4. Independence of dust mass and field voltage.
324
-------
Fabric
a) Starting field lines (efabrjc ~ 1>
Fabric
b) After charge accumulation
Figure 5. Predicted influence of charge accumulation
on the surface electric field.
1.6 -
R«-0.2
Pp- 2.164 B/cm3
R, - Reynolds No.-
q - particle charge;
Q « fiber charge/cm;
Pp • particle density;
Op * particle diameter;
U0 » upstream flow velocity;
- gas viscosity;
Of • fiber diameter; and
p « gas density.
100
t-***
IBuDp
Figure 6. Theoretical single-fiber collision efficiencies. (Inertia, interception,
and gravity with electrostatic forces. Lamb's flow field.) (Loeffler, 1978).
325
-------
IOO
,T 80
60
40
e 20
nocho,,. £ 93%»o,(l. 12-mit f.ber
/Qv 3-inch thick bed
"
o--:
01
05 10
Particle Diameter. dp . pm
Figure 7. Fractional efficiency of electrostatic fiberbed. (Yu and Teague, 1982).
100
80
60
40
20
'M I
O Electrical Mobility
Analyzer
* Laser Particle Counter
it L
0.01
0.05 0.1 0.5 1.0
Diameter, /zm
Figure 8. Filter efficiency as a function of particle size with and without an
electric field using naturally charged aerosols. (Bergman et al., 1979).
326
-------
ELECTRICAL ENHANCEMENT OF FABRIC FILTRATION: PRECHARGING VS. BAG ELECTRODES
by: R.P. Donovan
Research Triangle Institute
Research Triangle Park, NC 27709
and
L.S. Hovis and G.H. Ramsey
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC ,27711
ABSTRACT
Two distinct approaches to achieving electrically enhanced fabric fil-
tration are: 1) upstream precharging in a corona discharge; and 2) dust
filtration in an external electric field applied by electrodes woven into or
positioned adjacent to the bag. Both techniques have demonstrated perfor-
mance enhancement by some measure, but the mechanisms of enhancement may
differ; clearly the hardware does.
This paper reviews the hardware options and configurations available
with each basic approach and summarizes various mechanisms whereby each
approach brings about an electrical enhancement, including those which domi-
nate in combined precharger/bag electrode systems. From this background,
guidelines for matching configurations to source properties are suggested, as
well as a hybrid electrostatic precipitator "(ESP)/fabric filter design incor-
porating features that enable it to operate as either an electric-field aided
baghouse or a hybrid ESP/baghouse.
This paper has been reviewed in accordance with the U.S. Environmental
.Protection Agency's peer and administrative review policies and approved for
presentation and publication.
327
-------
As used in this paper, electrical enhancement of fabric filtration
refers to any technique for improving filter performance by the use of high
voltage electrodes, in either an ionizing or non-ionizing mode, at the fabric
surface or some upstream location. Even this restricted definition of elec-
trical enhancement now encompasses a growing number of configurations and
mechanisms. Rather than simplify and clarify choices, on-going research is
uncovering more mechanisms and options by which the performance of fabric
filters can be improved electrically. This paper reviews the present state
of understanding and design options now available.
The two basic variables that high voltage can control in particle col-
lection by a fabric filter are dust particle electrical charge and fabric
electric field (collector electrical charge).
Early electrical enhancement work concentrated on directly controlling
only one or the other of these quantities, while more recent research in-
cludes arrangements for controlling each independently. With the more elabo-
rate control capability has come not only more variation in control options,
but also improved understanding of the full range of electrical enhancement
mechanisms.
CORONA PRECHARGING
Typical hardware for controlling dust charge consists of a corona pre-
charger located upstream of the bags. Conventional wire-duct geometry—a
high voltage wire positioned along the centerline of a circular duct which
also serves as the low voltage electrode—lends itself to easy incorporation
in most installations, new or old. The role of the precharger is to add
electrical charge to the dust which certain researchers imply enhances fabric
filtration, although no plots of dust drag or dust cake resistance versus
collected dust charge/mass are in the literature. What typically appears is
a plot of AP, drag, or change of drag with dust load versus a measure of
corona power supply output or, more commonly, a comparison of precharger ON
and precharger OFF performance characteristics. Such data alone leave un-
answered questions regarding the enhancement mechanisms.
One obvious collection mechanism is electrostatic precipitation in the
precharging section—the precharger is a single-stage electrostatic precipi-
tator, albeit a low efficiency unit in most designs. Thus, dust charged by
the wire can be trapped in the same applied electric field that created the
corona conditions.
Even charged dust that is not trapped in the precharger section can be
precipitated on the duct walls or the baghouse walls prior to reaching the
bags. The electrical force driving the charged dust to the walls can be the
mutual repulsion between species with charge of the same sign (space charge
precipitation) or can be an image force created in the wall by each charged
dust particle. The latter is likely to be important only at low charge
densities, while the former dominates in high charge density regions more
typical of particulate control equipment design.
328
-------
Regardless of capture mechanism, the effect is to reduce the rate at
which dust arrives at the fabric, thus slowing the rate at which AP or drag
increases but not necessarily changing the dust cake specific resistance
(K2)- Thus, this enhancement mechanism is that of a prefilter. From a mass
viewpoint alone, a cyclone upstream of the bag does the same. (A cyclone
prefilter would affect the size distribution of the dust reaching the bags
differently than an ESP prefilter.)
A drawback of any prefilter concept is that two control sections must be
operated and maintained, including cleaning. Dust collected by the pre-
charger must also be accounted for and removed. Some clever sch :es exist
for minimizing this extra cost. For example, in the Apitron® (i, design
(Figure 1), the same air pulse that cleans the fabric also cleans the corona
wire and duct. The discharge from both sections is to a common hopper so
that, in principle, one cleaning action regenerates both sections.
STAST
CLEANING
RESUME
FILTRATION
Figure 1. Apitron® design and operation (1).
An alternative approach described by Lamb (2) is to locate the pre-
charger at the bag mouth so that the precipitated dust is collected on the
bag surface. In this arrangement, bag cleaning action is all that is neces-
sary since both the precipitated dust and the filtered dust are collected on
the bag surface.
329
-------
At the opposite extreme is the use of a conventional electrostatic pre-
cipitator (ESP) in series with a conventional baghouse. This option exists
in the field where regulations dictate upgrading of existing ESP equipment.
Simply adding a downstream baghouse while continuing to operate the ESP
would, in the context of electrical enhancement of fabric filtration, cause
the ESP to serve as a precharger for the dust entering the baghouse in ad-
dition to performing its usual collection function. Here, however, particu-
late control is clearly a tandem operation; both the ESP and the baghouse
would be operated and maintained as independent units. Because of the re-
duced load, the AP buildup across the fabric would be very slow, independent
of any charge-dependent increase in dust-cake permeability.
At least three cases of precharger/baghouse coupling thus exist. Table
1 classifies prechargers according to location with respect to the fabric
collection surface of the downstream baghouse. The role of a precharger is
to add electrical charge to the dust. All prechargers are assumed to perform
this function satisfactorily and to succeed in having most of the dust parti-
cles at near-saturation charge when they leave the precharger section. Then,
depending on the length of the run between the precharger and the geometry of
the intervening duct work, more or less charge is lost in transit.
For the close-coupled case, virtually all charge added to the dust by
the precharger ends up on the bag surfaces as the dust is collected. For the
independent units, considerable loss of charge and dust will occur by colli-
sion with the walls. Dust that contacts the walls but is subsequently re-
entrained will probably have lost most of its charge. The quantity of charge
reaching the fabric surface is likely to be a small fraction of that carried
away from the precharger by the dust.
No charge data exist for these two extreme cases. For the second case
listed in Table 1—the remote coupled—some measurements of charge deposited
on the bag do exist. By comparing these measured values with the total
charge that would be collected if all the particles collected carried the
saturation values of charge corresponding to the operating conditions of the
precharger, the observed charge is typically 2 to 5 percent of what is calcu-
lated to leave the precharger. (The procedure for calculating expected
charge per particle follows conventional ESP procedures and is outlined in
the Appendix of Ref. 3.)
Field coupling (Table 1) refers to a field component being present at
the fabric surface because of the location of the precharger. While a field
component at the fabric can be created not only by applying a voltage to the
corona wire but also by the charges added to the dust leaving the precharger,
the field coupling in Table 1 refers only to the direct coupling—that com-
ponent that is present when corona power is ON but no dust is flowing—and
only the close-coupled case produces such a fabric surface field component.
Common cleaning (Column 2, Table 1) implies that an additional separate
cleaning action is not required to remove the dust collected by the pre-
charger. For the separate units case (No. 1, Table 1), the ESP hoppers must
be emptied and the collection plates rapped. Common cleaning is not possi-
ble. For the close-coupled unit, there is no collection surface other than
330
-------
TABLE 1. TYPES OF PRECHARGER-BAGHOUSE COUPLINGS
3.
Electrical
Coupling
Charge Field
Common
Cleaning
Electrical
Enhancement of*
S; AS/At
K2'
1.
Two independent
separate units
(ESP upstream of the
baghouse)
Yes No No
(small)
Yes ?
(No data
available)
Remote-coupled units Yes
(Precharger in the
hopper inlet)
Close-coupled units Yes
(Precharger at the (high)
bag mouth)
No Yes/No
(Geometry
dependent)
Yes
Yes
Yes Yes/No
(Dust con-
ductivity
and size
dependent)
Yes Yes...butt
S = drag = AP/V; K2 = measured specific cake resistance = AS/AM (where
M = dust mass/unit area).
Yes, assuming uniform dust distribution, but dust distribution is known to
be highly non-uniform.
the fabric. When the fabric is cleaned, both the baghouse and the ESP col-
lectors are being cleaned, since they are the same.
The remote-coupled unit (Case 2, Table 1), as operated at EPA/RTP, does
not require additional cleaning because virtually all of the precipitated
dust collects in the hopper or the baghouse walls. No additional cleaning is
required as these surfaces slough off their dust loads in normal operation
without a special trigger. The precharger walls do not accumulate any dust
because of the high gas flow-through. Other larger remote installations may
not operate this way, and aperiodic rapping might be required.
The evidence showing filtration to be electrically enhanced consists of
reduced S and AS/At in all cases. Only for the close-coupled case is AS/AM
unambiguously reduced. This reduction in AS/AM appears to result primarily
from a highly non-uniform dust distribution along the bag length, the great-
est dust areal density being adjacent to the precharger end of the bag, where
the precharger contribution to the fabric surface electric field is concen-
trated (2).
331
-------
In the remote-coupled configurations, reports of K% enhancement conflict.
Ariman.and Helfritch (4), Chudleigh and Bainbridge (5), and linoya and Mori
(6) present data which show enhanced K2, while Donovan et al. (3) and Hovis
et al. (7) fail to find such a result. The key to this difference apparently
lies in the capability of certain collecting systems to accumulate electrical
charges as they collect dust, thereby building up an electric field on the
fabric surface. Filtration of charged dust in the presence of this self-
induced electric field produces a reduced K^. If> however, the collecting
surface is conductive, so that no significant electric field builds up, the
K.2 enhancement does not occur. At room temperature, increases in the rela-
tive humidity from 48 percent to 58 percent can convert a K£ enhancing system
to a non-enhancing system as reported in the data published by Chudleigh and
Bainbridge (5). Because of high flyash resistivity, utility baghouses oper-
ate under conditions that favor a K2 reduction in either the independent or
remote-coupled configurations, but this result also depends on particle size
being small (see Refs. 8 and 9).
Corona wires in all precharger versions must be cleaned, especially
those which are located between the bags and the hopper, such as the Apitron®
or a bottom-feed close-coupled case. During cleaning, the dust removed from
the bag cascades through the electric field on its way to the hopper, adding
an extra opportunity for dust collection on the corona wires. A top-feed
design avoids this potentially deleterious interaction.
FIELD ELECTRODES
The role of field electrodes is to create a strong external electric
field throughout the volume occupied by the filtration media and the near up-
stream volume of the fabric. Orientation of the electric field can be either
in the direction of gas flow through the filter--roughly perpendicular to the
fabric surface—or in the plane of the fabric surface—roughly perpendicular
to the direction of gas flow (Table 2). Different authors have preferred
different orientations based on analyses of a single fiber in an electric
field. For example, Havlicek (10) argues for the electric field in the
direction of the gas flow; Lamb et al. (11) prefer the electric field perpen-
dicular to the gas flow. Without discussion, Zebel (12) solves only the case
of electric field parallel to the gas flow.
In practice, both orientations have been used successfully. The practi-
cal advantages of building the electrodes into a cage or the fabric itself
have resulted in more experience with the electric field in the plane of the
fabric than along the direction of gas flow, but limitations in this arrange-
ment, chiefly with respect to power consumption with low resistivity dusts or
fabrics, may lead to more research with the alternative orientation and
arrangement. Primary emphasis here will be on configurations that orient the
field in the plane of the fabric—the ESFF (electrical stimulation of fabric
filtration) orientation developed at Textile Research Institute (11, 13).
332
-------
TABLE 2. FIELD ELECTRODE OPTIONS
Electric Field Perpendicular to Fabric Surface
a) center-line wire electrode (inside-out flow)
b) enclosing cage or housing wall (single bags)
Electric Field in the Plane of the Fabric Surface
a) electrodes upstream (dirty side)
b) electrodes downstream (clean side)
c) electrodes woven into the fabric
Even after the selection of the field orientation in the plane of the
fabric, options exist. The electrodes can be upstream (wholly on the dirty-
gas side of the fabric), downstream (wholly on the clean-gas side of the
fabric), or incorporated into the fabric itself (part upstream, part down-
stream, and part within).
The option recommended by Lamb and Costanza (14) (and still the prefer-
red from a field strength viewpoint) is the upstream location in which the
electrodes, an array of parallel wires of alternating polarity, are positioned
immediately in front of the fabric but not quite touching it. This location
creates a high electric field in the near-upstream region of the fabric just
prior to dust collection. The electrical enhancement is greater when the
fabric surface has a heavy nap finish which causes many surface fibers to
protrude into the upstream region. In the presence of an electric field
these fibers become more efficient capture centers than without the electric
field. Since dust collected in this low fiber density region makes a smaller
incremental contribution to fabric drag than dust collected deeper within the
fabric, the net result of turning ON the field is a lower K^. This model of
ESFF has been dubbed "the tall trees" model (15), and represents a depth
redistribution of collected dust mass. Areal mass density, averaged over
dimensions large with respect to fiber-to-fiber separation, is assumed con-
stant in this model. Dust mass distribution as a function of depth into the
filter is postulated to account for the improved performance.
A major advantage of the upstream location is that the electric field in
the near-upstream volume from the fabric is not attenuated by fabric elec-
trical properties. When the electrodes are shifted to the downstream clean
side, a layer of fabric shields this critical near-upstream volume from the
electrostatic field. With conductive fabrics (or conductive dusts which
become an important part of the filtration media as the dust accumulates in
and on the fabric), a significant reduction in the electrical field in the
near-upstream volume could occur. The "tall trees" model predicts that such
a reduction in field in this region should reduce the magnitude of the en-
hancement per unit applied voltage between the electrodes.
333
-------
In some reports switching from a high resistivity fabric such as Teflon
to a lower resistivity material such as polyester does in fact degrade the
observed electrical field enhancement, confirming qualitatively this predic-
tion of the tall trees model. By other accounts, however, this fabric depen-
dence is not observed. And, in the initial pilot-scale work on an industrial
boiler (16), switching from upstream electrodes to downstream electrodes
(Teflon fabric) had little effect on the electrical enhancement. Thus,
attenuation of the electrical field appeared unimportant here.
The major advantage of downstream electrodes is that they can be easily
incorporated into the support cages of outside-in bag filters, allowing ESFF
to be implemented with only minor modifications in the conventional hardware.
Rather than require the design of a special electrode harness as in the up-
stream option, the downstream option can be implemented by the simple addi-
tion of electrical insulators between appropriate cage members. This arrange-
ment has been shown to perform with impressive economic advantage on a pul-
verized-coal-fired industrial boiler (16), even though the fabric used in
this demonstration was the relatively expensive Teflon felt.
Electrodes woven into the fabric are a third option seemingly well
matched to the requirements of reverse-air cleaned baghouses. In this con-
figuration yarns made of conductive fibers become part of the fabric weave.
Choice of suitable weave pattern allows the conductive yarn to face primarily
upstream if that is deemed desirable or to be incorporated in any of the
variety of weaves available to textile processing. A major advantage of this
configuration is that it minimizes user preparation and demands. In routine
service the plan is to install and handle the ESFF bags much the same as
conventional bags. The only differences would be that the fabric is special
and an electrical contact, perhaps no different from conventional mechanical
attachment, however, must be made at each end of the bag. In operation the
ESFF power supplies must be monitored.
Experience with this option is limited to laboratory operation at pre-
sent but already some striking differences are apparent, the most pronounced
of which is preferred deposition of the collected dust on the electrodes, as
reported by Lamb (2). This mechanism of enhancement is a surface redis-
tribution phenomenon—the deposited dust does not coat the fabric surface
uniformly as in conventional filtration but piles up on the electrodes cre-
ating a distinct striped deposition pattern in the early stages of filtra-
tion. The dust pattern mirrors the electrode geometry, although one elec-
trode polarity generally appears to collect more dust than the other.
This kind of deposition pattern leads to an electrical enhancement
attributable to non-uniformity in dust deposition between electrodes, similar
to that previously reported by Penney (17) and by Chiang et al. (18). This
type of deposition non-uniformity is also similar to that previously associ-
ated with the close-coupled precharger. When the woven-in electrodes are in
the warp direction (along the bag axis) of a vertically suspended bag, the
deposition non-uniformity is in the lateral direction (the concentration
gradient is maximum in the horizontal direction). When combined with a
334
-------
close-coupled precharger (deposition non-uniformity in the vertical direc-
tion), as in Lamb's configuration, the resultant two-directional non-unifor-
mity produces dust deposits of drastically reduced K£ as compared to either
single-direction non-uniformity alone or the uniform deposition case (2).
Because of the parallel flow paths through a bag, the dust distribution
that results in the highest net drag is the uniform distribution. Any depar-
ture from uniformity produces a lower net drag because the flow through the
region of reduced drag increases with respect to the flow through all other
regions. This argument assumes that drag is proportional to areal mass
density. Decreasing the mass density in one area decreases that area's drag
and correspondingly increases the drag of all those areas to which the dis-
placed mass has been added.
Figure 2 illustrates the parallel flow concept. These curves, adapted
from Dennis et al. (19), plot net filter drag as a function of dust loading
for two distributions (Figure 3): 1) the uniform loading case, and 2) a
highly non-uniform loading case. The non-uniform case assumes that the dust
deposits on the fabric in increments of areal density equal to 700 g/m2 so
that the fabric consists of two areas: a loaded area and an unloaded area.
As deposition continues, the size of the loaded area (dust load always equals
700 g/m2) grows uniformly with time. In the uniformly loaded case, the dust
load per unit area increases uniformly with time.
10
Fraction of Filter Surface Covered
(Non-uniform loading curve only)
0.2 0.4 0.6 0.8 1.0
fraction of surface loaded to 700 g/m*
Nmin
15
dean surface drag
Sd = drag with 700 g/m2 loading
N min
Uniform
Loading Curve
Highly Non-uniform
Loading Curve
6 100 200 300 400 500 600 700
Average Fabric Loading, g/m2
Figure 2. Impact of non-uniform dust loading upon drag (19)
335
-------
Uniform Loading
Highly
Non-uniform Loading
a) Zero load
b) Initial deposition
c) Half final load
d) Final loading
— Fabric surface
Deposited dust
Figure 3. Loading schematic.
The curves of Figure 2 show that, for the drags assumed for the loaded
and unloaded areas, the net filter drag of the highly non-uniform case is not
only lower than that of the uniformly loaded case but (over much of the range
of dust loads) dramatically so. Not until the non-uniform case begins to run
out of area does its dS/dM exceed that of the uniform case. This rapid rise
in S brings the two curves together at the same final loaded condition (the
non-uniform case becomes uniform at 700 g/m2 loading when all its area is
loaded), but the route over which that same end point is reached differs
dramatically.
Non-uniform loading via electrical forces has the effect of changing the
conventional uniform loading drag characteristic into one more like the
non-uniform loading curve illustrated in Figure 2. Chiang et al. (18) have
mathematically modeled pressure drop reduction for both step changes in dust
cake areal density and bell-shaped distributions in support of non-uniform
deposition as a mechanism of electrically enhanced fabric filtration.
This non-uniform loading mechanism is distinct and independent from the
tall trees mechanism. Both the tall trees model and the non-uniform loading
model can predict the observed results, and some experimental confirmation of
each now exists in the literature. Which dominates when is important in
choosing hardware for electrical enhancement.
One further option should be resurrected—that of the electric field
aligned parallel to the gas flow (perpendicular to the fabric surface). An
336
-------
embodiment that is both easy to implement and compatible with reverse-air
baghouse design is the suspension of a corona wire down the center of a bag
which is itself either conductive or else incorporates conductive electrodes
woven into its fabric. This configuration can operate in either an ionizing
or non-ionizing mode; its main advantage, however, is that it is compatible
with dusts which otherwise, with electrodes in the plane of the fabric, would
draw prohibitively large currents. Note, however, that such a configuration
creates the highest field and the highest field gradient in the vicinity of
the corona wire. This feature means that this configuration will probably be
most effective when operated as an electrostatic precipitator (ESP), gener-
ating its own charges for the dust and drifting this charged dust by Coulomb
forces to the fabric surface for capture. In this mode the equipment then
operates as a conventional wire-duct ESP in which the collector surface is a
fabric filter and the air flow is through the collector plate rather than
parallel to it. It is in effect a hybrid ESP/fabric filter.
SUMMARY OF MECHANISMS
Mechanisms capable of predicting electrically enhanced filtration in-
clude:
1. Prefiltering, because of electrostatic precipitation.
2. The "tall trees" model, a depth distribution effect.
3. Non-uniform deposition, a surface distribution effect.
Other candidate mechanisms, such as electroclamping (20), exist.
Comparing these mechanisms with the choices of precharger configurations
(Table 1) and field electrode configurations (Table 2) leads to certain
general conclusions:
1. Prefiltering applies only to precharger location; the close-coupled
precharger minimizes its significance.
2. The tall trees model depends strongly on fabric properties and
electrode location. It postulates enhanced dust capture by up-
stream fibers, so depends on any variable that contributes an
electrical capture force such as fiber-scale electric field in the
upstream region, and dust electrical charge. This mechanism is
compatible with either Coulomb forces or polarization forces.
3. Non-uniform deposition also depends on electrical forces but re-
sponds over a bigger volume—the electrical forces in the upstream
regions remote from the near-fabric-surface volume, most likely
Coulomb forces, are important. Distinct from the tall trees mecha-
nism, it does not depend on fabric surface finish and should be
less dependent on fabric properties, a distinct advantage.
Any specific hardware design for electrical enhancement depends on
understanding the mechanism by which the electrical enhancement occurs. In
general this understanding does not yet exist. What typically is known is
337
-------
that electrifying a given conventional operation produces certain improve-
ments in performance. Reports of such improvements are frequently accom-
panied by speculation as to the enhancement mechanism, such as those previ-
ously listed, but established proof is scarce.
Against such a background and with the uncertainties it implies, a few
tentative guidelines can be suggested (Tables 3 and 4).
TABLE 3. PRECHARGER DESIGN CONSEQUENCES
Configuration
Dust
Remarks
Reduction
Close coupled
Any
Any
Small (<3-5 |Jm),
nonconductive
No K2 Reduction
Remote or separate Any size, conductive
Remote or separate Large (>5 Mm) >
nonconductive
Non-uniform deposition
Coulomb force aided (field
self-created by collected
particles)
No self-created field
Inertial capture forces
dominate Coulomb forces
TABLE 4. ELECTRODE DESIGN CONCLUSIONS
1. Downstream electrodes, incorporated into the support cage, match
up best with outside-in filters.
2. Woven-in electrodes are the preferred configuration for inside-out
filtration.
3. Upstream filters, while electrically advantageous, do not justify
their added cost at present.
1) While any of the precharger locations described can serve as a pre-
filter, the close-coupled configuration is the only one guaranteed to decrease
the dust-cake resistance because it is the only one which precipitates dust
on the bag surface and capitalizes on the non-uniform deposition mechanism
(2, 17, 18). For the special category of non-conductive small particles
338
-------
(less than say 3 to 5 (Jm), the self-induced electric field at the fabric
surface could also produce a reduced K^ using any of the precharger configu-
rations (8, 9, 21).
2) With remote or separate precharger configurations„ the only guaran-
teed mechanism of electrical enhancement is that of a prefilter. Only when
filtering small non-conductive dust particles is the electrical enhancement
likely to include a K2 reduction (9). The prefilter function is itself often
worthwhile, especially if it entails no additional cleaning hardware, hoppers,
or dust removal, as is true in the EPA/RTP pulse-jet baghouse.
3) When corona prechargers are used, they will be subject to operating
limitations similar to those of existing ESPs with regards to back corona.
Note, however, that poor particle charging and low-efficiency collection can
be better tolerated in a precharger than in an ESP. Thus, a precharger
should not be ruled out simply because an ESP has been ruled out.
4) Bag electrodes that create an electric field in the plane of the
fabric are simple to build and use when they are incorporated into the bag or
as part of the cage. This consideration favors the electrode/cage assembly
for pulse-jet operations and woven-in electrodes for reverse air (printed
electrodes, not yet successfully developed, also share this advantage [2]).
The performance advantages cited for upstream electrode placement do not
justify their added expense. Ruggedness and long field life would need to be
included in any demonstration to alter this conclusion.
5) Configurations that favor non-uniform deposition such as the woven-in
electrode fabric and the close-coupled precharger described by Lamb (2)
appear less dependent on fabric properties and electrode location than those
relying on the tall trees mechanism; that is, dust non-uniformities (large
with respect to fiber or dust particle dimensions) are less dependent on
fiber or dust properties than those that depend on dust-cake porosity changes.
The fields and electrical charges needed to create gross non-uniformities are
easier to reproducibly create in the laboratory than the fiber-scale fields
and charges.
6) A practical field-electrode configuration compatible with conductive
dust ESFF, while not yet demonstrated, appears to be available based on a
variation of one embodiment contained in Sharlit's patent (22) or as discus-
sed by Frederick (23). This concept is the suspension of a corona wire down
''the centerline of a reverse-air bag. The counter electrode is the bag itself
! (coated with conductive particles) or electrodes woven into the bag. No
conductive path forms between the high voltage and the bag electrodes. When
in corona, this configuration operates as a hybrid ESP/fabric filter.
ACKNOWLEDGEMENTS
It is a pleasure to acknowledge the contribution of George E.R. Lamb and
his coworkers at Textile Research Institute who hosted one of the authors
(R.P. Donovan) for an extended visit during which certain of the ideas ex-
pressed in this paper were crystallized or refined.
339
-------
REFERENCES
1. Helfritch, D.J. Apitron®. Product Brochure, Apitron Division, American
Precision Industries, 12037 Goodrich Drive, Charlotte, NC 28217, 1979.
2. Lamb, G.E.R. Influence of particulate precharging on the performance of
an electrically stimulated fabric filter (ESFF). Paper presented at
World Filtration Congress III, The Filtration Society, Downingtown, PA.
September 14-16, 1982.
3. Donovan, R.P., Hovis, L.S., Ramsey, G.H., and Abbott, J.H. Pulse-jet
filtration with electrically charged flyash. In: Third Symposium on
the Transfer and Utilization of Particulate Control Technology, Vol. I.
EPA-600/9-82-005a (NTIS PB 83-149583), July 1982.
4. Ariman, T. and Helfritch, D.J. Pressure drop in electrostatic fabric
filtration. In: Second Symposium on the Transfer and Utilization of
Particulate Control Technology, Vol. III. EPA-600/9-80-039c (NTIS PB
81-144800), September 1980.
5. Chudleigh, P.W. and Bainbridge, N.W. Electrostatic effects in fabric
filters during build-up of the dust cake. Filtration and Separation.
309-311, July/August 1980.
6. linoya, K. and Mori, Y. Experimental advances in fabric filtration
technology in Japan—effects of a corona precharger and relative humidity
on filter performance. In: Second Symposium on the Transfer and Utili-
zation of Particulate Control Technology, Vol. III. EPA-600/9-80-039c
(NTIS PB 81-144800), September 1980.
7. Hovis, L.S., Abbott, J.H., Donovan, R.P., and Pareja, C.A. Electrically
charged flyash experiments in a laboratory shaker baghouse. In: Third
Symposium on the Transfer and Utilization of Particulate Control Tech-
nology, Vol. I. EPA-600/9-82-005a (NTIS PB 83-149583), July 1982.
8. Loeffler, F. The influence of electrostatic forces and of the proba-
bility of adhesion for particle collection in fibrous filters. In:
T. Ariman (ed.), Novel Concepts, Methods and Advanced Technology in
Particulate-Gas Separation. University of Notre Dame, 1978.
9. Yu, H.S. and Teague, R.K. Performance of electrostatic fiberbed. Paper
presented at 1st Annual Meeting of the American Association for Aerosol
Research, Santa Monica, CA. February 1982.
10. Havlicek, V. The improvement of efficiency of fibrous dielectric fil-
ters by application of an external electric field. Int. J. Air and
Water Poll. 4(3/4): 225-236, 1961.
11. Lamb, G.E.R., Costanza, P.A., and O'Meara, D.J. Electrical stimulation
of fabric filtration. Part II: Mechanism of particle capture and
trials with a laboratory baghouse. Textile Research Journal. 48(10):
566-573, October 1978.
340
-------
12. Zebel, G. Deposition of aerosol flowing past cylindrical fiber in a
uniform electric field. J. Colloid Sci. 20, 1965.
13. Lamb, G.E.R. and Costanza, P.A. Electrical stimulation of fabric fil-
tration. Textile Research Journal. 47(5): 372-380, May 1977.
14. Lamb, G.E.R. and Costanza, P.A. A low-energy electrified filter system.
Filtration and Separation. 319-322, July/August 1980.
15. Greiner, G.P. Electrical stimulation of fabric filtration. Paper pre-
sented at APCA 74th Annual Meeting, Philadelphia, PA. June 1981.
16. Greiner, G.P., Furlong, D.A., VanOsdell, D.W., and Hovis, L.S. Elec-
trostatic stimulation of fabric filtration. J. Air Pollut. Control Assn.
31:10, 1125-1130, October 1981.
17. Penney, G.W. Using electrostatic forces to reduce pressure drop in
fabric filters. Powder Technology. 18: 111-116, 1977.
/•
18. Chiang, T-K., Samuel, E.A., and Wolpert, K.E. Theoretical aspects of
pressure drop reduction in a fabric filter with charged particles. In:
Third Symposium on the Transfer and Utilization of Particulate Control
Technology, Vol. III. EPA-600/9-82-005c (NTIS PB 83-149609), July 1982.
19. Dennis, R., et al. Filtration model for coal flyash with glass fabrics.
EPA-600/7-77-084 (NTIS PB 276489), August 1977.
20. VanOsdell, D.W., Donovan, R.P., Furlong, D.A., and Hovis, L.S. Permea-
bility of dust cakes collected under the influence of an electric field.
Paper presented at The Fourth Symposium on the Transfer and Utilization
of Particulate Control Technology, U.S. Environmental Protection Agency,
Houston, Texas. October 11-15, 1982.
21. Bergman, W., Hebard, H.D., Taylor, R.D., and Lum, B.Y. Electrostatic
filters generated by electric fields. Proceedings of The Second World
Filtration Congress. 1979.
22. Sharlit, I.E. Filter system. U.S. Patent 3,577,705 (to Hitco), May 4,
1971.
23. Frederick, E.R. Some effects of electrostatic charges in 'fabric filtra-
tion. J. Air Pollut. Control Assn. 24(12): 1164-1168, December 1974.
341
-------
PERMEABILITY OF DUST CAKES COLLECTED UNDER
THE INFLUENCE OF AN ELECTRIC FIELD
By: D. W. VanOsdell and R. P. Donovan
Research Triangle Institute
P. 0. Box 12194
Research Triangle Park, North Carolina 27709
D. A. Furlong
ETS, Inc.
Roanoke, Virginia 24018
L. S. Hovis
Industrial Environmental Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
The reductions in dust cake flow resistance and residual pressure drop
that can be achieved by electrostatic stimulation of fabric filtration (ESFF)
have been documented. The changes in the dust deposition characteristics
which cause these improvements are not well understood. This paper presents
results from three investigations into the nature of dust deposits collected
with ESFF. Permeability data collected at various axial positions along
ESFF and conventional bags were the basis of the first study. The second
study concerned the deposition patterns of dust collected on a laboratory
filter when ESFF was in use, and the third, the resistance to compression of
the collected dust with and without an electric field.
These investigations gave evidence that ESFF caused the formation of a
residual dust deposit which was not axially uniform (the permeability at the
bottom of the bag was half that at the top) and was more easily removed than
a conventional dust deposit. In the laboratory, it was shown that ESFF caused
newly collected dust to form very nonuniform deposits. The dust tended to
collect near the electrodes, in patterns which have not been observed in the
pilot units. In the third study, the ESFF electric field was found to improve
the resistance to consolidation of the dust cake collected in a pulse-jet
pilot unit.
This paper has been reviewed in accordance with the U. S. Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication.
342
-------
INTRODUCTION
The use of electrostatic fields to improve the performance of fabric
filters is a topic of considerable research interest today. Numerous re-
searchers have investigated the topic, applying electrostatics in different
ways. Particle charging, accumulation of charge on the fabric filter, and
the application of external electric fields (either perpendicular or paral-
lel to the gas flow) have all been used singly or in combination during
this research. In all cases, it has been found that the performance of the
filter has been improved in the sense that the drag due to the collected
particles was reduced. Generally, an improvement in collection efficiency
was also noted. Although much has been learned about the beneficial effects
of ESFF, the mechanismsjby which it operates in a baghouse are not well
understood. The experimental evidence is difficult to reproduce and some-
times contradictory. One important question is whether the ESFF filter de-
posit is nonuniform at the size scale of the bag, at the scale of the elec-
trode spacing, or at the scale of the particles. Treating the filter as an
assembly of resistances in parallel, at any scale', can account for the ESFF
effect mathematically.(1) Another question is whether the reduced dust cake
drag attributed to ESFF is due to the structure of the dust cake as col-
lected, or to some other phenomenon related to the electric field.
This paper describes three experiments with electrostatically stimulated
fabric filtration (ESFF) which cast light on the mechanisms by which ESFF
causes reduced filter drag. All three experiments utilized the same concept
of ESFF--no particle charging and an electric field applied at the filter
surface and oriented parallel to the surface. This concept was initially
developed by the Textile Research Institute(2,3,4) and was extended to pilot
scale by the Research Triangle Institute (RTI) and ETS, Inc.(5,6)
The first experiment to be described was conducted on bags from the
Waynesboro ESFF pilot unit. The permeability of the cleaned bags (both ESFF
and conventional) was determined at the top, bottom, and middle areas of the
bags. In addition, the particle size distribution of the dust collected at
the top and bottom of an ESFF bag was determined. The second experiment con-
sidered the appearance of dust deposits collected with and without electric
fields. A small laboratory filter was used to remove resuspended fly ash
from an air stream, the pressure drop characteristics were measured, and the
filter deposits photographed. The third experiment investigated the com-
pressibility of dust collected with and without an electric field. This ex-
periment was conducted on a pilot-scale pulse-jet fabric filter operating on
fly ash resuspended in ambient air.
FILTER BAG AXIAL PERMEABILITY MEASUREMENTS
EXPERIMENTAL ARRANGEMENT
The experiments reported in this section were conducted on bags from the
pilot-scale ESFF baghouse located at Waynesboro, Virginia. This baghouse has
been described in earlier papers(5,6), and only a brief description is pre-
sented here. The pilot unit included two identical baghouses operated as
parallel units with identical inlets. One baghouse utilized ESFF; the other
343
-------
was conventional. The pilot unit was operated on a slipstream from an indus-
trial bo.iler house. A schematic diagram of the pilot unit is given in
Figure 1. The boiler burned pulverized coal, and the coal supply was highly
variable; the average sulfur content was about 1.3 percent and the ash con-
tent was about 13 percent. Boiler operation was continuous. The average
inlet temperature to the baghouse was about 155° C (310° F), with an average
dust loading of about 0.9 g/sm3 (0.4 gr/scf).
The pilot baghouses (both conventional and ESFF) each contained four
bags in a dirty-gas-inside, reverse-air-cleaned configuration. The filter
bags were made of J. P. Stevens 648 fabric, a 570 g/m2 (16.8 oz/yd2) woven
fiberglass cloth with a 10-percent Teflon B coating. The initial permea-
bility of the fabric was 18 to 28 cm/s at 0.125 kPa (35 to 55 ft/min at
0.5 in. H20). The electrostatic augmentation of the ESFF bags was provided
by a vertical array of 3.2-mm (0.125-in.) wires, connected to provide alter-
nating grounded and high negative potential electrodes (Figure 2).
TEST PROGRAM
The pilot baghouses were used to test the long-term behavior of reverse-
air-cleaned fiberglass bags under ESFF conditions. The operational results
of this test have been reported(5), and this paper discusses permeability
results obtained after the bags had been removed from the baghouses. Re-
sults from two bag sets are reported. One bag set (to be referred to as JF)
was in use during January and February 1982. The second set (June) was in
use during June 1982. All bags were conditioned at 1 cm/s (2 ft/min) and
were operated at about 2 cm/s (A ft/min). The electrical field was applied
to the ESFF bags from start-up and was maintained continuously at about
3 kV/cm. During the JF test period, the conventional baghouse was used as a
control on the ESFF experiment. During the June test, both baghouses were
in use in different ESFF arrangements, and control bags were not available.
The permeability tests reported here were run on a standard bag permea-
bility test apparatus.(7) The JF bags were in a "just-cleaned" state when
tested. The June bags were removed before cleaning, taken to the laboratory
to remove dust samples, then extended to allow the remaining dust cake to
fall off prior to permeability testing. Permeability tests were run on the
fabric in an "as-received" condition and following vacuum cleaning at 3.75 kPa
(15 in. H20) vacuum and at 7.5 kPa (30 in. H20) vacuum. These tests were
done for both the conventional and ESFF bags.
The particle size distribution tests were run on dust samples from the
top and bottom of the June bags by resuspending and analyzing the dust in a
Bahco apparatus.
RESULTS
The results of the permeability tests are presented in Table 1, and the
results of the Bahco size analysis in Figure 3. Features of these results
which merit special attention are discussed below.
344
-------
UJ
\ OUTLET (TOP)
CONTROL
HOUSE
) INLET (BOTTOM)
,, INLET
(BOTTOM)
Figure 1. Schematic diagram of ESFF pilot unit,
OUTLET
(TOP)
-------
Top Electrodes at
High Negative Potential
Clamp
Insulating Block
Standard
Top Cap
Glass Fabric
Tape Electrode
Support
Approximate
Top of
Thimble
Bottom Electrodes
Grounded
Figure 2. Pilot-unit reverse-air bag electrode array.
346
-------
TABLE 1. PERMEABILITY RESULTS FOR ESFF AND CONVENTIONAL BAGS*
Pet-mobility Resurts-ft/min e 0.5 in. H^ AP
(cm/ie0.125kPaAP)
Simple Descriptor
As received:
Top
Middle
Bottom
Vacuumed @ 3.75 kPa
(15 in. H20):
Top
Middle
Bottom
Vacuumed @ 7.5 kPa
(30 in. H2O):
Top
Middle
Bottom
January- February (JF)
Conventional Bag
3.8(1.9)
2.6(1.3)
3.3(1.7)
3.8(1.9)
5.5 (2.8)
2.9 (1.5)
7.4 (3.8)
8.3 (4.2)
5.7 (2.9)
January-February (JF)*
ESFF Bag
5.5 (2.8)
3.6(1.8)
1.6(0.8)
11.8(6.0)
6.7 (3.4)
4.4 (2.2)
13.6(6.9)
9.7 (4.9)
7.5 (3.8)
June*
ESFF Bag
4.5 (2.3)
IM.M.I
3.6(1.8)
11.5(5.8)
N.M.§
10.5 (5.3)
17.5(8.9)
N.M.§
12.5(6.4)
* All test work performed by ETS, Inc.
* January-February results from one bag.
* June results are average from two bags.
§N.M. = not measured.
S
35
99
98
95
90
80
I-
1 »
I 40
| 30
5
20
10
5
Ash Sample from
Top of Bag / O
Ash Sample from
Bottom of Bag
3 45678
Particle Size, jim
10
20
Figure 3. Bahco size distribution of dust from top and bottom of
June ESFF bag.
347
-------
Axial Variation of Permeability of ESFF Bag
The permeability at the top of the ESFF bag was consistently greater
than at the bottom, both before and after vacuuming and for both ESFF bags.
This is especially evident for the JF ESFF bag. The permeability of the JF
conventional bag was lowest at the bottom, highest in the middle or top, and
did not vary as much between positions as did the JF ESFF bag.
Magnitude of Permeability
The average of the three permeability measurements for the ESFF bag in
the as-received state was 11 percent greater than that of the conventional
bag. It was 87 percent above the conventional bag permeability after the
first vacuuming and 44 percent higher after the second.
Ease of Cleaning
The JF ESFF bag had an average permeability before vacuuming of
3.6 ft/min (1.8 cm/s). Following vacuuming at 3.75 kPa, the average perme-
ability had risen to 7.6 ft/min (3.8 cm/s), an improvement of 11 percent.
For the JF conventional bag, the same treatment increased the permeability
from 3.2 ft/min (1.6 cm/s) to 4.1 ft/min (2.0 cm/s), an increase of
28 percent.
Axial Variation of Dust Deposit Size Distribution
The size distribution of the dust at the top of the ESFF bag differs
from that at the bottom (Figure 3). Mass mean diameter at the bag top was
2.6 JJm and at the bottom, 3.7 pm. As no corresponding data are available for
the conventional pilot plant bags and others(8) report similar trends for
conventional baghouses, this variation cannot be ascribed to the ESFF effect.
DUST COLLECTIONS ON A LABORATORY FILTER
EXPERIMENTAL APPARATUS
The laboratory filter deposition study was conducted on the apparatus
shown in Figure 4. As shown, the filter is oriented face downward and is
10 cm square. The electric field was provided by six parallel electrodes
spaced 2 cm apart and located on the clean side of the filter. The test dust
was resuspended fly ash, fed by a screw feeder into a sonic air jet, blown
against an irapaction plate, and then into the inlet duct. The tests were
conducted with ambient air as the carrier gas.
TEST PROGRAM
Fly ash from the Waynesboro pilot unit, resuspended in a sonic air jet
at about 4 g/m3 (1.8 gr/ft3), was fed to the filter at a face velocity of
2 cm/s (4 ft/min). Cascade impactor size analysis indicated a mass median
aerodynamic diameter of about 3.1 pro for the resuspended ash. An extractive
charge-to-mass measurement based on a filter within a Faraday cup indicated
an average net positive charge on the resuspended ash of 0.57 MC/8- The
348
-------
Sample ,
Point
——
Samp
Poin
r
$
'I
•f
?
1-1
,tefe
T
Dust
Injection
1
i
j[
•9
\
}
Ji
— r
i
i
i
ij
Test
Filter
Holder
Normal
. Gas
/^^ How
Valve
|
i
Ik
sss
ss
0
d
J IK -'II LL
\
r Valve |f
AF°I
Valve
l
]
l
L«_J Clean
in Air
" 1 Inlet
^
J Valve
L
II-' )
ji
nf=
i
= *t|— { Pump and
— 11— J Pinal Piltar
Valve
Figure 4. Laboratory filtration apparatus,
349
-------
filter was of glass fiber, and a new filter was used for each test. The
filter was weighed before and after the run to get an approximate measure
of the cake mass. Run time was 100 min for these experiments, and the total
dust cake mass was between 4 and 5 g.
RESULTS
The dust accumulations on the filter were nonuniform when the electric
field was ON, becoming increasingly nonuniform as the field increased. Fig-
ure 5 is a photograph of the dust cake when collected with the field OFF.
The faint surface lines were produced by turning the field ON after the dust
cake was collected, with clean gas flow through the filter. Figure 6 shows a
deposit collected with an electric field of 1 kV/cm, and Figure 7, a deposit
collected with a 4-kV/cm electric field. The pressure drop behavior of each
deposit is shown in Figure 8; as expected, the dust cakes with the field
ON have a lower pressure drop than with the field OFF at any given time.
Dust samples from the positive and negative electrodes of a 4-kV/cm
run were analyzed by a Coulter counter, along with a sample from between the
electrodes. The size distribution curves from the three sites were identi-
cal; the mean diameter was 2.37 pm from the Coulter analysis.
INFLUENCE OF AN ELECTRIC FIELD ON DUST CAKE COMPRESSIBILITY
INTRODUCTION
The compaction of filter dust cakes due to the increasing pressure drop
across the cake has been described.(9) Increased interparticle adhesion
within a particle bed due to an external electric field has also been docu-
mented and discussed.(10,11) Increased interparticle adhesion has the po-
tential of being an important ESFF mechanism. If a large fraction of the
pressure drop across a filter cake is caused by the continual compression of
the cake, then interparticle forces which prevent or reduce cake compression
would reduce the net pressure drop much as is observed with ESFF. Some evi-
dence for a mechanism of this sort can be observed in a pilot unit pressure
drop characteristic curve. Figure 9 compares the AP for the buildup of a dust
cake collected with the electric field ON to a conventional field-OFF curve.
While the reduced AP of the field-ON curve is evident throughout, the field-ON
curve exhibits a continually increasing slope which includes a series of steps
at high pressure drops—indicating dust cake consolidation by discrete cake
collapses. The field-OFF curve also shows these steps but not as many and
they are reduced in magnitude. This difference between the two curves sug-
gests that the field may increase interparticle adhesion sufficiently to
support a more porous dust cake initially but that at least part of this ad-
vantage can be lost through cake compression. The discussion below reports a
test series which investigates the drag of dust cakes collected with and
without an electric field when subjected to stress from increased filter AP.
EXPERIMENTAL APPARATUS
The experiments were conducted on a small pulse-jet fabric filter lo-
cated at the Industrial Environmental Research Laboratory of the U. S.
350
-------
Figure 5. Patch filter dust cake collected at 0 kV/cm.
'i *'
Figure 6. Patch filter dust cake collected at 1 kV/cm.
351
-------
, ,
• -<-
*U-
Figure 7. Patch filter dust cake collected at 4 kV/cm.
1.25 r-
i.oo
0.75
fe 0.50
0.25
OkV/cm
4 kV/cm
I i I
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Time From Start of Filtration Run, min
Figure 8. Pressure drop characteristics of laboratory filter
at constant dust rate.
352
-------
1.00
0.75 -
£
eC
0.50 -
0.25 -
10
20
30 40 50 60
Filtration Time, min
70
80
90
Figure 9. Long filtration time pressure drop curve.
Environmental Protection Agency in Research Triangle Park, North Carolina.
This particular pilot baghouse has been described previously.(12) In sum-
mary, the baghouse used 11.4-cm-diameter (4.5-in.) bags, 122 cm (4 ft) long,
made of Teflon felt. The test dust was resuspended fly ash at a concentra-
tion of about 6.4 g/ra3 (2.8 gr/ft3). Three bags were in use. All of the
work was at ambient conditions using air as the carrier gas.
TEST PROGRAM
The test program was a three-step procedure. First, a dust cake was
formed on the bags—with or without an electric field depending on the test.
The dust cake was built until a predetermined pressure drop had been reached
at the initial face velocity of 8.5 ft/min. The ESFF effect caused the dust
cake in the field-ON condition to require more time (and more fly ash) to
build than was required for an equivalent AP dust cake in the field-OFF condi-
tion. Once the dust layer was in place, the dust feed was stopped. The dust
cake was then stressed by increasing the flow rate to reach a new, higher,
predetermined AP. The flow rate at the increased stress level varied from
test to test; only the AP was fixed. Following a fixed time period at the
high AP level, the flow rate was reduced to the initial flow rate (the flow
rate at which the cake was built) and the AP recorded. The change in AP at
the initial flow rate is then indicative of the amount of cake compression
which took place during the stress period.
353
-------
RESULTS
The results of this test program are presented in Table 2. The Aug-
ust 6th data, which followed a thorough bag cleaning by vacuuming, permits
a fairly clear interpretation. Depositing a dust layer with the electric
field ON results in a layer which proves more compressible under a fixed
field-OFF AP loading (Run 5) than a similarly stressed layer deposited with
the field OFF (Run 6). In fact, the post-stress AP across a layer treated as
in Run 5 (deposit field ON, stress field OFF) is very close to what would be
predicted for depositing the same quantity of dust with the field OFF. It is
as though the air stress step eliminated the drag differences between the
field-ON and field-OFF depositions. However, it is notable that leaving the
field ON during the air stress step largely preserves the reduced drag of the
field-ON deposit.
CONCLUSIONS
In summary, the results of these three experiments are:
1. The residual permeability of an ESFF bag varied axially to a
greater extent than that of a conventional bag. The ESFF bag
was more easily cleaned by vacuuming and had a greater average
permeability.
2. The dust deposits collected with ESFF can be very nonuniform;
nonuniformities of this scale have not been observed in pilot
unit operation.
3. A given pressure drop caused less dust cake consolidation with
ESFF than without; the reduced drag due to ESFF can be elimi-
nated by sufficient dust cake compression.
TABLE Z STRESS DATA FOR FIXED AP LEVELS
Run
1
2
3
Date
7/28
7/28
7/29
Deposit AP
kP»
(deposit time)
0.91
(~14min)
0.91
(~28 min)
0.85
(~28 min)
Strati AP
kPa
1.74
1.74
1.87
Flow at
high stress
nrVrnin
4.87
5.64
Unstable
Field status
Deposition
Off
On
On
Stress
Off
On
On
Change in AP
after stress
kPa Remarks
0.31
0.17
0.30
0.25 kPa overshoot on high stress level.
Unstable layer; repeated stress cycling
produced continuously increasing AP.
4
5
6
7
8/6
8/6
8/6
8/6
0.67
(28 min)
0.62
(23 min)
0.62
(12 min)
0.62
(14 min)
1.50
1.50
150
1.50
{Bags Vacuumed, 7/30/82]
6.06 On On
6.20
630
6.20
On
OH
Off
OH
Off
On
024
0.40
0.21
0.19
Bag voltage read as 0 by MOO COUP
but had appropriate bag current for
field On.
354
-------
The conclusions which can be drawn from these experiments are tentative
because the range of applicability is limited and the number of experiments
small. Nevertheless, these experiments do increase the understanding of
the effect of ESFF on dust deposits and indicate fruitful subjects for
furthe r s tudy.
BAG PERMEABILITY VARIATION
The vertical variation in "cleaned-bag" permeability evidenced by the
ESFF bag--and the lack of variation for the conventional bag—can be ex-
plained in several ways. First, the difference could be due simply to a
smaller residual deposit of fly ash at the top of the ESFF bag—the electric
field causes the dust to be collected at a lower average height in this bot-
tom-feed unit. A second explanation would be that the same amount of fly ash
was present but that it was collected and held in a more porous structure at
the top than at the bottom. Another explanation would be that the dust frac-
tionates within the bag due to the electric field and that the fraction at
the top has characteristics which differ from those of the bottom fraction.
The top-to-bottom size difference noted supports this third mechanism, but
the phenomenon occurs in conventional bags as well, and the present test
data are very limited. The low permeability at the bottom of the ESFF bag in
the as-received condition is a strong indication that more of the residual
dust is at the bottom of the ESFF bag than for the conventional bag. This
implies that most of the dust collects at the bottom of the bag. The improved
cleanability of the ESFF bag is consistent with the premise that ESFF causes
the dust to collect on the upstream side of the fabric and not penetrate as
deeply into the fabric. This combination of the first and second hypotheses--
more dust at the bottom of the bag and collected in deposits more on the
upstream side of the fabric when compared to the conventional bag—seems best
supported by the data.
NONUNIFORM DUST DEPOSITS
The extreme nonuniformity shown by the laboratory dust cake deposited in
the presence of an electric field would certainly be adequate to account for
the observed ESFF effect. However, ESFF has also been observed with dust
deposits having no clear nonuniformity. This has been true for the pilot
units, and it may be that bag cleaning and the redeposition of the cake pre-
vents the extreme pattern shown by the laboratory filter. The size distribu-
tion analysis indicates that the nonuniformity of the deposits is probably
not due to fractionation bv particle size.
INTERPARTICLE ADHESION
The experiment on increased interparticle adhesion due to the electric
field demonstrates clearly that the reduced drag of an ESFF dust cake can be
eliminated by compressing the layer. That compression removes the reduced
drag advantage for ESFF argues that this ESFF effect is more of a bulk struc-
tural effect than just nonuniform deposition. The fact that the compression
is reduced by keeping the electric field ON is an indication that electro-
static forces are increasing interparticle adhesion. The role of the
355
-------
clamping in producing the initial reduced drag—the ESFF effect itself—was
not addressed by the experiment.
REFERENCES
1. Chiang, T- K., Samuel, E. A., and Wolpert, K. E. Theoretical aspects
of pressure drop reduction in a fabric filter with charged particles.
In: Third Symposium on the Transfer and Utilization of Particulate
Control Technology, Volume III, EPA-600/9-82-005c (NTIS PB 83-149609),
July 1982. pp. 250-260.
2. Lamb, G. E. R., and Costanza, P. A. Electrical stimulation of fabric
filtration. Textile Research Journal. 47:372, 1977.
3. Lamb, G. E. R., Costanza, P. A., and O'Meara, D. J. Electrical stimula-
tion of fabric filtration, Part II. Textile Research Journal. 48:566,
1978.
4. Lamb, G. E; R., and Costanza, P. A. A low-energy electrified filter
system. Filtration and Separation. 17:319, 1980.
5. VanOsdell, D. W., Greiner, G. P., Lamb, G. E. R., and Hovis, L. S.
Electrostatic augmentation of fabric filtration. In: Third Symposium on
the Transfer and Utilization of Particulate Control Technology, Volume I,
EPA-600/9-82-005a (NTIS PB 83-149583), July 1982. pp. 35-44.
6. Furlong, D. A., Greiner, G. P., VanOsdell, D. W., and Hovis, L. S.
Electrostatic stimulation of fabric filtration—an update. Paper
(No. 82-32.2) presented at the 75th Annual Meeting of the Air Pollution
Control Association, New Orleans, Louisiana, June 20-25, 1982.
7. ASTM D737-66.
8. Smith, W. B., Gushing, K. M., and Carr, R. C. Performance of a 10 MW
fabric filter pilot plant and comparison to full-scale units. Paper
No. 82-59.6 presented at the 75th Annual Meeting of the Air Pollution
Control Association, New Orleans, Louisiana, June 20-25, 1982.
9. Rudnick, S. N., and First, M. W. Dust cake compaction in fabric filtra-
tion. Paper No. 78-62.7 presented at the 71st Annual Meeting of the
Air Pollution Control Association, Houston, Texas, June 25-30, 1978.
10. Dietz, P. W., and Melcher, J. R. Momentum transfer in electrofluidized
beds. Air Pollutants—NO and Particulate Emissions. AIChE Symposium
Series, No. 174, 74:166.
11. Dietz, P. W. Cohesive force and resistivity between electrostatically
precipitated particles. Journal of Electrostatics. 6:273, 1979.
12. Turner, J. H. EPA research in fabric filtration: annual report on
IERL-RTP in-house program. EPA-600/7-77-042 (NTIS PB 267441). U. S. En-
vironmental Protection Agency, Research Triangle Park, North Carolina,
May 1977. 39 pp.
356
-------
"HIGH VELOCITY FABRIC FILTRATION FOR INDUSTRIAL COAL-FIRED BOILERS"
by: Gary P. Greiner and Shannon Delaney
ETS, Inc.
Suite C-103, 3140 Chaparral Dr., SW
Roanoke, VA 24018
Lou S. Hovis
U.S. EPA
IERL, MD-61
Research Triangle Park, NC 27711
ABSTRACT
Two parameters which dramatically affect the technical/economic
performance of a fabric filter system are gas/cloth (G/C) ratio (or
filtering velocity) and bag life.
Under an EPA sponsored contract, ETS, Inc. has been conducting a
study of state-of-the-art and experimental fabrics and finishes in
full-scale baghouses operating on coal-fired boilers at the Kerr
Finishing Plant, Travelers Rest, South Carolina. The objective of the
study is to operate, test, and evaluate performance at G/C ratios up
to 10/1 and screen various fabrics with respect to pressure drop, col-
lection efficiency, and fabric strength characteristics necessary for
technical/economic analysis. Included in the program are some revolu-
tionary fabrics which hold promise of technological breakthroughs.
This paper will discuss evaluation methodology, performance results,
and future program plans.
This paper has been reviewed in accordance with the U. S. Environ-
mental Protection Agency's peer and administrative review policies and
approved for presentation and publication.
357
-------
INTRODUCTION
ETS, Inc., through the sponsorship of EPA's Industrial Environmental
Research Laboratory, has been conducting a pioneering program for the appli-
cation of fabric filters to industrial coal-fired boilers since 1973. Many
of the early lessons concerning this application were learned during the
pilot program and subsequent full-scale implementation. Fabric compati-
bility, cleaning modes, component requirements and specifications, achie-
vable filtering velocities, and emission capabilities all were evaluated,
defined, and submitted to the industry. This information and the success
of the full-scale systems did much to accelerate the application of fabric
filters to industrial coal-fired boilers.
Now the program, continuing its pioneering approach, is seeking to
extend the capabilities of the fabric filter system in this application
through higher filtering velocities (G/C ratios) and longer fabric life,
while maintaining required emission levels.
SITE AND EQUIPMENT DESCRIPTION
The present program is being conducted at the Kerr Finishing Plant
in Travelers Rest, South Carolina. Figure 1 shows the general site lay-
out. Baghouse No. 1 controls emissions from a 60,000 Ib /hr, Babcock &
Wilcox (B&W) stoker boiler.; Baghouse No. 2 controls emissions from a
bank of four boilers (one 30,000 Ib/hr and three 20,000 Ib/hr). The coal
used at this site is typical Eastern low sulfur coal. The operation is
cyclic depending on production requirements, and is shut down each weekend.
The current project consists of two programs. The first, referred to
as the High G/C Program, utilizes Baghouse No. 1 arid seeks to raise the
stable filtering velocity operating point to provide lower overall cost.
The second, referred to as the Fabric Evaluation Program, utilizes Baghouse
No. 2 and seeks to evaluate the bag strength characteristics versus expo-
sure time of various fabric and finish candidates that could be applicable
to both pulse jet and reverse air collectors.
HIGH G/C PROGRAM
Boiler Description
The No. 1 baghouse is applied to a Babcock & Wilcox coal-fired boiler
rated at 60,000 Ib/hr steam. A stoker grade coal (1% X 1/4 in.) is fed
to a spreader type Detroit stoker from a central weighing scale into three
stoker sections and onto a riser type dump grate. Normal maximum boiler
operation provides 52,000 Ib/hr steam at 160 psig and saturation tempera-
ture. A multi-cyclone is located on the boiler outlet, but no economizer
or fly ash reinjection is provided. The combustion air flow is provided by
a combination of forced draft, overfire air, and induced draft fans. Nor-
mal operation produces 25-40,000 acfm and 400-450°F flue gas.
(*) Readers more familiar with metric units may use the conversion factors
at the end of this paper.
358
-------
Baghouse Description
The No. 1 baghouse accommodates up to 648 bags with a nominal size
of 5 in. diameter X 104 in. long, having a total cloth area of 6800 sq ft
for a maximum filtering velocity of 5.8 fpm with all bags installed. Its
cleaning energy is supplied by 80-100 psi compressed air pulses assisted
by low-pressure reverse-flushing air. The original design had 18 separate
bag plenums and outlet dampers, but with common inlet and hopper areas.
This provided a semi-off-line cleaning mode, but not of the type seen in
today's modular collectors. It also lacked the ability to monitor and
control the flow through each section. Modifications were performed to
sectionalize the inlet and hoppers. The sectional inlet dampars, test
ports, flow monitors, and differential pressure (A?) manometers permit
the control and monitoring of the gas flow into each section. The resul-
tant system provides a high degree of cleaning mode flexibility. Cleaning
can be on-line, pulse jet only; on-line, pulse jet with varying amounts of
reverse-flush assist; or off-line variations of the above. Normal opera-
tion is to take an individual cell off-line by closing the outlet damper,
and allowing a minimum volume of reverse gas to flow, primarily to block
the forward leakage flow am pulse the bags. This most closely simulates
today's modular off-line cleaning pulse jet collectors.
Fabric Description
All fabric in baghouse No. 1 is a form of Teflon fiber felt, sup-
plied by the E. I. Dupont de Nemours Company. A description of the fabrics
and bags utilized is shown .in Table 1. The prior EPA program utilized a
full house of 23 oz TefIon^felt bags. Section C, containing these ori-
ginal bags, has been retained to provide reference with past operation.
Section F contains an experimental 17 oz TefIon ^ felt bag set that is
constructed in an asymmetrical manner, such that most of the felt is on
the outside o£-sthe bag. Sections A, B, D and E contain bag sets of experi-
mental Teflon^ fiber felt of varying weights and construction. A charac-
teristic of Teflon^ is that it physically shrinks with increasing tempera-
ture. This results in not only a smaller bag,, but also a lower in-use
permeability as shown in Figure 2. Heat setting at sufficient time and
temperature will minimize this problem; however, these characteristics
must be accounted for in Jthe initial fabric and bag specifications. Note
that the original Teflon^bags were not oversized as much as the newer
bags (4-7/8 in dia. versus 5-1/4 in.). This was due to a lack of under-
standing of the actual shrinkage rate at the time (in 1976) that these
bags were constructed. Unfortunately, all of these fabric styles were
not obtainable at the beginning of the test program; they were installed
as they became available. The installation dates are included in Table 1.
Dupont TFE Fluorocarbon fiber
359
-------
TABLE 1. BAG AND FABRIC DESCRIPTION
Desdrlption
Baghouse
Section3
A
B
C X'
D
E
F
Fabric^
(Teflon^)
Exp. (Yl)
Exp. (X)
: 2363 (Std)
Exp. (Y2)
Exp. (X)
Exp . (Asym)
Wt.
oz/yd
21
23
27
19
23
17
Perm
fpm
36
21
27
42
21
59
Mullen
Burst
psi
354
340
306
324
340
231
Tensile St.
Ibs/in Width
Warp
87
88
115
84
88
83
Fill
78
70.5
93
74
70.5
61
Bag Size
dia. X length,
in.
5% X 108
5% X 108
4-7/8 X 108
5k X 108
5% X 108
5k X 108
Initial
Installation
Date
10/81
07/81
01/77
10/81
07/81
05/81
Location of section shown in Figure 4.
-------
TECHNICAL OBJECTIVES
The program's primary objective was to develop methods of raising
the stable filtering velocity operating level and, after demonstrating
that stability, detine the effect on drag, efficiency, and bap life.
The mechanisms evaluated for raising the G/C ratio were:
Cleaning energy; i.e., pulse jet versus reverse air
versus combinations;
Cleaning mode - on-line versus off-line;
Cleaning frequency; and
Fabric type and style.
PARAMETERS MONITORED AND APPROACH
The following describes the evaluation methods employed and the
parameters monitored.
Section Drag
The parameter used to define stable operation was average section
drag, which is the average tube sheet A? during a filtering cycle divided
by the section G/C:
S (Avg) = AP (Avg)
G/C
A section was defined as having short term stability if no consis-
tent upward movement in the average drag was observed over a minimum 2-3
day evaluation period.
The three cells of each section have pressure taps, so all 18 cells
of the baghouse may be monitored for pressure drop. Generally, the middle
section is monitored, and the AP is registered on a 15 in. W.G. Magnehe-
lic gauge. There is one Magnehelic for each section. The G/C ratio is
monitored by an Annubar located at each section inlet, and the flow is
registered on an incline manometer. The instruments are located on a
panel at the inlet test level.
The initial evaluation approach was to establish the cleaning vari-
able to be tested, adjust each section inlet damper to produce equal sec-
tion G/C, and monitor section AP. A method which proved more reliable
was to adjust for equal section AP and monitor section flows. Since the
total gas volume fluctuated with steam load demand, the average G/C typi-
cally varied between 5 and 8. The drag values were recorded several times
each day. Each condition was evaluated long enough to establish only short
term stability or instability. At various times during, the test program,
snapshots of AP versus G/C were obtained by varying the G/C from 6 to 10
for short periods and monitoring AP. This aided in verifying long term
data, and also gave a fabric drag versus exposure time profile.
Efficiency
Collection efficiency was to be determined at regular intervals dur-
ing the program. Due to the common outlet, testing of individual sections
361
-------
requires taking all other sections off-line. This has proved somewhat
difficult and provided questionable results. Due to budget constraints,
additional testing was deferred until optimum high G/C operation is
achieved and additional funds are available. No individual fabric emis-
sion testing has been done. Both baghouses have been tested and have met
compliance.
Fabric Characteristics
Bags were periodically removed from the baghouse for laboratory
testing. Mullen Burst was monitored to define fabric strength retention.
Permeability was monitored to define retention of flow characteristics.
RESULTS AND CONCLUSIONS
Time and funds dictated a fairly rapid initial evaluation of the
magnitude of the effect of the available options. Having the fabrics
enter the evaluation at different times further complicated the evalua-
tion; however, certain trends were consistently found.
Effect of Cleaning Energy on Fabric Drag
The original baghouse utilized two 1% in. diameter pulse jet valves
per cell to pulse 18 bags each. In order to achieve the higher G/C de-
sired without incurring high damper pressure drop losses, bags were re-
moved as shown in Figure 3. Since all active bags were now located in
four rows per cell, the cleaning energy supplied to each bag could easily
be increased by 50% by removing one blow pipe per valve. The operating
mode with four pipes is referred to as the High Pulse Jet mode, and with
six pipes, the Medium Pulse Jet mode. A much lower level of cleaning
energy was obtained by eliminating the pulse jet and cleaning with reverse
air only.
Table 2 shows the relative drag values observed during the three
cleaning energy levels. A significant effect was seen on all fabric types.
TABLE 2. EFFECT OF CLEANING ENERGY ON FABRIC DRAG
Typical Drag
Fabric Type _ R.A. Only Med. P.J. High P.J.
Std. Teflon IS
Asym. Teflon*
Exp. Teflon^5
L Unstable
I) Unstable
' 1.1
1.5
1.2
0.8
1.0
0.8
0.5
The experimental felts are grouped as shown, since no significant
difference was observed between styles during this experiment. Note that
both the standard and asymmetrical felts did not produce a stable opera-
ting point with reverse air only. This may be due to their longer bag set
exposure or differences in construction.
Effect of Cleaning Mode on Fabric Drag
Off-line cleaning is accomplished by activating a cell's outlet damper,
which closes to block the forward gas flow, pulsing all the bags in that
cell at one time, and maintaining the cell off-line for 15 seconds before
reopening the outlet damper and resuming normal flow. On-line cleaning
362
-------
was established by allowing the pulse jet cleaning to continue, but
deactivating the outlet dampers from the cleaning cycle. High energy
cleaning was employed throughout this experiment.
Table 3 shows the relative drag values observed during the two
cleaning modes and at high and low filtering velocities.
TABLE 3. EFFECT OF CLEANING MODE ON FABRIC DRAG
G/C Ratio
Cleaning Mode
Fabric Type/^
Std. Teflon^k
Asym. Teflon-Sj
Exp. Teflon^
Off -Line
1.0
0.8
0.5
Typical Drag
On-Line Off-Line
1.0 1.0
1.0 0.8
1.0 0.5
>6
On-Line
1.5 (Unstable)
1.5 (Unstable)
1.2 (Unstable)
This evaluation confirmed past operating experience at Kerr in that
stable operation at high G/C and with on-line cleaning is not possible.
The low G/C evaluation was not run over a long enough time to determine
if long term stability could be maintained. The dramatic effect on drag
is assumed to be due to the reduction of re-entrainment in the off-line
mode.
Effect of Cleaning Frequency on Fabric Drag
Theoretically, one tries to establish cleaning frequenty at the mini-
mum level that produces a reasonable stable pressure drop. This then
should minimize compressed air usage and extend bag, pulse-Aet, and damper
component life. Prior experience with the standard Teflon^ felt and asy-
mmetric felt was that a 6-minute cleaning frequency is required with the
medium cleaning energy level;" most data with those fabrics are at that
cleaning frequency. Initial evaluation of the new experimental fabrics
indicated that a lower cleaning frequency could be used: most data on
those fabrics was collected at the 20-minute cleaning frequency. With all
sections employing high energy cleaning, the results of Table 4 were
obtained.
TABLE 4. EFFECT OF CLEANING FREQUENCY ON FABRIC DRAG
Cleaning Frequency
Typical Drag
6 Min. 20 Min.
Fabric Type/gv
Std. TeflonWjv
Asym. Teflon-S^
Exp. Teflon^
1.0
0.8
0.5
^^•••fc.*-**^""^*— ^»»i»™*^"^B-™»^"^"""^^^^-"
1.2
1.0
0.5
Note that the fabric drag was not reduced by increasing the cleaning
frequency on the experimental felts. The pressure drop rise between clean-
ing has been both very low and extremely linear, indicating the possibi-
lity of further extending the cleaning frequency. The short term increase
in fabric drag of the standard and asymmetric felts under high cleaning
363
-------
energy and 20-minute cleaning was quite modest. Note that sufficient test
duration was not allowed to verify long term stability and that the more
rapid rise of pressure between cleaning indicates that the optimum clean-
ing frequency may be somewhere between 6 and 20 minutes.
Effect of Fabric Construction on Fabric Drag
Due to the variables of bag history, cleaning energyr^fnd cleaning
cycles employed, direct comparisons of the various Teflon ^-'felt con-^
structions have been difficult. Earlier tables have shown the Teflon^
styles for reference to indicate the relative effect of the cleaning vari-
able under study. The indicated trends are both real and significant.
Table 5 provides relative drag values for the four major felt con-
structions now under evaluation. Experimental construction _X differs
from construction Y_ in that Y^ was designed to have a 50% higher initial
permeability. Two different weight bags of construction Y_ were fabricated
(styles No. 402 and 403). No significant difference was seen between the
two in the drag results.
All of the prior evaluations were run with the system inlet having a
high degree of dilution due to ambient air in-leakage in the boiler stack,
expansion joints, and emergency dilution damper. This produced an inlet
grain loading of approximately 0.2 gr/acf. In-leakage was minimized
through most of these areas. While this has reduced the average G/C, the
boiler load fluctuates sufficiently over the course of a day that the same
G/C range of 6 to 10 can still be evaluated with the reduced cloth area.
Also, boiler operation has provided some variation in grain loading with
some data being obtained at 0.5 to 0.8 gr/acf. With this, the standard
fabric drags are now more in line with industrial experience at 4-5 G/C.
The short term A P versus G/C evaluations show a constant drag over this
range.
TABLE 5. EFFECT OF FABRIC TYPE ON FABRIC DRAG (short term evaluation)
Inlet Loading (gr/acf)
Std. Felt
Asym. Felt
Exp. Construction X
Exp. Construction Y
0.2 - 0
1.0
0.8
0.5
0.5
Typical
.3
Drag
0.5 - 0.
1.6
1.4
1.1
0.8
8
The drag data previously shown were a result of short term evalua-
tions which did not allow full stabilization of fabric condition and were
intended only for relative comparisons of fabric types. From May to
August 1982, an evaluation at fixed conditions revealed the results con-
tained in Table 6. While these drags are higher than previously reported,
they were consistent under normal boiler operation and represent what
might typically be expected. To further evaluate the drag values, the
system was operated at relatively constant conditions and modes over an
extended period of time. The baghouse was operated with the stack cap
bypass damper closed, the Vortex damper control on automatic off-line
364
-------
cleaning, a cleaning frequency of 6 minutes, and a typical grain loading
of 0.5 to 0.6 gr/acf. This evaluation produced the data of Table 6.
TABLE 6. EFFECT OF FABRIC TYPE ON FABRIC DRAG (long term evaluation)
Fabric Type
Standard Felt
Asym. Felt*
Exp.
Exp.
Exp .
Teflon
Teflon
Teflon
Const.
Const .
Const.
X
Yl
Y2
Typical
1
0
0
1
.6 -
.95-
.9 -
.1 -
1
1
0
1
Drag
.7
.0
.95
.2
CONCLUSIONS AS TO HIGH G/C CAPABILITY
The primary thrust of the evaluation to date has been to identify
parameters that produce lower drag values or improve pressure drop stabi-
lity so that higher G/C levels may be achieved with reasonable and stable
pressure drops. By necessity, the G/C levels have varied with the boiler
load, and specific evaluations have been short. Future program activity
is to select the optimum system and fabric parameters and evaluate them
over a longer period.
The results to date indicate that the optimum system operation is
with higher energy, off-line pulse jet-cleaning, a cleaning frequency of
6-20 minutes, and experimental Teflon^ felt Construction Y. Short term
stability up to a G/C of 10 has been demonstrated. During the long term
evaluation, the system was operated in a manner that allowed the section
pressure drops to self balance and the section gas volume flows to be
dictated by their fabric drags. This resulted in the experimental felt
Construction Y running typically at 7-9 G/C with no instability observed
and a drag typically less than 1.0 even at the highest inlet grain load-
ings during high boiler loads. Mullen Burst tests on samples from the
above test bags showed only a modest drop in strength for all fabrics
except experimental Construction Yl.
A longer term evaluation of the above conditions is recommended and,
if successful, could double the present filtering velocity now common in
this application.
FABRIC EVALUATION PROGRAM
Boiler Description
The No. 2 baghouse is applied to a bank of four coal-fired boilers
with a common manifold supplying a natural draft stack. The first three
boilers were built in 1928 by Walsh & Weidner Boiler Company and are
equipped with Combustion Engineering underfeed retort spreader type
* Note: Asymmetric felt was not included in this evaluation.
365
-------
stokers and side dump grates. The boilers are rated at 22,500 Ib/hr
steam at 250 psig and saturation temperature. The fourth boiler was
built by Combustion Engineering in 1941 and is equipped with Combustion
Engineering underfeed retort spreader type stoker and side dump grates.
It is rated at 130,000 Ib/hr steam at 250 psig and saturation temperature.
All four boilers are provided with force draft fans., and generally two
boilers at a time are operated. Outlet flue gas is directed to the bag-
house from the manifold. Normal operation produces 40,000 acfm and 400 -
450°F flue gas.
Baghouse Description
The No. 2 baghouse is identical to No. 1, except its cleaning method
is reverse air plenum pulse instead of pulse jet. The collector incor-
porates 648 bags with a nominal size of 5 in. diameter and 104 in. long,
having a total cloth area of 6800 sq ft. Normal filtering velocity is
5-6 fpm. Its cleaning energy is supplied by a 4000 acfm fan at 10 in.
W.6. pressure through cell outlet poppet dampers that simultaneously stop
the normal forward gas flow through one cell, and introduce the reverse
pulse of heated ambient air. A typical cleaning cycle is 10 seconds of
reverse flushing followed by 5 seconds of delay before moving to the next
cell. This produces a bag cleaning frequency of 4.5 minutes. The house
was divided into six sections and is instrumented in the same manner as
house No. 1.
Fabric Description
The fabric being evaluated consists ojk four bag sets: three woven
glass with varying finishes and one Nomex ^ felt with an acid resistant
finish. The woven 'glass bags all have the same basic fabric construction,
one that is presently being used in both pulse jet and reverse air collec-
tors on coal-fired boilers. Three different, finishes are being evaluated
on the glass bags: an experimental Teflon Abased finish developed by
Dupont, aad two state-of-the-art finishes, Burlington's style I6L&. 10%
Teflon B^, and Burlington's style ^|?5, Acid Flex^. ThagNomex ^bag set
utilized Globe Albany's 16 oz Nomex^felt with Permagard^ finish that
has shown success in Europe.
Fabric specifications are shown in Table 7. The locations of the
various fabrics in the baghouse are shown in Figure 4. A test cell incor-
porating all fabric types was established to normalize any biased effects
of section location.
TECHNICAL OBJECTIVE
This program's primary objective was to evaluate the operational
characteristics and service life of the various fabrics while exposed to
normal operating conditions of the Kerr coal-fired boiler and baghouse
system.
366
-------
TABLE 7. WOVEN GLASS FABRIC ANALYSIS
Fiber:
Construction:
ALL FABRICS
Class
Weave 3X1 Twill
Count W44 X F24
Warp Yarn DE 37-1/0 - (Singles Construction)
Fill Yarn DE 75-1/0 - (Plied, 2 Texturized,
1 Untexturized)
Virgin Cloth Perms 90 fpm @ 0.5 in. W.G.
Fabric
Finisher
Perm, Avg.
LOI, wt%
Oz/yd2
- Treated
- Untreated
Thickness, in.
A,
Acid/-v
Flex^
Burlington
77.8
4.6
12.6
12.0
0.0144
B,
Teflon^
Burlington
56.1
9.5
14.3
12.9
0.0142
x,
Exp'tal
Clark-Swabel
47.5
9.6
13.6
12.3
0.0158
Virgin Cloth Perm zSQ fpm
367
-------
PARAMETERS MONITORED AND APPROACH
The following describes the evaluation methods employed and the
parameters followed.
Fabric Drag
Filtering velocities were maintained as equal as possible between
test sections throughout the evaluation by monitoring the inlet Annubars
and manually adjusting the section inlet dampers. Normal filtering velo-
cities were maintained in the 4-6 fpm range. Section tube sheet AP was
recorded daily to compute average fabric drag values. Fabric permeability
was monitored on bags that were periodically removed for testing.
Fabric Properties
To evaluate the effect of operational exposure on bag life, the
fabric strength characteristics were monitored. Test bags were scheduled
for removal after exposure times of 1 hour, 1 day, 1 week, 1 month, 3
months, 5 months and 7 months. Fabric testing included Tensile, Mullen
Burst, MIT Flex,.Loss on Ignition (LOI) , and visual and microscopic in-
spection.
Efficiency Testing
Mass efficiency tests by FJ?A Method 5 were scheduled to be run after
stabilization of the system. Although difficult, individual section
emission testing was scheduled to provide comparative data between fabric
types. Testing currently has been deferred due to budget constraints.
BAG CONDITIONING AND OPERATING HISTORY
Conditioning of the bags was initiated on May 14, 1981, and continued
for the next 4 weeks. The flow was maintained at the low (1.5 - 2.5) G/C
for an extended time to both obtain some data at low G/C and ensure that
system control would prevent biasing of any fabric. The flow, temperature,
and section pressure drop were carefully monitored, and the bags were in-
spected visually.
Due to intermittent boiler operation and long weekend shutdowns, 7
full days of exposure required 1 month of calendar time (early in the pro-
gram). The plant was then shut down for the month of July. After resuming
operation, the baghouse was set to operate at 4-5 fprn- Malfunctions in the
damper system of section C, the experimental Teflon^ finish, resulted in
lower filtering rates in that section for 3 weeks. Again, due to limited
plant production which reduced the boiler needs, it would take approxi-
mately 3 months to obtain the next 2 months exposure time. The 3 month
test bags were scheduled for removal on October 16. Unfortunately, a
severe boil^ upset caused a fire in the baghouse, and all the experimen-
tal Teflon^finished bags and the test bag cell were destroyed on Octec-
ber 12. The baghouse remained off-line until new experimental Teflon™
finish bags could be obtained. The evaluation was restarted on May 5,
1982.
368
-------
RESULTS AND OBSERVATIONS
The cyclic nature of the baghouse during this evaluation dramatizes
the need for fabrics that can survive many start-ups, shutdowns, and
their resultant dew point excursions. Normal boiler operation was start-
up on the 2nd shift Sunday, and shutdown on the 2nd shift Friday. Due to
production limitations, several shorter operating cycles were run. In
addition, baghouse auxiliary component malfunctions forced the baghouse
off-line at times. Needless to say, this type of application intensifies
the need for preheat and purge capabilities (which exist) and fabrics that
can resist acid condensation.
AP Evaluation
Through June 25, the plant's summer shutdown, the filtering veloci-
ties were held constant and, for all practical purposes, the AP levels
of all sections were the same. Following the shutdown, Section C (exp.
finish) ran at lower G/C ratios and higher AP levels as a result of
cleaning system problems. These problems were intermittent, recurring
throughout August. After correction, its AP was comparable to that of
the other section.
No significant differences in fabric drag could be seen during the
evaluation between sections. Average drags ranged from 1.5 to 2.0,
reflecting the low energy cleaning system being employed.
Fabric Strength Tests
Results to date are plotted on Figures 5 through 8 and Table 8. The
test plan calls for the removal of three bags of each type after 1 hour,
24 hours, 1 week, 1 month, 3 months, 5 months, and 7 months of exposure
time. After removal, the top, middle, and bottom of each bag is tested
for its strength characteristics. The tests include permeability, tensile
strength, flex strength, and burst strength. Tests have been completed
through 7 months of exposure. In addition, all but the experimental bags
have seen over 6 months additional off-line exposure. It has been deter-
mined that there is no significant difference in the strength characteris-
tics of the top, middle, and bottom of any of the bags. Note that in-
creases in the value of a given parameter versus on-line time are not
expected. These increases are attributable to variations in the initial
strength characteristics of the bags, finish protection effectiveness, bag
location, or (in the case of permeability) handling.
The new, 1 hour, 24 hour, 1 week, and 1 month Acid Flex ^ Teflon B,v-'
and Nomex^ data were obtained before the October 12 fire. The 3-month
data were obtained from bags off-line for 6 months and were obtained in
May 1982, after restart-up. All the experimental Teflon data were obtained
on replacements for the original experimental Teflon bags. This evalua-
tion began in May 1982, and the replacement bags have strength characteris-
tics similar to those of the original bags.
Permeability
As previously stated with regard to the high G/C program, fabric
permeabilities were taken in the laboratory and in-situ cake conditions
369
-------
have been altered by removal from the baghouse (removal from cages) and
transferred to the laboratory. Again, this type of evaluation relies on
examining the relativity of the data rather than the absolute values.
As can be seen in Figure 9, the greatest decrease in the permeabili-
ty of all four fabrics occurs in the first hour of operation. Another
substantial decrease takes place from 1 hour to 24 hours. The Acid
Flex>-}bag set had the highest initial permeability, starting 35% higher
than the other woven glass fabrics. However, it drops teethe level of
the other fabrics (25-30 fpm) within 1 hour. The Nomex ^ fabric starts
with the lowest value; however, it exhibits the least decrease in perme-
ability, and is higher than the other fabrics up to 3 months. All the
fabrics are in the range of 4 to 10 fpm within 1 month. Dirty as received
permeability values of 4 to 10 are normal for fabrics operating at 4-8 in.
W.G. pressure drops at 4-6/1 G/C. At 3 months, all the woven fabrics
have roughly the same permeability. From 5 months on, all fabrics are
essentially the same and reflect the sameness in fabric drag seen in the
field. Note that no severe blinding has occurred in any of the fabrics
through 7 months.
Flex Strength
Figure. 5 is a plot of the flex strengths for the woven glass fabrics.
The Nomex ^ felt is considerably higher than any of the woven glass fab-
rics (initial value around 47,000).
The Nomex ^ showed its greatest decrease in flex strength between 24
hours and 1 week (from 42,000 to 9,000). The Nomex bag set showed dra-
matic loss of strength between 5 and 7 months, and several field failures
occurred during this period. The Teflon B*^fabric's initial superiority
was eliminated within 1 day, and this may be of significance to the
finisher. Also, the Teflon B^SJhad the lowest flex value by 1 month and
continued Xo have the lowest-at 5 months, although only slightly less than
Acid Flex vJ. The Acid Flex^fabric showed its greatest decrease to be
between 24 hours and 1 week. The"experimental Teflon^finished bags,
although starting at a lower value, have shown much less strength deterio-
ration, and bags tested at 5 and 7 months show virtually no reduction in
flex strength.
When the 1 month old bags were removed for testing, the samples loca-
ted on the outside walls were^ry wet. Table 8 is a summary of the ^£1 ex
results for the wet Acid Flex®, Teflon 8®, and experimental Teflon^
fabrics. The experimental. Teflon© showed no significant decrease in flex
strength. The Acid Flexv&) showed a significant loss of flex strength only
in the middle of the bag, and the Teflon BVS) showed virtually a total loss
of flex strength. These are good indications of the finish's ability to
encapsulate the fiber and resist moisture and acid attack.
Mullen Burst
Figure 6 shows the Nomex ^ to have: the lowest initial burst value,
as is typical of felts; a slow decline in strength; stability between 1 and
5 months; and raoid deterioration between 5 and 7 months. Of the glass fab-
rics, Acid Flex^and Teflon B^ show similar burst characteristics with
over 40% strength loss at 1 month, and only a slight decline to 7 months.
370
-------
, TABLE 8. WET BAG FLEX DATA FOR
1-MONTH OLD BAGS
Fabric
Number Of Flexes
Acid Flex^
Teflon B^
Tog
811
1384
31
26
Middle
37
34
9
14
Bottom
467
588
43
24
Experimental
Teflon
1025
1587
1242
912
614
820
The experimental TefIon^started with the same burst value as the other
woven fabrics (as expected, since all were fabricated with the same basic
fabric construction), but has shown superior retention of strength.
Tensile Strength
The Nomex^-^fabric shows noticeably lower (typical of felts) tensile
strength (Figures 7 and 8) than the Glass fabrics, but significant deterio-
ration after 5 months. The Glass fabrics show similar trends to each
other, with the greatest-decrease occurring between 24 hours and 1 week.
The experimental Teflon^ again has the highest value after 3 months.
General Observations
The study, by design, represents a short evaluation period when put
in the context by seeking greater than 2 years of bag life. The data are
unique in that they are the first documented evaluation of how rapidly
strength characteristics are lost (i.e., hours, days).
Fabric filters typically show an early strength loss of 50%, at which
time^they reach a plateau until the end of the bag life cycle. The Acid
Flex ^ and Teflon B ^ appear to be at this level within 1 week, and are
reasonably stable through 7 months on-line exposure plus the extended
off-line. The experimental Teflon B^ finish was designed to improve fab-
ric strength retention and has been successful through 7 months, indica-
ting that its finish may be protecting the fabric better.
The Nomex^bag set with its Permagard^6 finish showed superiority
371
-------
over the 1 month life previously obtained on untreated Nomex^-'bur
insufficient life to be practical in this application.
CONVERSION FACTORS
Readers more familiar with metric units may use the following
factors to convert to that system:
Non-metric Times Yields Metric
°F 5/9 (°F-32) °C
ft, 30.48 cm,
ft^ 0.09 m
ft 28.32 L (liter)
gr 0.06 g
in._ 2.54 cm_
in. 6.45 cm
Ib 0.45 kg
oz 28.35 g
yd 91.44 cm
372
-------
o
25
g
BOILER NO. 1
20.000 LB/HR
BOT.LKR NO. 2
20,000 LB/HR
ROH.KR NO. 3
20,000 LB/HR
nOTI.KR NO. 4
30,000 LB/HR
u
Bnii.ER NO. 5
60.000 LB/HR
g
CO
CQ
ts
i
u
Figure 1. Kerr site layout.
373
-------
60
50
00
5
e
P.
30
EM
3
PQ
20
o,.
B&E
Section A X
Section D Yl
Sections B A K Y2
Section C XT2363
Section F XT1661
10
0
100 200
TEMPERATURE, °F
300
400
Figure 2. Permeability vs. temperature.
374
-------
COVKS
PLATE
o
o
o
o o
o o
BLOWPIPE
PULSE JET
VALVE
u
"~ BAG
o
'
1
1
t
1
1
' ' 1
1 :
1 ;
I
.
i i i
i '
I •
i i
i I . _
. -i r ii •- •-
Flp,uri.> 3. High gas-to-cloth cell bag layout.
375
-------
BAG POSITION
SECTION
INLET
>
(f§) BURLI
14 OZ
3
© (iD
tfGTON STYLE
. WOVEN GLAS
2
© ©
»54
5-EXP .TEFLON
1 &0$
3
BURLINGT
14 OZ. WOV
2
3N STYLE 454
2N GLASS-TEF
1
•1615
.ON B
© ©
TEST
CELL
3
MISC.
FABRIC
2
3
BURL INC
14 OZ. V
2
TON STYLE 45
OVEN GLASS -A
MISC.
FABRIC
1
1
i-1625
]ID FLEX
MISC.
FABRIC
3
3
14 OZ
TEST
CELL
2
2
GLOBE ALBA1
NOMEX.FELT
MISC.
FABRIC
1
1
fY
-PERMAGARD
CELL
Figure 4. Kerr Baghouse - cell schematic - system 2.
-------
42.166
ui
-vj
4000 -
-J 3000 -
IOC -
New
7,616
o Teflon B
\
Notaex
- Teflon
1 i!r
24 Kr 1 Wk
ON-LINE TIME
1 y.
Mo
Mo
7 Mo
5. MIT Flex (fill) vs, on-line time
Kerr baghouse No. 2
fabric evaluation program.
-------
600 -,
to
•vl
00
CO
(X
1
«
2:
450-
300 -
150 -
New
1 Hr
• Acid Flex
o Teflon B
Nomex
Exp. Teflon
5 Mo
7 Mo
Figure 6. Mullen Burst vs. on-line time
Kerr baghouse No. 2
fabric evaluation program.
-------
500 -i
400 -
c
•H
<
OS
H
V3
CO —
-j i-:
\o i—
tr.
300 -
200 -
100 -*
New
3. Hr
24 Hr I Wk
ON-LINE TIM::
I Mo
3 Mo
• Acid Flex
O Teflon B
Nomex
Exp. Teflon
5 Mo
Figure 7. Tensile strength (warp) vs. on-line time
Kerr baghouse No. 2
fabric evaluation program.
7 Mo
-------
00
o
4-1
•u
•H
c
•H
400 _
o 300 _
fc-
C
ss
S-
u:
200
IOC -
• Acid Flex
O Teflon B
Noaex
Exp. Teflon
New
1 Hr
Figure 8. Tensile strength (fill) vs. on-line time
Kerr baghouse No. 2
fabric evaluation program.
-------
100 J
80 _
6
o-
60 .
cc
oo
20
10
5
• Acid Flex
O Teflon B
V Nomex
Exp. Teflon
New
1 Hr
24 Hr 1 Wk
1 Mo
3 Mo
5 Mo
7 Mo
ONLINE TIXZ
Figure 9. As received permeability vs. on-line time
Kerr baghouse No, 2,
fabric evaluation program.
-------
OPTIMIZING THE LOCATION OF ANTI-COLLAPSE RINGS
IN FABRIC BAGS
By: John G. Musgrove
Bechtel Power Corporation
P. 0. Box 2166
Houston, Texas 77252-2166
ABSTRACT
The number and placement of anti-collapse rings in fabric filter bags are
frequently left to the bag manufacturers. Bag manufacturers do not analyze
operating requirements for determination of ring location but merely sub-
divide the bag into sections of approximately equal length. Bags whose
section lengths are equal are suspected during reverse air cleaning of having
greater collapse (deflection) in lower sections than in higher sections.
An analysis has been developed to predict the amount of collapse (deflec-
tion) in all sections of a bag undergoing reverse air cleaning. The
methodology uses the catenary principle of a uniformly loaded suspension
cable to determine deflection. A computer program utilizing this methodology
has been developed to analyze the forces on a warp thread in a bag and to
determine the location of anti-collapse rings to assure equal bag collapse in
all sections. A companion program has been developed to predict bag section
collapse for bags with established ring locations.
382
-------
ACKNOWLEDGEMENTS
Special thanks is given to Research-Cottrell for providing the laboratory
space and support personnel for conducting the full scale test program:
Stephen Zierak, Southwest Regional Sales Manager; Stefan Negrea, Vice
President of Engineering for the Air Pollution Control Division; Dennis
Helfritch, Program Manager with Cottrell Environmental Sciences; and Chet
Gorski and Mike Adams, laboratory technicians with Cottrell Environmental
Sciences.
Thanks is also given to Houston Lighting & Power Company for providing spare
filter bags for use in the test program: Soli Irani, Program Manager.
The assistance provided by Research-Cottrell and Houston Lighting & Power
Company should in no way constitute their endorsement or agreement with the
scope or results of this study. The scope and conclusions are solely those
of the author.
INTRODUCTION
Fabric filter bags for reverse-air-cleaned baghouses, such as have been
installed on utility steam electric generating stations, have anti-collapse
rings located at intervals along their lengths to prevent excessive collapse
during reverse air cleaning, Figure 1.
During flue gas cleaning or forward flow, the bag flexes outward uniformly,
being restrained by the circumferencial stresses in the fill fibers. During
the reverse-air cleaning mode, the bag collapses in a multi-lobe pattern,
Figure 2. The fill fibers are no longer under tension and the stresses are
taken by the warp fibers. The fibers at the inside diameter of the lobes
undergo tension and form parabolic curves, Figure 3. Thus, the warp fibers
form a family of hyperbolic curves, which is the basis for this study. A
single section can therefore be presumed to form a hyperbolic shape, Figure A.
The deflection caused by reverse air flow in each section is believed to
increase from the top of the bag to the bottom of the bag as the tension
decreases (1). If the original bag tension is inadequate and the ash loading
on the inside of the bags is high, the deflection in the bottom section may
be so great as to cause "pancaking", wherein opposite sides of the bag touch
each other, Figure 5. This excessive deflection decreases bag life through
excessive bag flexures and through abrasion of the bag by ash falling down
through the bag.
The parabolic shape each fiber assumes under the uniform loading of the
reverse air flow is identical to the catenary shape taken by a suspension
cable under uniform loading. The purpose of this study is to develop a
mathematical model to predict the amount of bag deflection at the initiation
of reverse air cleaning so that the anti-collapse rings may be spaced so as
to permit equal deflection of each bag section during reverse air cleaning.
This study examines the catenary principle and applies it to the warp fibers
of filter bags.
383
-------
Effective Bog Rodluo, Inohee
Top Cuff « 2.5 in.
Section 1 - 56.13 in.
Cuff 1 " 2.0 in.
Section 2 " 54.5 in.
Cuff 2 - 2.0 in.
Section 3 » 51.75 in.
Cuff 3 * 2.0 in.
Section 4 - 49.75 in.
Cuff 4 - 2.0 in.
Section 5 " 48 in.
Cuff 5 •» 2.0 in.
Section 6 - 45.75 in.
Cuff 6 - 2.0 in.
Section 7 - 43.5 in.
Bottom
Cuff • 3.5 in.
-d
X.
8 £
84
BAG TYPE II
Figure 1. Filter bag with anti-collapse
rings.
Figure 2. Plan view of collapsed bag section.
0795h/0074h 394
-------
Effective Boq Radius. InoUa
uuu
oo
en
L 1 " 57 in.
L 2 - 58.2
L 3 - 53.5 in.
L 4 - 51.5 in.
L 5 - 49.7 in.
L 6 - 47.5 in.
L 7 • 44.3 in.
•"• •
*•:•
3 -
in
Figure 3. Section view of collapsed bag.
Figure 4. Downward view of collapsed bag section.
-------
Effaotive Boa
r3r2?
L 1 « 57 In.
L 2 « 56.2 In.
L 3 "53.5 in.
L 4 - 51.5 in.
L5-49.7 in.
L 6 - 47.5 in.
L 7 " 44.3 in.
BAG TYPE II
Figure 5. Section view of "pancaking" bag section.
CATENARY SUSPENSION
The equations for determining stress and sag of a single cable under
uniformly distributed load (2) are derived from the configuration of Figure
6. The forces and dimensions are as follows:
No load (initial configuration):
o I = horizontal distance between supports
o L = developed length of cable
o f = sag of cable
Loaded Cable:
o T = tension in cable at support caused by superimposed
loads
386
-------
o q. = weight of cable per unit length (assumed uniform)
o a = angle between horizontal and tangent to cable at
support
o V = vertical component of T at support
o H = horizontal component of T at support
o Af = increased sag in cable due to superimposed loads
.\
<_...
q q q
Figure 6. Catenary suspension force diagram.
The operational equations are:
L - 1*[ 1 + (8/3)*(f2 /I2)], in.
T = [(q*!2/8*f)]* SQR [1 + 16*(f2/A2)],oz.
tan a = 4*(f +-Af)/JL
V = T sin (a), oz.
H = T cos (a), oz.
(1)
(2)
(3)
(4)
(5)
The assumptions for these equations are that the initial sag, f, and distance
between supports, I are known, that the cable loading, q, is uniform, and
that the unloaded cable weight and resulting tension are negligible compared
to those of the loaded cable.
387
-------
INITIAL MATHEMATICAL MODEL
CATENARY FORCES
The approach to determining fabric deflection is to assume the warp fibers
undergo uniformly distributed loading, but in a vertical plane rather than a
horizontal plane, Figure 7. The fiber is assumed to have a diameter of .0227
inches and therefore an area/inch of .0227 in^/in. The fabric is assumed
to have a count of 44 fibers per inch.
wl+w2
FCI+1)
Figure 7. Bag warp fiber catenary force diagram.
388
-------
The forces that translate from the suspension cable are:
o 1 = L = vertical distance between rings, in.
o L = S = developed length of collapsed fiber, in.
o f = d = deflection of fiber from ring diameter, in.
o T = F = tension force in fiber, oz.
o q = p = force of reverse air, oz/in2/in.
o a = a = angle between vertical and tangent to fiber
o V = FJJ = horizontal component of F at ring, oz.
o H = Fy = vertical component of F at ring, oz.
Additional forces to consider include:
o Wl = unit weight of cloth, oz/in.
o W2 = unit weight of ash, oz/in.
o W3 = unit weight of ring and ring cuff, oz.
Equations 1 through 5 can thus be rewritten as:
S = L t 1 + ( 8/3)*(d2/L2)] (6)
F = [(p*L2)/(8*d )] * SQR [ 1 + 16*(d2/L2)] (7)
tan a = 4 * d/L (8)
FH - F * Cos (a) (9)
FV = F * Sin (a) = p * L2/2 (10)
Additional equations developed here include:
FVN - FVN+1 + W3 + SN *(W1 + W2) (11)
Where N denotes the bag section undergoing analysis.
The tension in the thread is thus:
TN = SQR [(FHN)2 + (Fvn)2] (12)
BAGHOUSE FORCES
The tension in the warp fiber is further influenced by the thermal expansion
of the baghouse casing, the reverse air force in the bag cap and the weight
of the accumulated ash. The tension due to thermal expansion is:
T4 = (T8-T6) * L' * K9 * (CE2 - CEi) (13)
389
-------
Where:
T8 = operating temperature, F
T6 = ambient temperature during bag hanging, F
L* » casing length ( —L + 16), in.
K9 = tensioning spring constant, Ib/in.
" Coefficient of thermal expansion of steel, in/in/F
- Coefficient of thermal expansion of glass (3), in/in/F
The relaxation of tension due to the reverse air force on the cap is:
T5 = A * .5781 * v *(R9)2/16 (14)
Where:
A = reverse air pressure, in. wg
R9 = bag radius, in.
The weight of ash per inch of fiber is:
W6 - 80 * Tl * * * D9 * (L9-6)/1728 (15)
Where:
Tl » thickness of ash layer, in.
D9 = bag diameter, in.
L9-6 = effective bag length (less ring cuffs)
The total fabric tension just prior to reverse air cleaning is:
T3 = T + T4 + T2 (16)
Where:
T2 « upward force on bag cap
The total fabric tension just after start of reverse air cleaning is:
T9 - T + T4 - T5 (17)
The tension created by the weight of the ash itself is not included, in T3 and
T9 because it is a force that acts solely downward from within the bag. That
is, since the support spring will deflect and relax the tension at the bottom
as the tension at the top increases, the net effect within the bag is
negligible and assumed to be zero.
390
-------
OPERATING EQUATIONS
The four fundamental equations for analysis then become
FH = F2(I) = p* [L (I)]2/2 (18)
N
FV = F1(I) = T9 * (16/1660)-K1 * (Wl + W2) - (19)
N W3 * (1-1)
SQR [(F1(I)) + (F2(I))] (20)
D(I) = p* L(I)2/(4 * SQR [(2* F(I))2- (p * L(I))2] (21)
The initial modeling effort, in using the catenary principles, had no method
for relating tension prior to uniform loading of the fiber to tension
following such loading since the catenary principal assumed initial tension
to be negligible. Development of this relationship was made by the
measurement of actual deflection under loading of two different fabric filter
bags.
LABORATORY TESTING
Research-Cot trell was kind enough to provide a high-bay test structure and
technical support for measuring the deflections of a filter bag undergoing
reverse air pressure. The bags were provided from spares purchased by
Houston Lighting & Power Company as part of a separate pilot baghouse test
program. The pilot test program was performed under a contract with Cottrell
Environmental Sciences.
LABORATORY TEST FACILITY
The test facility was erected in Research-Co ttrell's forty-foot high
precipitator electrode test bay. Individual fabric bags, each 30 '-6" long
and 11.5" in diameter were hung by a spring scale from the upper bay
supports. The bottom of the bags were attached to a single thimble which was
restrained by chains linked to earth anchors. The thimble was sealed at the
bottom and connected to a vacuum pump. An inclined manometer permitted
measurement of the differential pressure across the bag. A fixed scale was
placed adjacent to the thimble to indicate any bag movement due to
application of the air pressure, and a tape measure was hung beside the bag,
attached only at the top, to indicate the amount of bag shortening due to
application of reverse air pressure. A standard workman's scaffold was
erected around the bag and boards placed on it to permit convenient reading
of the support scale load and the convenient measurement of the collapsed
diameter of each section.
To assure attainment of reverse air pressures of up to 1 inch water guage
(in. wg) a tube of plastic film (5 mils thick) was installed over the bag as
it was hung to simulate the resistance of the dust cake. The weight of the
ash was simulated by hanging masses of approximately equal weight onto the
rings to simulate a uniform loading along the bag length.
391
-------
Bag deflection was measured by use of a standard outside diameter caliper and
a measuring rule to obtain the average minimum collapsed diameter of the bag.
Test Program
A total of ten test runs were conducted, as indicated in Table 1.
TABLE 1 - TEST PARAMETERS
Run, Bag Type, R.A. Pressure, Initial Tension, Ash Load,
Number Number in. wg pounds pounds
1 2 0.5 100 0
2 2 1.0 98 0
3 2 0.5 63 0
42 1.0 65 0
5 2 0.5 51 65.25
6 2 1.0 51 65.25
7 2 0.5 85 65.25
8 2 0.85 72 65.25
9 3 1.0 63 65.25
10 3 1.0 95 65.25
The data from Runs 1 through 4 indicated greater deflection at the top of the
bag than at the bottom of it. These data were subsequently discarded as being
the result of anomolies due to light loading (no ash) and the influence of the
plastic bag.
Test Runs 5 through 10 produced acceptable results - deflections increasing
down the bag - and were used in the subsequent analysis. Appendix I presents
the summary of test results.
DATA CORRELATION
TENSIONING DUE TO REVERSE AIR
The test program clearly identified a factor missing from equation 17, the
tension created by the collapse and shortening of the bag sections. In order
to calculate this tension, the resulting deflections were assumed to be solely
restrained by this tension. Figure 8 presents the relative distance between
rings as a percentage of the fiber length, a strong correlation. The total
bag reduction in length was apportioned to each section in proportion to its
unloaded length. Figure 9 presents the average reduced lengths for each test
reverse air pressure and the average for all the tests.
392
-------
LEGEND
RUN 15
BAG TYPE 2 8s
RUN IB
BAG TYPE 2
RUN 17
BAG TYPE 2
RUN IB
BAG TYPE 2
- -.
RUNI9
BAG TYPE 3
RUN #10
BAG TYPE 3
1O/3/B2
OS
FILTER BAG REVERSE AIR DEFLECTION TESTS
COLLAPSED RING SPACING v. BAG SECTION NUMBER
I 1
i
BAG SECTION NUMBER - TOP TO BOTTOM
Figure 8. Bag section shortening due R.A. pressure.
LEGEND g
AVERAGE FOR £
ALL 6 RUNS 8s
AVERAGE FOR |
DELTA P-.5
AVERAGE FOR
DELTA P». 85
_ _ •
AVERAGE FOR |
DELTA P- 1 *
FILTER BAG REVERSE AIR DEFLECTION TESTS
COLLAPSED RING SPACING v» BAG SECTION NUMBER
1 I i 1
i
BAG SECTION NUMBER - TO? TO BOTTOM
Figure 9. Average bag section shortening due to R. A. pressure,
393
-------
The shape of the curves suggested an exponential relationship for the
percentage shortening, one minus the reduced length. Figure 10 depicts the
exponential curves for each of the three test reverse air pressures.
Unfortunately, the data of the 0.85 in. wg case do not fall between those of
0.5 in. wg and those of 1.0 in. wg., perhaps due to measurement error or
insufficient data at that test condition. A single exponential curve was
then generated, Figure 11, to approximate the shortening of the hag
sections. The relationship obtained is:
L - (1 - [(.01) * (.005 * exp ((1-1) * 0.844) + .5673)])*S
(22)
Where:
L = distance between rings as a factor of fiber length S
1 = the bag section under analysis, numbered starting at the top
LEGBO
REVERSE AIR
DELTA P-.5
REVERSE AIR
DELTA P-.85
^^^HHI^^H* • ••••MHBIBHB
REVERSE AIR
DELTA M
FILTER BAG REVERSE AIR DEFLECTION TESTS
SECTION SHORTENING UNDER DELTA P
18WQ2
BAG SECTION MMER (MINUS ONE)
Figure 10. Averaged exponential curves for bag section shortening due to R. A.
pressure.
394
-------
LEGEND
FROM 6 TEST RUNS
Y-.M5«EXPC.844«X>9 '
FILTER BAG REVERSE MR DEFLECTION TESTS
SECTION SHORTENING UNDER DELTA P
.*-
.*-
.£-
BAG SECTION NUMBER (MINUS ONE)
Figure 11. Overall average exponential curve for bag section shortening due
to R.A. pressure.
FINAL MATHEMATICAL MODEL
By knowing the section deflections and the shortening caused by them, the
tensions resisting the deflection can be calculated using equations 22, 6, 8
and 10. Equation 10 was modified to include a pressure factor which was
varied to provide absolute equality. The pressure factors thus obtained, as
a function of bag section I, can be closely approximated by:
F9 (I) = .07833 + .00607 * I
(23)
Once the estimation of increased tension was obtained by the summation of the
values for each section, equations 10, 11, and 12 were then iteratively
calculated (with a new pressure factor in equation 10) to obtain a deflection
equal to that measured in the test program. The new pressure factor for
equation 10 is 0.673.
395
-------
The computer program 'DEFLEC' of Appendix II performs the computations of the
equations to calculate deflections of filter bags.when the section lengths
are known. Table 2 presents a sample printout of program 'DEFLEC1. Computer
program 'RINGOP' of Appendix III is a modification of 'DEFLEC1 which assumes
equal section lengths to start and then iteratively adjusts them until all
bag sections produce equal deflections for the stated conditions of initial
tension, ambient/operating temperature differential, support spring constant,
and ash cake thickness and weight. Table 3 presents a sample printout of
program 'RINGOP'.
Figure 12 presents the deflections of two bag designs, one with equal ring
spacings and one with optimized ring spacings, for an initial tension of 60
pounds and an ash weight of 57 pounds. The benefits of equal deflection can
be observed and are the result of designing the ring spacings to assure equal
deflection.
The'work described in this paper was not funded by the U.S. Environmental
Protection Agency and therefore the contents do not necessarily reflect the
views of the Agency and no official endorsement should be inferred.
10/12/82
Rodiu^ Inch..
n P t
u. d
Figure 12.
Section view of collapsed bag for both standard and optimum
spacing of anti-collapse rings.
396
-------
TABLE 2. SAMPLE OUTPUT OF PROGRAM "RINGOP".
This is prograM RINGOP to deterwine the optiwoM location of anti-collapse rings
in Filter bags to assure equal deflection in each section during reverse? air cleaning
11/2/1982
Bag
Length,
inches
396.0
Bag
Di a Meter,
inches
11,50
Operating
Temperature,
Degrees F
300
Ambient
Tenperature,
Degrees F
80
Dia, of
Ring wire,
inches
.1875
Tension
Prior to RA,
pounds
95.7
u>
vo
Weight
of cloth,
oz./sq.yd,
14
Initial
Tension,
pounds
60,0
Forward Gas
Pressure,
in.w.c,
4,00
Total Ash
Weight ,
pounds
40,8
Reverse Air
Pressure,
in ,w,c ,
.50
Number of Rings -• 1
Deflection greater than 1/3 diameter) further analysis terminated.
Tension
Start of RA,
pounds
139,5
Height Initial
of cloth, Tension,
oz./sq.yd. pounds
14 60,0
Forward Gas
Pressure,
in.w.c.
4.00
Total Ash
weight,
pounds
40.8
Reverse Air
Pressure,
in,w.c.
.50
Number of Rings -- 2
Deflection greater than 1/3 diameter; further analysis terminated,
Tension
Start of RA,
pounds
HI,2
Weight
of cloth,
oz./sq.yd.
14
Initial
Tension,
pounds
60.0
Forward Gas
Pressure,
in.w.c.
4.00
Total Ash
Weight,
pounds
40.8
Reverse Air
Pressure,
in,w,c.
,50
Nuttber of Rings • 3
Deflection greater than 1/3 dianeter; further analysis terminated,
Tension
Start of RA,
pounds
142,6
-------
TABLE 2. SAMPLE OUTPUT OF PROGRAM "RINGOP" (CONT.)
Weight
of cloth,
oz./sq.yd,
14
Initial
Tension,
pounds
60.0
Forward. Gas
Pressure,
in.w.c.
4.00
Total Ash..
Weight,
pounds
40.8
Reverse Air
Pressure,
in .w.c .
.50
Tension
Start of RA,
pounds
143.4
NuMber of Rings - 4
Section 1 .« 84.05 inches long (effectively); its deflection * 2.705 inches
Section 2 = 80,55 inches long (effectively); its deflection'•« 2.698 inchao
Section 3 = 77,8 inches long (effectively); its deflection - 2.742 inches
Section 4 •*. 75,05 inches long (effectively); its deflection ™ 2.795 inches
Section 5 = 71,5S inches long (effectively); its deflection =2.797 inches
Tension at the bottow of the bag (at the start of cleaning) - 91.3 pounds.
Measured (fabricated) section lengths are as follows (astuMing 2 inch ring covers)
Section 1- = 83,175 inches long
Section 2> 78,8 inches long
Section 3 •= 76,05 inches long
Section 4 = 73,3 inches long
w Section 5 ~ 70.675 inches long
CO
Weight
of cloth,
02 ,/sq.yd,
14
Initial
Tension,
pounds
60.0
Forward Gas
Pressure,
in.w.c.
4.00
Total Ash
Weight,
pounds
40.8
R&verse Air
Pressure,
in.w.c.
.30
Tension
Start of RA,
pounds
143.2
NuMber of Rings = 5
Section 1 = 7ft.04) inches long (effectively); its deflection - 1.881 inches
Section 2 - 67.666 inches long (effectively)J its deflection - V,SB inches
Section 3 = 65.291 inches long (effectively); its deflection = 1.88 inches
Section 4 - 64.416 inches long (effectively); its deflection - 1.971 inches
Section 5 = 61.916 inches long (effectively); its deflection = 1.971 inches
Section 6 ~ 57,416 inches long (effectively); its deflection - 1.972 inches
Tension at the bottoM of the bag (at the start of cleaning) - 90,6 pounds.
-------
TABLE 2. SAMPLE OUTPUT OF PROGRAM "RINGOP" (CONT.)
Measured (fabricated) section lengths are as follows (assuning 2 inch ring covers)
Section 1 = 69,166 inches long
Section 2 = 65.916 inches long
Section 3 = 63.541 inches long
Section 4 = 6S.666 inches long
Section 5 = 60.166 inches long
Section 6 = 58.541 inches long
u>
VO
10
Weight
of cloth,
oz ,/sq.yd.
14
Initial
Tension,
pounds
60.0
Forward Gas
Pressure,
in ,w.c .
4.00
Total Ash
Weight,
pounds
40.8
Reverse Air
Pressure,
in,w,c.
.50
Tension
Start of RA,
pounds
141.9
NuMber of
Section 1
Section 2
Section 3
Section 4
Section 5
Section 6
Section 7
Tension at
Rings = 6
= 60 inches, long (effectively); its deflection = 1.393 inches
- 58.25 inches long (effectively); its deflecfion - 1.393 inches
= 56.5 inches long (effectively); its deflection - 1.394 inches
~ 55.5 inches long (effectively); its deflection ~ 1,434 inches
= 54.625 inches long (effectively); its deflection^ 1.486 inches
- 52.75 inches long (effectively); its deflection = 1.487 inches
= 50.875 inches long (effectively); its deflection = 1.489 inches
the botton of the bag (at the start of cleaning) = 88.9 pound'-;.
Measured (fabricated) section lengths are as follows (assuwing 2 inch ring covers)
Section 1 = 59.125 inches long
Section 2 = 56.5 inches long
Section 3 = 54.75 inches long
Section 4 = 53.75 inches long
Section 5 == 52.875 inches long
Section 6 = 51 inches long
Section 7 = 50 inches long
-------
This it, pr-ogran DEFLEC to deternine the filter bag
deflection in each section during reverse air cleaning
Niwber of rings = 6
Bag
Length,
inches
396.0
Bag
Dianeter,
inches
11.50
Operating
Tenperature,
Degrees F
300
Anbient
Temperature,
Degrees F
80
Dia. of
Ring wire,
inches
.1875
Tension
Prior to RA,
pounds
95.7
Weight
of cloth,
oz./sq.yd.
14
Initial
Tension,
pounds
60.0
Forward Gas
Pressure,
in.w.c.
4.00
Total Ash
Weight,
pounds
10.2
Reverse Air
Pressure,
in.w.c .
.50
Tension
Start of RA,
pounds
141.5
Section 1 = 54.375 inches long (effectively)
Section 2 = 55.75 inches long (effectively))
Section 3 = 55.75 inches long (effectively);
Section 4 ~ 55.75 inches long (effectively);
Section 5 = 55.75 inches long (effectively);
Section 6 = 55.75 inches long (effectively);
Section 7 = 52.375 inches long (effectively)
Tension at the botton of the bag (just prior
; its deflection -
its deflection = 1
its deflection = 1
its deflection = 1
its deflection = 1
its deflection ~ 1
; its deflection =
to cleaning) =119.
1.147 inches
.234 inches
.264 inches
.296 inches
.329 inches
.365 inches
1.237 inches
5 pounds.
Bag
Length,
inches
396.0
Bag
Dianeter,
inches
11.50
Operating
Tenperature,
Degrees F
300
Anbient
Tenperature,
Degrees F
80
Dia. of
Ring wire,
inches
.1875
Tension
Prior to RA,
pounds
95.7
Weight
of cloth,
oz ,/sq.yd.
14
Initial
Tension,
pounds
60.0
Forward Gas
Pressure,
in.w.c.
4.00
Total Ash
Weight,
pounds
20.4
Reverse Air
Pressure,
in.w.c.
.50
Tension
Start of RA,
pounds
141.5
Section 1 - 54.375 inches long (effectively);
Section 2 - 55.75 inches long (effectively);
Section 3 = 55.75 inches long (effectively);
Section 4 - 55.75 inches long (effectively);
Section 5 = 55.75 inches long (effectively);
Section 6 - 55.75 inches long (effectively);
Section 7 = 52.375 inches long (effectively);
Tension at the botton of the bag (just prior
its deflection =
its deflection = 1
its deflection = 1
its deflection = 1
its deflection = 1
its deflection = 1
its deflection =
to cleaning) =109.
1.147 inches
.247 inches
.292 inches
.34 inches
.392 inches
.449 inches
1.332 inches
4 pounds.
400
-------
Bag
Length,
inches
396.0
Bag
Dianeter,
inches
11.50
Operating
Tenperature,
Degrees F
300
Awbient
Tenperature,
Degrees F
80
Dia. of
Ring wire,
inches
.1875
Tension
Prior to RA>
pounds
95.7
Weight
of cloth,
oz./sq.yd.
14
Initial
Tension,
pounds
60.0
Forward Gas
Pressure,
in.w.c.
4.00
Total Ash
Weight,
pounds
30.6
Reverse Air
Pressure,
in.w.c.
.50
Tension
Start of RA,
pounds
141.5
Section 1 -
Section 2 =
/•N Section 3 =
H Section 4 =
§ Section 5 =
^ Section 6 =
: Section 7 =
w Tension at
54.375 inches
55.75 inches
55.75 inches
55.75 inches
55.75 inches
55.75 inches
52.375 inches
the bottoM of
long (effectively);
long (effectively);
long (effectively);
long (effectively);
long (effectively);
long (effectively);
long (effectively);
the bag (just prior
its deflection =
its deflection =: 1
its deflection = 1
its deflection •- 1
its deflection = 1
its deflection •- 1
its deflection =
to cleaning) = 99.
1.147 inches
.26 inches
,321 inches
.387 inches
.461 inches
.544 inches
1.444 inches
3 pounds.
f
5
o
9
$
Bag
Length,
inches
396.0
Bag
DiaMeter,
inches
11.50
Operating
Tenperature,
Degrees F
300
Ambient
fenperature,
Degrees F
80
Dia. of
Ring wire,
inches
.1875
Tension
Prior to RA,
pounds
95.7
w Weight
« of cloth,
H oz,/sq,yd.
H
Initial
Tension,
pounds
60.0
Forward Gas
Pressure,
in.w.c.
4.00
Total Ash
Weight,
pounds
40.8
Reverse Air
Pressure,
in.w.c.
.50
Tension
Start of RA,
pounds
141.5
Section 1 = 54.375 inches long (effectively);
Section 2 = 55.75 inches long (effectively);
Section 3= 55.75 inches long (effectively);
Section 4 = 55.75 inches long (effectively);
Section 5= 55.75 inches long (effectively);
Section 6 = 55.75 inches long (effectively);
Section 7- 52.375 inches long (effectively);
Tension at the bottow of the bag (just prior
its deflection =
its deflection = 1
its deflection = 1
its deflection ~ 1
its deflection = 1
its deflection ~ 1
its deflection -
to cleaning) = 89,
1.147 inches
.273 inches
.351 inches
,438 inches
.538 inches
.652 inches
1.575 inches
2 pounds.
401
-------
Bag
Length,
inches
396.0
Bag
Dianeter,
inches,
11.50
Operating
Tewperature,
Degrees F
300
Anbient
Tenperature,
Degrees F
80
Dia. of
Ring wire,
inches
.1875
Tension
Prior to RA,
pounds
95.7
Weight
of cloth,
oz./sq.yd.
14
Initial
Tension,
pounds
60.0
Forward Gas
Pressure,
in.w.c.
4.00
Total Ash
Weight,
pounds
51.0
Reverse Air
Pressure,
in.w.c.
.50
Tension
Start of RA,
pounds
141 .5
Section 1 - 54.375 inches long (effectively);
Section 2 - 55.75 inches long (effectively);
^Section 3 = 55.75 inches long (effectively);
^•"Section 4 = 55.75 inches long (effectively);
g Section 5 = 55.75 inches long (effectively);
Q Section 6 =55.75 inches long (effectively);
^Section 7 = 52.375 inches long (effectively);
o Tension at the bottom of the bag (just prior
its deflection =
its deflection = 1
its deflection = 1
its deflection ~ 1
its deflection = 1
its deflection = 1
its deflection =
to cleaning) = 79.
1.147 inches
.287 inches
.382 inches
.493 inches
.622 inches
.777 inches
1.733 inches
1 pounds.
| Bag
o Length,
w inches
£ 396.0
Bag
Diane ter,
inches
11.50
Operating
TeHperature,
Degrees F
300
Anbient
Tewperature,
Degrees F
80
Dia. of
Ring wire,
inches
.1875
Tension
Prior to RA,
pounds
95.7
CO
M Weight
SJ of cloth,
jjjj oz./sq.yd.
14
Section 1 =
Section 2 -
Section 3 =
Section 4 -
Section 5 =
Section 6 -
Section 7 =
Tension at
Initial
Tension,
pounds
60.0
54.375 inches
55.75 inches
55.75 inches
55.75 inches
55.75 inches
55.75 inches
52.375 inches
the botton of
Forward Gas
Pressure,
in.w.c.
4.00
Total Ash
Weight,
pounds
61.2
Reverse Air
Pressure,
in.w.c.
.50
Tension
Start of RA,
pounds
141.5
long (effectively);
long (effectively);
long (effectively);
long (effectively);
long (effectively);
long (effectively);
long (effectively);
the bag (just prior
its deflection =
its deflection = 1
its deflection = 1
its deflection = 1
its deflection = 1
its deflection = 1
its deflection =
to cleaning) = 69.
1.147 inches
.301 inches
.415 inches
.552 inches
.717 inches
.922 inches
1.926 inches
1 pounds.
402
-------
BEFERENCES
1. Jensen, R.M., "Potential Improvements in Baghouse Design,"
Proceedings of the Fifth International Fabric Alternatives Forum,
Scottsdale, Arizona, January 15-16, 1981.
2. Gaylord and Gaylord, "Structural Engineering Handbook", McGraw-Hill
Book Co., New York, 1979.
3. Personal telephone call to Steve McCluskey of Burlington Mills,
3-16-81.
403
-------
APPENDIX I - TEST RESULTS
This is data for Run 5 of the bag deflection testing.
The Bay Type is * 2
The bag is 366.4 in. long, 11.38 in. in diaweter and weighs 252 oz
The initial bag tension was 51 pounds.
The bag tension at the start of R.A. cleaning was 215 pounds.
The R.A. pressure was .5 inches w.g.
The support spring constant was 90 pounds/inch.
The ash weight was 65.40 pounds.
The bag length prior to R.A. application was 31.4479 feet,
The bag length after R.A. application was 31.3646 feet.
Effective(straight)
56.99 in. long.
56.24 in. long.
53.49 in. long.
51.49 in. long.
49.74 in. long.
47.49 in. long.
44.36 in. long.
The bag
Ring 1
Section
Ring 2
Section
Ring 3
Section
»*» - .
o Ring 4
Section
Ring 5
Section
Ring 6
Section
Ring 7
Section
Ring 8
section dimensions are as follows;
Measured Effectiwe(arc) E
1,50 in, long, 1.50 in. long.
1 56.13 in. long, 57.00 in. long,
2,00 in. long, .25 in. lonci.
2
3
4
5
6
7
54
2
51
2
49
2
48
2
45
2
43
3
,50
.00
,75
.00
,75
.00
.00
.00
.75
.00
.50
,50
in,
long,
56
in, long,
in,
in,
in,
in,
in.
in.
in.
in.
in.
in.
Bag deflections are
Section
Section
Section
Section
Section
Section
Section
Section
Dia,
i;
2;
3;
4;
5)
6)
7;
6.
6.
6.
6.
6.
5.
5.
at
625
438
540
688
188
875
438
long,
long,
long,
long,
long,
long,
long,
long,
long,
long,
as follows:
deflection
in .
in.
in.
in.
in .
in.
in .
53
51
49
47
44
3
.25
,25
,50
,25
,50
.25
.75
,25
.50
,25
.38
.50
in. long.
l
in. long.
in, long
in, long
in, long
in. long
in, long
in. long
in, long
in. long
in. long
in. long
i
i
i
;
i
)
i
Catenary deflection
2.375 in
2.469 in
2.418 in
2.344 in
2.594 in
2.750 in
2.969 in
i
i
i
i
Equalizing deflection
1.783 in.
1.846 in,
1.811 in.
1,761 in,
1.929 in,
2,031 in,
2.169 in,
-------
APPENDIX I - TEST RESULTS (CONT.)
of the bag deflection testing,
11.38 in. in dianeter and weighs 252 oz,
This is data for Ron
The Bay .Type is t 2
The bag is 366,4 in, long,
The initial bag tension was 51 pounds.
The bag tension at the start of R'.A, cleaning was, 261 pounds
The R.A. pressure was 1.0 inches w.g.
The support spring constant was 90 pounds/inch,
The ash weight was 65.25 pounds.
The bag length prior to R,A. application was 31.4583 feet.
The bag length after R.A. application was 31.3229 feet.
The bag
Ring 1
Section
Ring 2
Section
Ring 3
Section
Ring 4
o Section
01 Ring 5
Section
Ring 6
Section
Ring 7
Section
Ring 8
section dinensions are as follows:
Measured Effectiwe(arc) E
1
2
3
4
5
6
7
1.
56.
2.
54.
2.
51,
2.
49,
2.
48.
2,
45,
2.
43.
3,
50
13
00
50
00
75
00
75
00
00
00
75
00
50
50
in .
in,
in,
in,
in.
in,
in,
in,
in .
in.
in,
in.
in ,
in,
in ,
long,
long,
long,
long,
long,
long,
long,
long,
long,
long,
long,
long,
long,
1 orig ,
long,
1
57
56
53
51
49
47
44
3
.50
,00
.25
.25
.25
.50
,25.
.50
.25
.75
.25
,50
,25
,38
,50
in
in
in
in
in
in
in
in
in
in
in
in
in
in
in
. long
, long
, long
. long
, long
. long
. long
. long
, long
, long
. long
, long
. long
, long
. long
t
i
.
,
.
,
•
,
4
,
.
,
t
Bag deflections are as follows;
Section
Section
Section
Section
Section
Section
Section
Section
Dia.
1}
2}
3;
4j
5;
7<
5
5
5
5
5
5
5
at deflection
.250
.438
,604
,688
.313
,063
.125
in,
in.
in.
in.
in,
in,
in .
Catenary deflection
3,
2,
2.
2.
3,
3,
3.
063 in
969 in
886 in
844 in
031 in
156 in
125 in
.
t
1
,
f
t
1
Effective(straiglit)
56,98 in, long,
56,23 in. long,
53.48 in, long.
51,48 in. long.
49,73 in. long.
47,48 in. long,
44,36 in. long.
Equalizing deflection
2.226 in.
2.169 in.
2.117 in,
2,090 in,
2,207 in,
2.282 in,
2.264 in.
-------
8
APPENDIX I - TEST BESULTS (CONT.)
This is data for Run 7 of the bag deflection testing,
The Bag Type is I 2
The bag is 366,4 in. long, 11,38 in, in dianeter and weighs 252 oz
The initial bag tension was 85 pounds,
The bag tension at the start of R.A, cleaning was 203 pounds.
The R.A, pressure was ,5 inches w,g.
The support spring constant was 90 pounds/inch,
The ash weight was 65.25 pounds,
The bag length prior to R.A. application was 31,4583 feet,
The bag length after R,A, application was 31.4323 feet.
The bag section dimensions
Measured
Ring 1 1.50 in. long,
Section 1 56.13 in, long.
Ring 2 2,00 in. long,
Section 2 54,50 in, long,
Ring 3 2,00 in. long,
Section 3 51,75 in. long,
Ring 4 2.00 in. long,
Section 4 49.75 in. long,
Ring 5 2.00 in. long,
Section 5 48.00 in, long,
Ring 6 2,00 in. long,
Section 6 45,75 in, long,
Ring 7 2,00 in. long,
Section 7 43,50 in, long,
Ring 8 3.50 in. long,
are as follows:
Effective(arc)
1.50 in, long.
57,00 in. long,
.25 in. long.
56.25 in, long,
,25 in. long.
53.50 in. long,
.25 in. long.
51.50 in. long,
.25 in, long,
49,75 in. long,
.25 in. long.
47,50 in, long,
.25 in. long.
44.38 in. long,
3.50 in. long.
Effective(straight)
57.00 in. long,
56,25 in, long,
53.50 in, long.
51.50 in. long.
49,75 in. long,
47.50 in. long.
44,37 in, long,
Bag deflections are as follows:
Section
Section 1;
Section 2}
Section 3;
Section 4;
Section 5;
Section 6}
Section 7;
Dia. at deflection
6.938 in.
7.438 in.
7.646 in.
7.436 in,
7.244 in.
7.063 in.
6.813 in.
Catenary deflection
2.219 in.
1.969 in.
1.865 in.
1.970 in.
2.066 in.
2.156 in.
2.281 in.
Equalizing deflection
1,676 in.
1.500 in.
1.426 in,
1.501 in.
1.569 in.
1.632 in.
1.719 in.
-------
3
APPENDIX I - TEST RESULTS (CONT.)
This is data for Run 8 of the bag deflection testing.
The Bag Type is 4 2
The bag is 366.4 in. long, 11.38 in, in diameter and weighs 252 oz,
The initial bag tension was 72 pounds.1
The bag tension at the start of R.A. cleaning was 211 pounds..
The R,A. pressure was .9 inches w,g,
The support spring constant was 90 pounds/inch,
The ash weight was 65,25 pounds,
The bag length prior to R.A. application was 31.4792 feet,
The bag length after R.A. application was 31.4063 feet.
The bag section dimensions
Measured
Ring 1 1,50 in, long,
Section 1 56.13 in, long,
Ring 2 2.00 in, long,
Section 2 54.50 in, long,
Ring 3 2,00 in. long,
Section 3 51,75 in, long,
Ring 4 2,00 in, long,
Section 4 49.75 in. long,
Ring 5 2.00 in. long,
Section 5 48.00 in. Ipng,
Ring 6 2.00 in, long,
Section 6 45.75 in. long,
Ring 7 2.00 in. long,
Section 7 43.50 in, long,
Ring 8 3.50 in, long,
are as follows:
Effective(arc)
1,50 in. long
57.00 in, long,
.25 in, long
56,25 in, long,
,25 in, long,
53,50 in, long,
,25 in. long
51.50 in. long,
.25 in. long
49.75 in. long,
.25 in. long.
47.50 in. long,
.25 in. long.
44,38 in, long,
3,50 in, long.
Bag deflections are as> follows:
Section
Dia, at deflection
Section
Section
Section
Section
Section
Section
Section
1}
2}
3;
4}
5}
6}
7)
6.
6,
6,
6,
6,
6,
6,
250
500
646
667
330
438
229
in.
in.
in .
in.
in,
in,
in.
Effective(straight)
56,99 in. long.
56.24 in. long,
\,
53.49 in, long.
51.49 in. long.
49.74 in. long,
47,49 in. long.
44,36 in, long,
Catenary deflection
2.563 in.
2,438 in.
2,365 in,
2,354 in.
2,523 in,
2,469 in,
2,573 in,
Equalizing deflection
1,908 in.
1,825 in,
1.776 in.
1.768 in.
1,882 in.
1,846 in,
1,915 in.
-------
o
00
APPENDIX I - TEST RESULTS (CONT.)
This is data for Run 9 of the bag deflection testing,
The Bag Type is * 3
The bag is 366.4 in. long, 11.25 in. in diameter arid weight, 256 oz
The initial bag tension was 63 pounds.
The bag tension at the start of R.A. cleaning was, 229 pounds.
The R.A, pressure was 1,0 inches w.g.
The support spring constant was 90 pounds/inch.
The ash weight was 65.25 pounds.
The bag length prior to R.A. application was 31,3646 feet,
The bay length after R,A. application was 31.2813 feet.
The bag section dinensions
Measured
Ring 1 1.50 in. long,
Section 1 54.25 in, long,
Ring 2 2.00 in. long,
Section 2 51.50 in. long,
Ring 3 2,00 in. long,
Section 3 50.00 in, long,
Ring 4 2.00 in. long,
Section 4 50.00 in. long,
Ring 5 2.00 in. long,
Section 5 49.00 in. long,
Ring 6 2.00 in. long,
Section 6 47.63 in. long,
Ring 7 2.00 in. long,
Section 7 47.80 in. long,
Ring 3 3.50 in. long,
are as follows:
Effective(arc)
1.50 in. long.
55.13 in. long,
,25 in, long,
53.25 in, long,
.25 in. long.
51.75 in. long,
.25 in, long.
51.75 in. long,
,25 in. long.
50.75 in. long,
.25 in. long.
49,38 in. long,
.25 in. long.
47.88 in. long,
3.50 in, long.
Effective(straight)
55.11 in, long.
53.24 in. long.
51.74 in. long.
51,74 in, long,
50,74 in. long,
49.36 in. long.
47.86 in. long.
deflections are as follows;
Section
Section Ij
Section 2;
Section 3)
Section 4;
Section 5;
Section 6}
Section 7j
Dia, at deflection
6.250 in.
6.021 in.
5.962 in.
6.313 in.
5.896 in.
5.771 in.
4,808 in,
Catenary deflection
2.500 in.
2.615 in.
2.644 in,
2.469 in.
2.677 in.
2.740 in,
3,221 in,
Equalizing deflection
1.865 in .
1,940 in.
1.959 in,
1.844 in.
1.981 in,
2.021 in.
2.316 in,
-------
o
vo
APPENDIX I - TEST RESULTS (CONT.)
is data for Run 10 of the bag deflection testing.
The Bay Type is 3 3
The bag is 366.4 in. long, 11.25 in. in diacieier arid weight, 256 02
The initial bag tension was 95 pounds.
The bag tension at the start of R.A, cleaning was 241 pound*.
The R.A, pressure was 1,0 inches w,g,
The support spring constant was 90 pounds/inch.
The ash weight was 65.25 pounds.
The bag length prior to R.A. application was 31,3854 feet.
The bay length after R.A. application was 31.3438 feet,
Effecliue(straight)
55,12 in. long,
53,24 in. long,
51,74 in , long,
51,74 in. long.
50.74 in. long.
49.37 in, long.
47.87 in. long.
The bag
secti
on difiensions are
Measured
Ring 1
Section
Ring 2
Section
Ring 3
Section
Ring 4
Section
Ring 5
Section
Ring 6
Section
Ring 7
Section
Ring 8
1
2
3
4
5
6
7
1
54
2
51
2
50
2
50
2
49
2
47
2
47
3
.50
.25
,00
.50
.00
.00
.00
.00
.00
.00
.00
.63
.00
.00
.50
in .
in.
in ,
in.
in ,
in,
in,
in.
in,
in,
in.
in.
in .
in.
in ,
long,
long,
long,
long,
long,
long,
long,
long,
long,
long,
long-,
long,
long,
long,
long,
as follows:
Effective(arc)
1
55
53
51
51
50
49
47
3
.50
,13
,25
,25
.25
,75
,25
,75
,25
.75
.25
.38
.25
.88
.50
in .
in ,
in ,
in.
in ,
in,
in ,
in,
in .
in .
in .
in.
in .
in.
in .
long
long
long
long
long
long
long
long
long
long
long
long
long
long
long
.
>
,
;
i
>
,
)
•
)
.
>
>
>
,
Bag deflections are ai, follows;
Section
Dia. at deflection
Section
Section
Section
Section
Section
Section
Section
1
2
3
4
5
6
7
i
;
i
i
)
]
i
6.
6,
6.
6.
6.
6.
5.
750
417
475
708
313
359
641
in,
in,
in.
in.
in.
in.
in.
Catenary deflection
2,250 in,
2.417 in.
2.388 in,
2,271 in.
2.469 in.
2.446 in.
2.805 in,
Equalising deflection
1.696 in.
1.809 in,
1.789 in,
1.710 in.
1.844 in.
1.828 in.
2.062 in,
-------
APPENDIX II - COMPUTER PROGRAM "DEFLEC"
10 REM This is prograM DEFLEC developed by J. Musgrove of Bechtel Power Corporation, Hou&ton, TX
20 REM This prograM was developed as part of 'OptiMizing the Location of Anti-Collapse Rings
30 REM in Fabric Filter Bags,' Paper 89, presented to The Fourth Synposiun on the
40 REM Transfer and Utilization of Participate Control Technology
50 REM Houston, Texas, October 11-15, 1982
60 REM This prograM listing is current as of Nove«ber 2, 1982
70 PRINTER IS 701,140
80 OPTION BASE 1
90 DIM L(9>,F<9),F1(9>,F2(9>,D<9),L1(9),DK9>,F9(9>,P9(9),T(9),T1(1>
100 MAT L=ZER@ MAT F=ZER8 MAT F1=ZER8 MAT F2=ZER
110 MAT D'ZERS MAT LITERS MAT D1=ZER6 MAT F9=^ZER
120 MAT T=ZER@ MAT Tl-ZERS MAT P9*ZER
130 PRINT "This is prograM DEFLEC to deternine the filter bag"
140 PRINT "deflection in each section during reverse air cleaning"
150 PRINT
160 CLEAR
170 DISP "What is the bag dianeter, inches"
180 INPUT D9
190 R9=D9/2
200 DISP "What is the total bag length, inches?"
210 INPUT L9
220 DISP "How nany anti-collapse rings are in the bag?"
230 INPUT 12
240 PRINT "NoMber of rings = ")I2
250 L8=L9-6~.25*12
260 K=I2-H
270 REDIM L(K),F(K),F1(K),F2(K),D(K),L1(K),D1(K),F9(K),P9(K),T(K)
280 DISP USING 290 ; K
290 IMAGE "Input all ",2D," individual section lengths (excluding ring cover dimensions)"
300 DISP "top to bottoM. Be sure to separate data by coMnas (i.e. 56,52,55,60)"
310 MAT INPUT L
320 LU)=LUH.875
330 L(K)=L(K)+.875
340 FOR Z^2 TO 12
350 L(Z)=L(Z>+1.75
360 NEXT Z
-------
APPENDIX II - COMPUTER PROGRAM "DEFLEC" (CONT.)
370 PRINT g PRINT
380 DISP "What is the bag support spring constant, Ib/irt?"
390 INPUT K9
400 DISP "What is the operating te«p,, F?"
410 INPUT T8
420 DISP "What is the awbient tenp., F, during initial bag tensioning?"
430 INPUT T6
440 T7-T8-T6
450 DISP "What is the weight of the bag Material, oz./sq.yd,?"
460 INPUT C
470 DISP "What is the nuMber of fill yarns per inch of fabric (44 is typical)?1
480 INPUT Cl
490 C2=C1*D9*PI
500 C3-1/C1
510 DISP "What is the dianeter of the ring wire, inches?"
520 INPUT R
530 DISP "What is the reverse air pressure, in, H20?"
540 INPUT A
550 T4=T7*(L9-H6)*K9*. 0000038
560 T5-Ax.5781*PI*R9A2/16
570 CLEAR
580 DISP e DISP e DISP e DISP e DISP
590 DISP "DO NOT DISTURB!!!"
600 DISP "PROGRAM IN PROGRESS"
610 FOR II•=! TO 3
620 FOR 13=1 TO 6
630 ASM
640 T2=A2*.5781*PI*R9A2/16
650 P=A,*,5781*C3
660 T8(l)=0
670 FOR 1=1 TO K
680 Y6=.OQ5*EXP«I-1)*.844>
690 Y7=Y6+.5673
700 Y8=Y7/100
710 Y9=1-Y8
720 L2(I)=Y9#L(I)
-------
APPENDIX II - COMPUTER PROGRAM "DEFLECT (CONT.)
,?3D D9a>=SOR<,375*L2~L2U»)
740 A9"ATN<4KD9/L2(I))
750 F9(I>*,07833+,00407«I
760 T«F9+T(I>
780 NEXT I
790 T*6Q*(Il+3>/4
800 Ai-00
810 T1=.015625KI3
820 U2=A1*T1*C3*16/1728
830 W5*RA2*PIA2*,284*U*12/<4*C2>
840 W1*C/(9*C1*144)
850 W4=B*W1
860 W3^W4+W5
870 W6=T1*PI*D9*a9-6>*Al/l728
880 T3*T+T4+T2
890 TS(1)=T8U)*C2/16
900 T9"T+T4~T5+T8<1>
910 PRINT " Bag Bag
920 PRINT " Length, Dianeter,
930 PRINT " inches inchet
940 PRINT USING 950
Operating
Teciperature,
Degree?, F
Ambient
Tenperature,
Degrees F
L9,D9,T8>T6>RJ3
950 IMAGE 2X,3D.D,8X,D».DD,10X,3D,12X,3D,10X,D,4D,9X,3D,D
960 PRINT g PRINT & PRINT
970 PRINT " Weight
980 PRINT " of cloth,
990 PRINT "o:./sq.yd.
1000 PRINT USING 1010
Initial
Tension,
pound
Forward Ga
Pressure,
in.w.c.
Total Ash
Weight,
pounds
C,T,A2,W6,A,T9
1010 IMAGE 4X,2D,9X,3D.D,BX,2D.2D,10X,3D,D,11X,D,2D,1QX,3D.D
1020 GOSUB 1220
1030 PRINT USING
1040 NEXT 13
10SO PRINT USING
1060 NEXT II
•J07D PRINT USING
1080 CLEAR
"K" ; CHR*(27) ; "
"K
"K
CHR$(11)
CHR$(11)
Dia. of
Ring wire,
inches
Reverse Air
Pressure>
in.w.c.
Tension"
Prior to RA,"
poondc"
lension"
Start of RA,"
pounds"
-------
APPENDIX II - COMPUTER PROGRAM "DEFLEC" (CONT.)
10?0 D1SP " J 000 BBEB"
1100 DISP ' J 0 0 E B"
1110 DISP " J 0 0 BBBB"
1120 DISP " J JO OB B"
1130 DISP " JJJ 000 BBBB"
1140 DISP •"
1150 DISP ""
1160 DISP " EEEEE N N DDDD"
1170 DISP ' E NN N D D"
1180 DISP " EEE N N N D D"
1190 DISP "' E N NN D D"
1200 DISP " EEEEE N N DDDD"
1210 END
1220 PRINT
1230 FOR 1=1 TO K
1240 P9-^P*,673
1250 F2/2
1260 ON I GOTO 1270,1280,1290,1300,1310,1320,1330
1270 Kl'O @ GOTO 1340
1280 Kl-L(l) 6? GOTO
1290 K1=L(1)+L(2) 6 GOTO 1340
1300 K1-L(1)+L(2)+L(3) 8 GOTO 1340
1310 K1-L(1)+L(2)+L(3)+L<4) 8 GOTO 1340
1320 K1=L(1)+L(2)+L(3)+L(4)+L(5) P GOTO 1340
1330 K1=L(1)+L(2)+L(3)-+L<4)+L(5)+L(6) 9 GOTO 1340
1340 F1(I)*T9*(16/C2)-K1*(W1+W2)-U3*(I-1)
1350 F1-T9*<16/C2)-L8*(WHW2)-W3K(1-1)
1360
1370
1380 IF FKO THEN F=-F
1390 D(I)=P9-x-L(I)A2/(4-x-SQR((2'X-F(I))A2-(P9*L(D)ft2))
1400 NEXT I
1410 HAT Ll^ZER
1420 MAT D1=ZER
1430 FOR J=l TO K
1440 LKJ)«INT(100flKL/1000
-------
APPENDIX II - COMPUTER PROGRAM "DEFLEC" (CONT.)
145& DHJ)
!1£! PRINT "Seeti°nV;HrlljL1R9 THEN GOTO 1480 ELSE 1490
1480 PRINT " This section is pancaking!!"
1490 NEXT J
1500 PRINT USING 1510 ; F
1510 IMAGE "Tension at the botton of the bag (just prio^ to cleaning) =",3D.D," pounds."
1520 IF F<0 THEN GOTO 1530 ELSE 1540
1530 PRINT "BAG IS BLOWING INTO THIMBLE DUE TO EXCESSIVE ASH OR INSUFFICIENT TENSION!!"
1540 RETURN
-------
Ul
APPENDIX III - COMPUTER PROGRAM "RINGOP"
10 RE.M This is progran RINGOP developed by J, Musgrove of Bechtel Power Corporation, Houston, TX
20 REM This prograM was developed as part of "Optimizing the Location of Anti-Collapse Rings
30 REM in Fabric Filter Bags,' Paper 89, presented to The Fourth SyMposiuM on the
40 REM Transfer and Utilization of Particulate Control Technology
50 REM Houston, Texas, October 11-15, 1982
60 REM This prograM listing is currerrt as of NoveMber 2,1982
70 REM "This prograM RINGOP determines the optiMUM location of anti-collapse rings"
80 REM "in filter bags to assure equal deflection in each section during reverse air cleaning"
90 PRINTER IS 701,140
100 OPTION BASE 1
110 DIM L(9),F(9),F1<9),F2<9>,D(9),P(9),F9<9>,T<9>,T8(1),L2(9)
120 PRINT "This is prograM RINGOP to deternine the optiMUM location of anti-collapse rings"
130 PRINT "in filter bags to assure equal deflection in each section during reverse air cleaning"
140 MAT L-ZERQ MAT F~ZER@ MAT Fl^ZERP MAT F2~ZER
150 MAT D=ZER@ MAT P^ZERS MAT F9=ZER
160 MAT T=ZER(? MAT L2=ZERP MAT T8=ZER
170 CLEAR
180 DISP "What is today's date? iWdd/yyyy"
190 INPUT At
200 PRINT A$
210 PRINT 6 PRINT
220 CLEAR
230 DISP "What is the bag dianeter, inches"
240 INPUT D9
250 R9=D7/2
260 DISP "What is the total bag length, inches?"
270 INPUT L9
280 DISP "What is the bag support spring
290 INPUT K9
300 DISP "What is the operating teMp., F?
310 INPUT T8
"What is the artbient teMp., F,
constant, lb/iri?'
320 DISP
330 INPUT T6
340 T7=T9-T6
350 DISP "What is the weight of the bag
360 INPUT C
during initial bag tensioning?"
Material, oz./sq,yd.?"
-------
APPENDIX III - COMPUTER PROGRAM "RINGOP" (CONT.)
370 DISP "What is the number of fill yarns per inch of fabr-ic (44 is typical)?"
380 INPUT Cl
390 C2=C1*D9KPI
400 C3=1/C1
410 DISP "What is the diameter of the ring wire, in.?"
420 INPUT R
430 DISP "What is the reverse air pressure, in. H20?"
440 INPUT A
430 P=A*,5781XC3
460 T4~T7*
610 W1=C/(9*C1*144)
620 U4=8*U1
630 U3=U4+W5
640 W6=T1»PI»D9K-(L9-6)*A1/1728
650 T3==T+T4fT2
660 PRINT " Bag Bag
670 PRINT " Length, Dianeter,
680 PRINT " inches. inches
690 PRINT USING 700
Operating
Temperature,
Degrees F
Artbient
Temperature,
Degree's, F
Dia. of
Ring wire,
inches
L9,D9,T8,T6,R,T3
700 IMAGE 2X,3D.D,8X,DD.DD,10X,3D,12X,3D,10X,D.4D,10X,3D.D
710 PRINT B PRINT 8 PRINT
720 FOR 12=1 TO 6
Tension"
Prior to RA,"
pounds"
-------
APPENDIX III - COMPUTER PROGRAM "RINGOP" (CONT.)
730 LB=L9-6-,25*12
740 T8(l )=0
750 K=I2+1
760 REDIM L(K),F(K)>F1(K),F2(K),D(K),P(K)>F9(K),T(K),L2(K)
770 FOR J=l TO K
780 L(J)=L8/K
790 NEXT J
800 FOR Zl=l TO K
810 Y6=,005*EXP«Z1--1)*.844>
820 Y7=Y6+.5673
830 Y8-Y7/100
840 Y9=1-Y8
850 L2*Y9*L(Z1)
860 D9*-L2=F9(Z1)*L
910 NEXT Zl
920 T8U)=T8<1)*C2/16
930 T9-T+T4-T5+T8U )
940 GOSUB 1160
958 NEXT 12
960 PRINT USING "K" ; CHR*U1)
970 NEXT 13
980 PRINT USING "K" ; CHR*(11)
990 NEXT II
1000 PRINT USING "K" ; CHR$(11)
1010 PRINT A*
1020 CLEAR
1030 DISP " J 000 BBBB"
1040 DISP " J 0 0 B B"
1050 DISP " J 0 0 BBBB"
1060 DISP * J JO OB B"
1070 DISP " JJJ 000 BBBB"
1080 DISP ""
-------
APPENDIX III - COMPUTER PROGRAM "RINGOP" (CONT.)
1090 DISP ""
1100 DISP " EEEEE N N DDDD"
1110 DISP • E NN N D D"
1120 DISP " EEE N N N D D"
1130 DISP " E N NN D D"
1140 DISP " EEEEE N N DDDD"
1150 END
1160 PRINT 6 PRINT
1170 PRINT " Height Initial Forward Gas Total At,h Reverse Air Tension"
1180 PRINT " of cloth, Tension, Pressure, Weight, Pressure, Start of RA,"
1190 PRINT "oz./£,q,yd. pounds in.w.c. pounds in.w.c. pounds"
1200 PRINT USING 1220 j C,T,A2,W6,A,T9
1210 PRINT
1220 IMAGE 4X,2D,9X,3D.D,8X,2D.2D,10X,3D.D,11X>D.2D,11X,3D,D
1230 J9=0
1240 PRINT "Nuwbei" of Rings = ";I2
1250 FOR 1=1 TO K
1260 P9«P«.673
1270 F2(I)=P9*L(I>/2
1280 ON I GOTO 1290,1300,1310,1320,1330,1340,1350
1290 Kl-=0 S GOTO 1360
1300 K1-~L(1) 8 GOTO 1360
1310 K1«L(1)+L(2) 6 GOTO 1360
1320 K1~LU)+L(2)+L(3> § GOTO 1360
1330 Kl==L(l)-H-<2)+L<3>+L<4) 8 GOTO 1360
1340 K1=L(1)H.(2)-H.<3)+L(4)-H.<5) 6 GOTO 1360
1350 Kl-L(mL(2)+L<3)H.<4>+L<5>+L<6)
1360 FKI)=T9*(16/C2>-K1»(W1+W2)-W3*(I-1)
1370 Fl=T9*(16/C2)-L8*(Wl+W2)-U3x(I-l)
1380 F(I)=SQRA2->
1430 IF D(I»D9/3 THEN GOTO 1650
1440 NEXT I
-------
APPENDIX III - COMPUTER PROGRAM "RINGOP" (CONT.)
1450 Xl-AMAX(D) S X2=*AMAXROW
1460 Y1=AMIN(D) S Y2-AMINROW
1470 Z=X1-Y1
1480 IF Z<.1 THEN 1540
1490 L(X2)=L-.125 8 L(Y2)=L(Y2)+,125
1500 J9=J9+1
1510 IF J9>200 THEN 1530
1520 GOTO 1250
1530 PRINT "Iterative lirtit reached at 200"
1540 FOR J=l TO K
1550 L(J)=INT(100Q*L(J))/1000
1560 D/1000
1570 PRINT "Section";Jj"=";L(J);"inches long (effectively); its deflection =";D(J);"inches"
1580 NEXT J
1590 PRINT USING 1600 ; Fl
1600 IMAGE "Tension at the bottoM of the bag (at the start of cleaning) = ",3D,D," pounds,,"
1610 IF FKO THEN GOTO 1620 ELSE 1630
1620 PRINT "BAG IS BLOWING INTO THE THIMBLE DUE TO EXCESSIVE ASH OR INSUFFICIENT TENSION!!"
1630 GOSUB 1670
1640 RETURN
1650 PRINT "Deflection greater than 1/3 dianeterj further analysis terminated."
1660 RETURN
1670 L(l)=L(l)-.875
1680 L
-------
PULSE JET ON-LINE CLEANING FILTER FOR FLY ASH
by: Wayne G. Wellan
Carter-Day Company
Minneapolis, Minnesota 55432
ABSTRACT
This paper describes the development and performance of
on-line cleaning by pulse jet filters with needled felt. This
pulse jet filter design is in operation on various types of
coal-fired boilers. Considerable research and testing was in-
volved in finding a fiber that would endure continuous pulsing,
high temperatures, and a sulfur environment. Additional re-
search was required to develop the proper filter media in the
areas of needling and scrim design. Also, analysis of various
types of fly ash was conducted to determine particle size and
shape, and what effect it has on the performance of the filter.
420
-------
INTRODUCTION
Eight years ago it was decided by our management that
Carter-Day should 'develop a filter for the flue gas filtration
market, specifically industrial and utility coal-fired boilers.
The market trends at that time showed a tremendous growth due
to the oil embargos and increase in prices of oil and gas.
Also, other people in the baghouse manufacturing industry had
tested a reverse air fiberglass baghouse with success.
Our first test site was a power plant at Granite Falls,
Minnesota operated by Northern States Power. This power plant
generated 50 megawatts of electricity from a pulverized coal-
fired boiler. They burned a blend of western and Kentucky coal.
This plant is a peaking station. They operate 8 to 16 hours
per day. Therefore, they pass through the acid dew point on
start-up and shut-down, which is a severe test for the fiber.
We installed two slip-stream filters. One was a 24RJ96 filter,
the other was a high velocity filter. The volume on the RJ
filter was 2,500 cfm and the volume for the high velocity filter
was 3,000 cfm. These units were put on stream in February,
1974.
Some of the first fibers we evaluated were Nomex, Polyi-
mide, Kynol, DrayIon T, Orion, CS treated Nomex, acrylics, woven
fiberglass, which all failed. We also tried teflon felt. This
withstood the high temperature gas and sulfur conditions, but
had very poor collection efficiency due to the on-line cleaning.
Extensive testing was conducted on stainless steel fibers in an
air laid web and also in a sintered construction, and none of
these were successful. We also evaluated a woven fiberglass
bag in the RJ filter which was in the round bag configuration.
This did withstand the operation for a few months, but the bags
eventually failed. Tests were also conducted on the high
velocity filter using teflon in a felt and pile construction.
None of these were successful.
We then installed a 48RF96 filter and a standard Pulse Jet
filter, 14PJ96, at the Dairyland Co-op Power Generating Station
at Alma, Wisconsin. This is an 80 megawatt power generating
system that has a pulverized coal-fired boiler which burned high
sulfur eastern coal, low sulfur western coal, and petroleum
.coke. This plant is a base loaded utility station which
operates 24 hours a day. The RF filter was designed to handle
3,500 cfm and the volume for the Pulse Jet is 600 cfm. The
slip-stream filters are designed to operate around the clock.
We have our own ash conveying system which is tied into their
system. These filters are operated unattended, other than
having the filter pressure drop, air flow and cleaning pressure
recorded and mailed to us once a week. Both of these systems
421
-------
are designed for on-line cleaning. We did not have a backup
filter for off-line cleaning. Our design and premise of this
whole program is with on-line cleaning. So, the bags are pulsed
while the flue gas is entering the filter. We again tried some
of the same felt we had at Granite Falls which failed in the
same manner. We also evaluated other medias such as Kermel,
carbon, graphite, ceramics, and Hyglass, which all failed. We
have evaluated a newer Hyglass with a 3/4 inch mesh round cage
design, and failed after 9 months of operation.
During this testing interval we came upon our new fiber
called Ryton and we installed 3 bags in our Pulse Jet unit and
proceeded to life test the bags. After 9 months we found that
there was no deterioration or failures.
We acquired additional fiber to completely fit out the
Pulse Jet filter and the RF filter at Alma. These bags were
evaluated initially for collection efficiency and also for bag
life. The collection efficiency was not satisfactory on the
first set of bags and were replaced with a heavier weight Ryton
in order to meet the new proposed EPA code, 0.03 to 0.05 Ibs
per million BTU. We also changed the bag frame by adding addi-
tional vertical and hoop wires to reduce the flexing of the
media, which would cause the media to fail. We tested the new
media and our collection efficiency did improve. Our results
were .018 pounds per million BTU.
We continued testing the bags to determine bag life, and
after 2 years without any bag failures, we proceeded to work
with our fiber supplier to get a commitment to supply us with
fiber for future jobs.
When we first contacted Phillips they were in the develop-
ment stages of their fiber grade Ryton resin; therefore, they
were not ready to release their fiber for commercial use.
Carter-Day got involved and developed a method to spin the Ryton
resin into fiber for the baghouse systems that are currently in
operation.
You probably have seen Ryton advertised in various maga-
zines. There are 2 different types of Ryton resin. One is
their molding grade resin which is solely used for injection
molding, and the other is fiber grade resin. The molded grade
resin cannot be spun into fiber. It is a highly cross-linked
resin which makes the fiber brittle.
FIBER TESTING
During the course of this program we develope'd a labora-
tory accelerated acid and temperature test to evaluate fibers
for the flue gas environment. The fibers are wicked in an acid
422
-------
solution of 75% H^SO^ at 350°F for 3 hours. Then tensile tests
are conducted on individual fibers and are compared to fibers
before the acid treatment.
This test procedure has worked out quite well to screen out
new fibers, and also serves as a quality control check on pro-
duction lots of Ryton fiber. If the Ryton fibers are not pro-
cessed properly they will deteriorate in a flue gas environment
in a very short time.
Another procedure we use to evaluate fibers is our labora-
tory microscope with a special lens. This enables us to see any
deterioration caused by acid attack and also shows us high stress
concentration in the fiber caused by over crimping the fiber
during the fiber spinning process.
In 'addition to the lab tests, bags are removed from the
slipstream filters, and from various installations in the field.
Physical tests are conducted on these bags such as tensile,
tear, and mullen burst. See Figure 1. From this data we can
project the bag life in various flue gas conditions. Also, all
new production lots of Daytex felt are checked for physical
strength, and one or two bags from each production lot is placed
in our slip-stream filter.
NEEDLING AND SCRIM TECHNOLOGY
During the test program on our slip-stream filters, we were
able to evaluate the long term effect on the filter bags. After
2 1/2 years in a flue gas environment, we have seen evidence of
delamination of the fiber from the scrim which is related to how
well the felt is needled. So, our next step was to evaluate the
needling technology and develop ways to improve the process.
After a long series of tests we were able to come up with an
optimum procedure that doubled the tensile and mullen burst
strength of the felt. The Ryton felt that is currently being
produced is needled with the latest technology.
The main mode of failure is caused by flexing which is
accelerated by the sulfuric acid and high temperature environ-
ment. In order to exceed the 2 year bag life, we had to improve
the needling and scrim design. The scrim is a woven screen made
of a fluorocarbon base fiber. This is placed in the center of
the felt to give it additional strength for flexing. A series
of tests were conducted on various scrim designs to improve the
strength of the felt. These series of tests were conducted at
the same time we evaluated the needling techniques. So, with
the combination of better needling and a stronger scrim we will
exceed our 2 year guarantee.
423
-------
100
80
£ 60
I-
s o
40
20
(A) TENSILE (FILL)
(B) MULLEN
(C) TEAR (WARP)
12
4-
30 36 42
18 24
MONTHS
FIGURE I- AGING EFFECTS OF OAYTEX IN A FLUE GAS ENVIRONMENT
-------
"What is Daytex"
This is the filter media developed by Carter-Day for fly
ash applications, and consists of:
1. Ryton fiber - (Polyphenylene Sulfide) resin developed
by Phillips Chemical.
2. Scrim - fluorocarbon support grid optimized by Carter-
Day.
3. Needling technology - optimized by Carter-Day.
PARTICLE SIZE VS FILTER PERFORMANCE
One of the other areas that determine a successful
operating filter system on fly ash is knowing the particle size
and shape. We have done extensive work in analyzing various fly
ashes from boilers around the country, and also reviewed the
operating performance of the baghouse at these locations. From
this we compared it to the data we have accumulated from our
field installations and slip-streams, and are able to recommend
the proper baghouse size for future fly ash applications.
We classify the various fly ashes into 3 major groups. One
is fly ash from an inefficient fired stoker, such as dump
grates, some underfeed stokers, or a stoker that is operating
with high excess oxygen, or upset conditions. Under these con-
ditions, the stoker will create a very fine soot or unburned
hydrocarbons which are sub-micron. See Figure 2. Therefore,
you have to reduce your air to cloth ratio in order to maintain
2-5 inches of pressure drop across the baghouse.
Another type of stoker we refer to is an efficient
operating stoker, such as a traveling grate, where there is no
upset conditions in the fire bed. This type of stoker will pro-
duce large unburned carbon particles, lOOy to 200y MMD. See
Figure 3. This type of fly ash is very easy to filter, there-
fore we can size our baghouse to operate at a higher air to
cloth ratio.
The other major type of fly ash is from a pulverized coal-
fired boiler. The particle size will range from 4p to 8y MMD
when burning western coal and lOy to 20u MMD when burning other
coals. See Figure 4. The biggest concern is the spherical
shape of the particles which can bleed through the media during
the cleaning cycle if the media is not designed properly. The
Daytex will eliminate the seepage of these particles during the
cleaning cycle.
425
-------
FIGURE 2- FLY ASH FROM INEFFICIENT FIRED STOKER
-------
6
FIGURE 3-FLY ASH FROM EFFICIENT FIRED STOKER
-------
to
co
FIGURE 4- FLY ASH FROM P-C FIRED
-------
BAGHOUSE DESIGN FEATURES
The filter used for the industrial and utility boiler
applications is the RF filter. See Figure 5. The gases enter
at the bottom of the filter, and the fly ash is collected on the
outside of the filter tube. The dust cake is released from the
filter tube by a high energy pressure wave. This pressure wave
is released from a quick opening diaphragm valve, then passes
through special orifices on a rotating manifold. The cleaning
system is a high volume, low pressure design (7.5 psi).
The weight of the Daytex filter tube is 22-25 oz/yd2 and
has proven to have excellent acid and temperature resistance up
to 400°F. The length of the filter tubes are 8 to 14 feet and
are removed from the top side of the filter.
SUMMARY
Ryton fiber with proper needling and scrim design is an
excellent felt for the Pulse Jet filters on industrial and
utility coal-fired boilers.
The particle size and shape of the various fly ashes has a
major effect on the performance of the baghouse.
As for bag life, none of the bags at the Coors facility
have failed, which have been in service for 33 months.
The RF filter system can operate with "on-line cleaning"
or "off-line cleaning."
The work described in this paper was not funded by the U.S.
Environmental Protection Agency and therefore the contents do
not necessarily reflect the view of the Agency and no official
endorsement should be inferred.
429
-------
SOLENOID
VALVE
DRIVE MOTOR
FOR REVERSE PULSE
AIR CLEANING SYSTEM
DIAPHRAGM
VALVE
AIR RESERVOIR TANK
FOR REVERSE PULSE
AIR CLEANING SUPPLY
FILTER TUBE SHEET
FOR HOLDING FILTER
TUBE CAGE ASSEMBLY
CEA'Carter-Day
DAYTEX FILTER TUBES
AIR INLET
COLLECTED
PARTICULATE
OUTLET
FACTORY INSTALLED
INSULATED TOP ACCESS
DOORS FOR FILTER
TUBE INSPECTION SERVICE
HIGH TEMPERATURE
ACID-RESISTANT
GASKETING
CLEAN
AIR OUTLET
CLEANING AIR
MANIFOLD
NOTE:
FILTER BODY AND HOPPER
INSULATED IN FIELD
HOPPER ACCESS
HOLE NOT SHOWN
60° CONICAL HOPPER
FIGURE 5 - CARTER-DAY RF FILTER
FOR COAL FIRED BOILERS
430
-------
TOP INLET VERSUS BOTTOM INLET BAGHOUSE DESIGN
by: Robert M. Jensen
Bechtel Power Corporation
San Francisco, California 94119
ABSTRACT
This paper compares top inlet with bottom inlet baghouse designs for bags
that collect on the inside. The comparison includes performance data for top
inlet baghouses in service with comparable bottom inlet performance.
The paper explains why pressure loss predictions for top inlet may be
more reliable than for bottom inlet. Other advantages claimed for top inlet
are higher cloth ratio for the same pressure loss as bottom inlet; use of
longer bags; better utilization of unreacted reagent after a spray dryer; and
more efficient cleaning with the reverse air.
The most important advantage of top inlet design may be the apparent
ability for all bags to operate at a constant face velocity in contrast to
bottom inlet bags in which the face velocity diminishes with time on line
after each cleaning.
Until a few years ago it was assumed that no correlation existed between
the pressure loss of a baghouse on coal-fired flue gas and the properties of
the coal burned and its ash. We now have ample evidence that a correlation
does exist but we do not know what it is. One consequence of this situation
is that we cannot predict accurately the pressure loss for a baghouse on
coal-fired flue gas.
We do know that a correlation exists between pressure loss and particle
size. We know that pressure loss increases as the median particle size
decreases. Anything that causes the median particle size to decrease will
increase pressure loss and should be avoided.
For baghouses on coal-fired flue gas, it is possible that the degree of
unpredictability of pressure loss is the same, or nearly the same, as the
degree of unpredictability of the amount of shift in particle size toward the
smaller size caused by hopper fallout and reverse gas recirculation. If we
431
-------
eliminate the size shift, we may also eliminate the uncertainty in predictions
of pressure loss. One way to avoid, or reduce, the shift may be to use top
inlet instead of bottom inlet baghouse design.
Bottom inlet baghouses decrease the median particle size entering the
bags by allowing coarse particles to fall out on entering the hopper and by
recirculating the reverse gas flow with its fines back to the inlet duct.
This shift toward smaller' particles can be avoided if we eliminate hopper
fallout and if we filter the reverse gas in a separate system.
This paper compares top inlet baghouse design with bottom inlet baghouse
design for cylindrical bags that collect on the inside. The schematic
arrangements of these two types of baghouses are shown for the filtering mode
in Figure 1 and for the cleaning mode in Figure 2.
V
hO-
V
BOTTOM INLET TOP INLET
Figure 1. Filtering mode.
When reliable application data are available, a bottom inlet baghouse can
be designed with assurance that it will work within the predicted pressure
loss and not exceed the allowable emission limit. Reliable application data
have been available for many years for a great variety of industrial
applications. Until recently, it has been assumed that baghouses can be used
to remove fly ash from coal-fired flue gas using design criteria that have
been satisfactory for similar materials and operating conditions. This
assumption is now in question as a consequence of the inability of some
baghouses on coal-fired flue gas to meet predicted pressure loss and, in some
cases, to stay within allowable emission rates.
One reason for this situation may be the highly variable nature of fly
ash. Another reason may be some undesirable inherent characteristics of
bottom inlet baghouses.
432
-------
V
o-
BOTTOM INLET TOP INLET
Figure 2. Cleaning mode.
Two of these undesirable characteristics of bottom inlet baghouses are,
first, the unavoidable fallout of coarse material when the dirty gas enters
the hopper and, second, the unavoidable dead end at the top of the bag. They
cannot be corrected for obvious reasons. An explanation of why they are
undesirable follows.
Consider the formula used for the pressure loss across cloth and cake:
C.
AP = k'V +
x
7000
v
x
x V
x v
k'V
(1)
where
AP =
Ci =
t =
V =
the pressure loss across the cloth and cake in one compartment just
before cleaning, inches of water gauge (in. wg)
3
grain loading at bag inlet, grains/actual ft (gr/acf)
time between cleanings, minutes
air-to-cloth ratio = face velocity, ft/min, fpm
= resistivity of new cloth (negligible),
in. wg
(fpm)
433
-------
k' = resistivity of filter cake, ln' Wg—=-
(fpm)(lb/ftZ)
111 W£?
kl = resistivity of cloth and cake after cleaning, -T-^—x
3 " (fpm)
where
k = Kozeny-Carman constant, dimensionless
= 5 (generally accepted value)
n = gas viscosity, lb(mass)/ft • sec
o
3
6p = specific particle density, Ib/ft
2
Ap = surface area of particle, ft
3
Vp = volume of particle, ft
e = porosity = void fraction = 1 - a
bulk density of particle ,. ,
a = TT~.—j—.„ e „. , > dimensionless.
specific density of particle
The terms kJV and k'V are usually neglected; k|V because it is very small
and k'V because it is the pressure loss immediately after cleaning which does
not have to be guaranteed and is not used in evaluations. Our basic formula
is thus:
Ci 2
AP = k2 X 7000 X C X V '
If we assume spherical particles and typical values for the terms in
Equation (2), we can solve for k' for different particle sizes. The
relationship of k' and particle size is shown in Figure 3. For the assumed
conditions, Figure 3 shows that k' is less than unity and therefore reduces
pressure loss for particles greater than 10 microns and that k! and pressure
loss increase rapidly as the particle size gets smaller below 10Tmicrons.
The value of k' used in Equation (3) is usually the median particle size
entering the baghouse. It should be the median particle size for the size
distribution entering the bottom of the bag. We do not know and we cannot
determine the particle size distribution entering the bag. This is true for a
number of reasons.
434
-------
100.0
10.0-
5
K
PARTICLES:
SPHERICAL, d DIAMETER
BULK DENSITY 40 LB/FT3
SPECIFIC DENSITY 160 LB/FT3
0 10 20 30
SPHERICAL PARTICLE DIAMETER. MICRONS
figure 3. Variation of kfc with particle size.
When writing a baghouse specification or when designing a baghouse it is
customary to calculate the grain loading entering the baghouse. This is done
by calculating the total ash, using the coal burn rate and the percentage of
ash in the coal. An estimate is then made for the amount of total ash that
will become fly ash. This gives us an approximate rate for the quantity of
fly ash.
The gas volume entering the baghouse is also an approximation. It
includes estimates of inleakage, airheater leakage, ash hopper evaporation,
and margin. The number of grains per actual cubic foot is thus the result of
dividing one approximation by another.
On the other hand, if we are working with a baghouse in service we can
make a traverse of the inlet duct and determine grain loading and particle
size distribution entering the baghouse or a compartment. This information
does not serve our purpose for several reasons. First, it differs from the
conditions at the inlet to the bags. Second, it may not include the fly ash
that is recirculated during reverse air cleaning. And third, it includes the
coarse fraction that will fall out when entering the hopper.
435
-------
We cannot determine the grain loading and particle size distribution
entering the bags because we cannot prevent nor measure hopper fallout. The
gas flow rate into a compartment varies; it is highest immediately after
cleaning and lowest immediately before the next cleaning. Hopper fallout will
therefore vary because of the variable inlet velocity.
As the gas flow rate into a compartment diminishes, so will the gas
velocity entering the bottom of the bags. A lowered velocity entering the
bags means a reduced ability to carry solid particles. As a consequence,
hopper fallout increases and the mean particle size entering the bags
diminishes.
A bottom inlet design is thus seen to cause a shift in the particle sizes
entering the bags toward the smaller sizes. As the median size of the
particles entering the bags diminishes, k' increases and so does pressure
loss. This is an undesirable inherent characteristic of bottom inlet designs.
Another inherent characteristic of bottom inlet design is the dead end at
the top of the bag. When the dirty gas enters the bottom of a bag, it has a
certain vertical velocity. As the gas passes through the bag, the volume
remaining and rising in the bag diminishes and, therefore, so does the
vertical velocity. As the vertical velocity diminishes, so does the ability
to carry solid particles. As the vertical velocity approaches zero at the top
of the bag, only the smallest particles will be carried upward. As a result,
fines accumulate at the top of the bag.
Several investigators report that they have determined experimentally
that there is no accumulation of fines in the top of bottom inlet bags. In
each of these experiments there was a time lapse between taking the bags out
of service and inspection of the bag. During that time, the compartment was
opened to the atmosphere, it was cooled and purged by pulling outside air into
it, and there was reverse flow through the cloth and down-flow in the bags.
During this time there was a gradual reduction in temperature, which reduced
the chimney effect in the bags. This time was equivalent to a null period
which some favor as part of a cleaning cycle to allow dust to fall. It is
very likely that some dust fell out of the bags during the cool-down period.
It is possible that the predicted accumulation of fines in the top of the bag
fell out during this period.
The actual inspection of the bags necessitated some agitation of the
bottoms of the bags by the people moving about on the tubesheet. Sampling the
cake on the inside of the bags by cutting swatches with the bags in place or
removing the bags and cutting them open would also dislodge and/or
redistribute the dust cake.
The conclusion that these experiments prove that there is no accumulation
of fines in the tops of bottom inlet bags is not warranted. These experiments
are inconclusive; they do not prove that there is no accumulation of fines at
the top of bottom inlet bags.
The value of k' at the top of the bag may be higher if the median
particle size is smaller. The path of least resistance for gas flow will then
436
-------
be through another part of the bag where k', and therefore flow resistance, is
lower. In other words, the face velocity, or cloth ratio, is variable over
the length of the bag.
None of the conventional cleaning methods is effective in removing the
fines from the top of the bag. During cleaning, some of the fines stick to
the top of the bag, some agglomerate and fall to the hopper, and some float in
the motionless gas until the bag goes on line again, at which time they return
to the bag.
Shaking the top of the bag will cause some agglomeration, but some fines
will still remain to go back on the bag. One or more deflation flows during a
shake/deflate cleaning cycle will help to move the dislodged material to the
hopper. However, deflation flow will take the path of least resistance, which
is in the lower part of the bag, and therefore is not effective for removing
fines from the top of the bag.
In some baghouses, reverse air cleaning has not been effective either.
It, too, takes the path of least resistance through the lower part of the bag,
bypassing the top of the bag.
The dead end at the top of the bag is thus seen to be a cause of
ineffective cleaning that entails pressure loss, increased frequency of
cleaning, and reduced bag life. There is no way to avoid this because the
submicron material will not settle by gravity. A half-micron size particle
of fly ash in still air will reach its terminal velocity in a fraction of a
second. At that velocity, it will take weeks for it to settle by gravity 30
feet from the top of the bag to the hopper. It actually will not settle any
distance because its fall will be opposed by the convection currents, or
chimney effect, in the bag.
We can assume that the accumulation of fines in the dead end will cause
some cloth area at the top of the bag to go blind. Loss of filter area for
this reason would necessitate a higher cloth ratio and therefore a higher
pressure loss in the balance of the bag if nothing else changed. What
actually happens can be inferred from the curve shown in Figure 4, which is
typical for bottom-inlet bags.
If we assume reasonable values for kl and C. and if we take values of AP
and t from the curve, we can solve Equation (3]T for V, cloth ratio. If we
then plot V versus the same time span, we have the relationship shown in
Figure 5. Although k* and C. are also likely to vary with time, we can assume
that k' will probably increase as the size of the particles entering the bags
becomes smaller and that its increase will be offset, to some extent, by a
decrease in C.. It is probably safe to assume that the shape of the curve in
Figure k is caused mostly by changing cloth ratio and that Figure 5 is a
reasonable representation of the variation of cloth ratio with time on line.
Figure 5 confirms the well-known fact that the gas flow rate through a
bottom inlet compartment decreases while on line. The total flow through a
baghouse is therefore the sum of the variable flows through the compartments.
This can be illustrated by the electrical analogy of Figure 6.
437
-------
k'-|V -I-
Figure 4. Pressure loss versus time on line for bottom
inlet baghouses.
Figure 5. Cloth ratio versus time on line for bottom
inlet baghouses.
438
-------
Q = TOTAL GAS FLOW
q = COMPARTMENT GAS FLOW
d = DAMPER RESISTANCE
b = CLOTH AND CAKE RESISTANCE
AP = SYSTEM PRESSURE LOSS
AP-
Figure 6. Electrical analogy.
In this diagram, Q is the sum of all the compartment flows, q, which
varies depending on how long each compartment has been on line since its last
cleaning. The system pressure loss, AP, will vary with Q, the number of
compartments in service, and during the cleaning cycle. However, at any one
moment, the sum of the resistances of the two dampers and the cloth and cake
for any compartment will be the same as the system AP.
The pressure loss across the cloth and cake, b, will vary as shown by
Figure 4. As b increases, q decreases and, as q decreases, so does the
resistance through the two dampers, d. Baghouses are said to be "self-
leveling" in the sense that the gas seeks the compartment with the lowest
pressure loss. This may explain the initial steepness of the curve in Figure
4 when the bags are newly cleaned. Note that "self-leveling" for bottom inlet
baghouses is on a compartment basis.
Figure 5 also illustrates the well established fact that cloth ratio for
bottom inlet baghouses diminishes with the time on line. The design cloth
ratio may be the average cloth ratio of an on-line compartment for the time
span between cleanings. This average may be different for each compartment.
Variable cloth ratio over the length of each bag and in every compartment is a
consequence of this undesirable inherent characteristic of the bottom inlet
design. Variable cloth ratio is undesirable because it means that the cloth
is not filtering at a constant rate. For bottom inlet baghouses, cloth ratio
can be thought of as the baghouse size; it is not the face velocity except
momentarily . at various places in some bags in one compartment when the
baghouse is in service.
Remember, cloth ratio is the amount of gas to be cleaned divided by the
area of the cloth used to clean it, or:
CFM ft
ft
min
= face velocity.
439
-------
If V diminishes between^ cleanings as shown by Figure 5, it could be
because CFM diminishes or ft increases. We know that cloth area does not
increase and therefore CFM must decrease. However, if the bag goes blind at
the top, we are reducing ft which would make V increase. If the bag goes
blind at the top and if V decreases, we conclude that CFM diminishes at a rate
that compensates for the reduction in filtering cloth area at the top of the
bag.
We thus have a vicious circle. As the compartment flow diminishes, the
velocity entering the bag diminishes and the proportion of fines entering the
bag increases. This accelerates the blinding at the top of the bag which
accelerates the reduction in the rate of flow which reduces the velocity
entering the bag, and so on. This may explain why the curve in Figure 4 shows
the rate of AP increase slowing down as the compartment nears the time for its
next cleaning. This vicious circle is unavoidable with bottom inlet
baghouses.
Top inlet baghouses do not have the two undesirable inherent
characteristics of bottom inlet baghouses. There is no hopper fallout and
there is no dead end in the bag. As a result, the relationship of pressure
loss versus time on line is as shown by the plots of data from top inlet
baghouses in service in Figure 7. The data sources are described in the
references.
AP,
IN. WG
6-
5-
4-
3-
2-
1-
-REFERENCE 1
^REFERENCE 2
10
50
60
20 30 40
t. MINUTES
Figure 7. Pressure loss versus time on line for top
inlet baghouses •
440
-------
Substituting values from Figure 7 in Equation (3) and solving for the
corresponding cloth ratio, we have the relationship for cloth ratio versus
time on line as shown in Figure 8.
Figure 8 shows that top inlet baghouses probably operate at constant
cloth ratio. If so, this would mean that every bag and every compartment
would have the same face velocity regardless of how long it has been on line.
For this to be so, we have to infer several things. First, k' does not vary
over the length of the bag. Second, the rate of gas flow is the same in all
compartments all the time on line. Third, top inlet baghouses may be self-
leveling on a per bag basis rather than on a compartment basis as is the case
for bottom inlet baghouses.
Figure 8. Cloth ratio versus time on line for top inlet
baghouses.
These inferences should be checked experimentally. The bases for them
are as follows:
For k' to be constant, the permeability, or porosity, of the filter
cake must Be the same in all bags, over the length of each bag, and in all
compartments -even though the cake gets thicker with time on line. This is
possible within limits. Contrary to some reports in the literature, there is
no hopper fallout with top inlet. As the vertically downward velocity in the
bag diminishes, we have a downward velocity component approaching zero and a
constant horizontal velocity component (face velocity) with a resultant
component approaching the horizontal. In other words, the particles move
toward the bag and reach the bag before they fall out of the bottom of the
bag.
441
-------
With no hopper fallout, the cake will have the full range of particle
sizes and the median size will be larger. With coarse material in the cake,
it will be more porous, k' will be smaller, and pressure loss will be smaller.
For a typical grain loading leaving a coal-fired boiler and for a typical
cleaning cycle, the filter cake will usually be only two or three hundredths
of an inch thick. With a porous cake and a small pressure differential across
it, there may be a negligible increase in k' as the cake builds from say 0.01
inch to 0.02 inch. In addition, there is very little likelihood of cake
collapse, which would make it less porous. The first inference is thus
plausible.
The second and third inferences are related. For the flow to be the same
in all compartments all the time, the resistance to flow must be the same in
all compartments all the time. This is possible if k', C , and V are all
constant, which may be the case for top inlet bags. For this to be so, we
have to assume that the cake in every bag is of uniform thickness, composition,
or porosity over the full length of the bag. This is possible if cake
deposition in the bag is uniform as a consequence of the gas seeking the path
of least resistance in each bag.
As dirty gas enters the top of a clean bag, it will deposit cake at the
top. The path of least resistance will then be below that cake. As the cake
builds, the path of least resistance moves down the bag. This process
probably happens very quickly over the length of the bag. The result may be a
cake of uniform porosity for the full length of the bag. Any deviation from
uniformity may be immediately corrected as the gas seeks the easiest way out.
The self-leveling process may thus be on a per bag basis for top inlet. This
could be a very significant difference from the per compartment basis for
bottom inlet.
There is a possibility that top inlet bags perform differently from
bottom inlet bags. Figure 9 may be evidence for this difference in
performance. Both curves are for baghouses in service on cement kiln
exhausts. The top inlet baghouse has a nominal cloth ratio of 3 ft/min and
the bottom inlet baghouse has a nominal cloth ratio of 2 ft/min. These
pressure losses were measured across cloth and cake in one compartment.
2.6
TOP INLET
7.4
BOTTOM INLET
IOAM
O5
IPER OPEN
1.0
1.5 2.0
SECONDS
2.5
3.0 3.5
Figure 9. Initial Pressure Loss Versus Time (Reference 3)
442
-------
Note the abrupt rise in pressure loss for the bottom inlet compartment
and the gradual rise for the top inlet compartment. Note, too, that the
pressure loss after the first three seconds on line for the bottom inlet
compartment is about three times the loss for the top inlet compartment.
During this test, the pressure loss for the bottom inlet compartment rose to
8.6 to 9.0 in. wg when it was on line and one of the other compartments was in
the cleaning"mode. The pressure loss in the top inlet compartment rose to 3.1
to 4.5 in. wg when it was on line and one of the other compartments was in the
cleaning mode.
The bottom inlet curve may be a confirmation of the theory that fines
floating in the top of a bottom inlet bag jump back on the bag as soon as it
goes back on line after cleaning. The top inlet curve may be a confirmation
that the filter cake formation in a top inlet bag is at a uniform rate.
The most significant difference between top and bottom inlet designs is
the fact that the bottom inlet design has hopper fallout and dead ended bags
and that the top inlet design has neither. Top inlet has other advantages
that are certainly worth consideration.
Some evidence exists that top inlet baghouses will operate at a higher
cloth ratio for the same pressure loss as bottom inlet baghouses or that they
will operate at a lower pressure loss for the same cloth ratio as bottom inlet
baghouses. This is illustrated by Figure 10.
Figure 10 shows that top inlet provides a saving in capital cost at the
same pressure loss or a saving in operating cost (AP) at the same cloth ratio
when compared with bottom inlet.
TOP INLET.
Figure 10. Comparison of top versus bottom inlet baghouses.
443
-------
Top inlet bags can be longer than bottom inlet bags because they do not
have a dead end to go blind. Making bottom inlet bags longer accelerates the
rate of blinding at the top and aggravates the vicious circle described
earlier. Contrary to common belief in the field of fabric filtration, high
entrance velocity at the bag inlet is not necessarily the cause of bag failure
at the inlet end. Failure of bottom inlet bags near the bottom at low (240
fpm) inlet velocity is more likely caused by inadequate bag tension. On the
other hand, top inlet bags are in service on cement kiln exhausts that have
inlet velocities of approximately 680 fpm that do not fail from abrasion at
the inlet.
One advantage of longer bags is the lower cost per square foot of cloth
and lower installation labor cost per square foot of cloth. In addition,
longer bags mean fewer bags and less plan area.
If the baghouse is to be used downstream of a spray dryer, the hopper
fallout in a bottom inlet design will contain some unreacted material that
would remove additional sulfur dioxide if it got into the cake on the bags.
With a top inlet design, this material would not be wasted.
To sum up:
Bottom inlet baghouses
o Suffer unnecessary pressure loss because particle size entering the bags
is reduced in several ways, and smaller particles entail higher pressure
loss.
o Operate at variable cloth ratio within each bag and among compartments.
o Are difficult to clean because of the accumulation of fines in the dead
end top of the bag.
Top inlet baghouses
o Have lower pressure loss because the filter cake contains the full range
of particle sizes.
o May operate at a constant cloth ratio within each bag and in all
compartments all the time.
o Clean more readily because the cake includes the coarse particles and
because the reverse flow does not bypass the top of the bag.
o Can operate at higher cloth ratios.
o Can use longer bags.
o Improve utilization of reagent and fly ash alkalinity in dry flue gas
desulfurization systems.
444
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency and therefore the contents do not necessarily reflect the
views of the Agency and no official endorsement should be inferred.
ACKNOWLEDGEMENTS
In addition to the persons listed in the references, the author is most
grateful to John Musgrove, William Lane, Ernest Purcell, and Betty Winkelman,
all of the Bechtel Power Corporation, for their assistance in various ways in
the preparation of this paper. Their assistance does not imply that they
agree or disagree with this paper.
REFERENCES
1. Estopinal, E., and Noone, G., of Columbian Chemical Co. Data from carbon
black baghouse in Mohave, California. Top inlet bags are 10-in. diam-
eter, 31 ft 6 in. long. Private communication.
2. Graham, H., of Giant Portlandx& Masonry Cement Co. Data from cement kiln
exhaust baghouse in Harleyville, South Carolina. Top inlet bags are
8-in. diameter, 62 ft 10 in. long. Private communication.
3. Brumagin, C., of Fuller Co. and Foster, R., of Whitehall Cement Manufac-
turing Co. Data from cement kiln exhaust baghouses in Cementon,
Pennsylvania. Top inlet bags are 11.5-in. diameter, 37 ft 10 in. long.
Private communication.
445
-------
UPGRADE OP PLY ASH COLLECTION CAPABILITY AT THE CROMBY STATION
by: T.J. Ingram
R.J. Biese
Gilbert/Commonwealth
Reading, Pennsylvania
R.O. Jacob
Philadelphia Electric Company
Philadelphia, Pennsylvania
ABSTRACT
A description is given of measures to upgrade the dust collection cap-
ability of 28-year old Cromby Unit 1 in the interim period before installa-
tion of an S02 scrubber. A novel approach was taken to enhance the per-
formance of outdated equipment, resulting in continued usage of existing me-
chanical collectors, upgraded electrostatic precipitators and the addition
of a sidestream baghouse to accomplish the objective. The equipment was in-
stalled and started up within a year of writing specifications, and com-
pliance was attained. The paper discusses some problems experienced,
notably bag blinding which has resulted in a severe maintenance problem;
steps in progress to resolve the problem are discussed.
INTRODUCTION
In August 1979, Philadelphia Electric Company signed a consent agree-
ment with the Environmental Protection Agency and the Pennsylvania Depart-
ment of Environmental Resources to bring Unit 1 at their Cromby Station into
compliance with particulate emission limitations of 0.65 Ib per million Btu
input pending installation of a wet particulate scrubber and an SC>2
scrubber. Gilbert/Commonwealth (G/C) was consulted to make recommendations
for achieving this goal and to do the engineering.
This 28-year old unit utilized mechanical dust collectors and elec-
trostatic precipitators in series to remove dust from the flue gas, and it
was found that emissions would have to be reduced by approximately 50 per-
cent to attain the limitation of 0.65 Ib/MMBtu. Although this limit is not
strict by today's new source standards, the old and worn equipment, typical
of such units, was not judged capable of achieving this performance with
simple reconditioning. On the other hand, extensive replacement was not
warranted because the scrubber installation would be complete within four
years.
446
-------
The program instituted was one of compromises, which was necessitated
by a schedule calling for compliance by April 1981, and by the usual space
limitations in an older plant. By the time initial tests could be run to
establish base performance, there were only sixteen months remaining to the
compliance date. To meet it, a gas flow model study was conducted, the
existing equipment was reconditioned and modernized, and a small baghouse
was added on a side stream.
Overall success of the program was demonstrated by achieving compli-
ance. However, there has been a problem with early blinding and shortened
bag life. A number of measures have been taken and the solution of this
problem is well on the way to resolution.
PROBLEM DEFINITION
HISTORY
Cromby Unit 1 went into service in 1954. It is a 165 megawatt (MW)
unit, designed to burn eastern pulverized coal in a front-fired, radiant,
Babcock & Wilcox boiler. The original design was for pressurized operation
and omitted induced draft fans. However, before construction was complete
it was recognized that there were operating circumstances where a negative
pressure in the boiler would be required for periods of time and reduced
size induced draft fans were included. Casing leaks eventually forced con-
tinuous use of these fans and reduced power output. However, full size fans
have recently been installed in conjuction with the installation of the S02
scrubber.
This unit was one of only three in the Philadelphia Electric system
that operated on coal when this project was begun in 1979. It was, there-
fore, considered economical to maintain it in service to take advantage of
the cheaper fuel, and to find means to comply with imposed environmental
regulations.
INTERIM CONSENT DECREE
As a result of discussions with the regulating authorities a consent
order was issued which required Cromby 1 to burn lower sulfur coal and to
reduce particulate emissions during the interim period to a maximum of 0.65
ib/MM/Btu input while the particulate and 862 scrubbing system was being
installed.
A series of tests performed in October 1979 showed average particulate
emissions of 1.08 Ib/MM/Btu input which was well in excess of the estab-
lished limit.
An upgrade schedule for achieving compliance by April 1981 was sub-
mitted in accordance with the consent order.
THE SITUATION
EXISTING EQUIPMENT DESCRIPTION
The flue gas cleanup equipment that existed at the plant at the begin-
447
-------
ning of the modifications was unique by today's standards. In each of two
parallel gas streams, between two tubular air heater stages, there was a
mechanical collector (Figure 1) to remove most of the ash. The mechanical
collector was the horizontal multi-tubular type. The flue gas entered the
tubes through vanes which imparted a spin. The ash, being heavy, was thrown
toward the wall" and exited along with some gas through an annulus in the
tube about half way through. It then dropped into a collecting duct and was
drawn off through two sets of cyclones in series by a conveying fan. The
conveying gas was reintroduced to the main gas stream just upstream of a
precipitator as shown in Figure 2.
The ash transport gas from the mechanical collectors was approximately
eight percent of total gas flow. Ash removed from this stream by the
cyclones dropped into a dust silo through rotary valves. Ash collected in
the precipitator hoppers was picked up by a stream of flue gas drawn from
the precipitator inlet plenum and carried to small cyclones by a secondary
conveying fan. The clean secondary stream was then combined with the clean
primary ash collection stream before emptying into the precipitator inlet.
The precipitator, or actually two precipitators in parallel, were old
units, each having 31 gas passages 8-3/4 inches wide, and two fields of col-
lecting plates 20 feet tall by 9 feet long. The collecting plates were
constructed of double sheets of expanded metal, and were rapped from the
Figure 1. Mechanical dust collector.
448
-------
MECHANICAL
COLLECTOR
CYCLONE
COLLECTORS
ID FANS
DUST
SILO
AIR
HEATER
ELECTROSTATIC
PRECIPITA
AIR
HEATER
FROM
ECONOMIZER
GRADE
Figure 2. Schematic - fly ash collection original installation.
449
-------
side by magnetic .impulse rappers. The discharge electrodes were weighted
wire type, using air vibrators for rapping. Power was supplied from trans-
formers through mechanical rectifiers. Single perforated plates were in-
stalled at both the inlet and outlet to obtain satisfactory gas distri-
bution.
POTENTIAL FOR PERFORMANCE IMPROVEMENT
Condition of Existing Equipment
At the outset, the existing equipment was examined to determine the
potential for upgrade. The mechanical collectors were found to be badly
worn: several of the spinners showed advanced wear, and there were holes in
a number of the tubes. The precipitator had several wires broken or
missing, and a heavy ash buildup on those remaining. The collecting plates
were heavily coated with ash. The perforated plates front and rear had some
buildup, though were not plugged. The lower stabilizing frame was subject
to sway. There were many ash deposits in the inlet and outlet indicating
problems with gas distribution, and there were several leaks in the casing.
The largest primary cyclones had inspection ports, and when examined
appeared to be in satisfactory condition. The smaller ones, having no
inspection ports, were not examined.
Improvement Requirements
The emission limitation of 0.65 pounds of particulate per million
Btu's required about 90 percent overall removal efficiency of the dust con-
trol equipment. The results of the initial tests indicated overall
efficiency of just under 88 percent, so the required improvement, at least
on the surface, did not appear to be too difficult to obtain. Average
efficiency of the mechanical collectors was 79 percent and of the precipi-
tators 57 percent. The question then was, could the existing equipment be
suitably modified or overhauled; or should it be replaced with new equip-
ment, in view of the age of the equipment and the required performance in-
crease? Although only a nominal overall improvement was actually required,
in order to be sure that all tests would show compliance and in order to
maintain a desired low particulate carryover for the future scrubber down-
stream, a somewhat higher target for overall efficiency was set. In fact, it
was decided to try to upgrade the existing equipment to get the best
possible performance improvement.
There were three choices for the mechanical collector. One, it could
be completely overhauled with new tubes and low pressure drop spinners to
minimize wear. Two, it could be completely overhauled as originally fur-
nished with high pressure drop spinners to maximize performance. Or three,
it could be gutted and not used.
The precipitator presented another problem. It suffered from a very
low specific collection area, about 89 square feet per thousand ACFM, and a
very high velocity of about 9.2 feet per second. In addition, the aspect
ratio was only 0.9. The deposits of ash observed in the inlet and outlet
during the initial inspection, and a brief examination of the ductwork con-
450
-------
figuration, indicated that a gas flow model study should be made to
determine changes to improve the velocity distribution through the pre-
cipitator. It was also obvious from the type of power supplies and controls
that much improvement could be made with modern electrical gear. Finally, it
appeared that the rappers were not doing an adequate job of cleaning the
electrodes. The wires that remained in the unit were in many cases kinked or
bent so that clearances to the plates were considerably decreased. The con-
dition of the precipitator casing was also questionable because of the
number of holes that were observed during inspection.
Alternatives for Improvement
There was a valid question as to whether it might be more economical
to install either new additional precipitator capacity or a baghouse, rather
than upgrade the old equipment. There were also several other considerations
in determining an alternative. The station suffered from a severe space
limitation, typical of many backfits. Upon looking at a photograph of the
plant before these modifications were made (Figure 3), it is obvious that
there was no space available to put a longer precipitator between the duct-
work coming from the air heater and the ID fan inlet. There was also a
concrete deck extending part way over the existing precipitator which
Figure 3. Cromby Station before modification.
451
-------
effectively blocked installation of equipment above. Space upriver from the
precipitator was obstructed by the No. 2 Unit and the area downriver from
the precipitator was limited due to space allotment for the new scrubber, as
well as to an existing substation. In addition to the space limitations,
there were severe schedule restrictions. The consent agreement had sfipu-
lated that compliance must be achieved by April 1981, which left only about
16 months to make any improvements, to test to make sure those improvements
were in fact doing the job intended and, if not, to make further improve-
ments. There were also, of course, requirements for a minimum duration
outage to make the modifications because of the cost of replacement gener-
ation.
SELECTED ALTERNATIVE
As a result of all these considerations, G/C recommended that the
mechanical collector be refurbished, the precipitator upgraded, and a pulse-
jet baghouse added to polish the dust conveying gas stream from the mechani-
cal collectors (Figure 4). This option required replacement of the conveying
fan to overcome the additional pressure drop imposed by the baghouse. G/C
engineers were confident that the difficult schedule could be met. Looking
back from the compliance date of April 1, 1981, testing the unit was planned
for the end of February, which called for start-up around January 1. Since
it was already mid-December 1979, only a little over a year remained to com-
plete the project.
Several specific measures were considered for improving precipitator
performance. Power could be maximized by replacing all wires with a later
design and using heavier weights, replacing the old transformer-rectifiers
with modern types having solid-state controls, and installing stabilizing
insulators for the lower guide frame. Rapping could be improved by replacing
all rappers with new magnetic impulse type having solid-state controls.
Effective gas velocity could be improved by optimizing distribution and
relocating the dust conveying gas duct discharge downstream of the precipi-
tator. This would require installation of a baghouse to remove dust passed
by the cyclones. It would have the effect of increasing the SCA and
decreasing the velocity through the precipitator by about eight percent.
The latter was felt to be critical because the scalping velocity was
probably being exceeded.
PROGRAM EXECUTION
SCHEDULE
The planned schedule is shown in Figure 5. After the base performance
test results were analyzed in mid-December 1979, the specification was writ-
ten for model testing. It was decided that these test results were needed
before the precipitator contract could be awarded, so a date of March 9,
1980 was requested for a preliminary report. In order to assure completion
by the end of the year, it was necessary to award contracts for all major
equipment by April 1. Although the initial stages occurred about a week
late, the contracts were awarded on time. Erection was accomplished during a
452
-------
MECHANICAL
COLLECTOR
CYCLONE
COLLECTORS
\
ID FANS
X
DUST
SILO
ELECTROSTATIC
PRECIPITATOR
AIR
HEATER
AIR
HEATER
FROM
ECONOMIZER
Figure 4. Schematic - fly ash collection modified installation.
453
-------
five-week outage, followed by checkout and test prior to start-up at the end
of the year.
MODEL STUDY
The first task was the model study which was necessary to determine if
modifications would be required in the gas flow path. Accordingly, the
specification was issued on December 24, 1979 and bids were due January 11,
1980.. The study was subsequently awarded to NELS, Inc., of St. Catharines,
Ontario.
Surprisingly, the velocity distribution which had been anticipated to
be very bad turned out to be quite good, with RMS deviation of about 12
percent. Although it appeared improvement might still be made, NELS1 experi-
ments with turning vanes did not indicate sufficient improvement to justify
the installation.
EQUIPMENT CONTRACTS
With the short bid time allotted, it was expected that it would be
difficult to evaluate all the bids and award the contracts on schedule.
However, with the number of bids received, the work was completed on time.
Philadelphia Electric internally expedited its procurement process and the
required approvals for purchase were immediately prepared.
1979
1980
NOV
DEC
JAN
FEB
MAR
NOV
DEC
MODEL STUDY
STARTUP
PRECIPITATOR UPGRADE
M ^•QggygQQQd
BAGHOUSE
(BASE
(PERFORMANCE
iTEST RESULTS
CONVEYING FANS
• ••raaaaaaSMMM
ASH REMOVAL SYSTEM
SPECIFICATIONS
BIDDING
EVALUATION
FABRICATION
ERECTION
PRELIMINARY REPORT
Figure 5. Project schedule.
454
-------
Precipitator
The precipitator upgrade work was awarded to Belco Pollution Control
Company. Their proposal covered the work requested, plus some innovative
ways to increase sectionalization and improve rapping.
Shortly after Belco began design, the scrubber engineer informed the
project engineers that the new ID fans which were to be installed to handle
the scrubber pressure drop would impose a vacuum of about 20 inches of water
on the precipitator. Accordingly, it was necessary to make an examination to
see if the negative pressure could be withstood especially in view of the
poor condition of the casing. Belco made this examination and recommended
adding stiffeners on the inside of the casing. This was necessary since the
outside of the casing between the precipitators was not accessible. To in-
stall the stiffeners, the collecting plates next to the walls would be slid
or swung out of the way far enough to gain access to the casing wall. Belco
subsequently was awarded an addition to their contract for this work, which
was scheduled to take place concurrently with the other modifications.
Baghouse
Enviro-Systems & Research was awarded the baghouse. They proposed a
four-module, pulse-jet unit with a gross air-to-cloth ratio of 3.86/1.
Design dust loading was 3 grains per actual cubic foot. The bags were to be
nominal 5 inch by 10 foot fiberglass having an acid resistant finish, with a
fabric weight of 16 ounces per square yard, and a double-beam weave.
Other Equipment
The new single-inlet, airfoil vane, main conveying fans (one operating
and one spare) were awarded to Westinghouse-Sturtevant. A vacuum ash con-
veying system was awarded to Allen-Sherman-Hoff.
DELIVERY, ERECTION AND START-UP
Equipment delivery and installation turned out to be a real challenge
for both the project expediters and the fiel.d workers. It is a credit to
all concerned that there was good cooperation among Philadelphia Electric's
engineering, production, construction and start-up personnel, G/C engineers,
United Engineers (the construction manager) and the vendors. There were
many long days worked through the 1980 Christmas holidays in order for the
equipment to be placed in service on schedule.
Erection was made more difficult by the weather. The winter was one of
the area's colder ones, with several days during the erection and start-up
periods being around 0°F. Besides making work arduous, moisture condensation
inside the equipment was a problem, making start-up tedious.
The unit went back into service on December 31, 1980. Because it had
been a long cold winter and the work had been done very rapidly, a number of
problems were revealed immediately after start-up. There was excessive
455
-------
sparking in the precipitator and there were problems with stuck butterfly
dampers and malfunctioning controls in the baghouse. There were also prob-
lems with the ash system, primarily because the weather was below freezing
and there was moisture in the ash lines which resulted in plugging. It was
found that the sparking in the precipitator was caused by bits of welding
rod and other debris which had been dropped inside the precipitator and
which lodged in the expanded metal collecting plates. There were also some
warped collecting plates next to the outside walls which had been bent when
they were swung out of the way during installation of the casing stiffeners.
The debris was easily removed; however, it was decided that the collecting
plates, since they were expanded metal, could not be satisfactorily
straightened. Therefore, it was necessary to remove a number of wires in
the affected gas passages and to weld steel strips to the perforated plate
ahead of those gas passages which were short on wires in order to restrict
the amount of dirty gas that would flow through them. Although this was
undesirable because it increased the velocity through the remainder of the
precipitator, the project staff felt it was necessary in order to obtain
satisfactory performance.
The sticking baghouse inlet and outlet dampers were trimmed in order
to keep them from binding and much time was spent on the controls in order
to secure proper cleaning sequences. Although the baghouse is equipped with
hopper heaters, the ash lines plugged repeatedly because of moisture and
improperly installed controls.
Start-up of the baghouse was done carefully. Since it was on a side
stream, its use was not necessary until the unit was burning coal, so flow
was not established during light-off with No. 2 oil. The conveying fan was
started and the baghouse dampers were opened after two of the four mills
were operating. Meanwhile, the baghouse hopper heaters had been on for over
a day.
COMPLIANCE TESTING
Problems with the equipment were corrected to the point that the unit
could be tested for compliance. This was done early in March of 1981;
fortunately, the results were nearly as anticipated. The unit averaged a
little over 0.5 pounds of particulate per million Btu's during three tests.
This worked out to a little over 93 percent overall efficiency.
UNIT PERFORMANCE AND PROBLEMS
Concurrently with the compliance tests, the unit was also tested for
equipment performance. These results are compared with the base data tests
in Figure 6.
Although the unit passed the required tests, all was not completely
satisfactory. As shown in Figure 6, the mechanical collector had actually
decreased in performance, probably because of the lower loss spinners, and
was performing at 67 percent. However, the precipitator, bent plates and
all, was performing at up to 85 percent efficiency. The baghouse was
-------
100-1
o
s
LU
cc
80-
o
g 60
o
LL
LL
LU
40-
20-
87.8
(>OSYSTEM<
579.2
tot
BASE
LINE
TEST
1979
93.2
SYSTEM
66.
SOS
;LU
SO?
85.3
98.6
COMPLIANCE
TEST
1981
93.4
>SYSTEM
74.1
O
LU:
o
UJ
73.8
.LU
LU
RETEST
1982
Figure 6. Performance test results.
4S7
-------
reaching 98.6 percent efficiency, but was experiencing early high pressure
drop and it was difficult to remove the cages and bags for inspection. The
latter problem was finally traced to an internal cuff in the bags combined
with oversized cage end caps, which prevented cages from being extracted
from the bags. This delayed an investigation of exactly what was causing the
high pressure drop because a bag could not be removed with the baghouse in
service. By the time a bag could be removed and sent to the laboratory for
investigation, the bags were completely blinded, after less than four months
use. A further investigation by TexLab, Inc., of trial bags of different
material indicated that not only the original bags but also two Huyglas and
two Teflon felt bags were blinded with fine particulate which was pene-
trating the cloth to the degree that it could not be removed with normal
cleaning. Tests for particle size distribution were made indicating a mean
particle size of about seven microns, by Bahco analysis. There was also a
significant quantity of fines. The reason for this is apparent when one
examines the system schematic in Figure 4. The baghouse receives the finest
particulate because the gas has previously passed through cyclone col-
lectors. Cyclones, being most efficient on large particles, leave the small
particles to be removed by the baghouse. Although this was anticipated, the
number of fines was larger than expected. In order to alleviate r.-he
problem, it was decided that a partial bypass around the large cyclones
would be installed, taking dust directly from the mechanical collectors so
that more of the coarse particulate would be introduced to the baghouse, at
the slight expense of higher dust loading.
The testing program had also disclosed that the secondary cyclones
which remove the ash conveyed from the precipitator hoppers were passing a
large quantity of material and in fact appeared to be leaking. Therefore,
when a 12-week outage was scheduled for the end of 1981 to install new ID
fans, a number of different tasks were planned to improve the interim modi-
fications. The baghouse had to be rebagged, and it was decided that trie
original fabric was still the most suitable choice for this, but with a re-
vised bag cuff design. Also, new cages having smaller diameter end caps were
furnished by the vendor, and the partial bypass around the cyclones was in-
stalled. The secondary cyclones were scheduled for refurbishment. When they
were dismantled, a great deal of pluggage and erosion was found.
Once again start-up took place around the first of a year, in late
January 1982. The baghouse was precoated using pulverized limestone and
started using similar procedures to the previous year. However, this time a
program was followed to establish the best cake possible in order to keep
the fines from penetrating the cloth. The baghouse was started with on-line
cleaning, and pulse air pressure was set at 55 psig. Cleaning was initiated
when overall pressure drop reached about 5 inches of water.
Initially, things looked much better than they did the year before,
and a new test was performed the end of February. Overall performance re-
sults were very good, indicating that the unit was emitting about 0.35
pounds particulate per million Btu's or again a bit over 93 percent effi-
ciency. However, the baghouse, in spite of the work done on it, again began
to creep up in pressure drop until the bags again blinded. Although the
458
-------
cyclone bypass appeared to have increased the mean particle size to the bag-
house, it had apparently done nothing to decrease the overall high number of
fines.
At the time of this writing the baghouse has recently returned to
service (9/20/82) after a one week outage during which new woven fiberglass
bags were installed, Staclean diffusers were installed, the pulse pipe
sequence was changed to random and the pulse air pressure setting was raised
to 110 psig.
Additional measures which may be implemented, if found necessary, are
to enlarge the pulsing system and to use bags made of a felted fabric.
CONCLUSION
The modification program was successful in achieving compliance with
the required emission limit. The blinding problem, although troublesome,
appears to be solvable. The large amount of work that was performed in a
short period of time demonstrates that a cooperative effort can accomplish a
great deal. The fact that the unit is in compliance at full rated Load
points to the real success of the project.
The work described in this paper was not funded by the U.S. Environ-
mental Protection Agency and therefore the contents do not necessarily re-
flect the views of the Agency and no official endorsement should be infer-
red.
459
-------
HIGH SULFUR FUEL, FABRIC FILTER STARTUP EXPERIENCE
by: Phil Hanson
Power Production Manager
City Power and Light Department
Independence, Missouri 64056
Larry Adair
Project Manager/Senior Engineer
Phelps and Phillips
Independence, Missouri 60455
Robert N. Roop
Product Manager
Research-Cottrell, Inc.
Somerville, New Jersey 08876
Robert B. Moyer
Manager Fabric Filter Applications
Research-Cottrell, Inc.
Somerville, New Jersey 08876
ABSTRACT
The long term operation of fabric filters on low sul-
fur fuels has been, with a few exceptions, well demonstrated.
Similar experience on high sulfur coals has not yet been
gained. As utilities continue to convert to coal, reliabil-
ity of fabric filters for high sulfur coal service will be
of paramount concern.
The City of Independence, Missouri, recently put into
service a fabric filter designed for three to five percent
sulfur coal flue gases. Details of the design will be
discussed. Anticipating the opportunity to test in real
460
-------
world conditions, a bag evaluation program was initiated and
is in progress. Physical and mechanical properties of the
bag fabric are being tested to determine the effects of high
sulfur service and a possible relationship to bag life.
Results of that bag test program, as obtained to date, and
its implications will be presented.
PLANT HISTORY
The two unit Missouri City Power Station, now owned and
operated by the City of Independence, Missouri, was orig-
inally owned and operated by the N.W. Electric Power Cooper-
ative, Cameron, Missouri, a member of Associated Electric
Cooperative, Inc., Springfield, Missouri. Two identical
generating units of 22 MW each were put into operation in
1954 at a site located adjacent to the Missouri River
approximately eight miles northeast of Independence.
Foster Wheeler boilers each generated 220,000 Ibs/hr of
steam by oil and coal firing to drive Westinghouse turbine
generators. Mechanical cyclones provided particulate col-
lection. In addition to the boilers, turbine generators
and their auxiliary equipment, the site included coal and
oil unloading and storage facilities, ash pond, cooling
water intake structure, well water storage reservoir, rail-
road sidings, transmission lines and a substation for power
transformation and distribution. The substation and trans-
mission lines remain the property of N.W. Electric Power
Cooperative, Inc.
Air Pollution emission limitations entitled "Air
Quality Standards and Air Pollution Control Regulations for
the Kansas City Metropolitan Area" were adopted by the
Missouri Air Conservation Commission on January 5, 1969,
and were applicable to the Missouri City Plant. The state
cited the Missouri City Plant for violation of visible
emission limitations on March 25, 1970.
The N.W.E.P.C. was granted variance to burn coal by the
Air Conservation Commission of the Missouri Department of
Natural Resources (DNR), but soon afterwards the EPA issued
an administrative order requiring compliance with the regu-
lations by the installation of electrostatic precipitators.
In early 1975, N.W.E.P.C. requested that the state
variance and the federal administrative order for ESP's
461
-------
be withdrawn. The request was based on escalating coal
prices. While at the time, coal was costing $0.62 per
million BTU and oil $1.80 per million BTU, it was projected
that coal would increase to $1.77 per million BTU by 1978.
Considering the projected rapid escalation of coal cost,
the cost of ESP's and the limited use of the Missouri City
Plant, N.W.E.P.C. wanted to revert to oil firing to comply
with the air pollution regulations. The DNR and EPA amended
their orders to allow for such.
In September, 1975, the Missouri City station exper-
ienced an electrical fire which destroyed the plant's high
voltage switchgear and much of the adjacent plumbing, wiring
and mechanical systems. The N.W.E.P.C. decided not to
repair the damages and closed the plant; Associated Electric
Cooperative had available other generating equipment better
suited to its system.
During the middle 70's rapidly escalating oil and
natural gas prices prompted a need for the City of Inde-
pendence to look for additional base and intermediate load
coal-generated capacity. The city was operating three coal-
fired units at its Blue Valley Station and several oil and/
or gas-fired combustion turbines throughout its system.
Phelps, Hogland & Phillips Engineering Company was hired to
study the City's power needs and various possible power
supplies. These studies and much negotiating resulted in
the City's purchase of the Missouri City Power Station in
1979. Engineering for renovation of the plant and the
installation of necessary particulate collection equipment
to permit coal firing was awarded to Phelps, Hogland &
Phillips.
FABRIC FILTER SELECTION
Three major factors contributed to the selection of a
fabric filter for controlling particulates from the Missouri
City Power Station:
UNKNOWN COAL SUPPLIES
Engineering to provide quick reactivation of the plant
was begun as soon as it became apparent Independence would
become the new owner. The contract to provide air pollution
control equipment was directly on the critical path, but
at that time the City had no long term coal supply with a
well defined quality. The ability of the boilers to burn
a variety of fuels dictated that the air quality system also
have that capability.
462
-------
ALLOWANCE FOR FUTURE REQUIREMENTS TO REDUCE SULFUR EMISSIONS
WITHOUT HARMING PARTICULATE COLLECTION
The City's Blue Valley Power Station had been re-
stricted on allowable sulfur emissions. A need to make
further restrictions to control ambient air quality in the
Kansas City area was a possibility. Any reductions re-
quiring the burning of low sulfur coal would not inhibit
the collection capability of a fabric filter.
ON-LINE MAINTENANCE
Since one of the reasons the Missouri City plant was
being renovated by the City was to replace oil and gas-fired
generation, outages that required utilizing combustion tur-
bines would be costly; continuity of service with on-line
maintenance of the air pollution control equipment was an
important criterion. However, funds for renovation were
not unlimited. The flexibility of a single multicompartment
baghouse to serve both units while providing the capability
of on-line maintenance was a good fit.
The specification for the fabric filter was published
for bidding on September 4, 1979, with bids received
November 5, 1979. The contract was awarded in December,
1979. Construction began in January, 1981 and was completed
in December, 1981. Start-up was in late March, 1982;
testing was conducted in July, 1982.
FABRIC FILTER DESIGN
The contract was placed with Research-Cottrell (R-C)
for complete system responsibility. R-C work commenced at
the discharge of the existing I.D. fans atop the roof of
the boiler house, included dismantling of the tie-duct
between the fan and the old stub stacks, and terminated at
the discharge of two new booster I.D. fans into a new common
chimney.
The scope encompassed design, engineering, fabrication
and erection of the inlet ductwork, including modulating
I.D. fan discharge dampers from the existing fans, bypass
system, fabric filter, discharge ductwork, booster I.D.
fans, ash handling equipment, insulation and the erection
of all supplied equipment.
The gas cleaning equipment for the unit is a Research-
Cottrell reverse air fabric filter consisting of eight (8)
modules, 4 deep x 2 wide. Each module contains 336 filter
bags, 23' long x 8" diameter, which result in a total
463
-------
filter area of 126,928 ft (111,062 ft2 with one module
off). Gas-to-cloth ratio with 8 modules on line is 1.78
ACFM/ft2 cloth and 2.27 ACFM/ft2 with 7 modules on line
including R/A flow. Filter bags are constructed of 14 oz /
yd2 woven fiber glass cloth with an acid resistant finish
of polymer/silicone-graphite-teflon. Table 1 shows the
fabric filter design data and Table 2 describes the filter
bags.
TABLE 1. FABRIC FILTER DATA
Design
Flue Gas Volume (ACFM) 226,000
Temperature ( F) 310
Inlet Dust Loading (Gr/ACF) 4.64
Gas-to-Cloth Ratio (Gross) 1.78:1
Net (w/o Reverse Air) 2.035:1
Net (w/Reverse Air) 2.27:1
Outlet Loading (Gr/ACF) 0.010
Number of Compartments 8
Gross Cloth Area 126,928
Total Bags 2688
TABLE 2. FABRIC FILTER BAGS
TYPE: Fiberglass (ECDE) with Acid Resistant Finish
2
SPEC: Weight - nominal 13.5 oz/yd
Permeability - 40-55 CFM/sq. ft. @ 0.5" W.G.
Count - 44 x 24
Weave -3x1 twill
Fiber Thickness - 0.015 in.
Weave - Warp: 37 1/0(F)
Fill: 75 1/2(T) + 75 1/0(F)
DESIGN: Bag Size - 8" dia. x 23'-0" (L/D = 34.5:1)
Bottom Attachment - Snap Ring
Top Design - Banded Top With Disposable Cap
Rings - 4 carbon steel anti-collapse rings/bag
464
-------
Cleaning of the modules is accomplished by a reverse
air system which pulls clean gas from the common outlet
duct and discharges it into an isolated module, collapsing
the filter bags and dislodging the collected particulate.
Reverse air (gas) is supplied by two (2) reverse air (R/A)
fans (one operating, one standby), each rated at 26,000
ACFM requiring 125 HP motors.
System booster fans are located downstream of the
fabric filter and are rated at 135,600 ACFM, 300 HP each.
Each fan's inlet louvre damper is positioned and operated
by a Bailey Control System. The Bailey Control signal is
based on the static pressure at the outlet side of the
boiler I.D. fans.
Hopper ash removal is accomplished by a Hydro Ash
System. Ash is pneumatically conveyed from each hopper
in sequence, mixed with the vacuum-producing water and
gravity-flowed to a nearby ash pond. The controls sequence
the compartment discharge feeder immediately after the
compartment cleaning cycle.
The fabric filter is designed to treat 226,000 ACFM
of flue gas at 310°F. The collection efficiency is to be
99.8% (based on a minimum inlet loading of 4.64 GR/ACF)
with one (1) module out of service. System pressure drop
is guaranteed to be 8 in. W.C. or less across the baghouse,
ductwork and dampers.
FABRIC FILTER STARTUP
PRE-COATING
Since the plant's flyash is generated from burning
high sulfur fuel, it was Research-Cottrell's recommendation
that the filter bags be pre-coated to add protection for
the bag fabric. An inert, moderate particle size diameter,
low sulfur ash was chosen for the pre-coat. The selected
ash was trucked to the site and conveyed via 4" diameter
hoses to individual compartments using a blower on the
truck.
The pre-coat operation was started with one booster
fan operating, all compartment inlet dampers closed, and the
baghouse bypass dampers closed. Ash was injected to each
compartment through the hopper access door. Initially
there was little increase in AP on module photohelic gauges
using this method. It was obvious some ash was getting onto
the bags because the shutting of the outlet dampers caused
ash to fall back into the hoppers. This method was continued
until the ash truck was empty. There was an ash level in
465
-------
each hopper up to the door. Continued operation of the
booster fan and air lancing of the ash in the hoppers re-
entrained it in the air flow to the bags. The second
booster fan was then operated with minimal inlet damper
opening and 90% on the first fan damper. One compartment
inlet damper at a time was opened (also one bypass damper)
and air lancing of the hoppers continued. This operation
was maintained until at least three-quarters of the ash
was held on to the bags.
After all hopper doors and inlet and outlet dampers
were closed, there was no evidence that a significant amount
of ash fell off the bags. It took only a very short time
to pull the remaining ash from hoppers through the ash
system. The maximum AP indicated on any module during this
pre-coat procedure was 1" W.C.
INITIAL BAGHOUSE OPERATION
The following conditions were required for initial
baghouse operation: one boiler operating at full load (20MW),
for a minimum of one (1) hour at steady state conditions.
Baghouse conditions prior to putting first module on
line are summarized below:
o One booster fan operating with damper 60% open
(Note: at this time fan damper operation was in
manual and was not being controlled at desired
1" W.C. boiler outlet pressure).
o Bypass damper open
o Outlet dampers open
o Inlet dampers closed
o Reverse air (R/A) dampers closed
o Reverse air (R/A) fans off
o Hopper heaters on (24 hrs.)
o Module AP = 0" W.C.
The No. 1 module was placed on-line by opening the
inlet damper. Indicated module AP increased to 0.5".
After approximately 1-1/2 hours of operation, no obvious
increases in AP were., noted. At this time Module No.. 5 was
placed in service, a AP of 0.5" was observed.
466
-------
After about one more hour of operation, Modules No. 2
and No. 6 were placed into service. One bypass damper was
then closed. The AP across in-service modules rose to 1".
Modules No. 3 and No. 7 were placed in service followed by
the remaining modules No. 4 and No. 8.
The time taken to bring all modules in service and to
close the second bypass damper was five hours.
After closing of the last bypass damper the increase
in module AP became more obvious. Within one hour, the
maximum AP was up to 3".
At this time, modules with the highest pressure drop
were nulled (outlet damper closed) one at a time and then
placed back on line. The effect was significant. When
each compartment was placed back on line its pressure drop
had been reduced to between 0.5" and 1". The conclusion
was that most of the new filter cake could be removed
during a null without reverse flow and therefore the bags
were not blinding. No reverse air was used during this
time.
After approximately 32 hours of operation the baghouse
was placed back into the bypass mode and the modules were
taken out of service (inlet and outlets closed) in order to
begin startup of the second boiler.
When the second boiler was up to full load and oper-
ating at stable conditions with no oil for one hour, the
baghouse was placed back into service for both boilers.
The normal startup method just described was employed with
minor exceptions. All inlet and outlet dampers were opened.
Time for modules to stabilize was allowed and then both by-
pass dampers were closed.
Indicated AP across the modules began increasing at a
faster rate than with just one boiler as would be expected.
After a short period when AP had increased to around 2",
nulling was tried to see what effect it would have. Little
or no effect was observed. At this time a reverse air fan
was started.
Initial operation with the R/A fan, with manual control
cleaning of a module, showed when cleaning was started at
3.5" to 4" AP, cleaning for a few seconds would drop the AP
in the clean module by 1". During this period great diffi-
culty in opening the reverse air dampers was encountered due
to high static pressure in the closed reverse air duct.
After overcoming the static pressure, R/A poppets would
surge open nearly half way. Cleaning in this method re-
sulted in puffing even with long settling times.
467
-------
REPAIRS AND MODIFICATIONS
Stack puffing was deduced to be caused by the in-rush
of gas into a module from the R/A system when a R/A poppet
opened. The expansion of the gas in the system and lag
time before the fan automatically throttled-down would force
gas out the inlet duct into the other modules. A stack puff
would then occur. This puff was eliminated when the module
inlet damper was closed at the start of cleaning. Although
satisfactory, a different solution was sought.
To reduce the R/A duct static pressure and the initial,
fully opened damper-related surge, modifications were made
to the fan damper control system within the Bailey controls.
Originally one damper set point existed which controlled the
dampers to maintain a given fan motor current. With all R/A
dampers closed (between cleaning periods), the dampers would
open fully in response to a control signal to increase fan
current. This philosophy resulted in high static pressure
in the R/A duct, and full volume condition when cleaning
started. A device was backfit into the Bailey system to
supersede this setpoint until a R/A poppet was fully open.
The device drives the R/A fan damper fully closed. When
the R/A poppet is fully open (limit switch contact made),
damper control seeks the motor amp set point and R/A volume
increases.
Because the fan dampers are not totally gastight,
enough gas is still passed to develop a moderate static
pressure in the R/A duct. A second modification was made
to bleed off this additional gas and keep the static pressure
low. Three 4-inch pipes were installed between the R/A and
outlet ducts. These allowed a small volume to bleed off
but did not affect the cleaning volume. Experimentation
indicated only one pipe was necessary; caps were screwed
onto the other two.
TUNING
Initial cleaning adjustments were made June 8-10, 1982
at 75% boiler load. Final tuning was accomplished during
the week of July 12-16 at 100%.
Full 100% load operation was observed at this time and
final adjustments were made. Figure 1 shows operation of
the cleaning system with a 4.5" W.C. start-cleaning set
point, 60% opening of R/A inlet damper and approximately
45 seconds of R/A cleaning time per compartment. Figure 2
shows the same 4.5" W.C. start point but with longer clean-
ing time (180 seconds indicated) and R/A inlet damper 100%
open.
468
-------
In figure lr it can be seen that the inverval between
cleaning cycles decreases with time and that the clean cloth
AP increases from 2.8 to 3.8. After adjustment to the R/A
cleaning duration and flow, clean-down has consistantly held
between 3.0-3.5" AP and the duration between cleaning cycles
has remained the same.
PERFORMANCE
On July 14, 1982, the fabric filter system was tested
for compliance and performance. Fuel analysis for the tests
and design values are noted in Table 3. The inlet test ports
for AP readings were located just downstream of the boiler
#2 I.D. fan discharge after combining the flows from boilers
1&2. The outlet pressure taps are located just beyond the
outlet flange of the baghouse prior to the inlet of the new
booster fan inlets.
TABLE 3. FUEL ANALYSIS
Average
Constituents Range %
Sulfur
Ash
Moisture
Carbon
Hydrogen
Nitrogen
Oxygen
Heat Value
2.
5.
2.
57.
3.
•
5.
(BTU/#) 10,
0
0
0
6
5
7
5
- 4.
-19.
-17.
-70.
- 5.
- 1.
- 9.
500-13
1
0
0
5
0
8
0
,500
Design %
2
12
7
64
4
1
6
11,
.7
.8
.4
.6
.7
.4
.4
700
As Tested %
4
11
8
67
4
0
10
12,
.05
.88
.75
.97
.87
.8
.53
800
At maximum tested gas flow of 210,000 ACFM, the maximum
system AP was measured at 6.9" W.C., when the first compart-
ment was isolated during the cleaning cycle. This AP is
dependent on the AP utilized to initiate the cleaning cycle
and was within the guaranteed value. A summary of perform-
ance data is included in Table 4.
469
-------
TABLE 4. FABRIC FILTER PERFORMANCE DATA
Design
Tested*
Flue Gas Volume (ACFM) 226,000
Temperature 310°F
Inlet Dust Loading (Gr/ACF) 4.64
Gas-to-Cloth Ratio (Gross) 1.78:1
(Net) 2.035:1
Outlet Loading (Gr/ACF) 0.010
Removal Efficiency (%) .998
Opacity (%) 20
System AP ("W.C.) 8.0
*Average of 4 runs
210,000
342°F
2.025
1.65:1
1.89:1
0.0081
.996
3
6.9
BAG TESTING PROGRAM
For this first baghouse to see high sulfur fuel utiliz-
ing a bag finish of polymer/silicone-graphite-teflon, it
was determined that a bag fabric testing program would be
undertaken. The acid-resistant-finish fiberglass cloth had
shown excellent life on low sulfur with occasional high
sulfur exposure at United Power Association's Elk River
station, but Independence would be the first high-sulfur-
only exposure. The testing program was designed to examine
bags in operation for increasing periods of time to determine
the cumulative effect of high sulfur exposure. Standard
fabric tests including weight, permeability, breaking
strength, Mullen burst, MIT flex and LOI were run. In
addition, the pH of the dust cake was measured.
Bag weight and permeability measurements can be indi-
cative of bag blinding. Breaking strength, Mullen burst and
MIT flex tests are measures of chemical, physical or thermal
degradation of the cloth. LOI (loss on ignition) shows the
presence or absence of the finish. For comparison purposes,
new bags are used as the base for determination of any de-
gradation. Weight and permeability are measured with "the
dust cake on the fabric and after it is removed by vacuuming
and washing. Figures 3, 4 and 5 graphically summarize the
MIT flex tests and breaking strength results. All of the
test results are enumerated in Table 5.
470
-------
TABLE 5. CITY OF INDEPENDENCE TEST REPORT
Weight (oz/yd2)
Top
Bottom
NEW
13.5
APRIL
18.4/13.6
18.1/13.7
JULY
23.3/13.6
22.2/13.6
AUGUST
23.2/13.6
22.3/13.6
SEPTEMBER
22.2/13.7
22.1/13.7
Permeability
(cfm/ft )
Top
Bottom
47.0
Strength (psi)
(WARP/Fill) 592/292
Top
Bottom
Mullen Burst
(lb/in2)
Top
Bottom
805
4.73
5.20
601/280
599/278
725
739
1.79
2.65
424/191
431/193
538
537
1.73
2.19
385/162
380/159
470
448
1.93
2.06
384/149
393/151
476
485
MIT (cycles)
(WARP/Fill) 40,038/3,517
Top
Bottom
L.O.I.
PH*
35,020/1,625
36,881/1,643
4.6 4.5
9.8
- 6,705/631
6,666/629
4.6
3.72
4,659/584
4,779/589
4.3
3.7
4,785/515
4,465/513
4.4
•Surface pH on bag after cleaning cycle
Note: Values in "new" condition reflects fabric tested to
determine initial values and not published data.
The permeability in the "as received" condition
averaged between 1.8 - 2.65 CFM/sq. -ft. for bags tested
during the last three test periods. These values corres-
pond with values obtained from Research-Cottrell's Elk
River installation employing identical acid-resistant-finish
bags. At these permeabilities, the differential across the
cloth is a maximum of 4" dirty and approximately 2V
immediately after cleaning.
After vacuuming the "as received" fabric,-the vacuumed
permeability ranged between 11.2 - 12.6 cfm/ft . This is
somewhat lower than expected. Under microscopic examination,
the cloth interstices show nodules with a high sulfate con-
tent which had encapsulated the surface filaments thus pre-
venting proper dust release. The low pH of the dust cake
solution supports this observation. However, as previously
stated, it appears that it is not significantly affecting
471
-------
the cloth resistance while operating within the baghouse.
The washed permeability returned to the values in the "new
condition".
The L.O.I, of the acid-resistant finish is specified
at a minimum of 4% by weight of the greige goods. Since
after seven months of operation, the tested L.O.I, values
exceeded the minimum of 4%, the results clearly indicate
its stability.
The M.I.T. flex cycle tests are indicative of the
number of flexes-to-failure which provides a comparison with
other fabric flex tests. This M.I.T. flex endurance test is
probably the most misinterpreted and misunderstood test as
applied to bag life in baghouses today.
Many variables affect M.I.T. flex-to-failure. These
variables consist of yarn construction, twist levels, method
of texturizing, finishing, treatment procedures, finish cure
time/temperature, operation time and temperature in a bag-
house, removal efficiency of volatile matter and embedded
abrasive particles within the yarn structure.
A fabric with a high flex value compared to a fabric
with a low flex value does not necessarily mean that one is
superior to the other. A good way to compare one fabric to
another is to evaluate the percent loss of M.I.T. flex
cycles based upon the environment to which the fabric is
exposed.
Fiberglass is the primary fabric in the utility and
industrial fly ash particulate market. That fabric will
usually stabilize after 4-6 months of operation at approxi-
mately 50% reduction in M.I.T. flex cycles compared to new
fabric. This reduction is primarily caused by physical and
thermal deterioration upon first exposure to flue gas.
Upon reaching equilibrium, the fabric M.I.T. cycles will
remain virtually constant until the fiber begins to show
additional fatigue or breakdown due to chemical, thermal, or
additional physical deterioration. Upon drastic degradation
from equilibrium, the rate of fiber fatigue/breakage due to
physical deterioration may indicate that the fabric is about
to fail and bags should be considered for replacement.
If chemical or thermal deterioration also occurs, the
equilibrium value of flexes will be lower than fabric flex
values caused by physical deterioration only. Chemical and
thermal deterioration usually occurs rapidly and can be
easily measured. Once chemical or thermal adverse conditions
cease to exist, fabric degradation stops. Physical deterior-
ation generally occurs where movement in the fabric exists.
472
-------
For example, reverse air cleaning causes local physical
damage at the flex zone. Chemical deterioration occurs
throughout the entire fabric.
By comparing flex values, between "flex fold" and
"non-flex fold" areas, levels of chemical vs physical
deterioration can be established. As an example, consider
the flex cycle from the July test. The new fabric flex
cycle was 3517. Assuming the normal 50% reduction to reach
equilibrium, 1759 flex cycles would have been expected.
However, due to chemical, thermal and physical deterioration,
the actual equilibrium norm was at 630 cycles, (test on flex
fold position) , which equates to an additional 32% loss.
= 82%'" 82% ~ 50% = 32% additional reduction
Therefore, we can presume that the fabric deteriorated
the equivalent of 1129 additional flex cycles due to chemical
and physical deterioration (1759 - 630 = 1129) . From further
testing of the fabric on the "off- fold" section, we obtained
a flex cycle value of 790 cycles vs the 630 cycles from the
"on fold" section. Subtracting this "off- fold" value from
the equilibrium norm we can speculate that the difference in
flex cycle loss was due strictly to chemical attack (1759 -
790 = 969 flexes) or 86% of the additional loss was due
from chemical deterioration 1 - 1129 - 969 _ Q,. and 14%
- TTTo — ~ 6%
was due to physical deterioration (1.00 - .86 = 14%).
BAG FAILURES
Knowing that most baghouse outlet loadings usually are
below 0.005 GR/ACF on utility applications, it was quite
surprising to observe loadings up to 0.0081 Gr/ACF during
the performance test. Approximately a week after the per-
formance tests were conducted, the boilers came down.
During an inspection of the baghouse 34 failed bags, over
of the total installed quantity, were found.
To date, a total of 93 bags have been replaced due to
mechanical fiber fractures, as shown in the photomicrograph
Figure 6. Of the 93 bags replaced, 34 were completely failed
where the bag had actually separated from the cap, and the
additional 59 appeared ready to separate; the top band was
visible, but was still in contact with the cap. (Figure 7) .
It appears from numerous microscopic examinations that when
the caps were installed in the bags by the bag manufacturer,
the packers apparently pulled on the top cuff area incor-
rectly when seating the band and fabric against the cap seat,
thus severing the fibers.
473
-------
Further inspection revealed that this top band was not
wrapped with a protective cover nor was the body of the bag
folded over to give a separate layer of fabric between the
band and bag body.
To date, ten complete examinations have been performed
on used bags and two new spare bags and only four of the ten
did not reveal this fiber fracture. In one instance,
several yarns had all 1632 filaments severed, all in the
position coinciding with the band top edge. All fractures
were at the top interior side of the bags, thus making it
impossible to observe the defects without removing the cap
and cutting the fabric at the band.
Observing the fractured filaments also in the new spare
bags confirms that the damage originated prior to installing
the bags in the baghouse.
A survey of bag suppliers was undertaken to question
methods of installing these bands in bags. The majority
responded that their standard construction for 8" or 12"
dia. bags includes a pre-wrapped band or the bag body folded
over to protect the bag itself from the band. As a recom-
mendation to end users and O.E.M.'s to eliminate problems
such as this, specifications issued to bag manufacturers
should specifically require that all bands and ring covers
be independently wrapped or that multiple layers of
fabric exist between any metal object and the bag body itself.
CONCLUSIONS AND RECOMMENDATIONS
1. The City of Independence fabric filter is sur-
passing all performance guarantee requirements
while high sulfur fuel is being burned. The air
quality system has not caused any boiler outages.
2. Startup was uneventful with only minor modifi-
cations required.
3. Pre-coating the bags with inert flyash appears
to have reduced the degree of bag blinding from
chemical attack to a nominal level.
4. Reduction in MIT flexes-to-failure and tensile
strength occurred after startup. An equilibrium
condition has been reached.
5. Continued monitoring of bag condition is recom-
mended to predict baglife.
474
-------
6. Bag construction specifications should include
provisions to prevent any metal and hardware
contac-b with the bag cloth.
ACKNOWLEDGEMENT
Research-Cottrell wishes to acknowledge the Environ-
mental Consultant Company of Phoenix, Arizona for their
assistance in performing physical/mechanical, chemical
tests and for the photo micrographs of bag fabric.
DISCLAIMER
Findings and projections regarding fabric filter
performance and durability which appear herein are strictly
for the purpose of scientific and technical inquiry and
should not be construed as representations or warranties
regarding characteristics of fabric filter systems marketed
by Research-Cottrell.
The work described in this paper was not funded by the
U.S. Environmental Protection Agency and therefore the
contents do not necessarily reflect the views of the Agency
and no official endorsement should be inferred.
475
-------
Figure 1
City of Independence
Interval Between Cleaning Cycles
Decreased Down to Near 20 Minutes Between Cycles
cr.
4.5" AP Initiation of cleaning
7/12/82 60% Inlet damper opening RA fan
45 Sec. (indicated) RA poppet opening
Full load
5 - reached
6 11 AM
12 N 1 PM 2 PM 3 PM 4 PM 5 PM 6 PM 7 PM 8 PM 9 PM 10 PM 11 PM 12 PM 1 AM
Time
-------
o
8
Figure 2
City of Independence
Cleaning Cycle After Final Tuning
4.5" A P Initiation
7/14/82 100% Inlet damper opening RA fan
180 Sec. (indicated) RA poppet opening
SAM 9AM 10AM 11AM 12 N 1PM 2PM 3PM 4PM 5PM 6PM 7PM 8PM
Time
-------
Figure 3
Tensile Strength
400 -t
600- -
500- -
a.
I
£ 400
o>
I
«*
CO
"3
| 300-
co
£
A,
9
1200-
100-
21
Warp yams
Fill yams
I
April
I
June
I
Aug.
I
Oct.
March May July Sept.
1982 (Months)
478
Nov.
-------
Figure 4
MIT Flex to Failure
(Filling Yarns)
4000
o
o
0)
Cycles
10,000
1,000
100
]? 3000- J
V)
JO)
u
>.
u
X
0)
2000- -
Warp
Fill
*\
V
0 Hours 10 ' ioo1 ' '1000
Days 1 7 21 32
1000- -
Expected physical equilibrium norm
(50% degradation)
Physical & chemical deterioration
(on flex fold)
Month
-------
Figure 5
MIT Flex to Failure
(Warp Yarns)
40,000
5,000--
~ »« B « I ' • I | I
Mar. April May June July Aug. Sept. Oct. Nov. Dec.
1982 (Months)
480
-------
FIGURE 6. GLASS FIBER FRACTURE
FIGURE 7. FABRIC FAILURE AT TOP SNAP BAND
481
-------
FUNDAMENTAL STRATEGIES
FOR
CLEANING REVERSE AIR BAGHOUSES
by: GENERAL ELECTRIC ENVIRONMENTAL SERVICES. INC.
M.G. KETCHUCK
M.A. WALSH
O.F. FORTUNE
M.L. MILLER
ADAPCO
M.A. WHITTLESEY
ABSTRACT
In order to choose among the various cleaning cycle strategies,
(Batch, Continuous, and Distributed) used for utility reverse air cleaned
baghouses, it is necessary to understand the fundamentals of the cleaning
process. This paper describes both analytical and experimental investiga-
tion of fundamental mechanics common to all cleaning strategies.
482
-------
INTRODUCTION
Considerable debate has occured as to what is the best way to clean
reverse air baghouses. Continuous cleaning, Batch cleaning and Distributed
cleaning are all being used in operating utility baghouses with varying
degrees of success. GEESI's operating experience indicates that a Batch
cleaning provides both reasonable pressure losses, and makes provision
for normal upsets in plant operating conditions, such as wet coal piles,
steam soot blowing, low load flame stabilization, oil/gas supplemental
firing, minor boiler tube leaks, etc.
Another issue which generates considerable discussion is whether it
is better to allow a very heavy filtercake to build up on fabric filter
bags before cleaning them, or to clean them, lightly loaded, as frequently
as possible. Thick filtercakes seem to produce better cleaning in practice,
but if overdone can lead to slack bags and fabric damage.
In order to insure that our baghouse cleaning cycles are fully
optimized, GEESI has committed to on-going experimental and analytical
programs designed to gain more insight into the fundamentals of reverse
air cleaning, and to get factual answers to the above questions.
In this paper we would like to report the results of two of these
programs:
(1) An analysis of the basic mechanics of the reverse air cleaning
process, using the DELTAP computer program.
(2) Experimental measurements, using load cells, of the change in
bag tensioning during the cleaning process.
THE FINITE ELEMENT MODEL
Observation of filter bags in reverse gas type fabric filter units
has long established that the phenomenon of filter bag collapse and
suspension spring bottoming occur during cleaning. To gain further
insight into the interdependent effects of the reverse gas flow and the
static/dynamic response of the total suspension system a Finite Element
model was developed utilizing the ANSYS program. The following parameters
were studied to determine the effects on filter bag cleaning, filter bag
life, and structural loadings on the fabric filter structure.
o Reverse Air Flow
o Internal Decompression
o Initial Tension
o Dynamic Snap Back
483
-------
The Finite Element model shown in Figure 1 reflects the stiffness of
the suspension support structure, the suspension spring, suspension hard-
ware, and the thimble/tube sheet structure. The fabric filter bag was
modeled as two dimensional spars spanning ring nodes. In the Dynamic
Analysis portion of the study the ring mass was lumped at the ring nodes
and the dust mass was attached to the ring nodes via breakable links. The
breakable links offered the ability to study various dust shedding schemes
ranging from simultaneous release at all node points to a cascade release
as dust cake from upper levels impinged upon lower regions of the filter
bags. A gap element was utilized to analyze the effect, of slack bags.
TENSION
SPRING
SPAN
I
SPAN
SPAN
3
SPAN
4
UPPER BEAM
SUBSTRUCTURE
SPAN
6
SPAN
7
.RING NODES -
HELD LATERALLY
55
•BAG NODES
47
3 I
23
16
, SUBSTRUCTURE
609 Pressurtzotlon Model
Used To Determine
Deflections And Tenstonlng
Spring Loads
I g Q DUST MASS
Snap Bock Dynastc
Analysis Model
Figure I. Finite Elvient Modelling Schematics
484
-------
REVERSE GAS FLOW/INTERNAL DECOMPRESSION
The effects of Reverse Air Flow and Internal Decompression were
studied utilizing the Finite Element model combined with a fluids
approach. The collapse of the filter bag under the external pressure
loading of reverse gas flow was studied using a Static Analysis to deter-
mine the deformed filter bag geometry and the bag spring rate.
A Dynamic Analysis was performed to determine the vertical mode
frequency response, as it was initially felt that the rapid load increases
in the suspension spring was possibly related to dynamic response of the
Bag-Spring System as the dust cake detached and rapidly unloaded the
filter bag and springs. The results of the Modal Analysis are shown in
Table 1. It is important to note that even in the case where the filter
bag lower span is collapsed, the fundamental frequencies are well above
the observed frequency of oscillation of .125 to .167 HZ, and therefore,
would not make significant contributions to the observed peaks in filter
bag loadings.
TABLE 1. MODAL ANALYSIS SUMMARY
Case Description Mode Freq.(HZ)
a
b
c
Bag assumed to be connected to a
rigid tubesheet and rigid upper
support beam.
Bag supported on a flexible
tubesheet and upper support
beam.
Bag supported by a flexible upper
beam and unsupported at the bottom
(i.e. bottom bag span collapsed).
1
2
3
1
2
1
2
3
5.58
12.77
20.92
5.49
7.58
2.58
9.01
16.69
A Fluids Analysis approach was used to study the effects of the
falling dust plug. The effect is similar to that of a piston and cylinder,
however, in this case the cylinder walls (the filter bag) has permeability
and the piston (the dust plug) has porosity. Figure 2 shows the concept
used in this portion of the study
485
-------
0
1
i
1
1
1
1
|l
>
1
©
>
1"
S [_
, , . / ' / ' / ' /
^ — BAG WALL
^f
1 " "~4}
ii- -£'
!i
ir~^— CUNIHUL VULUMb
I. '!
ll
1 ]l
JC l
"7"
^/Z/fa—L
i
i I
i
' ^^ FALLING DUST CAKE
P
r
•
*
r
|-*— D— *•]
Figure 2. Schematic of Pressure Rel oi lonsh (ps
For Fluid Analysts Model
The solution to the problem involved establishing equivalent "G"
loadings on the spar elements, deformed filter bag geometry behind the
falling dust plug, and inputing the stress strain properties of the filter
bag material. Due to the nonlinear material properties, large deformations
of the bag structure, and stress stiffening effects, an iterative procedure
was required to converge on the solution.
Initially the reverse gas pressure loading was applied and the
structure was allowed to deform to equilibrium geometry. The dust cake
release was initiated at the top of the filter bag, and as shown in Figure
3 the porous dust plug began to fall towards the filter bag outlet. As
the dust plug cascaded down the length of the filter bag the incremental
dust mass was released. Maximum suspension spring loading occurred as
the porous dust plug cleared the bottom of the filter bag. The results
of one analysis is shown in Figure 4. Clearly, spring loadings of 2 to 3
times the initial filter bag tension can occur due to the internal
decompression behind the falling porous dust plug.
Filter bag suspension spring, suspension hardware, and support
structures must be designed to accommodate these increased loadings and
proper design considerations must be made to eliminate structural failure
due to low cycle fatigue.
486
-------
INWARD MOTION DUE TO
SUCTION BEHIND FALLING
DUST PLUG
DUST PLUG
Figure 3. Idealization of ihe Falling
Dust Cake Upon Cleaning
487
-------
250T
DUST PLUG
CLEARS BOTTOM
CALCULATED -
BASED ONI 25X
POROSITY PLUG
BASED ON FIELD
DATA
TIME
Figure 4. Comparison Between Calculated and Field
Measured Peak Spring Tension
488
-------
INITIAL BAG TENSION
The effects of Initial Bag preload on the filter bag deformed geometry
and filter cake fracture was studied using the Finite Element model
developed. The case of a fabric bag pretensioned to 75 Ibs. H^ 15 Ibs. was
studied. The results of the analysis are presented in Figure 5. Nodal
displacements for the upper most and lowest span are presented for both
the 90 Ibs. and the 60 Ibs. tension cases, thereby bracketing the initial
target tension of 75 Ibs.
The effect on filter bag nodal displacement and subsequent cake
cracking appears to be narrowly banded. This indicates that the require-
ment for precise initial bag tension is currently over emphasized by
equipment specifications and in fact may be detrimental from a maintenance
point of view, as the requirement for precision filter bag tension breeds
complicated hardware with increased cost.
However, the case for an adequate magnitude of initial tension should
not be diminished as too low an initial tension can lead to early filter
bag failure.
5.0
« 4.0 +
c
£
o
c
c 3.0
o
o
o
a
s"
o
o
2
1.0 +
BOTTOM SPAN
15 LBS. OVER SPECIFIED
BAG TENSION
15 LBS. UNDER SPECIFIED
BAG TENSION
1.0 2.0
A Pressure Cinches H203
Figure 5. Maximum Span Displacements for the
Top and Hollo* Spans Versus A P
for A Range of Bag Tensions
3.0
489
-------
DYNAMIC SNAP BACK
The effect of a low initial filter bag tension may result in the lower
region being untensioned and slack. In order to study this condition the
Finite Element model was modified by placing a Gap Element in the bottom
bag span. The filter bag model was then subjected to an initial dust cake
loading with the bottom span in a slack condition. The dust cake was
released and the filter bag was allowed to respond.
Figure 6 shows the force-time history response of the bottom span.
The fabric material is subjected to a severe, short duration pulse loading.
The effect on subsequent filter bag spans diminishes as one moves up the
bag. At the upper suspension point the peak load is non existent and the
system oscillates about the new equilibrium without the dust loading.
As previously discussed, the precise magnitude of initial bag tension
is unimportant, but to avoid damaging effects to filter bag material
sufficient minimum tension must be applied to prevent Dynamic Snap Back.
320
240--
1/1
CO
-J
LJ
U
s
ISO--
BO ••
TIME
Figure 8. Snap-Bock Analysis -- Force Time
Htslory For the BoilOB Bag Span
490
-------
EXPERIMENTAL FIELD DATA
Visually observed tension spring and bag movements along with the
Finite Element model predictions of bag, tension spring, bag support
structures and filter cake responses during the reverse air cleaning cycle
prompted field "Insitu" testing to further quantify and support the
proposed theories.
Testing was performed at a utility flyash baghouse operating at a flue
gas temperature of 270°F., with cleaning cycle initiation at a baghouse
compartment pressure drop of approximately 4 inches I^O. The three acid
resistent coated filter bags tested were 12 inches diameter and 34 + feet
long. Testing was accomplished via the setup schematically shown on Figure
7, where a specifically designed load cell was placed in the bag suspension
system between the tensioning hardware and the top bag cap. Therefore,
monitoring of the spring/top of bag forces was enabled.
A typical measured bag tension history is shown on Figure 8, with the
outlet, reverse air, and reverse air relief valve movements shown. Several
important items to note include:
(1) The gradual rise in spring tension at the onset of the reverse
air valve opening.
(2) The sharp gradient starting at the reverse air valve fully opened
position.
(3) The high (approximately 2.5 times initial bag tension) peak
spring/top of bag tension.
(4) The spike-like peak duration most probably due to the force
generated by the falling dust cake.
r
STANDARD BAG
TENSIONING ASSEMBLY
LOAD CELL
TO OSCILLO6RAPH1C
RECORDER
12 INCH. DIA..
FILTER BAG
Figure 7. Schematic of Bag Tension Measurement Setup
491
-------
U
2
OUTLET VALVE OPENED
RA VALVE CLOSED
RA RELIEF VALVE OPENED
RA RELIEF VALVE CLOSED
RA VALVE OPENED
OUTLET VALVE CLOSED
100 200
SPRING TENSION CLBS.3
300
Figure 8. Typical Bag Response During Cleaning
492
-------
CONCLUSIONS
Both reverse gas flow and the falling porous dust plug contribute to
filter bag loadings during typical cleaning cycles.
The internal decompression effect contributes significantly to
structural loadings. Proper attention in the design process is required
to insure against low cycle fatigue failure of the suspension structure
and hardware.
The effect of variances in initial tension is narrow banded making
economic justification for hardware capable of precision adjustment
difficult. Precision adjustment hardware breeds maintenance headaches.
The magnitude of initial tension is extremely important in reducing
dynamic snap back of the fabric filter bag and damage to the material
fibers.
The work described in this paper was not funded
by the U.S. Environmental Protection Agency and
therefore the contents do not necessarily reflect
the views of the Agency and no official endorsement
should be inferred.
493
-------
DESIGN CONSIDERATIONS FOR BAGHOUSE - DRY S02 SCRUBBER SYSTEMS
by: Owen F. Fortune
and
Richard L. Miller
General Electric Environmental Services, Inc.
Lebanon, Pennsylvania 17042
ABSTRACT
The combination of a reverse air cleaned baghouse and a dry sulfur
dioxide scrubber is attractive because of the ease with which the baghouse
can remove over 99.9% of the high solids loading in the gas stream exiting
the scrubber, and because sulfur dioxide scrubbing continues to occur in the
filter-cake on'the bags. However, in order to avoid having gypsum deposits
shortening the usable life of the bags, close attention has to be paid to
several design considerations. Among them are system response to boiler tube
leaks, approach to saturation temperature, reheat system, avoidance of
condensation in reverse air cleaning system, plant maintenance problems,
changing boiler loads, and changing sulfur dioxide concentrations. Experi-
ence-based design strategies to deal with these issues are discussed in this
paper.
494
-------
INTRODUCTION
In less than ten years, the reverse-air cleaned, fiberglass bag fabric
filter has become accepted, reliable technology for power plant dust emis-
sions compliance. In the last few years, the utility baghouses1 role has
been expanded to become part of unified fly ash/sulfur dioxide removal
systems. When a "wet" S02 scrubber is used, the baghouse remains immediately
downstream of the air preheaters, and design modifications are not needed.
However, when a "dry" spray absorber S02 removal system is used, the
baghouse(s) are placed between the scrubber outlets and the induced draft
fans, and must be designed to function for the dust loadings and close
approach to saturation temperature of the spray absorber exit flows.
In 1980, we at General Electric Environmental Services, Inc. (GEESI) and
Anhydro A/S of Copenhagen, contracted to supply the City of Marquette,
Michigan, acting through Lutz, Daily, and Brain, with a spray absorber/
baghouse system for their new 44 Megawatt Shiras No. 3 plant. Some of the
system design parameters are given in Table 1, and the completed unit is
shown in Figure 1. Since Shiras No. 3 will be coming on-line in the next
month, it is timely to recount the issues that were considered in the design
of the baghouse, and how they would apply to spray absorber/baghouse systems
for boilers as large as 800 megawatts.
TABLE 1. CITY OF MARQUETTE SHIRAS NO. 3
• PROCESS
44 Megawatts
226,000 CFM at 265° F to spray absorber
167,000 CFM at ]75° F to baghouse
1500 PPM S02
15% Ash
80% S02 removal guaranteed
• Baghouse
75' x 45' in plan; 62* high baghouse compartments
168 bags per compartment
bags are 12 inch diameter x 35 feet long
140,000 square feet of filtering surface
• Spray Absorber Vessel
36* diameter x 71' high
Central 7800 RPM rotating wheel atomizer
200 HP vertical AC motor
495
-------
Figure 1. GEESI Spray Absorber/Baghouse System, City of Marquette, Michigan
Shiras No. 3 Station
CHOICE BETWEEN BAGHOUSE AND ELECTROSTATIC PRECIPITATOR
Perhaps the first issue to be considered is why one should use a
baghouse instead of an electrostatic precipitator in the first place. Since
the gas temperature and humidity out of the spray absorber are essentially
constant regardless of boiler load, it would seem to simplify the process of
sizing of a precipitator for this application. Also, the very high gas
moisture content is beneficial to the electrostatic precipitation process.
Indeed, it seemed that SCAs in the range of 400-500 would be sufficient to
insure dust emission levels under 0.03 Ib/MBTU.
However, in investigating the issue, we found two strong reasons that
will almost always weigh the decision in favor of a baghouse for a new power
plant. The first is that the dust removal equipment must be capable of
maintaining compliance, even if the S02 removal system is off-line. For low
sulfur western coals, this means SCAs in the 600-800 range. Compared to
precipitators in that size range, a baghouse will almost always be the more
economical choice (1).
496
-------
The second strong argument for teaming baghouses with spray absorbers
is that about three times as much S02 is removed from the gas passing
through a baghouse as through a precipitator operating at the same approach
temperature. In other words, a baghouse will account for about 12% of the
total S02 removal across the system, while a precipitator will only account
for about 4% of the reductions in S02 emissions.
This may seem like a small consideration, but present worth evaluation
factors on the order of ten million dollars per ton per hour of lime consumed
are currently being used for utility boiler S02 scrubbing systems. For an
800 MW boiler burning 1% sulfur coal, approximately 6 tons of lime will be
consumed each hour at full load. Thus, being able to operate with an 8%
reduction in lime consumption would result in operational savings valued in
today's dollars at 5 million dollars.
If the station were burning a 3% sulfur eastern coal, the lime consump-
tion would increase by a factor of six (as required S02 removal efficiency
increases from 70% to 90%). Thus, the financial incentives towards using a
baghouse become even more pronounced, even though a smaller precipitator
sizing (i.e., 450 SCA) would be appropriate.
Some air pollution equipment vendors (2) maintain that precipitators
can be reliably operated at lower approach to saturation temperatures than
baghouses, but other vendors, including GEESI, who use central rotating atom-
izers for fragmenting and dispersing the lime slurry into the boiler gas flov^
have not had this problem (3,4,5,6).
Once the decision has been made to use a baghouse, the two main baghouse
design factors to be considered are:
(a) To properly size the baghouse to handle the higher density,
higher dust loading, gas exiting the spray absorber(s),
and
(b) To design the system so that operation will always be
safely above the saturation temperature.
BAGHOUSE SIZING
At first glance, the fact that the gas entering the baghouse will be
100° to 150° F cooler than the gas exiting the air preheaters - and , hence,
seems to have only 80% its flow volume - would argue that no modifications
need to be made to the usual baghouse sizing procedures for the spray absorb-
er FGD application. Reality is, however, that
(a) the mass flow rate of the boiler gas is increased by as much
as 6% by the water evaporated in the spray absorber(s), and
497
-------
(b) a considerable amount of new solids are formed in the spray
absorber, due to the combination of the gaseous sulfur diox-
ide and calcium from the atomized slurry, to form calcium
sulfite and calcium sulfate. In addition, as much as 25%
reductions in lime consumption are realized in a recycle
process by first slurrying and then reinjecting as much as
50% of the solids collected in the spray absorber and bag-
house hoppers.
(c) Finally, the density of the gas entering the baghouse has
increased by at least 20%, and will influence the baghouse
pressure drops proportionately.
These factors can combine to easily double the dust loading to the baghouse
and, thus, result in either substantially higher baghouse pressure losses, or
substantially larger baghouses.
A positive factor in holding down baghouse sizes is to design the spray
absorber so that a considerable portion of the solids enter and formed in the
vessel are captured in the spray absorber hopper. The standard Anhydro reac-
tion vessel is shown in Figure 2. A tapered inlet scroll and guide vanes are
used to set up a cyclonic gas flow pattern which is reinforced by the rotary
momentum of the atomized slurry spray and a near horizontal flow exit is
positioned at two-thirds the height of the-conical hopper. The controlled
cyclonic flow pattern, combined with the Vortex holding effect of the exit
duct, results in
(a) solids mechanical collection efficiencies of from 45% to 55%,
and
(b) several hours of hopper storage time to insure vessel maintain-
ability.
The use of this type of spray absorber vessel, rather than a vessel that
passes on all, or most, of the solids to the baghouse(s), reduces the bag-
house inlet dust loading to levels similar to that at the air preheater exit.
For instance, in the Shiras No. 3 design, the maximum load dust loadings are:
STATION DUST LOADING (tons/hour)
Air Preheater Exit 4.4
Spray Absorber Vessel Exit 5.2
Baghouse Inlet 5.3
Hence, sizing the baghouse using arbitrary rules (such as gross air-to-cloth
ratios of 1.5:1 or 2.0:1 net-net with reverse air), that ignore the processes
occurring in the FGD vessels is poor practice. It is necessary to understand
the consequences to the baghouse(s) of the various different types (multi
498
-------
Figure 2. Anhydro Spray Absorber Vessel
499
-------
two-fluid nozzle, multi-rotating atomizer, and central rotating atomizer) of
atomizer and hopper configurations, worst case coal sulfur content, and their
resulting impact on baghouse design.
REHEAT
Another factor which can have a significant impact on baghouse design is
the practice of adding ,hotter gas to the spray absorber exit flow for fear of
sub-dewpoint operations occurring in the baghouse or induced draft fans.
Many different-ways of "reheating" the FGD vessel outlet flows have been
proposed. A few of the more logical ones are:
• Bypass of part of the economizer outlet flow around both the air
preheaters and spray absorber vessels.
• Bypass of part of the air preheater outlet flow around the spray
absorber vessels.
• Reheat of baghouse exit gas, using a "waste" steam heat exchanger
to elevate temperature.
Considering that air pollution control manufacturers have been able to
demonstrate stable rotating atomizer spray absorber operation at 20° F
approach temperatures, and that about 10° F of reheating occurs across an
induced draft fan operating at a 20" W.G. head, and that a well built
baghouse - particularly one whose compartments have been pressure tested
before being insulated - will have a temperature drop of less than 10° F, it
is difficult to see why it is necessary to boost the spray absorber exit
temperature before entering the baghouse when the boiler is operating at, or
near, full load. Particularly since doing so will decrease the amount of
S02 removed across the baghouse and, hence, the amount of lime consumed in
the spray absorber.
However, as the mass flow through the system decreases, thermal losses
through the baghouse will increase and can reach 15 to 20° F, even in a low
inleakage, well insulated system. And at low boiler, lime consumption is
less of an issue.
These factors argue for selection of a low capital cost, minimal capa-
city (10e F) reheat system. GEESI feels that the best way of supplying this
reheat capacity is to bypass a few percent of the economizer outlet around
the air preheaters. This insures that 450° F to 650° F temperature differ-
ential is available to elevate the baghouse gas inlet temperature with little
capital cost beyond thermocouples, well insulated ductwork, and isolation and
modulation dampers. Neither steam piping and valves nor electric heaters
need be purchased and maintained.
PREVENTION OF THERMAL LOSSES
Using baghouse gas temperatures 20° F above saturation reinforces the
500
-------
importance of minimizing air inleakage, and external and internal insulation
during construction of the baghouse is emphasized. Pressure leak testing of
the compartments before they are insulated, durable door gasketing and door
insulation shields, and minimizing the number of baghouse expansion joints,
all become even more important than they are for straight fly ash collection
baghouses.
In particular, the issue of compartment internal insulation becomes of
greatly increased interest and concern. If a compartment with uninsulated
internal walls is taken off line for inspection, then the wall temperature in
the adjoining compartments will go below the gas dewpoint and there will be
condensation on the on-line compartment side of the wall. This argues that
full height internal insulation is needed on compartment walls to prevent
wall and tubesheet corner condensation.
In a similar manner, heating the bottom of the baghouse hoppers to
prevent condensation and insure free flowing ash removal becomes more
important than in a straight fly ash baghouse, as does the importance of
emptying the hoppers once a shift. In areas with severe winter climates,
consideration should be given to insulated hopper enclosures in order to
minimize the worst case sizing of the hopper heaters.
LOW LOAD OPERATION
For boilers larger than 200 MW, it is probable that more than one struc-
tural baghouse will be used per boiler. Discussions have appeared in the
literature (7) recommending that at low load, entire baghouses be taken off-
line and isolated in order to minimize thermal losses due to having greatly
reduced mass flow rate per square foot of insulated external wall in each
baghouse. It is true that a baghouse with, say, a 10° F loss at 100% boiler
load, will have about an 18° F loss at 25% load. However, this low load
condition can be compensated for by either operating the on-line spray
absorber vessels at a higher approach temperature, or using a minimal capa-
city reheat system.
The reason that this is preferable to simply, isolating baghouses, or
compartments, is that a typical sized, 4000-bag fabric filter will have about
two tons of water vapor in it when brought off line. Within a few hours, all
of the water will condense as the baghouse cools down, unless an automatic
auxiliary purge system is added to the baghouse to replace the trapped boiler
gas with ambient air. Then, when boiler load increases, the baghouse must be
reheated before being brought back on line.
It is difficult to see what is gained through this elaborate procedure,
which significantly increases the capital cost of the baghouse by adding
massive (i.e., 15' x 12') inlet and outlet isolation guillotine dampers, as
well as the new ductwork and many-compartment dampers of the automated purge
system, while doing nothing to improve system performance.
501
-------
RESPONSE TO SYSTEM UPSETS
The presence of a spray absorber upstream of a baghouse protects the
baghouse against the consequences of many of the minor upsets that occur in
boiler operation during the course of a year. For instance, a properly
designed spray absorber system minimizes the consequences of minor boiler
tube leaks, since it will reduce the amount of recycle, or dilution, water as
the boiler gas humidity rises in order for the spray absorber outlet temper-
ature to stay at the specified approach to saturation temperature. Also, if
one air preheater wheel stops turning, the ability of the spray absorber
system to reduce the gas temperature by at least 150° F, will keep the aver-
age gas temperature entering the baghouse below 500° F and, hence, not even
require bypass of the baghouse.
Operational upsets in the spray absorber systems - such as a lime slurry
valve failing open - that could adversely impact the baghouse by causing sub-
dewpoint operating conditions, can be controlled by vessel outlet temperature
and humidity monitoring. For example, the GEESI Two-Loop Control system,
shown in Figure 3, would respond by closing off the vessel recycle flow, as
well as simply alarming the failure.
/ u. %
DRY FLUE GAS DESULFURIZATION SYSTEM
SPRAY ABSORBER/FABRIC FILTER
Figure 3. GEESI Two-Loop Control System
502
-------
BAG FABRICS AND COATINGS
It will still be necessary for the baghouse to be able to reduce dust
emissions to below 0.03 Ibm/MBTU, regardless of whether or not the'spray
absorber is on-line. Thus, the bag fabric must be able to withstand gas
temperatures above 300° F, as well as near saturated gas operation at 140 to
170° F. These dual requirements lead to fiberglass remaining as the most
cost effective bag material in utility fabric filters.
Since baghouse dust inlet loadings will be higher in the combined S02/
particulate removal system, GEESI believes that 13-1/2 ounce, rather than
9-1/2 ounce, fabric should be used because of its greater durability, as
shown by flex testing.
The value of the bag coating now chiefly lies in its ability to act as
a lubricant between glass fibers. Thus, even in the combined systems'
highly alkali filter cakes, the so-called "Acid Resistant" coatings are
preferred, since they flow more evenly during application and, hence, cover
a higher percentage of the glass fiber surface area than do the earlier
coatings, such as silicone-graphite-tefIon, or pure Teflon-B.
BAGHOUSE MAINTENANCE CONSIDERATIONS
The operation of either a wet or dry S02 scrubbing system imposes a
significant load on a plant's Maintenance and Operations Departments and,
hence, it is of increased importance that baghouse maintenance issues be
minimal.
The best way to insure this is to invest the time needed to keep the
system automatic temperature, S02, and humidity monitors in good calibration.
In particular, humidity sensors require periodic attention. GEESI believes
that, at present, extractive sampling monitors give the most reliable infor-
mation from dust laden flows. Thus, care should be taken that insulation and
heat tracing of the sampling lines is adequate, as well as frequently check-
ing the system's calibration.
The payoff for doing this will be avoidance of hopper pluggage and bag
damage problems caused by sub-dewpoint operation.
SUMMARY
One very general conclusion that can be drawn from working out the
design of a 'composite dry spray absorber/baghouse system is that the same
factors which lead to good straight baghouse design, simply become more
important for the composite system:
• Sharply evaluate the design point dust loading, mean particle size,
gas density, etc., for sizing the baghouse, rather than just rely-
ing on rule-of-thumb formulas.
503
-------
• Use and maintain gas temperature and humidity sensors to avoid
sub-dewpoint operation for even short periods of time.
• Use insulation on compartment internal walls to prevent local-
ized condensation.
• Heat hopper bottoms to head off potential pluggage problems.
• Pressure test compartments during erection to eliminate sources
of ambient air inleakage due to skipped welds, or poor door
gasketing.
• Instrument the system so that a mimi-computer can control feedback
loops to avoid operational problems. Set up agreed upon calibra-
tion procedures to insure that the computer inputs are reliable.
• Use fiberglass bags with protective coatings.
A baghouse designed along these lines should be, a high-reliability
component in a composite spray absorber SC>2 removal/fabric filter system.
The work described in this paper was not funded
by the U.S. Environmental Protection Agency and
therefore the contents do not necessarily reflect
the views of the Agency and no official endorsement
should be inferred.
-------
REFERENCES
1. Campbell, K.S., et al., Economics of Fabric Filters Versus Precipita-
tors. Electric Power Research Institute FP-775, June, 1978.
2. Wilkinson, J. M. and Tonn, D. P., Baghouse vs. Precipitator for Dry
Scrubber Systems — Pilot Study Results. 4th International Coal Utili-
zation Exhibition & Conference, Houston, Texas, December 1981, Vol. 3,
p. 168.
3. Parsons, E. L. , Hemenway, L. F., Kragh, 0. T. , and Brna, T. G., SO
Removal by Dry FGD. Proceedings: Symposium on Flue Gas Desul-
furization — Houston, October 1980, Vol. 2, p. 801.
4. Parsons, E. L. , Boscak, V., Brna, T. G., and Ostop, R. L., S02 Removal
by Dry Injection and Spray Absorption Techniques. Third Symposium on
the Transfer and Utilization of Particulate Control Technology, March
1981: Volume 1. Control of Emissions from Coal Fired Boilers, p. 303.
5. Samuel, E. A., Lugar, T. W., Lapp, D. E. , and Fortune, 0. F., Dry FGD
Pilot Plant Results: Lime Spray Absorption for High Sulfur Coal and
Dry Injection of Sodium Compounds for Low Sulfur Coals. Paper pre-
sented at 1982 Symposium on Flue Gas Desulfurization, Hollywood,
Florida.
6. Meyler, J. A. and Felsvang, K., One Year of Operation of the Riverside
Dry Scrubber. Paper presented at the 44th American Power Conference,
April, 1982.
7. Kaplan, S. M., et al., Dry Scrubbing at Northern States Power Company
Riverside Generating Plant. Paper presented at 1982 Symposium on Flue
Gas Desulfurization, Hollywood, Florida.
505
-------
RESULTS OF BAGHOUSE AND FABRIC TESTING AT RIVERSIDE
Bys H. W. Spencer III, Y. J. Chen, M. T. Quach
Joy Manufacturing Company
Western Precipitation Division
Los Angeles, California 90039
ABSTRACT
This paper presents the results of one year of baghouse and fabric
testing at the Riverside Dry FGD Demonstration Facility during 1981. Opera-
ting parameters and baghouse performance data are summarized. The results
of the fabric evaluation test are discussed. Pressure drop measurements are
reported and pressure drop predictions for various fabric filters based on
the experimental data are compared. Pressure drop predictions are compared
with overall baghouse pressure drop. Good agreement between the predicted
values and the actual measurements are reported. Measurements reported in
the paper cover operation with three different coals.
506
-------
INTRODUCTION
This paper presents the results of one year of baghouse and fabric tes-
ting at the Riverside dry FGD demonstration facility during 1981. The
Riverside dry FGD demonstration facility is located in Minneapolis, Minnesota,
at Northern States Power's Riverside Station. Operating parameters and bag-
house performance data are summarized for the period January, 1981 to December
1981. During this period, the Western Precipitation reverse air baghouse was
successfully operated downstream from the NIRO spray dryer.
EQUIPMENT DESCRIPTION DRY FLUE GAS DESULFURIZATION PLANT
Figure 1 is a schematic of the JOY/NIRO dry FGD Demonstration Plant. The
design conditions and technical data for the fabric filter portion of the
plant are given in Table I.
PROGRAM OBJECTIVES
The objectives of the JOY/NIRO test program at Riverside are listed
below:
1. Confirming the scale-up factors from pilot to full-scale operation.
2. Confirming the sensitivity of various operating parameters in a
full-scale system.
3. Demonstration of the system's operational flexibility, including
turn-down, start-up, shut-down, and normal operation.
4. Demonstration of long-term, low-temperature operation.
5. Evaluation of various fabric filter media with dry FGD application.
6. Demonstration of operation with various fuels, including low sulfur
Western coal and higher sulfur Eastern or Midwestern coals.
7. Demonstration of particulate and S02 removal capabilities with either
a baghouse or an electrostatic precipitator.
8. Characterization of particulate matter emitted from the system.
9. Determining the characteristics of the dry FGD waste product with
respect to disposal and utilization.
10. Characterization of the system operation and maintenance costs.
11. Optimization of energy and reagent consumption.
12. The establishment of system availability and reliability data.
13. The development of operating and maintenance procedures.
14. The training of power plant operators.
During 1981, a majority of these objectives were met. The test and oper-
ating record for the facility during 1981 is given in Figure 2. As shown,
tests were conducted with three fuels having sulfur contents ranging from .8%
to 3.2%. The characteristics of these fuels are summarized in Table II.
Photomicrographs of isokinetically-collected dust samples at the inlet of the
SDA and the inlet of the baghouse are displayed in Figures 3 and 4. Larger
507
-------
particle sizes were found at the baghouse inlet with the high sulfur fuel.
This is most likely explained by the higher alkalinity of the slurry that was
being spray-dried. Typical operating conditions for each of the coals are
given in Table III. The Bahco particle size measurements of the baghouse
dust samples contained in this table are consistent with the photomicrograph
data. Dust loadings given in the Table were measured at the inlet of the
baghouse.
Typical operating baghouse parameters with and without .the spray dryer
in service are given in Table IV. When the spray dryer is in operation, the
baghouse runs at a considerably lower temperature than for a non-FGD instal-
lation. Dust loadings are higher and collection efficiencies are also
higher. Data in Table IV shows increased emissions with broken bags. Bag
failures are discussed later in this paper and were confined to certain
types of bags. Figures 5 and 6 are the plots of the operating parameters
during selected periods which include the high sulfur tests.
FABRIC EVALUATION
One of the major emphases of the JOY test program is to evaluate a
variety of fabric materials in search of materials with longer life, lower
cost, and lower pressure drop characteristics. The bag arrangements in the
twelve compartment baghouses for the first phase and second phase of the test
programs are given in Figures 7 and 8. The design characteristics for each
of these materials are given in Table V. All the bags tested were fiberglass
except for the ones in Compartment 9, which were polyester. The polyester
bag type was not in service long enough for full test evaluation due to its
temperature limitation.
The test program was planned to evaluate the fabric strength and bag
life. New bags and bags which have been in service for certain periods of
time were removed for evaluation tests, including weight tests, breaking
strengths, Mullen burst, MIT flex test and permeability tests. In-situ bag
weights were measured to check the dust build-up trends on the bag surface.
A representative sample of in-situ bag weight of three bag types is plotted
in Figure 9. Significant variation was found between bag material types with
the texturized fiberglass bags having a much higher residual dust layer than
the non-texturized bags. Very little variation was seen between the front
and back of compartments. Measurements of the variation in bag weight from
compartment to compartment was also made and again, there was no significant
difference.
An example of decrease in breaking strengths and Mullen burst strengths
with time is plotted in Figure 10. This plot is for a non-texturized light-
weight teflon-finished fiberglass bag. As shown in a later plot, this bag
had low failure rates even though there was a measurable decrease in the
strength of the bag.
In Figure 11, the permeabilities of bag samples and corresponding bag
materials weights are compared for as received, vacuumed, and washed samples.
508
-------
The data shows that washing restored original permeabilities and weights.
Vacuuming the material only partially restored the permeability. As expec-
ted, the permeabilities of the material in service are considerably less than
the new materials.
BAG LIFE
Bag life is a major factor in determining the economics of fabric fil-
ters for utility applications. Figure 12 presents the cumulative bag fail-
ure rate as experienced at Riverside. Two of the bag materials, which have
teflon coatings, showed very low failure rates; failures were apparently
limited to bags damaged during initial installation. However, two bag
materials with silicone, graphite, teflon tri-coated finish had unacceptable
failure rates. The finish apparently did not protect the bags as well as the
teflon coating. The bags with high failure rates also have lower strength in
terms of MIT flex cycles. Fabrication errors also contributed to the fail-
ures which occurred at the top and lower bands, ring cover stitchings and
outer ring cover areas. The tri-coated bags have now been removed from the
Riverside baghouse. We are continuing to monitor the life of the teflon-
coated bags and other bag materials.
BAGHOUSE PRESSURE MEASUREMENTS
One of the major factors controlling the design of a fabric filter is
minimizing pressure drop. The Riverside baghouse has been instrumented to
measure flange-to-flange pressure drop, and pressure drop across the indivi-
dual compartments, outlet valves and inlet dampers. Pressure drop has been
monitored over a range of air-to-cloth ratios by reducing the number of
compartments in service under various boiler loads. Pressure drop was
measured during full flow periods, settling periods (when one compartment is
out for cleaning), and reverse air cleaning (when one compartment is out for
cleaning with the reverse air flow), corresponding to the gross, net and net-
net air-to-cloth ratios respectively. A sample trace of flange-to-flange
pressure drop is given in Figure 13.
Individual compartment flow rate and pressure drop were measured for
determination of filter drag, which is the ratio of compartment pressure drop
and air-to-cloth ratio. By measuring the dust concentration to the baghouse
and assuming uniform distribution of the dust between all compartments, the
incremental areal cake density was calculated for each compartment. The
incremental areal cake density is equal to the quantity of dust collected per
unit area between the cleaning cycles. A knowledge of the dependence of the
filter drag on areal cake density can be used to determine the effective drag
coefficient, SE, and specific dust-resistant coefficients K2 for each speci-
fic bag material as shown in Figure 14. Knowledge of these coefficients can
be used to predict pressure drops as a function of air-to-cloth ratio, dust
concentrations, and cleaning cycles. In Figure 14, filter drag is plotted as
a function of the incremental areal cake density for Type B bag materials at
Riverside. Residual drag, which is the real drag of a filter after cleaning,
is plotted at the incremental areal cake density just prior to cleaning
509
-------
instead of at the zero incremental areal cake density. The extrapolation of
the filter drag curve at zero cake density point, Sg, does not coincide with
the measured residual drag value, however, it is within the expected range.
The slope of the curve determines the value of K£. Another method using a
computer model developed by Western Precipitation was used to reduce the data
and calculate the effective drag, SE, and specific dust-resistant coefficient
K2- The pressure drops prediction based on these coeffficients as a function
of air-to-cloth ratio is presented as an example in Figure 15. Two pressure
drop comparisons are given. The solid curves show pressure drops across
different kinds of material without taking into account the effect of clean-
ing process. Considerable differences are found at higher air-to-cloth
ratios. The open symbols compare the pressure drop including the effect of
the cleaning process of a multi-compartment system. In this case, only minor
differences between different materials are found.
In Figure 16, total baghouse pressure drop at Riverside is plotted as a
function of air-to-cloth ratio and compared with computer predictions. Data
are given for three coals with and without SDA operation. There is no appar-
ent trend with respect to the three different coal types. The data shown
correspond well with pilot plant data and with the computer-predicted pres-
sure drops.
SUMMARY
In summary, the baghouse at Riverside has operated well. Baglife for
teflon-coated bags has been good, pressure drops have been predictable and
reproducible. The baghouse as a particulate collector for dry scrubber
should be considered as a demonstrated technology.
REFERENCES
1. Kaplan, Steven M., Chen, Yang-Jen, Sannes, Carl A., Jr., "Dry Scrubbing
at Northern States Power Company Riverside Generating Plant," EPA/EPRI
Symposium on Flue Gas Desulfurization, May 17-20, 1982, Hollywood,
Florida.
2. Spencer, H. W. Ill, Brown, Bert, Chen, Yang-Jen, "Experience with Bag-
house for Dry FGD Service," 74th APCA Meeting, June 21-26, 1981,
Philadelphia, Pennsylvania, Paper No. 81.95.
The work described in this paper was not funded by the U.S. Environmental
Protection Agency and therefore the contents do not necessarily reflect the
views of the Agency and no official endorsement should be inferred.
510
-------
TABLE I
THERM-0-FLEX FABRIC FILTER DATA SUMMARY
PLANT NAME: Northern States Power Company
Riverside Generating Plant
LOCATION: Minneapolis, Minnesota
DESIGN CONDITIONS (FGD)*
Gas Volume (max acfm) 420,000 (540,000)
Design Temperature (°F) 500
Outlet Loading (gr/106 Btu) 210
Pressure Drop (in. VWC) 6
Design Pressure (in. VWC) -30 to +20
Total Filter Area (sq. ft.) 321,200
Total Filter Ratio 1.3111 (1.68:1)
TECHNICAL DATA
No. of Fabric Filters 1
No. of Rows 2
No. of Compartments /Row . ., 6
Filter Bags:
Total 3000
Per Compartment 250
Diameter (in.) 12
Length (ft.) 35
* Values in parenthesis are for non-FGD system operation. Where none are
shown, values can be assumed to be the same.
TABLE II
FUELS USED IN SDA/BH TESTING
Average
Sulfur %
Aah%
Moisture %
BTU/lb.
Normal
85%CoMrip'
(Rotebud, Montani)
(McKay Seam)
15% Petroleum Coke
1.37
8.33
22.0
9,500
Hi-Sulfur
•HnoteCoal
3.19
10.36
12.76
10,930
Low-Sulfur
Sarpy Creek Coal
(Sarpy Creek, MT)
(Robinson Seam)
0.79
9.20
25.33
8,614
511
-------
TABLE III
TYPICAL OPERATING CONDITIONS
• AT Ad. Sat (T)
- Feed Rate (GPM)
• FeedSottd*
• Dust Loading (gr/ACF)
• Bulk Penalty (tapped) («W)
• Bahco Mast Mam (pm)
• Flow Rate (ACFM)
M-8
19.2
1.17
13&5
35.49
10.2
SOX)
17
390,560
Sarpy Creek
17
1.72
140.33
42.42
11.8
02.5
9.4
347,200
Colstrip
17.3
1.34
121.85
44.77
9.9
56.2
8.4
379,950
TABLE IV
BH Operating Range
—WHhSDA:
• Inlet Temperature (T) 140—ITS
• MetGMVokmio(ACFM) 219.000 — 444,000
• Inlet DuttLoadtog (gr/ACF) 4J7 —12.1
• Outlet Dutl Loading (gr/ACF)* 0.0010 — 0.0192-
• EMctoncy (%)• MM' — MM
-Without SOA:
• tale! Temperature fF)
• mlel Gee Volume (ACFM)
• Met Oust LoedhtgfT)
270 — 310
182,000 — 458,000
1.1 — IS
•Mgh outlet toadMtg end tow efficiency ere due to broken beg*.
TABLE V
FABRIC SPECIFICATIONS
Avg. Finish
Wt.
Oz./Sq. Yd.
FABRIC A
FABRIC B
FABRIC C
FABRIC D
FABRIC E
FABRIC F
FABRIC G
FABRIC H
FABRIC J
FABRIC K
8.4
9.3
8.8
8.4
9.3
9.5
13.5
9.5
9.5
9.5
Silicone
Graphite
Teflon
Teflon
Teflon
Grafosil
—
Teflon B
Acid Res.
Teflon B
Acid Res.
Teflon
Material
Glass
Glass
Glass
Glass
Polyester
Glass
Glass
Glass
Glass
Glass
512
-------
FIGURE 1
FLOW DIAGRAM - DRY SCRUBBING SYSTEM AT RIVERSIDE
FIGURE 2
TESTING & OPERATION RECORD
SARPY CREEK
TRANSITION O-
HIGH-S —-
SDA-ESP •»«•
DEMO.
NSP EQUIP. DOWN • -
PARA. • —
INTERPOLL O
FLY ASH -O
O
O
O O CX">O
1st 1st 1st 1st 1st 1st 1st 1st 1st 1st 1st 1st
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC 81
Note:
One day event
513
-------
FIGURE 3
Hi Sulfur
BH INLET
DUST SAMPLES
Colstrlp
Sarpy Creek
FIGURE 4
Hi Sulfur
SDA INLET
DUST SAMPLES
Colstrlp
Sarpy Creek
514
-------
FIGURE 5
FGD OPERATION PARAMETERS
AUG SEP SEP OCT NOV
21 25 26 3 89 10 11 22 23 24 26 27 1 2 23 3 9 10 11 12
160
130
400
200
100
80
60
40
20
0
40
30
20
10
BH OUTLET TEMP.
BH OUTLET SO,
AT AD. SAT.
Jim
IS II .
FIGURE 6
BH OPERATION PARAMETERS
JULY AUG SEP OCT
. 2« 30 4 6 10 13 20 24 3 9 10 21 23 25 _ I 6 15
INLET GH ACF
FEED SOLID %
L
CO
O
co
PARA
-HIGH SULFUR-
515
-------
FIGURE 7
NSP, RIVERSIDE BAGHOUSE
BAG CONFIGURATION
March - Dec., 1981
1
A
3
C
5
A
7
0
9
E
11
A
2
B
4
B
6
B
8
B
10
B
12
B
lil
FIGURE 8
NSP, RIVERSIDE BAGHOUSE
BAG CONFIGURATION
Second Start-up — June, 1982
1
H
and
F
2
B
3
C
4
B
5
K
6
B
7
K
9
J
8
B
10
B
11
G
12
B
lil
516
-------
FIGURE 9
BAG WEIGHT
FRONT VS. REAR OF COMPARTMENT
50
w^
g 40
i-
ui 3°
O
to 20
MEW
FRONT REAR
APR 6
JUN9
1981
TEST DATE
A (Comp. 1)
SEP 15 FEB11
1982
700
600
500
0
FIGURE 10
TYPE C
FLEX CYCLES
WARP
BREAKING STRENGTH
1580 2790
BH OPERATING TIME
4450 MRS
517
-------
1500 2790
BH OPERATING TIME
MRS
FIGURE 12
0
BAG FAILURES
Jan. Feb. Mar. Apr. May Jun ,)ul
DISCOVERY DATE
Aug. Sep. Oct. Nov. Dec.
1981
518
-------
FIGURE 13
TYPICAL BAGHOUSE SYSTEM PRESSURE DROP
O
M
Q
01
W
(Q
-------
o
CN
a
w
m
10
§
U
O
«
o
w
w
w
w
FIGURE 15
COMPARISON OF PRESSURE DROP PREDICTION
Fabric A
Fabric B
Fabric C
A O
B A
C D
AIR-TO-CLOTH RATIO
FIGURE 16
AP VS. AIR-TO-CLOTH
O J
(N
X
z
H
0,
o
B
a
i
U)
w
o, •
U
U)
o •
0
FULL
FLOW
COLSTRIP 0
HIGH SULFUR *
SARPY CREEK O
FLY ASH ONLY f
SETTLING
O
A
0
_|_
REVE
AIR
0
A
O
_i_
PREDICTED PRESSURE
DROP
MAX
UT?* 11
AIR-TO-CLOTH RATIO (FT/MIN.)
520
-------
REACTIVITY OF FLY ASHES IN A SPRAY DRYER/FABRIC FILTER
FGD PILOT PLANT
by
Wayne T. Davis, Randal E. Pudelek, and Gregory D. Reed
Department of Civil Engineering
The University of Tennessee
Knoxville, Tennessee 37996-2010
ABSTRACT
/
This paper summarizes the results of a study in which the reactivities of
23 fly ashes were evaluated in a pilot plant spray dryer/fabric filter sulfur
dioxide removal system. The primary objective was to determine the ability of
each fly ash (including lignite, subbituminous, and bituminous eastern and
western ashes) to remove S02 when placed in a water-based slurry and atomized
by a spinning disk atomizer into a spray dryer located on a slipstream from a
stoker-fired boiler.
Data are summarized in both tabular and graphical form including chemical
kinetic data as well as S02 removal efficiency indicating the enhancement in
efficiency resulting from use of fly ash. The S02 removal efficiency ranged
from 10-50% with only fly ash in the slurry.
INTRODUCTION
In the last five years the use of spray dryers to remove sulfur dioxide
from flue gas has moved from pilot plant to full scale applications. The
system consisting of a spray dryer and particulate removal device has
generally employed a Ca(OH)2 slurry which was atomized in the drying chamber
via a nozzle or spinning disk atomizer. It has been observed in these systems
that the fly. ash may also play an important role in minimizing the reagent
requirements, particularly if recycle is employed. The objectives of this
study, sponsored by the Department of Energy, were to 1) collect a number of
different fly ashes from various coal-fired power plants, 2) quantify the
ability of each ash to react with S02 in a spray dryer pilot plant, and 3)
identify the parameters responsible for the reactivity. To this end, 23
different ashes were studied, the results of which are reported herein.
521
-------
DESCRIPTION OF THE TEST FACILITY
The spray dryer/fabric filter pilot plant was installed on a Riley
spreader-stoker coal-fired boiler at the University of Tennessee steam plant.
A slipstream of flue gas with a nominal flow of 1,000 ACFM, taken from the
main ductwork, entered the system which consisted of a spray dryer with a
spinning disc atomizer followed by a fabric filter collector. The slipstream
was pulled through the pilot system by an induced-draft fan. A layout of the
system is shown in Figure 1.
The ductwork leading to the spray dryer contained ports for the injection
of ambient air and/or S02. Other ports labeled one, two and three respec-
tively, enable the monitoring of temperature, static pressure, and S02 con-
centration at the inlet and outlet of the spray dryer find the outlet of the
fabric filter collector, respectively.
Once the desired inlet S02 concentration and temperature was achieved the
flue gas then proceeded to the spray dryer inlet where it encountered the
atomization machinery. Atomization was accomplished by a Stork-Bowen high
speed AA-6 Spray Machine and a centrifugal atomizer. The water and fly ash
slurry passed from the pumping system through the spray machine housing into
the atomizer. Rotating at a high speed (17000-18000 RPM), the disc atomized
and distributed the slurry as a uniform, fine mist into the hot flue gas
entering the drying chamber.
Flue gas exiting the boiler entered the drying chamber through a set of
vanes concentric with the spray machine. These vanes imparted an angular
downward swirling motion to the air. The rotation of the swirling air was
opposite to that of the feed leaving the atomizer thus insuring intimate
mixing of the hot air and the fine particulate mist. The drying chamber was a
7 foot diameter vessel with a standard conical bottom which contained a rotary
valve for the removal of the spray dryer product.
The particulate-laden gas stream upon leaving the spray dryer entered the
fabric filter collector. This collector removed both the suspended fly ash-
S02 product and any other type boiler ash contained in the flue gas stream.
The bag house consisted of one filtration compartment containing four 32 ft. x
12 inch fiberglass bags (14.5 oz./sq. yd.). A low energy shaker mechanism was
used to clean the bags. It was operated manually at the completion of or
prior to a test.
The control room contained 1) a panel for the control of atomizer rpm and
2) other devices for the monitoring of the system flow rate and the slurry
feed rate. Also, monitored in the control room were the following:
1. S02 concentration (3 ports) Lear-Siegler SM800
2. Temperature (thermocouples)
a. Inlet/outlet spray dryer
b. Inlet bag house
c. Ambient temperature
d. Slurry feed temperature
522
-------
3. Slurry feed rate
4. Concentration of additive (i.e. fly ash percent)
5. Static pressure inlet/outlet of spray dryer/bag house
When the desired range of S02 concentration cannot be achieved, a manual-
ly controlled supplemental S02 injection system is used. This system was
located about 80 feet upstream of the spray dryer in the inlet duct to allow
for sufficient mixing. Inlet temperature control was maintained by a manually
operated
port.
ambient air dilution damper located upstream of the S02 injection
Monitoring of S02 was accomplished by extracting samples through heated
trace lines into the control room where the flue gas sample was drawn through
the sampling cell of a Lear-Siegler SM800. The concentration of S02 was
measured on a wet basis by this instrument. A dry basis concentration was
calculated later after the moisture content was determined.
Slurry preparation in this study consisted of gravimetrically weighing
and mixing of a known quantity of a fly ash into a known quantity of water.
The detention time was 25-40 minutes prior to conducting each test.
LABORATORY TEST PROCEDURE
The potential S02 removal capability of fly ash was
twenty-three fly ashes of different rank, i.e., lignite,
bituminous eastern and western ashes (see Table 1).
investigated using
subbituminous, and
Prior to the actual testing of each fly ash, laboratory analyses were
performed to quantify both the physical and chemical characteristics of the
ash. Table 2 shows the physical and chemical properties measured and the
instruments that were used. Chemical analyses conducted for total alkalinity,
calcium and sodium content, pH, and total hardness yielded information re-
garding the behavior of each fly ash in a water-based slurry. These analyses
were conducted at a temperature of 140°F(60°C) and a fly ash concentration of
1.45 pounds per gallon of water. These conditions were chosen in order to
simulate the condition to be tested at the spray dryer facility.
In addition, laboratory preparation of the fly ashes, according to ASTM
procedures (5), followed by analysis with an atomic absorption spectro photo-
meter determined the levels of calcium oxide (CaO), sodium oxide (Na20),
magnesium oxide (MgO), and potassium oxide (K20) as received. Using these
data an estimate of total alkaline metal oxides (TAMO) is obtained as follows:
Total g-moles Alkaline
Metal Oxides/lOOg. ash
g. CaO/lOOg. + g. Na^O/lOOg. + g. MgO/lOOg.
56.1 g./g-mole 62.0 g./g-mole 40.3 g./g-mole
q. K?0/100g.
94.2 g./g-mole
523
-------
Table 1. Origin of the Fly Ashes Investigated
Fly Ash
I.D.
Power Plant
of Origin
Type of Fuel
(approximate)
Source
ui
6
10
13*
14
ISA
18B
23B
24
27
34
36
37
38B
38C
41
42
43
44
45*
46
49
50
Belews Creek Steam Station bituminous
Bowen Steam Plant bituminous
Clay Boswell Unit 4 subbituminous
Nebraska Public Power Station subbituminous
Gerald Gentleman Station
Pacific Power and Light subbituminous
Wyodak Plant
Hunter Steam Plant bituminous
Cherokee #4 bituminous
Gallagher Station II bituminous
Public Service of Indiana
Texas Utilities Generating Co. lignite
Big Brown
New Madrid Power Plant #2 bituminous
Ohio Edison - Gorge bituminous
Harrington Station: subbituminous
Southwestern Public Service Co.
Laramie River Station subbituminous
Basin Electric Powwer lignite
Cooperative - Unit 1
Milton R. Young Station lignite
Center Unit 1
United Power Association lignite
Cooperative Power Association
Black Hills Power subbituminous
and Light Company
Otter Tail Power lignite
Hoot Lake Station Unit #2
Minnesota Power and Light Company subbituminous
Clay Boswell Station
Monifer Resources subbituminous
San Antonio Public Service
University of Tennessee bituminous
Steam Plant
Marshall Steam Station bituminous
Low sulfur eastern coal
eastern Kentucky
Big Sky Mine - Col strip Montana
Black Thunder Mine
Campbell County, Wyoming
Wyodak Resources - Wyodak Mine
Wilberg Mine - Emery County, Utah
Colorado western slope
Amax Aryshlre
Freestone County
Southern Illinois, Seam #6
Ohio strip mine
Black Thunder Mine
near Gillette, Wyoming
Cordero Mine
Consolidated Coal Co.
Stanton, North Dakota
Bankol-Noonan Mine
Center, North Dakota
Falkirk Mine
Underwood, North Dakota
Wyodak Mine - Wyoming
Knife River Coal Mining Co.
Beulah, North Dakota
Big Sky Mine.
Colstrip, Montana
Cordero Mine, Wyoming
eastern Kentucky
Low sulfur eastern coal
*These fly ashes were obtained from the same source, however, their production resulted from different
combustion conditions.
-------
Table 2. Fly Ash Physical and Chemical Characteristics
to
Ln
Fly Ash
I.D.
6
10
13
14
ISA
18B
23B
24
27
34
36
37
38B
38C
41
42
43
44
45
46
49
50
A A M
44M -
MMDfl
Cum)
12.0
14.8
9.4
9.5
15.0
7.8
14.9
12.3
9.5
9.2
15.3
7.4
11.5
15.6
12.0
9.4
18.0
11.0
10.0
8.5
12.9
14.0
C 1 . . -• ~ U
r ly asn
Geometric
Deviation
2.3
2.2
1.9
1.8
1.9
1.9
2.8
2.2
2.5
1.7
2.0
2.5
2.2
2.6
1.8
1.9
2.4
1.9
2.0
1.8
1.9
2.1
.
44 ground in a
Particle
Density
(g/cm3)
2.7
2.1
1.5
2.7
2.5
2.3
2.2
2.7
2.4
2.5
2.6
2.7
2.6
3.2
2.7
2.6
2.5
2.7
2.5
2.5
2.1
2.2
U-»11 m-!!!
oa i I mill
Surface
Area
(m2/g)
1.31
2.14
0.21
2.09
0.73
2.14
2.85
2.88
0.74
1.23
0.22
1.15
1.66
0.76
3.27
0.29
1.03
2.37
3.39
1.48
12.52
1.47
PH
5.2
6.7
11.2
11.0
10.6
11.2
11.3
11.4
10.3
4.4
10.5
11.3
11.2
10.8
10.7
11.4
10.9
10.9
10.7
10.9
3.6
4.8
T f\ Q
iU * y
Alkalinity
mg/1 . as
CaC03
200
500
3800
3500
1200
3400
1800
2900
1700
0
1400
1800
1500
1600
1100
3700
2100
1300
1600
1600
0
0
9i nn
Ca,
mg/1.
472
265
962
132
130
849
264
1151
415
604
245
85
123
274
321
887
179
340
302
128
175
321
3CI9
Na,
mg/1.
40
26
48
299
54
88
51
91
300
300
49
200
54
4200
8000
62
75
8200
58
100
150
52
ocnn
O3UU
Total
Hardness,
mg/1 . as Ca
547
-r
1094
210
160
1000
340
1208
547
717
283
170
245
368
396
981
274
396
396
170
--
358
AT)
*f / £.
Mass mean diameter and geometric standard deviation (Coulter Counter Model TAII)
Helium/air Pyncnometer (Micromeritics Model 1302)
•*
"Micromeritics Surface Area Analyzer
-------
This determination was made based on previous studies in which it was shown
that TAMO was related to sulfur retention in fly ash (Fiedler, 1). Table 3
contains both the percent of each compound in the ash (wet basis) and the
estimate of TAMO. Another term used in some of the correlations was the value
of TAMO less the sulfur in the ash:
This term was used to account for reaction that had already occured in the
as-received ashes.
FLY ASH TESTING IN THE SPRAY DRYER
Fly ash testing at the spray dryer facility was accomplished by mixing a
known mass of fly ash in a known quantity of water with sufficient residence
(mixing) time (greater than 30 minutes) to allow any soluble metal oxides to
go into solution in the slurry. Figure 2 is a typical example of the dissolu-
tion kinetics exhibited by each fly ash. This information was used to deter-
Table 3. Fly Ash Mineral Analysis Summary
Fly Ash
I.D.
6
10
13
14
ISA
18B
23B
24
27
34
36
37
38B
38C
41
42
43
44
44M
45
46
49
50
% S
0.43
0.32
0.49
0.73
0.70
0.45
1.87
1.22
0.66
1.04
0.36
0.78
0.72
0.
4.06
1.02
0.66
0.75
0.75
0.68
0.75
2.19
0.41
% CaO
7.79
5.44
13.96
22.91
18.96
9.72
7.88
8.66
14.99
6.76
0.98
24.95
21.81
13.95
18.01
18.96
6.96
23.80
23.80
11.63
28.90
4.02
5.75
% Na20
10.71
9.42
10.96
10.96
10.78
10.89
10.25
9.94
11.19
11.79
12.69
10.98
13.48
35.88
16.68
10.97
10.76
17.71
17.71
10.66
14.94
6.97
12.47
% MgO
0.97
0.91
2.79
3.49
3.79
2.14
1.18
1.91
2.40
0.97
0.39
3.59
2.97
5.58
3.98
3.79
1.00
4.57
4.57
2.71
3.98
0.94
1.44
%K20
2.53
2.36
1.79
1.29
1.50
1.94
1.68
0.76
1.70
2.90
5.56
1.00
1.29
3.69
1.52
1.90
2.49
1.14
1.14
1.45
1.39
1.94
2.78
TAMO
(moles/lOOg. Ash)
0.36
0.30
0.52
0.69
0.62
0.42
0.35
0.35
0.53
0.37
0.29
0.72
0.70
1.01
0.71
0.63
0.35 -
0.84
0.84
0.46
0.87
0.23
0.37
526
-------
mine the required duration of mixing and the effect of slurry preparation and
temperature. After mixing, the slurry was pumped to a final dilution tank
where it awaited final transfer into the spray dryer. Prior to and during the
slurry production procedure the following conditions were maintained for all
fly ashes tested at the spray dryer/fabric filter facility:
1. Inlet temperature range (300 ± 10°F)
2. Inlet S02 concentration (600-800 ppm)
3. Gas volumetric flow rate (1,000 acfm)
4. Slurry preparation temperature (130-140°F)
5. Slurry concentration (1.45 IDS fly ash/gallon H20)
6. Approach to saturation in spray dryer (20°F)
The total S02 removed by the system (spray dryer plus bag house) was
determined from inlet and outlet concentration data obtained from a Lear
Siegler SM800 S02 analyzer. These data were converted to mass/time and
reported in pounds/minute of S02 removed as well as g-moles/sec of S02. By
defining the quantity of S02 removed in this manner the existence of any
significant correlation with alkaline metal oxide content of the ash could be
identified.
Table 4 provides a summary of the actual efficiencies of S02 removal
across the spray dryer (EFFSD) and the system (EFFSYS) which included the
spray dryer plus the baghouse. As can be seen, the system ranged from 0-30%
for the fly ashes mixed in the slurry. A ball-milling of fly ash 44 increased
the efficiency from 18% up to 46%. It was interesting to note that this
improvement was a result of improved efficiency in the baghouse.
Two important checks were conducted periodically on the system during the
course of this investigation to minimize the errors involved in actual test-
ing. First, S02 balances were conducted at test conditions by measuring
concentrations at all three ports without any fly ash injection. A consistent
four percent reduction in the level of S02 was observed at the system's outlet
due to air inleakage. Secondly, water injection (no fly ash), via the spin-
ning disk atomizer, resulted in a two percent S02 efficiency. The values in
Table 4 and following tables represent values corrected for air leakage and
water.
RESULTS OF STATISTICAL ANALYSES
PRELIMINARY ANALYSES
The primary objective of this study was to identify the potential reac-
tivity of various types of ashes and determine their capability for S02
removal in the spray dryer/fabric filter system. It was hoped that a high
level of correlation could be found between the parameters and the S02 removal
rates in the system. However, the chemical analyses of the slurries, as shown
in Table 2, were measured on the filtrate from the slurries and did not re-
flect any reactivity which might occur on the surface of the suspended fly
ash. Figure 3 shows that the actual number of moles of S02 removed per second
527
-------
Table 4. S02 Removal Efficiency by a Fly Ash Slurry
Fly Ash
I.D.
6
10
13
14
ISA
18B
23B
24
27
34
36
37
38B
38C
41
42
43
44
45
46
49
50
EFFSD
0
12
7
35
22
26
4
7
3
6
2
27
19
13
19
10
23
17
13
14
13
2
EFFSYS
1
7
17
30
21
23
2
6
5
6
0
26
26
15
21
9
26
18
24
13
29
1
44M 17 46
was generally equal to or greater than the calcium plus sodium feed rates into
the spray dryer, (expressed in moles/sec as Ca).
The ASTM analyses of total alkaline metal oxides (TAMO-S), on the other
hand, were a measure of the total values contained in the fly ashes and as
such over-predicted the amount available for reaction due to the insolubility
of the ashes. This is evident in Figure 4 which shows a graph of S02 removed
in the system versus TAMO-S. The amount of S02 removed was typically only
one-tenth of the total amount which could have been theoretically removed
assuming a complete reaction with TAMO-S.
A preliminary statistical analysis of sulfur removed versus individual
parameters such as TAMO, alkalinity, % calcium, % sodium, % calcium & sodium,
hardeness, hardness + sodium, and surface area yielded correlation coeffi-
cients, R2, in the range of .03-0.28. Upon dividing the fly ashes into the
three coal ranks (lignite, subbituminous, and bitumminous), the lignite coal
yielded values of R2 of 0.61, 0.61, and 0.85 when linear correlations were
conducted of sulfur retained in the system versus slurry sodium, slurry
calcium + sodium, and surface area, respectively. Thus, further analyses were
528
-------
concentrated on multiple regressions in which the fly ashes were separated by
coal rank. Also in an effort to account for the previous history of the fly
ash it was decided to use the parameter total sulfur (TS) removal (sulfur in
the ash as received plus sulfur removed in the spray dryer/fabric filter
system) rather than just the latter.
PREDICTION OF TOTAL SULFUR RETENTION
As discussed earlier, various sorbents containing one or all of the
alkaline metal oxide compounds (CaO, Na20, MgO, and K20) have demonstrated
success in the removal of S02. These metals have been suggested as the
primary constituents that occur in the formation of sulfate or sulfite end
products [Rosenberg, et.al. (2) and Ando, et.al. (3)]. Davis and Fiedler (4)
used this information to quantify the ability of fly ash to retain sulfur in
the boiler after the coal was burned. Figure 4 shows the curve developed by
Davis, et.al. (4) with the fly ashes used in this study plotted in the same
manner. The ashes in this investigation exhibited less sulfur retention than
found in the previous study presumably due to the lack of lignite data in the
previous study.
It was this concept that provided the final basis for the modelling of
total sulfur retention. Expressing the amount of S02 removed by the spray
dryer/fabric filter system as grams of sulfur retained per 100 grams of ash
and summing that with the quantity of sulfur already in the fly ash as re-
ceived (see Table 2), determined the total sulfur retained. Table 5 contains
the sulfur retention data for both the fly ash as received and for the fly ash
slurry injected into the system.
The total sulfur retention (the sums of the two components) is included
in Table 5 and is referred to as TSL, TSSB, and TSB for the lignite, subbi-
tuminous, and bituminous data, respectively.
Multiple regression analyses were conducted to determine the correlation
and. relationship between TS and the three parameters of surface area of fly
ash, slurry alkalinity and TAMO. The following equations were found to ade-
quately describe the sulfur retention for the lignite and bituminous fly
ashes:
TSL = 0.019 (SAI) + 125.320 (ALKAL) - 0.507 (TAMO) - 1.478 (Eq. 1)
TSB =0.005 (SAI) + 53.397 (ALKAL) + 9.758 (TAMO) - 3.336 (Eq. 2)
where
TSL = total sulfur retained by lignite ash (g/lOOg. fly ash)
TSB = total sulfur retained by bituminous ash (g/lOOg. fly ash)
SAI = surface area injected (m2/100g. of fly ash)
ALKAL = slurry alkalinity injected (moles/lOOg. of fly ash)
529
-------
Table 5. Sulfur Retention by Fly Ash
Sulfur Retained
Fly Ash by Boiler
I.D. g./100g. fly ash
Lignite
27
38C
41
42
44
44M
Subbituminous
13
14
18A
37
38B
43
45
46
Bituminous
6
10
18B
23B
24
34
36
49
50
0.66
0.00
4.06
1.03
0.75
0.75
0.49
0.73
0.70
0.78
0.72
0.66
0.68
0.75
0.43
0.32
0.45
1.87
1.22
1.04
0.36
2.19
0.41
Sulfur Retained
by SD/BH System
g./lOOg. fly ash
0.35
1.33
2.23
0.88
1.81
4.61
1.44
3.35
2.46
2.76
2.87
2.89
2.48
1.21
0.11
0.53
2.95
0.15
1.05
0.49
0.00
3.12
0.25
Total Sulfur
Retention
g./lOOg. fly ash
1.01
1.33
6.29
1.91
2.56
5.36
1.93
4.08
3.16
3.54
3.59
3.55
3.16
1.96
0.54
0.855
3.40
2.02
2.27
1.53
0.36
5.31
0.66
TAMO = total alkaline metal oxides injected (moles/lOOg. of fly ash)
The correlation coefficients, R2, for equation 1 and 2 were 0.90 and
0.95 respectively. The R2 for the subbituminous data was only 0.23,
suggesting that further analysis is needed to quantify the relationship.
DISCUSSION
The data and results presented herein support the concept that reactivity
is controlled by three factors, none of which is totally independent. For
example, the TAMO inherently affe.ts the concentration of calcium and sodium
ions which are available to go into solution. Likewise, the ajkalinity is a
measure not of the cations, but rater the anions (OH and C03 ). Thus, the
530
-------
physical significance of the developed equations is somewhat obscurred. It is
certainly reasonable that, as the alkalinity and/or TAMO are increased, the
potential for reaction in the spray dryer is enhanced. Likewise, as the
surface area is increased, the possibilities of physical adsorption and/or
surface reactions are increased. To achieve a higher level of confidence in
the above correlations, it is necessary to determine the mechanisms that are
controlling the reactivity.
The possible effect of surface area is best illustrated in fly ash 49
which had very low values of alkalinity, and TAMO, the lowest of any of the
ashes. Yet it had the third highest sulfur retention (5.31 g/100 gram). This
can be attributed to physical adsorption due to the highest observed surface
area (12.5 m2/g).
Fly ash 44 also provides some insight into the mechanisms that are acting
to remove S02- A light ball milling of this ash resulted in a 17% increase in
surface area and a 62% increase in alkalinity. The sulfur retention was in-
creased by 150% due to this action. It is unclear whether the alkalinity or
surface area enhanced the removal. It is clear however that the effect is
non-linear.
The failure to arrive at a suitable" equation for the subbituminous coal
(R2=0.23) may be due in part to the narrower range of values of total sulfur
retention for the fly ashes studied. The eight values only varied by a factor
of 2 whereas the lignite and bituminous data varied by factors of 6 and 10
respectively. The subbituminous data were input into equation 1 and 2 to
determine if the measured values fell between the values predicted by the two
extreme coal ranks. In all but one case, the data fell within or equal to the
values predicted by equations 1 and 2 within the errors of measurement.
CONCLUSIONS
The major conclusions of this effort are summarized below:
1. It was demonstrated that S02 efficiencies of up to 30% could be
achieved by simply mixing fly ash in water followed by injection
into a spray dryer/fabric filter FGD system. The feed rate of 1.45
Ibs per gallon is typical of the fly ash collected at a typical
loading of 3 grains/CF.
2. Although of a preliminary nature, the data suggest that the surface
area, slurry alkalinity and total alkaline metal oxides content
affect the removal. Equations were developed through multiple
regression analysis to illustrate that these parameters were of
value in predicting reactivity.
3. The fly ash preparation technique (ball mill vs. simple mixing) was
shown to have a significant effect on reactivity for one lignite fly
ash. The removal across the FGD system was improved from 18% to 46%
due to ball milling of the fly ash.
531
-------
This study indicates that it is possible to develop statistical relation-
ships which can be used to determine directions to be taken to enhance S02
removal efficiency. The alkaline metal content and surface area of fly ashes
has in many cases gone untapped as a source of reactive reagents. Further
research is needed to determine the following:
1. The optimum fly ash ratio for S02 removal in a spray dryer.
2. Improved extraction methods to extract alkaline metal oxides from
fly ash to increase reaction potential of the slurry.
3. Document the advantages of grinding the fly ash in a ball mill and
thus, produce corresponding increases in S02 removal.
REFERENCES
1. Fiedler, Mark A. Sulfur Retention in Fly Ash. Masters Thesis, Univer-
sity of Tennessee, 1980.
2. Rosenberg, H.S. et al. "The Status of S02 Control Systems." Chemical
Engineering Progress. 71:5, May, 1975.
3. Ando et al. "Sulfur Dioxide Removal from Waste Gases: A Status Report
Japan." Pollution Engineering and Scientific Solutions, Plenum Press,
1973.
4. Davis, W.T., and M.A. Fiedler. "The Retention of Sulfur in Fly Ash from
Coal-Fired Boilers," JAPCA. Vol. 32, No. 4, 1982, pp. 395-397.
5. "Gaseous Fuels; Coal and Coke; Atmospheric Analysis." Annual Book of ASTM
Standards. Part 26, 1978.
ACKNOWLEDGEMENT
The research conducted in this study was funded under a contract with the
Department of Energy located at the Grand Forks Energy Technology Center. The
contents do not necessarily reflect the views of DOE, nor does the mention of
trade names or commercial products constitute endorsement or recommendation
for use.
The work described in this paper was not funded by the U.S. Environmental
Protection Agency and therefore the contents do not necessarily reflect the
views of the Agency and no official endorsement should be inferred.
532
-------
Fluegas
Ui
U)
CO
Waste
S02
P
T
W
PORT 3
-- OP
CD
To Fan
Pure S02 injection
Dilution air port
Flyash injection
S02/02monStoring port
Static pressure port
Temperature port
Wet bulb port
OrFfice plate
Gas flowrate control damper
Direction of gas flow
Waste
FIGURE 1 I SCHEMATIC OF SPRAY DRYER/FABRIC FILTER SYSTEM,
-------
U>
t-
<
{£
^
z
ui
tJ
O
u
5 10 15 20 25 30 35 40 45
TIME (MIN)
FIGURE 2; DISSOLUTION KINETICS OF HOOT LAKE FLY ASH
' IN DISTILLED WATER AT ?4°c.
10
,-3
s
1
10
1* M
LIGNITE
SUBBITUMINOUS
BtTUMIMOUS
JL.
J—L
5 lO"* Z 5 W 2
SLURRY CALCIUM + SODIUM (MOLES/SEC AS CAO)
FIGURE 3: S02 REMOVED VERSUS CALCIUM PLUS SODIUM,
-------
Ul
U>
U1
u
ui
CO
CO
LU
Q
UJ
i
UJ
ae.
o
to
LIGNITE
SUBBITUMINOUS
BITUMINOUS
l i I i I I I
10"3 2
'2
5 10' 2
TAMO-S (MOLES/SEC)
>
cc
13
FIGURE 4: S02 REMOVED VERSUS TOTAL AVAILABLE METAL OXIDES
LESS SULFUR (AS CAO),
FIGURE 5! SULFUR RETENTION VERSUS TOTAL ALKALINE
METAL CONTENT (REPORTED AS CAO),
-------
FABRIC FILTRATION - AS IT WAS, HAS BEEN, IS NOW AND SHALL BE
by: Edward R. Frederick, Technical Director
Air Pollution Control Association
Pittsburgh, PA 15230
ABSTRACT
Since the first baghouse patent issued in 1852, commercial filtration
technology has progressed significantly with advances in both collector design
and in the performance capability of filter fiber and fabric. With the re-
sulting extended and expanded service of fabric filters, critical issues have
evolved concerning electrostatic involvement. Even though the natural charges
present on gas entrained particles and on the collecting media interact to
play a major role in essentially all filtration operations, more intense
interest and study is being devoted to electrical augmentation as a means for
optimizing these features and, thereby, all collection parameters. These ob-
servations have also stimulated further interest in and the production of new
types of electrets, the electrified fibers that retain charges for extended
service even under adverse conditions.
Although zinc oxide was first suggested as an absorbant for SC>2 during
the last century, more economic reagents now serve with fabric filters to
control this emission contaminant commercially by dry and wet/dry scrubbing.
Further advances in this technology through the addition of special catalysts/
additives and/or with special processing aids, increase S(>2 removal efficiency
and also control NOX effectively.
Normal use temperature filter fabrics are being improved and even better,
although more exotic high temperature products are available. Improved pro-
cessing, finishing and treating practices are in use to offer special value
in extending bag life, chemical resistance and cleanability. The major yet
unheralded problem of nodule formation, deserves special research consideration
in view of the check valve effect that these "dingleberries" have on the fil-
tration process. Corrective measures for this condition and expansion of
waste heat/product recovery and utilization will certainly lead to further
expansion of fabric filtration technology for S02 as well as particulate
matter control.
536
-------
INTRODUCTION
Those of you involved with fabric filtration have had to find the devel-
opments during the recent 8 to 10 years very exciting. Technology in this
particulate control method has advanced further in this period than during the
preceding decades except, perhaps,at the time when synthetic fibers and changes
in processing methods were introduced during the 40's and 50's.
The current advancement in the art of fabric filtration, as well as in
other control practices, is due in very large measure to the research support
and publicity provided by the U. S. Environmental Protection Agency through
both in-house and contracted research. Never before were the air pollution
control methods so thoroughly and extensively examined and so favorably ex-
ploited.
These and independent R & D efforts in particulate control technology
have led to very favorable predictions. Market Analysts^ have noted that
"Equipment and media sales in the U. S., totaling $2.7 billion in 1980, were
dominated by particle emissions collectors, which accounted for 23 percent of
the total market. Electrostatic precipitators and fabric filters comprise
the bulk of this category.
This dominance is expected to remain through 1995, when particle emis-
sions collectors will total nearly $3 billion, with fabric filters gaining
some market share as the result of advances in fabric technology (especially
for high temperature applications) and innovations such as electrostatically
charging filter fabrics to increase particulate removal efficiency."
HISTORY
Fabric filtration is as old as the art of textile production; but, in-
dustrial particulate collection in baghouses began only during the middle of
the last century. Despite this early beginning and more than 130 years of
commercial service, fabric filtration remains an art and will not emerge to
the sophisticated status of a science until more is known about media, par-
ticulates, and their interrelationships.
The commercial baghouse received stimulus early from the nonferrous
metals industry for two principal reasons: to recover valuable raw material
and to avoid court actions by neighboring farmers. It is especially inter-
esting to note that early in this period (late 19th century) of baghouse de-
velopment in a lead smelter operation, H. H. Alexander treated woolen bags
with titanium' chloride to achieve effective acid (sulfuric) resistance. Even
before this, Sprague applied zinc oxide as an acid neutralizer in smelter
smoke control. This, whether the advocates of dry scrubbing realize it or not,
was the beginning of what they now advocate as an important new phase of
fabric filtration technology.
Modern baghouse history began with the installation of massive plant-
built collectors and continued with the transition to "manufactured" bag-
537
-------
houses and the introduction of an array of synthetic fibers. In between these
events, a major filtration study—the "1929 Investigation" was undertaken.
This is significant as the first extensive field study of filter media,
cleaning methods, particulate variables, and other filtration related para-
meters. According to Armand Labbe3 who had been responsible for ASARCO'S
baghouse operations, this commercial scale evaluation demonstrated the sup-
eriority of a fulled, twill weave woolen fabric over a variety of other
natural fiber, smooth surfaced, tightly woven materials; the critical impor-
tance of the shaking operation, the advantage of a horizontal shake of suit-
able magnitude, and especially, "that in the collection of two smokes of
identical composition and concentration, one could be filtered at 6 times
the rate of the other." This difference in performance was ascribed to var-
iations in particle crystallinity and size of the otherwise similar dusts.
Other findings revealed, as we now know, that the rate of filtration depended
upon the amount and type of dust collected. Also noted was the value of low
over high A/C ratios in limiting blinding and for providing often needed
reserve capacity. While a primary goal of the investigation was to find the
key to high ratio operations, the results demonstrated that such could be
realized with substantial initial savings but the end result was increased
bag wear with shortened bag life, lower efficiency, and no reserve capacity
for unpredictable operational peaks.
During this same historical period, the pressure versus vacuum baghouse
contoversy developed. Despite the use of the fan on the dirty side, Labbe
favored the pressure type for a number of important reasons.
Subsequently, but much later, studies in our own laboratory demonstrated
that not only did different fibers, yarns, and fabrics perform differently
but that the more exotic weaves offered real advantages. The common plain
weave was found to require tight yarn packing to provide suitable efficiency;
whereas, the same yarn woven in twill or sateen configuration leaked less
dust even at higher permeability. In the filtration process, particle move-
ment tends to be normal to the fabric surface, passing through visible holes
in plain weave fabrics, but, not holes visible only at an angle in a twill
or sateen weave medium. At this same time, inherent fiber properties that
were not normally considered important were found in most cases to have a far
more critical influence on the collection process than fabric construction.
For example, the electrostatic character of the media eluded to earlier by
several investigators, was found to be influenced by the same property of the
collected particulate matter. Electrostatic effects were found to control
the filtration process4. More will be said about this and similar new devel-
opments .
Despite repeated attempts to develop universally acceptable high ratio
collectors and the successes realized in some applications, the Labbe guide-
lines of low A/C ratios and good cleanability have been proven to be indis-
pensible for realizing high efficiency and long bag life. To my knowledge
the 30 to 1 ratios achieved with the Hersey or Hersey type reverse air jet
collectors are the only commercial devices operating successfully at such
levels but only with some dusts and the "right" fabrics. Likewise, and also
in a number of less critical applications, many pulse jet high ratio collect-
ors are performing well, generally at less than half this A/C ratio. Here,
538
-------
too, felt type, depth filtration media with suitable qualities, are required
for effective performance. The development of, and progressive improvement
made in these media have led to expanded use of high ratio collectors
whether constructed to. employ tubular or pocket type filters.
Conventional shaker and reverse air type collectors continue to serve
extensively as the low ratio (up to ^6 cfm/ft2 of cloth) collectors. Bag
life has been and continues to be a critical operating criteria with 5 years
being about the normal expectance. Even the fragile glass bags have reached
this level of longevity when optimum conditions prevail but, even much long-
er periods of use have been reported for organically based fiber systems.
The utility industry has adopted baghouse technology to overcome problems
associated with fuels and ash by employing mostly low ratio reverse air type
systems. By and large the results have been extremely favorable and the
development of novel techniques promises to further expand baghouse usage
here and elsewhere.
FABRIC DEVELOPMENTS
The key to an effective and efficient filtration operation, assuming
the equipment hardware is functioning satisfactorily, is the filter medium.
Fiber suppliers, fabric designers, producers, and finishers alike have
searched desperately for answers by developing new fibers and processes for
fabrics and even for different, better cleaning systems. As a result, the
applications of fabric filtration have expanded significantly into regions
of higher temperature and conditions of chemical insult never before even
considered possible.
FIBERS
During the last 40 years an array of new fibers, semisynthetic and
synthetic, have been added to the natural fibers available for fabric fil-
tration. These materials, and especially those of more recent development,
provide unique .properties and extend performance and operating temperatures
up to and even beyond 500°F. Actually, more exotic new products extend the
range beyond 1000°F with better chemical resistance.
The fibers commonly available today cover a wide spectrum of properties
from those offered by the natural materials like wool and cotton to those
given by nylon, polyester, acrylic, polyolefin, aramid, Teflon and glass.
They are prepared in a variety of woven, knitted and needled filter media.
These filter fabrics offer almost unlimited opportunities for conducting the
diverse collection operations needed to recover valuable products, by-
products or waste materials and to limit harmful or nuisance type emissions.
Opportunities to vary the base fiber have also brought about a realiza-
tion among users and suppliers that a given fiber, however desirable other-
wise, is not entirely effective in collecting different dusts. This inability
to specify desirable fibers unilaterally, especially in certain critical
539
-------
applications, has led to the acceptance of a gray region in our filtration
theory. Now, because it is becoming more evident that the electrical re-
lationship between particulate matter and filter media plays a major if not
dominant role in the process, an extension to the theory is evolving.
Recognizing the critical influence of the nature of the cake formed on
the collecting fabric and the fact that most particulates may be deposited
in a porous structure under the influence of favorable electrostatic con-
ditions, new and/or more rational media prescription practices will evolve.
In the interim, electrical augmentation or articifical charging of particles
and applied electric fields at the fiber surface in pilot scale studies are
indicating clearly the interdependence of these conditions. Substantially
higher air to cloth ratios, consistently higher efficiencies, better fabric
cleanability, less particle penetration and reduced fabric blinding with
promises of extended bag life are all evident through electrical augmentation.
Couple such developments with the practical experience already gained with
fabric filtration in dry scrubbing and in controlling tarry or corrosive
emissions, and it becomes apparent that this technology has come a very long
way.
By adding neutralizing, absorbing or adsorbing reagents to the filter
surface or by including appropriate agents with the reactive emissions, the
fabric filter/scrubber process has expanded into critical control operations
of aluminum production, coking, galvanizing and flue gas desulfurization.
Other opportunities in difficult to control processes and for upgrading those
already in use by the ferrous, nonferrous, cement, clay, glass, food pro-
ducts, coal, asphalt, clay, and chemical industries are now more apparent.
Even so, this control technology like all of the others, will become even
more useful as the by-product and energy reduction product recovery and
material utilization processes become more feasible.
ECONOMIC CONSIDERATIONS
An attempt to predict the capital, operating, and maintenance costs of
any operation is risky and always subject to correction. One issue about
economics, however, seems to be crystal clear even though considerable var-
iation in the predictions occurs here too. I refer to the cost/benefit
relationship of particulate control. The analyses carried out by EPA, The
National Wildlife Federation, The National Academy of Sciences, and others
show without question, that air pollution control, for example, pays off
substantially with very significant benefits above costs. The specific ad-
vantages, costwise, of one control technology over another are quite diffi-
cult to determine without consideration of all specifics. Nevertheless,
when a choice is possible, a few, very "broad-brush" guidelines are available
from various sources.
An EPA partially funded elaborate 1978 study of the "Capital and Operat-
ing Costs of Selected Air Pollution Control Systems"5 is available. Two
years earlier, an extensive review of "The Environmental Control Industry"6
was prepared for CEQ. An attempt is made in these presentations to compare
540
-------
the various costs as well as some other features of fabric filters with elec-
trostatic precipitators and scrubbers. Quite obviously, it is necessary to
acknowledge the questionable reliability of some methods used to equate costs.
In general, and this may be the best way to relate the information con-
cisely, it seems evident that fabric filters:
0 are capable of controlling very fine particulate matter at very high ef-
ficiency,
0 perform at lower cost on a flow throughput basis,
0 operate with less energy input than scrubbers but somewhat more than that
needed for ESPs,
0 show a trend for lower increases in flange-to-flange prices than indicated
for the other control methods,
0 show less of an advantage in installation, total operating, and maintenance
costs than ESPs,
0 are less costly to operate than scrubbers.
The filter medium is critically important in its influence on baghouse
performance and represents a large part of the capitol, operating, and main-
tenance costs. In any economic analysis, therefore, a comparison is also
necessary.
Substantial variations occur between the price for very light weight
filament materials, varying weight woven, spun or combination fabrics, and
heavy needled felts. These differences and the fact that the different air
to cloth ratios indicate again how difficult and unreliable baghouse cost
generalizations can be.
DEVELOPING MATERIALS, PROCESSES AND TECHNIQUES
ELECTRETS
The electrical analog of a magnet, the electret, is a recent commercial
development that also serves to verify the value of electrical involvement
in the fabric collection process. The Hanson electrostatic filter developed
in the 40's for the war effort in England was an outgrowth of Heaveside's
work in about 1890, Eguchi's in 1920, and Gemants in 1935. His rosin impreg-
nated wool has since been modified and refined by including acrylic fibers
with wool and by using zinc resinate instead of rosin.
Within the last decade an entirely new and different approach has been
used. One product is prepared by imposing an electric field on heated poly-
olefin film after the drawing/orientation stage of the process and then
fibrillating the charged film to form durable electrified staple fiber. Fiber
systems of this type, usually needled, now offer a high level electrostatic
field of long durability even in the presence of moisture and other discharg-
ing environments. They have found extensive use as efficient respiratory
filters. As far as I know, electrets have not yet been used extensively in
commercial filtration service.
541
-------
DRY AND WET-DRY SCRUBBING
Interest in, and commitment .to, powder injection and spray drying for
S02 and particulate control has increased dramatically in view of the claims
for lower costs, dry wastes, lower energy requirements and simplicity as well
as the acceptability of needed lower SOa removal from low sulfur coals.
Since removal efficiency depends upon many factors such as stoichiometric
ratios, temperature and agent reactivity; studies of this technology are
continuing at an accelerated pace. For example, considerable effort is being
directed toward improvements in removal not only of S02 but also of NO^.
More favorable parameters and the use of catalysts', promoters** and other
additives" are being investigated for overall increased control.
NOX and SOz Control
In addition to conventional current scrubbing applications of fabric
filters for SOa control, recent studies have demonstrated the value of
electron beam radiation following the spray dryer for removing NO^ and op-
timizing SOa control. The high energy electrons oxidize 862 and NOX to
higher oxidation states which react more readily with excess lime from the
spray dryer or with alkaline fly ash. The net effect is enhanced S(>2 control
plus highly efficient NOx removal. The practicability and economics of the
method are being investigated at several laboratories.
An especially interesting new process offered by the National Bureau of
Standards" employs ozone and propylene to oxidize S02 and NOx f°r easier
removal of the resulting sulfate and nitrate with included ammonia.
ELECTRICAL AUGMENTATION
Artificial charging, as I prefer to identify the practice of charging
particulate matter and/or applying an electric field to the filter medium,
has received very considerable attention recently at both the bench and
pilot scale level in investigation. The results have been consistently
favorable. Lest there be unfavorable results coming from any such study, it
must be made clear that the principle benefit of electrical augmentation
resides in the porous nature of the deposited material that occurs in the
presence of the electric field. Most types of particulate material respond
in this way, but I believe that a few do not. Can you imagine the disen-
chantment that would occur by a report of negative data from studies of
electrical augmentation? Recognizing that not all particulate matter are
created equal, it seems evident that an attempt should also be made to class-
ify dusts according to their ability to form a porous deposit as well as how
they respond to the electrical augmentation process.
Quite possibly, the most serious issue facing the would-be users of
542
-------
electrical augmentation involves the durability of the circuitry. The
question Is, "how can an electric field be generated and maintained on
the collecting surface for very extensive periods of use?" EPA's modifi-
cation of TRI's charging electrodes has converted the outside fine wire
girdle to a more substantial charging cage. These and future developments
will certainly hasten the commercial adaptation of electrical augmentation
to the filtration operation.
Gaylord Penney11 has proposed a different approach, one that may offer
a simpler practice. By insulating all or half of the bags in a collector,
using slightly conductive media (<10 Q/Q) and by applying charges of op-
posite polarity to opposing bags, Penney creates a field at the collecting
surfaces. He also suggests that it may be advantageous to impose a bi-
polar charge on the particles entering this augmented collector. But this
approach needs further study.
Whatever practice is ultimately devised for providing an electrical
field at the filtering surface, it will be apparent that mechanical simplic-
ity and durability will be the keys to its successful commercial utilization.
The incentives for adopting electrical augmentation for collecting appro-
priate particulates are truly significant. The achievable improvements are
real and substantial. Energywise, for example, the pressure drop reductions
alone can reduce annual fan energy costs by at least a quarter million dol-
lars for very large collectors. Add to this the savings in cleaning energy
and bag replacement costs and it will be apparent that electrical effects
deserve serious attention.
NATURAL CHARGES AND ELECTRIC FIELDS
A significant portion of the improved filtration performance realized
by electrical augmentation or artificial charging is achievable simply by
balancing particulate and fabric charges. All industrial processes produce
particulate matter with charges, positive and/or negative. Thus, particles
entering conventional collectors are charged, sometimes far more extensively
than at other times but mostly of mixed polarity. The type of process gen-
erating particles greatly influences the magnitude of the charge, with
grinding and other energy intensive operations producing particulate matter
with extremely high levels of charge.
The electrical field, its polarity and magnitude, that develops on the
filter medium by particle contact is determined by the inherent properties
of the fibers that make up the fabric, by the construction of the fabric, by
the particulate itself, and by operating conditions. It is possible, there-
fore, under the ideal conditions realized by use of a filter fabric of
favorable electrostatic properties relative to those of the particulate being
collected and, of course, for a particulate capable of alteration, to depos-
it a low air flow resistant cake without electrical augmentation. In other
words, the natural triboelectrification properties of the medium may be
employed to approach the ultimate level of filtration performance now
achieved by augmentation.
The opportunities to utilize natural electrostatic effects fully are
543
-------
limited to some extent by the availability of appropriate media and more so
by the variability of commercial fabrics. The most serious problem, however,
is the limitation that prevails because of our restricted knowledge of the
electrical properties of particulates and fabrics. We need to know more
about these characteristics separately and how they relate to each other.
Here, then, is an area for further study, the results of which could easily
lead to more accurate detection and correction of filtration problems and,
thereby, even more reliance on this technology by reliable prescription.
In addition to the measurement problems, there are inherent features of
particulate and media that need to be dealt with. For example, there tends
to be a variation in fiber qualities, even among those of the same or sup-
posedly the same commercial origin. My studies, for instance, showed very
significant differences in the electrostatic properties of polyester fibers.
Even among those of one type, for example, Dacron fabrics, I've found tri-
boelectric polarities ranging from the very electropositive region char-
acteristic on nylon, to the electronegative levels of the polyolefins. No
explanation has been offered for these very significant differences but
there is considerable evidence, both from experimental and industrial exper-
ience to indicate that they are responsible for otherwise unexplained per-
formance variations.
Other less obscure differences among filter media, regardless of similar-
ity in fiber make-up, are also evident at times and these too influence be-
havior. Consider what happens when a residual amount of the fiber producer's
antistatic finish is allowed to remain on the fabric, even in a trace amount.
Lowered electrical resistivity occurs and adverse side effects may be expect-
ed. Even though the fast charge bleed1 off features of this antistatic agent
may sometimes be desirable, just as often it tends to be unfavorable since
it reduces the available charge below that needed to optimize porous cake
formation. Furthermore, these agents are heat sensitive, fugitive quater-
nary ammonium compounds that degrade to an amine which conveys electropositive
properties to the fabric before it too burns off.
These effects and the common problems associated with electrostatic be-
havior have contributed to the uncertainty and even doubt about the part
that electrostatics play in the filtration process. Nevertheless, filter
media users, producers, suppliers, and researchers alike are all well aware
that electrostatic charges do exist on particulate matter and on the collect-
ing bags. The electrical augmentation studies have now rekindled the issue
and seem to make it necessary, to clearly define the role on natural charges.
Since some doubt remains, let's adopt a practice that can be accepted
without bias to provide relevant data. I propose that from now on, filter
fabric specifications include two additional parameters. These are, the
electrical resistivity and triboelectric data. By offering this information
and by making a comparative analysis of performance characteristics with
other physical properties of the media, it should be possible to determine
whether, how, and to what extent the collection process depends upon elec-
trical phenomena. To be more completely effective in such an analysis,
electrical and electrostatic data should also be available for the particulate
matter being collected.
544
-------
RESEARCH NEEDS
High Temperature Fabrics
The growth of the fabric filtration market depends upon current and new
developments in several other important areas. Better high temperature fab-
rics with more favorable chemical and mechanical properties tend to be on
everybody's wanted list. While I agree that opportunities exist for media
to be used in the range of 750° to 1000°F, a continuing demand remains for
acid resistant, durable fibers capable of operating in the 300° to 500°
range, I fail to see a very large market for exotic, high cost fibers al-
though a moderate demand will certainly remain for those super fibers now
being produced.
For many high temperature processes an alternative to high temperature
filter media is possible within certain limits, even when dewpoint problems
exist. I refer specifically to the prospects for reducing the temperature
of emissions through improved heat transfer conversion systems. The more
effective use of conventional heat exchangers and more efficient pyroelec-
trical methods that can make the conversion of heat to other forms of energy
more appealing, would make some moderately high temperature filtration pro-
cesses more viable substitutes for those that are now considered necessary
at the higher temperatures.
The need for improvements in the currently available commercial high
temperature media will remain. Superior protective and certain special
custom finishes for glass fabrics, for example, can be expected to increase
performance as well as bag life. Conceiveably too, aramid fabrics might be
made more acid resistant than permitted by current treating practices. It
must be appreciated, however, that because the filter media market is some-
what limited, certainly as compared to that for apparel fabrics, the incen-
tives, research budgets and overall efforts to effect improvements in media
will be limited.
Fabric Variations
Other filter fabric developments deserve mention and appropriate con-
sideration. Seamless filter bags made possible by the tubular knitting
technique offer new and different features for a variety of collection oper-
ations. Further development in the knitting processes, better fabric stabil-
izing methods, improved fibers, yarns, weaving, and/or needling and finishing
and treating practices will surely provide better media for some better fil-
ter service.
In felts and felt-like materials, a major breakthrough can be expected
with expanded production of durable and stable scrimless or very low scrim
content needled fabrics. I doubt that we fully realize the restrictions
imposed by the scrim component of needled fabrics. Felts of appropriate
fibers in such new constructions will advance the performance of high ratio
collectors substantially. No-scrim Nomex felts are now available. The real
value of this change will be soon be apparent.
545
-------
Filter Cleaning
More effective cleaning of any filter is a continuing goal of everyone
involved in this technology. While the introduction of scrimless felts is
one approach, other techniques have been applied and will be used for im-
proving cake arrd plug removal. The prospect for realizing significantly
better cleaning is not restricted to fabric changes and treatment. It also
applies to the adoption of improved collection procedures, the rigorous
application of electrostatic principles, and the use of other cleaning as-
sists like those offered by magnetic and acoustic devices and others.
Nodules (Refer to Figure 1)
For many filtration processes, effective control of nodules would lead
to substantially better overall performance.- The problem presented by
nodules, although not always evident, was most clearly noted by Penney^
in his description of these so called dingleberries as check valves (note
Figure 2) that restrict air transfer and have a serious influence on the
cleaning process. The check valves simply swing away from the surface under
the influence of reverse air but close the surface voids during the filtra-
tion cycle.
Nodule formation, I contend, is in part another electrostatic phenomenon.
The initial accumulation of particles at the ends of fibers is easily ex-
plained in this way. Conversion of these originally soft deposits to a
hard mass seems to occur as a result of a phase or chemical change with a
cementing action. Some questions for researchers are, therefore:
1. What mechanisms account for the accumulation of nodules?
2. Is moisture alone critical in the hardening process, and in causing
them to adhere tenaciously to fiber(s)?
3. What practices may be followed to avoid, eliminate or negate the
influence of these deposits on the filtration process? and
4. What type, if any, treatment may be applied to the fibers/fabric
to control nodules without impairing the filtration process?
Waste Recovery/Utilization
Last but not least among the ways for the fabric filtration industry as
a whole to improve its usefulness and its image is to promote an energetic
program of waste recovery and/or utilization. This applies to the utiliza-
tion of waste heat as well as to the recovery of waste or seemingly low
value by-products. Mountains of fly ash will become another problem unless
the contained minerals and included elements are reclaimed or another use
can be made of the ash itself. Other, so called waste products are or will
be classified as hazardous and will become a serious liability unless they
too are reclaimed.
As one example, consider with me the product collected in baghouses on
some electric furnaces. With a 20% zinc, 10% lead and trace metal content;
reuse of this dust in the furnace was deemed undesirable. As such, it was
treated as a waste and disposed of, until recently, by burial. Now in view
546
-------
of the recognized potential hazard of leaching, the burial practice has been
forbidden. Would it not be safer and more compatible with the conservation
trend to recover the raw materials? If health effects and economics dictate
policy, this action appears justified, especially in view of the potential
danger of lead and the fact that zinc ore is imported.
In closing, let me show you how different particulate control methods,
including fabric filtration, have been useful for pollution control in
Pittsburgh. Figure 3 is a photograph of the city taken in 1945. The second
was taken in 1956, just eleven years later, from the same location looking
toward the downtown area. It will be apparent that Pittsburgh was trans-
formed to a clean city during this period from its well established status
as the "Smokey City".
The work described in this paper was not funded by the U. S. Environ-
mental Protection Agency and therefore the contents do not necessarily re-
flect the views of the Agency and no official endorsement should be inferred.
547
-------
FIGURE 1. Nodule Attached to Fiber
Dust Cake
Reverse Air Flow
(Cleaning)
Dust Cake
Particulate/Gas
F10V7
Forward Air Flow
(Filtering)
The Check Valve Effect - G. W. Penney, Electrostatic
Effects in Fabric Filtration — EPA— 600/7-78-142a
September 1978
FIGURE 2. A Model of Nodule Action
548
-------
J*rfu*<:- .... I
FIGURE 3. View of Downtown Pittsburgh - 1945
FIGURE A. View of Downtown Pittsburgh - 1956
549
-------
REFERENCES
1. Predicasts, Inc. News Release, August 1982.
2. E. S. Godsey, American Smelting and Refining Company, Inc., Salt Lake
City, Utal, Personal Communications, July 1980.
3. A. Labbe, Consultant, Personal Communication,
4.' E. R. Frederick, How Dust Filter Selection Depends on Electrostatics,
Chemical Engineering 68:107, June 26, 1961.
5. R. B. Neveril, J. U. Price and K. L. Engdahl, Capital and Operating
Costs of Selected Air Pollution Control Systems, APCA RS-9 Reprint
Serie's. 1979 from J. Air Pollution Control Association 28. 1978.
6. J. A. Klein and K. C. Leung, The Environmental Control Industry,
Allanhold Osmun & Co., Montclair, N. J. 1976.
7. M. Linne1, J. Klinspor, H. T. Karlsson, and I. Bjerle, Limestone Based
Wet-Dry Scrubbing to Form Gypsum, Chem. Eng. Sci., _5, (37), 807, 1982.
8. H. T. Karlsson, J. Klingspor, M. Linne1 and I. Bjerle, Activated Dry
Scrubbing of SOa, to be published in the J. Air Pollution Control Assoc.
9. H. T. Karlsson, J. Klingspor and I. Bjerle, Adsorption of Hydrochloric
Acid on Solid Slaked Lime for Flue Gas Clean Up, J. Air Pollution
Control Assoc. 11, 31, 1981.
10. P. L. Feldman and D. J. Helfritch, Particle Properties and Feasibility
of E-Beam Precharging and R. H. Davis, Agenda Overview - E-Beam Work-
shop on The Combined Removal of S02, NOx and Particulate Matter from
Stack Gas by Electron Beam Treatment, Arlington, VA, February 25 and
26, 1982, Sponsored by the U. S. Department of Energy.
11. G. W. Penney, Electrical Augmentation of Fabric Filters, U. S. Patent
3910779 October 7, 1975 and U. S. Patent 3966435, June 29, 1976.
12; G. W. Penney, Electrostatic Effective Fabric Filtration, Volune 1.
Fields, Fabrics and Particles, EPA-600/7-78-142A, September, 1978.
550
-------
AUTHOR INDEX
AUTHOR NAME PAGE
ADAIR, L 1-460
ADAMS, R.L 11-35
ANDO, T 11-474
ARMSTRONG, J III-241
ARSTIKAITIS, A.A 11-194
BALL, C.E III-370
BANKS, R.R 1-37, 1-62
BARRANGER, C.B 1-132
BAYLIS, A.P 11-384
BELTRAN, M.R 11-51
BENSON, S.A 111-97
BERGMAN, F . III-154
BIESE, R.J 1-446
BOSCAK, V 111-66
BRADBURN, K.M 11-499
BRADLEY, L.H 11-369
BRINKMANN, A III-211
BUCK, V III-335
BUMP, R.L 11-17
CAPPS, D.D 1-121
CARR, R.C. 1-148
CHAMBERS, R 1-226, 1-239
CHANG, R III-271
CHEN, F.L III-347
CHEN, Y.J 1-506
CHIANG, T 11-184
CHRISTENSEN, E.M 11-243
CHRISTIANSEN, J.V 11-243
CILIBERTI, D.F III-282, III-318
CLEMENTS, J.S 11-96
COE,JR, E.L 11-416
COLE, W.H III-l
551
-------
COOK, D.R 11-349
COWHERD,JR, C III-183
OOY, D.W III-370
CRYNACK, R.R II-l
CUSCINO, T III-154
GUSHING, K.M 1-148
DAHLIN, R.S 1-192
DABBY, K 11-499
DAVIS, R.H 11-96
DAVIS, W.T 1-521
DAVISON, J.W III-166
DELANEY, S 1-357
DEMEAN, A 111-66
DENNIS, R 1-22, 111-81
DIRQO, J.A 111-26, 111-81
DISMDKES, E.B 11-444
DONOVAN, R.P 1-77, 1-107, 1-316, 1-327, 1-342
DORCHAK, T.P III-114
DRENKER, S III-271, III-282
DRIQGERS, G.W 11-194
DUBARD, J.L 11-337
DUFFY, M.J 11-489
DURHAM, M 11-84, III-241
EBREY, J.M 11-349
ENGLEHART, P.J III-183
ENSOR, D.S III-347
FAULKNER, M.B 11-204, 11-337
FINNEY, W.C 11-96
FORTUNE, O.F 1-482, 1-494
FOSTER, J.T 1-37, 1-91
FREDERICK, E.R 1-536
FRISCH, N.W III-114
FURLONG, D.A 1-287, 1-342
GARDNER, R.P 1-77, 1-107
GAWRELUK, G.R 11-17
GELFAND, P 11-35
552
-------
GIBBS, J.L 11-430
GILES, W.B 111-41, 111-53
GOLAN, L.P III-226
GOLDBRUNNER, P.R 11-401
GOLIGHTLEY, R.M 1-164
GOOCH, J.P 11-444
GOODWIN, J.L III-226
GRANT, M.A 111-81
GREEN, G.P 1-192
GREINER, G.P 1-287, 1-357
GRONBERG, S III-141
GRUBB, W.T 1-62, 1-91, 1-179
HALL, H.J 11-459
HALOW, J.S 11-96
HANSON , P , 1-460
HARMDN, D 1-226, 1-239, III-131
HAWKINS, L.A 11-194
HERCEG, Z 11-489
HOVIS, L.S 1-22, 1-77, 1-107, 1-287, 1-316, 1-327,
1-342, 1-357, 111-81, III-347
HOWARD, J.R 1-164
INGRAM, T.J 1-446
ISAHAYA, F 11-154
ITAGAKI, T 11-322
JACOB, R.0 1-446
JENSEN, R.M 1-431
JONES, R .- 1-303
KASIK, L.A 11-430
KETCHUCK, M 1-482
KINSEY, J III-154
KOHL, A.L III-300
KUBY, W III-271
KUNKA, S 1-239
KUTEMEYER, P.M III-211
LAMB, G.E.R 1-303
LARSEN, P.S 11-243
553
-------
LAWLESS, P.A 11-271
LEE, W 1-303
LEITH, D 111-26
LEONARD, G.L 11-230
LEWIS, M 1-179
LIPPERT, T.E III-280, III-318
LOGAR, T.W 11-184
MARCHAWT,JR, G.H 11-444
MASON, D.M III-256
MASUDA, S 11-139, 11-169, 11-322, III-386
MATSOMDTO, Y 11-474
MATOLEVICIUS, E.S III-226
MXAIN, J.D III-198
MOOOLLOR, D.P 111-97
MCDONALD, J.R 11-204
MCKENNA, J.D 1-210
MCLEAN, K.L 11-489
MENARD, A 1-255
MILLER, M.L 1-482
MILLER, R.L 1-494
MILLER, S.J 111-97
MITCHNER, M 11-230
MOSLEM, G.B 11-288, 11-306
MOSLEY, R.B 11-204
MDYER, R.B 1-460
MOSGROVE, J 1-382
MYCOCK, J.C 1-210
NAKATftNI, H 11-169
NG, T.S 11-489
NOVOGORATZ, D 11-349
OGLESBY, S 11-534
O'ROORKE, R III-318
PEARSON, G 1-121
PETERS, B.J 1-179
PIDLLE, W 11-401
PONTIUS, D.H 11-65
554
-------
PUDELEK, R.E 1-521
PUTTICK, D.G 11-126
QUACH, M.T 1-506
RAMSEY, G.H 1-316, 1-327
RANADE, M.A III-347
REED, G.D 1-521
REHMAT, A III-256
REIDER, J.P III-183
REISINGERr A.A 1-179
RICHARDS, R.M 1-255
RICHARDSON, J.W 1-210
RINARD, G 11-84, III-241
ROOP, R.N 1-460
ROSS, D.R 1-164
RUGENSTEIN, W.A , 11-430
RUQG, D 11-84, III-241
RUSSELL-JONES, A 11-384
SAIBINI, J 1-132
SAMUEL, E.A 1-1, 11-218
SANDELL, M.A II-l
SAWYER, J III-271
SEARS, D.R 1-192, 111-97
SELF, S.A 11-230, 11-228, 11-306
SHACKLETON, M III-271
SHISHIKUI, Y 11-139
SMITH, W.B 1-148
SORENSON, P.H III-362
SPARKS, L.E 11-204, 11-271, 11-337
SPENCE, N 1-132
SPENCER,III, H.W 1-506
STELMAN, D III-300
STOCK, D.E 11-261
SUHRE, D III-335
SUOTER, T.C 1-48
SURATI, H 11-51
TACHIBANA, N 11-474
555
-------
TASSICKER, O.J III-271, III-282
THOMPSON, C.S 111-12
THOMSEN, H.P 11-243
TOKUNAGA, 0 11-96
TREXLER, E.C 11-96
TRILLING, C.A III-300
TSAO, K.C III-256
VANN BUSH, P 11-65
VANQSDELL, D.W 1-287, 1-342
WALSH, M.A 1-482
WEBER, E 11-111
WELLAN, W.G 1-420
WEXLER, I.M 11-521
WHITTLESE?, M 1-482
WILOOX, K III-154
WILLIAMSON, A.D III-198
YAMAMDOD, T III-241
YEAGER, K.E III-xv
556
-------