United States
Environmental Protection
Agency
Office Of Air Quality EPA-454/R-98-015
Planning And Standards September 1997
Research Triangle Park, NC 27711
Air
OFFICE OF AIR QUALITY
PLANNING AND STANDARDS
(OAQPS)
FABRIC FILTER
BAG LEAK DETECTION GUIDANCE
-------
FABRIC FILTER BAG LEAK DETECTION GUIDANCE
-------
Prepared for
U. S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Emissions, Monitoring and Analysis Division
Emission Measurement Center (MD-19)
Research Triangle Park, NC 27711
September 1997
-------
FABRIC FILTER BAG LEAK DETECTION GUIDANCE
Prepared for:
U. S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Emissions, Monitoring and Analysis Division
Emission Measurement Center (MD-19)
Research Triangle Park, NC 27711
Attn. Mr. Daniel Bivins
Prepared by:
Midwest Research Institute
Suite 100
5520 Dillard Road
Cary,NC 27511
September 1997
-------
TABLE OF CONTENTS
Page
1.0 APPLICABILITY 1
2.0 EMISSION SOURCE AND CONTROL DEVICE DESCRIPTIONS 2
2.1 FABRIC FILTERS 2
2.2 EMISSION SOURCES 4
3.0 MONITORING SYSTEM 5
3.1 PRINCIPLE OF OPERATION 6
3.2 FACTORS THAT AFFECT TRIBOELECTRIC MONITOR
PERFORMANCE 7
3.2.1 Composition of PM and Probe 7
3.2.2 Velocity 7
3.2.3 Particle Size 7
3.2.4 Charge 7
3.2.5 Accumulation of PM on the Probe 8
3.2.6 Particle Shape 8
3.2.7 Temperature 8
3.2.8 Relative Humidity 8
3.3 SIGNAL MONITORING AND ALARMS 9
4.0 SYSTEM MATERIAL SELECTION AND PROBE LOCATION 9
4.1 SENSOR ASSEMBLY MATERIAL SELECTION 9
4.2 SENSOR LOCATION 10
4.3 SIGNAL PROCESSING ELECTRONICS 10
5.0 MONITORING SYSTEM OPERATION 10
5.1 APPROACH TO MONITOR SET UP 11
5.2 MONITOR SET UP PROCEDURES 15
5.3 MONITORING SYSTEM ADJUSTMENTS 17
5.4 RESPONSE TEST 17
5.5 ELECTRONICS DRIFT CHECKS 18
6.0 QUALITY ASSURANCE PROCEDURES 18
6.1 SENSOR INSPECTION AND CLEANING 18
6.2 MONTHLY CHECKS 19
6.2.1 Response Test 19
6.2.2 Electronics Drift Check 19
6.3 ANNUAL INSTRUMENT SET UP 19
6.4 RECORDKEEPING 19
7.0 REFERENCES 20
-------
LIST OF FIGURES
No. Page
Figure 1. Reverse-air cleaning method 3
Figure 2. Shaker-type cleaning method 3
Figures. Pulse-jet cleaning method 4
Figure 4. Monitoring system schematic 6
Figure 5. Installation location for a negative-pressure fabric filter application 11
Figure 6. Effects of sensitivity adjustment 11
Figure 7. Effect of response time on a typical baghouse cleaning peak 13
LIST OF TABLE
No. Page
TABLE 1. COMMON INDUSTRIAL APPLICATIONS FOR FABRIC FILTERS ... 5
VI
-------
FABRIC FILTER BAG LEAK DETECTION GUIDANCE
This document provides guidance on the use of triboelectric monitors as fabric filter bag leak detectors. It does
not impose regulatory requirements. The guidance addresses only one suggested approach to the use of bag leak
detectors. However, proper setup and operation of a bag leak detector can vary with site-specific conditions and those
conditions may dictate variances from the approach suggested in this guidance.
This document includes fabric filter and monitoring system descriptions; guidance on monitor selection,
installation, set up, adjustment, and operation; and quality assurance procedures. The monitoring system description and
information on monitor selection and installation was taken primarily from information received from one instrument
vendor.1 The monitor set up procedure in this guidance was developed based on testing conducted on shaker and pulse-
jet baghouses; however, the guidance is expected to apply to reverse-air baghouses as well.2'3
1.0 APPLICABILITY
Several types of instruments are available to monitor changes in paniculate emission rates for the purpose of
detecting fabric filter bag leaks or similar failures. The principles of operation of these instruments include electrical
charge transfer and light scattering. This guidance applies to charge transfer monitors that use triboelectricity to detect
changes in particle mass loading. Charge transfer monitors based on electrostatic induction are also potentially
applicable, but sufficient information was not available to include them in this guidance.
The set up procedures described in this guidance are intended to allow the operator to identify upset conditions
within the baghouse (e.g., torn bags) using real time data. This guidance is not intended to evaluate changes in the long
term performance of the baghouse system, nor does it apply to applications in which the monitoring system attempts to
quantify emission rates. The guidance assumes an emission source with relatively constant exhaust gas flow rate and
paniculate matter (PM) characteristics. This guidance is not appropriate for applications in which these factors vary
significantly. In addition, only fabric filters (both positive and negative pressure) with exhaust gas stacks are covered by
this guidance.
2.0 EMISSION SOURCE AND CONTROL DEVICE DESCRIPTIONS
This section contains information on the different types of fabric filters and the types of emission sources they are
used to control. Information on fabric filter types and fabric filter operation was taken from References 4 and 5.
2.1 FABRIC FILTERS
Fabric filters are one of the most widely used devices for controlling emissions of PM. A fabric filter system
typically consists of multiple filter elements, or bags, enclosed in a compartment, or housing. The process stream
1
-------
typically enters the housing and passes through the filter elements, and PM accumulates as a dust cake on the surface of
the bag. This dust layer becomes the effective filter medium. The filter elements are cleaned periodically to remove the
collected dust. A short-duration spike in paniculate emissions occurs immediately following cleaning due to the loss of
the dust cake.
Fabric filters generally are classified by cleaning method. The four types of cleaning methods are reverse-air,
shaker, pulse-jet, and sonic cleaning. Reverse-air fabric filters are cleaned by back-flushing the filters with low pressure
air flow, which is provided by a separate fan. Figure 1 depicts the reverse-air cleaning method. In shaker-type systems
(Figure 2), a reciprocating motion is mechanically applied to knock the filter cake off the bags. Pulse-jet fabric filters
use high-pressure compressed air, which creates a shock wave that travels along the bag, thereby loosening accumulated
dust from the filter material (see Figure 3). Sonic cleaning employs a sonic horn to induce acoustic vibrations in the
fabric. This method generally is used to enhance shaker and reverse-air cleaning systems.
Fabric filters also can be can be classified as either positive- or negative-pressure designs, depending upon the
location of the fan(s) that provides the motive force for the exhaust stream through the unit. The fan is located upstream
of the filter housing in a positive-pressure (forced-draft) unit, and downstream of the filter housing in a negative-pressure
(induced-draft) unit. Positive-pressure baghouses require no ductwork or exhaust stack downstream of the unit, making
bag leak detection more difficult. As such, this guidance does not apply to positive pressure baghouses without exhaust
ductwork or an exhaust gas stack.
Fabric filters are capable of extremely high control efficiencies of both coarse and fine particles; outlet
concentrations as low as 20 mg/dscm (0.01 gr/dscf) can be achieved with most
-------
Figure 1. Reverse-air cleaning method.4
SONIC
VIBRATION
HORIZONTAL
VERTICAL
Figure 2. Shaker-type cleaning method.4
-------
BLAST
OF AIR
Figure 3. Pulse-jet cleaning method.4
fabric filter systems. Fabric filters are not suitable for use if the emission stream contains hygroscopic materials, a high
moisture content, or sticky substances; clogging (blinding) of the filter media can occur in these conditions. Gas stream
temperatures in excess of approximately 288°C (550°F) must first be cooled, unless special ceramic or refractory fiber
bags are used. Either of these modifications can add significantly to the cost of the control system. In addition, fabric
filters generally are not preferred for use on highly corrosive exhaust streams or to remove high levels of soluble gases
from exhaust streams. Charge transfer monitors are particularly suited to the same type of applications that use fabric
filters for control of particulate emissions.
2.2 EMISSION SOURCES
Fabric filters are used in a wide variety of industrial applications for which efficient removal of PM from
relatively dry exhaust streams is desired. In the mineral product industries, fabric filters are commonly used for emission
control and product recovery for milling operations such as crushing, grinding, and screening. Fabric filters also are the
preferred control device for mineral product pyroprocesses such as cement and lime kilns. In the metallurgical
industries, fabric filters are often used to control emissions from furnaces and boilers. Table 1 lists some of
-------
TABLE 1. COMMON INDUSTRIAL APPLICATIONS FOR FABRIC FILTERS
Industry
Steel
Foundries
Nonferrous metals
Grain handling
Mineral processing
Cement
Asphalt concrete
Glass
Chemical
Power plants
Waste disposal
Sources
Electric arc furnaces3
Sintering plants3
Boilers3
Cupolas3
Lead furnaces3
Copper smelting furnaces3
Zinc furnaces3
Cleaning operations
Grinding mills
Mixers and blenders
Material transfer
Crushers
Grinding mills
Screening operations
Air classifiers
Dryers
Kilns3
Calciners3
Raw mills
Kilns3
Finish mills
Drum mixers
Melting furnaces3
Dryers
Grinding mills
Coal-fired boilers3
Incinerators3
"Cooling of the gas stream or use of refractory fiber bags may be required.
the more common industrial applications for fabric filters. Fabric filters generally are not used with sources
characterized by moist and/or sticky exhaust streams, such as those from wood product dryers.
3.0 MONITORING SYSTEM DESCRIPTION
Triboelectric monitoring systems typically consist of one or more in-stack probes, a cable from the sensor
assembly to the main instrument box, and signal-processing electronics housed in the main box. An example monitoring
system is shown in Figure 4.
-------
PROCESSING
ELECTRONICS
SENSOR
DIRECTION OF
PARTICULATE FLOW
4-20 mA ANALOG
OUTPUT
OR
ALARM VIA
RELAY CONTACT
Figure 4. Monitoring system schematic.1
The following sections describe the principles of operation of triboelectric monitoring systems, factors that affect
the performance of these systems, and signal monitoring and alarms.
3.1 PRINCIPLE OF OPERATION
When two solids come into contact, an electrical charge is transferred between the two bodies. This charge
transfer is known as the triboelectric principle, or contact electrification. As particles in a gas stream collide with a
sensor placed in the stream, the charge transfer generates a current that can be measured using triboelectric monitoring
equipment. The current signal produced by the triboelectric effect is generally proportional to the paniculate mass flow,
though it can be affected by a number of factors as described below. The current, which can be as low as 10"13 amperes,
is amplified and transmitted to the processing electronics. The processing electronics are tuned to the specific
installation and configured to produce a continuous analog output (i.e., 4-20 mA signal) and/or an alarm at a specific
signal level.
All fabric filter bags allow some amount of PM to pass through; this constant bleedthrough is used to establish a
baseline signal. The monitoring system detects gradual or instantaneous increases in the signal from the baseline level.
According to vendor literature (see Reference 1), triboelectric monitoring systems have been shown to detect baseline
emissions as low as 0.1 mg/dscm (0.00005 gr/dscf).
-------
3.2 FACTORS THAT AFFECT TRIBOELECTRIC MONITOR PERFORMANCE
The effects of various PM and gas stream parameters on the triboelectric signal are discussed below. The
discussion is based on information obtained primarily from one vendor of triboelectric monitors.
3.2.1 Composition of PM and Probe
The materials that compose the triboelectric probe and the PM in the gas stream have direct bearing on the
triboelectric signal generated. The farther apart the probe and PM materials are on the triboelectric table, the greater the
charge generated by their contact. Generally, contact between a good electrical conductor and a good insulator
produces the greatest signal. With the standard stainless steel triboelectric probe (a good conductor), a stronger signal is
generated by PM composed of insulating materials than by metallic PM.
3.2.2 Velocity
The greater the velocity of a given particle, the greater the signal generated. Depending on the materials
involved, the relationship of signal to velocity ranges from linear to exponential. Observed exponents have ranged up to
a power of 2 (i.e., triboelectric signal increases with the square of velocity). Thus, the signal output can be very sensitive
to changes in gas stream flow rate.
3.2.3 Particle Size
All other factors being equal, the triboelectric signal per unit mass is greater for smaller particles. Small particles
have a greater surface area per unit mass of material, allowing for more efficient charge transfer. Thus, up to a point, the
triboelectric monitor is more sensitive to small particles. However, at some point in the submicron range, particles no
longer strike the probe because they lack sufficient momentum to break out of the gas stream as it flows around the
probe. The aerodynamic diameter at which this phenomenon occurs varies with the material; denser materials are
detected at smaller sizes than less dense materials.
3.2.4 Charge
Charged particles generate a signal independent of the triboelectric effect when they strike the triboelectric
probe. As a result, the instrument is more sensitive to charged particles than to particles without charge. Conditions
that cause variations in the charge on the PM will result in variable sensitivity.
3.2.5 Accumulation of PM on the Probe
When material accumulates on the surface of the probe, the sensitivity of the triboelectric monitor may be
reduced. Harder materials tend to accumulate slowly, if at all, while softer, stickier materials accumulate more rapidly.
Accumulation of conductive PM on the probe can also cause an electrical bridge between the probe and ground,
generating a large signal.
7
-------
3.2.6 Particle Shape
Particle shape is likely to have some effect on triboelectric signal because, as discussed above for particle size,
shapes with greater surface area per unit mass are expected to generate a greater signal than those with a lower surface-
to-mass ratio. No data, however, are available to quantify what effect, if any, particle shape has on the signal.
3.2.7 Temperature
Gas stream temperature has no direct effect on the signal as long as the temperature remains above the dew point
and below about 1100°F. The triboelectric current generated in the probe is so small that the resistance of the probe is
insignificant, making temperature-induced variation in the conductivity of the probe insignificant. If the temperature
drops below the dew point, water droplets generate a signal in addition to the PM signal. In addition, liquid water on
the probe causes PM to accumulate. Above about 1100°F, the standard stainless steel probe begins to generate
electrons, interfering with the triboelectric signal; this effect increases as temperature increases.
If gas stream temperature affects the nature of the PM, indirect effects on triboelectric signal may occur. For
example, temperature effects on the chemical composition or particle size of the PM would be expected to result in
variations in triboelectric signal. Changes in the gas stream temperature could also indicate a change in process
conditions that could have an effect on PM characteristics.
Any affect of ambient temperature on the electronic components of the instrument can be compensated for
automatically.
3.2.8 Relative Humidity
No direct gas stream humidity effects have been observed as long as the temperatures of the exhaust gas is above
the dew point. If the temperature of the gas stream prior to the monitor drops below the dew point, condensation may
occur and cause false alarms. Indirect effects are possible when the PM is hygroscopic or the PM characteristics are
otherwise sensitive to humidity.
3.3 SIGNAL MONITORING AND ALARMS
Triboelectric monitors include on/off (switch type) and analog designs. These designs differ in the output signal
generated by the electronics. On/off systems operate only with an alarm relay output that is activated at a pre-set level
to indicate a high emission level. Analog systems operate with a continuous 4 to 20 mA signal that corresponds directly
to the relative paniculate emission level. Analog systems usually also include one or more alarm relays. The simplest
analog monitor has an analog gauge with a needle indicating the current signal (percent of scale) and an on/off relay that
is tripped when the input signal reaches the level set by the user. Other monitors may include analog output signals and
gauges, low and high alarms, digital readouts, internal diagnostics, and quality assurance functions. Analog systems are
-------
recommended over on/off systems, so baghouse activity (baseline signal and cleaning peaks) may be tracked visually and
recorded.
4.0 SYSTEM MATERIAL SELECTION AND PROBE LOCATION
The following sections provide guidance on sensor material selection, probe location, and signal processing
electronics.
4.1 SENSOR ASSEMBLY MATERIAL SELECTION
The materials for the probe and insulator should be selected based on the service environment, and selections
should be approved by the manufacturer. Material selection for the insulator is especially important. The insulator is
positioned between the probe and the housing to electrically isolate the probe, and this isolation must be maintained to
assure valid signal transmission. If PM accumulates on the probe sufficiently to bridge over the insulator to the housing,
the current will flow from the housing to the probe, generating false alarms.
Several materials of construction are available for sensors. Probes are often made from stainless steel for
standard applications. Other mateirals that may be used are tungsten carbide for abrasive applications or Inconel for
corrosive applications. Insulators may be made from Teflon (e.g., for abrasive, noncorrosive applications), high-
performance polymers (e.g., for moist gas streams), or ceramics (e.g., for high temperature and/or pressure
applications). Air purge can be used to minimize the buildup of particulate matter on the insulator.
4.2 SENSOR LOCATION
The sensor, or probe, is designed to be mounted directly on the ductwork downstream of the fabric filter
housing. Where practicable, the probe should be installed so that it extends at least halfway across the duct cross-
sectional area. The maximum probe length may be limited (for example, 36 inches). For large ducts (greater than 72
inches), multiple sensors can be installed and electrically connected in parallel. The insulator sleeve should be flush with,
or protrude slightly from, the inner duct wall; it should not be recessed within the duct wall.
The probe should be located, where practicable, in a length of straight duct, a minimum of 2 duct diameters
downstream and one-half duct diameter upstream from any flow disturbance, such as a bend, expansion, or contraction
in the stack or duct. A velocity traverse is recommended, in order to insure the probe is sited in a location that has
similar flow characteristics to the overall exhaust gas stream. In nonmetallic ducts, an electrostatic (Faraday) shield
should surround the duct and be electrically connected to the probe along with an earth ground to isolate the signal from
stray electrical fields. It is important that the probe is well grounded. In addition, the probe should not be installed in a
location that experiences excessive vibration or is in close proximity to a high voltage or current source.
-------
To avoid potential build-up of particles around the probe, it should not be installed at the bottom of horizontal
ducts or pipes. The location should allow ready access for maintenance and allow for removal of the sensor from the
duct for inspection and cleaning. An example installation location for a negative-pressure fabric filter application is
shown in Figure 5.
4.3 SIGNAL PROCESSING ELECTRONICS
The signal processing electronics can be connected directly to the sensor assembly or located at a distance using
coaxial cable. The electronics should not be exposed to temperatures outside the range specified by manufacturers. The
electronics should be protected from excessive vibration and physical damage and accessible for maintenance. The
display should be visible to the operator.
5.0 MONITORING SYSTEM OPERATION
The following sections provide guidance on monitor set up (sensitivity, response time, and alarm levels) and
operation. Methods for checking system response and drift are also included.
BAGHOUSE
PC SOFTWARE
& REMOTE
RECORDING
Figure 5. Installation location for a negative-pressure fabric filter application.
5.1 APPROACH TO MONITOR SET UP
After installation, the sensitivity and response time of the signal processing system are adjusted to establish signal
levels for baseline operation and alarms. Sensitivity is the amplification, or gain, of the system, and this adjustment is
used to establish the baseline signal level as a percent of the system full-scale (for analog systems). The scale is simply a
relativescale from 0 to 100 percent, and the relationship of the signal to the particulate mass emission rate is linear. The
selected baseline level determines the full scale level.
10
-------
Increasing the sensitivity decreases the range to be measured; decreasing the sensitivity increases the range to be
measured. For example, if the sensitivity is set so that baseline emissions are at 2 percent of scale, 100 percent of scale
corresponds to an emission rate of 50 times baseline. However, if the sensitivity is set so that baseline is at 10 percent,
full scale is only 10 times the baseline emission rate. Figure 6 illustrates these effects of sensitivity adjustments.
Decreasing the sensitivity to lower the baseline level results in smaller scale reading changes for a given change in
the input signal level, which reduces the system's ability to detect small changes in PM levels (e.g., changes due to small
bag leaks). A better approach is to use a short response time, discussed below, to smooth the cleaning peaks.
Conversely, increasing the sensitivity to raise the baseline setting results in larger scale reading changes for a given
change
11
-------
80
70
60
LJJ
0
03
^ 40
30
O
99
20
10
0
TIME
Figure 6. Effects of sensitivity adjustment.
12
-------
in input signal level, which can result in nuisance alarms from small changes in PM levels (e.g., from emission spikes
associated with normal cleaning cycles) or cause the cleaning cycle spikes to exceed the scale of the instrument. The
sensitivity is typically set so that normal baseline PM loading is at some level near the bottom of the scale, usually less
than 10 percent.
With a baseline greater than 10 percent, moderate to high cleaning peaks may leave no room for an adequately
high broken bag alarm on scale. Sensitivity is best set so a typical cleaning peak reaches around 30 percent of scale,
leaving plenty of room for an broken bag alarm as a multiple of the typical cleaning peak height, while still allowing
medium and high cleaning peaks to stay within the scale of the graph.
Response time has a smoothing effect on the output signal by allowing the system to average the signal over a
small period of time, thus lessening the effects of a momentary high signal. On a chart recording of the output, a longer
response time results in lower, broader peaks, while a shorter response time results in taller, narrower peaks. In either
case, the area under the curve is identical, and adjusting the response time does not alter the indicated emissions levels.
The shortest response time setting shows the sharp peaks associated with the filter bag cleaning cycle, and the
signal can be used to identify the row or compartment of bags that may require maintenance. However, false alarms may
result from momentary high signals that do not correspond to cleaning cycle peaks. Increasing the response time from
the minimum setting results in a dampening of momentary high signal spikes and smooths cleaning cycle peaks. Long-
term trending of bag wear and overall emissions increases is best monitored by using a long response time; however, a
response time of 5 to 10 seconds is typically recommended by the manufacturer for most filter types because it smooths
momentary high signal spikes while still providing a good representation of baghouse cleaning cycle activity.
Based on data analyzed by the EPA, a response time of 5 seconds typically serves to smooth the baseline and
dampen momentary high signals not associated with a cleaning cycle peak, but still provides an accurate depiction of the
baghouse activity. Figure 7 depicts a typical cleaning peak at 1, 5, 10, and 15 seconds of response time. At a 1 second
response time, the signal is very jagged. At 5 seconds, it is smoothed out well, without overly dampening the cleaning
13
-------
peak. The re&ptpse Lime of 15 seconds piovides the most smoothing, but deueases the lieiglil of thiij particular cleaning
peak from arou
id 20 percent of scale to approximately
Time
Figure 7. Effect of response time on a typical baghouse cleaning peak.
14
-------
15
-------
11 percent of scale. A long response time, such as 15 seconds, may permit a ruptured bag to go unnoticed for a longer
time, while the 5 and 10 second response times prevent false alarms by dampening momentary high spikes very well and
only slightly decreasing the height of the cleaning peak.
Some instruments can be set to incorporate a delay time. When a delay time is used, the monitor does not
indicate an alarm until some set time after an emission increase is detected. The alarm is only activated when the signal
remains above the alarm level for the full delay period.
5.2 MONITOR SET UP PROCEDURES
The following procedures provide a recommended set up when applicable to a given site. Changes to these
procedures or alternate procedures may be necessary to address site-specific conditions.
The baseline level is established as a percentage of output scale by adjusting the sensitivity and response time of
the output signal from the sensor assembly. The alarm level is then set based on the baseline emission level and/or
cleaning cycle peaks. Operating characteristics vary for each baghouse, and these settings are unique to each
installation. The general procedures for setting the baseline and alarm levels for analog systems are given below. The
procedures for on/off systems are similar.
The general procedures for setting the baseline and alarm levels for analog type systems are as follows:
1. Ensure that the process is operating normally with air and particulate flow past the probe and that the fabric
filter system is in good repair (filter bags in good condition, pressure drop normal, etc.).
2. Set the response time to minimum, and delay time to zero.
3. Adjust the sensitivity setting until the baseline emissions are 5-10 percent of scale and typical spikes during
filter bag cleaning are below 50 percent of scale.
4. Increase the response time so that the baseline signal is smoothed and momentary high signals are damped,
but the cleaning peaks can still be seen; a response time of 5-10 seconds is recommended.
5. Set the alarm level at 2 times the maximum height of a typical cleaning spike for bag leak detection. (For
example, if the maximum height of a typical cleaning cycle peak is 30 percent of scale, the alarm level should be set to 60
percent of scale.) If there are no discernable cleaning peaks, the alarm level may be set as a multiple of the baseline, such
as three times the baseline.
Some triboelectric monitors have the capability for dual alarm levels. One level may be set as a multiple of the
cleaning peak height with no delay time to detect broken bags, and a second level may be set as a multiple of baseline
emissions with a delay time set at least as long as the cleaning cycle in order to detect increases in the baseline emission
level.
16
-------
For on/off systems, the alarm level may be fixed at some percent of full range. Therefore, the alarm level is
effectively adjusted by adjusting the sensitivity to a level which results in normal cleaning peaks occurring below the
alarm level and high cleaning peaks triggering the alarm. A response time of 5-10 seconds is also recommended for
on/off type systems so momentary high spikes do not cause an alarm.
Since a short response time is recommended for use in dampening momentary high signals and the alarm level is
recommended to be set as a multiple of the typical cleaning peak height (once sensitivity is adjusted), the use of delay
time is not recommended. This guidance addresses the use of triboelectric monitors as bag leak detectors, not as means
of measuring a mass emission rate. Therefore, the alarm must prompt maintenance of the baghouse and must be able to
detect an abnormally high cleaning cycle peak. The use of delay time may prevent a high cleaning cycle peak from
activating the alarm.
Alternate procedures to set alarm levels may be needed to address site specific conditions. For example, during
one EPA study 3, the monitor response to a bag leak was predominantly seen in the baseline signal. In cases such as this
one, it may be appropriate to consider an alarm level that is a multiple of the baseline level and incorporates a delay time
and a longer response time. For this particular study, setting the baseline at 10 percent of scale, the response time at
2 minutes, the alarm level at 30 percent (three times the baseline), and incorporating a delay time of 1 minute was
appropriate. This setting produced alarms during simulated bag leaks. Again, however, monitor setup details will be
site specific.
Another example of an alternate procedure may be when high humidity conditions cause false alarms. In this
case, a procedure to detune the monitor or otherwise prevent the false alarms may be appropriate. Such procedures
should clearly define when the period that alarms are prevented starts and ends.
5.3 MONITORING SYSTEM ADJUSTMENTS
An initial 30-day trial period is recommended to verify that the set up of the instrument is appropriate, in order to
prevent frequent false alarms and ensure that the instrument has sufficient detection capability. Another reason such a
trial period is recommended is to verify the system selected will perform reliably in the application and environment to
which it is exposed. Some monitors may have higher sensitivity upon initial installation, but over a period of several
days will stabilize and remain repeatable. The monitor lacks the ability to compensate for a buildup of paniculate on the
probe, so conditioning the system to the process environment is critical to reliable and repeatable operation.
After the sensitivity, response time, alarm levels, and alarm delay (if applicable) have been set and undergone the
30-day trial period, they should not be readjusted unless normal process conditions change in a manner that affects the
characteristics of the particles or exhaust gas stream, such as:
17
-------
1. Change out of filter bags, repair of leaks, or other process improvement that would reduce particulate
emissions;
2. Slow drift of signal due to environmental factors such as humidity. If the sensitivity drifts more than -50 to
100 percent from the initial set up, the monitoring system and control device should be inspected and any necessary
repairs performed.
3. Equipment is taken out of service for repair, replacement, or upgrading.
5.4 RESPONSE TEST
The response test is meant to be a check on the operational status of the monitor; it is not an accurate measure of
electronic drift. The system should be tested monthly to ensure a repeatable and reliable response. A test port should be
installed upstream of the probe where a known quantity of dust can be injected into the exhaust gas stream to simulate a
broken filter bag. A specified dusty material and injection procedure should be prescribed that will always be used for
this test. Various quantities of the selected material should be injected until the amount necessary to trigger the alarm is
determined. This quantity of dust should be doubled and used to test the system monthly, in order to verify operation of
the monitor. If the monitor is equipped with a continuous output, the signal response during the dust injection test
should be recorded and compared to testing conducted during previous months. If signal levels differ significantly from
the initial response test, action should be taken to investigate the cause of the discrepancy.
5.5 ELECTRONICS DRIFT CHECKS
The electronics drift checks are meant to be an accurate measure of the monitoring system's electronic drift. A
zero drift check can be conducted by disconnecting the sensor or shielding it from particulate. A sensitivity check can be
conducted with an instrument which generates a low level current similar to the signal generated by the sensor. The
sensor is disconnected from the electronics (or the process is shut down) and the signal generator is connected in its
place. The instrument is then used to send a controlled input signal to the electronics to test the accuracy of the system.
Some models perform automatic internal drift checks at specified time intervals. The electronics should be adjusted if
the drift is greater than 20 percent, or as specified by the manufacturer. Manufacturer's instructions should be consulted
for procedures specific to each model.
6.0 QUALITY ASSURANCE PROCEDURES
Quality assurance (QA) is a critical element of any environmental data collection. It is a system of management
activities designed to ensure that the data collected are of the type and quality needed by the data user. QA procedures
should include the necessary checks of the monitor's functioning, measurement performance criteria, maintenance
18
-------
procedures, and documentation to assess and document the continuing functioning and accuracy of the bag leak
detection monitor. The following QA procedures are suggested to ensure proper monitoring system operation.
6.1 SENSOR INSPECTION AND CLEANING
Each sensor should be inspected at regular intervals to remove any build-up of material that may collect on the
probe or insulator. A build-up of material on the probe may dampen or decrease the signal strength, and material on the
insulator can form a conductive electrical bridge across the insulator, increasing the signal strength and resulting in a
high alarm.
The rate of material buildup on the sensor assembly is dependent upon many factors and will vary for each
installation. Thus, the interval between inspections or probe cleaning may vary considerably among installations.
Inspection and cleaning of the probe and insulator should be in accordance with the manufacturer's recommendations.
6.2 MONTHLY CHECKS
Monthly QA checks should be performed to ensure the monitor is operating properly. If the results of the
response test or electronics drift check are not favorable, the cause should be investigated and any malfunctions
corrected.
6.2.1 Response Test
According to the procedures specified in section 5.4, inject the previously determined type and quantity of dust
into the port installed in the duct to test the operation of the triboelectric monitor and alarm. A specific injection
procedure and dust type should be defined on a case-by-case basis during the set up of the monitoring system. The
output signal response should be recorded and compared to the reading obtained during the initial monitor set up. If the
readings differ significantly, corrective action should be initiated.
6.2.2 Electronics Drift Check
According to the procedures specified in section 5.5, a signal generator should be used, with signal strengths
that match those determined when the monitor was initially set up, to check the baseline and alarm level readouts. A
zero drift check should be conducted; the readouts should be within 20 percent of the set levels. If the readouts do not
meet this criteria, corrective action should be initiated.
6.3 ANNUAL INSTRUMENT SET UP
If the monitor's settings have not been adjusted within a year's time, an annual instrument set up should be
performed. The set up procedures given in section 5.2 should be repeated and documented.
19
-------
6.4 RECORDKEEPING
A record that includes the date, time, condition of each sensor as-found, and a description of any actions taken
should be maintained of all inspections (e.g., probe/insulator cleaning). Records should also be maintained for all drift
checks and response tests performed. Each entry in the log should be signed by the person conducting the inspection,
testing, or maintenance.
The initial instrument set up procedures should also be documented so the annual instrument set up will be
performed consistently. Documentation should include values for the baseline (sensitivity) setting, response time setting,
and alarm level(s) and a description of how each was established. If process changes require the system parameters to
be adjusted (see Section 5.3 of this guidance), the date, adjustments, and reasons for the adjustments should be
documented and signed by the personnel responsible for the modifications. The instrument set up procedures should
then be revised accordingly.
7.0 REFERENCES
1. Auburn International, Triboflow and Triboguard Dust Emission Monitors and Broken Bag Detectors. General
Guidelines for Operation. April 1995.
2. Midwest Research Institute, Evaluation of Triboelectric Monitors. Final Test Report, prepared for U.S.
Environmental Protection Agency, Emission Measurement Center, March 1997.
3. Midwest Research Institute, Evaluation of Triboelectric Monitors on Pulse Jet Fabric Filters, prepared for U.S.
Environmental Protection Agency, Emission Measurement Center, September 1997.
4. U.S. Environmental Protection Agency, APTI Course 413: Control of Particulate Emissions. Student Manual.
EPA 450/2-80-066, Research Triangle Park, NC, October 1981.
5. U.S. Environmental Protection Agency, Operation and Maintenance Manual for Fabric Filters.
EPA/625/1-86/020, Research Triangle Park, NC, June 1986.
20
------- |