-------
-
11 11 1 1 1 1 1 1 1
Range of
Measured Res
sponses
/
j /
/ /
/ /
/
~
/
-
//
//
/s
//
y
4
f
/' /
\ ' J '
' / /
/ /
/
t /
/
/
/
/
-
Average
Measured Res
ponses
4
/
/ '
f / /
' / /
r /
/
~
/
/
f
/ /
/' '''
/ ' /
/ /
S
/ y
A
\
_ + 3% C
Erro
pacity
r Band
/
/
*j
'/ / -
// /
/
/ /
/
L
*
/
V Expect
ed Responses
—
/ j
~ M
,'/A
> ' /
/ /
/
~
~
r
/
/
/
r
1 II 1 | 1 1 II
l l l l | M l I
1 1 1 1 | 1 1 1 1
1 1 111 II II
ii ii | i i i i
1
0 10 20 30 40 50
Actual Opacity (% Opacity)
Figure 5. Observed Opacity Responses for All
Audited Monitors: 89 Monitors
16
-------
C. Monitor Maintenance
1. Approximately 20% of the audited monitors did not
respond correctly to the internal zero and span
functions.
2. For seven monitors, monitor zero and span
adjustments were made without cleaning the exposed
optical surfaces.
3. Approximately 20% of the audited monitors had
excessive dust deposited on the optical surfaces.
4. Four monitors were not aligned properly, and three
of these monitors were out-of-calibration.
D. Monitor Calibration*
1. Approximately 10% of the audited monitors were not
properly calibrated at low range levels (~ 10%
opaci ty) .
2. Approximately 15% of the audited monitors were not
properly calibrated at the midrange level (~ 20%
opaci ty) .
3. Aproximately 20% of the audited monitors were not
properly calibrated at the high range level (~ 40%
opaci ty) .
4. The monitor's response to the internal zero and
span functions is not a good indication of monitor
calibration.
Before the calibration of a monitor was checked, all
exposed optics were cleaned. In addition, the monitors'
preset pathlength correction factor, whether it was
correct or incorrect, was used to correct the audit
filters for stack exit conditions. Thus, the calibration
error determinations exclude errors due to dust
deposition and inaccurate preset pathlength correction
factors.
17
-------
4. DISCUSSION OF RESULTS
4.1 GENERAL OPERATION
At the initiation of the transmissometer field audit
program, all of the monitors were believed to be functioning
properly; however, of the 93 monitors audited, only 89 were
fully operational. In addition, 22 of the operational
monitors audited (25%) had significant biases (>3% opacity)
in the recorded opacity data.
4.1.1 Stack Exit Correlation Analysis
An error in the pathlength correction factor affects
the monitored opacity exponentially. The most common error
in computing the optical pathlength is the use of the
flange-to-flange distance, rather than the stack/duct inside
diameter, for the transmissometer pathlength. Monitoring
systems adjusted for an incorrect pathlength ratio will
generate inaccurate opacity data; however, these data can be
easily adjusted to obtain true stack exit opacities.
Therefore, previously recorded opacity data which are
inaccurate due to an improper preset pathlength ratio can be
corrected and used for their intended purpose.
18
-------
4.1.2 Fault Lamp Analysis
The fault lamps provided a good indication of the
current operational status of the monitoring system. All of
the monitors with activated fault lamps (i.e., five
monitors) had calibration errors in excess of 10% opacity.
4.1.3 Panel Meter Analysis
The control panel meter analysis was not a good
indication of monitor performance, particularly for the Lear
Siegler monitors audited. The control panel meter errors
for the optical density and input scales were usually found
to be outside the recommended range. However, these scales
are not normally used by the sources, and thus, the accuracy
of these scales is not paramount. The opacity scale
correction factor was usually found to be within the
specified range; however, there was no correlation between
the panel meter factor and the calibration error
determination results. Even though the opacity scale on the
panel meter is usually accurate, the opacity responses
displayed on the chart recorder should be used during
monitor adjustments.
For cases where the panel meter (opacity scale) was
determined to be accurate and where the chart recorder
biased the data, a calibration curve can be constructed to
present the panel meter responses (accurate values) versus
the chart recorder responses. This curve could then be used
19
-------
to adjust previously recorded opacity
correct values. Seven monitors (8%)
offsets.
readings to reflect
had chart recorder
4.1.4 Reference Signal Analysis (Lear Siegler)
The reference signal for all of the Lear Siegler
monitors was within the recommended range; one monitor had a
reference signal outside the recommended range, but the
panel meter was not properly adjusted and the monitor was
calibrated correctly. The reference signal value did not
vary significantly for monitors which performed either well
or poorly during the calibration error determination;
therefore, no meaningful relationship could be found between
the reference signal and monitor performance.
4.1.5 Internal Zero and Span Analysis
The monitor zero and span checks did not appear to
provide a direct correlation to monitor calibration.
Seventy-two monitors had accurate zero and span responses;
however, twelve of these monitors had calibration errors
outside the recommended range. Also, seventeen monitors had
zero and/or span responses outside the recommended range,
but only ten of these (~ 60%) monitors had calibration
errors outside the recommended range.
A separate problem concerning zero and span adjustments
was identified at two sources (seven monitors); these
sources had adjusted the zero and span without cleaning the
optics. The zero mode is designed to simulate the amount of
20
-------
light that would be returned to the transceiver under clear
stack conditions. If there is any dust accumulation on the
transceiver window when the zero adjustment is made, then a
lesser amount of light is being returned to the detector,
and thus, an improper adjustment will be made. This
improper adjustment may or may not bias the recorded
opacity. The zero compensation circuit is automatically
reset during the zero adjustment, and thus, the circuit will
respond to the dirty optics as if they were clean.
4.1.6 Monitor Optical Alignment Analysis
Four monitors audited were not properly aligned; three
of the four monitors were out-of-calibration. For two of
these monitors (Dynatron), the source indicated that when
they were properly aligned, the monitors could not be
zeroed. An alternate alignment was determined off-stack
during calibration, and this alternate alignment was
duplicated when the monitor was installed on the stack; both
of these monitors were out-of-calibration.
4.1.7 Zero Compensation Analysis (Lear Siegler)
During the performance audit program, the results of a
comparison of the amount of dust deposition on the
transceiver optics and the corresponding level of zero
compensation showed that the zero compensation circuit
accurately indicates the amount of accumulated dust and
appropriately adjusts the measurements obtained by the
monitoring system. The zero compensation circuit of the LSI
21
-------
opacity monitor indicates the level of dust accumulation on
the transceiver measurement optics and adjusts the monitor
output to compensate for its effect. It does not indicate
the level of dust accumulation on the optical surfaces of
the retroreflector. In addition, the audit results showed
that the zero compensation levels for eight of the monitors
audited (Lear Siegler) were outside the specified allowable
range of 4% opacity.
4.1.8 Dust Accumulation Analysis
Quantitative assessments of dust deposition on exposed
optical surfaces were performed for eighty-five monitors;
sixteen monitors (19%) had excessive dust on the optical
surfaces. It is impossible to determine the correct opacity
levels recorded for times before the audit, since the dust
deposition does not occur at a linear rate. The audits also
showed that dust accumulation does not necessarily occur at
the same rate for the transceiver and retroreflector
measurement optics. The audit data indicate that dust
deposition on the monitor's optics is site-specific and
monitor component-specific.
4.1.9 Calibration Error Analysis
For those instruments which failed the calibration
error test, the calibration error increased with increasing
opacity. Execpt for two monitors, the monitors which failed
the the midrange calibration error test are a subset of
those monitors which failed the high range calibration error
22
-------
test. All of the monitors which failed the low range
calibration error test also failed the midrange calibration
error test.
For five of the monitors that were miscalibrated, the
sources had adjusted the monitor to read neutral density
filters without correcting the filter values for stack exit
conditions. In cases where this problem was the sole cause
of inaccurate data, the past data can be adjusted to reflect
the true stack exit opacities.
Section 60.13(c)(2), Part 40 of the Code of Federal
Regulations states that sources which entered into a binding
contractual obligation prior to September 11, 1974 to
purchase an opacity monitoring system shall demonstrate that
the monitor is capable of measuring emission levels
(opacity) within + 20% with a confidence level of 95%.
Within the confines of this auditing program, this
requirement was determined to mean that any installed
monitor should be able to measure the opacity values of the
audit filters within + 20%. The average calibration errors
for all the monitors audited were well within this + 20%
specification.
Twenty of the twenty-two monitors having calibration
errors > 3% opacity were biased high. Based upon this
experience, it can be postulated that opacity monitors which
are not performing well usually report high-biased opacity
data. Therefore, it is in the source's best interest to
keep its transmissometer well calibrated and maintained.
23
-------
4.1.10 Combining Duct Opacities
During the auditing program, thirty transmissometers
were audited which were installed in exhaust ducts that
transported effluent streams to a common exhaust stack.
None of these common stacks had an installed
transmissometer; thus, the sources had to determine the
stack exit opacity by combining the duct opacities measured
by the respective transmissometers. None of the combined
duct opacities were weighted according to the duct flowrate,
and a few sources only arithmetically-averaged the duct
opacities to obtain a stack exit opacity.
4.1.11 General Survey Analysis
The general audit survey addressed a variety of other
aspects of monitor operation (e.g., the accessibility for
maintenance, the supply of spare parts, etc.); for the most
part, these factors had little correlation with monitor
performance. However, one factor found to be important was
whether the monitor maintenance was performed by plant
personnel or by a monitor servicing contractor. Nearly all
(six of seven monitors) of the monitors serviced by
monitor-servicing contractors had calibration errors of £ 3%
opacity.
Monitor availability, as determined by reviewing the
previous month's data, was 100% for sixty-nine of the
ninety-three monitors audited; the average monitor
availability was approximately 95% for all of the monitors
24
-------
audited. The general survey also indicated that there were
no specific monitor components that had a history of
failure.
4.2 MONITOR SPECIFIC AUDIT PROCEDURE RESULTS
Several problems which were encountered during the
transmissometer field audits were peculiar to one type of
monitor. The problems were not viewed as design problems,
but were caused by source maintenance practices.
4.2.1 Lear Siegler Transmissometers
One problem observed at nine of the fifty-eight Lear
Siegler monitors audited was the accumulation of excessive
dust on the optical surfaces. These monitors have a zero
compensation circuit which is used by most sources to
monitor the dust accumulation and to determine cleaning
intervals. There were a few problems which were identified
during the audits with using the zero compensation circuit
as an indicator of dust accumulation. The zero compensation
addresses only the dust accumulation on the transceiver
optical surfaces, and thus, excessive dust accumulation can
occur at the retroreflector without the source being aware
of the problem. Also, some sources were not aware of the
specified limit of zero compensation, or of what that limit
meant. Performance Specification 1 (Appendix B, 40 CFR 60)
indicates that the zero compensation shall not exceed 4%
opacity. The value displayed on the zero compensation scale
25
-------
is in units of optical density, and some sources are not
aware that a zero compensation limit of 4% opacity equates
to an optical density limit of 0.018 OD.
The final problem identified which was related to the
zero compensation analysis was due to monitor calibration
practices. Zero adjustments were made without cleaning the
optical surfaces. When the zero is adjusted during
conditions of dirty optical surfaces, the zero compensation
is artifically reduced, leading to an incorrect optical
surface dust accumulation analysis by the zero compensation
ci rcuit.
4.2.2 Dynatron Transmissometer
Although only five Dynatron monitors were audited, some
problems were observed which may affect other installed
Dynatron transmissometers. Two of the monitors audited had
fault indications (measurement lamp output was outside the
specified limits) which were not noticed by the source. The
fault conditions were not noticed because the dust cover for
the control unit partially obstructs the view of the fault
lamp. The other problem which was observed was that two of
the monitors were not aligned according to the
manufacturers' recommendations. The source indicated that
the monitor could not be zeroed when aligned as recommended;
an alternate alignment was determined off-stack, and it was
duplicated on-stack.
26
-------
4.2.3 Contraves Goerz and Esterline Angus Monitors
No specific problems were observed that were
attributable to the source maintenance or set-up of the
twenty-seven Contraves Goerz and three Esterline Angus
monitors audited.
27
-------
5. CONCLUSIONS AND RECOMMENDATIONS
5.1 CONCLUSIONS
A. Monitor Performance
1. Installed transmissometers provide accurate opacity
data, and they operate at a high level of avail-
ability.
2. Monitors which are not properly calibrated almost
always introduce a positive bias to the reported
opacity data.
3. The audit results obtained from performing internal
zero and span checks did not necessarily provide
valid indications of transmissometer performance.
Several of the excessive errors resulted because
the sources failed to rename the internal filter
values after the transmissometers had been
recalibrated. Nevertheless, while being but
marginally effective for auditing purposes, these
internal parameters can provide the source with
valuable information regarding the status of the
monitor's stability and calibration.
B. Monitor Operation and Maintenance
1. The auditing program revealed that the intensive-
ness of operation and maintenance practices was
directly related to transmissometer performance.
Those transmissometers which were calibrated and
cleaned with greater frequency provided opacity
data of higher quality. In addition, the audits
indicated that transmissometers maintained under a
service contract significantly out-performed those
maintained by the source alone.
2. There are no generally applicable specific time
intervals for monitor maintenance that ensure
quality data; the maintenance intervals are
site-specific.
3. Dust deposition was found to be site-specific and
component-specific, and the depositions do not
occur at a linear rate; therefore, opacity data
recorded over periods of time during which the
optics are dirty cannot be adjusted to reflect true
stack exit opacities.
28
-------
4. Incorrect pathlength correction factors were due to
the use of the flange-to-flange distance, rather
than the stack/duct inside diameter, for the
transmissometer pathlength.
5.2 RECOMMENDATIONS
A. Monitor Performance
1. The source should ensure that the opacity
monitoring system provides high quality data;
inaccurate opacity data usually are high-based, and
thus, the source may assume that online particulate
control equipment is not performing efficiently,
and the source may report invalid instances of
excess opacity emissions.
B. Monitor Operation and Maintenance
1. The source must implement a comprehensive operation
and maintenance program for their monitoring
systems; this program may be implemented inhouse or
through a monitor-servicing contractor.
2. Each source should perform a trial and error
analysis to determine the appropriate maintenance
intervals for each of its monitors. Initially, a
reasonable time interval for performing zero/span
adjustments, purge air filter cleaning/replacement,
and exposed optics cleaning would be every other
three months. The zero and span functions should
not drift more than 2% opacity between service
checks, and the total dust deposition between purge
air filter cleaning/replacement and optics cleaning
should not exceed 4% opacity for monitors with zero
compensation or 2% opacity for monitors without
zero compensation. If an initial time interval is
insufficient to provide the above results, halve
the time interval of interest.
3. Monitor zero and span responses should not be used
as the sole indicator of monitor calibration.
4. At a minimum, the source should adhere to the
manufacturer's recommendations regarding monitor
specifications.
29
-------
C. Monitor Calibration
1. Care must be taken to ensure that the neutral
density audit filter values are corrected for stack
exit conditions before using them to calibrate the
moni tor.
2. Sources should check that the preset pathlength
correction factor for the monitor is correct; the
monitor pathlength is the inside diameter of the
stack/duct at the monitor location, not the
flange-to-flange distance.
3. No monitor adjustments should be made without prior
cleaning of the exposed optics.
4. Monitor panel meters should not be used to provide
information during the adjustment of monitor
parameters; however, the panel meter should be
adjusted to display correct monitor responses.
30
-------