FABRIC FILTRATION EXPERIENCE DOWNSTREAM FROM
ATMOSPHERIC FLUIDIZED BED COMBUSTION BOILERS
Kenneth M. Cushing
Southern Research Institute
P.O. Box 55303
Birmingham, Alabama 35255-5305
Victor H. Belba
4758 Edison Lane
Boulder, Colorado 80301
Ramsay L Chang
Thomas J. Boyd
Electric Power Research Institute
P.O. Box 10412
Palo Alto, California 94303
For presentation at the Ninth Symposium on the
Transfer and Utilization of Particulate Control Technology
Williamsburg, Virginia
-------
FABRIC FILTRATION EXPERIENCE DOWNSTREAM FROM
ATMOSPHERIC FLUIDIZED BED COMBUSTION BOILERS
ABSTRACT
In recent years atmospheric fluidized bed combustion (AFBC) has become a
commercially accepted technology for solid-fuel-fired power generation systems, with
about 10 GWe of power generating capability either in operation or under construction.
The continued development and commercialization of AFBC technology has also
included evaluations of balance-of-plant systems, including particulate control
methods, required to support a power generating system. These evaluations have
included research on the physical and chemical characteristics of AFBC fly ash.
Results of this research show AFBC fly ash to be generally more difficult to handle
than pulverized coal (PC) fly ashes.
As part of their research efforts to support the development of particulate control
technology for the utility industry, EPRI has evaluated the performance of electrostatic
precipitators and baghouses downstream of AFBC boilers. EPRI has recentfy
completed studies of fabric filtration during two utility AFBC demonstration plant
projects in Colorado and Kentucky. Also completed recently is a worldwide survey of
pulse-jet baghouses including those applied to both utility and industrial AFBC boilers.
Results from these studies are presented in this paper. These data include baghouse
performance data (air-to-cloth ratio, pressure drop, emissions, bag life), as well as
information on baghouse operation and maintenance experience including startup and
shutdown procedures, common maintenance problems, etc. These data show that
although there is still some uncertainty as to the exact interaction of AFBC ash
characteristics and filtration performance, the overall experience of plants employing
-------
FABRIC FILTRATION EXPERIENCE DOWNSTREAM FROM
ATMOSPHERIC FLUID1ZED BED COMBUSTION BOILERS
INTRODUCTION
In recent years atmospheric fluidlzed bed combustion (AFBC) worldwide has
increased sharply - about 10 GWe of generating capacity is currently either in
operation or under construction. The advantages of this technology for new plant
construction, retrofit at older plants, and plant life extension have motivated this
increased interest In the retrofit area alone, EPRI estimates that 20,000 MW of
pulverized-coal (PC) boilers are candidates for AFBC retrofits (1). Besides the utility
market, there are a-number of small AFBC boilers installed around the world for the
purpose of process steam generation or co-generation.
Key advantages of AFBC (bubbling bed and circulating bed) technology include
reduced S02 and NOx emissions, compared to those of PC boilers, as well as the
possibility of utilizing a wide variety of fuel sources (coal, peat, culm, bark, wood
waste, hog fuel, etc.). However, AFBC particulate emissions may tax conventional
particulate control systems.
In electrostatic precipitators (ESP), AFBC ash is more difficult to collect than PC ash
because AFBC coal ash has a higher electrical resistivity and the use of cyclones
(muWclones) for recycling, inherent with the AFBC process, tends to reduce exit gas
stream particle size. In the case of fabric filters, the effects of AFBC ash
characteristics on filtration performance have been difficult to estimate because of
limited experience. For both ESPs and fabric filters the high particulate mass
concentration exiting AFBC boilers (typically two to four times higher than PC boilers)
can also provide an emissions control challenge.
In spite of these concerns, the general particulate control experience of AFBC plant
operators has been favorable, especially in cases where fabric filtration has been used.
This paper reviews the status of fabric filtration applied to AFBC boilers. Current fabric
filter control technology is discussed, the unique filtering characteristics of AFBC ash
are presented, and baghouse performance, operation, and maintenance issues are
highlighted. Specific performance data from EPRPs two AFBC Demonstration Plant
baghouses are included.
-------
UNIQUE CHARACTERISTICS OF AFBC FLY ASH
Some of the features of the AFBC process have a direct bearing on the particulate
properties that most strongly influence baghouse performance in terms of filtering
pressure drop (2). Operating at temperatures in the range of 1500-1600*F, an AFBC
boiler does not reach the ash fusion temperature as does a PC boiler at 2200-2500* F.
As a result, AFBC combustion of coal forms irregularly shaped fly ash particles
compared to the spherical PC fly ash particles formed by melting and condensation.
The co-combustion of coal and limestone in most AFBC boilers yields a distinctly
different fly ash chemistry than that of PC ash. The high alkalinity of the AFBC ash
afters the eohesivity and, consequently, the porosity or thickness of the dustcake, or
both. The AFBC process also involves inertial segregation of the fly ash with cyclones
(multiclones) In order to recycle the larger particles through the furnace to improve fuel
and sorberrt utilization. This recycling decreases the mean diameter of the ash
particles that are passed to the fabric filter. These snail particles tend to create a
dustcake structure with a high intrinsic pressure drop.
Figure 1 compares the range of values for several physical properties of AFBC fly ash
to those for PC and spray dryer residue ashes. These data form part of the EPRI
Fabric Piter Database. Specifically called out in the figure are the values for ashes
from the two EPRI AFBC Demonstration Plants. These plants are the Colorado Ute
Electric Association's (CUEA) 110-MWe Nucla Unit 4 located near Nucla, Colorado
and Tennessee Valley Authority's (TVA) 160-MWe Shawnee Unit 10 located near
Paducah, Kentucky. The Nucla facility has a circulating bed AFBC boiler and the
Shawnee facility has a bubbling bed AFBC boiler. Data for fly ashes from the TVA 20-
MWe AFBC Pilot Plant and the Texas/New Mexico 150-MWe TNP Unit 1 are also
presented in this figure.
On average, the AFBC particles are smaller in size, have higher specific surface area,
and higher porosity relative to PC ashes. Although the higher porosity of AFBC ash
helps to compensate for the smaller particle size and higher surface area, the net
effect is higher pressure drop attributable to the small pore diameters within the
dustcake caused by the small, irregularly shaped particles. For comparable residual
dustcake area! densities, AFBC dustcakes will have a higher inherent pressure drop
than PC dustcakes at the same filtering air-to-cloth ratio. Even though AFBC residual
dustcakes tend to be lighter (lighter bag weights) compared to PC dustcakes, the bag
cleaning methods employed must be energetic enough to maintain a low residual
dustcake areal density. This will insure a low to moderate operating pressure drop.
This is demonstrated in Rgure 2 which shows the effect on baghouse performance at
the TVA 20-MWe AFBC (bubbling bed) Pilot Plant when the bag cleaning method was
changed from reverse gas to reverse gas with sonic assistance (3). The filtering
air-to-cloth ratio was doubled without a pressure drop penalty.
-------
CURRENT FABRIC FILTRATION EXPERIENCE
The use of AFBC worldwide has Increased dramatically as the technology has become
more widely accepted and commercialized at the industrial-sized level. Over 150 units
are in operation or under contruction with sizes ranging up to 165 MWe. Except for
two of the larger utility AFBC units, all of the remaining AFBC boilers worldwide use
fabric filtration for final flue gas cleanup.
There are two major types of fabric filters, low ratio and high ratio. Low ratio
baghouses typically operate with filtering air-to-cloth ratios in the range of 1.5 to 3.0
ft/min and employ one of three different bag cleaning methods: reverse gas, reverse
gas with sonic assistance, and shake/deflate. High ratio baghouses generally operate
with filtering air-to-cloth ratios in the range of 3.0 to 5.0 ft/min and use one of three
pulse-Jet cleaning methods: high pressure/low volume (40-100 psi), intermediate
pressure/intermediate volume (15-40 psi), or low pressure/high volume (5-15 psi)
(4). Traditionally in the United States pulse-jet cleaned baghouses have been used
most commonly on smaller industrial-sized AFBC units where the flue gas volume to
be cleaned is small and the small size and site area common to the units make them
attractive. Typically, the larger utility AFBC installations in the U.S. use low-ratio
baghouses that employ the more gentle bag cleaning methods, although, as has been
noted above, more agressive cleaning methods are generally required with AFBC ash
in order to control residual dustcake buildup in the bags and correspondingly high
pressure drops.
Over the past several years, EPRI has conducted baghouse performance evaluations
at both full-scale and pilot-scale fabric filters downstream from AFBC boilers aid has
surveyed fabric filter applications on industrial and utility AFBC boilers around the
world in order to assess baghouse performance characteristics and create a database
of operating data to be used in predicting fabric filter performance (5) (6). Figure 3
summarizes the performance of several types of fabric filter installations downstream
from AFBC boilers. These data demonstrate that fabric filters are able to maintain
reasonable pressure drops at typical filtering air-to-cloth ratios and that the more
energetic cleaning methods (shake/deflate and pulse-jet) are best for maintaining low
pressure drops at high air-to-cloth ratios.
The database of information from these plants also indicates that both low-ratio and
high-ratio baghouses with the proper bag fabric can maintain emission levels at or
below regulatory limits. Table 1 summarizes the baghouse performance experience at
a representative selection of AFBC baghouses worldwide. EPRI has recently
completed extensive evaluations of two full-scale, low ratio fabric filter installations at
their AFBC Demonstration Plants in Colorado and Kentucky; performance data from
these units has been included in Figure 3 and Table 1. A more detailed review of the
fabric filter installations is presented below.
-------
LOW RATIO FABRIC FILTRATION
CUEA Nucla 110-MWe AFBC (circulating been Demonstration Plant
The Nucla Station originally consisted of three 12-MWe stoker-fired boilers. In 1974, in
order to meet more stringent emissions regulations, a baghouse was installed
downstream from each boiler. These three, six-compartment, shake/deflate cleaned
baghouses (Unite 1, 2, and 3) were built by Wheelabrator-Frye. In 1984 it was
decided to retrofit the Nucla Station with an AFBC boiler to replace the three stoker
boilers. To accomodate the additional gas flow generated by the new 110-MWe boiler,
Research-Cottrell was contracted to build a new baghouse to supplement the three
older baghouses. The new twelve-compartment baghouse (Unit 4) also uses
shake/deflate cleaning. Figure 4 shows the general layout of the four baghouses in
relationship to the boiler house. Table 2 presents the design data for the fabric filter
system.
Fabric Filter System Performance. The baghouses were placed in service in July 1987
and accumulated over 12,500 hours in service during the demonstration project that
ended in June 1990. Two sources of coal were burned during this operating period
(Salt Creek and Peabody). Considering the new boiler technology on which they were
applied, the Nucla baghouses have had a very good performance record, comparable
to other full-scale utility baghouses downstream from puh/erized-coal boilers. There
have been no periods of uncontrollable high pressure drop, and the shake/deflate
cleaning system, in general, has worked well in cleaning the bags, maintaining low bag
weights, and controlling pressure drop under a variety of coal and sorbent feed rates.
An extensive baghouse evaluation was performed on the new (Unit 4) baghouse after
approximately 9500 hours of service. This baghouse was able to provide a low mass
emission rate (0.007 Ib/MBtu (no bag failures)) at a fairly high filtering air-to-cloth ratio
(2.4 to 2.9 ft/min) with low to moderate tubesheet (3.7 to 5.2 in. H2Q) and
flange-to-flange (5.0 to 6.5 in. H20) pressure drops at near full-load conditions. The
shake/deflate cleaning system was able to maintain low bag weights (0.23 lb/ft2
residual dustcake areal density) even though the inlet particulate concentration was
high (8.9 gr/dscf) compared to values typically experienced at baghouses downstream
from utility pulverized-coal boilers (7).
Figure 5 compares the performance of the Nucla Unit 4 baghouse with baghouses
located downstream from other utility AFBC boilers. Figure 6 compares Nucla Unit 4
baghouse performance to data from full-scale utility baghouses downstream from
puh/erized-coal boilers. In both instances the Nucla baghouse performance compares
-------
Filter Baa Performance. The filter bags used at Nucla were constructed from a 10
oz/ydz, woven fiberglass fabric with a nominal 10% Teflon B finish. Fabric integrity
was generally good during the 12,500 hours of operation during the demonstration
project, except for failures associated with fabric abrasion near the bottom of the bag.
Beginning at about 4500 hours of service and for the next 8000 hours of operation, the
baghouses at Nucla experienced 377 bag failures (out of 4176 total bags installed), the
majority of which were due, either primarily or secondarily, to ash abrasion on the
lower two feet of bag fabric just above the tubesheet Ash abrasion at this location
occurred because the bags are mounted inside the thimbles and sealed in place with
snap bands sewn into puckered bottom cuffs. The resulting pleats and folds in the
fabric are exposed to the ash-laden flue gas passing into the bags. Our theory is that
the abrasion problem was compounded by ash that accumulated between the
opening in the tubesheet and the outside surface of the bag as a result of the poor
snap band/cuff seal to the thimble. This accumulation of ash forced the fabric further
into the gas stream. Figure 7 illustrates the ideal snap-band bag attachment versus the
actual situation occurring at Nucla. Utility baghouses normally have bags attached on
the outside of the thimbles that are mounted on the top of the tubesheet.
Of these bag failures, 82% occurred in the Unit 2 baghouse. It was determined that
the large number of early bag failures in the Unit 2 baghouse was exacerbated by
overdeflation (also occurring in the Unit 1 and Unit 3 baghouses, but to a lesser
extent) and by the fact that most of the shaker mechanisms in this baghouse were
working properly (causing additional stress to the abraded fabric). Adjustment of the
deflation flow rate and the replacement of the failed bags led to a substantial decrease
in the rate of bag failures in the three older baghouses. Figure 8 shows the total
number of bag failures for each of the last seven quarters of baghouse operation
during the demonstration project (the first bag failures were found during the third
quarter of 1988).
In the spring of 1990 a recommendation was made to CUEA to install a compartment
of new bags modified to reduce fly ash abrasion on the bottom of the bags. The
modification incorporated the installation of an anti-collapse ring approximately eleven
inches above the bottom of the bag. The purpose of the anti-collapse ring was to limit
the movement of the fabric into the flue gas stream entering the bag (especially during
resumption of filtering after a cleaning sequence when the bags are normally collapsed
due to deflation and shaking), thus reducing the likelihood of abrasion in this region of
the bag. The modified bags have been in service about 7500 hours. No bag failures
have occurred to date.
On two occasions during the demonstration project, bags were removed from the
baghouse and submitted for strength testing and evaluation. This was done after 5,000
and 11,000 hours of operation. Also, a new, unused bag was submitted to determine
its properties for comparison to the used bags. For the bags in service for 11,000
-------
hours the Mullen Burst strength averaged 302 lb/in2, a 49% loss from the average
value for the unused bag (592 lb/in2). This compares to 362 lb/in2, a 39% loss, for
the bags tested after 5,OCX) hours of service. Fiberglass fabric normally loses a
significant fraction of its original strength during the irrtial period of service (up to 2,000
hours), then the rate of loss tapers off to a gradual decline with continuing service
under stable operating conditions. Figure 9 graphically shows the behavior of the
Mullen Burst values over the first 11,000 hours of service. It was determined from this
and other data that the current (11,000 hour) level of strength retention was still quite
serviceable, and that bag life could easily be extended to 25,000 hours; assuming no
losses from abrasion and if filtration characteristics of the fly ash remain acceptable.
TVA Shawnee 160-MWe AFBC (bubbling bed) Demonstration Plant
The TVA Shawnee Steam Plant originally consisted of ten, 175-MWe putverized-coal
boilers. Rue gas cleanup equipment consisted of a ten-compartment, reverse-gas
cleaned, fabric filter following each boiler (installed 1979 -1981 by General Electric
Environmental Services, Inc.). Following the successful completion of the test
program at TVA's 20-MWe AFBC (bubbling bed) Pilot Plant at Shawnee (3), a new
160-MWe AFBC (bubbling bed) boiler was constructed adjacent to the old Unit 10
pulverized-coal boiler. The Unit 10 PC boiler was shut down and the original turbine
and fabric filter were then used in conjunction with the new AFBC boiler. In the
mid-1980s the Shawnee baghouses were retrofitted with sonic horns to supplement
the normal reverse-gas cleaning system. Figure 10 shows the general layout of the
Shawnee plant Table 3 presents the design data for the fabric filter system.
The Unit 10 baghouse began filtering AFBC fly ash during the third quarter of 1988.
Through August 1991 the baghouse has accumulated over 12,500 hours of service.
The Shawnee Unit 10 baghouse has had a very good performance record,
comparable to the other nine PC baghouses at Shawnee. There have been no
periods where boiler operation has been limited by baghouse pressure drop, and the
sonic-assisted reverse-gas cleaning system has worked well in controlling pressure
drop and maintaining moderate bag weights.
The baghouse has provided particulate emission rates less than the current EPA New
Source Performance Standard (NSPS) of 0.03 Ib/MBtu while operating with no bag
failures. At full-load conditions and at a moderate filtering air-to-cloth ratio (1.6 ft/min)
the baghouse has operated at a moderate flange-to-flange pressure drop of 8.0 in.
H20 with reverse-gas cleaning and 6.5 in. HzO with reverse-gas cleaning with sonic
assistance. The sonic-assisted, reverse-gas cleaning system has been able to
maintain moderate bag weights (0.55 lb/ft2 residual dustcake area! density) even
though the inlet mass concentration has been high (unmeasured) compared to values
typical for pulverized-coal boilers.
-------
Figure 11 compares the flange-to-flange pressure drop performance of the Shawnee
Unit 10 baghouse during several operating periods. Until July 1990 the baghouse
operated with reverse-gas cleaning only. As can be seen in this figure, this low energy
cleaning method was not able to control pressure drop. At full-load conditions the
pressure drop increased from 5.1 in. H20 (May - Sept 1989) to 7.0 in. H20 (Oct 1989 -
Feb 1990) to 8.0 in. HzO (June 1990). In July 1990 the sonic horns were placed in
service for the first time. By August 1990 the pressure drop had stabilized at about
6.4 in. HzO. In late 1990 new horns (same fundamental frequency, more powerful
(based on the manufacturers specifications)) were installed in each of the baghouse
compartments. Although the new horns were able to reduce, on average, the residual
dustcake area! density from 0.78 to 0.55 lb/ft2, there was not a significant
improvement in pressure drop. As of July and August 1991 the pressure drop was
still about 6.5 in. H20 at full-load conditions. The important point to consider, though,
is that pressure drop has been stabilized through the use of the more energetic
reverse-gas cleaning with sonic assistance.
Figure 5 compares the performance of the Shawnee Unit 10 baghouse with low-ratio
baghouses located downstream from other utility AFBC boilers. Figure 6 compares
Shawnee Unit 10 baghouse performance to data from full-scale, low-ratio utility
baghouses downstream from pulverized-coal boilers. In both instances the Shawnee
baghouse performance compares favorably.
HIGH RATIO FABRIC FILTRATION
The pulse-|et (high ratio) fabric filter performance data for the ten units listed in Table 1
represent both domestic and overseas plants. These baghouses are operating on
relatively small bubbling bed and circulating bed boilers (11.5 to 55 MWe). High
pressure/low volume pulse-jet cleaning is most common among the ten units. Both
woven glass and needlefelts (Ryton and Nomex) fabrics are used in these pulse-jet
facilties; however, the woven glass is of a heavier weight (16 to 22 oz/yd2) compared
to typical woven glass fabrics for low-ratio baghouses (10 to 14 oz/yd2). These
heavier woven glass fabrics are better able to withstand pulse forces and abrasion.
These heavier, woven glass fabrics are able, in most cases, to simulate needle felts,
for which, contrary to the case for low-ratio collectors, reliance for filtering is placed on
the fabric and not the dustcake. The data indicate that, on average, at the time of the
visits to these sites, the units were operating at air-to-cloth values of about 3.0 ft/min.
Although two of the units were designed for air-to-cloth values of about 4.5 ft/min,
none of the ten units were operating with air-to-cloth values in the range of 4.0 to 5.0
ft/min.
As shown in Figure 3, except for one plant, the flange-to-flange pressure drops are
moderate, similar to values for the low-ratio baghouses. These data suggest that
pulse-jet fabric filters (PJFFs) can operate at air-to-cloth ratios up to two times higher
-------
than those of conventional reverse-gas baghouses without a pressure drop penalty.
The potential reduction in plan area requirement can result in pulse-jet baghouse
dimensions which are as little as 1 /2 to 2/3 of that needed by a conventional low-ratio
baghouse. This can be an important consideration for retrofit applications where there
are spacial limitations.
The particulate emission rates for these fabric filters are very good, averaging well
below 0.02 Ib/MBtu. Bag life has been fair to good for these plants, ranging from 1 to
3 years. As noted in Table 1, several plants have had bag failures due to a variety of
problems including abrasion, S03 attack, and misaligned pulse pipes. In some cases
bag life has been increased by the use of lower air-to-cloth values requiring less
frequent pulse cleaning. Overall, there appears to be no operating limitations due to
selection of pulse cleaning pressure or fabric type, or application on bubbling or
circulating bed boilers.
A worldwide survey of pulse-jet baghouse performance funded by EPRI and the
Canadian Electric Association (6) has revealed that the concern that pulse-jet
baghouses have inherently high maintenance costs is unfounded. Large utility, PC
boiler installations of PJFFs are currently in place in other parts of the globe, and they
have not demonstrated such problems, provided reliable and sturdy components are
used and intelligent design and fabrication details are observed.
An attractive aspect of pulse jet baghouses is the potential retrofiting of pulse-jet
cleaned baghouse modules within old electrostatic precipitator (ESP) casings. This is
particularly attractive to utilities considering plant life extension by replacing their old
PC boilers with an AFBC boilers. If the old plant were originally equipped with an ESP,
it can be cost effective to install pulse-cleaned modules within the ESP box. Longer
bags (in excess of 20 feet in length) have been successfully applied in pulse-jet
cleaned baghouses (6). The use of long bags reduces plan area requirements and
often is essential for this type of retrofit into ESP casings.
Ultimately, the selection of one type of baghouse technology over another can be
made only after the options have been compared on a technical and economic basis
for the site specific conditions. Not only must capital cost be evaluated, but also
significant operating and maintenance costs, such as filter bag replacement, must be
considered. The imposition of an ultralow emissions limitation theoretically might
require a low-ratio baghouse. However, the worldwide survey (6) indicates, as noted
previously, that PJFFs can provide outlet emissions performance at levels comparable
to low-ratio baghouse technologies at well below the NSPS of 0.03 Ib/MBtu.
-------
STARTUP AND SHUTDOWN PROCEDURES
A critical factor for baghouses is the adherence to proper startup and shutdown
procedures. The performance of a filter bag is sensitive to the bag's history, and
improper procedures followed during startups and shutdowns can affect the
performance of the cloth for its life. The primary concern for any boiler application is
protecting the fabric from moisture and acid condensation brought about by dewpoint
excursions during startups as well as the deposition of hydrocarbons from oil
combustion during startup. The use of auxiliary fuels during startups for lengthy
periods of bed heat-up and other operational quirks make such procedures even more
important for an AFBC baghouse.
The procedure recommended by most baghouse manufacturers and followed at most
of the sites visited (6) is the precoating of any new fabric prior to startup and/or
admitting any flue gas to that fabric. Many different substances are used, such as
inert fly ash, pulverized limestone (CaCOg), diatomaeeous earth, and hydrated lime
(Ca(OH)2) (not pebble lime, quicklime, or lime (CaO) that is not hydrated). In any
event, the precoat material must be coarser than the ash or dust to be filtered and
must be inert and not react (in a harmful way) with moisture, acids, or other flue gas
constituents to form a dustcake that becomes sticky, breaks down, hardens, or in
other ways becomes difficult to remove.
In AFBC units auxiliary fuels must be burned to preheat the bed material prior to
adding coal to the boiler. In heating the limestone, its calcination process begins and
unhydrated lime will carry over to the baghouse. Additional moisture results from
burning gas or oil, which, in combination with reacted or unreacted lime, can produce
a tenacious dustcake on bags that have not been protected with a precoat Many sites
in North America use the shovel method to load the inert precoat through hatches in
the ductwork while the ID fans are in operation. European operators typically use a
precoat system which makes the process easier and perhaps more uniform. Many
sites, and especially those in Europe that operate seasonally, precoat as part of every
major shutdown for the season and prior to the startup for a new season.
The importance of developing proper startup and shutdown procedures is exemplified
by the experiences of two pulse-jet fabric filter installations. Their problems related to
unexpected acid-dewpoint excursions resulting from carbon-catalyzed dissociation of
calcium sulfate during slumped-bed operation (8). At the first site, premature failure of
felted Nomex bags was traced to temporarily high levels of SOx. At the other site,
relatively high levels of SOx at low gas temperatures resulted in deposition of these
adds on the filter cake, with the resulting reaction causing a hard, impermeable
dustcake to develop on the Ryton needlefelt bags.
-------
During shutdowns at the first site with the Nomex bags, the PJFF was normally left
on-line since, once coal firing ceased, the baghouse could be purged with hot air from
the boiler. However, early in the plant's history, temporary short-term shutdowns were
occasionally required for numerous reasons. Typical S02 concentrations during
normal operation were at the 110 ppm level. However, the plant discovered that very
high S02 levels, on the order of 500 ppm, could occur for periods up to 5 hours or
longer when the bed was slumped during these temporary shutdowns. These
excursions in S02 concentration were occurring during periods when moisture and
acid would condense on the bags because of the low gas temperature. These
conditions eventually were determined to be the cause of acid attack on and
premature failure of the Nomex bags.
The plant developed a theory as to why the SOx excursions occurred. During the
slumped bed condition, the bed remains hot, and consists mainly of burned and
unbumed coal and primarily spent sorbent (CaSO^. Without the addition of fresh
sorbent, the S02 concentrations in the flue gas would soar. This was not perceived,
initially, to be a problem since the plant's SOx limits are in terms of pounds per day,
and their overall daily emission limit would not be significantly affected, since despite
the high concentration in the flue gas, the volume of the flue gas was quite low at 5 to
15% of the full-load volume. It was theorized that in the hot reducing atmosphere of
the slumped bed dissociation of the calcium sulfate occurred, thus generating the high
SOx concentrations. Since discovering the problems associated with such a slumped
bed mode of operation, the plant has started to add limestone to the bed during
periods of temporary outages.
During the startup and early operational period at the second AFBC unit, achieving
efficient carbon burnout of the fuel proved difficult Analyses of ash carryover to the
pulse-jet baghouse indicated loss on ignition (LOI) figures typically above 30% and
often as high as 50%. Eventually, pressure drop across the baghouse climbed until the
unit had to be shut down. To resolve the immediate high pressure drop problem, the
Ryton filter bags were removed and cleaned off-site by means of a gentle washing
cycle in industrial washing machines. These bags subsequently were hung to dry.
Testing of the washed fabric revealed an increase in permeability and a 25% increase
in outlet emissions when compared to new bags. However, the original emissions
were quite low and the result was that the plant was still in compliance with its
limitations. Reasonable pressure drops were restored to this facility.
Initially it was suspected that the numerous, long-term startups encountered in the
plant's early operating phase on natural gas resulted in excessive moisture excursions.
However, theory and investigations eventually indicated that the high carbon in the
recycled ash and spent/unspent sorbent mixture acted as a catalyst to dissociate the
calcium sulfate into its components during the numerous slumped bed periods. The
resulting relatively high emissions of SOx at low gas temperatures resulted in
deposition of these acids on the filter cake. The chemical reactions within the
-------
dustcake caused a hard, impermeable ash layer to develop on the bags. Subsequent
efforts to permanently solve the problem have been aimed at reducing LOI values, with
some success.
SUMMARY
Fabric filtration has worked well on both industrial and utility-sized AFBC boilers.
Fabric filtration has been applied to AFBC boilers up to 165 MWe in size. While the
physical and chemical properties of the AFBC fly ash cause this ash to be somewhat
more difficult to handle, properly designed fabric filtration systems providing adequate
bag cleaning energy have worked well in maintaining low to moderate filtering
pressure drops without a degradation in outlet emission rates. With proper attention
to baghouse design, fabric selection, startup and shutdown procedures, and other
O&M considerations, bag life has been good. Evidence of good performance by
fabric filters downstream from AFBC boilers in a full-scale utility setting has been
provided at EPRl's Nucla aid Shawnee Demonstration Plants, and at pulse-jet facilities
worldwide.
ACKNOWLEDGEMENTS
The data presented in this paper were obtained under three Electric Power Research
Institute projects: RP1179-19 (Project Manager - Mr. Thomas J. Boyd), RP2303-21
(Project Manager - Mr. Thomas J. Boyd), and RP1129-21 (Project Manager - Dr.
Ramsay L Chang). The assistance of Mr. Richard Carson of the Tennessee Valley
Authority and Mr. Thomas Heller and Mr. Robert Melvin of Colorado Ute Electric
Association are gratefully acknowledged.
REFERENCES
1.) Technical Brief No. TB.GS.88.11.89, "AFBC Fabric Filter Monitoring," Electric Power
Research Institute, Palo Alto, CA.
2.) P.V. Bush, T.R. Snyder, and W.B. Smith, "Filtration Properties of Fly Ash from
Fluidized Bed Combustion," JAPCA 37 (11):1292 (1987).
3.) K.M. Gushing, P.V. Bush, T.R. Snyder,"Fabric Filter Testing at the TVA Atmospheric
Fluidized-Bed Combustion (AFBC) Pilot Plant," Electric Power Research Institute
Report No. CS-5837, May 1988.
-------
4.) D.H. Pontius, K.M. Gushing, R.R. Wilson, "Proceedings: Workshop on Pulse-Jet
Baghouse Technology," Electric Power Research Institute Report No. GS-6210,
January 1989
5.) K. M. Gushing, R. L Merritt, "Design, Performance, Operation, and Maintenance of
Fabric Filters in the Utility Industry," Electric Power Research Institute Report No.
GS-7287, April 1991.
6.) V.H. Belba, W.T. Grubb, "Pulse-Jet Baghouses: User's Survey," Electric Power
Research Institute Report No. GS-7457, August 1991.
7.) K.M. Cushing et al, "Fabric Filter Monitoring at the CUEA Nucla AFBC
Demonstration Plant," Presented at the Eighth EPA/EPRI Symposium on the Transfer
and Utilization of Particulate Control Technology, San Diego, CA, March 20-23, 1990.
8.) Unpublished Site Visit Reports by V.H. Belba regarding sites 19 and 31 for EPRI
Project RP1129-21.
-------
Table 1. AFBCBaghouse Design and Performance Qata
*
Plant
Location
Type
UWe
Design Flowrate
Claaiilne
Fabric
Design
M (insured
f-f
Tubesheet
Particulate
(acfm)(000)
Method
Alr-to-Clolh
Airto-Clotll
Pressure
Pressure
Emission Rate
Ratio
(•vmln)
Ratio
ffUmint
Drop
(In. H20)
Drop
(In. H20)
(ib/MBlu)
1
Ulfflly (Pllol)
USA(KY)
Bubbling
20
100
no
14 oz Woven Glass -
i*
aa-ia
2.3-7.8
2.1-11.7
<0.03
1
Utility (Pilot)
USA(KY)
Bubbling
20
100
mm
14 oz Woven Glass
15
1.0-2.0
2.0-6.7
ie-e.0
<0.03
2
Utility
USA(KV)
Bubbling
160
613
RG
10 oz Woven Glass
1.6
14
6
6.1
«0.03
2
Utility
USA(KY)
Bubbling
160
513
RG/S
lOoz Woven Glaw
1.6
1A
6.6
4.6
<0.03
3
Uiillty
USA (CO)
Circulating
110
215
SO
14 oz Woven Glau
2.4
2.4-2*
5-6.6
3.7-6.2
0.007
4
Utility
USA(TX)
Circulating
ISO
650
so
10 ozWwan Glass
2.2
2.15
4.25
3.5
0.008
S
Chemical Plant
Japan
Bubbling
11.6
#7
PJ (IPAV)
Nomex
4.2
3.87
5.6
-
0.0041
a
Manufacturing Plant
Japan
Bubbling
•
68
PJ (LPiHV)
Nontax
2.87
3.28
•
2.36
0.005/
7
CoGan Facility
USA
Circulating
55
203
PJ (HPA.V)
22 oz Woven Glass
3.6
3.86
8.4
7.3
0.0064
s
CoGen F scHliy
USA
Bubbling
19
81
PJ (HP/LV)
Nontax
3.2
»
•
-
0.0168
8
CoGen Faculty
USA
Bubbling
IS
ai
PJ (HP/LV)
Ryton
3.2
3.S6
4.6
2.8
0.0185
10
Relinery
USA
Bubbling
-
145
PJ (HP/LV)
16 oz Woven Glass
4.S
2.6
4.6
3.6
-
11
Steam Plant
USA
Circulating
-
182
PJ (HP/LV)
22 oz Woven Glass
3.2
2.7
4.7
3.3
0.0018
12
CoGen Facility
Germany
Circulating
37
111
PJ (HP/LV)
Ryion
4.6
3.4
6.6
-
0,0114
13
Power/Heating
Germany
Bubbling
41
161
PJ (1P/1V)
Nome*
3.8
1J
2.41
1.41
0.0095
14
CoGen Facility
Germany
Circulating
48
185
PJ (HP/LV)
Rvlon wl qlaze
3.04
2.37
6.02
-
0.0095
NOTE: BQ ¦ Reverse Gat Cleaning
RG/S • Rw«h 6« Cisa/ilng wilh Sonic Assistance
SO ¦ Shake-OeltaM Cleaning
PJ - Puis* Jet Cleaning
(Source: RalsrancM 3.5, 6, 7)
Table 1. AFBC Baghouse Design and Performance Oala (continued)
«
Plant
Location
Type
UWe
Bag Lite
m\
Expected Life
(y«>
Comments
1
UUMy (Pilot)
USA (KY)
Bubbling
20
4
4
Nat currently In operation
1
Utility (Pilot)
USA (KY)
Bubbling
20
4
4
Not currently In operation
2
UttUiy
USA(KV)
Bubbling
160
1 * (FBC)
5t
Bags originally Installed In mid-1880s lor PC boiler
2
Utility
USA (KY)
Bubbling
160
H(FBC)
5t
New horns Installed last quarter o( 1090
3
Utility
USA (CO)
Circulating
110
2»
4t
Fabric abrasion (tubesheet area) on some bags
4
Utility
USA(TX)
Circulating
150
1 +
4t
Very good performance record lo date
S
Chemical Plant
Japan
Bubbling
11.S
2,5t
3
.
6
Manutacturing Plant
Japan
Bubbling
•
1*
3
Sand only In bed
7
CoGan Faculty
USA
Circulating
55
1.25t
Oust abrasion due to How distribution
8
CoGen Facility
USA
Bubbling
19
1.5
Abrasion-misaligned plpes/S03 attack
8
CoGen Facility
USA
Bubbling
18
It
.
10
Rehnery
USA
Bubbling
28*
4* years but at low cap. factor, 2.8 service yrs so tar
11
Steam Plant
USA
Circulating
•
It
.
12
CoGen Facility
Germany
Circulating
37
1.25t
Bags washed In machine
13
Powei/I tutting
Germany
Bubbling
41
1.13
2
14
CoGon Facility
Germany
Circulating
48
1.17*
•
-
-------
Table 2
DESIGN INFORMATION FOR THE CUEA NUCU AFBC BAGHOUSES
Baqhpyy? ft, #3, & #3
Baahouse #4
Baghouse manufacturer
Number of compartments per baghouse
Bags per compartment
Bag size
Bag manufacturer and model number
Bag fabric
Bag fabric finish
Bag cleaning method
Cloth arm per bag
Cloth area per compartment
Cloth area per baghouse
Total cloth area
Design filtering air-to-cloth ratio,
for ail 30 compartments ami
fulMoad flow of 414,000 acfm
Wheelabrator-Frye
6
112
8 in. x 22 ft
Fabric FBters #504
3x1 twill, warp out,
10% Teflon B
Shake/deflate
44.31 ft2
4,963 ft2
29,778 ft2
Research-Cottrell
12
180
8 in. x 22 ft
Fabric Filters #504
3 x 1 twill, warp out,
10% Teflon B
Shake/deflate
44.31 ft2
7,978 ft2
95,712 ft2
185,046 ft2
2.24 acfm/ft2 (gross)
2.50 acfm/ft2 (net)
2.76 acfm/ft2 (net-net)
Bag cleaning initiation
Bag cleaning set point
Compartment cleaning frequency
Compartment cleaning sequence
Deflation air-to-cloth ratio
Shake frequency
Shake amplitude
Pressure drop
5.0 in. H20 (slow mode)
6.0 in. H2Q (fast mode)
360 s (slow mode)
10 s (fast mode)
Null (25 s). Deflate (45 s)
Shake (5 s, 15 s after
deflation starts).
Null (15 s)
0.3 acfm/ft2
4 Hz
1 in.
Pressure drop
6.0 in. HzO (slow mode)
7.0 in. H20 (fast mode)
360 s (slow mode)
10 s (fast mode)
Null (25 s), Deflate (45 s)
Shake (5 s, 15 s after
deflation starts),
(Null 15 s)
0.3 acfm/ft2
3 Hz
1 in.
High pressure drop alarm
High pressure drop bypass
Bag tension
7.0 in. HzO
No bypass available
TO lb
8.0 in. H20
9.0 in. HaO
60 lb
Compartment isolation available
Compartment vent system
High inlet temperature bypass
Low inlet temperature bypass
Hopper size per compartment
Yes
No
No bypass available
No bvpass available
98 fr
Yes
Yes
320"F
180-F
230 ft3
-------
Table 3
DESIGN INFORMATION FOR THE TVA SHAWNEE AFBC BAGHOUSE
Unit to
Baghouse manufacturer
Number of compartments per baghouse
Bags per compartment
Bag size
Bag manufacturer and model number
Bag description
Bag cleaning method
Cloth area per bag
Cloth area per compartment
Total cloth area
Design filtering air-to-cloth ratio,
based on full-load flow of 472,000 acfm
Buell-Envirotech (GEESI)
10
324
12 in. x 35 ft
Fabric Filters, Midwesco,
W.W. Criswell
3 x 1 twill, warp out,
9% Teflon B, 14 oz/yd2
Reverse-gas w/sonic assistance
108 ft2
34,992 ft2
349,920 ft2
1.63 acfm/ft2 (net)
Bag cleaning initiation
High pressure drop alarm
High pressure drop bypass
Bag tension
Compartment isolation available
Compartment vent system
High inlet temperature bypass
Pressure drop
Yes
Yes
75 lb
Yes
Yes
Yes
-------
I I I I i I I I
to 15 20 2B 20 2.2 2.4 2,#
SPECIFIC SURFACE AREA, rr^/g PARTICLE DENSITY. g/ar>3
n tp s m s
so
70
80
OUSTCAKE POROSITY. %
90
COULTER MM0. jjm
5 10 1i
RELATIVE OAS FLOW RESISTANCE, m
P*S T
2 3
DRAG EQUIVALENT DIAMETER, nm
J
20
RAWOl FOR ALL PtAHTS
CS3 SPRAY DRYER
¦¦AFBC
f '1 PC
10
NS
20 30
MORPHOLOGY FACTOR
50
ssa
7 9
EOUtUBRtUM pM
11
13
SPtOWC AP»C HANTS
T'tWORI fGfWC - 1S0MWI
N - CUCA NUCLA 4CFBC ~ 110 MW|
S-TVA4HA¥W#EEfAF*C-160MW>
P • TVA PHOT tAFIC - 20 MW)
Figure 1. Range of values for eight physical properties of dustcake ashes in the EPRI
Fabric Rlter Database. Data for four AFBC plants are highlighted.
REVERSE GAS WITH SONIC ASSISTANCE
FILTERING AIR-TO-CLOTH RATIO, acfm/ft2
Figure 2. Comparison of tubesheet pressure drop versus filtering air-to-cloth ratio at
the 20-MW TVA AFBC Pilot Plant baghouse for reverse-gas cleaning and reverse-gas
cleaning with sonic assistance.
-------
0
N
1
a
o
a
c
a
C
I
o
a
c
a
£
10
9
8 H
7
6 -
3 -
4
3 -
2 -
1 -
0
AFBC Baghouses Worldwide
Praraur* Drop Parfomwmc*
/
/
/
a R*wr*« Gas (Rant #1)
+ Qaa/S (Plant #1)
© Rmtn Sas (Plant #2)
* Rmni Gas/S (Plant #2)
A Shak»/D«f!at» (Plant #3)
A Sbok*/D*fl
-------
0
N
1
a.
o
a
c
a
E
I
o
a
c
a
£
10
9
8
7
6
9
4
3
2
1
FULL-SCALE St PILOT-SCALE AFBC BAGHOUSES
BaghouM Performance Data
~
+
O
A
X
7
TVA 20—WW Pilot (RG)
TVA 20—MW Pflot (RG/S)
TVA 160—MW D«mo (RG)
TVA 160—MW 0«mo (RG/S)
CUBA 110—MW Dtmo (S/D)
TOP 150—MW
¥
7
1 1 1 1 J 1 1 1 1 1 1 1 1
0 0.4 0.8 1.2 1.6 2 2.4 2.8
Filtering Air-to—Cloth Ratio, ft/min
Figure 5. Performance data for fabric filters downstream from four U.S. utility AFBC
boilers.
0
N
1
a
o
•
v
c
o
C
9
e
o
G
10
9
8
7
6
5
4
3
2
1
PC & AFBC BAGHOUSE PERFORMANCE
Fuil—Seal# UtSKy Fabric Filters
AA
~ Nucla 110—MW AFBC (SO)
O TVA 160—MW AFBC (RG)
+¦ TVA 160—MW AFBC (RG/S)
« Pulverized Coal (RG)
A Pulverized Coal (RG/S)
x Pulverized Cool (SO)
1 1 1 1
t 2
Air—to—doth Ratio, ft/mln
Figure 6. Comparison of baghouse performance at the two EPRI AFBC
Demonstration Plants and a variety of pulverized-coal utility boilers.
-------
SNAP-RING BAG ATTACHMENT
CYLINDRICAL BAG.
NO FOLDS OR PLEATS.
NO ASH ON FLOOR
EXCELLENT BOTTOM
CUFF MANUFACTURING,
NO PUCKERS
^16
ASH-LADEN
FLUE GAS
(nucla)
LOWER BAG NOT CYLINDRICAL.
FOLDS AND PLEATS EXPOSED
TO FLY ASH
ASH INFILTRATION
ASH ON FLOOR
REDUCED DIAMETER
POOR BOTTOM CUFF
MANUFACTURING -
MANY FABRIC PUCKERS
Figure 7. Schematic drawing showing ideal snap-ring bag attachment and actual
snap-ring attachment experienced at the CUEA Nucla Demonstration Plant baghouse.
120
2 100
GC
3
Q
<
CD
>
-I
cc
111
I1-
K
<
3
O
_i
2
o
AW,
4th Q 1st Q 2nd Q 3rd Q
1989
^ 1988 ^
<4
4th Q
1st u
2nd u
1990
Figure 8. Number of bag failures per quarter for the last seven quarters of baghouse
operation during the CUEA Nucla Demonstration Plant project {October 1988 throuqh
June 1990).
-------
700
O 500
100
NEW
5.000
SERVICE TIME, h
10,000
Figure 9. Effects of service time on Mullen Burst strength of fabrics from the CUEA
Nucla Demonstration Plant baghouse.
TV A SHAWNEE STEAM PLANT
STACK
STACK
10
OLD PC BOILERS (1-9 STILL IN OPERATION)
TURBINE ROOM (1-10)
Figure 10. General layout of TVA's Shawnee Fossil Steam Plant. Unit 10 is the new
160-MW AFBC (bubbling bed) Demonstration Plant.
-------
TVA 160 MW AFBC DEMONSTRATION PLANT
Dojhowi Pwfonflorci Dulu
O
M
z
•
a
c
a
E
I
0
1
•
9
C
a
E
3 ~
2 -
0,4
i
0.S
T
1.2
Jim 90 (RG)
Oct 89—F«fa 90 (RG)
Jul 91 (RG/S)
Aug 90 (RG/S)
May—S«pt 89 (RG)
1.6
2.4
Ali—to—Cloth Ratio, ft/min
Figure 11. Baghouse performance data for the TVA Shawnee AFBC Demonstration
Plant during five operating periods, three with reverse-gas cleaning and two with
reverse-gas cleaning with sonic assistance.
------- |