United States
Environmental Protection
Agency
Atmospheric Research and
Exposure Assessment Laboratory
Research Triangle Park NC 27711
Research and Development
EPA/600/S3-89/019 Sept. 1989
Project Summary
Evaluation of HCI Continuous
Emission Monitors
Scott A. Shanklin, J. Ron Jernigan, and Scott C. Steinsberger
The final report summarized herein
presents the findings obtained from
the field evaluation of commercially
available HCI monitoring equipment
at a municipal waste-fired boiler not
equipped with HCI emission control
equipment. The analyzers were oper-
ated continuously during a two-
month test period.
The measurement techniques em-
ployed by the evaluated HCI monitors
were IR gas filter correlation, specific
ion electrode, wet chemical colori-
metric, dry reaction colorimetric, and
gas membrane galvanic cell.
Except for the gas membrane
galvanic cell monitor, the HCI mon-
itoring equipment produced effluent
measurements in good agreement
with concurrent reference measure-
ments. The results comparing the
CEM data to the reference wet-chem-
istry measurement data indicate no
biases in any of the monitor meas-
urement techniques resulting from
analytical interferences present in
the effluent of this municipal refuse-
fired boiler. Further, both in-stack
dilution systems and the nondilution,
heat-traced sampling system were
found to reliably provide represen-
tative effluent samples to the
analyzers.
This Project Summary was devel-
oped by EPA's Atmospheric Research
and Exposure Assessment Laboratory,
Research Triangle Park, NC, to an-
nounce key findings of the research
project that is fully documented in a
separate report of the same title (see
Project Report ordering information at
back).
Introduction
Certain U.S. Environmental Protection
Agency (EPA) regulations (40 CFR 264,
Subpart 0, Sections 264.340 - 264.351)
suggest that hazardous waste incineration
facilities must monitor HCI on a contin-
uous basis to economically demonstrate
continuous compliance with the HCI
emissions requirement of Section
264.343(b). Already, California and cer-
tain northeastern states (e.g..Connecticut,
Massachusetts, New York, New Jersey,
and Pennsylvania) already require oper-
ating permits for new or proposed refuse-
fired boilers to include a provision that
HCI continuous emission monitors
(CEMs) be installed and operated to
demonstrate HCI removal requirements
when HCI CEM systems become
commercially available. However, not yet
documented are identification of the
various types of commercially available
HCI CEMs or the demonstration of their
effectiveness in continuous monitoring
effluent HCI emissions from hazardous
waste incinerators or from refuse-fired
boilers.
The Quality Assurance Division (QAD)
of the Environmental Monitoring Systems
Laboratory (EMSL), Research Triangle
Park, North Carolina, is responsible for
assessment of environmental monitoring
technologies and systems. QAD has
initiated a field test program to assess
the performance of commercially avail-
able HCI CEMs. The major objectives of
the project were (1) to evaluate the
reliability of multiple HCI analyzers in
terms of the accuracy, precision, and
availability of measurement data, and (2)
to determine the adequacy of sampling
-------
systems for reliably providing effluent
samples to the various HCI analyzers.
The field evaluation was conducted at a
municipal refuse-fired boiler with HCI
effluent concentrations ranging, typically,
between 200 and 500 ppm.
The sampling location was the outlet
duct from the electrostatic precipitator
(ESP). Six HCI analyzers, involving four
different detection techniques and three
sampling systems, were evaluated. Table
1 presents information concerning the
specific analyzers/monitoring systems
selected for evaluation. The reader will
note that the analyzers manufactured by
MDA Scientific, Lear Siegler Instruments
(LSI) and CEA Instruments, designed for
ambient air applications, are not
equipped with a sample conditioning/
handling system.
Test Procedures
The field evaluation utilized concurrent
operations of the various analyzers and
sampling systems to determine factors
affecting the reliability of the equipment.
Performance tests for relative accuracy,
calibration drift, calibration error, and
response time were conducted according
to procedures outlined in "Gaseous Con-
tinuous Emission Monitoring Systems -
Performance Specification Guidelines for
S02, N02, C02, and TRS," EPA-450/3-
82-026, October 1982. Manual sampling,
using a wet-chemical impinger sampling
train, was also conducted. To facilitate
quantification of HCI sample line losses,
flue gas sampling was conducted
simultaneously at two locations - the duct
and the CEMs' common manifold. The
manifold distributed diluted flue gas
sample to the CEMs delivered from one
of the two dilution probes located at the
duct.
The HCI in the flue gas samples was
collected with a sampling train similar to
an EPA Reference Method 6 train. The
absorbing reagent, 15 mL of 0.1 N NaOH,
was added to each of the first two
impingers. The reagent was used in the
manifold impinger trains at 0.001 N
NaOH because of the lower HCI content
after dilution by the dilution probe. The
third impinger (a Mae West design) was
filled with calcium sulfate (Drierite) to
protect the Singer dry gas meter from
moisture. The desired sampling rate
during the relative accuracy testing was 2
L/min with a sampling time of 20 min.
The sampling systems used for the
duct sample consisted of all-glass com-
ponents that contacted the stack gases. A
glass-lined probe and glass components
were used to convey the stack gas to the
duct impinger train. A three-way glass
valve was mounted in-line directly up-
stream of the first impinger.
The techniques used for analysis of
impinger samples were a mercuric nitrate
titration procedure (EPA Method 325.3 -
Method for Chemical Analysis of Water
and Waste, EPA 600/4-79-020) and ion
chromatography (1C). The mercuric ni-
trate titration procedure was used for
analysis of samples in the field to provide
rapid feedback on the operation of the
HCI CEMs. The 1C analysis was restricted
to the laboratory and provided con-
firmation of the titration results.
At the beginning of each test period, all
CEMs were calibrated (at zero and one
upscale calibration point) to the same
standard. Triplicate sampling and ana
ysis, using wet-chemical impinger sam|
ling and mercuric-nitrate titrations, wei
conducted on the calibration gases
verify their concentrations. The thre
analyzers not supplied with a probe ar
sample conditioning/ handling systei
(i.e., the LSI, MDA, and CEA analyzer
were supplied gas samples from eithi
the TECO or Compur dilution systems.
Results And Discussion
Operational problems were exper
enced at the startup of the analyzers ar
monitoring systems. The test personn
and equipment vendors spent the fir
few months of the field evaluation idei
tifying and resolving problems affectir
the reliability of HCI monitor operation.
The CEA Model TGA-410 HCI an;
lyzer, initially provided for the fiel
evaluation, did not respond to changes
sample concentration. CEA represent!
tives suspected a bad electrochemic
gas sensor within the analyzer an
replaced the analyzer. However, the ne
analyzer performed similarly to the one
replaced. As a result, CEA withdrew it
analyzer from the evaluation.
Evaluation results, generally, indical
acceptable HCI CEM performance durin
the relative accuracy test periods. We
chemical impinger sampling was cor
ducted to collect "reference" HCI coi
centration measurements to compare I
the HCI monitoring measurements so thi
relative accuracy could be determine*
The computations of relative accurac
were performed using the procedures
Performance Specification 2, 40 CFR &
Appendix B. A relative accuracy specil
Table 1. HCI Continuous Emission Monitors
Manufacturer
Thermo Electron
instruments (TECO)
MDA Scientific, Inc.
Model
15
Series 7100
Measurement Technique
NDIR gas filter
correlation
Colorimetric
(chemically-treated
cassette tape)
Sampling System
Dilution probe
None
Measurement
Ranges
Variable ranges from
0-5 ppm to 0-5000 ppm
0-100 ppm or less
Available Calibration
Techniques
Cylinder gases
Stain card, cylinder
gases
Lear Siegler
Instruments (LSI)
Compur
CEA Instruments
Bodenseewerk
(BSWK)
TGM-555
677 IR
Colorimetric
(liquid reagent)
None
4150 Ion-sensitive electrode Dilution probe
TGA-410 Gas membrane
galvanic cell
NDIR gas filter
correlation
None
Heated probe, sampling
line, and sampling pump
to heated gas cell
0-50 ppm or less
0-3353 ppm
0-50 ppm or less
0-1000 ppm
Liquid standards, cylindei
gases
Liquid standards, cylindei
gases
Liquid standards, cylindei
gases
Cylinder gases, internal
sealed-gas cell
-------
ation of s 20% was adopted for this
evaluation. GEM wet-basis measurement
data were corrected to dry-basis so that a
direct comparison could be made to the
dry-basis impinger results.
Table 2 presents the relative accuracy
results for each of the CEMs.
Other findings from the relative accu-
racy determinations, not shown in Table
2, are briefly described below:
• The results of the calibration checks
did not always indicate monitor
performance during the relative accu-
racy tests. In each of the four cases
where the relative accuracy result
exceeded the < 20% specification,
the magnitude of the mean difference
term indicated the reason for
exceeding the specification may have
been caused by an improper adjust-
ment to the analyzer calibration. In two
of these four cases, there was good
agreement between the calibration gas
values and the GEM responses to the
calibration gas injections. In one case,
the calibration check results were not
consistent with the relative accuracy
mean difference term (i.e., positive
drift was noted from the post-test
calibration check, but the relative
accuracy mean difference term indi-
cated a negative bias in the flue gas
measurement). On one occasion, a
significant amount of drift was noted
(3.7% relative to instrument span, or
29% relative to the pre-test response),
whereas the relative accuracy result
(7.4%) indicated acceptable monitor
performance.
• The relative accuracy results indicate
that any effects of HCI line losses can
be minimized by injecting the HCI
calibration gases through the entire
sampling system when performing
dynamic calibrations.
• During the final relative accuracy test
conducted on April 29, 1987, the
sampling rate for the last five manifold
impinger samples was reduced from 2
L/min (the prescribed sampling rate) to
1 L/min to determine if the sample
flow rate had an effect on the results.
Significantly lower impinger sample
results were produced at the lower
sampling rate. The stainless steel
hardware mounted on the manifold
could possibly have contributed to
these lower impinger sample values
by adsorbing HCI at the lower flow
rate.
A seven-day calibration drift evaluation
was not performed according to the
procedures of promulgated Performance
Specification 2. However, daily calibration
checks were performed over 3- and 4-
day periods according to the calibration
drift test procedures. These test data
were compiled to demonstrate the capa-
bilities of these HCI analyzers to maintain
daily drift within the 2.5% of span
specification of Performance Specifica-
tion 2 adopted for this evaluation.
Table 3 presents the maximum daily
calibration drift observed for each of the
analyzers.
The calibration drift test was conducted
on the Bodenseewerk monitoring system
for seven consecutive days by using the
data afforded by the automatic calibration
check the Bodenseewerk performs using
zero air and the internal gas-filled cell.
This procedure did not require test
personnel to remain on-site for seven
consecutive days.
The calibration drift data indicate that
the MDA and Compur analyzers may
have difficulty producing repeatable
results to satisfy the adopted calibration
drift requirement.
Calibration error determinations were
performed on four of the five monitors.
The calibration error test procedure
involved performing five nonconsecutive
injections of the zero air and two HCI
calibration gases through the entire moni-
toring system. The differences between
the monitor responses and the known
concentrations of the calibration gases
were recorded. The calibration error
determinations were computed by sum-
ming the absolute value of the mean
difference and the 95% confidence inter-
val determined for the five injections.
Table 4 presents a comparison of the
monitor calibration error determinations
Table 2. Summary of Relative Accuracy Determinations
Relative Accuracy Results
Test Date
LSI
BSWK
TECO
MDA
Compur
conducted on the two dilution sampling
systems. The Bodenseewerk monitoring
system did not undergo a calibration
error test because only one of the
available HCI calibration gases (the 503
ppm cylinder gas) was within the 0-1000
ppm operating range of the analyzer.
Periodic injections of the single gas
yielded responses that were in excellent
agreement with the gas value.
The following observations can be
made from the results:
• Most of the calibration error results
exceeded the < 5% specification
adopted for this evaluation. All the
CEM responses to the 503 ppm gas,
except for one MDA response, were
within ± 8% of the cylinder gas value.
For comparison purposes, all the
monitors produced responses to the
503 ppm gas injections that would
meet the Appendix F, Procedure 1
acceptable accuracy requirement of ±
15% using the Cylinder Gas Audit
(CGA) procedure. The CGA procedure
requires three gas injections of each of
two audit gases and specifies that the
average of the three responses should
be used in determining accuracy.
Monitor imprecision is not accounted
for because the confidence interval is
not included in the Procedure 1
accuracy determination.
• Because the LSI and MDA analyzers
were calibrated to the 503 ppm HCI
calibration gas, the LSI and MDA
results for the high-level check (1556
ppm) are excessive as a result of the
nonlmearity associated with the
measurement techniques of these two
monitors.
• The magnitude of the confidence inter-
val terms for some of the analyzers
reflects a high degree of monitor
imprecision during these tests. These
analyzers had difficulty producing
repeatable responses to the same
calibration gas. This same problem
also affected the daily calibration drift
results.
The response time is defined as the
amount of time required for the meas-
3/19/87
3/20/87
4/27/87
4/29/87
4
24
17
8
8
14
6
3
9
31
19
4
27
6
4
7
13
8
23
11
-------
Table 3. Summary of Maximum Daily Calibration Or/ft*
Analyzer Low Range
Table 5.
High Range
Summary of Response Times to HC
Gas Inactions
LSI
Bodenseewerk
TECO
MDA
Compur
2.2% (0/14)
0.3% (0/10)
1.3% (0/15)
0.1% (0/15)
2.7% (1/10)
4.2% (1/14)
0.7% (0/10)
1.4% (0/15)
9.9% (9115)
4.7% (2/10)
System
Response
Times (mm)
"Numbers in parentheses are number of days on which the drift exceeded the 2.5%
specification/number of test days.
Table 4.
Summary of Calibration Error Tests
Calibration Error"
Analyser
Mid-Level
(503 ppm)
Compur Sampling System (40:1 dilution)
LSI ( + ) 3.5%
TECO ( + ) 7.9%
MDA ( + ) 13.9%
Compur ( + ) 7.0%
High-Level
(1556 ppm)
Bodenseewerk Monitoring Systems 3
Compur Monitoring System 5
Compur Sampling System
LSI 15
TECO 4
MDA 4
TECO Sampling System
LSI 16
TECO 5
MDA 6
37.7%
(-) 16.3%
( + ) 10.2%
TECO Sampling System (45:1 dilution)
TECO (•>
MDA (i
3.4%
8.6%
"The sign within parentheses denotes whether the mean of the monitor responses was
greater than (+) or less than (-) the known concentration of the injected calibration
gas.
+ /Vof tested on TECO sampling system because of the long response time of the
analyser and also the shortage of available calibration gas.
urement system to display 95% of a step
change in gas concentration on a data
recorder. The response times were deter-
mined by injecting zero air and HCI
calibration gases through the entire
monitoring system. The response times
of the LSI, TECO, and MDA analyzers
were determined first for samples de-
livered by the Compur sampling system,
and then for samples provided by the
TECO sampling system. The results are
presented in Table 5.
The results afforded by the various
performance tests conducted on both
dilution sampling systems indicate no
significant difference in the ability of
either sampling system to deliver a
representative sample to an HCI CEM.
The two primary physical differences
between these two similar sampling sys-
tems are: (1) the Compur dilution probe
sample critical orifice is constructed of
stainless steel and is electrically heated
to maintain a constant temperature,
whereas the TECO dilution probe critical
orifice is constructed of glass and is not
heated except by the flue gas; and (2) the
Compur system delivers diluted sample
flow to the CEMs at a rate of approx-
imately 33 L/min, whereas the TECO
system delivers sample at approximately
6 L/min.
The dilution ratios of both probes were
checked periodically by using a CO
analyzer and CO calibration gases. The
dilution ratio never changed without
indicating a significant change in either
the sample orifice vacuum or the dilution
air delivery pressure. The dilution ratios
were also verified immediately before the
initiation of each of the four relative
accuracy tests.
The differences noted during three of
the four relative accuracy test periods are
apparently due to HCI loss in th
sampling system. Both sampling system
transport the diluted flue gas sample <
relatively high flow rates through hea
traced tubing (300°F). So-called memor
effects" (HCI losses due to wall adsorf
tion of HCI) are less likely to occur unde
these conditions. The apparent differenc
between the duct and manifold sampl
results may be attributed to HCI loss I
the stainless-steel components on th
manifold, which included a few type 31
stainless-steel fittings and ball valve
The sample delivered by either samplin
system to the common glass manifol
passed through the stainless-steel har<
ware before it entered the HCI impingi
tram. Therefore, because of the potenti
for HCI loss, the use of stainless-ste
components was minimized whereve
possible.
-------
Conclusions
The following conclusions are based on
ine results that were obtained from the
project.
• Except for the CEA Model TGA-410,
the HCI monitoring equipment pro-
duced effluent measurements in good
agreement with concurrent wet-chem-
istry measurements for uncontrolled
HCI emissions from a municipal waste
combustion source.
• The TECO 200 and Compur sampling
systems (which employ dilution
probes), as well as the Bodenseewerk
monitoring system (which uses a non-
dilution, heat-traced sampling system)
can reliably provide representative
effluent samples to the analyzers. The
relative accuracy results indicate that
the effect of HCI line losses can be
minimized by injecting the HCI
calibration gases through the entire
sampling system when performing the
dynamic calibrations.
• For some of the analyzers, the calibra-
tion gas injections and the relative
accuracy tests do not always provide
the same indication of CEM perform-
ance.
• The relative accuracy test data do not
indicate biases for any of the monitor
measurement techniques because of
analytical interferences present in the
effluent of this municipal boiler.
-------
Scott A. Shanklin, J. Ron Jernigan, and Scott C. Steinsberger are with Entropy
Environmentalists, Inc., Research Triangle Park, NC 27709.
Roosevelt Rollins is the EPA Project Officer (see below).
The complete report, entitled "Evaluation of HCI Continuous Emission Monitors,"
(Order No. PB 89-161 863/AS; Cost: $21.95, subject to change) will be available
only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
United States Center for Environmental Research
Environmental Protection Information
Agency Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S3-89/019
000085833 PS
0 S ESflfi F80TICTICH AGEflCY
REGION 5 LIBBAif
230 S BEABBOHH STBIE1
CHICAGO II 60604
------- |