EPA-600/2-76-171
June 1976
Environmental Protection Technology Series
EVALUATION OF MONITORING SYSTEMS FOR
POWER PLANT AND SULFUR RECOVERY
PLANT EMISSIONS
Environmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into five series. These five broad
categories were established to facilitate further development and application of
environmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the ENVIRONMENTAL MONITORING series.
This series describes research conducted to develop new or improved methods
and instrumentation for the identification and quantification of environmental
pollutants at the lowest conceivably significant concentrations. It also includes
studies to determine the ambient concentrations of pollutants in the environment
and/or the variance of pollutants as a function of time or meteorological factors.
This document is available to the public through the National Technical Informa-
tion Service. Springfield, Virginia 22161.
-------
EVALUATION OF MONITORING SYSTEMS FOR
POWER PLANT AND
SULFUR RECOVERY PLANT EMISSIONS
By
Malbone W. Greene R. Neal Harvey
Robert L. Chapman Glen A. Heyman
Samuel C. .Creason William R. Pearson
BECKMAN INSTRUMENTS, INC.
ADVANCED TECHNOLOGY OPERATIONS
ANAHEIM, CA 90620
Contract No. 68-02-1743
Project 1363-2678-800
Project Officer
James B. Homolya
Emissions Measurement and Characterization Division
Environmental Sciences Research Laboratory
Research Triangle Park, NC 27711
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
RESEARCH TRIANGLE PARK, NC 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental
Sciences Research Laboratory, U.S. Environmental Pro-
tection Agency, and approved for publication. Approval
does not signify that the contents necessarily reflect
the views and policies of the U.S. Environmental Pro-
tection Agency, nor does mention of trade names or
commercial products constitute endorsement or
recommendation for use.
ii
-------
ABSTRACT
This project was conducted to evaluate a number of commercially available
extractive-type sampling and monitoring systems for monitoring sulfur
dioxide and hydrogen sulfide source emissions. Evaluation testing was
performed at a Fossil Fuel-Fired Power Plant and at a Claus Sulfur
Recovery Plant to obtain representative ranges of stack gas temperature,
water and particulate loading, and corrosiveness. Tests of calibration
error, relative accuracy, two- and twenty-four-hour zero and calibration
drifts, response time, and operational period were performed in accordance
with published EPA guidelines, The performance in each test was judged
against the published EPA performance criteria (EPA-650/2-74-013),
The detailed field test results, the complete Work Plan, sampling inter-
face drawings, the results of evaluations of Compliance Test Methods 6
and 11, miscellaneous observations and results, and a discussion of the
calculation and reporting instructions of the EPA guidelines are given
in the appendices. A detailed description of the systems evaluated,
summaries of the field test results, and relevant comments concerning
the results are given in the body of the report.
Because the sites chosen for evaluation testing provided wide ranges of
sample temperature, solids and water loading, and SC>2 concentration, the
results are considered to be relevant to most stack-gas monitoring prob-
lems. Results at the Claus Plant site were somewhat compromised by
equipment failures, but two general conclusions may be drawn. First,
systems available commercially are capable of meeting the requirements
of the guidelines, at least for short-term testing. Secondly, the cost
of correctly installing and maintaining such monitoring equipment is
probably considerably higher than is generally estimated by the
manufacturers.
iii
-------
CONTENTS
Page
Abstract ill
List of Figures vi
List of Tables vii
Sections
I CONCLUSIONS 1
II RECOMMENDATIONS 8
III INTRODUCTION 10
Purpose of Contract 10
Scope of Contract . 11
General Approach and Background Information ..... 12
The Fossil-Fuel-Fired Power Station Test Site .... 14
The Glaus Sulfur Recovery Plant Site 16
Brief Description of Contract Phases 23
IV DESCRIPTION OF SYSTEMS EVALUATED 25
Mechanical Dilution (Disc Diluter) and Electro-
chemical Monitor 25
Permeation Dilution Sampling and Monitoring System. . 28
Refrigeration Drying and Electrochemical
S02 Monitor 35
Reflux Probe, Permeation Drying, and Infrared
S02 Monitor 37
H2S Monitoring at Glaus Plant' (Before Incinerator). . 42
Data Acquisition System 45
V BRIEF OUTLINE OF WORK PLAN OBJECTIVES 50
Comment on Error Calculation 51
Results of Testing at Fossil-Fuel-Fired Power Plant . 52
Results of Testing at Glaus Sulfur Recovery Plant . . 64
Summary of Performance Criteria and Test 75
Comments on Sampling/Interface Systems, and
Field Performance . 78
VI APPENDICES 87
-------
FIGURES
No.
1 Power Plant Test Site 17
2 Power Plant Test Site, Elevation View 18
3 Claus Plant Site, Plan View 20
4 Claus Plant Site, Elevation View . . 21
5 Claus Plant Site 22
6 Disc Diluter Assembly 26
7 Disc Diluter, Gas Flow Diagram 27
8 Permeation Diluter System 31
9 Meloy Model FSA-190-2A Dependence Upon Carrier
Flow Rate 34
10 Refrigerated Condenser System 36
11 Reflux Probe and Permeation Dryer Flow Diagram .... 39
12 Reflux Probe and Permeation Dryer Sampling
Interface Enclosure 40
13 Reflux Probe, Permeation Dryer, and NDIR System. ... 41
14 Claus Plant H2S Sampling—Interface System 44
15 Power Plant Test Site Data Acquisition System 47
VI
-------
TABLES
No. Page
1 Summary of Performance Criteria and Test Results
for Power Plant Tests 2
2 Summary of Performance Criteria and Test Results
for Glaus Plant Tests 3
3 Calibration Error of Systems Evaluated at Power
Plant 53
4 Relative Accuracy of Systems Evaluated at Power
Plants 55
5 Summary of Two-Hour Zero and Calibration Test
Results for Power Plant Test Site 57
6 Summary of Twenty-Four-Hour Zero and Calibration
Test Results for Power Plant Test Site 58
7 Summary of Response Time Test Results Obtained at
Power Plant Site 60
8 Summary of Performance Criteria and Test Results
for Power Plant Tests - 62
9 Calibration Error Test Results Obtained Before
Transport to and Installation at the Glaus Plant
Test Site 65
10 Summary of Relative Accuracy Results Obtained with
Three Systems Monitoring the Claus Sulfur Plant
Emissions after Incineration 67
11 Summary of Two-Hour Zero and Calibration Test
Results Obtained at the Claus Sulfur Recovery
Plant 69
12 Summary of Twenty-Four-Hour Zero and Calibration
Test Results Obtained at the Claus Sulfur
Recovery Plant 71
13 Summary of Response Time Test Results Obtained at
Claus Plant 74
14 Summary of Performance Criteria and Test Results
for Claus Plant Tests 76
15 Test Results—Low-Concentration Gases 153
16 Test Results—High-Concentration Gases 153
17 Method 11(^5) Test Results
18 Results of Preliminary Method 6 Correlation Testing
at Power Plant 180
19 Example Results for Claus Plant Tests 192
vii
-------
SECTION I
CONCLUSIONS
The results obtained at the two test sites are summarized in Tables 1 and
2. The results obtained at the Glaus Plant, Table 2, should be qualified
by noting that only the NDIR System was operating correctly at the start
of the test. The other systems' malfunctions were not directly relatable
to the application, and were primarily being monitored for degradation by
the Claus Plant tail gas sample. They are considered to have performed
satisfactorily in this regard, even though many of the errors given in
Table 2 exceeded the test performance criteria.
The two following conclusions may be drawn from the experience gained in
planning the evaluation tests:
1. The commercially available extractive-type sampling and moni-
toring systems evaluated are capable of meeting most or all of
the requirements of the published guidelines (EPA-650/2-74-013).
2. Obscure problems related to the installation at a given site
can result in poor performance and/or failure of any or all
of the systems (and peripheral test equipment) evaluated. A
great deal of "custom engineering" may be required if good
performance and low maintenance costs are to be achieved.
These general conclusions are derived from more specific conclusions in
the following paragraphs. Discounting failures of breadboard quality
circuit boards, production models of all monitors evaluated will be
satisfactory, provided proper sample conditioning exists. The monitors
evaluated were as follows:
-------
Table 1. SUMMARY OF PERFORMANCE CRITERIA AND TEST RESULTS FOR POWER PLANT TESTS
Test
Performed
Cali-
bration
Error
Relative
Accuracy
Two-Hour
Zero Drift
Twenty-
Four-Hour
Zero Drift
Two- Hour
Calibra-
tion Drift
Twenty-
Four-Hour
Calibra-
tion Drift
Mean
Response
Time
Oper-
ational
Period
Test Performance
Criteria from
EPA-650/2-74-013
5%
20%
2%
4%
2%
5%
95% in
600 s
168 hours with-
out special
maintenance
Refrigerated
Condensor and
Dynasciences
Model SS-330
3.1%
1.5%
2.4%
5.3 to
11.5%
0.7%
0.9%
(avg)
2.7%
(avg)
2.6%
(avg)
150 s
Maintenance
only between
two 168-hour
tests
Beckman
Reflux Probe,
Permeation Dryer,
Model 865 NDIR
4.5%
0.6%
0.6%
5.3 to
11.4%
0.9%
1.4%
(avg)
1.5%
(avg)
4.5%
(avg)
45 s
Maintenance
only between
two 168-hour
tests
Permeation
Diluter and
Dyna sc ienc e s
Model SS-310
(Disc Diluter
used for this
test only)
4.5, 4.4, 2.7%
6.3 to
10.9%
0.5%
1.7%
(avg)
3.0%
(avg)
7.3%
(avg)
120 s
Maintenance
only between
two 168-hour
tests
Permeation
Diluter and
Flame Photometric
(Meloy)
4.0%
4.4%
2.1%
6.4 to
15.9%
0.2%
1,5%
(avg)
3,5%
(avg)
9.0%
(avg)
140 s
Maintenance only
between two 168-
hour tests
UV Absorption
N/A
8.3 to
12.7%
N/A
N/A
N/A
N/A
N/A
N/A
in situ
N/A
6.6%
(only 3
data
sets
N/A
N/A
N/A
N/A
N/A
N/A
-------
Table 2. SUMMARY OF PERFORMANCE CRITERIA AND TEST RESULTS FOR
GLAUS PLANT TESTS
Test
Performed
Calibration
Error
Relative
Accuracy
Two-Hour
Zero Drift
Twenty-Four-
Hour Zero
Drift
Two-Hour
Calibration
Drift
Twenty-Four-
Hour Calibra-
tion Drift
Appropriate
Mean Re-
sponse Time
Operational
Period
Test Performance
Criteria
EPA-650/2-74-012
5%
20%
2%
4%
2%
5%
95% in
600 s
168 hours without
special
maintenance
Beckman
Reflux Probe,
Permeation Dryer,
Model 865 NDIR
(S02 only)
12%
2%
0.6%
4 to 10%
0.8%
1%
(avg)
2%
(avg)
10%
(avg)
30 s
No special main-
tenance required
Permeation
Diluter,
Dynasciences
Model SS-310
(S02 only)
5%
9%
6.5%
10 to 19%
4.9%
(avg)
6%
(avg)
13 . 5%
(avg)
36%
(avg)
150 s
No special
maintenance
given
Permeation
Diluter,
Flame Photometric
(Meloy)
(S02 only)
3%
7%
8%
10 to 27%
2.8%
(avg)
5%
(avg)
30%
(avg)
40%
(avg)
200 s
No special main-
tenance given
Existing PGC
Sample Tap,
Houston Atlas
Diluter /Monitor
(H2S only)
5.4%
2.2%
38 to 44%
(against PGC)
6%
20%
(avg)
50%
98%
(avg)
960 s
No special
maintenance
given
-------
• Dynasciences Model SS-310 Ambient Level S0? Monitor used at
both test sites. (Sensing cell is easily damaged by pressure
variations, including altitude during shipment.)
• Dynasciences Model SS-330 Source Level SCL Monitor used at
Power Plant site only. (Rugged and quite reliable.)
• Meloy Flame Photometric Ambient Level Total Sulfur Monitor
used at both sites on SCL. (Production model expected to
perform satisfactorily.)
• Beckman Model 865 NDIR Source Level S02 Monitor (with all
revisions that took effect prior to November 1974) used at
both sites on SCL.
• Houston/Atlas, Inc., Ambient Sulfur Analyzer used on H_S at
Glaus Plant only. (Response is inherently slow, but satis-
factory on slowly varying sample.)
The major conclusions with regard to sampling interface system perform-
ance are as follows:
• Mechanical dilution appears to be entirely feasible. The Disc
Diluter requires further development, but the Houston Atlas
Diluter was satisfactory for the two-week evaluation.
• Permeation dilution is feasible, but the Diluter evaluated
(Meloy Dyfusatron) requires improved protection from pressure
surges.
• Permeation drying is feasible, but the construction materials
of the Beckman system evaluated (manufactured by Permapure
Products, Inc.) must be improved to achieve an operational
life of more than about 60 days on 250 ppm SCL.
• The conventional Beckman refrigerated condensor sampling sys-
tem operated With no maintenance and showed only moderate
corrosion after about 60 days at the Power Plant.
-------
• The removal of free sulfur in two condenser stages was not
entirely satisfactory, but better adjustment of operating
parameters is all that appeared to be required.
As examples of obscure sampling problems and failures of peripheral test
equipment that were not anticipated, the following specific problems are
cited:
• One pneumatically operated ball valve (of three evaluated)
tended to stick open, allowing some sample leakage during zero
and calibration cycles.
• Water condensed in the permeation diluter sampling system at
the Power Plant stack wall at a flow rate of 100 ml/min, but
this did not occur in two similar probes handling 10 liters/
min of sample. Heating of the mounting flange corrected the
difficulty.
• The Esterline-Angus Model D-2020 lost a power supply after
about one week of operation at the Power Plant. The reason
for the failure is unknown, but it is probable that high
ambient temperature and line voltage transients caused it.
• A 1.5-ampere "Slo-Blo" fuse in the teletype failed at about
the mid-way point of the Power Plant tests. It was replaced
with a one-ampere standard fuse, and the equipment operated
satisfactorily for the balance of the Power Plant tests. No
reason for the fuse failure was discovered, but a very large
line voltage transient may have been the cause.
• Random failures of digital components in the Beckman-designed
automatic zero and calibration console occurred at the Power
Plant. These failures can be explained by abnormally large
voltage transients, combined with a dc power voltage too near
the maximum for this type of component. The dc voltage was
reduced and no failures occurred at the Glaus Plant.
-------
An objective of this contract was to evaluate systems after an installa-
tion and troubleshooting effort which could reasonably be expected of a
typical industrial customer. Problems such as those enumerated above
caused significant delays in the commencement of the formal evaluation
testing. While all were actually minor failures, they resulted in a
great deal more start-up effort than would be anticipated by an indus-
trial purchaser of such "standard systems." Furthermore, problems such
as the sticking valve could go undetected indefinitely. The problem was
discovered when it was found that different monitor outputs were obtained
when standard gases were passed directly into the monitor, rather than
being introduced at the stack-mounted sampling, interface housing. A
conclusion that such "unexpected" problems should be considered normal
for a typical industrial plant seems inescapable.
Compliance Test Method 4 (HO) was employed at both sites with question-
able results. While the probe was heated as required by the procedure
given in the Federal Register, it was concluded that condensate trapped
in and/or dislodged from the short, unheated sampling probe to cooled
impinger interconnection could have large effects upon the volume of
water collected. A revision of the method to include suggested vari-
ations in technique would be beneficial to minimize this source of
variance in results.
Method 6 (S0~) gave acceptable results for up to 0.9% SCL in nitrogen.
The method could be revised to include procedural modifications similar
to those described in Appendix D to extend the range of usefulness.
Method 11 (H S) is not suitable for H-S concentrations above about 250
ppm in nitrogen. A significant revision of Method 11 would be required
to extend the range of applicability to 1% H~S.
-------
In general, the guideline procedures and performance criteria (EPA-650/2-
74-013) appear to be adequate. However, the speed-of-response require-
ment is probably not adequately covered by a single value (such as 95%
response in 600 seconds) since some combustion processes can be accom-
panied by more rapid significant changes. Specific requirements are best
determined for particular applications.
Finally, a potential interpretation ambiguity of the calculation and
reporting instructions of EPA-650/2-74-013 was discovered. The two
interpretations result in reported values (sum of absolute values, mean
deviation, and 95% confidence interval) which do not differ signicantly.
However, the mean values are smaller and the confidence intervals are
larger when calculated by the second interpretation rather than by that
used for all values reported in the body of this report. The two in-
terpretations are discussed in Appendix F.
-------
SECTION II
RECOMMENDATIONS
Compliance Test Method 6 (SO-) should be revised to include the minor
procedural changes necessary for extension of the range of applicability
to 1% SO ( which may occur in Claus Plant tail gases). Additional
testing to demonstrate that the results are valid is, of course, in
order.
Compliance Test Method 11 (H-S) was shown to be inadequate for use on
concentrations of H~S exceeding several hundred parts per million (ppm).
It is recommended that a method capable of determining H?S up to 1% be
developed.
The guidelines for monitoring source emissions published in EPA-650/2-
74-013, should be revised to eliminate a potential ambiguity with regard
to calculation and reporting instructions. See Appendix H for a dis-
cussion of two possible interpretations.
No recommendation for revision of the guideline tests, test procedures,
or requirements is justified by the results obtained on this contract.
Failure of the systems to meet the requirements in some of the tests
appeared to be the result of random component failures which were not
related to the application.
It is recommended that any future test of this type be continued for a
much longer time. There were signs of corrosion in parts of the sampling
systems tested at the Power Plant after about 60 to 70 days of continuous
operation. It was concluded that pumps with better corrosion resistance
-------
are desirable for the application, for example, but there was no clear
evidence that pump corrosion would have caused system failures within a
given operating time. Semi-automatic operation of the systems for sev-
eral more months, followed by another two-week intensive testing phase,
would have provided much more realistic information with regard to the
probable cost of maintenance of the systems.
-------
SECTION III
INTRODUCTION
PURPOSE OF CONTRACT
It was the purpose of this contract to evaluate a number of extractive
sampling and monitoring systems on both the stack gas of a fossil-fuel-
fired power plant employing wet limestone scrubbing for S09 removal and
on Glaus Sulfur Recovery Plant emissions (H?S, SO,,, and total sulfur)
before and after incineration. The major goal of this contract was to
provide the Environmental Protection Agency (EPA) with information upon
which minimum specifications for continuous extractive-type sampling
interface systems could be based. To achieve this goal various sampling
systems were to be coupled to appropriate analyzers utilizing various
principles of measurement. Thus the sampling and monitoring systems
evaluated at each site were to be representative of the commercially
available extractive-type sampling interface systems. In addition, the
initial checkout and debugging of the systems after field installation
were to be limited to essentially the level-of-effort that could reasonably
be expected for an ordinary industrial customer.
The basic approach chosen was to evaluate a number of sampling interface
systems concurrently, as the most cost effective means of obtaining
directly comparable test results. The current EPA Compliance Test Methods
were to be used for obtaining reference values against which the instru-
mental results could be judged. The results, therefore, were to be in-
dicative of those which could reasonably be expected by an industrial
user of commercially available sampling interface systems and of the EPA
Compliance Test Methods.
10
-------
The specific instrumental tests were to be performed using published
preliminary EPA procedures, techniques, and performance requirements
to determine the degree to which the systems evaluated were capable of
meeting the relevant published requirements for fossil-fuel-fired steam
generators. The evaluation test results would, therefore, be of direct
value in the formulation of minimum performance specifications for ex-
tractive-type sampling interface systems along the lines outlined in
the previous EPA publication.
SCOPE OF CONTRACT
The scope of the contract effort was limited to the work necessary to
accomplish the goals outlined above. Available government furnished
equipment (GFE) was utilized to the maximum possible extent in the
interests of cost effectiveness. Other necessary equipment was leased
whenever possible to limit the material expenditures to items that were
essentially expendable. Design and fabrication were limited to those
items of peripheral equipment which were not already available but were
required for interfacing the monitors to the sampling point. For example,
a requirement for automatic standardization made it convenient to design
sample handling interfaces with common solenoid valve actuation require-
ments. This approach also met a requirement for introducing standard
gases as close to the actual sampling point as possible. Sample probes
and interface housings were designed (where necessary) and fabricated
under the contract except for a GFE dilution probe system. Similarly,
the data acquisition system was designed to use available GFE and con-
tractor test equipment at no cost to the contract other than the data
system design, interconnection, testing, and maintenance.
Instrumental output data were acquired on both analog strip chart re-
corders (contractor furnished) and by a digital data logger with Teletype
interface including a punch for generating a punched paper tape record of
all digitized data (GFE). The contract scope permitted final data
11
-------
reduction from either (or both) analog or digital data, to permit
selection of the most cost effective approach. The two data systems
were independent, providing redundancy to reduce the probability of
lost time due to the failure of such peripheral test equipment.
Location of test sites and arranging for owner permission to use the
sites was a contractor responsibility. The contract scope encompassed
designing, fabricating, and/or modifying the test facilities as necessary
to interface with the test equipment, and restoration of the sites at the
conclusion of the test program.
Contractor effort in the hardware design and familiarization phase in-
cluded training of field personnel in the use of EPA Compliance Test
Methods k (1^0), 6 (S02>, and 11 (H S). Significant results obtained
in high-range testing of Methods 6 and 11 are discussed in Appendix D
of this report.
GENERAL APPROACH AND BACKGROUND INFORMATION
The type of sites chosen for evaluation testing had particular merit with
regard to the pertinence of the evaluation. It is anticipated that wet
limestone scrubbing for S02 removal will become common practice, although
at present (1975) there are only a few pilot plants in operation. The
sample may be characterized by the following:
• Relatively low S02 levels (150 to 250 ppm).
• Low temperature (60° to 70°C).
• Low H2SO^ mist content.
• Relatively low particulate content (except for a possible
carryover of limestone powder and/or slurry).
• Water content relatively low (saturated at about 50°C).
12
-------
The Glaus Sulfur Recovery Plant emissions before incineration may be
characterized by the following:
• Moderate temperature (150°C) .
• High water content (saturated at 72°C).
• High free sulfur content (condensation occurs below about
• High H_S and S02 content (0.5 and 0.25%, respectively),
making the sample quite reactive.
The emissions after incineration may be characterized by the following:
• High temperature (above 550°C) .
• High water content (saturated at 60°C).
• High SOo and/or sulfuric acid mist content, making the sample
very corrosive.
• High S02 content (0.4%).
The evaluation, therefore, included representative and/or nearly worst-
case industrial ranges of such parameters as sample temperature, pollutant
concentrations, water content, particulate content (including sulfuric
acid mist) , troublesome condensibles content (free sulfur) , and various
degrees of corrosiveness associated with combinations of pollutants at
various concentrations in the presence of varying amounts of moisture
at different temperatures.
The Field Evaluation Work Plan was designed to determine the performance
of the sampling and monitoring systems in terms of the requirements and
guidelines published in two EPA documents, the relevant portions of which
are reproduced in Appendix A. The documents are: EPA-650/2-74-013, Per-
formance Specifications for Stationary-Source Monitoring Systems for Gases
and Visible Emissions, and the Federal Register, Wednesday, September 11,
1974, Vol. 39, No. 177, Part II, Stationary Sources, Proposed Emission
13
-------
Monitoring and Performance Testing Requirements. In addition, char-
acterization studies of the emissions from Claus Sulfur Recovery Plants
were designed to provide data on the emission levels and to address the
feasibility of continuous monitoring of the process gases. The analyses
provided by the various continuous monitors were compared to those ob-
tained by standard manual EPA Compliance Test Methods to the extent that
the methods were applicable. The results of this evaluation, therefore,
are indicative of the applicability of the Compliance Test Methods (6
and 11) to the test sites, and of the ability of currently available
commercial extractive-type sampling and monitoring systems to perform
their intended function of providing a continuous record of pollutant
concentrations to assist the plant operating personnel in complying with
federal (or state and local) emission control requirements. While
strictly applicable only to the two types of plants selected for the
test (and to S0_ and H_0 monitors), these test sites provided such a
wide range of variation in the critical interface parameters that some
of the results are broadly applicable to many common industrial stack
gas sampling applications.
THE FOSSIL-FUEL FIRED-POWER STATION TEST SITE
The plant began operation in 1962. It is rated at 125 megawatts, burning
about 1,360 metric tons per day (1,500 tons per day) of western coal,
containing 0.48% sulfur. The flue gas is split and blown into twin
scrubbers. Before entering the stack, the gases pass first through a
flooded disc scrubber, then an absorber, and finally a reheat unit. The
original purpose of the scrubbers was to reduce particulates only, since
the low sulfur coal used here assured S02 emissions below the stated
limits. Suitable raw material being available nearby, one of the
scrubbers was modified in 1973 to use limestone slurry to remove S02
as a research and development project.
14
-------
The most suitable sampling point for the evaluation was located after
the two streams (limestone-scrubbed and water-scrubbed effluents) were
mixed in the stack. Consequently, the evaluation was performed on about
a 50/50 mixture of limestone-scrubbed and water-scrubbed sample. An
ultraviolet (UV) absorption type SO. analyzer had been installed to mon-
itor two extractive samples—one taken before limestone scrubbing and
the second after blending the scrubbed and unscrubbed gases. The stack
was breached for the evaluation test probes near the level used for the
second sampling tap of the UV analyzer. It appeared possible, therefore,
to correlate the evaluation test results with those of the UV analyzer.
In addition, an in-situ type monitor was installed prior to December,
1973, at a higher level on the stack. This instrument was being leased
by the test site and was being maintained on a contract basis. Further
cross correlation with the in-situ analyzer was, therefore, possible.
(Unfortunately, due to improper maintenance and/or for other reasons
discussed in Section V, relatively little meaningful correlation with
either site instrument was possible.)
The major sample parameters at the sampling point were as follows:
• SC>2 concentration of 100 to 250 ppm.
• Some fly ash but essentially no sulfuric acid mist.
• Saturated with water at 49°C (120°F), which is the operating
temperature of the absorber.
• Sample temperature about 70°C (160°F), which is the reheater
effluent temperature.
The ambient temperature at the test site was expected to vary between
about -35 and +30°C (-30 to +86°F) during the months of November through
January. One of the factors involved in the selection of the 13.4-meter
(44-ft) level for stack penetration was the availability of space for
erection of a platform to support an instrument shed adjacent to the
15
-------
stack. This location of the monitors reduced the length of heat-traced
sample line required to interconnect some of the sample probes and con-
ditioning systems to the monitors. The sampling ports, conditioning
systems, and instrument shed are shown in Figure 1. Figure 2 shows the
basic plant and stack structures, the position of the in situ monitor,
the position of the existing UV analyzer sampling probe (about 1.5 meters
[5 ft] above which the evaluation probes were installed), and the loca-
tion of the instrument shed that was erected for the evalustion test.
THE GLAUS SULFUR RECOVERY PLANT SITE
Brief Description of the Glaus Process
The Claus Sulfur recovery process was developed about 1890, and signif-
icantly modified in about 1937. Briefly, it is based on the reaction of
a portion of the H2S to SO- by precise combusion with atmospheric oxygen
to achieve near stoichemetric balance of H-0 and SO- for the reaction
The plant normally involves several stages, each followed by a coalescer
in which the sulfur vapor is condensed to liquid sulfur and drained to
storage. The final tail gas, containing both H-S and SCL, is mixed with
air and fuel and incinerated to oxidize the residual H-S to S0_ and SO-
^> £. J
before discharge to the atmosphere.
Description of the Claus Plant Test Site
The plant that was available for the evaluation test was eight to nine
years old. It originally exhausted to the atmosphere through a 50-meter-
high (160 feet) stack with a water spray system at its base. The old
stack was closed in recent years and the gas is now diverted to a much
taller new stack through a long duct. The duct also carries gas from a
newer Claus Plant to the new stack. A Beckman Process Gas Chromatograph
16
-------
MANUAL
SAMPLE
PROBE
REFLUX PROBE AND
PERMEATION DRYER,
PERMEATION
DILUTER
PROBE
INSTRUMENT
SHED
REFRIGERATED
CONDENSER
PROBE
DISC
DILUTER
MANUAL
SAMPLE
TRAIN
PROBE
SUPPORT
RAIL
SPECIAL
TEST
PLATFORM
GAS CYLINDERS
Figure 1. Power Plant Test Site
17
-------
4.7 M
SAMPLING
PLATFORM
1.2 M (4l)
W/3.04 M ( 1
(
154')
WIDE
0' ) EX!
4 PORT
•s i
REHEAT
ABSORBED TUBES~\
\
INSTRUMENT\ \
SHEDx
s \S
FLOODED NX.D
DISC >
SCRUBBER^
1 .8 X 1
(6' X 6
INSIDE
1
8.22 M
\
.8 M
r B
(29' )
A
K
N
L
\
,-—
c
\
_
T
H
~t
\
4
h
\
MX
FAN
xv^^^\\\\\x
3.6M
N
1
"
S
5.4M
18')
(12')
t
1 ID
1
ID •
1
1
t t
.
^
IOOOO
1 w~~\
5.7M
(19')
6M(20')
18 M (171 )
76.2 M (
50.2 M (
THIS SIDE
LIKE OTHER
SIDE
LOCATION
OF
IN SITU PROBE
'i
5 PROBES |
^~ ADDED 1
FOR TEST
21 .
.VV ANALYZER PROBE
4.26 M (14* )
if
\\\\\\\\\\\\\\\\\\\N
M (27* )
12.2 M (
1
3 M
40' )
^\\\\\\^
70' )
165' )
>X\XXN^XX
250' )
Figure 2. Power Plant Test Site, Elevation View
18
-------
(PGC) monitors 1^8 and SC>2 in the tail gas effluent from the sulfur
coalescer to the incinerator. In addition, an ultraviolet (UV) S02 ana-
lyzer is used to monitor S02 in the effluent from the incinerator, using
an extractive sampling system.
At the incinerator inlet the following normal operating conditions exist:
• The temperature is about 150°C (300°F).
• The pressure is essentially atmospheric.
• The sample gas composition is normally about as follows:
Water 37%
H2S 0.2 to 0.5%
S02 0.1 to 0.25%
COS 0.05%
CS2 0.03%
Total 82+84+85 1% maximum
NH3 Traces have been detected
N2 Balance
At. the incinerator outlet the following conditions exist:
• The temperature is 550°C (1000°F) or more
• The pressure is slightly below ambient atmosphere
• The sample composition is normally about as follows:
Water 15%
H2S Nil
802 P-4%
N2 71% (varies ±10% with feed gas)
62 11% (varies with burner operation)
C02 2.4%
NH3 Traces have been detected.
Figures 3 and 4 show the basic Glaus Plant structures, the location of
the evaluation test sampling points, and the location of the temporary
instrument shed erected for the evaluation test. Figure 5 is a photograph
showing the location of the sampling probes installed on the incinerator
outlet duct.
19
-------
SULFUR
PIT
HOUSTON ATLAS (H S)
FEED FROM GC TAP
PIPE
. RACK
I
I LOW
1 PLATFORM
l
| PGC
j_SHED_
I HIGH PLATFORM^
cri~~~ po
INCINERATOR
| '
I THERMOWELL
ELEVATION
SEE- FIGURE
___
L^H
l_ L J
yp
J UV ANALYZER "
SANPLE TAP
TO NEW STACK
DRAFT
CONTROL
LOCATION OF TWO NEW
TAPS FOR SO2 ANALYZERS
AND FOR MANUAL PROBE
FROM NEW SULFUR UNIT
Figure 3. Glaus Plant Site, Plan View
20
-------
1-0
LOCATION OF EXISTING PGC SAMPLE TAP,
SAMPLE TAPS
ADDED FOR
TEST OF SO
OUTLET INCINERATOR
FROM WHICH SAMPLE WAS TAKEN FOR THE
HOUSTON-ATLAS H S SYSTEM
I 1.06 M (3%' )
Figure 4. Glaus Plant Site, Elevation View
-------
DUCT
INCINERATOR,
OLD STACK
NJ
S3
REFLUX PROBE
AND PERMEATION
DRYER
UNHEATED
SAMPLE LINE
(FROM
INCINERATOR
INLET)
INSTRUMENT
SHED
HOUSTON-
ATLAS
INTERFACE
PERMEATION
DILUTION PROBE
MANUAL PROBE AND
SAMPLING TRAIN
Figure 5. Glaus Plant Site
-------
BRIEF DESCRIPTION OF CONTRACT PHASES
The contract work was performed in two phases. Phase I included the
following:
• Procurement of equipment as required.
• Locating .suitable test sites and obtaining owner permission
to perform field tests.
• Obtaining test site process and emission characterization data
necessary for designing the sample interface systems and for
preparation of the detailed Work Plan.
• Designing and fabricating sampling systems and peripheral
equipment.
• Design, assembly, and functional testing of data handling
equipment.
• Laboratory familiarization with all equipment and with EPA
Compliance Test Methods 4(H20), 6(S02), and 11(H2S).
• Preparation of Field Test Work Plan for approval by the EPA
Project Officer.
The second phase consisted of field testing at the two selected sites.
The Work Plans are included as Appendix B of this Final Report. The
following sampling and S02 monitoring systems were evaluated at the
Fossil-Fuel-Fired Power Plant.
• Refrigerated condensation type system with a filter on.the probe
inlet coupled to an electrochemical source level S0_ analyzer.
• Permeation dilution type system with a filter on the probe
inlet coupled to an ambient level flame photometric total
sulfur analyzer.
• Permeation distillation drying system with a "reflux" probe
coupled to a nondispersive infrared (NDIR) source level analyzer.
• Permeation dilution-type system coupled to an ambient level
electrochemical SO- monitor (connected in series with the flame
photometric analyzer).
23
-------
It was originally intended that the ambient level electrochemical ana-
lyzer be used to monitor the sample delivered by the Disc Diluter, but
when the Disc Diluter failed mechanically in preliminary testing, this
monitor was connected to the permeation dilution system to provide re-
dundant monitoring. (The available Disc Diluter test results are
summarized in Appendix E.)
The original Work Plan called for evaluation of three systems each on
the Glaus Sulfur Recovery Plant emissions before and after the tail gas
incinerator. Several factors such as test site constraints and program
costs made it necessary to alter the Work Plan to provide for evaluation
of the following systems at the Glaus Plant:
• Before incinerator, monitoring was accomplished with an
existing steam-cooled-free sulfur knockout assembly (used
for existing PGC sampling) followed by a chilled-water-cooled
sulfur and water knockout assembly, coupled to a lead acetate
paper tape H S analyzer.
• After incinerator, a permeation dilution system with a filter
on the probe inlet was coupled to an electrochemical ambient
level SCL analyzer and a flame photometric total sulfur ana-
lyzer (in series).
• After incinerator, a permeation distillation drying system with
a reflux probe was coupled to an NDIR source level SCL analyzer.
-------
SECTION IV
DESCRIPTION OF SYSTEMS EVALUATED
MECHANICAL DILUTION (DISC DILUTER) AND ELECTROCHEMICAL MONITOR
The Disc Diluter was available as GFE. The original Work Plan called for
evaluation at both sites and at all three sampling points, provided that
it was capable of providing sustained operation with reasonable mainte-
nance. The system is shown in some detail in Figures 6 and 7, which are
reproductions of Monsanto Research Corporation illustrations. The prin-
ciple of operation is adequately described in other publications, and is
not given here.*
The original Work Plan was revised after repeated mechanical failure of
the Disc Diluter made it impractical to attempt to continue the field
evaluation. During the laboratory familiarization tests and in prelim-
inary field testing, the Disc Diluter showed promising results when it
operated correctly, but the mechanical drive train failed frequently.
It became apparent that it would be impossible to make the required mod-
ifications of the mechanical drive within the scope of the contract
effort. The major problems identified were as follows:
• Excessive leakage of the disc occurred when compressive loading
was light enough to permit smooth operation by the drive train.
• Oscillatory drive of the disc occurred when compressive loading
was adequate to reduce leakage, indicating that the drive train
friction load system was oscillating in a mode similar to that
of a torsion rod pendulum.
*Monsanto Research Corporation, Operation and Design Manual,
"Construction and Field Testing of a Commercial Prototype Disc
Diluter," prepared for EPA under Contract No. 68-02-0716.
25
-------
'/VACUUM FILTER
DISK DILUTER
DRIVE MODEL
0.6M (24") EXTENSION
(2.4M (81) AS USED)
xROTATING DISK DILUTER
HEAD ASSEMBLY
DISK DILUTER
CONTROL MODULE
TACHOMETER
READOUT
VACUUM FLOW
INDICATOR
DILUTE GAS
FLOW INDICATOR
STANDARD GAS
FLOW INDICATOR'
Figure 6, Disc Diluter Assembly
-------
NJ
SO2 ANALYZER
CONTROL MODULE
SO2 ANALYZER
LEGEND
1 Disc diluter head assembly 10
2 Disc diluter extension tube 11
3 Vacuum line (polyethylene) 12
4 Diluent air line (polyethylene) 13
5 Calibration gas line (Teflon) 14
6 Diluted stack gas line (Teflon) 15
7 Electrical cable for drive motor 16
8 Vacuum on-off valve 17
9 Diluent air switching valve 18
•^^
Diluted stack gas outlet
Disc diluter calibration gas inlet
Diluent gas inlet
Calibration gas tank
Diluent ai r tank
Diluted stack gas inlet
Purge gas inlet
SO_ analyzer calibration gas inlet
SO analyzer outlet
Figure 7. Disc Diluter, Gas Flow Diagram
-------
• The friction load and leakage varied with changing probe tem-
perature and operating time.
• When the disc was wet with condensate, the leakage rate and
friction loading were compatible with the drive train design.
This suggested that deliberate water lubrication of the disc
might be a useful design approach. Further, it is probable
that the compressive loading was adjusted with the disc wet on
some occasions, and that subsequent oscillatory drive problems
were simply due to evaporative drying of the disc when the
probe attained operating temperature.
• There was some evidence of disc (Delrin) galling after oper-
ating for many hours, suggesting abrasion by fly ash was
occurring.
In summary, the Disc Diluter problems observed were due to details of
design and construction, and were not surprising in view of the fact that
the system tested was an initial breadboard design. The concept defi-
nitely appears to be worthy of further development. All test results are
documented in Appendix E of this report.
In the revised Work Plan, the Dynasciences Model SS-310 S0? Monitor, which
was to have been used with the Disc Diluter, was used with the Meloy per-
meation dilution system to provide redundant monitoring for the only
dilution sampling system still under evaluation at the power plant site.
The details of this system are given in the next paragraph.
PERMEATION DILUTION SAMPLING AND MONITORING SYSTEM
A Meloy Laboratories Inc. (Meloy) Model FSA-190-2A with a built-in
permeation dilution sampling interface (Dyfusatron*), available as GFE,
*A registered trademark of Meloy Laboratories, Inc.
28
-------
was used at both test sites. This instrument uses a flame photometric
total sulfur analyzer with very high sensitivity but with linear response
to only about 2 ppm total sulfur. It is, therefore, inherently an am-
bient-level sulfur monitor. The permeation diluter is used to reduce
source level concentrations by roughly three to five orders of magnitude
to match the diluted sample concentration to the basic 0-1 ppm total sul-
fur range of the flame photometric analyzer. The model evaluated was
fabricated by Meloy under contract to the EPA. While it utilized many
basic components of the commercially available Meloy flame photometer, the
analyzer included revised and/or repackaged circuit boards of breadboard
quality. It is believed that the many intermittent electronic failures
experienced after a brief initial satisfactory laboratory evaluation were
a direct result of the breadboard quality of these circuit boards.
After failure of the Disc Diluter sampling probe, the Dynasciences (now
Environmental Products Division of the Whittaker Corporation) Model SS-310
Electrochemical Ambient Level S0~ Monitor was reassigned the role of re-
dundant monitor for the Meloy permeation diluter.. The Model SS-310
consumes less than 0.1 percent of the S0~ present in the sample at normal
flow rates. It was possible, therefore, to install it between the Dyfusa-
tron output and the Meloy flame photometric analyzer burner. This was
done for all formal field evaluation testing of the Meloy System for two
primary reasons: (1) it permitted maximum use of the available monitoring
and data acquisition equipment, and (2) by the time the Model SS-310 be-
came available, the Meloy flame photometric analyzer was becoming very
unreliable because of several intermittent and one or more permanent elec-
tronic failures. The addition of the Model SS-310, therefore, provided a
backup means of monitoring the performance of the Meloy permeation diluter
independent (almost) of the operation of the Meloy flame photometric sul-
fur analyzer.
29
-------
The complete system Is illustrated in simplified schematic block diagram
form in Figure 8. More complete details of the system are shown in
drawing C/AE-14856 of Appendix C, particularly the details of the sampling
interface enclosure and of the Meloy Analyzer Gas Flow System. Referring
to Figure 8, Meloy pump No. 1 draws sample into the 5-micrometer filtered
probe, through the heated sampling interface enclosure and heated sample
line, through the source-sample side of the Dyfusatron, and discharges to
the atmosphere. Meloy pump No. 2 draws ambient air through a charcoal
cleanup trap (for removal of sulfur compounds and hydrocarbons) to pro-
vide carrier air for the diluted-sample side of the Dyfusatron. The
carrier air then passes through the backup monitor (Dynasciences Model
SS-310), through the hydrogen flame burner, and to exhaust. A fixed-
orifice air inlet near the inlet of each pump allows the pumps to operate
near volumetric capacity, reducing the negative pressure that can be
applied to the Dyfusatron membrane if the intakes are blocked. The
details of the flow controls and monitors are omitted from Figure 8 for
clarity of explanation of the basic functioning of the system.
The sampling interface enclosure contains valves, flowmeters, heaters,
air-pressure regulators, etc., as necessary to accomplish the following
functions:
• Selection of sample, zero, or span gases by remote command of
solenoid valves.
• Probe cleaning by blowback air during zero gas portion of cali-
bration cycle.
• Automatic sample shutoff in case of failure of either electri-
cal power or instrument air pressure (shutoff accomplished by
use of a pneumatically operated, spring-closed valve in the
sample line).
• Temperature control above sample dewpoint by heated-air-bath
technique.
30
-------
u>
I-1
HEATER ON FLANGE AS REQUIRED TO
PREVENT CONDENSATION IN THE PROBE
STACK WALL
-•• INSTRUMENT AIR IN 413.6 - 689.4 KPa
(60 - 100 PSI6)
-«-ZERO GAS IN, AIR, 68.9 KP3 (10 PSIG)
< HYDROGEN SUPPLY
MELOY MODEL FSA-190-2A
•SPAN GAS IN, PPM SO-
IN N2> 68.9 KPa
(10 PSIG)
SAMPLING
INTERFACE
ENCLOSURE
71 « 3°C
(160 ! 5°F)
5 MICRON FILTER
WITH BLOW-BACK AIR
CLEANING DURING
INSTRUMENT ZERO
GAS CYCLE
OUTPUT TO
DATA
ACQUISITION
SYSTEM
FLOW ABOUT 100 ML/MIN
ELECTRICALLY HEATED LINE (75C) I
ABOUT 12.1 M (40') LONG, 0.63 CM |
(i") O.D. TEFLON
ATMOSPHERIC EXHAUSTS
FOR CARRIER AIR CLEAN-UP
OUTPUT
TO DATA
ACQUISITION
SYSTEM
Figure 8. Permeation Diluter System
-------
• Provisions for delivering zero and span gases to the sample
line at essentially stack pressure.
As shown in Figure 8, it was also necessary to heat the flange that
mounted the interface enclosure to the stack (or duct) to prevent conden-
sation of water in the probe. This was not necessary for similar inter-
faces using about 10-liter-per-minute sample flow rates, but the low
sample rate tolerated by the Meloy System (about 100 milliliters per
minute) did not provide adequate flange heating due to the flow of warm
sample.
The sample line was electrically heated by a variac built into the Meloy
analyzer. This line was 0.63 cm (1/4 in.) OD Teflon surrounded by the
electrical heater wire, thermal insulation, and an external jacket. The
sample moisture was not allowed to condense at any point in the system
unless, perhaps, it condensed just ahead of the excess air inlet orifice
where the temperature was decreased to a few degrees above ambient. Di-
lution by ambient air at this point lowered the dewpoint to below ambient,
precluding condensation in pump No. 1.
The carrier air serves as the combustion air for the burner of the Meloy
analyzer. The burner is operated with a controlled flow rate of hydrogen
and a selectable (within limits) flow rate of carrier air, using the
forced draft (suction) of pump No. 2 to bring in the carrier air and to
remove the products of combustion. The operating characteristics are
such that the carrier air flow rate is about 35% higher with the flame
out than it is with the flame on. The flame photometric principle is
such that the output signal is semi-proportional to the number of sulfur
atoms entering the burner per unit time over a fairly broad range of car-
rier flow rates (for fixed H2 flow rate). The permeation dilution tech-
nique provides a constant rate of input of sulfur atom containing mole-
cules (of a given molecular species such as S02) for a fixed partial
32
-------
pressure of sulfur compound on the source-sample side almost independent
of carrier flow rate. Consequently, the S0~ (for example) concentration
(actually, partial pressure) in the carrier gas is proportional to the
carrier flow rate, but the Meloy output is only slightly dependent upon
carrier flow rate when set at the optimum flow rate. The output versus
carrier flow rate curve of Figure 9 illustrates this characteristic of
the Meloy analyzer output.
The Dynasciences Model SS-310 Electrochemical SC^ Monitor responds to
sample partial pressure of S02 (PsOo)- ^-n view of the above discussion,
the Model SS-310 output was inversely proportional to the Meloy carrier
gas flow rate when sampling in the manner indicated in Figure 8. The use
of the Model SS-310 as a backup monitor for the Meloy Dyfusatron output
was complicated by this effect whenever the flame of the Meloy analyzer
went out—causing a large increase of carrier flow rate. Otherwise, the
carrier flow rate was sufficiently stable to make the Model SS-310 and
Dyfusatron a stable system. Furthermore, the known change in carrier
flow rate for flame-on to flame-out conditions and the simple dependence
upon flow made it possible to make rough corrections of the Model SS-310
output for changes in carrier flow rate.
The output of the Dyfusatron is a sample with sulfur-containing molecule
partial pressure proportional to that of the wet sample before dilution.
This is a consequence of maintaining high temperature throughout the sam-
pling system, which makes the partial pressure of water (PHOO) °f tne
sample in the Dyfusatron equal to that at the probe inlet. Since there
is negligible drop in the sample total pressure from probe to Dyfusatron
and the sum of all partial pressures is constant (ambient atmosphere ±25
cm of water), the presence of water vapor pressure reduces all other par-
tial pressures proportionately. This dilution effect may be compensated
for by dividing the wet basis output of the analyzers by the quantity
(l-FI^O), where FI^O is the mole fraction of water in the sample. This
33
-------
cvj
O
(A
a.
Q.
r-
m
1.0
0.9
0.7
P 0.6
LU
g 0.5
to
UJ
£ 0.4
a 0.3
0.2
0.1
TEST PERFORMED 12 NOV. 1974, ABOUT TWO
HOURS BEFORE START OF THREE GAS CALIBRA-
TION TEST. FLOW OF 7 WAS SELECTED FOR
TEST. RELATIVE OUTPUT HAD DROPPED TO
0.87 WHEN CALIBRATION TEST WAS PERFORMED,
BUT NO TEST OF FLOW EFFECT WAS MADE.
I
Figure 9.
3 4 56 7 8 9
CARRIER FLOW RATE, ARBITRARY UNITS
Meloy Model FSA-190-2A Dependence upon
Carrier Flow Rate
10
34
-------
discussion assumes that the analyzers are calibrated on dry gases, which
was the case for all results reported. This correction, then, is neces-
sary before comparison to the results obtained with the standard EPA
Compliance Test Methods, which are reported on a dry basis. It was
necessary, therefore, to employ Compliance Test Method 4 to determine the
mean water vapor content of the sample for use in correcting the outputs
of the Meloy and the Dynasciences Model SS-310 to a dry sample basis for
comparison to the correlation test data from the Compliance Test Methods.
REFRIGERATION DRYING AND ELECTROCHEMICAL S02 MONITOR
This system was tested at the power plant site only. It represented the
most conventional system in the group evaluated. The system is shown in
simplified schematic block diagram form in Figure 10 and in detail in
drawing B/AE-14866 of Appendix C.
The probe filter was 20 micrometers to permit 10-liter-per-minute sam-
pling with a reasonable pressure drop. The sampling interface enclosure
was basically identical to that used with the Meloy System described in
detail in the last paragraph, except for the inclusion of a second filter
element to remove particles down to about 1-micrometer size. The heat-
traced sample line was also of the same type but with a 0.95 cm (3/8 in.)
OD Teflon tube to accommodate a higher sample flow rate. The line tem-
perature was controlled with a second variac located in the Meloy System.
Referring to Figure 10, the sample is drawn through the inlet filter,
probe, sample interface enclosure, and heated line by the sample pump.
The sample pump discharges through a short unheated line into the refrig-
erated condenser. Condensate is removed by a polyvinyl chloride (PVC)
trap, equipped with a float valve for automatic dumping of condensate to
the drain. The dried sample (dewpoint of about 1°C) then passes to a
68.9 kPa (10 psig) relief valve and to the analyzer flow control valve
connected in parallel. Most of the sample normally vents to atmosphere
35
-------
STACK WALL
PROBE
20 MICRON FILTER
WITH BLOW-BACK
CLEANING DURING
INSTRUMENT ZERO
GAS CYCLE
HEATER ON FLANGE AS REQUIRED TO
'PREVENT CONDENSATION IN THE PROBE
^-INSTRUMENT AIR IN, 413.6 - 689.4 KPa (60-100 PSIG)
•ZERO GAS IN, AIR, 68.9 KPa (10 PSIG)
•SPAN GAS IN, PPM SO IN N
68.9 KPa (10 PSIG)
2*
FLOW ABOUT 100 ML/MIN
ELECTRICALLY
ABOUT 12.1 M
0.63 CM (i")
HEATED LINE
(40') LONG,
O.D. TEFLON
SAMPLING
INTERFACE
ENCLOSURE
71
(160
3°C
°'
RELIEF VALVE
(10 PSIG)
REFRIGERATED
CONDENSER
rSPAN GAS, IN, PPM SO-, (10 PSIG)
•ZERO GAS. IN AIR, (10 PSIG)
•SAMPLE VENT
VAL!E
[V]WATER TRAP (PVC)
Tf
DYNASCIENCES
MODEL SS330
SO MONITOR
0- 500 PPM
0-1500 PPM
0-5000 PPM
T
OUTPUT
TO DATA
ACQUISITION
SYSTEM
DRAIN
Figure 10. Refrigerated Condenser System
-------
through the relief valve. The balance of the sample passes through a
5-way manual valve to the Dynasciences Model SS-330. The 5-way valve
permits manual introduction of zero and span gases downstream of the
refrigerated condensor if and when desired. This feature is of consid-
erable advantage in terms of troubleshooting and in determining the
response delay due to the components in the sampling interface housing,
heated lines, refrigerated condensor, and water trap.
The output of the Dynasciences monitor is on a dry basis, the sample
having nearly the same residual water content as that left by the ice
bath used on the sample train of the Compliance Test Methods. With no
correction for the water content difference, the output should be within
less than 1/2% of the Compliance Test Method determination.
REFLUX PROBE, PERMEATION DRYING, AND INFRARED S02 MONITOR
This system was tested at both the power plant and at the Claus Sulfur
Recovery Plant (after incineration). Both.the reflux probe and the perme-
ation distillation dryer are relatively new developments. Briefly, the
reflux probe withdraws, filters, and returns (refluxes) most of the sam-
ple to the stack through the inner of two essentially concentric tubes
of different length. The relatively high velocity of the reflux gas dis-
charged inside of the larger diameter outer probe produces an axial dis-
charge of high velocity, causing a flow disturbance of sample passing by
the end of the probe. The net effect is to cause the high velocity stack
gas stream lines to deflect around the probe tip, which carries particu-
lates away from the probe inlet. The gas drawn into the outer probe is
from the relatively low velocity vortex created by the reflux flow, and
is relatively free of particulates. A small percentage of the sample
passes on the analyzer, the majority being discharged through the reflux
(inner) probe.
37
-------
The system is illutrated in Figure 11, which is greatly simplified to
facilitate discussion. More complete details are given in drawing B/AE-
115010 of Appendix C.
Figure 12 is a photograph of the reflux probe interface housing with the
door open, showing some of the components contained within the heated
zone of the housing.
Figure 11 also illustrates the operation of the permeation dryer. That
part of the sample which is not refluxed passes through a backup filter,
a differential pressure flow controller, through the permeation dryer (a
bundle of small Permapure* plastic tubes), and out through the sample line
to the monitor. The space surrounding the dryer tubes is purged by rela-
tively dry air (counterflow). Because the tubes are very permeable to
moisture, the sample leaving the dryer can be almost as dry as the purge
gas entering it. The residual sample water also depends upon the ratio
of purge gas to sample gas flow rates—normally 2 to 1 or greater. The
use of a differential pressure-actuated flow controller, as indicated
in Figure 11, protects the dryer tubes from excessive internal-to-external
pressure differences.
The complete system evaluated is shown in simplified block diagram form
in Figure 13. More complete details of the entire system are in drawing
B/AE-115010 of Appendix C. Referring to Figure 12, the permeation drier
sample flows through an unheated 0.63 cm (1/4 in.) OD Teflon tube to the
Instrument Shed. Near the instrument the sample line passes through a
solenoid valve that is closed by the action of a pressure switch in the
instrument air line. This feature protects the analyzer and permeation
dryer from moisture which would accumulate if the purge air failed in the
^Manufactured by Permapure Products, Inc., P.O. Box 70, Oceanport, N.J.,
07757
38
-------
Kl^l
STACK OR Vs
,^L
^st^==
i
REFLUX SAMPLE
PROBE
— HEATER ON FLANGE AS SPAN
RFOiiTRFn rn PRFVFWT ^ , IN N
CONDENSATION IN THE (10 f
PROBE
SAMPLING INTERFACE ENCLOSURE
MAINTAINED AT 71 t 3°C 1 r^
(160 ± 5°F) X — I/I
FLOW CONTROLLER, T J -
DIFFERENTIAL _^
TO ZERO GAS PRESSURE ACTUATED ~~^
VALVE AND CONTROL 1
FILTER, PUMP NO. 1 £y^
f TOTAL f~~\ \ FIITFR, f \
4-I1 SAMPLE FLOW ' ^ 1 10 MICRON \
— ill, \ / WET PURGE" PERMEATION
I 1 ^' I /" \ | AIR OUT 1 DISTILLATION
REFLUX FLOW \*J DRYER TUBES
PUMP NO. 2 1 REFL-UX SAMPLE
* T FLOW
l57
^ RELIEF VALVE,
ADJUSTABLE TO
68.9 KPg (10 PSIG)
1 r
PURGE AIR
VENT
GAS IN, PPM SO-
,68.9 KPa
3SIG)
DRY PURGE AIR IN
413.6 - 689.4 KPa
(60 TO 100 PSIG)
(ALSO USED FOR
ZERO GAS)
•-PRESSURE
REGULATOR AND
GAGE 68 9 KPa
( 10 PSIG)
DRY SAMPLE
SO2 MONITOR
Figure 11. Reflux Probe and Permeation Dryer Flow Diagram
-------
PERMAPURE DRYER
INSIDE COVER
SAMPLE
HANDLING
SOLENOID
VALVES
AIR ACTUATED
SAMPLE VALVE
REFLUX
PROBE
SAMPLE
SPAN
ZERO
SAMPLE
PUMP
INSIDE
COVER
DIFFERENTIAL
FLOW
CONTROLLER
MANUAL
.VALVE
FLOW METERS
SAMPLE
FILTER
ELECTRONIC
CONTROLS
INSIDE BOX
Figure 12. Reflux Probe and Permeation Dryer
Sampling Interface Enclosure
40
-------
STACK OR
DUCT WALL
•HEATER ON FLANGE AS
REQUIRED TO PREVENT
CONDENSATION IN THE
PROBE
PRESSURE
SWITCH
INSTRUMENT AIR IN, 413.6 - 689.4 KPa
-«- (60 TO 100 PSIG) (USED FOR ZERO GAS
AND DRIER PURGE)
SPAN GAS IN, PPM S
68.9 KPa ( 10 PSIG)
IN N
PERMEATION
DRIED SAMPLE
OUT
SOLENOID
VALVE,
ACTUATED BY|
PRESSURE
SWITCH
EXCESS
SAMPLE
VENT
SAMPLE VENT
SAMPLE LINE 0.68 CM (i")
O.D. TEFLON, ABOUT 12.1 M
(40') LONG
SPAN
GAS IN
SAMPLING
INTERFACE
ENCLOSURE
71 • 3°C
(160 ' 5°F)
OUTPUT. TO
DATA ACQUISITION
SYSTEM
Figure 13. Reflux Probe, Permeation Dryer, and NDIR System
-------
event air pressure is lost. A major portion of the sample is bypassed
to vent, while the balance passes through a manual valve manifold to
a Beckman Model 865 Nondispersive Infrared (NDIR) S0? Monitor with lin-
earized output. The manual valves permit introduction of zero, span, or
sample gas to the analyzer, independent of the sample interface enclosure,
permeation dryer, and sample lines. This feature is very convenient when
troubleshooting the sampling system monitor.
The output of the Beckman System is on a dry sample basis. The error in
reported data (about 0.5% water) due to neglecting the difference in water
content of Method 6 samples and the sample delivered by this system (about
0.1% water or less) is less than 1/2 percent.
H S MONITORING AT GLAUS PLANT (BEFORE INCINERATOR)
In the revised work plan, this was the only system to be evaluated for H S
monitoring ahead of the Glaus Plant tail gas incinerator. Advantage was
taken of an existing sampling system being used with a Beckman process gas
chromatograph for monitoring the H2S and 862 ahead of the incinerator.
Referring to Figure 14, the sample for the evaluation test was taken
after the existing steam temperature controlled sulfur knockout pot
(coalescer). The sample line was maintained at 121°C (250°F) up to a
chilled-water-cooled sulfur and water knockout pot, which was designed
on this contract. The refrigerated condenser used at the power plant for
sample drying, was used in a circulating loop for chilling the water, as
shown in Figure 14. Details of the chilled-water-cooled sulfur and water
knowkout pot design are shown in drawing C/AE-14874, Appendix C. It was
manually cleaned periodically by heating with steam, which melted the
accumulated sulfur and carried it out the drain.
After leaving the water-cooled sulfur knockout pot, the sample passed
through unheated Teflon lines, through a filter, the sample pump, and on
to a solenoid valve manifold used for zero, span, or sample selection
42
-------
(see Figure 14). As in the other systems, the sample control valve was
a pneumatically operated, spring-closed valve to provide sample shutoff
if either power or instrument air failed. (More complete details of
the system described thus far are shown in drawing B/AE-14774 of Appendix
C.) The sample then passed through another filter to the Houston Atlas,
Inc., (HA) Dilution and Monitoring System.
The Houston Atlas, Inc., lead acetate paper tape sulfur monitor is in-
herently capable of analyzing samples containing up to about 40 ppm of
H«S. For applications to higher concentrations, HA provides a dilution
system very similar in concept to that of the EPA-developed Disc Diluter.
Briefly, a GC-type slide valve is pneumatically operated by a solenoid
valve driven by a clock motor-cam-microswitch timer to inject fixed
volumes of sample at a fixed repetition rate into a carrier gas (instru-
ment air with absorber to remove sulfur components). The carrier gas
(diluted sample) passes through a mixing chamber to a standard Houston
Atlas lead acetate paper-tape sulfur analyzer. The dilution factor is
controlled by sizing the volume of the slide valve (sample injector), and
by changing the number of lobes on the cam and/or the clock-motor speed
to control the rate of injection of fixed volumes of sample into the
carrier gas.
The HA lead acetate paper-tape analyzer is based upon the reaction of H2S
(and other reduced sulfur compounds) with lead acetate (white) to form
lead sulfide (black). The analyzer actually measures the rate of change
of tape optical reflectance with time, which is proportional to the rate
of formation of lead sulfide. When sampling directly, the output is pro-
portional to both sulfide concentration and to sample flow rate. However,
when coupled to the HA diluter it was found that carrier and sample flow
rate changes had only transient effects on the analyzer output. This is
consistent with the observation that the lead acetate tape removes all
from the sample, which means that, within limits, the steady-state
43
-------
REFRIGERATION
UNIT puMp
WATER AT
(40°F)
STEAM IN
FOR CLEANING
STEAM FOR
TEMPERATURE
CONTROL
TO PGC
AIR IN, 413.6 TO 689.4 KPa
(60 TO 100 PSIG)
INSTRUMENT AIR
IN FOR DILUTION
EXCESS
FT, TTTR
FILTER
PUMP
SULFUR AND
WATER
KNOCK-OUT
POT
ZERO GAS IN
AIR 68.9 KPa
(10 PSIG)
SULFUR
KNOCKOUT
(EXISTING
)
SPAN GAS JN>
PPM H S IN N
: —th—
-&-
SOLENOID
VALVES
( 10 PSIG)
DRAIN
OUTPUT TO DATA
ACQUISITION
SYSTEM
DILUTED
SAMPLE
HOUSTON ATLAS
LEAD ACETATE
TAPE ANALYZER
LINE TO INCINERATOR
SAMPLE VENT
Figure 14. Glaus Plant
Sampling — Interface System
-------
output should depend only on raw sample H^S concentration and on the
diluter slide valve volume and repetition rate. Consequently, it is only
important that the sample flow rate be high enough to purge the slide
valve injection volume between cycles, and that the carrier gas flow be
stable and within the rate acceptable by the tape sampler.
The output of the HA System as tested was essentially on a dry basis—
there being little more residual moisture in the sample leaving the
chilled-water knockout pot (4°C) than there is from the ice bath used in
the Compliance Test Method sampling train. Compliance Test Method 11
was not applicable to the concentration of H~S present in the Glaus
Plant tail gas. Consequently, the evaluation test results could be
compared only to the PGC analysis, and the vendor analysis of the span
gas was the sole basis for calibration of the HA analyzer. The span gas
was dry and the sample contained about 0.8% water (saturated at 4°C),
meaning that the error due to neglecting the sample residual water was
less than 1%.
DATA ACQUISITION SYSTEM
All monitors under evaluation provided analog outputs of 0-100 millivolts,
or were easily modified for 0-100 millivolt output. Texas Instruments,
Inc., strip chart recorders were available (contractor-furnished) for
continuous monitoring and recording of four data channels. An Easter-
line-Angus (E-A) Model D-2020 Data Logger with 20 channels was available
(GFE) for providing analog-to-digital conversion (ADC). A teletype (TTY)
for both printout and punched paper-tape data storage was also available
(GFE) for interconnection to the E-A Data Logger. Consequently, the work
plan called for data acquisition by analog strip chart recorders, TTY
printout, and on punched paper tape (by TTY punch). The system was re-
dundant in the sense that the required data reduction could be accomp-
lished from either analog or digital data (or both).
45
-------
The data acquisition system and the systems evaluated at the power plant
are shown in block diagram form in Figure 15. It was possible to cali-
brate all systems for 0-500 ppm (referred to the sample before dilution)
except the Meloy—which has no provision for adjustment of either zero or
span by the operator. Each of the monitor outputs (except the Dyna-
sciences Model SS-310) was connected to a strip chart recorder input, to
an input channel of the E-A Data Logger (20 channels available), and to
an input of an automatic zero-cal module. The Model SS-310 was not con-
nected to the auto zero-cal unit because it was originally intended that
it be used with the Disc Diluter probe (dashed line in Figure 15) which
was not to be automatically zero and span adjusted. When the Disc
Diluter failed and the Model SS-310 was moved to the Meloy Sampling
System (solid lines of Figure 15) it was not connected for auto zero-cal
because of field test time constraints. The three active auto zero-cal
module outputs were connected to three more channels of the E-A Data
Logger to provide digital output of both raw and auto zero-cal stand-
ized monitor outputs. The E-A Data Logger provided a digital display of
any channel selected, TTY printout of any or all channels when desired,
plus a TTY-generated punched paper tape for all data output by the E-A
unit. This data acquisition scheme had the following advantages:
• Direct analog recording of all raw monitor outputs, inde-
pendent of the operation of the digital acquisition system
and of the auto zero-cal modules.
• Acquisition of a punched-tape record permitting introduction
of data into a digital computer for more sophisticated data
reduction, if desired, and/or generation of duplicate TTY
printout of digital data to facilitate manual data reduction.
• Extra E-A data logging channels were available for monitoring
temperatures, system mode identification signals, etc.
46
-------
REFRIGERATED
CONDENSER
SAMPLING
SYSTEM
REFLUX PROBE
AND PERMEATION
DISTILLATION
DRIED SAMPLE
SYSTEM
PERMEATION
DILUTION
SAMPLING
SYSTEM.
(MELOY
DYFUSATRON)
DISC DILUTER
(FAILED IN
PRELIMINARY
TESTING)
DYNASCIENCES
MODEL SS-330
SOURCE LEVEL
S02 MONITOR
(0-500 PPM)
BECKMAN
MODEL 865
NDIR
SO MONITOR
(0-500 PPM)
LINEARIZED
MELOY FLAME
PHOTOMETRIC
TOTAL SULFUR
MONITOR, MODEL
FSA-190-2A
(NOT POSSIBLE
TO CALIBRATE,
BUT LINEARIZED)
DYNASCIENCES
MODEL SS-310
AMBIENT LEVEL
S02 MONITOR
(CALIBRATED FOR
0-500 PPM
AT SOURCE)
(USED WITH
DISC DILUTER
IN PRELIMINARY
TESTS )
TEXAS
INSTRUMENTS, INC.
DUAL PEN ANALOG
RECORDER NO. 1
STACK
TEMPERATURE
BECKMAN
DUAL PEN
RECORDER .
(ALSO USED
FOR TROUBLE
SHOOTING
AND BACK-UP)
(NO CONNECTION)
TEXAS
INSTRUMENTS,
INC.
DUAL PEN
ANALOG
RECORDER
NO. 2
AUTO
ZERO-CAL
SOLENOID
VALVE DRIVE
OUTPUT-
(NO CONNECTION)
AUTOMATIC
ZERO AND
SPAN CONSOLE
•AUTO ZERO-CAL
MODULES
•TEMPERATURES
•CHANNEL
IDENTIFICATION
•TRIGGER
CIRCUITRY
•TIMERS
•SOLENOID VALVE
DRIVERS
ESTERLINE-ANGUS
MODEL D2020
(20 CHANNELS)
RAW DATA
FROM MONITORS
AUTOMATIC
ZERO AND
SPAN ADJUSTED
MONITOR DATA
AMBIENT TEMPERATURE
TRIGGER FOR D-2020
TEMPERATURE PROBES
Figure 15. Power Plant Test Site Data Acquisition System
-------
The auto zero-cal console contained four auto zero-cal modules (one
spare) plus all peripheral circuitry. The peripheral equipment in-
cluded the following:
• Signal conditioning for two temperature probes that monitored
stack gas and ambient temperatures.
• Mode identification output to the E-A Data Logger, i.e., a
voltage indicating zero, span, or sample status of solenoid
valves (on one channel) and a voltage indicating manual, two-
hour or twenty-four-hour cycle status .of a manual cali-
bration mode selector switch on another channel.
• Two clock motors for control of auto zero-cal standardization
in either two-hour or twenty-four-hour mode, plus triggering
of the E-A Data Logger for data acquisition at desired inter-
vals when in either automatic mode.
» Solid-state relays for solenoid valve operation upon command
from either the auto zero-cal module, or from a manual zero,
span, sample selector switch when in the manual operating mode.
• Line to twenty-four-volt transformers for solenoid valve
operation.
• DC power supplies for all circuitry contained in the auto-
matic zero-cal console.
The original Work Plan utilized all of the data acquisition capability
described above. However, failures of both timer motors and of one type
of digital component of the auto zero-cal modules during preliminary-
tests at the power plant site made it impossible to operate the system
in either the two-hour or twenty-four-hour automatic modes during the
formal evaluation test at the power plant site. Accordingly, two-hour
and twenty-four-hour standardizations were performed manually using the
manual mode of the automatic zero-calibration console at both test
sites. Automatic digital data acquisition was accomplished when de-
sired with the built-in, timed, trigger capability of the E-A Data Logger.
48
-------
The systems evaluated at the Glaus Plant site are essentially illustrated
in Figure 15. The only change is the substitution of the Houston Atlas,
Inc., diluter and lead acetate sampler for the refrigerated condenser—
Dynasciences Model SS-330 system shown in Figure 15.
-------
SECTION V
FIELD TEST RESULTS
BRIEF OUTLINE OF WORK PLAN OBJECTIVES
The Field Evaluation Work Plan was designed to determine the performance
of the sampling and monitoring systems in terms of the requirements and
guidelines published in two EPA documents, the relevant portions of
which are reproduced in Appendix A for convenience of reference. The
documents are EPA-650/2-74-013, Performance Specifications for Station-
ary Source Monitoring Systems for Gases and Visible Emissions, and the
Federal Register, Wednesday, September 11, 1974, Vol. 39, No. 177, Part
II, Stationary Sources, Proposed Emission Monitoring and Performance
Testing Requirements.' The analyses provided by the various continuous
monitors were compared to those obtained by standard, manual EPA Com-
pliance Test Methods to the extent that those methods were applicable.
The tests required by EPA-650/2-74-013 are more comprehensive than those
specifically required by the subject contract. In the interests of ob-
taining the maximum useful information from the funds required for field
testing, the Field Evaluation Work Plan included all eight of the follow-
ing desirable tests from EPA-650/2-74-013.
1. Calibration Error; determined by repetitive cycling between
three known standard gases (except that only two gases were
used for one system, as allowed by the Federal Register,
Sept. 11, 1974).
2. Relative Accuracy; obtained by comparison to Compliance Test
Method results (where applicable).
3. Two-hour zero drift.
4. Twenty-four-hour zero drift.
5. Two-hour calibration (or "span") drift.
6. Twenty-four-hour calibration drift.
50
-------
7. Response time.
8. Operational period.
In the original Work Plan the Disc Diluter with Dynasciences Model SS-310
Monitor was to participate in tests 1, 2, and 7 only, because this system
was not to be included in the automatic calibration scheme. Test 1 was
performed at the Beckman facility in Anaheim, California, before moving
to the test sites, in accordance with the original plan. After failure
of the Disc Diluter and of the automatic mode timers in preliminary tests
at the power plant, the Test Plan was revised to include tests 2 through
7 for all monitors, including the Model SS-310 which was then connected
to the Meloy Permeation Dilution Sampling System.
All reasonable effort (within contract scope and funding) was made to ob-
tain optimum performance from all systems in the field installation and
preliminary testing stage, and to maintain all systems throughout the
formal testing. In accordance with the procedures of EPA-650/2-74-013,
it was necessary to restrict the frequency of adjustment of the systems
while performing drift and accuracy tests. Such details are given in the
abridged Work Plan of Appendix B.
COMMENT ON ERROR CALCULATION
In reviewing all data prior to issuance of this report, a possible ambi-
guity in interpretation of the guidelines given on pages 56-60 of EPA
650/2-74-013 (Appendix A of this report) was discovered. All of the cal-
culations had been made using the sum of the mean of the absolute values
and the 95% confidence interval. The referenced guidelines actually re-
quire reporting of the sum of the absolute value of the mean and the 95%
confidence interval. The method employed in calculating the results
reported here results in a larger absolute value of the mean error, but
a smaller confidence interval. Sample calculations were made to estab-
lish the impact of calculating the other way upon the reported results.
51
-------
In all cases examined the significance of the results would not be
altered, and in many cases (wherever all deviations were of the same
sign) there is no difference at all. In the interest of consistency,
therefore, all results reported were calculated using the sum of the
mean of the absolute values of the deviations and the 95% confidence
interval. Appendix F provides some examples of the difference of re-
ported errors that would occur if the other interpretation had been
used.
RESULTS OF TESTING AT FOSSIL-FUEL-FIRED POWER PLANT
Calibration Error (Before Field Installation)
This test was performed in accordance with the original Work Plan for the
power plant given in Appendix B. The test gases were introduced through
the span gas valves which were mounted in the sampling interface housings
(except for the Disc Diluter System). Introduction of gases through the
sample probes would have entailed excessive consumption of test gases
since two of the four systems were designed to operate on about 10 liters/
minute of sample, and additional flow would have been required to flood
the probes with gas. Introduction of gases at this point included all of
the systems except the stack probe, and leakage of critical valves would
have influenced the results. The test, therefore, adequately demonstrated
system performance to the extent practical in the laboratory. In the case
of the Disc Diluter, the test gases were also introduced through the span
gas inlet, with the disc rotating in the reversed mode as it would be for
an in situ calibration. The primary difference between the laboratory
test and a field test was, therefore, the operating temperature of the
Disc Diluter probe.
The results of this test are summarized in Table 3. This test was per-
formed at the conclusion of the laboratory familiarization phase just
prior to transport and field installation at the power plant test site.
52
-------
Table 3. CALIBRATION ERROR OF SYSTEMS EVALUATED AT POWER PLANT. Test
was performed in the laboratory immediately before transport
to and installation at the power plant test site.
Test Gas
Percent of Error* Calculated for Each System
Refrigerated
Condenser and
Dynasciences
Model SS-330
Beckman
Reflux Probe,
Permeation Dryer,
and Model 865 NDIR
Disc Diluter
and
Dynasciences
Model SS-310
Permeation
Diluter and
Flame Photometric
(Meloy)
Ul
157 ppm
S02 in N2
309 ppm
S02 in N2
458 ppm
S02 in P
3.1
4.5
4.5
4.0
1.5
0.6
4.4
4.4
2.4
0.6
2 ..7
2.1
*Errors reported are the sum of the mean of the absolute values and the 95%
confidence interval (computed the same way), expressed as a percentage of
the known test gas S02 concentration. Some of these errors would change,
but insignificantly, if computed using the absolute value of the mean of
the deviations.
-------
Since the percent of error requirement of EPA 650/2-74-013 for this test
is 5% or less, it is concluded that all four systems (including the Disc
Diluter with Model SS-310) passed this test.
Relative Accuracy
Thirty-two valid data sets for the determination of relative accuracy,
as defined in EPA-650/2-74-013, were obtained at the power plant site.
These data sets were divided chronologically into four groups of nine
sets each. The last set of data in each group of nine sets is the first
set of the next group, in accordance with the test procedure published
in the Federal Register.
The results of relative accuracy tests, reported on a dry-basis, with
95% confidence interval included, are summarized in Table 4. All four
systems were within the accuracy requirements of EPA 650/2-74-013 (20%
or less) for all four data sets. The available preliminary results ob-
tained with the Disc Diluter are given in Appendix E.
Two-Hour Zero and Calibration Drift
Two sets of data allowing determination of two-hour zero and calibration
drifts were obtained in the power plant test site. The results are ex-
pressed as percent of the emission standard, as required by the published
EPA procedures. The current emission standard is 0.54 kilograms (1.2
pounds) S02/2.3 x 10~° joules (million BTU) which may be shown to be
equivalent to about 450 ppn 862 using the following equation and
assumptions:
54
-------
Table 4. RELATIVE ACCURACY OF SYSTEMS EVALUATED AT POWER PLANT
Data Set
No.
1
2
3
4
Percent of Relative Accuracy*
Refrigerated
Condenser and
Dynasciences
Model SS-330
11.5
9.8
6.5
5.3
Beckman
Reflux Probe,
Permeation Dryer,
and Model 865 NDIR
11.4
10.2
7.9
5.3
Permeation Diluter
and Dynasciences
Model SS-310
10.9
13.5
8.2
6.3
Permeation
Diluter and
Flame Photo-
metric (Meloy)
6.4
11.5
12.6
15.9
*Reported values are after correction to dry basis where required, and are the sum
of the mean of the absolute deviation values and the 95% confidence level (com-
puted the same way). All of these results are unchanged if the absolute value of
the mean deviation is employed instead.
-------
?OQO
E = CF x l~ 54° PPm S°2
where
E = emission in kilograms S02/2.3 x 109 joules heat input
(pounds S02/million BTU heat input)
%02 = 5% by assumption for this calculation
C = S02 concentration in kilograms (pounds)/ meter3 (scfd)
F = fuel factor in scfd/2.3 x 1010 joules (10^ BTU) assumed to
be 100 (for coal) for this calculation
The results of the two-hour zero and calibration drift tests are sum-
marized in Table 5.
The requirements of EQP-650/2-74-013 for this test are that the drift be
2% or less for both zero and calibration. All four systems had acceptable
two-hour zero drifts in both tests (data sets). Only the Reflux Probe/
Permeation Dryer/NDIR System remained within specifications for both two-
hour calibration drift tests. The conventional sampling (refrigerated
condenser) and Model SS-330 System were acceptable in the first calibra-
tion drift test but failed the second. Both monitors operating on the
permeation diluter sample failed both two-hour calibration drift tests.
Twenty-Four-Hour Zero and Calibration Drifts
Two sets of data allowing determination of the twenty-four-hour zero and
calibration drifts were obtained at the power plant test site.
The discussion of the method of calculating and reporting results given
above also applies to the twenty-four-hour drift test results. The re-
sults of the twenty-four-hour zero and calibration tests are summarized
in Table 6.
The requirements of EPA 650/2-74-013 for these tests are 4% for zero drift
and 5% for calibration drift in twenty-four hours. All systems passed
56
-------
Table 5. SUMMARY OF TWO-HOUR ZERO AND CALIBRATION TEST RESULTS FOR POWER PLANT
TEST SITE
Data Set
No.
1
2
Percent of Emission Standard Zero and Calibration Drift* in Two Hours
Refrigerated
Condenser and
Dyna sciences
Model SS-330
Zero
0.7
0.7
Calib
1.2
3.2
Beckman
Reflux Probe,
Permeation Dryer,
and Model 865 NDIR
Zero
0.9
0.9
Calib
1.5
1.4
Permeation Diluter
and Dynasciences
Model SS-310
Zero
0.5
0.4
Calib
3.4
2.6
Permeation
Diluter and
Flame Photo-
metric (Meloy)
Zero Calib
0.2 3.9
0.2 3.2
*Reported values are the sum of the mean of the absolute values of the deviations and the
95% confidence interval (computed the same way). Some of these drifts would change, but
insignificantly, if the absolute value of the mean deviation had been used.
-------
Table 6. SUMMARY OF TWENTY-FOUR-HOUR ZERO AND CALIBRATION TEST RESULTS FOR
POWER PLANT TEST SITE
Data Set
No.
1
2
Percent of Emission Standard
Zero and Calibration Drift* in Twenty-Four Hours
Refrigerated
Condenser and
Dynasciences
Model SS-330
Zero
0.8
1.0
Span
3.2
2.0
Beckman
Reflux Probe,
Permeation Dryer,
Model 865 NDIR
Zero
0.7
2.0
Span
3.2
5.7
Permeation Diluter
and Dynasciences
Model SS-310
Zero
2.0
1.4
Span
8.4
6.3
Permeation
Diluter and
Flame Photo-
metric (Meloy)
Zero Span
1.6 8.7
1.4 9.4
00
*Reported values are the sum of the mean of the absolute values of the deviations and
the 95% Confidence Interval (computed the same way). Some of these drifts would
change, but insignificantly, if the absolute value of the mean deviation had been
used.
-------
both twenty-four-hour zero drift tests, but only the Refrigerated Conden-
sor/Dynasciences Model SS-330 passed both calibration drift tests. Both
monitors on the Permeation diluter failed both calibration drift tests,
and the Reflux Probe/Permeation Dryer/NDIR System failed one of the cal-
ibration drift tests.
Response Time
The response times of all systems evaluated at the power plant site were
determined twice. The results are summarized in Table 7. The require-
ment of EPA 650/2-74-013 for this test is 95% response in 600 seconds or
less, with a deviation of less than 15% between the up and downscale
responses. It may be concluded that all systems had acceptable response
times. Only the refrigerated condensor system displayed more than 15%
difference in the average upscale and downscale response times. This
difference could have been the result of different calibrations of the
zero and calibration gas flowmeters.
Operational Period
All systems operated continuously for over 168 hours in each of two
correlation test cycles without special corrective maintenance, which
satisfies the requirement of EPA 650/2-79-013. The errors reported
above for relative accuracy and for zero and calibration drifts were
obtained during these two test cycles.
Correlation Data with UV Absorption and in situ SO Analyzers at the
Power Plant Site
The power plant was equipped with both UV absorption and in situ SO
Monitors, as discussed in Section III. It was an objective of the Test
Plan to obtain as much data from these analyzers as possible within the
Work Plan time constraints to permit correlation of these systems with
the Method 6 results. Unfortunately, it was not always possible to ob-
tain average values of the S0_ concentrations indicated by these
59
-------
Table 7. SUMMARY OF RESPONSE TIME TEST RESULTS OBTAINED AT POWER PLANT SITE
Test
Parameter
(Average
Value
Reported)
Upscale
Downscale
Percent
Deviation
from
Slowest
Time to 95% Response in Seconds
Refrigerated
Condenser and
Dynasciences
Model SS-330
Test 1
171
132
23
Test 2
179
134
25
Beckman
Reflux Probe,
Permeation Dryer,
Model 865 NDIR
Test 1
43
38
12
Test 2
47
41
13
Permeation Diluter
and Dynasciences
Model SS-310
Test 1
119
126
6
Test 2
118
118
0
Permeation Diluter
and Flame Photo-
metric (Meloy)
Test 1
139
146
5
Test 2
150
144
4
-------
monitors during the time intervals of collection of Method 6 samples.
Furthermore, many of the cross-checks made during the preliminary
testing indicated that some results were probably erratic, and others
did not correlate well. Through personal correspondence with the man-
ufacturer's technical personnel after completion of the evaluation tests,
it was learned that the UV analyzer was found to have been out of cali-
bration during the months of the test, possibly due to an operator error.
This known error required a reduction of about 15% in the reported
values. After such correction and conversion to a dry basis, the aver-
age UV analyzer errors—including the 95% confidence interval^-for three
groups of nine data sets each, correlatable with Method 6 tests, were
11.4, 12.7, and 8.33%, respectively. Only three correlatable in situ
analyzer data sets were obtained. They indicated an accuracy of 6.6%
for the in situ analysis, after dry basis correction and with the 95%
confidence interval included.
Summary of Performance Criteria and Test Results for Power Plant Tests
The results of all power plant testing are summarized in Table 8, along
with the relevant performance criteria extracted from EPA 650/2-74-013.
Average values and/or ranges of errors are used in Table 8 in the interest
of achieving brevity without complete sacrifice of detail. While the
above discussions of each test result contain factual comments of a
passed-or-failed nature, examination of the summary data of Table 8 makes
it apparent that no gross failures were observed for any of the systems
evaluated. It should be noted in particular that the relative accuracy
error ranges for all five instrumental systems show coincidence; that is,
data groups 1 through 4 showed decreasing errors for all five systems
(including the UV analyzer). This strongly suggests that undetected
problems with the Method 6 analysis were involved, resulting in improved
Method 6 results with time.
61
-------
Table 8. SUMMARY OF PERFORMANCE CRITERIA AND TEST RESULTS FOR POWER PLANT TESTS
Test
Performed
Cali-
bration
Error
Relative
Accuracy
Two-Hour
Zero Drift
Twenty-
Four-Hour
Zero Drift
Two-Hour
Calibra-
tion Drift
Twenty-
Four-Hour
Calibra-
tion Drift
Mean
Response
Time
Oper-
ational
Period
Test Performance
Criteria from
EPA-650/2-74-013
5%
20%
2%
4%
2%
5%
95% in
600 s
168 hours with-
out special
maintenance
Refrigerated
Condenser and
Dynasciences
Model SS-330
3.1%
1.5%
2.4%
5.3 to
11.5%
0.7%
0.9%
(avg)
2.7%
(avg)
2.6%
(avg)
150 s
Maintenance
only between
two 168-hour
tests
Beckman
Reflux Probe,
Permeation Dryer,
Model 865 NDIR
4.5%
0.6%
0.6%
5.3 to
11.4%
0.9%
1.4%
(avg)
1.5%
(avg)
4.5%
(avg)
45 s
Maintenance
only between
two 168-hour
tests
Permeation
Diluter and
Dynasciences
Model SS-310
(Disc Diluter
used for this
test only)
4.5, 4.4, 2.7%
6.3 to
10.9%
0.5%
1.7%
(avg)
3.0%
(avg)
7.3%
(avg)
120 s
Maintenance
only between
two 168-hour
tests
Permeation
Diluter and
Flame Photometric
(Meloy)
4.0%
4.4%
2.1%
6.4 to
15.9%
0.2%
1.5%
(avg)
3.5%
(avg)
9.0%
(avg)
140 s
Maintenance only
between two 168-
hour tests
UV Absorption
N/A
8.3 to
12.7%
N/A
N/A
N/A
N/A
N/A
N/A
in situ
N/A
6.6%
(only 3
data
sets
N/A
N/A
N/A
N/A
N/A
N/A
NJ
-------
It should also be noted that the two Dynasciences Analyzers and the Meloy
Flame Photometer had rather large thermal coefficients, and that the
twenty-four-hour zero and calibration drifts were not, in general, cum-
ulative. The major sources of these drifts were as follows:
• Incomplete response within 10 minutes, on zero and span gases
(especially the refrigerated condenser system).
• Ambient temperature dependence of the monitors.
The Meloy flame photometer was known to have a defective component which
caused abnormally large thermal errors. Prom personal conversation with
Dynasciences technical personnel it appears that the models evaluated
are now obsolete and that smaller thermal errors should be expected from
currently available designs. The Beckman Model 865 S09 Monitor was the
only monitor evaluated having a potential direct interference by water
vapor. The data indicate that the sample was sufficiently dry at all
times (permeation dryer) but that this analyzer has a larger thermal
coefficient than would be desired for installation in an uncontrolled
environment. The typical Model 865, measuring 0-500 ppm S0?, with a
linearized output, will hold within about ±1% calibration for ±11°C
(20°F) ambient temperature variations. (While this was the smallest
coefficient of any of the analyzers evaluated, a two- or three-fold
improvement is desirable.) In addition, during preliminary testing, it
was found that the Model 865 gave erratic results when the Instrument Shed
temperature reached about 55°C (131°F). This fault could be corrected by
removing a circuit board cover within the analyzer. Apparently the lin-
earizing (or another) board overheated if covered when the ambient temp-
erature reached about 55°C.
The basic thermal sensitivity of the monitors, plus test personnel comfort
considerations, necessitated incorporation of a shed ventillation fan with
thermostat on/off control. This fan reduced the maximum shed temperature
to about 30°C (86°F) with the doors closed. The minimum shed temperature
63
-------
during the tests reported was about 10°C (50°F), which occurred during
early morning and late evening when the doors were open for performance
of tests. The ambient variations during the two 168-hour test cycles
reported were, therefore, 10 to 30°C (50 to 84°F), with rapid changes
occurring primarily at the beginning and end of each day of Method 6
correlation testing. The effects were especially pronounced on the
twenty-four-hour zero and calibration draft data sets, which were always
obtained early in the morning when the rate of change of ambient temper-
ature was large.
RESULTS OF TESTING AT GLAUS SULFUR RECOVERY PLANT
Calibration Error (Before Field Installation)
This test was performed at the conclusion of the laboratory familiariza-
tion phase before transport to and installation at the Glaus Sulfur Plant
test site. Gases were introduced through the span gas valves in the
sampling interface housing. The results are summarized in Table 9.
Because the response of the Houston Atlas System is very slow (over 10
minutes), it was tested on only two gases, as permitted in the newer EPA
Procedure (Federal Register, Sept. 11, 1974). In the final Work Plan
the Houston Atlas was the only system used for sampling of the tail gas
before incineration.
From the results summarized in Table 9, it appears that none of the sys-
tems tested met the calibration error specification of EPA 650/2-74-013
(5% or less) but it is important to note that the gases were not of pre-
cisely known concentration on all of the gases tested. The Model 865
NDIR gave acceptably reproducible results on the repeat readings, meaning
that either the linearization was in error or the -tank mixtures were not
as labeled. The Houston Atlas system (two H«S gas mixtures only) came
very close to "passing" this test, against tank labels. Similar drifts
of the Dynasciences Model SS-310 and the Meloy Analyzer outputs for
64
-------
Table 9. CALIBRATION ERROR TEST RESULTS OBTAINED BEFORE TRANSPORT TO
AND INSTALLATION AT THE GLAUS SULFUR PLANT TEST SITE
Test Gas,
Percent
Pollutant
in N
Percent Error* Calculated for Each System
Beckman
Reflux Probe,
Permeation Dryer,
Model 865 NDIR
(S02 Only)
Permeation Diluter
with Dynasciences
Model SS-310
(S02 Only)
Permeation Diluter
with Flame Photo-
metric (Meloy)
(S02 Only)
Houston Atlas
Diluter and
Monitor
(H2S Only)
0.375
S02
0.554
S02
0.881
S02
0.400
H2S
0.904
H2S
11.9
1.8
0.6
4.9
9.2
6.5
3.0
6.6
8.1
5.4
2.2
*The reported errors are the sum of the mean of the absolute values of the deviations and the
95% confidence interval (computed the same way). Two errors in this Table would be altered,
but insignificantly, if the absolute value of the mean deviation had been used.
-------
repeat analyses during the test suggest that the Dyfusatron temperature
was changing slightly during the test. Since the "standards" were
questionable, there appeared to be no merit in repeating this test.
Relative Accuracy of Systems after Incinerator (SO? Only)
Thirty-five valid sets of data were obtained for correlation to compli-
ance Test Method 6 (SO,,) for each of the three monitors sampling for S0«
after tail gas incineration. These thirty-five sets were divided into
four groups of nine data sets each, according to the approach outlined
in the Federal Register. Since the test site's UV monitor readings were
not taken simultaneously, they are not included in these results. The
results (corrected to a dry basis and with the 95% confidence interval
included) for the three monitors for the four groups of nine data sets
are given in Table 10.
The requirement of EPA-650/2-74-013 for this test is an accuracy of 20%
or less. Only the Meloy Permeation Diluter and Flame Photometric Analyzer
failed this test (3 out of 4 tests). The third data group includes one
set of data taken during a transient plant upset. During the upset
period, the S02 concentration exceeded 1% briefly, which was over-range
for all three monitors. It is possible that the linear ranges of monitor
operation were exceeded for this one set of data (out of nine sets).
Relative Accuracy of System before Incinerator (H?S)
Compliance Test Method 11 for H2S was found to be unusable on the high
(0.25% and more) H2S concentrations encountered. Consequently the only
correlation possible was against the H2S indicated by the existing test
site PGC. Details of the problems encountered with Method 11 are con-
tained in Appendix D. The valid sets of data for correlation between
the HA and the PGC were divided chronologically into two groups of nine
data sets each for calculating relative accuracy. In summary, the
66
-------
Table 10. SUMMARY OF RELATIVE ACCURACY RESULTS OBTAINED WITH
THREE SYSTEMS MONITORING THE GLAUS SULFUR PLANT
EMISSIONS AFTER INCINERATION
Data
Group
No.
1
2
3
4
Percent Relative Accuracy* of Systems Evaluated
Beckman
Reflux Probe,
Permeation Dryer,
Model 865 NDIR
3.8
5.4
9.6
3.0
Permeation
Diluter with
Dynasciences
Model SS-310
16.9
10.4
15.3
18.7
Permeation
Diluter with
Flame Photo-
metric (Meloy)
27.1
10.5
27.0
21.2
*Reported values are after correcting the dry basis where re-
quired, and are the sum of the mean of the absolute deviation
values and the 95% confidence interval computed the same way.
About one-half of these results would be altered by com-
puting with the absolute value of the mean of the deviations,
but the changes would not affect the conclusions that might be
drawn. See Appendix F for examples of these results computed
by each method.
67
-------
relative accuracies determined for the HA Diluter and Monitor were 44%
and 38%, respectively, for the two groups of nine data sets each. The
Houston Atlas System indication was lower than the process gas chromat-
ograph in every case, making the 6% difference for the two groups of
data significant with regard to stability and reproducibility. This is
an arbitrary determination of the relative accuracy of the HA Diluter and
Monitor; the accuracy of the process gas chromatograph used for reference
could not be rigorously .validated within the scope of this evaluation.
Two-Hour Zero and Calibration Drift
The general EPA practice of reporting these drifts in terms of the per-
centage of the relevant emission standard is inappropriate for the results
of this evaluation. The reason is that the instrument ranges were made 0
to 1% SC>2 (and H2S) to be consistent with the actual range of sample con-
centration. However, the applicable emission standards are equivalent to
about 15 to 20 ppm 862 and 0.15 ppm ^S concentrations, which are small
fractions of 1% of the instrument full-scale ranges. To resolve this
situation in a manner which is in keeping with the basic EPA reporting
philosophy, the calculations were made as if mid-scale instrument output
coincided with the emission standards, i.e., 0.5% of both SC>2 and ^S.
The results for the two-hour drift tests, reported upon this basis, are
summarized in Table 11. The reported results are dry-basis corrected,
and the 95% confidence interval is included.
The two data groups contained 15 data sets each for the SCL Monitors
sampling the incinerated tail gas. Data Group No. 1 for the Houston
Atlas System monitoring H2S before incineration contained 12 data sets.
The requirements of EPA 650/2-74-013 for two-hour zero and calibration
drift are both 2%. Only one system (NDIR) was within specification for
both zero tests, and of the other systems only the Meloy was within
specifications for one of the two zero tests. All but one of the systems
68
-------
Table 11. SUMMARY OF TWO-HOUR ZERO AND CALIBRATION TEST RESULTS OBTAINED
AT THE GLAUS SULFUR RECOVERY PLANT
Data
Group
No.
1
2
Two-Hour Zero and Calibration Drifts,* Percent of Mid-Scale
(0.5% S02 or H2S) Monitor Output
Beckman
Reflux Probe,
Permeation Dryer,
Model 865 NDIR
(S02)
Zero
0.8
0.8
Calib
1.8
2.2
Permeation Diluter
Dynasciences
Model SS-310 (S02)
Zero
4.6
5.2
Calib
13
14
Permeation Diluter,
Flame Photometric
(Meloy) (S02)
Zero
1.4
4.2
Calib
33
27
Houston Atlas
Diluter-Monitor ,
before
Incinerator (H2S)
Zero
6.0
-
Calib
50
-
*These drifts are the sum of the mean of the absolute values of the deviations and the 95%
confidence interval computed in the same way. Most of these drifts are altered, but
insignificantly if computed with the absolute value of the mean deviation.
-------
(NDIR) failed both of the calibration drift tests, and it failed one of
them. An explanation'of these results in terms of intermittent elec-
tronic failures is given in the following discussion of the twenty-
four-hour drift test results.
Twenty-Four-Hour Zero and Calibration Drift
Two data groups for determination of twenty-four-hour zero and calibra-
tion drifts were obtained for the three systems monitoring the Glaus
Plant tail gas after incineration and the one monitoring before inciner-
ation. These drifts were calculated and are reported on the basis out-
lined in the previous paragraph for the two-hour drift data. The data
are summarized in Table 12.
The requirements of EPA 650/2-74-013 for twenty-four-hour drifts are 4%
for zero and 5% for calibration. Only one system (NDIR) met the zero
drift specification for both tests, while the other three failed both
tests. All systems failed the calibration drift requirement on both
tests.
Reference to the raw data indicates that the Model 865 NDIR readings on
successive morning calibration cycles did not shift more than 4% of
reading in any case, and that the drifts on upscale gas (about 90% of
full scale) were generally both plus and minus. This raw data (mV out-
put) result is inconsistent with the calculated results for calibra-
tion drifts upon which the data of Table 12 are based. While a repeat
calculation would be possible, it would not significantly alter the
results; rather, raw data for the other systems reflect the erratic
behavior described further below. No attempt was made to assess the
validity of the calculated values of Table 12 for the erratic systems.
The Meloy Dyfusatron and photomultiplier tube (PMT) temperature controls
had both become unreliable by the time these (Glaus Plant) tests were
70
-------
Test 12. SUMMARY OF TWENTY-FOUR-HOUR ZERO AND CALIBRATION TEST RESULTS OBTAINED
AT THE GLAUS SULFUR RECOVERY PLANT
Data
Group
No.
1
2
Twenty-Four-Hour Zero and Calibration Drifts,* Percent of Mid-
Scale (0.5% S02 and H2S) Monitor Output
Beckman
Reflux Probe,
Permeation Dryer,
Model 865 NDIR
(so2)
Zero
1.2
1.0
Calib
12
7.2
Permeation Diluter,
Dyna sciences
Model SS-310
(so2)
Zero
5.8
5.8
Calib
36
36
Permeation Diluter,
Flame Photometric
(Meloy) (S02)
Zero
5.2
5.0
Calib
40
.40
Houston Atlas
Diluter-Monitor
before
Incinerator (H~S)
Zero
20
20
Calib
61
135
*The drifts reported are the sum of the mean of the absolute values of the deviations and the
95% confidence interval computed in the same way. These drifts are altered if calculated
using the absolute value of the mean deviation, but not to the extent that conclusions
drawn from the. results would be affected.
-------
performed. Examination of the two-hour and twenty-four-hour drift raw
data indicates that the two analyzers monitoring the Dyfusatron diluted
sample drifted randomly, and they also did not track each other. The
Dynasciences Model SS-310 output would be affected by the Dyfusatron
temperature (dilution ratio changes) only, while the Meloy Flame Pho-
tometric Monitor output would be affected by both the Dyfusatron and
the PMT temperature. The two effects, therefore, probably account for
both the random and the non-tracking nature of the two outputs.
Unfortunately, the Houston Atlas System had also become intermittent by
the time these parts were performed. The problem was not located and
corrected primarily because intermittent failures are very difficult to
isolate and identify. The raw data reflect the nature of the problem
encountered, and also indicate that periods of static performance were
observed. The raw data may be employed for new calculations if satis-
factory criteria for rejection of some data are available. For example,
if only the data for April 6, 7, and 8 are considered, the Houston Atlas
System performed very well.
It was decided that the tests should be continued after consideration
of the following factors:
s The major objective of the test was to evaluate the per-
formance of sampling interface and monitoring systems at
the site.
» The intermittent electronic failures (of three of the
four monitors) did not relate directly to the test site
application. That is, they had apparently developed
during, and had also effected the laboratory performance
noticeably prior to field installation.
« The erratic behavior was random, and the general impression
was that the results obtained when normal operation prevailed
were satisfactory.
72
-------
In summary, random failures (probably electronic) of three of the four
monitors adversely affected the evaluation results, but did not preclude
accomplishment of the primary objective. With no indication of degra-
dation of any component of the systems due to the nature of the special
application, the tests were continued. The large errors observed should
not be construed to mean that any given system proved to be unsatis-
factory since all evidence indicated that the causitive electronic
failures were not brought about by the evaluation testing.
Response Time
The response times of three systems evaluated at the Glaus Plant site
after incineration were determined twice. That of the Houston Atlas
(before incineration) was determined once. The results are summarized
in Table 13. The requirement of EPA-650/2-74-013 for this test is 95%
within 600 seconds, with a deviation of less than 15% for up and down-
scale responses. All systems except the HA exceeded the response time
requirement. In addition, the difference in up and downscale response
times for two systems (NDIR and Houston Atlas) were within the required
15%, while the two monitors on the Meloy Permeation Diluter failed the
response time difference requirement. This could have been due to a
simultaneous thermal transient (loss of control) and/or due to conden-
sation within the sampling system which escaped notice. It is probable
that this result was spurious, since reasonably fast and symmetrical
response was normal for the Dyfusatron System.
Operational Period
Following several sampling system revisions during start-up, the Reflux
Probe/Permeation Dryer/NDIR System operated for over 168 hours without
maintenance beyond that specified in the operating manual.
The Permeation Diluter (Meloy) Dynasciences System operated for over
168 hours, but was erratic in output. The erratic behavior is believed
73
-------
Table 13. SUMMARY OF RESPONSE TIME TEST RESULTS OBTAINED AT GLAUS PLANT.
All times except for Houston Atlas are for 95% response.
Test
Parameter
(Average
Values
Reported)
Upscale
Downscale
Percent
Deviation
from
Slowest
Time to 95 Percent Response in Seconds
(90 Percent and Minutes for Houston Atlas)
Beckman
Reflux Probe,
Permeation Dryer,
Model 865 NDIR
(S02)
Test 1
31
26
10
Test 2
27
28
4
Permeation Diluter ,
Dynasciences
Model SS-310 (S02)
Test 1
200
237
16
Test 2
92
99
7
Permeation Diluter,
Flame Photometric
(Meloy) (S02)
Test 1
191
255
25
Test 2
60
225
73
Houston Atlas
Diluter and
Monitor (H2S)
Test 1
17 min
16 min
6 min
Test 2
—
—
•-J
-e-
-------
to have been the result of Meloy Dyfusatron temperature control failures
and/or to Meloy Flame Photometric Analyzer intermittent problems, as
detailed above.
The Meloy System also developed problems with the photomultiplier tem-
perature controller, which further degraded the stability of the flame
photometric analyzer. It is believed that the rapid decrease in circuit
reliability of the Meloy System observed between the time of first tests
and the Claus Plant tests was a consequence of the breadboard quality of
the circuit boards used in the particular analyzer evaluated.
The Houston Atlas Diluter and lead acetate paper tape sampler-type monitor
operated erratically for unknown reasons. The problem that caused sub"-
standard performance of the Houston Atlas System during most of the field
testing was not identified, but had the characteristics of an intermittent
electronic failure.
None of the electronic problems appeared to be related to the character-
istics of the Claus Plant samples, which means that with respect to the
basic objective of the evaluation the systems could be said to have met
the 168-hour maintenance-free performance criteria.
SUMMARY OF PERFORMANCE CRITERIA AND TEST
Results for Claus Plant Tests
The results of all Claus Plant testing are summarized in Table 14,
together with the applicable performance criteria extracted and/or
adapted from EPA 650/2-74-013. Comparison of this summarized data to
that of Table 8 for the Power Plant test results indicates that the per-
formance of the systems was generally worse at the Claus Plant. However,
it must be noted that failure of temperature controls in the Meloy
75
-------
Table 14 . SUMMARY OF PERFORMANCE CRITERIA AND TEST RESULTS FOR
CLAUS PLANT TESTS
Test
Performed
Calibration
Error
Relative
Accuracy
Two-Hour
Zero Drift
TWenty-Four-
Hour Zero
Drift
Two-Hour
Calibration
Drift
Twenty-Four-
Hour Calibra-
tion Drift
Appropriate
Mean Re-
sponse Time
Operational
Period
Test Performance
Criteria
EPA-650/2-74-012
5%
20%
2%
4%
2%
5%
95% in
600 s
168 hours without
special
maintenance
Beckman
Reflux Probe,
Permeation Dryer,
Model 865 NDIR
(S02 only)
12%
2%
0.6%
4 to 10%
0.8%
1%
(avg)
2%
(avg)
10%
(avg)
30 s
No special main-
tenance required
Permeation
Diluter,
Dyna sciences
Model SS-310
(S02 only)
5%
9%
6.5%
10 to 19%
4.9%
(avg)
6%
(avg)
13.5%
(avg)
36%
(avg)
150 s
No special
maintenance
given
Permeation
Diluter,
Flame Photometric
(Meloy)
(S02 only)
3%
7%
8%
10 to 27%
2.8%
(avg)
5%
(avg)
30%
(avg)
40%
(avg)
200 s
No special main-
tenance given
Existing PGC
Sample Tap,
Houston Atlas
Diluter /Monitor
(H2S only)
5.4%
2.2%
38 to 44%
(against PGC)
6%
20%
(avg)
50%
98%
(avg)
960 s
No special
maintenance
given
-------
Analyzer affected both monitors operating from the Dyfusatron equally,
and had an additional affect upon the Meloy Flame Photometric Analyzer.
The Houston Atlas Mechanical Diluter and lead acetate paper tape ana-
lyzer became rather erratic after several days of testing, and the
problem was not identified. The behavior appeared to be characteristic
of intermittent electronics, but the summary data of Table 14 indicate
that the results of correlation testing (against the PGC only) were much
less sporadic than were the two- and twenty-four-hour zero and calibra-
tion drift test results. This suggests that there may have been a leak
in the zero calibration control valves (mounted in the contractor-
furnished sampling interface housing) which adversely affected the zero
and calibration drift results. In this regard, it should be noted that
the Houston Atlas readout is based upon the rate of change of paper tape
reflectance, and any large transients in concentration at the time this
rate was determined would result in an erratic output.
The Beckman Model 865 was apparently improperly linearized, since it was
12% high at 37.5% of scale in the calibration error test. It could have
been adjusted to give good agreement with the three test gases, but this
would have violated the purpose of the evaluation. It was linearized
against the original calibration data supplied in the instruction manual,
just as it would have been for a typical customer. The relatively large
twenty-four-hour calibration drift of the Model 865 NDIR shown in Table
14 (10%, in terms of mid-scale SO^ concentration) raises a question with
regard to the accuracy of data reduction. Referring to the source data
for Table 14 results, there was an apparent calibration drift of 0.08%
SO- for the second entry. This drift was about four times as large
as the average of the other five entries, and may have inadvertently
included a manual resetting of the calibration which should have been
excluded. Examination of the raw data supports the contention that an
intentional recalibration made at 1030 hours on 4-7-75 was inadvertently
77
-------
included as a calibration drift. No large drifts are reflected by the
raw (millivolts output) data. Elimination of this one data point would
reduce the average error in Table 14 to 7%. The difference in results
is not considered to be large enough to justify repeating the calcu-
lations for the one analyzer.
Finally, it should be noted that all zero and calibration drift data are
referred to mid-scale output. Consequently, all drifts are only about
one-half as large if calculated as the percentage of the calibration gas
(93% of scale).
In summary, it is believed that the results presented in Table 14 do not
indicate significant failures to comply with the requirements when proper
allowance is made for the known deficiencies in the analyzers under test.
The results should not be construed to mean that the normal commercial
analyzers of the types tested would suffer from similar problems. Perhaps
a longer test period would have indicated a significant degradation in one
or all systems, but there was no indication that the systems tested were
degrading with time because of the environment or because of the nature
of the sample streams.
COMMENTS ON SAMPLING/INTERFACE SYSTEMS, AND FIELD PERFORMANCE
A major objective of the subject contract was the evaluation of sampling/
interface systems. Accordingly, the salient design features and field
performance results for each system are summarized in the following
paragraphs.
Refrigerated Condensor-Type System (Power Plant Only)
System Features -
An overview of the salient features of this system is given by the
following:
78
-------
• Probe using 20-micrometer filter with automatic air blowback
cleaning, followed by a heated one-micrometer fiber glass
filter.
• Heated Teflon sample line to condenser.
• Refrigerated condenser (0.5% water vapor in effluent) with
corrosion-resistant automatic drain trap (PVC).
• Materials contacting sample were alloy 20, 18/8 stainless
steel, Teflon, glass, Kynar,* PVC, Viton,** and Teflon-coated
stainless steel.
• An electrochemical S0_ monitor (Dynasciences Model SS-330)
analyzing on a dry basis.
Field Performance —
Comments on the field performance and the degradation observed by visual
inspection after 60 to 70 days of continuous testing are as follows:
• Probe filter welded joint corroded and filter fell off by the
end of the Power Plant tests. The time to failure is unknown.
• One-micrometer filter (heated) collected a great deal of dry
fly ash.
• Refrigerated condenser and PVC.trap both operated continuously
and there was no evidence of corrosion of the 316 stainless
steel parts.
• Diaphragm sample pump with Teflon-coated 316 stainless-steel
diaphragm showed evidence of corrosion occurring at pump outlet
only.
* Registered trademark of the Penn Walt Corporation
**Registered trademark of the duPont E.I. DeNemours and Company
79
-------
Meloy Permeation Diluter System (Both Sites)
System Features
The sampling probe and interface were of contractor design. The system
was similar to the others in all matters of choice. The important fea-
tures are as follows:
• Probe using 5-micrometer filter with automatic air blowback
cleaning, but no additional filter because of the low Meloy
sample flow rate.
• Heated Teflon sample line up to Meloy Analyzer.
• Permeation dilution by diffusion of sample constituents through
a Teflon membrane into a carrier air stream (Meloy Dyfusatron).
• Materials contacting the undiluted sample upstream of the
Meloy Analyzer were 18/8 stainless steel, Teflon, glass,
Kynar, and Viton.
• An electrochemical S0~ monitor (redundant), analyzing on a wet
sample basis (Dynasciences Model SS-310).
• A flame photometric total sulfur monitor (Meloy), analyzing on
a wet sample basis.
Field Performance -
Comments on the field performance and the degradation observed by visual
inspection after 60 to 70 days of use are as follows:
• Evidence of probe filter corrosion was found.
• Dyfusatron membrane was replaced several times due to puncturing
by pressure surges (see comments following).
• Low sample flow rate necessitated field addition of heating of
the stack-to-interface housing flange to avoid condensation of
moisture in the probe.
• After the Glaus Plant evaluation, white-green-yellow deposits
were found in the probe, but the Alloy 20 probe did not corrode.
80
-------
The failures of the Dyfusatron Teflon membrane require additional com-
ment. The Dyfusatron membrane (0.05-mm-thick Teflon at up to 208°C) will
not tolerate differential pressures of any magnitude. After experiencing
a permeation membrane failure in the initial laboratory, familiarization
tests, support screens (stainless steel mesh), and suitable positioning
rings were added to provide improved support for the membrane. (Meloy
has added supports to later, models of the Dyfusatron.) In retrospect,
very fine screens or sintered discs would have been superior and would
not have affected the membrane permeation rate or speed of response sig-
nificantly. Numerous membrane failures still occurred, as detailed below.
The Meloy Analyzer was designed for sampling from an excess of sample gas
at approximately atmospheric pressure, which the power plant stack and
Glaus Plant duct provided. Zero and calibration gases had to be delivered
to the Meloy unit on the same basis, which necessitated the use of excess
gas flow vented to atmospheric pressure at the Meloy unit zero and cali-
bration gas inlet ports. Since it was an objective of the program to
introduce standardization gases as close to the sampling point as possible,
the built-in Meloy standardization solenoid valves were disconnected and
duplicated in the contractor-furnished sampling interface housing—as
they were for the other systems evaluated. The Meloy unit included a
negative pressure actuated relief valve near the Dyfusatron to protect
the permeable membrane (raw sample side) from excessive negative pressures.
After the second membrane failure in laboratory testing, the Meloy relief
valve was replaced with a valve that operated at a lower differential
pressure and had a lower flow impedance. A similar positive pressure
actuated relief valve was used in the contractor-designed interface
housing to protect against positive pressure surges due to introduction
of standard gases. In the preliminary power plant tests, it was found
that both relief valves had a tendency to stick (either open or closed).
Both were moved into the Meloy unit and located in a lower temperature
zone than was the original Meloy relief valve. However, failures of the
81
-------
membrane were still experienced.. The carrier gas flow system of the
Meloy is much less complicated than is the raw sample system; consequently,
no carrier-side pressure transients were anticipated. It is possible that
unrecognized factors caused failures due to carrier-side pressure tran-
sients, but it is more probable that the very light relief valves (set
for operation on a differential of only about 25 cm of water column) were
sticking even when maintained near ambient temperature.
Permeation Drying System (Both Sites)
System Features
• Reflux probe (rather than a probe filter) and a 1-micrometer
fiber glass filter for the total sample flow.
• Permeation distillation dryer to reduce effluent moisture to
less than 0.1%.
• Materials in contact with the sample were Alloy 20, 18/8 stain-
less steel, Teflon, glass, Kynar, Viton, Teflon-coated stain-
less steel, and the "plastic" permeation element tubes, tube
headers, and the cement used to bond the tubes to the headers.
• Model 865 NDIR S0_ Monitor, analyzing on a dry sample basis.
Field Performance -
Comments on the field performance and the degradation observed after 60
to 70 days of continuous operation are as follows:
• Shrinkage of length of bundle of water-permeable tubes caused
inlet side tube header to crack and then crumble.
• Coalescing of sulfuric acid at the outlet end of the permea-
tion dryer resulted in degradation of the header and/or
element, forming a black, rubber-like mass.
• A.stainless-steel relief valve (Viton seat) stuck closed,
eliminating the reflux action and resulting in intake of
beyond normal amounts of fly ash.
82
-------
• One-micrometer filter (heated) collected a large amount of
dry fly ash.
• Brown acidic droplets were visible throughout the part of the
system mounted in the interface housing, indicating inadequate
housing heater operation.
• Inadequate heating caused condensation in and corrosion of the
Teflon-coated stainless-steel diaphragm sampling pump.
• White, water-soluble material accumulated in the cooler zones
of the reflux system at the Glaus Plant site. (See comments
below.)
• The Beckman Model 865 NDIR SO- Analyzer showed some problems
with regard to ambient temperature effects. In addition, one
defective circuit board was replaced during the laboratory
familiarization phase prior to the power plant tests, and the
latest design revision was incorporated to provide inde-
pendence of adjustment of the zero and calibration controls.
An attempt was made to analyze a sample of the white deposit found after
testing at the Glaus Plant. Briefly, the material was highly soluble in
water. A barium chloride test indicated that the material was not a sul-
fate. An infrared spectrophotometer absorption scan was made on a sample
(formed into a KBr pellet) with inconclusive results. When wetted with a
drop of KOH solution, the salt turned yellowish and the odor of ammonia
was observed—proving that it was an ammonium salt. No attempt was made
to identify the specific salt, but ammonium chloride power condenses at
340°C, is quite soluble in water, and is white. Deposits found in the
probe used with the permeation diluter system were faintly colored, but
were probably the same salt. According to the test site personnel,
traces of ammonia are found in the tail gas, both before and after the
incinerator.
83
-------
Condenser, Mechanical Diluter, Lead Acetate Tape (Glaus Plant Only)
System Features -
• Sample was obtained from an .existing tap used for a process gas
chromatograph (PGC) after removal of free sulfur by a steam
temperature-controlled condenser.
• Sulfur and water removal by a chilled-water-cooled condenser
with manual drain and steam cleaning features, followed by
submicrometer filtering.
• Materials contacting the sample up to the mechanical diluter
(Houston Atlas) were 18/8 stainless steel, Teflon, glass,
Viton, and carbon steel.
• Mechanical dilution of sample by carrier air (Houston Atlas).
• Lead acetate paper tape type H S analyzer (Houston Atlas)
analyzing on a dry sample basis.
Field Performance -
The major results of field testing and the observations made by visual
inspection during and after testing for about 30 days are as follows:
• Sulfur condensed in the electrically heated line between the
existing steam temperature-controlled condensor and the
chilled-water-cooled condensor, indicating that a higher
temperature was desirable.
• Steam cleaning of the chilled-water-cooled condensor did not
remove all sulfur from the cooling coil, indicating that a
higher steam pressure is desirable.
• The submicrometer filter element turned from white to green
color, indicating a possible reaction with the sample.
• There was no evidence of corrosion except for rusting of the
carbon steel chilled-water condensor parts.
84
-------
Summary of Desired Improvements in Systems
The following brief comments summarize the field performance results from
the standpoint of features known to require improvement.
Refrigerated Condenser System -
• Better corrosion resistance of the stainless-steel probe filter
welds is desirable.
• The diaphragm pump (stainless steel with Teflon-coated stain-
less steel disphragm) corroded significantly where condensate
was allowed to form. Better corrosion resistance of the pump
is desirable.
• Location of the condenser near the probe to shorten heated line
length would improve the speed of response.
Permeation Diluter (Meloy) System) -
• Provision of improved support and/or pressure surge protection
for the permeable membrane is highly desirable.
• A sampling system permitting a higher flow rate would signifi-
cantly improve the response time.
• A built-in scheme for conversion of results to a dry basis
analysis would facilitate correlation to Compliance Test
Method data and to the emission standards, which are also on
a dry basis. (This point must be balanced against the pos-
sible advantages of direct, wet sampling and the difficulty
of making the conversion to dry basis internally.)
Permeation Drying System -
• Materials of construction of the Permapure Products, Inc.,
Dryer should be changed for better compatibility to SO-, sul-
furic acid, and other common flue gas constituents.
85
-------
• Allowance for the large shrinkage of the dryer tube material
must be made in the design of the assembly. Otherwise, this
special material appeared to be satisfactorily compatible with
the gases sampled.
• Better corrosion resistance of the diaphragm pump, which
corroded significantly where condensation occurred, is
desirable.
• A more uniform controlled temperature for the sampling/inter-
face housing, and especially for the sample pump, would be
beneficial.
Disc Diluter (Not Fully Evaluated Due to Failure) -
• Redesign of the rotating disc drive train would permit con-
tinuous operation with acceptable disc leakage.
• Redesign of disc thrust loading mechanism would allow for
greater buildup of fly ash and/or operating temperature
changes without creating galling and binding.
• Choice of a disc material more compatible to the application
than Delrin or Teflon. In particular, a high temperature
material is required for general sampling applications.
Chilled-Water Condenser -
e Better corrosion resistance (than galvanized iron pipe) is re^
quired for long-term service.
o Closer coupling to the high temperature source and/or higher
temperature for the line to the condenser to prevent conden-
sation of sulfur is required for longer service.
« Empirical selection of steam cleaning parameters, and auto-
mation of the cleaning cycle, would be important with regard
to maintenance requirements.
0 The Houston Atlas System responds very slowly but no specific
recommendations for improvement can be made because of the
brief period of the evaluation.
86
-------
SECTION VI
APPENDICES
A. ORIGINAL AND REVISED WORK PLANS FOR FIELD TESTS
AT POWER PLANT ' 88
B. EPA DOCUMENTS USED IN DESIGN OF FIELD WORK PLANS. . . . 115
C. MONITORING SYSTEMS DRAWINGS 139
D. RESULTS OF AN EVALUATION OF EPA COMPLIANCE TEST
METHODS 6(S02) AND 11(H20) IN THE RANGE OF 0.15
TO 0.9% OF EACH IN NITROGEN 1A8
E. MISCELLANEOUS TESTS, CALCULATIONS, AND RESULTS 167
F. COMMENTS ON INTERPRETATION OF CALCULATIONS AND
REPORTING INSTRUCTIONS OF EPA-650/2-7A-013 188
87
-------
APPENDIX A
ORIGINAL AND REVISED WORK PLANS FOR
FIELD TESTS AT POWER PLANT
88
-------
CONTENTS
Page
Original Work Plan for Field Tests at Power Plant 90
Revised Work Plan for Field Tests at Power Plant 107
Revised Field Test Plan for Glaus Plant no
89
-------
October 7, 1974
(Revised and abridged June 25, 1975,
for inclusion in Final Report)
ORIGINAL WORK PLAN FOR FIELD TESTS AT POWER PLANT
1.0 INSTALLATION AT POWER PLANT SITE
The stack will be breached at the 13.41-meter (44-ft) level where probes
and associated gear for the four analytical systems and for manual sam-
pling will be located. The instruments and peripheral gear will be
located in a shelter that will be erected on a platform near the probes.
2.0 SAMPLING/INTERFACE/MONITORING SYSTEMS
All systems except the Disc Diluter system will be equipped with auto-
matic zero and span modules. Detailed drawings of the complete systems
are located in Appendix B and the systems are described in Section IV
of the Final Report.
2.1 Analytical System Operation and Maintenance
All four systems will be operated continuously for the 30-day test period
during which all but the Disc Diluter System will be zeroed and spanned
automatically, normally on a 24-hour basis. The Disc Diluter System will
be zeroed and spanned manually and only at those times that manual samples
are being taken.
Reasonable, normal maintenance—as specified in the manufacturers' instruc-
tion manuals—will be performed. However, a major breakdown in any of the
analytical systems requiring extensive field or factory service or repair
may result in cessation of testing of that particular system. This is a
necessary condition of the Field Test Work Plan since the period of testing
is too short to allow abnormal service or repair and still allow obtaining
all of the specified, correlated data.
90
-------
2.1.1 Disc Diluter
The Disc Diluter provided for this program was designed for sampling
stacks for relatively short periods and not for continuous operation.
This is implicit in the maintenance instructions supplied with the
equipment, summarized as follows:
NOTE
Due to insufficient continuous testing, exact
data for recommended maintenance schedule are
lacking.
1. Disc resurfacing: The Delrin disc required resurfacing at one-
hour intervals for stack conditions of 148.8°C (300°F), 3%
moisture, and heavy fly ash loading. The Karak disc was still
OK after 10 hours at temperatures from 148.8° to 387.7°C (300°
to 550°F).
2. Disc replacement: Replacement should occur when disc thickness
is less than 6.4 mm (0.250 inch)
3. Vacuum filter: The filter was never replaced. Expected inter-
vals of replacement are every 24 hours except under very heavy
particulate loading.
4. Silica gel dryer: The dryer requires changing every 3 hours
at moisture level of 3% and 148.8°C (300°F).
5. Disc head lubrication: Intervals have not been determined.
Lubricate with silicone grease whenever disassembled.
It will be impossible to resurface the disc at such frequent intervals.
If such is required, the attempt to operate the Disc Diluter System on a
continuous basis will be abandoned, It is hoped that under the conditions
of the test site stack the disc will provide reasonable service and con-
tinuous operation will be possible. Also, it would be impossible to re-
place the silica gel every 3 hours. To avoid this, an aspirator will be
substituted for the normal pump in the system, eliminating the need for
the silica gel dryer. The effect this may have on the overall performance
of the system is unknown.
91
-------
3.0 INFORMATION TO BE ACQUIRED
The contract states, "Systems outlined under 3.a. shall be installed and
operated for a thirty-day period. Each system shall have the capability
for automatic zero and calibration every 24 hours. System response shall
be correlated with EPA Standard Compliance Test Method 6 for SC^. A min-
imum of 27 Method 6 samples shall be taken at intervals of nine samples
per week over the thirty-day period such that each system can be corre-
lated with at least 18 different manual samples." Summarized, this calls
for 27 Method 6 analyses of which 18 (or more) shall be correlated with
readings of each monitoring system. In view of the published EPA per-
formance specifications for stationary-source monitoring systems (EPA-
650/2-74-013), it would enhance the value of this study if it were
possible to include the determination of all of the parameters listed
there:
1. Relative Accuracy (obtainted from the correlated readings)
2. Calibration Error
3. Two-Hour Zero Drift
4. Twenty-Four-Hour Zero Drift
5. Two-Hour Calibration Drift
6. Twenty-Four-Hour Calibration Drift
7. Response Time
8. Operational Period
The relative accuracy is, of course, the information that is required by
the above-quoted portion of the contract. The 24-hour zero drift and the
24-hour calibration drift can be determined by utilizing the specified
capability for automatic zero and calibration. The other five parameters
are beyond the original scope of the contract but it appears that they can
be included to a large extent in the tests performed at home and at the
steam plant without increasing the total cost of the project.
All 8 parameters will be determined for the conventional Sample Handling/
Dynasciences System, the Reflux Probe/NDIR System, and the Meloy Permeation
Diluter/Flame Photometric System. Parameters 1, 2, and 7 will be determined
92
-------
for the EPA Disc Diluter/Dynasciences System. The 2-hour and 24-hour zero
and span drifts will not be determined for the latter system since it will
not have the automatic zero and span capability. The standard information
on operational period will be replaced by a full report on the maintenance
log for this system.
4.0 DATA TO BE ACQUIRED FOR THE VARIOUS PARAMETERS
4.1 Relative Accuracy
At least 27 manual stack samples will be taken for analysis by the wet
chemical Method 6.
NOTE
The standard Method 6 will be followed
except that stack velocity will not be
measured since it is not required for
determining concentration.
Simultaneous output signal recordings will be obtained with all four ana-
lytical systems during 18 (or more) of the manual samplings. The zero
and span readings will be checked on all analytical systems before and
after taking each manual sample.
Immediately before or after taking each Method 6 sample, the water vapor
content of the stack gas will be determined by Standard Reference Method 4.
4.2 Calibration Error
During checkout testing of the analytical systems before taking them to
the field site, the following procedure will be performed:
• Each system will be calibrated with S02/N2 gas blends repre-
senting approximately 30, 60, and 90 percent of span. A
series of 5 nonconsecutive readings will be made at each con-
centration (Example: 30%, 90%, 60%, 90%, 30%, 90%, 60%, etc.)-
93
-------
This is the only parameter that will be determined at home. All others
will be done at the field site.
4.3 Two-Hour Zero and Span Drift
Readings will be obtained on three of' the four systems for zero and cali-
bration gases at 2-hour intervals until 15 sets of data are obtained.
During the period of this test, the automatic zero and span modules will
be set to operate every two hours. The Disc Diluter system will not
participate in this test.
Calibration checks will be made with test gas concentration between 70
and 90 percent of span.
4.4 Twenty-Four-Hour Zero and Span Drift
Readings will be obtained on three of the four systems for zero and cali-
bration gases at 24-hour intervals throughout the 30-day test period in
similar manner to that described in the 2-hour test. The Disc Diluter
system will not participate in this test.
4. 5 Response Time
Each of the four systems will be flushed with zero gas, introduced as
close to the sample interface as possible. After the reading has stab-
ilized, the upscale calibration gas of concentration between 70 and 90
percent of span will be quickly substituted for the zero gas. The time
from switching gases to final stable response will be recorded. After
the system response has stabilized at the upper level, zero gas will be
quickly substituted for the upscale gas and the time from switching gases
to final stable response will be recorded. This test sequence will be
performed three times.
94
-------
4.6 Operational Period
Detailed maintenance logs will be kept for each analytical system.
5.0 DATA ANALYSIS AND REPORTING
5.1 Relative Accuracy
This calculation uses the 18 (or more) Method 6 test data and the ana-
lytical system recordings obtained during the manual sampling. The
recordings are first corrected for any zero or span drift occurring
during the sampling period, then the average reading over the sampling
period is determined (by eye) and converted to concentration by refer-
ence to the zero and span readings taken immediately before and after
the sampling period. Finally, the concentration figure is converted
to dry basis in the cases of the Disc Diluter system and the Permeation
Diluter System, using the result of the water vapor determination (Method
4) performed immediately before or after the sampling period. The con-
centration figure is already on the dry basis in the cases of the conven-
tional system and the Reflux System.
The Method 6 concentration is subtracted from the analytical system con-
centrations. This is repeated for all 18 (or more) sets of data. Using
these data, the mean difference and the 95% confidence interval will be
computed for each analytical system. The sum of the absolute mean value
plus the 95% confidence interval will be reported, as a percentage of
the mean reference value, for each analytical system.
5.2 Calibration Error
Using the data from Section 4.2, the known value will be subtracted from
the value shown by each analytical system for each of the 5 readings at
each span test concentration. The mean of these difference values and
the 95% confidence interval will be calculated for each of the analytical
95
-------
systems. The sum of the absolute mean value plus the 95% confidence in-
terval will be reported, as a percentage of each test gas concentration,
for each analytical system.
5.3 Two-Hour Zero Drift
Using the zero readings from Section 4.3, the drift occurring during each
2-hour period during the test will be calculated for each analytical sys-
tem. From these sets of data, the mean value and the 95% confidence
interval will be calculated for each analytical system. The sum of the
absolute mean value plus the 95% confidence interval will be reported,
as a percentage of the full-scale concentration range, for each of the
analytical systems.
NOTE
"Full-'scale concentration range" refers to the
range of concentration in the stack that gen-
erates full-scale output from the analytical system.
5.4 Twenty-Four-Hour Zero Drift
Using the zero readings from Section 4.4, the differences between the zero
reading after adjustment and the zero reading 24 hours later, just prior
to zero adjustment, will be calculated for each set of data, for each ana-
lytical system. Using these data, the mean value and the 95% confidence
interval will be calculated for each analytical system. The sum of the
absolute mean value plus the 95% confidence interval will be reported, as
a percentage of the full-scale concentration range, for each analytical
system.
5.5 Two-Hour Calibration Drift
Using the calibration readings from Section 4.3, the differences between
the readings and the test gas value will be calculated. These values
will be corrected for the corresponding zero drift during the 2-hour
96
-------
period. The mean and the 95% confidence interval of these corrected
values will be calculated and the sum of the absolute mean value plus the
95% confidence interval will be reported, as a percentage of the full-
scale concentration range, for each analytical system.
5.6 Twenty-Four-Hour Calibration Drift
Using the calibration readings from Section 4.4, the differences between
the readings and the test gas will be calculated. These values will be
corrected for the corresponding zero drift during the 24-hour period.
The mean and the 95% confidence interval of these corrected values will
be calculated and the sum of the absolute mean value plus the 95% confi-
dence interval will be reported, as a percentage of the full-scale con-
centration range, for each analytical system.
5.7 Response Time
Using the data from Section 4.5, the time intervals from concentration
switching to 95% of the final stable values for all upscale and downscale
tests will be calculated. The mean of the three upscale test times and
the mean of the three downscale test times will be calculated. The slower
of the two times will be reported for each analytical system,
5.8 Operational Period
From the maintenance logs kept on the analytical systems, it will be de-
termined how long each system operated without requiring corrective main-
tenance, repair, replacement, or adjustment other than that specified in
the respective instruction manuals, as routine and expected in a 1-week
period. From the rest of the data, these operating periods will be cor-
rected downward, if necessary, to the period within which each system
displayed no more than the following drift or error for any one parameter:
97
-------
a. Relative Accuracy 20% of mean reference value
b. 2-Hour Zero Drift 2% of full scale
c. 24-Hour Zero Drift 4% of full scale
d. 2-Hour Calibration Drift 2% of full scale
e. 24-Hour Calibration Drift 5% of full scale
The Disc Diluter System will not be judged in this manner since it is
not intended to be used for continuous monitoring. Instead, its detailed
maintenance log will be reproduced in full in the report.
6.0 CALCULATIONS
6.1 Calculating Mean Values
The mean value of a data set is calculated according to equation (E-l)
_ , n
x = - X) x± CE-1)
n ^^ •*-
iFl
where x. = individual values
Ic = mean value
n = number of data points
E = sum of the individual values
98
-------
6.2 Calculating 95% Confidence Intervals
The 95% confidence interval (two-sided) is calculated according to
equation (E-2).
where
r T - fcl-a/2
L» • J- •
Qt-
95 /S n(n-l)
Ex. ?=. sum of all data points
/n = square root of the number of data points
l-ot/2 = .975 for n samples from a table of percentages
of the t distribution
C.I. s = 95% confidence interval estimate of the average
mean value
Typical Values for tl-a/2
n
2
3
4
5
6
t.975
12.706
A. 303
3.182
2.776
2.571
n
7
8
9
10
11
'.975
2.447
2.365
2.306
2.262
2.228
n
12
13
14
15
16
t.975
2.201
2.179
2.160
2.145
2.131
The values in this table are already corrected for n-1
degrees of freedom. Use n equal to the number of
samples as data points.
99
-------
7.0 DATA ACQUISITION SYSTEM
7.1 Analog Recorders (Strip Chart)
Analog recorders (Beckman supplied) shall be used to record the output of
each monitor. The recorders shall be connected directly to the basic ana-
lyzer output before the auto zero and calibration module to provide a
continuous direct recording of the output of each monitor. Appropriate
hand-written notes shall be made directly upon the recorder charts to
provide a permanent record of the test results.
While not contractually required, such real-time records of monitor per-
formance will be used as aids to test personnel. Additionally, all re-
quired test data will be available on'the permanent records, providing
immediate visual assurance that significant results are being obtained
and/or immediate indication of failure of one or more systems. To the
maximum extent allowed by availability of test equipment, significant
parameters such as ambient temperature and stack gas temperature will be
monitored and recorded on strip-chart recorders when and as necessary or
relevant.
7.2 Digital Data Acquisition
An Esterline-Angus Model D2020 and teletype (both GFE) shall be employed
to provide periodic typed outputs for all analog data channels. In
addition, the teletype shall be used to generate a punched tape of all
data output by the Model D2020. The punched tape will be suitable for
use in computer data analysis. Data shall be acquired at 1-minute inter-
vals while collecting samples for Method 6 analysis, at 2-hour intervals
for zero and calibration 2-hour drift analysis, and at 1-hour intervals
for the overall 30-day test.
As a minimum, all monitor outputs (direct), and all monitor outputs from
the auto zero-cal module shall be connected to the Model D2020 for tele-
100
-------
type printout and for punched-tape recording. In addition, the Model
D2020 automatically provides the time at which the data are obtained with
1-minute resolution. Since the stack gas temperature directly affects
the dilution ratio of the Disc Diluter, a stack temperature monitor will
be provided by Beckman to provide a stack gas temperature input to the
Model D2020 and on the punched tape.
To the maximum extent possible such parameters as ambient temperature shall
also be recorded on the teletype print-out and on the punched tape. At the
option of the test personnel, for example, the stack gas temperature may be
removed from the stack and utilized in monitoring the temperature of heat
traced sample lines.
The objectives of data acquisition shall be as follows:
1. To obtain all necessary data required to fulfill the requirements
of Sections 2,0 through 6.0 and 9.0 of this Work Plan as a first
priority, and
2. To fully utilize on-site time and available equipment to obtain
maximum information relevant to the accomplishment of the first
objective. In this regard, the field test personnel shall have
broad latitude with respect to objective 2.
Finally, the raw data obtained using EPA Compliance Test Method 6 shall be
manually entered on the punched tape through the teletype to facilitate
computer analysis of all data. The entry procedure will be added to the
modified test protocol as the final step, which will include an identi-
fying code followed by date, time of day, and the necessary data.
101
-------
8.0 DATA ANALYSIS
8.1 Analysis of Data from Strip Chart Recorders
Sections 2.0 through 6.0 of this Work Plan can be accomplished with ana-
log strip-chart recordings of the relevant data coupled with manual data
reduction. Since such recordings are the normal data output for source
pollution monitoring systems, this method of analysis shall be considered
acceptable for the first priority objective of this test program. Accord-
ingly, at least preliminary data analysis after each Method 6 test shall
be performed as soon as possible using the analog strip-chart recordings
to ensure that meaningful results were obtained. Within the constraints
of time, test personnel shall have the option of repeating any tests for
which such preliminary analyses indicate a radical deviation between in-
strumental and Method 6 results which might indicate a gross error in
procedure or technique.
8.2 Analysis of Data by Computer Technique
The secondary objective of this program can best be accomplished by acqui-
sition of all raw data on a punched tape suitable for analysis by a dig-
ital computer. Once all relevant data are on punched tape, the possi-
bilities of sophisticated data analysis are limited only by available
funding. In this regard, the recording of ambient temperature and critical
monitor temperatures would even permit analysis of thermal errors for each
monitor tested. As noted above, all relevant data beyond the contractual
scope of this program shall be recorded on punched tape within the limits
of equipment availability and test personnel time. This will permit sub-
sequent extensive data analysis by the EPA as desired. This Work Plan,
however, is restricted to analysis by computer programming as defined in
the following paragraph
A computer program capable of performing the data analysis required by
Sections 5.1 through 5.6 and 6.1 through 6.2 shall be prepared and
102
-------
functionally tested prior to commencement of field testing in accordance
with this Work Plan. In addition, the computer program shall provide for
calculation of daily average SC>2 concentrations for each monitor (based
upon data acquired every hour) for the entire thirty-day test period.
The punched-tape data accumulated after start-up shall be analyzed as
soon as possible to ensure that the system is functioning properly. At
the conclusion of each set of nine (9) Method 6 tests, the punched-tape
data shall be computer analyzed to provide the data required by Sections
5.1 through 5.8 and 6.1 through 6.2 of this Work Plan.
8.3 Final Report Data Analysis
At the option of the Beckman Program Manager, the final report shall contain
either or both data analyzed by manual techniques and data analyzed by com-
puter techniques.
9.0 SCHEDULE AND MAN LOADING
The field work is expected to require 7 weeks after the site is fully prepared
to accept the analytical systems and all equipment has been delivered to the
site. The following schedule indicates the expected step-wise performance of
the tasks and the man loading to perform them:
Nov. 11-15
Install and start up all analytical systems; all but Disc Diluter
start continuous operation by Nov. 15.
Tech. - 5 days; one A.E. - 5 days; other A.E. - last 3 days
Nov. 18
Set up chemical lab. and run one Method 6, not necessarily corre-
lated with analytical systems.
Tech. - 1 day; A.E. - 1 day; Scientist- 1 day
103
-------
Nov. 19
Run 1 Method 6 correlated with all systems except Disc Diluter, and
1 Method 6 correlated with all 4 systems.
Same man loading as Nov. 18
Nov. 20
Run 2 Method 6's correlated with all systems; Disc Diluter starts
continuous operation.
Same man loading as Nov. 19
Nov. 21
Reduce data obtained so far.
Same man loading as Nov. 20
Nov. 22
Start 30 day run.
Same man loading as Nov. 21
Nov. 25-28
Run 2 Method 6's per day correlated with all 4 systems, if possible,
plus 2 Method 4's per day.
Tech. - 4 days; A.E. - 4 days
Nov. 29
Run 1 Method 6 correlated with all 4 systems, if possible, plus
1 Method 4.
Tech. - 1 day; A.E. - 1 day
Dec. 2
Check all analytical and data acquisition systems for proper
operation.
Tech. - 1 day; A.E. - 1 day; Scientist - 1 day
104
-------
Dec. 3
Start 2-hour zero and calibration tests.
Same man loading as Dec. 2 .
Dec. 4
Complete 2-hour zero and calibration tests.
Tech - 1 day; A.E. - 1 day
Dec. 5
Run response tests.
Same man loading as Dec. 4
Dec. 6
Continue 30-day run.
Same man loading as Dec. 5
Dec. 9-12
Run 2 Method 6's correlated with all 4 systems plus 2 Method 4's
each day.
Tech. - 4 days; A.E. - 4 days
Dec. 13
Run 1 Method 6 correlated with all 4 systems plus 1 Method 4.
Reduce week's data.
Tech. - 1 day; Scientist - 1 day; A.E. - 1 day
Dec. 16-19
Same as Dec. 9-12.
Tech - 4 days; A.E. - 4 days
Dec. 20
Run 1 Method 6 correlated with all 4 systems plus 1 Method 4.
Tech. - 1 day; Scientist - 1 day; A.E. - 1 day
105
-------
Dec. 23-24
End of 30 day run. Shut down all systems.
Tech. - 2 days; A.E. - 2 days
Jan 6-10
Remove equipment and restore site.
Tech. - 5 days; A.E. - 5 days
106
-------
December 6, 1974
(Revised and abridged June 25, 1975,
for inclusion in Final Report)
REVISED WORK PLAN FOR FIELD TESTS AT POWER PLANT
1.0 FACTORS NECESSITATING REVISION
Because of numerous equipment failures in the initial two weeks of effort
at the Power Plant Test Site, it was necessary to revise the Work Plan to
ensure completion of the evaluation within available funding. Failures of
both the digital and analog data acquisition systems have occurred. Both
are now operational, but it was necessary to adhere to the formal test
schedule as long as either system was operational. Effort toward
development of the computer program for computer reduction of punched
paper-tape data was terminated because the equipment reliability problem
made the cost unjustifiable. Punched tape was generated for possible
future use to the extent that it entailed no cost impact. Emphasis was
placed upon the acquisition of data for manual reduction, and upon adher-
ence to the revised Detailed Daily Schedule, Further equipment failures
necessitated termination of testing of the related system, rather than
delaying the schedule.
2.0 SPECIFIC REVISIONS
The following items are deleted from the original Work Plan:
• Last paragraph of Section 7.2
• Section 8.2
• Section 9.0
Section 9.0 of the original Work Plan is replaced by the Detailed Daily
Schedule of this revision to the Work Plan for the Power Plant Test Site.
107
-------
REVISED FIELD TEST PLAN FOR POWER PLANT
DETAILED DAILY SCHEDULE
Date Time Step
Day (Dec) (Hr;Min) No. Activity
0 8 NA NA Run speed of response tests on all
instruments.
Run three Method 6 tests on span gas.
1 9 0:00 1 Run.zero and span calibrations on all
instruments. Then set all sample handling
systems to "SAMPLE" mode and data logger
to internal timer set for one scan every
two minutes.
0:15 2 Acquire two manual samples for Method 6
analyses, collected in sequence.
2:00 3 Repeat step 1
2:15 4 Repeat step 2
4:00 5 Repeat step 1
4:15 6 Repeat step 2
6:00 7 Repeat step 1
6:15 8 Repeat step 2
8:00 9 Repeat step 1, except set data logger for
internal one-hour scan (or to external
timer set for one scan every 48 minutes
if system remains operational).
8:15 10 Acquire a manual sample for Method 4
analysis.
2 10 0:00 1 Repeat step 9 of Day 1.
2:00 2 Repeat step 1
4:00 3 Repeat step 1
6:00 4 Repeat step 1
8:00 5 Repeat step 1
8:15 6 Repeat step 10 of Day 1.
All Day 7 Analyze all manual samples taken on Day 1.
3 11 All Day NA Repeat all steps of Day 1.
4 12 All Day NA Repeat all steps of Day 2.
108
-------
REVISED FIELD TEST PLAN FOR POWER PLANT (Continued)
DETAILED DAILY SCHEDULE
Day
5
Date
(Dec)
13
Time
(Hr:Min)
0:00
0:15
2:00
2:15
Remainder of
Step
No.
1
2
3
4
5
Day
6 14 0:00
7 15 0:00
8 thru 14 16-22 NA
1
1
NA
Activity
Repeat step 1 of Day 1.
Repeat step 2 of Day 1.
Repeat step 9 of Day 1.
Repeat step 10 of Day 1.
Analyze manual samples taken this day.
Maintain/repair equipment as time and
material permit.
Repeat step 9 of Day 1.
Repeat step 9 of Day 1.
Repeat all steps of Day 1 through Day 7,
Instrument Systems to be Utilized
1. Dynasciences Model SS-330 with the conventional (refrigerated) sampling
system.
2. Dynasciences Model SS-310 and the Meloy Analyzer with the Meloy
Dyfusatron Sampling System.
3. Beckman Model 865 NDIR with the reflux sampling Probe and Permeation
Dryer Sampling System.
NOTE
Manual 'samples have been increased from 27
(18 samples correlated) to 36 samples all
correlated.
109
-------
March 28, 1975
(Revised and abridged June 25, 1975,
for inclusion in Final Report)
REVISED FIELD TEST PLAN FOR CLAUS PLANT
(Supercedes Original Plan Entirely)
1.0 SITE DESCRIPTION
The Glaus Unit that will be used for these tests has a typical, unmodified
incinerator. The gases entering the incinerator contain approximately 0.2 to
0.5 percent hydrogen sulfide and 0.1 to 0.25 percent sulfur dioxide plus total
S2 + S + S, up to 1 percent, at a termperature of about 148.8° F (300° F) and
atmospheric pressure. The exit gases are at about 537.7° C (1000° F) and 248
pascals (1 inch water) vacuum.
2.0 SAMPLING-INTERFACE-MONITORING SYSTEMS
2.1 Incinerator Inlet
The following system will be used on the incinerator inlet stream:
Houston Atlas H~S Analyzer with sample dilution unit for
0-1% range and a conventional sampling system designed
specifically for this stream.
2.2 Incinerator Outlet
The following three systems will be used on the incinerator outlet:
• Beckman Reflux Sampling Probe and Permeation Dryer with
Beckman NDIR SO Analyzer.
• Meloy Total Sulfur Analyzer and the Dynasciences Model
SS-310 with a Meloy Dyfusatron Sample Diluter.
110
-------
2.3 Analytical System Operation and Maintenance
The system identified in Paragraph 2.1 will be operated continuously for
ten days on the incinerator inlet and the systems identified in Para-
graph 2.2 will be operated continuously for ten days on the incinerator
outlet. During the operating period, all systems will be zeroed and
spanned on a 24-hour basis. Refer to the Detailed Daily Schedule of
this Work Plan for details.
Reasonable, normal maintenance will be performed. However, a major break-
down in any of the analytical systems, requiring extensive field or fac-
tory service or repair will require cessation of testing of that particular
system.
3.0 DATA TO BE ACQUIRED
3.1 Relative Accuracy after Incinerator (S02)
Thirty-seven manual samples will be taken for analysis by the wet chemi^
cal Method 6. Simultaneous output signal recordings will be obtained with
all analytical systems during all of the manual samplings. The zero and
span readings will be checked on all analytical systems before taking
manual samples.
3.2 Twenty-Four-Hour Zero and Span Drift
Readings will be obtained on the analytical systems for zero and calibra-
tion gases at 24-hour intervals throughout the test period.
3.3 Two-Hour Zero and Span Drift
Readings will be obtained on the analytical systems for zero and calibra-
tion gases at two-hour intervals during the periods of manual sampling.
3.4 Operational Period
Detailed maintenance logs will be kept for each analytical system.
Ill
-------
4.0 DATA ANALYSIS AND REPORTING
4.1 Relative Accuracy
This calculation is as described in the "Work Plan for Field Tests at
Power Plant."
4.2 Twenty-Four-Hour Zero and Span Drifts
These calculations will also be as described in the "Work Plan for Field
Tests at Power Plant."
4,3 Two-Hour Zero and Span Drifts
These calculations will also be as described in the "Work Plan for Field
Tests at Power Plant."
5.0 CALCULATIONS
Calculation of mean values and 95% confidence intervals will be per^
formed as described in the "Work Plan for Field Tests at Power Plant."
6,0 DATA ACQUISITION SYSTEM
Data acquisition will be performed by the same system as used at the
Power Plant.
7.0 DATA ANALYSIS
Data analysis will be performed in the same manner as at the Power Plant,
The Final Report will contain either or both data analyzed by manual
techniques and data analyzed by computer techniques, at the option of
the Beckman Program Manager,
112
-------
REVISED FIELD TEST PLAN FOR GLAUS PLANT
DETAILED DAILY SCHEDULE
Day
0
2-5
6
7
8
Time Period
0800-0830
NA
0800-0830
0845-0900
1000-1030
1030-1045
NA
1200-1230
1400-1430
1445-1500
NA
1600-1630
1630-1645
1700
NA
1800-1830
1830-1845
1845-1900
NA
2000-2030
0800-0830
0800-0830
0800-0830
Activity
Calibrate all instruments.
Set data logger to one-hour cycle.
Run Method 4 on incinerator outlet.
Run zero and span gases on all instruments.
Set data logger to one-minute cycle.
Take manual sample for Method 6 on incinerator outlet,
Set data logger to one-hour cycle.
Analyze samples by Method 6.
Run zero and span gases.
Set data logger to one minute.
Take sample for Method 6.
Set data logger to one hour.
Analyze samples by Method 6.
Run zero and span gases.
Lunch
Run zero and span gases.
Set data logger to one minute.
Take sample for Method 6.
Set data logger to one hour.
Analyze samples by Method 6.
Run zero and span gases.
Set data logger to one minute.
Take sample for Method 6.
Set data logger to one hour.
Analyze samples by Method 6.
Run zero and span gases.
Set data logger to one minute.
Take sample for Method 6.
Set data logger to one hour.
Analyze samples by Method 6.
Run zero and span gases.
Run Method 4.
Identical to Day 1.
Run zero and span gases.
Run zero and span gases.
Run zero and span gases.
Set data logger to one minute.
113
-------
REVISED FIELD TEST PLAN FOR GLAUS PLANT (Continued)
DETAILED DAILY SCHEDULE
Day Time Period Activity
0830-0845 Take sample for Method 6 on outlet.
Set data logger to one hour.
NA Analyze sample.
1000-1030 Run zero and span gases.
Set data logger to one minute.
1030-1045 Take sample for Method 6.
Set data logger to one hour.
NA Analyze sample.
1200-1230 Run zero and span gases.
Run Method 4.
Lunch.
1400-1430 Run zero and span gases.
Set data logger to one minute.
1430-1445 Take sample for Method 6.
Set data logger to one hour.
NA Analyze sample.
1600-1630 Run zero and span gases.
Set data logger to one minute.
1630-1645 Take sample for Method 6.
Set data logger to one hour.
NA Analyze sample.
1800-1830 Run zero and span gases.
Run Method 4.
9,10 Identical to Day 8.
11 0800-0830 Run zero and span gases.
The plan described above will produce the following:
• Eleven days continuous operation
• Eleven 24-hour zero and calibration drift checks
• Forty-five 2-hour zero and calibration drift checks
• Thirty-seven Method 6 analyses on incinerator outlet
• Twelve Method 4 analyses on incinerator outlet
114
-------
APPENDIX B
EPA DOCUMENTS USED IN DESIGN OF
FIELD WORK PLANS
115
-------
CONTENTS
Page
1. Performance Specifications for Stationary-Source
Monitoring System for Gases and Visible Emissions,
EPA-650/2-74-013, January 1974, Appendix E, pp 49-60 . . . 113
2. Federal Register, Environmental Protection Agency,
Stationary Sources, "Proposed Emission Monitoring
and Performance Testing Requirement" Volume 39,
No. 177, Part II, pp 32864-32871 126
116
-------
EPA-650/2-74-013
PERFORMANCE SPECIFICATIONS
FOR
STATIONARY-SOURCE
MONITORING SYSTEMS
FOR GASES AND VISIBLE EMISSIONS
by
John S. Nader, Fredric Jaye,
and William Conner
Chemistry and Physics Laboratory
Program Element No. 1AA010
U.S. ENVIRONMENTAL PROTECTION AGENCY
National Environmental Research Center
Research Triangle Park, N. C. 27711
January 1974
117
-------
APPENDIX E.
EXAMPLE:
PERFORMANCE SPECIFICATIONS AND SPECIFICATION
TEST PROCEDURES FOR MONITORS OF POLLUTANT
GAS EMISSIONS FROM STATIONARY SOURCES
Performance specifications for continuous measurement systems
for pollutant gases are given in terms of critical operating para-
meters. Test procedures are given to test the capability of the
measurement systems to conform to the performance specifications.
1. Principle and Applicability
1.1 Principle - Pollutant gases are sampled continuously
in the stack emissions,and the gas concentration is
analyzed continuously as a function of time. Sam-
pling may include either the extractive or non-ex-
tractive (in-situ) approach.
1.2 Applicability - The performance specifications are
given for continuous pollutant gas measurement systems
applied to specific source-pollutant combinations. The
following discussion is addressed to SC^ and NOX emissions
from coal-burning power plants. Instrument system should
be capable of operation within performance specifications
at particulate loadings and in a temperature range corre-
sponding to those of the environment of the installation.
2. Apparatus
2.1 Calibration Gas Mixture - Mixture of a known concentra-
tion of the pollutant gas in oxygen-free nitrogen.
118
-------
Nominal concentrations of 30 percent, 60 percent, and
90 percent of span are recommended. It is strongly
recommended that the gas mixture be analyzed by a
reference method prior to use.
2.2 Zero Gas - A gas containing no more than 1 ppm of
the pollutant gas.
2.3 Equipment for measurement of pollutant gas concentration
using the reference method.
2.4 Strip Chart Recorder - Analog strip chart recorder, in-
put voltage range compatible with analyzer system out-
put, full scale (per travel) in two seconds or less.
2.5 Continuous measurement system for pollutant gas.
3. Definitions
3.1 Measurement System - The total equipment required for
the determination of a pollutant gas concentration in
a given source effluent. The system consists of three
major subsystems:
1. Sampling Interface - That portion of the measure-
ment system that performs one or more of the fol-
lowing operations: delineation, acquisition, trans-
portation, and conditioning of a sample of the source
effluent or protection of the analyzer from the
hostile aspects of the sample or source environment.
2. Analyzer - That portion of the system which senses
the pollutant gas and generates a signal output that
is a function of the pollutant concentration.
119
-------
3. Data Presentation - That portion of the measure-
ment system that provides a display of the output
signal in terms of concentration units.
3.2 Span - The value of pollutant concentration at which the
measurement system is set to produce the maximum data
display output. For the purposes of this method, the
span shall be set at a pollutant gas concentration of
1.5 times the emission standard or the pollutant gas
concentration of interest.
3.3 Accuracy (Relative) - The degree of correctness with
which the measurement system yields the value of gas
concentration of a sample relative to the value given
by a defined reference method. This accuracy is ex-
pressed in terms of error which is the difference be-
tween the paired concentration measurements. The error
is expressed as a percentage of the reference mean
value.
3.4 Calibration Error - The difference between the pollutant
concentration indicated by the measurement system and
the known concentration of the test gas mixture.
3.5 Zero Drift - The change in measurement system output
over a stated period of time of normal continuous
operation when the pollutant concentration at the time
of the measurements is zero.
3.6 Calibration Drift - The change in measurement system
output over a stated period of time of normal contin-
uous operation when the pollutant concentration at
120
-------
the time of the measurements is the same known up-
scale value.
3.7 Repeatability - A measure of the measurement system's
ability to give the same output reading(s) upon re-
peated measurements of the same pollutant concentra-
tion (s) .
3.8 Response Time - The time interval from a step change
in pollutant concentration at the input to the measure-
ment system to the time at which 95 percent of the
corresponding final value is reached as display on
the measurement system data presentation device.
3.9 Operational Period - A minimum peridd of time over
which a measurement system is expected to operate
within certain performance specifications without
unscheduled maintenance, repair or adjustment.
4. Measurement System Performance Specifications
The following performance specifications shall be met
in order that a measurement system shall be considered
acceptable under this method.
Parameter
a. Accuracy (relative)
b. Calibration Error
c. Zero Drift (2 hour)
d. Zero Drift (24 hour)
Specification
<_ 20% of mean reterence value
<_ 5% of each test gas value*
<_ 2% of emission standard*
< 4% of emission standard*
*Absolute mean value + percent confidence interval.
121
-------
Parameter Specification
e. Calibration Drift (2 hour) < 2% of emission standard*
f. Calibration Drift (24 hour) 15% of emission standard*
g. Response Time 10 minutes (maximum)
h. Operational Period 168 hours
*Absolute mean value + 95% confidence interval.
5. Specification Test Procedures
5.1 Calibration Error and Repeatability Test - Set up and
calibrate the complete measurement system according to
the manufacturer's written instructions. Record the
readings of calibration gas concentrations of
approximately 30, 60, and 90 percent of span. Make a
series of five nonconsecutive readings at each con-
centration (Example: 30 percent, 90 percent, 60 percent
90 percent, 30 percent, 90 percent, 60 percent, etc.)
Convert the measurement system output readings to ppm.
5.2 Field test for accuracy (relative), zero drift, cal-
ibration drift, and operation period<0
5.2.1 Set up and operate the measurement system in
accordance with the manufacturer's written in-
structions and drawings. Operate the system
for an initial 168 hour conditioning period.
During this period the system should measure
the pollutant gas content of the effluent in a
normal operational manner.
5.2.2 After completion of this conditioning period, the formal
168 hour performance and operation test period shall
begin. The system shall be monitoring the source efflu-
ent at all times when not being zeroed, calibrated, or
122
-------
backpurged. During this 168 hour test period, make a
minimum of nine (9) pollutant gas concentration measure-
ments using the reference method at intervals of not less
than 1 hour. For S02> each of the nine tests shall con-
sist of a Method 6 S02 concentration measurement, plus
the instrumental measurement. For NOX, each of the nine
tests shall consist of a set of three Method 7 NOX con-
centration measurements, plus the instrumental measure-
ment. In each set of three Method 7 measurements, three
samples should be taken concurrently or within a 3-minute
interval. The sampling location for the reference method
shall be as prescribed. The sampling location for the
monitoring method shall 'be as close to the location of
the reference method sampling point as conditions will
permit to demonstrate the specified accuracy. The re-
ference method data shall be compared with simultaneously
collected monitoring data for calculation of accuracy.
Before and after each reference method test, record the
values given by both zero and calibration concentrations.
Record the values given by zero and calibration concentra-
tions at two hour intervals until 15 sets of data are ob-
tained. This two-hour period need not be consecutive but
may not overlap. Zero and calibration corrections and
adjustments are allowed only at 24 hour intervals or at
such shorter intervals as the manufacturer's written
instructions specify. Automatic corrections made by
the measurement system without operator intervention or
123
-------
initiation are allowable at any time. During the entire
test period, record the values given by zero and cali-
bration pollutant gas concentrations before and after
adjustment at 24 hour intervals. Calibration checks
are made with one test gas concentration between 70
and 90 percent of span.
5.3 Field Test for Reponse Time
5.3.1 This test shall be accomplished using the entire
measurement system as installed including sample
transport lines if used. Flow rates, line diameters,
pumping rates, pressures, etc., shall be at the nominal
values for normal operation as specified in the manu-
facturer's written instructions. In the case of cyclic
analyzers, the response time test shall include one
cycle. This test shall be repeated for each sampling
point of multi-sampling point systems.
5.3.2 Introduce a zero concentration of pollutant gas into the
measurement system sampling interface or as close to the
sampling interface as possible. When the system output
reading has stabilized, switch quickly to a known concen-
tration pollutant gas at 70 to 90 percent of span. Re-
cord the time from concentration switching to final stable
response. After the system response has stabilized at the
upper level, switch quickly to a zero concentration of pol-
lutant gas. Record the time from concentration switching to
124
-------
final stable response; perform this test sequence three (3)
times. The strip chart recorder charts or copies of them
from this test should be included in the data submission
with time scales and up and down scale values clearly
marked.
6. Calculation, Data Analysis and Reporting
6.1 Procedure for Determination of Mean Values and Confidence Intervals
6.1.1 The mean value of a data set is calculated according to
equation E-l.
. n
I = - V x.
n . Equation E-l
where x. = individual values Z = sum of the indi-
_ vidual values
x = mean value
n = number of data points
6.1.2 The 95% confidence interval (two-sided) is calculated according
to equation E-2.
Equation E-2
where Ex. = sum of all data points
i/n~ = square root of the number of data points
tj ,_ = t g7e. for n samples from a table of percentages
of the t distribution.
C.I.gs = 95% confidence interval estimate of the average mean
value.
V./2
^T
\
\
n(Zx2.) - (Z
v2
n(n-l)
125
-------
Typical Values for *1 - «/2
n
2
3
4
5
6
t.975
12.706
4.303
3.182
2.776
2.571
n
7
8
9
10
11
t.975
2.447
2.365
2.306
2.262
2.228
n
12
13
14
15
16
t.975
2.201
2.179
2 .160
2.145
2.131
The values in this table are already corrected for n-1 degrees of freedom.
Use n equal to the number of samples as data points.
6.2 Data Analysis and Reporting
6.2.1 Accuracy (relative) - This calculation uses the reference method
test data and the measurement system concentrations recorded
at the times the reference method tests were run. Subtract
the reference method test concentration from the measurement
system concentration.* Repeat for all nine test pairs. Using
this data, compute the mean difference and the 95% confidence
interval using Equations E-l and E-2. Report the sum of the
absolute mean value and the 95% confidence interval as a per-
centage of the mean reference value.
6.2.2 Calibration Error - Using the data from Section 5.1 subtract
the known value from the value shown by the measurement
system for each of the 5 readings at each span test concen-
tration. Calculate the mean of these differences values
and the 95% confidence interval according to Equations E-l
and E-2. Report the sum of the absolute mean value and the 95%
confidence interval as a percentage of test gas concentration.
*For S02, subtract the Method 6 values from the corresponding instrumental
value. For NOX, subtract the mean value of the set of three Method 7
values from the corresponding instrumental values.
126
-------
6.2.3 Zero Drift (2 Hour) - Using the zero concentration values
measured during the field test, calculate the mean value and
the confidence interval using Equations E-l and E-2. Report
the sum of the absolute mean value and the confidence inter-
val as a percentage of the emission standard.
6.2.4 Zero Drift - (24 Hour) - Using the zero concentration values
measured every 24 hours during the field test, calculate the
differences between the zero point after zero adjustment and the
zero value 24 hours later just prior to zero adjustment. Calibrate
the mean value of these points and the confidence interval
using Equations E-l and E-2. Report the sum of the absolute
mean value and confidence interval as a percentage of the
emission standard.
6.2.5 Calibration Drift (2 Hour) - Using the calibration values
obtained at two-hour intervals during the field test, cal-
culate the differences between the readings and the test
gas value. These values should be corrected for the cor-
responding zero drift during that two-hour period. Calcu-
late the mean and confidence interval of these corrected
values using Equations E-l and E-2. Report the sum of the
absolute mean value and confidence interval as a percentage
of the emission standard.
6.2.6 Calibration Drift (24 Hour) - Using the calibration values
measured every 24 hours during the field test, calculate
the differences between the calibration concentration reading
after zero and calibration adjustment and the calibration
127
-------
concentration reading 24 hours later after zero adjustment but
before calibration adjustment. Calculate the mean value of
these differences and the confidence interval using Equations
E-l and E-2. Report sum of the absolute mean value and confi-
dence interval as a percentage of the emission standard.
6.2.7 Response Time - Using the charts from Section 5.3 calculate
the time interval from concentration switching to 95% of the
final stable value for all up scale and down scale tests.
Report the mean of the three up scale test times and the mean
of the three down scale test times. The two average times
should not differ by more than 15% of the slower time. Report
the slower time as the system response time.
6.2.8 Operational Period - During the 168 hour performance and
operational test period, the measurement system shall not
require any corrective maintenance or repair or replacement
or adjustment other than that clearly specified as required
in the operation and maintenance manuals as routine and
expected during a one-week period. If the measurement
system operates within the specified performance parameters
and .does not require corrective maintenance, repair, replace-
ment or adjustment other than specified above, during the 168
hour test period, the operational period test will be success-
fully concluded. Failure of the measurement to meet this
requirement shall call for repetition of the 168 hour test
period. Portions of the test which were satisfactorily
completed need not be repeated. Failure to meet any perfor-
mance specifications shall call for a repetition of the one
128
-------
week performance test period and that portion of the testing
which is related to the failed specification. All maintenance
and adjustments required shall be recorded. Output readings
shall be recorded before and after all adjustments.
7. Sunplemental References
Experimental Statistics, National Bureau of Standards, Handbook 91,
1963, P. 3-31, Paragraph 3-3.1.4.
129
-------
WEDNESDAY, SEPTEMBER 11, 1974
WASHINGTON, D.C.
Volume 39 • Number 177
PART II
ENVIRONMENTAL
PROTECTION
AGENCY
STATIONARY SOURCES
Proposed Emission Monitoring and
Performance Testing Requirements
No. 177—Ft. II- 1
130
-------
PROPOSED RULES
Date of Test
Soan Filter
Analyzer Span Setting. , ._ , ,
1
Upscale 2
3
XOoadty
% Ooadty
seconds
seconds '
seconds
Downscale
1
2
3
Average response,
. seconds
. seconds
seconds
seconds
. Figure 1-2. Response Time Test
Zero Setting.
Span Setting .
. (See paragraph 8.2.1)
Date Zero Reading Span Reading Calibration
and (Before cleaning Zero Drift (After cleaning and zero adjustment Drift
Tine end adjustment) (aZero) but before span adjustment) (aSpan)
• Zero Drift • IHean Zero Drift* t CI (Zero)
« Emission Standard! x 100 • .
Calibration Drift • [Mean Span Drift* * CI (Span).
* Emission Standard] x 100 •.
Absolute value
Figure 1-3. Zero and Calibration Drift Test
PERFORMANCE SPECIFICATIOK 2—PERFORMANCE
SPECIFICATIONS AND SPECIFICATION TEST PRO-
CEDURES FOB MONITORS Or SO, AND NOZ FROM
STATIONARY SOURCES
1. Principle and Applicability.
1.1 Principle. Oases are continuously sam-
pled In the stack emissions and analyzed for
either sulfur dioxide or oxides or nitrogen
by a continuously operating emission meas-
urement system. Sampling may Include either
the extractive or non-extractive (
-------
PROPOSED RULES
32865
TABU: 2-1.—PIBTOKMANCX SPECIFICATIONS
Parameter SpeeifeaHon
1 Accuracy* «£20% of reference mean value.
t. calibration error- <**>% « each (60%. 80%) calibration gas
mixture value.
3. Zero drift (3 hours) • —- ^3% of emission standard.
4. Zero drift (24-hour)'•_ —— =54% of emission standard.
6. Calibration drift (2 hours) • — «3% of emission standard.
6. Calibration drift (24-hour) • — ^8% of emission standard.
7. Response time —• 16 minutes maximum.
8. Operational period 168 hours minimum,
•Expressed as sum of absolute mean value plus 96 percent confidence Interval of a series
of tests.
6.1.2 The 95 percent confidence Interval
(two sided) Is calculated according to equa-
tion 2-2.
5. Performance Specification Teat Pro-
cedures. The following test procedures shall
be used to determine conformance with the
requirements of paragraph 4:
5.1 Calibration test.
5.1.1 Analyze each calibration gas mixture
(60%, 90%) using reference methods 6 for
sulfur dioxide and 7 for oxides of nitrogen,
and record the results on the example sheet
shown In Figure 2-1. This step may be omit-
ted for non-extractive monitors where dy-
namic calibration gas mixtures are not used
(See 6.1.2).
6.1.2 Bet up and calibrate the complete
measurement system according to the man-
ufacturer's written instructions. This may
be accomplished either In the laboratory or
In the field. Make a series of five non-consec-
utive readings with span gas mixtures alter-
nately at each concentration (e.g.. 60%, 90%,
50%, 90%, 60%, etc.). For non-extractive
measurement systems, this test may be per-
formed using procedures and two or more
calibration gas concentrations differing by
a factor of two or more, certified by the man-
ufacturer. Convert the measurement system
output readings to ppm and record the re-
sults on the example sheet shown In Figure
2-2.
6.2 Field Test for Accuracy (Relative), Zero
Drift, Calibration, and Drift—"Install and op-
erate the measurement system In accordance
with the manufacturer's written Instructions
and drawings as follows:
6.2.1 Conditioning Period—Offset the zero
setting at least 10 percent of span so that
negative zero drift can be quantified. Operate
the system for an Initial 168-hour condition-
ing period In a normal operational manner.
6.2.2 Operational Test Period—Operate
the system for an additional 168-hour
period. The system shall monitor the source
effluent at all times except when being
zeroed, calibrated, or backpurged. For meas-
urement systems employing extractive
sampling, It Is recommended that the meas-
urement system and the probe tips be
placed adjacent to each other In the duct.
Record the reference methods test data and
measurement system concentrations on the
example data sheet shown In Figure 2-3 for
the -tests given as follows:
5.2.2.1 NO, Monitoring Systems. Make
twenty seven NO, concentration measure-
ments using the applicable reference
method. No more than three measurements
shall be performed in any one hour, and
any set of three measurements shall be
performed concurrently or within a 3-
minute Interval and the results averaged.
5.2.2.2 SO, Monitoring Systems. Make
nine SO* concentration measurements using
the applicable reference method. No more
than one measurement shall be performed
In any one hour.
5.2.3 Field Test for Zero Drift and Cali-
bration Drift. Determine the values given by
zero and span gas pollutant concentrations
at 2-hour intervals until 15 sets of data are
.obtained. Alternatively, for non-extractive
measurement systems, determine the values
given by an electrically or mechanically pro-
duced zero condition and by Inserting a
certified calibration gas concentration
equivalent to not less than 300 ppm Into the
measurement system. Record these readings
on the example sheet shown In Figure 2-4.
These 2-hour periods need not be consecu-
tive but may not overlap. The zero and span
determinations to be made under this para-
graph may be made concurrent with the
tests under 6.2.2. Zero and calibration cor-
rections and adjustments are allowed only
at 24-hour Intervals or at such shorter In-
tervals as the manufacturer's written In-
structions specify. Automatic corrections
made by the measurement system without
operator Intervention or Initiation are al-
lowable at any time. During the entire 168-
hour operational test period, record the
values given by zero and span gas pollutant
concentrations before and after adjustment
at 24-hour Intervals In the example sheet
shown In Figure 2-5.
6.3 Field Test for Response Time.
6.31 This test shall be accomplished using
the entire measurement system as Installed,
including sample transport lines if used.
Flow rates, line diameters, pumping rates,
pressures (do not allow the pressurized cali-
bration gas to change the normal operating
pressure in the sample line), etc.. Shall be
at the nominal values for normal operation
as specified In the manufacturer's written
Instructions. If the analyzer Is used to
sample more than one pollutant source
(stack). this test shall be repeated for each
sampling point.
5.3.2 Introduce zero gas Into the measure-
ment system sampling Interface or as close
to the sampling Interface as possible. When
the system output reading has stabilized,
switch quickly to a known concentration of
pollutant gas at 70 to 90 percent of span.
Record the time from concentration switch-
ing to final stable response. After the sys-
tem response has stabilized at the upper
level, switch quickly to a zero concentration
of pollutant gas. Record the tune from con-
centration switching to final stable response.
Alternatively, for nonextractlve monitors,
the highest available calibration gas con-
centration shall be switched Into and. out
of the sample path and response times
recorded. Perform this test sequence three
(3) times. For each test record the results
on the example sheet shown In Figure 2-6.
6. Calculations, Data Analysis and Report-
ing.
6.1' Procedure for determination of mean
values and confidence Intervals.
6.1.1 The mean value of a data set Is cal-
culated according to equation 2-1.
-_i
~n
Equation 2-1
where:
x, ^individual values,
£=sum of the Individual values,
xWmean value, and
n = number of data points.
Equation 2-2
where:
2X< = sum of all data points,
«.975='J-a/2, and
C.I.^85 percent confidence Interval esti-
mate of the average mean value,
VALTTES TOR '.975
n '.975
2 ................................. 12.708
3 ...... . ....... . ................. 4.303
4 ................................. 3.182
6 ..... . ........................... 2.778
6 .................... - ............ 2.671
7 _______ .......................... 2.447
8 ... .............................. 2.368
9 ......... . ........ . ............ .. 2.306
10 ______ ....... ____________________ 2.282
11 ________ ................. . ....... 2.228
12 ____________________________ ..... 2.201
13 „ ...... _______ ...... . ...... 2.179
14 ... ...... ___________ .......... »-- 2.180
16 ................................. 2.146
16 _________ ............... - ........ 2.131
The values In this table are already cor-
rected for n-l degrees of freedom. Dee n equal
to the number of samples as data points.
6.2 Data Analysis and Reporting.
8.2.1 Accuracy (Relative) For each of the
nine reference method testing periods, de-
termine the average pollutant concentration
reported by the continuous measurement
system. These average concentrations shall be
determined from the measurement system
data recorded under 6.2.2 by Integrating the
pollutant concentrations over each of the
time Intervals concurrent with each refer-
ence method test, then dividing by the
cumulative time of each applicable reference
method testing period. Before proceeding to
the next step, determine the bases (wet or
dry) of the measurement system data and
reference method test data concentrations.
If the bases are not consistent, then a mois-
ture correction shall be applied to either the
reference method concentrations or the
measurement system concentrations as Is
appropriate. The correction factor shall be
determined by moisture tests concurrent
with the reference method testing periods.
The moisture test method and the correc-
tion procedure employed shall be reported.
For each of the nine test runs, subtract the
respective reference method test concentra-
tions (use average of each set of 3 measure-
ments for NO,) from the continuous moni-
toring system average concentrations. Using
these data, compute the mean difference and
the 95 percent confidence Interval using
equations 2-1 and 2-2. Accuracy Is reported
as the sum of the absolute value of the mean
difference and the 05 percent confidence In-
terval expressed as a percentage of the mean
reference method value. Use the example
sheet shown in Figure 2-3.
6.2.2 Calibration Error — Using the data
from paragraph 6.1, subtract the measured
pollutant concentration determined under
paragraph 6.1.1 (Figure 2-1) from the value
shown by the measurement system for each
of the 5 readings at each concentration meas-
ured under B.I 3 (Figure 2-2). Calculate the
mean of these difference values and the 95
percent confidence Intervals according to
equations 2-1 and 2-2. The calibration error
is reported as the sum of the absolute value
of the mean difference and the 95 percent
confidence Interval as a percentage of each
respective calibration gas concentration. Use
example sheet shown In Figure 2-2.
6.2.3 Zero Drift (2-hour) — Using the zero
concentration values measured each 2 hours
FEDERAL REGISTER, VOL 39, NO. 177—WEDNESDAY, SEPTEMBER 11, 1974
132
-------
32866
PROPOSED RULES
during the field test, calculate the differences
between consecutive 2-hour readings express-
ed in ppm. Calculate the mean difference
and the confidence Interval using equations
3-1 and 2-2. Report the zero drift as the
sum at the absolute mean value and the
confidence Interval as a percentage of the
emission standard. Use example sheet shown
in Figure 2-4.
6.2.4 Zero Drift (24-hour)—Using the zero
concentration values measured every 24
hours during the field test, calculate the dif-
ferences between the zero point after zero
adjustment and the zero value 24 hours later
Just prior to zero adjustment. Calculate the
mean value of these points and the confi-
dence Interval using equations 2-1 and 2-2.
Beport the zero drift (the sum of the abso-
lute mean and confidence Interval) as a per-
centage of the emission standard. Use exam-
ple sheet shown In Figure 2-5.
6.2.5 Calibration Drift (2-hour)—Using
tlie calibration values obtained at 2-hour In-
tervals during the field test, calculate the
differences between consecutive 2-hour read-
Ings expressed as ppm. These values should
be corrected for the corresponding zero drift
during that 2-hour period. Calculate the
mean and confidence Interval of these cor-
rected difference values using equations 2-1
and 2-2. Do not use the differences between
non-consecutive readings. Beport the cali-
bration drift as the sum of the absolute mean
and confidence Interval as a percentage of
the emission standard. Use the example sheet
shown In Figure 2-4.
6.2.6 Calibration Drift (24-hour)—Using
the calibration values measured every 24
hours during the field test, calculate the
differences between the calibration concen-
tration reading after zero and calibration ad-
justment and the calibration concentration
reading 24 hours later after zero adjustment
but before calibration adjustment. Calculate
the mean value of these differences and the
confidence Interval using equations 2-1 and
2-2. Report the sum at the absolute mean
and confidence Interval « a percentage of
the emission standard. Use the example sheet
shown In Figure 3-5.
6.2.7 Response Time—Using the charts
from paragraph 53, calculate the time Inter-
val from concentration switching to 96 per-
cent to the final etabto value for aQ Trpscal*>
and downseale tests. Beport tbe mean of th»
three upscale teat tiznea and the mean of
the three downseale test tlmee. The two av-
erage; time* should not differ by more than
It percent at the dower tune. Beport tb*
slower tune a* the system response Ume.
Use the example sheet shown In Figure 2-6.
6.2.8 Operational Test Period—During the
168-hour performance and operational test
period, the measurement system shall not
require any corrective maintenance or repair
or replacement or adjustment other than
that clearly specified as required In the op-
eration and maintenance manuals as routine
and expected during a 1-week period. If the
measurement system operates within the,
specified performance parameters and does
not require corrective maintenance, repair,
replacement or adjustment other than as
specified above during the 168-hour test pe-
riod, the operational period will be success-
fully concluded. Failure of the measurement
to meet this requirement shall call for a
repetition of the 168-hour test period. Por-
tions of the test which were satisfactorily
completed need not be repeated. Failure to
meet any performance specifications shall
call for a repetition of the 1-week perform-
ance test period and that portion of the
testing which is related to the failed speci-
fication. All maintenance and adjustments
required shall be recorded. Output readings
shall be recorded before and after all adjust-
ments. '
7. Reference!.
7.1 Monitoring Instrumentation for the
Measurement of Sulfur Dioxide in Station.
ary Source Emissions, Environmental Protec-
tion Agency, Research Triangle Park, N.C,
February 1973.
7.2. Instrumentation for the Determina-
tion of Nitrogen Oxides Content of Station-
ary Source Emissions, Environmental Protec-
tion Agency, Research Triangle Park, N.C,
Volume 1, APTD-0847, October 1971; Volume
2, APTD-0942, January 1972.
13 Experimental Statistics, Department of
Commerce, Handbook 91, 1963, p. 3-S1, par-
agraphs 3-3.1.4.
7.4 Performance Specifications for Station-
ary-Source Monitoring Systems for Oases
and Visible Emissions, Environmental Protec-
tion Agency, Research Triangle Park, N.C.
EPA-650/2^74-013, January 1974.
Reference Method ttsed
Date IMd-Rartae'CAllbratlon Gas WxtQre
Sample 1 ppm
Sample 2 ppn
Sample 3 ppm
Average ppm
HKih-Range (swft) Calibration Gas Mixture
Sanplel
Sample 2
Sample 3 .
Avertgt _
_
.. ', , .. ..P"»
pn»
PBB
.pw»
Figure 2-1. Analysis of Calibration Gas Ntxtms
FEDEIAL UOISTHI, VOL n. NO, J7T—WEDNESDAY SEFIEMMt 11. OT4
133
-------
PROPOSED RULES
32867
O
Calibration Go Wxtura Mter CFrw Figure W)
C6fttMtmtai>
Jtean dlffwincs.
ConfJdenM Interval
Calibration ertw • -itey Wftraiq 2-* c.i. ^..-
Average calibration 64s concentration
Calibration gw ceneeittratlon • neasuranent systen
Absolute Wlw
Figure 2-2. Calibration Error Deteratnatlon
M1d Nigh
+ + ^
J %
Dai* Reference Method Sanies
«nd Test Saitfel S«?le 2 Satrie 3 Anna
Tine No. (DM) (DM) (DM) (DDK)
Analyzer
*'**" Averts** Difference)**
(DM) (DM)
1
?
3
4
5
6
7
8
9
Mew dlfferaice
ssi cufUum literal •
HEIH Reference Mttod Ma •
^ExpUtn ttthod iB«d to dettnriM ivenot,
Wfferenci • th* 1-Mv tvengi itm the raferam HtM tvtrigi.
«, "rtjr.
Ft gun 2-3. Accuraqr DaMnriMttoi
NO 177— pt.n
raMMt *retsn*, voi. n, HO. TTT— WWNEWAY, SEPRMMI n,
134
-------
32868 PROPOSED RULES
Bit* Zero CHlbntton
g! ,25* fi% & U*«>
Zero Drift • £nean Zero Drift* + CI (Zero)' Emission Suncurdj x 100 •
C«Hbr«t1cn Drift • [Mew Spin Prlft* + CI (Span) Emtiston SUndard] x 100 •
•Absolute Value.
Date Zero Span Calibration
and Zero Drift Reading Drift
Time Reading (AZer-p) (After zero adjustment) (ASpan)
Zero Drift • [Mean Zero Drift* + CI (Zero)
Emission Standardj x 100 •
Calibration Drift • [Mean Span Drift* * CI (Span
Emission Standardj x 100 = _____
•Absolute value
Figure 2-5, Zero and Calibration Drift (24-Hour)
FEDERAL REGISTER, VOL 99, NO. 177—WEDNESDAY, SEPTEMBER 11, 1974
135
-------
nOtOSED HUES
32869
Date of Test
Span Gas Concentration.
Analyzer Span Setting .
OBB
Upscale
1.
2.
3
sewn*
seconds
seconds
Average upscale response ,
seconds
Downs cale
seconds
seconds
seconds
Average downscale response .
System response time » slower time
seconds.
seconds
deviation from slowest time » average upscale minus average downscale x 100%
slower time
Figure 2-6. Response Time
PERFORMANCE SPECIFICATION 3—PERFORMANCE
SPECIFICATIONS AND SPECIFICATION TEST PRO-
CEDURES FOB MONITORS OF OXYGEN FBOM
STATIONARY SOURCES
1. Principle and Applicability.
1.1 Principle. Gases are continuously sam-
pled In the stack. Emissions are analyzed for
oxygen by a continuously operating measure-
ment system.
1.2 Applicability. This method Is appli-
cable to the Instrument systems specified
by subparta for continuously monitoring
oxygen. Specifications are given In terms of
performance. Test procedures are given to
determine the capability of the measurement
system to conform to the performance spec-
ifications prior to approving the systems In-
stalled by an affected facility. Sampling may
Include either the extractive or non-extrac-
tive (in situ) approach.
2. Apparatus.
2.1 Calibration Gas Mixtures. Mixture of
Snown concentrations of oxygen in nitrogen.
Nominal oxygen concentrations of 60 percent
and 80 percent ot the Instrument span set-
ting are required. The 90 percent gas mixture
Is to be used to set and to check the Instru-
ment span and Is referred to as span gas. If
the Instrument span setting Is higher than
21 percent O, (consistent with paragraph
3.2), ambient air may be used In place of the
90 percent calibration gas mixture.
2.2 Zero Gas. A gas containing less than 10
ppm of oxygen.
2.3 Chart Recorded. Analog chart re-
corder, Input voltage range compatible with
analyzer system output, fun scale (per travel)
In 2 seconds or less.
2.4 Continuous Measurement System for
Oxygen.
3. Definitions.
3.1 Measurement System. The total equip-
ment required for the determination of oxy-
gen in a given source effluent. The system
consists of three major subsystems:
3.1.1 Sampling Interface—That portion of
the measurement system that performs one
or more of the following operations: delinea-
tion, acquisition, transportation, and con-
ditioning of a sample of the source effluent
or protection of the analyzer from the hostile
aspects of the sample or source environment.
3.1.2 Analyzer—That portion of the mea-
surement system which senses the pollutant
gas and generates a signal.output that Is a
function of the pollutant concentration.
3.1.3 Data Recorder—That portion of the
measurement system that provides a per-
manent record of the output signal In terms
of concentration units.
3.2 Span. The value of pollutant concen-
tration at which the measurement system is
set to produce the maximum data display
output. For the purposes of this method, the
span shall be set approximately 1.5 to 2 times
the normal oxygen concentration In the stack
gas at the affected facility.
3.3 Zero Drift. The change In measurement
system output over a stated period of time
of normal continuous operation when the
oxygen concentration at the time for the
measurements Is zero.
3.4 Calibration Drift. The change. In mea-
surement system output over a stated period1
time of normal continuous operation when
the oxygen concentration at the time of the
measurements Is span gas.
3.5 Operational Test Period. A minimum
period of time over which a measurement sys-
tem Is expected to operate within certain
performance specifications without unsched-
uled maintenance, repair, or adjustment.
4. Measurement System Per/ormance Spec-
ifications
A measurement system must meet the per-
formance specifications in Table 3-1 to be
considered acceptable under this method.
5. Performance Specification Test Proce-
dures. The following test procedures shall be
used to determine conformance with the
requirements of paragraph 4:
5.1 Calibration Check. Establish a calibra-
tion curve for the measurement system using
zero, mid-scale gas (50%), and span gas
mixtures. Verify that the resultant curve of
instrument reading vs calibration gas value
la consistent with the expected response
curve as described by the Instrument manu-
facturer. If the expected response curve Is
not produced, additional calibration gas
measurements shall be made, or additional
steps undertaken to verify the accuracy of
the response curve of the analyzer.
5.2 Field Test for Zero Drift and Calibra-
tion Drift. Install and operate the measure-
ment system In accordance with the manu-
facturer's written Instructions and drawings
as follows:
KDERAL KGWTU, VOt »«, NO, 177—WEDNESDAY, SEPTEMBSt II. 1*74
136
-------
32870
PROPOSED RULES
TABLE 3-1.—PEHFOUMANCa BPZCmCATIDKB
Parameter Specification
1. Zero drift (2 hours)* — =£1% of (pan.
2. Zero drift (24-hour)' £3%of»pan.
3. Calibration drift (a hours)' • <£i% OS span.
4. Calibration drift (24-hour)' -=3% of span.
S. Operational period — — 168 hours minimum
•Expressed as sum of absolute mean value plus 95 percent confidence Interval of a aerie*
of tests.
5.2.1 Conditioning Period. Offset the zero
setting at least 10 percent of span so that
negative zero drift may be quantified. Oper-
ate the system for an initial 168-hour con-
ditioning period in a normal operational
manner.
5.2.2 Operational Test Period. Operate the
system for an additional 168-hour period.
The system shall monitor the source effluent
at all times except when being zeroed, cali-
brated, or backpurged.
5.2.3 Field Test for Zero Drift and Calibra-
tion Drift. Determine the values given by
zero and span gas concentrations at 2-hour
Intervals until 15 sets of data are obtained.
Alternatively, for non-extractive measure-
ment systems, determine the values given by
an electrically or mechanically produced
zero condition by inserting a certified cali-
bration gas concentration equivalent to not
leas than 10 percent O3 Into the measurement
system. Record these readings on the ex-
ample sheet shown In Figure 3-1. These 2-
hour periods need not be consecutive but
may not overlap. Zero and calibration cor-
rections and adjustments are allowed only
at 24-hour Intervals or at such shorter Inter-
vals as the manufacturer's written Instruc-
tions specify. Automatic corrections made by
the measurement system without operator
Intervention or Initiation are allowable at
any time. During the entire 168-hour test
period, record the values given by zero and
span gas concentrations before and after
adjustment at 24-hour Intervals In the ex-
ample sheet shown In Figure 3-2.
8. Calculations, Data Analysis and Report-
ing.
6.1 Procedure for determination of mean
values and confidence Intervals.
6.1.1 The mean value of a data set is cal-
culated according to equation 3-1.
,
ft? Equal ion 3-1
where:
x,=lndlvldual values,
1 :=8um of the Individual values,
"5 =mean value, and
n = number of data points.
6.1.3 The 06 percent confidence Interval
(two sided) Is calculated according to equa-
tion 3-2.
Equation 3-2
..
nvn— 1
where:
IX=aum of all data points,
tW=t,-"/2. and
C.I.M=95 percent confidence Interval esti-
mate of the average mean value.
VALUES FOB t.m
n t.m
2 ............ . .......... . ....... — 12.708
3 ....... ________ ..... . ........ ____ 4.303
4 _____ .......... ______ ............ 3.182
5 ... .............................. 2.776
6 ________________ ................. 2.571
7 ......... ______ .................. 2.447
8 ...... ______ ........ . ............ 2.365
9 _________________ ........ _. ...... 2.306
10 ... ..... ... .......... ... ......... 2.262
11 ....... ___________ ......... . ..... 2.228
12 ............. _____ ............ ... 2.201
:3 .. .................. . ..... . ..... . 2.179
14 ......... ------- ..... ______ ...... 2.180
15 ____ ........ _____________________ 2.145
16 ............... „ .......... . ..... 2.131
The values in this table are already corrected
for n-1 degrees of freedom. Use n equal to the
number of samples as data points.
6.2 Data Analysis and Reporting.
6.2.1 Zero Drift (2-hour)— Using the zero
concentration values measured each 2 hours
during the field test, calculate the differences
between consecutive 2-hour readings ex-
pressed In ppm. Calculate the mean difference
and the confidence Interval using equations
3-1 and 3-2. Report the zero drift as the sum
of the absolute mean value and the confi-
dence Interval as a percentage of the Instru-
ment span. Use example sheet shown In Fig-
ure 3-1.
6.2.2 Zero Drift (24-hour) — Using the zero
concentration values measured every 24 hours
during the field test, calculate the differ-
ences between the zero point after zero ad-
justment and the zero value 24 hours later
Just prior to zero adjustment. Calculate the
mean value of these points and the confi-
dence Interval using equations 3-1 and 3-2.
Report the zero drift (the sum of the abso-
lute mean and confidence Interval) as a per-
centage of the Instrument span. Use example
sheet shown In Figure 3-2.
6JJ.3 Calibration Drift (2-hour)—Using the
calibration values obtained at 2-hour in-
tervals during the field test, calculate the
differences between consecutive 2-hour
readings expressed as ppm. These values
should be corrected for the corresponding
zero drift during that 2-hour period. Calcu-
late the mean and confidence Interval of
these corrected difference values using equa-
tions 3-1 and 3-2. Do not use the differences
between non-consecutive readings. Report
the calibration drift as the sum of the ab-
solute mean and confidence Interval as a
percentage of the Instrument span. Use the
example sheet shown In Figure 3-1.
6.2.4 Calibration Drift (24-hour)—Using
the calibration values measured every 24
hours during the field test, calculate the
differences between the calibration con-
centration reading after zero and calibration
adjustment and the calibration concentra-
tion reading 24 hours later after zero adjust-
ment but before calibration adjustment.
Calculate the mean value of these differ-
ences and the confidence Interval using
equations 3-1 and 3-2. Report the sum of
the1 absolute mean and confidence Interval
as a percentage of the Instrument span. Use
the example sheet shown in Figure 3-2.
6.2.5 Operational Test Period—During the
168-hour performance and operational test
period, the measurement system shall not
require any corrective maintenance or n-
palr or replacement or adjustment other
than that clearly specified as required In
the operation and maintenance manuals as
routine and expected during a 1-week
period. If the measurement system operates
within the specified performance parameters
and does not require corrective mainte-
nance, repair, replacement or adjustment
' other than as specified above during the 168-
hour test period, the operational period will
be successfully concluded. Failure of the
measurement system to meet this require-
ment shall call for a repetition of the 168-
hour test period. Portions of the test which
were satisfactorily completed need not be
repeated. Failure to meet any performance
specifications shall call for a repetition of
the 1-weeK performance test period and
that portion of the testing which Is related
to the failed specification. All maintenance
and adjustments required shall be recorded.
Output readings shall be recorded before
and after all adjustments.
7. References.
7.1 Per/ortnonce Specification* for Sta-
tionary Source Monitoring Systems for Oases
and Visible Emissions, Environmental Pro-
tection Agency, Research Triangle Park,
N.C., EPA-650/2-74-013, January 1974.
7.2 Experimental Statistics, Department
of Commerce, National Bureau of Standards
Handbook 91, 1963, p. 3-31, paragraphs
3-3.1.4.
FEDERAL MOISUK, VOL. 39, NO. 177—WEDNESDAY, SEPTEMBER 11, 1974
137
-------
PROPOSED RULES
32871
Cati Zero Ctllbnttn
Tim Set Zero Drift Span Drift
Date Begin End No. Rending (izaro) Reading (iSp«n) '
1
13
15
Zero Drift • [Mean Zero Drift* + CI (Zero)
Calibration' Drift • [Mean Span Drift* + CI (Span) _
•Absolute Vtlue.
. Instrument'Span] x 100 •
_ Instrument Span] x 100 *
Figure 3-1. Zero and Calibration Drift (Z-Hour)
Date Zero Span Calibration
and Zero Drift Reading Drift
Time Reading feZero) (After zero adjustment) {4Span)
Zero Drift - [Mean Zero Drift* .
. + CI (Zero)
Instrument Span] x 100
Calibration Drift - [Mean Span Drift* * CI (Span)
• Instrument Span] x 100 • .
•Absolute value
Figure 3-2. Zero and Calibration Drift (24-Hour)
[FB Doc.74-20578 Piled »-10-74;8:45 am)
[40CFRPart51]
[PEL 239-8)
REQUIREMENTS FOR THE PREPARATION,
ADOPTION AND SUBMITTAL OF IMPLE-
MENTATION PLANS
Emission Monitoring of Stationary Sources
JULY 15, 1974.
Sections 110(a) (2) (F) (ii) and (ill) of
the Clean Air Act, as amended In 1970,
require State implementation plans to
contain legally enforceable procedures
which require the owners or operators
of stationary sources to install equip-
ment to monitor emissions from such
sources and to report the data obtained.
EPA's regulations currently require that
states have legal authority to require
such monitoring and recording (40 CFR
51.11(a)(6)>, but do not require that
state plans contain legally enforceable
procedures mandating monitoring and
recording. This is because at the time
EPA's regulations for implementation
plan preparation were published, 'the
Agency had accumulated little data on
the availability and reliability of con-
tinuous monitoring devices. The Agency
believed that the state-of-the-art was
such that it was not prudent to require
existing sources to install such devices.
Since that time, much work has been
done by the Agency and others to field
test and compare various emission moni-
tors. As a result of this worty the Agency
now believes that for certain sources,
general specifications for accuracy, re-
liability and durability can be established
for continuous emission monitors of
oxygen, sulfur dioxide, and nitrogen
dioxide and for the continuous measure-
ment of opacity for particulate matter
emissions. Accordingly, the Agency is
proposing to amend 40 CFR 51.19, Source
Surveillance, by adding a new paragraph
(e) which will require States to revise
their implementation plans to require
sources to install monitoring instruments
and to report the resulting data to the
appropriate State Agency.
State plans submitted in response to
40 CFR 51.19(e) must (1) require own-
ers or operators of specified categories
of stationary sources to install emission
monitoring equipment within one year of
plan approval, (2) specify the categories
of sources subject to the requirements,
(3) identify for each category of sources
the pollutant(s) which must be moni-
tored, (4) set forth performance speci-
fications for monitoring instruments, (5)
require that such instruments must meet
performance specifications through on-
site testing by the owner or operator, and
(6) require that data derived from such
monitoring be summarized and made
available to the State on a quarterly
basis. The minimum data requirements ,
set forth by EPA require that for gaseous
monitors all hourly values reduced to the
units for the applicable standard, not
just the excesses above the applicable
standard, should be reported to the State.
By having more than just the excess
emissions, the State will have a more
FEDERAL REGISTER. VOL 39, NO. 177—WEDNESDAY, SEPTEMBER 11, 1974
138
-------
APPENDIX C
MONITORING SYSTEMS
DRAWINGS
139
-------
CONTENTS
Drawing No. Page
B/AE-115010 Sampling System 141
B/AE-14866 Condensation-Type Conditioner 144
B/AE-14774 Houston Atlas Sampling System 145
C/AE-14856 Meloy Labs Model FSA-190-2A with Dyfusatron
Monitoring System 146
C/AE-14874 Water Cooled Condenser 147
140
-------
MAINTAINED A
)60°F
C71 °e±
NOTE 6
VENV.
W0LES TAPPED FoC l" *3"
LEGEND : i so LB . FLA ME
RMi — REFLUX MODULE .
V/T — VoLUMETANK
RD —REGENESAT
P£i —PRESSURE
FI -Fz —FILTERS
PRi —PRESSURE REGULATOR
c,i —PRESSURE&A
£l —
s. FLoWMETEe WITH \/AL\/E
— IN6 OR EQUAL FflR. PuR
-------
Drawing Beckman
Ref. No. Part No.
1
2
4 856181X
5,14
6,15 835891
7A,B
10
11
12,17
Description
Reflux probe per C/AE 14120 (8 foot length)
Ball valve with pneumatic actuator air to
open with spring return, alloy 20
Ball Valve, alloy 20, teflon, manual
X = teflon coated with spring polypropylene
coated. (1/2") port, 100 PSIG, SS body,
1 micron, glass element, teflon gasket
Gauge snubber
Compound press-vac gauge
15 psig-0-30 "Hg vac, 2-1/2" dial
1/4" lower male connection,
316 stainless steel socket and tube, drawn
steel case
Dual head diaphragm pump, one head teflon coated,
115 VAC 60 Hz, 1/8 HP 180 watts, 1/4" NPT
connections
Filter with nominal 10 micron kynar element,
polypropylene, viton, 1/4" O.D. tube
connections
Relief valve with teflon coated spring & viton
0-ring, adjusted for 10 psig, 1/4" NPT
connections
Differential pressure regulator with 13 psi
spring on top of diaphragm, teflon coat lower
body, 1/4" NPT connections
Polypropylene fittings, neoprene outer shell,
1/4" O.D. sample connection 3.8" O.D.
tube purge connection
Flowmeter with needle valve on outlet,
480-4800 cc/min air at 70° F & 1 ATM, 316 SS,
viton, 1/4" NPT connections
Quantity
1
Beckman*
INSTRUMENTS. INC.
CODEIDENTNO.
05721 !
SCALE
SIZE
A
Reflux Probe, Permeation Drying,
Sampling System
AE-115010
: 1
SHEET 2 of 3
X27-72.71
E-249
142
-------
Drawing
Ref. No.
13
19,24
18
16
20,21
22,23
25,26
Beckman
Part No.
Description
Flowmeter with needle valve on inlet, 480-4800
cc/min air at 70° F & 1 ATM, 316 SS, viton,
1/4" NPT connections
Pressure regulator with gauge, 10-180
PSI range, 1/4" NPT female connections
Solenoid valve with mounting bracket,
24 VAC 60 Hz, brass, buna N, CV = 0.15,
1/4" NPT connections'
Two way normally closed solenoid valve,
viton, 24 VAC 60 Hz, 1/4 inch NPT connections
Check valve 316 SS, buna N, 10 psi cracking
pressure 1/4" swagelok connections, 1/4 inch
NPT connections
3-way solenoid valve, 24 VAC 60 Hz, stainless
steel, viton CV = 0.31, 1/4" NPT connection
Check valve, 1/4 inch NPT connections
Quantity
1
Beckman*
INSTRUMENTS. INC.
CODE 1 DENT NO.
05721 !
SCALE
SIZE
A
• Reflux Probe, Permeation Drying,
Sampling System
AE-115010
: 1
SHEET 3 of 3
X27-72.71
143
-------
MAINTAIN/ED AT
• -(.-ASSEMBLED
INSULATE AS
TO MAINTAIN _
DEWPOINT TEMPEBATUBE I
Z"OR 3"(50L&. FLANGE
tf—e
V-'/4"
/
'A"c.D POLYETHYLENE. OR EQUAL
^{60-looPSIC. INSTRUMENTAL
'/4"or> TEFLON
MBINTAINED ABOVE170°F
r
I
u
^—3")5oLB FlAHUEWlTU HoLEiTAPPED FoB
•5 PPMSOiSPANGAS.IOPSlG
4 AIR. ZERO GAS. JOPSIG
ATMOS.
PRESS.
VENTS
( DYMASCIENCES
fi.S-330. SOi
3
li
5, 6
7
a
9, 10
11, ait,
12, 13,
Ik, 15
16
17
18, 19
20
21, 22
23
25
- Probe Filter (20 micron)
- Pro"be Assembly
- Air Operated Ball Valve
- 1 Micron Fiber Class Filter
- 3 Way Solenoid Valves for Air Service
- Air Operated Valve
- Two Way Calibration Solenoid Valve
- Air Pressure Regulators with Gauge
- Flowmeters
27, 28 - Check Valves
- Three Way Calibration Solenoid Valve
- Heated Weather Resistant Enclosure
- 3/8" O.D. Teflon Electrically -Traced Line
- Vacuum Gauge with Snubber
- Teflon Coated Diaphragm Pump
- Refrigerated Condenser with PVC Float Trap
- Relief Valve (10 psi)
- Five Uay Manual Calibration Valve
ATMOS. i
DRAIN
16
BY
DR BK
APPD
DATE
REV
SOLENOID WATTS VAC. SAMPLE ZERO SPAN MANUALbloWBAeic
8
20
10.5
24
24
OFF
OFF
OFF
DESCRIPTION
OFF
ON
ON
OFF OFF
OF?
OFF
ON
OFF
OFF
ON
OFF
OFF
OFF
DWG NO.
-------
135 LB.
TO
lOPSI SPAM 3—C*-Nh
LIME TO
INClWERAToe
LE&END :
Va -Block \/«L\/E /"BV CUSTOMER)
R/C -f?EFElGERAT£D CONDENSER.
Fi,Fz -FILTERS
6 -AIR OPERATED VflLl/E
W-Va -2 WAY SOLENOID ,n,_,,^
V\o -SWAVSOLEWOIJ? VAL\/E.
VH - METERING VALVE
Pi.Pz - PuMPS
DESCRIPTION
DATE
APPD
INSTRUMENTS. !NC
TITLE" HOUSTON/ ATLAS
- -14774
• 1T-IJ.JI TOOI41 Printed la U.S.*
-------
MELoy LASS. FSA-i3o-2A
WITH DYFUSATRON® "LutfiAS SULPHUR
(* WJTH MODl
r
INSULATE «s ?s
MAINTAIN SAMPLE A50I/E
CEWPOJNT T "
4*
l^lTY (7) II • |
_\ .OUf^U liorAMETSl' i_|._/'£RRJ; ^'ij.^
c NEEDLE V'ALVE —Jh fi i i
_AMBlEMT*IR
V. 1-tl.l t n J
SVt
p*]
c£ IM.C
^PRESSURE
REGULATOR
f
Mil
LElfTERWAL
VENr
i"o«
LEGEN
j-?Rose FILTER
i-PRoBE ASSEMBLY
3-AIR OPERATED E.ALL VALVE
4,5- 3WAYSOLENOID VALVei 611R SERVICE)
6 - AlR OPERATED VALVE
7- ik/AY CflLISRATiort S 3LENOID VALVE
8,9-«IR PRESS REGULATORS V/ITH GAUGE
10.II- FLoWMETERS
IZ.15,18 (13- SHECk VAU/ES
14.15- 3WAY C'ALISe ATION SoLEWOlD VflU/ES
l( - HEATED wearMEe ftsiisr«NT EN
-------
CaoLi
.p-
--J
.OUTLET
ITEM
QT/
DESCRIPTION
3'/l" NPT PIPE PLUG >'3L
-------
APPENDIX D
RESULTS OF AN EVALUATION OF EPA COMPLIANCE TEST METHODS
6(SO?) and 11(H2S) IN THE RANGE OF 0.15 TO 0.9% OF EACH
IN NITROGEN
148
-------
CONTENTS
Page
Method 6 Tests and Results 150
Method 11 Tests and Results 157
149
-------
METHOD 6
TESTS AND RESULTS
SUMMARY OF METHOD 6 TESTS AND RESULTS
A series of tests were run on two sets of gases containing from 150 to
460 ppm and from 4000 to 9300 ppm sulfur dioxide in nitrogen. Good re-
sults were obtained for the set of gases that contained the lower con-
centrations. Reasonable results were obtained for the other set, but
in three of four cases the concentrations that were determined by using
Method 6 were lower than expected.
TEST METHOD
The test used was Method 6 as given in the Federal Register, Volume 36,
Number 247, pages 24890-24891.* Briefly, the Method specifies that the
gas to be analyzed be passed through a train of four midget impingers of
30 ml capacity each. The first is to contain 15 ml of 80% 2-propanol,
each of the next two is to contain 15 ml of 3% hydrogen peroxide, and
the last is to be empty. The gas that is to be analyzed is passed through
the train and any sulfur dioxide present is captured and oxidized to sulfate
ion.- An air purge is used to flush dissolved S02 from the first impinger to
the second, third, and fourth impingers. The contents of the second, third,
and fourth impingers are collected and diluted to a standard volume. Aliquots
of this solution are mixed with 2-propanol and titrated with barium perchlorate
solution to a thorin endpoint. A blank is run with each series of samples. From
a knowledge of the amount of barium perchlorate consumed and the amount of gas
passed through the train, the concentration of sulfur dioxide in the gas can be
calculated.
*These pages are attached
150
-------
o
The sampling rate and sampled volume were nominally 0.056 mj (2 cu ft)
per hour and 0.028 m^ (1 cu ft), respectively. Exact values varied from
run to run.
WORKING SOLUTIONS
Working solutions were as specified in the Method. Standardized hundredth-
normal sulfuric acid was purchased and used without further analysis.
MODIFICATIONS TO THE BASIC METHOD
No modifications were made to the basic method; however, those gases that
contained higher levels of sulfur dioxide were not analyzed in exactly the
same way as those that contained lower levels. The difference involved
the extent to which the contents of the absorbing impingers were diluted
in preparation for titrations. In the first phase of the program, gases
with levels of sulfur dioxide below 500 ppm were employed. As specified
in the Test Method, the contents of the absorbing impingers were diluted
to 50 ml, and 10-ml aliquots of the solution were titrated (after being
mixed with 40 ml of 2-propanol). Under the existing conditions, between
7 and 20 ml of barium perchlorate solution was used in a typical titration
and the endpoint was reproducible to within a drop or two. Using exactly
the same dilution, something in excess of 100 ml of barium perchlorate
solution was required in the analysis of a gas that contained one of the
higher levels of sulfur dioxide, and the endpoint of a typical titration
was difficult to judge accurately. The reproducibility was not nearly so
good as for the samples that contained lower levels of sulfur dioxide.
A series of preliminary experiments established that much better reproduc-
ibility could be obtained (with no loss of accuracy) if the contents of
the absorbing impingers were diluted to 250 ml (when higher concentrations
were employed) and 10-ml aliquots were analyzed. The endpoint is sharper
and more easily recognized. This is not a modification of the method
151
-------
since the volume to which the sample is diluted is shown as a variable in
equation (6-2) (see attached) rather than being included as a constant.
TEST GASES
A total of 17 tests were run on 5 gases that contained from about 150 to
460 ppm sulfur dioxide in nitrogen. All these were obtained from an out-
side vendor—Liquid Carbonic. An additional ten tests were run on 4 gases
in which the concentration ranged from about 3700 to about 9400 ppm in
nitrogen. Two of these were supplied by the gas blending facility of
Beckman Instruments, Inc., and 2 were supplied by Liquid Carbonic.
The concentrations of gases provided by Liquid Carbonic were established
by controlling the pressure of each component during the blending opera-
tion. Each blend was then analyzed by an iodine-thiosulfate wet-chemical
analysis, The estimated accuracy was from 3 to 5% for the analysis.
The concentrations of the gases obtained internally were established by
controlling the pressure of each component during the blending operation.
The expected accuracy for this operation is approximately 2%.
TEST RESULTS
The results obtained for the gases containing less than 500 ppm sulfur
dioxide are shown in Table 15.
In all cases except one the mean concentration, as determined with Method
6, lies near the center of the range that was reported by the vendor. The
one exception involved cylinder No. 39098, and even in that case the dis-
agreement is only slight since the highest value obtained by using Method
6 is only 2 ppm below the range reported by the vendor.
The results obtained for the gases containing more than 3500 ppm sulfur
dioxide are shown in Table 16.
152
-------
Table 15. TEST RESULTS—LOW-CONCENTRATION GASES*
METHOD 6 VENDOR
Cylinder
55256
7212
39098
31661
270310
Concentration ,
Mean (ppm)
157
310
425
458
451
Experimental
Values, Range
155-161
302-316
420-429
454-462
450-453
Concentration
ppm 1 Date
158
312 (10/29)
33 (11/07)
442 (10/29)
442 (11/07)
458 (11/04)
458 (11/07)
455 (11/04)
455 (11/07)
Range, Assuming
5% Accuracy
154-162
305-319
325-341
431-453
431-453
447-469
447-469
444-466
444-466
*Liquid Carbonic was source for all gases.
Table 16. TEST RESULTS—HIGH-CONCENTRATION GASES
Cylinder
316513
14940
2708
6817
Source
Beckman
Beckman
Liquid
Carbonic
Liquid
Carbonic
METHOD 6 VENDOR
Concen-
tration,
Mean
(ppm)
3750
5540
8815
9310
Experi-
mental
Values
Range
3710-3780
5530-5550
8810-8820
9290-9340
Concen-
tration
(ppm)
4000
6000
9160
9310
Range
Assuming
5% Accuracy
3800-4200
5760-6240
8710-9610
8850-9770
153
-------
In three out of four cases the mean concentrations, obtained by using
Method 6, are lower than expected—falling below the expected range (cyl-
inders No. 316513 and No. 14940) or near the lower edge of the expected
range (cylinder No. 2708). Only in one case (cylinder No. 6817) does the
mean concentration, obtained by using Method 6, fall within the expected
range. Thus, there is at least some reason to believe Method 6 may be
less than satisfactory for use at higher concentrations.
The reason for this is not immediately obvious. Experiments were per-
formed to determine if some of the sulfur dioxide was passing untrapped
through the sampling train. Apparently this is not the case. For example,
a third impinger of hydrogen peroxide was added to the end of the sampling
train and the gas from cylinder No. 2708 (9160 ppm) was passed through the
train. The contents of the impinger were diluted to 25 ml, and a 10-ml
aliquot was analyzed. The endpoint of the titration was reached after
only one drop of titrant was added, indicating insignificant carryover.
Further, in a run involving the gas from cylinder No. 14940 (6000 ppm),
the second impinger of hydrogen peroxide was found to contain the equiva-
lent of 26 ppm, indicating that most of the sulfur dioxide had been cap-
tured in the first impinger.
154
-------
24890
RULES AND REGULATIONS
PLANT.
DATE
RUN NO.
CONTAINER
NUMBER
1
2
TOTAL
WEIGHT OF PARTICULATE COLLECTED,
mg
FINAL WEIGHT
^x.
TARE WEIGHT
;x.
WEIGHT GAIN
FINAL
INITIAL
LIQUID COLLECTED
TOTAL VOLUME COLLECTED
VOLUME OF LIQUID
WATER COLLECTED
IMPINGER
VOLUME,
ml
SILICA GEL
WEIGHT,
9
9* | ml
CONVERT WEIGHT OF WATER TO VOLUME BY DIVIDING TOTAL WEIGHT
INCREASE BY DENSITY OF WATER. (1 g ml):
= VOLUME WATER, ml
(1 g/ml)
Figure5-3. Analytical data.
6.8.2 Concentration in Ib./cu. ft.
nrliero:
c,= Concentration of purtirulutc mutter In stui-k
pas, Ib./s.c.f., dry btisis.
453,600=Mg/lb.
equation 5-5
Mi^TiituI amount of pai'liculuto matter collected,
mg.
Vmi(,,|s=Voluinc of gas sample through dry gas motnr
{standard conditions), cu. ft.'
6.7 IsokineLlc variation.
7,rV'o<*n,o
1=Tj~ M^o
wlicro: '
I»iVrcunt of isokinotic sampling.
Vif=Totul volume of liquid collected Iti impingiTS
.unil silica pel (Soo Fin. 5-3), ml.
PH*o™l)wsity of wuier, 1 g./ml.
K = I(lcal uns ronstant, 21.83 inches Hg-cu. ft./lh.
innlo-°H.
Mii1osaMoltiailar wufjrlit of water, 18 Ib./lb.-molr.
Vn,» Volume of irassiitiiplo through the dry gas incd'r
(meter conditions), cu. ft.
T,,.=Alis:>lute uvcriiKt; soluto stuck gas pressure. Indies lit:. •
Ai,«(.'i'oss-s«ii:lioiial area of nozzle, sq. ft.
6.8 Acceptable results. The following
range sets the limit on acceptable tsoklnetic
sampling results;
If 90 % < I < 110 %, the results are acceptable.
otherwise, reject the results and repeat
tho test.
7. Reference.
Addendum to Specifications for Incinerator
Testing at Federal Facilities, PHS, NCAPC,
Dec. 6. 1967.
Martin, Robert M., Construction Details of
Isokinetlc Source Sampling Equipment, En-
vironmental Protection Agency, APTD-0581.
Rom, Jerome J.. Maintenance, Calibration,
and Operation of Isokinetlc Source Sam-
pling Equipment. Environmental Protection
Agency, APTD-0576.
Smith, W. S.. R. T. Shlgehara, and W. F.
Todd, A Method of Interpreting Stack Sam-
pling Data, Paper presented at the 63d An-
nual Meeting of the Air Pollution Control
Association, St. Louis, Mo., June 14-19, 1970.
Smith, W. S., et al., Stack Gas Sampling
Improved and Simplified with New'Equip-
ment, APCA paper No. 67-119, 1967.
Specifications for Incinerator Testing at
Federal Facilities, PHS, NCAPC. 1967.
METHOD 6 DETERMINATION OF SULFUH DIOXIDE
EMISSIONS FROM STATIONARY SOURCES
1. Principle and applicability.
1.1 Principle. A gas sample is extracted
from the sampling point in the stack. The
acid mist, Including sulfur trioxlde, is sepa-
rated from the sulfur dioxide. The sulfur
dioxide fraction is measured by the barium-
thorin tltration method.
1.2 Applicability. This method is appli-
cable for the determination of sulfur dioxide
emissions from stationary sources only when
specified by the test procedures for determin-
ing compliance with New Source Performance
Standards.
2. Apparatus.
2.1 Sampling. See Figure 6-1.
2.1.1 Probe—Pyrex ' glass, approximately
5 to 6 mm. ID, with a heating system to
prevent condensation and a filtering medium
to remove particulate matter including sul-
f.uric acid mist.
2.1.2 Midget bubbler—One, with glass
wool packed in top to prevent sulfuric acid
mist carryover.
2.1.3 Glass wool.
2.1.4 Midget impiiigers—Three.
2.1.5 Dry-'ig tube—Packed with 6 to 16
mesh indicating-type silica gel, or equivalent,
to dry the sample.
2.1.6 Vnlvo—Needle valve, or equivalent,
to adjust flow rate.
2.1.7 Pump—Lenk-free, vacuum type.
2.1.8 Rate meter—Rotameter or equiva-
lent, to men.'iure a 0-10 s.c.f.h. flow range.
2.1.9 Dry gas meter—Sufficiently accurate
to measure the sample volume within 1%.
2.1.10 Pitot tube—Type S, or equivalent.
Equation 5-6 1 Trade names.
FEDERAL REGISTER, VOL. 36, NO. 247—THURSDAY, DECEMBER 23, 1971
155
-------
Ul
necessary only If a sample traverse Is re-
quired, or if stack gas velocity varies with
time.
2.2 Sample recovery.
2.2.1 Glass wash bottles—Two.
2.2.2 Polyethylene storage bottles—To
store implnger samples.
2.3 Analysis.
PROBE (END PACKED
WITH QUARTZ OR .^ STACK WALL
PYREXWOOL) >\^ MIDGET BUBBLER MIDGET IMPINGERS
J GLASS WOOL
SILICA GEL DRYING TUBE
II*. VI
\
TYPE S PITOT
DRY GAS METER ROTAMETER
Figure 6-1. SOj sampling train.
a.3.1 Pipettes—Transfer type. 5 ml. and
10 ml. sizes (0.1 ml. divisions) and 25 ml.
size (0.2 ml. divisions).
2.3.2 Volumetric flasks—50 ml., 100 ml.,
and 1.000 ml.
2.3.3 Burettes—5 ml. and 50 ml.
2.3.4 Erlenmeyer flask—125 ml.
3. Seagents.
3.1 Sampling.
3.1.1 Water—Delonlzed, distilled.
3.1.2 Isopropanol, 80%—Mix 80 ml. of Iso-
propanol with 20 ml. of distilled water.
3.1.3 Hydrogen peroxide, 3%—dilute 100
ml. of 30% hydrogen peroxide to 1 liter with
distilled water. Prepare fresh dally.
3.2 Sample recovery.
3.2.1 Water—Delonlzed, distilled.
3.2.2 Isopropanol, 80%.
3.3 Analysis.
3.3.1 Water—Delonlzed, distilled.
3.3.2 Isopropanol.
3.3.3 Thorln Indicator—l-(o-arsonophen-
ylazo)-2-naphthol-3,6-aisulfonlc add, dlso-
dlum salt (or equivalent). Dissolve 0.20 g. In
100 ml. distilled water.
3.3.4 Barium perchlorate (0.01 N)—Dis-
solve 1.95 g. of barium perchlorate
[Ba(C10.),.3H3O] In 200 ml. distilled water
No. 247—Pt. n 3
velocity. Take readings at least every five
minutes and when significant changes In
stack conditions necessitate additional ad-
justments In now rate. To begin sampling,
position the tip of the probe at f the first
sampling polnj. and start the pump. "Sam-
ple proportionally throughout the run. At
the conclusion of each run, turn off the
pump and record the final readings. Remove
the probe from the stack and disconnect It
from the train. Drain the Ice bath and purge
the remaining part of the train by drawing
clean ambient air through the system for 15
minutes.
4.2 Sample recovery. Disconnect the 1m-
plngers after purging. Discard the contents
of the midget bubbler. Pour the contents of
the midget Implngers Into a polyethylene
shipment bottle. Rinse the three midget Im-
plngers and the connecting tubes with dis-
tilled water and add these washings to the
same storage container.
4.3 Sample analysis. Transfer the contents
of the storage container to a 50 ml. volu-
metric flask. Dilute to the mark with de-
lonlzed, distilled water. Pipette a 10 ml.
aliquot of this solution Into a 125 ml. Erlen-
meyer flask. Add 40 ml. of Isopropanol and
two to four drops of thoiin indicator. Titrate
to a pink endpolnt using 0.01 N barium
perchlorate. Run a blank with each series
of samples.
5. Calibration.
5.1 Use standard methods and equipment
which have been approved by the
trator to calibrate the rotameter. pltot tube,
dry gas meter, and probe heater.
5.2 Standardize the barium perchlorate
against 25 ml. of standard suUuric add con-
taining 100 ml. of Isopropanol.
8. Calculations.
6.1 Dry gas volume. Correct the sample
volume measured by the dry gas meter to
standard conditions (70* P. and 29.92 Inches
Hg) by using equation 6-1.
17.71 r
"R
_
in. Hg \
/ V.Pb.
')
equation 6-1
where:
VmitJ = Volume of gas sample through the
dry gas meter (standard condi-
tions) , cu. ft.
Vm = Volume of gas sample through the
dry gas meter (meter condi-
tions), cu. ft.
T.,d = Absolute temperature at standard
conditions, 530* R.
Tm = Average dry gas meter temperature,
°R.
pbAra Barometric pressure at the orifice
meter, Inches Hg.
P.,a"= Absolute pressure at standard con-
ditions. 29.92 inches Hg.
6.2 Sulfur dioxide concentration.
and dilute to 1 liter with Isopropanol. Stand-
ardize with eulfurlc acid. Barium chloride
may be used.
3.3.5 Sulfurlc acid standard (0.01 W) —
Purchase or standardize to ±0.0002 N
against 0.01N NaOE which has previously
been standardized against potassium acid
phthalate (primary standard grade).
4. Procedure.
4.1 Sampling.
4.1.1 Preparation of collection train. Pour
15 ml. of 80% Isopropanol into tbe midget
bubbler and 15 ml. of 3% hydrogen peroxide
into each of the first two midget Implngers.
Leave the final midget implnger dry. Assem-
ble the train as shown In Figure 6-1. Leak
check the sampling train at the sampling
site by plugging the probe inlet and pulling
a 10 Inches Rg vacuum. A leakage rate not
in excess of 1% of the sampling rate is ac-
ceptable. Carefully release the probe inlet
plug and turn off the pump. Place crushed
Ice around the Impingers. Add more Ice dur-
ing the run to keep the temperature of the
gases leaving the last Implnger at 70" F. or
less.
4.1.2 Sample collection. Adjust the sam-
ple now rate proportional to the stack gas
/ ib i \
= f7.05xlo-'-^-r )
\ g.-ml./
where:
0*0.,= Concentration of sulfur dioxide
at standard conditions, dry
basis, Ib./cu. ft.
7.05 X 10-== Conversion factor, Including the
number of grams per gram
equivalent of sulfur dioxide
(32 g./g.-eq.), 453.6 g./lb., and
1,000 ml./l.. Ib.-l./g.-ml.
Vt — Volume of barium perchlorate
titrant used for the sample,
mi.
V,,. = Volume of barium perchlorate
titrant used for the blank, ml.
W = Normality of barium perchlorate
titrant, g.-eq./l.
V,0,a = Total solution volume of sulfur
dioxide, SO ml.
Vm = Volume of sample aliquot ti-
trated, ml.
Vm,,a = Volume cf gas sample through
' the dry gas meter (standard
conditions), cu. ft., see Equa-
tion 6-1.
>
O
FEDSRAL REGISTER, VOL. 36, NO. 217-•T.'.'l.W.Vf, D:cE.MCrR 33, 1971
O
r~
equation 0-2 i
O
7. Re/erencea.
Atmospheric Emissions from Sulfurlc Acid
Manufacturing Processes, US. DHEW, PHS.
Division of Air Pollution. Public Health Serv-
ice Publication No. 999-AP-13, Cincinnati.
Ohio, 1965.
Corbett, P. P., The Determination of SO.
and 8OS In Flue Gases, Journal of the Insti-
tute of Fuel, 24:237-243, 1961.
Matty, R. E. and E. K. Dlehl. Measuring
Flue-Gas SO, and SO,, Power 10.1:94-97, No-
vember, 1957.
Patton, W. P. and J. A. Brink, Jr.. New
Equipment and Techniques for Sampling
Chemical Process Gases, J. Air Pollution Con-
trol Association, 13, 162 (1963).
METHOD 7—DETERMINATION OT N1TBOGBN OXIDE
EMISSIONS' FROM STATIONABT SOUBCES
1. Principle and applicability,
1.1 Principle. A grab sample Is collected
in an evacuated flask containing a dilute
sulfuric acid-hydrogen peroxide absorbing
solution, and the nitrogen oxides, except to
CO
-------
METHOD 11
TESTS AND RESULTS
SUMMARY OF METHOD 11 TESTS AND RESULTS
It was necessary to evaluate Method 11 for use on the Glaus plant tail
gas before incineration, which contained up to about 1% H~S during
transient upsets, and normally operated in the 2000-ppm ^S range at
the selected test site.
A series of tests were run on gases containing from 200 to 9040 ppm hydro-
gen sulfide in nitrogen. Good results were obtained, as expected, for the
gas containing 200 ppm; however, results for the gases containing higher
concentrations (beyond that which may be determined using Method 11 un-
modified) were unacceptably low. The data are presented and discussed,
including a possible reason for unacceptable results.
TEST METHOD
The basic test used xvas Method 11, as given in the Federal Register, Vol-
ume 39, Number 47, pages 9321-9323.* Briefly, the method specifies that
the gas to be analyzed be passed through a train of 5 midget impingers
of 30-ml capacity each. The first is to contain 15 ml of freshly pre-
pared 3% hydrogen peroxide, each of the next three is to contain 15 ml
of cadmium hydroxide absorbing solution, and the last is to be empty.
Gas is to be passed through the train at 1.13 1/min for a minimum of
10 minutes. The hydrogen sulfide is captured as cadmium sulfide. The
sulfide is determined by adding the contents of the 3 absorbing impingers
to 100 ml of acidified iodine solution. This process regenerates hydro-
gen sulfide which then reacts with iodine. After sufficient time has
elapsed for all the hydrogen sulfide to react with iodine, the remaining
*These pages are attached.
157
-------
iodine is titrated with thiosulfate to a starch endpoint. A blank is
titrated similarly. From a knowledge of the amount of gas passed through
the train and how much thiosulfate was consumed in each titration, th.e
concentration of hydrogen sulfide in the gas may be calculated.
WORKING SOLUTIONS
Some care was exercised to ascertain that the reagents and solutions used
were in satisfactory condition. Pre-standardized 0.1 N sodium thiosul-
fate solution was purchased. The standardization was checked against
primary-standard-grade potassium dichromate and found to be as stated,
within experimental error. Hundredth-normal (nominal) iodine solution
was prepared and standardized daily. Although there are no restrictions
stated in the Method concerning the lifetime of the cadmium hydroxide
solution, only 21 days elapsed between the time the solution was prepared
and the time the final test was run. Further, a duplicate test was run
in one instance to assess the effect of the freshness of the absorbing
solution.
MODIFICATIONS TO THE BASIC METHOD
Given the constraints that are specified in Method 11 concerning sampling
rate and time, volume and normality of iodine solution, and assuming an
ideal blank, the absolute maximum concentration of hydrogen sulfide that
can be determined is about 530 ppm. However, the stream that was to be
analyzed and the span gases that were to be used in instrument calibration
contained approximately 5000 ppm hydrogen sulfide and 2500 ppm sulfur di-
oxide, and 9000 ppm hydrogen sulfide, respectively. Hence, the method
could not be used without modification. By mutual agreement with the
cognizant Project Technical Officer, it was agreed that the sampling time
would be reduced to extend the upper limit of H2S concentration theo-
retically detectable.
3
A volume of 0.0014 m (0.05 cu ft) was judged to be about the smallest that
could be read with accuracy on the dry gas meter that was available for
158
-------
the tests . (The meter did comply with the specifications that were
applicable to the unmodified Method 11). To be sure that no significant
error was introduced by sampling this relatively small volume, the sam-
pling time was measured to the nearest tenth second. After the sample
was taken, but before the rotameter setting, temperature, or pressure
changed significantly, the time required to pass a relatively large vol-
ume of ambient air (purge) through the system was also determined. A
further crude check on the sampled volume was obtained in this manner by
comparing times and indicated volumes.
o
Assuming that .0014 m (0.05 cu ft) is the volume of gas that is sampled,
and no other modifications are made to the procedure, then the highest
concentration of hydrogen sulfide that can be determined is approximately
4250 ppm. Thus, in many of the tests discussed, not only was the smaller
volume of gas employed, but additional iodine was used in the workup of
the sample and blank as well. When the gas being sampled contained
4000 ppm or more hydrogen sulfide, additional impingers containing cadmium
hydroxide were inserted into the sampling train to ascertain that all the
hydrogen sulfide was scrubbed from the gas stream.
At the time the tests were run, questions arose concerning the tendency
of the hydrogen peroxide in the first impinger to irreversibly remove
hydrogen sulfide from the gas streams in which it was present at high
concentrations. Thus, a few tests were run without hydrogen peroxide
in the train. All the tests run in the program were run on gases that
contained only hydrogen sulfide in the nitrogen, but in a few of the
tests in which hydrogen peroxide was used a small amount of sulfuric
acid was added to it to simulate the effect upon pH of passing a mixture
containing sulfur dioxide through the train.
159
-------
TEST GASES
A total of 15 tests were run on gases that contained from 200 to 9040 ppm
hydrogen sulfide in nitrogen. Two of the gases (200 and 4000 ppm) were
supplied by the gas blending facility of Beckman Instruments, Inc. The
concentrations are established by controlling the pressure of each com-
ponent during the blending operation. An accuracy of 2% is to be expected
in this operation and that figure is appropriate for the gas containing
4000 ppm. However, two such operations were used in order to produce the
gas containing 200 ppm. Hence, an accuracy of 4-5% is appropriate in that
case. The remaining gases, 9000 and 9040 ppm, were supplied by an outside
vendor—Liquid Carbonic. No data are available to indicate the accuracy
of these numbers. In no case was an independently obtained, wet-chemical
analysis available. However, the Houston Atlas, Inc., System indicated
that the relative values of the various tanks were within a few percent
of the values on the tank labels. Since the Houston Atlas System normally
provides a linear output, its analysis is deemed to strongly support the
conclusion that the Method 11 results on the high concentrations were
erroneous.
TEST RESULTS
The data from the tests are presented in Table 17.
Runs 1 through 6 were made using the gas containing 4000 ppm hydrogen sul-
fide. In no case did the results approach the expected value. The re-
sults of runs 1, 2, and 3 were approximately 40% low. Omitting the
hydrogen peroxide from the train,(runs 4 and 5) may have produced some
benefit, but not nearly enough. Using acidified hydrogen peroxide (run
6) did not help; however, the results from runs 3, 5, and 6 do give a
clue as to where the problem may lie. The fact that 3 impingers contain
sufficient cadmium hydroxide to capture all the hydrogen sulfide that
will react with 50 ml of iodine solution is implicit in the unmodified
160
-------
Table 17.
METHOD 11(H2S) TEST RESULTS
Run 1
Cylinder Contents, ppm
Date
T ime
Samp 1 e
Time, Finish
Time. Start
Duration, Seconds
Cas Meter, Finish, ft3
(las Meter, Start, ft3
Temperature, Finish, °F
Purge
Time, Finish
Time, S art
r.as Met r. Finish, ft3
Timed V lume. ft3
Time. S ronds
Computed Sample Time, SIT.
Kotameter, ft3/hr
Titrations
13- in Sample, ml
S.O "~ to Titrate, ml
13- in Blank, ml
SO " to Titrate, ml
I..- Standardization Volume
S 0 ~^ to Titrate, ml
ppm. Method 11
Remarks
1
4000
Test
1/6/75
1100
79.0
0.69
0.64
70
69
29.66
1125
1)10
1.02
0.
139.1
35
2
50
19.30
50
49.01
ml 25
"•^ 54
22.50
22.40
2551
1H2°2
bed (Oil >2
2
4000
1 ni c i.'i 1
Test
1/6/76
1300
79.9
0.25
0.20
73
73
29.67
0. 20
322.9
30.7
2
50
20.65
50
48.42
25
'*'' 74
22.70
22.60
2400
IH20,
6Cd((lH),
3
4000
3/19/75
1700
72.0
0.40
0.35
75
75
29.63
0. 20
239.4
72.4
2.4
100
7 1 . 18
50
49.64
25
''3 23
23. 13
23.00
2423
1H20,
6Cd(OH),
Slight
yellow
in *4
(Cd(OH)2.
None
in <>5.
4
4000
"" "^
3/18/75
1130
72.7
0.88
0.81
71
70
29. 78
1355
1340
0.20
104.8
76. 1
2.4
100
64 . 60
100
99.75
25
''3 08
23.15
3012
6C.K1IH),
.'.'inn
2 .•
1/25/75
1227
71.0
0.05
0.0"
74
7 1
29.11
1242
1227
0.20
284.5
71. 1
2.4
50
16.49
50
49. 19
25
•' ' 76
22.99
22.88
2844
6CdOIH>,
Slight
yellow
inJ4
Cd(OH)2.
None
in »5.
6 7 8
•'"""_ 9"4" 90'°
11,0. Test T.-sl
1/25/75 ,/>!/75 1/13/75
M45 lf>H! 0930
72.8 81.6 HI. 9
0.35 0.65 1 .00
0.10 0.60 0.95
72 70 66
;•' 70 66
"J.5I 29.41 29.80
1216 1 nn ,
120! 1635 '.i'»SO
0. 10 0.20
144.7 311.8
72.1 31.1
2.4 2 2
50 100 100
16. 50 31. 16 31 . 96
50 100 50
49.1? 97.85 50.56
25 25 25
'".7b 22.82 23.44
22.99 22.79 21.46
22.88 22.76 23.54
2815 5605 5871
111,", 1H20, 111,0.,
6Cd(011), 6Cd, 6i:d(OH),
Slight
yellow
in 04
(Cd(OH)2.
None
in 85
I 10
9040 OOOno
22 22
3/14/75 1/20/75
1600 1415
7i.:' 68.'.
0. 10
0.05
71 72
71 7'
2". 65 29.62
0.20 0.20
28J 286.6
7fi.5 71.6
2.4 2.4
100 100
28.35 14.25
IM'l 100
97.90 98.51
25 25
22.80 2'J.98
22.89 23.11
21.01
5992 5553
6Cd(OH)2 7Cd«IH),
Slight
yellow
in ?5
Cd(OH>2.
None
in "6.
11 12
9H0.1 9000
- ' 2 H2"2
3/21/75 1/21/75
1510 1010
0.6 74.5
.35 0.75
.10 0.70
5 71
5 71
29.66 29.72
1055
1552 1040
0.'20
297.8
74.4
2.4 2.4
100
76.76
"'2V
6i:d(0!0 2drops H*
empty 6Cd(OH)
Slight 1 empty
yellow Slight
in M yellow
Cd(OH)2. in J4
None Cd(OH),.
in «5. None
in i'5.
13 14
9000 200
1I2«, mark
1/21/75 3/26/75
1400 0910
0941
0936
71.7
0.55 1.06
0.50 0.66
70 65
70 61
29.71 29.50
1427
1412
2.4 2.4
50
26.73
50
48.23
25
22.63
22.61
229
111,1',+ 111,11.,
2drops H*
6Cd(OH), 3rd (OH),
1 empty ! emp[v
Slight Slight
yellow yellow
in *4 1" "J
cd(OH)2. "2
None
in 15.
15
200
inf solution
1/25/75
2010
1.05
0.65
69.5
68
29.19
2047
2.4
50
26. 50
50
48.85
25
22.76
22.99
22.38
241
'":":
ICd(ol')
1 empty
Slight
ve 1 1 ov
In ?'3
C.I(Oll),
-------
Method 11. In runs 3, 5, and 6 no more than 35 ml of iodine solution was
consumed by hydrogen sulfide, based on the amount of thiosulfate solution
that was used in the titrations. However, in each of these runs there
was a distinct yellow color in the fourth impinger of cadmium hydroxide—
indicating that some hydrogen sulfide was carried beyond the third. Thus,
there is some question as to how effectively the cadmium hydroxide scrubs
hydrogen sulfide when the latter is present in relatively high concentra-
tion. The lack of an entry in this respect for runs 1, 2, and 4 indicates
only that no notation was made in the log book. These runs were made early
in the series, before this aspect of the significance of carryover was fully
appreciated.
Runs that were made on the gas containing 9040 ppm produced results that
were about 45% low regardless of whether hydrogen peroxide was used (runs
7 and 8) or not (run 9). There may have been carryover past the third
impinger containing cadmium hydroxide in these runs.
Runs that were made on the gas containing 9000 ppm also produced unsatis-
factory results. When no hydrogen peroxide was used, the result was
about 45% low (run 10). Based on the appearance of the cadmium hydroxide
that was used in run 11 (sam e conditions as run 10), an analysis would
have produced results that were little or no better. Using acidified
hydrogen peroxide (runs 12 and 13) likely produced worse results. Based on
the amount of thiosulfate solution consumed in the titration of run 12, the
analysis would have been low by an additional factor of two. In all cases,
there was carryover past the third impinger containing cadmium hydroxide.
A benchmark test was run on gas containing 200 ppm hydrogen sulfide to
establish that the techniques involved were being performed properly.
The test (run 14) was performed in strict accordance with Method 11.
The result, 229 ppm, is perhaps a bit farther from 200 ppm than would
have been preferred, but it was the first result that was higher than or
162
-------
even reasonably near the expected value. The test was repeated with fresh
absorbing solution (run 15) and a value of 241 ppm was obtained. The mean
of the two values, 235 ppm, and the range, ±6 ppm, are considered
satisfactory.
163
-------
RULES AND REGULATIONS
9321
10, Bibliography.
10.1 McElroy, FranK, The Intertech NDIR-CO
Analyzer, Presented at llth Methods
Conference on Air Pollution, University
of California, Berkeley, Calif., April 1,
1870.
10.3 Jacobs, M. B., et al., Continuous Deter-
mination of Carbon Monoxide and Hy-
drocarbons In Air by a Modified Infra-
red Analyzer, J. Air Pollution Control
Association, S(2):110-114, August 1959.
10.3 MSA LIRA infrared Gas and Liquid
Analyzer Instruction Book, Mine Safety
Appliances Co., Technloal Products Di-
vision, Pittsburgh, Pa,
10.4 Models 2 IS A, 318A, and 4 IB A Infrared
Analyzers, Beckman Instruments, Inc.,
Beckman Instructions 1638-B, Fuller-
ton, Calif., October 1967.
10.S Continuous CO Monitoring System,
Model A8611, Intertech Corp., Princeton,
N.J.
10.6 UNOB Inffared Gas Analyzers, Bendlx
Corp., Honceverte, West Virginia.
ADDENDA
A. Performance Specifications for NDIR Carbon Monoxide Analyzers.
Range (minimum) 0-1000ppm.
Output (minimum) 0-10mV.
Minimum detectable sensitivity ZOppm.
Rise time, 90 percent (maximum) 30 seconds.
Fall time, 90 percent (maximum). — 30 seconds.
Zero drift (maximum) f 10% In 8 hours.
Span drift (maximum) 10% In 8 hours.
Precision (minimum) ± 2% of full scale.
Noise (maximum) ± 1 % of full scale.
Linearity (maximum deviation) 3% of full scale.
Interference rejection ratio COs—1000 to 1, HO—500 to 1.
B. Definitions of Performance Specifica-
tions.
Range—The minimum and maximum
measurement limits.
Output—Electrical signal which Is propor-
tional to the measurement: Intended for con-
nection to readout or data processing devices.
Usually expressed as millivolts or mllllamps
full scale at a given Impedance.
Full scale—The maximum measuring limit
for a given range.
Minimum detectable sensitivity—The
smallest amount of Input concentration that
can be detected as the concentration ap-
proaches zero.
Accuracy—The degree of agreement be-
tween a measured value and the true value;
usually expressed as ± percent of full scale.
Time to 90 percent response—The time In-
terval from a step change in the input con-
centration at the instrument inlet to a read-
ing of 90 percent of the ultimate recorded
concentration.
Rise Time (90 percent)—The Interval be-
tween initial response time and time to 90
percent response after a step Increase in the
Inlet concentration.
Fall Time (90 percent)—The Interval be-
tween initial response time and time to 90
percent response after a step decrease in the
inlet concentration.
Zero Drift—The change in Instrument out-
put over a stated time period, usually 24
hours, of unadjusted continuous operation
when the input concentration Is zero; usually
expressed as percent full scale.
Span Drift—The change In Instrument out-
put over a stated time period, usually 24
hours, of unadjusted continuous operation
when the input concentration is a stated
upscale value; usually expressed as percent
full scale.
Precision—The degree of agreement be-
tween repeated measurements of the same
concentration, expressed as the average de-
viation of the single results from the mean.
Noise—Spontaneous deviations from a
mean output not caused by Input concen-
tration changes.
Linearity—The maximum deviation be-
tween an actual instrument reading and the
reading predicted by a straight line drawn
between upper and lower calibration points.
METHOD 11—DErrERUWATION OF HYDROGEN ST7L-
FIDE EMISSIONS FROM STATIONARY SOURCES
1. Principle ana. applicability.
1.1 Principle. Hydrogen sulflde (HSS) is
collected from the source in a series of midget
implngers and reacted with alkaline cad-
mium hydroxide [Cd(OH),I to form cad-
mium sulflde (CdS). The precipitated CdS
Is then dissolved in hydrochloric acid and
absorbed in a known volume of Iodine solu-
tion. The Iodine consumed is a measure of
the H,8 content of the gas. An implnger con-
taining hydrogen peroxide is Included to re-
move SO, as an Interfering species.
1.2 Applicability. This method is applica-
ble for the determination of hydrogen sul-
flde emissions from stationary sources only
when specified by the test procedures for
determining compliance with the new source
performance standards.
2. Apparatus.
2.1 Sampling train.
2.1.J Sampling line—6- to 7-mm (}4-lnch)
Teflon' tubing to connect sampling train to
sampling valve, with provisions for heating
to prevent condensation. A pressure reduc-
ing valve prior to the Teflon sampling line
may be required depending on sampling
stream pressure.
2.1.2 Implngers—Five midget Implngers.
each with 30-ml capacity, or equivalent.
2.1.3 Ice bath container—To maintain ab-
sorbing solution at a constant temperature.
2.1.4 Silica gel drying tube—To protect
pump and dry gas meter.
2.1.6 Needle valve, or equivalent—Stainless
steel or other corrosion resistant material, to
adjust gas flow rate.
2.1.6 Pump—Leak free, diaphragm type, or
equivalent, to transport gas. (Not required
if sampling stream under positive pressure.)
2.1.7 Dry gas meter—Sufficiently accurate
to measure sample volume to within 1 per-
cent.
2.1.8 Rate meter—Rotameter, or equivalent,
to measure a flow rate of 0 to 3 liters per
minute (0.1 ft'/mln).
2.1.9 Graduated cylinder—25 ml.
2.1.10 Barometer—To measure atmospheric
pressure within ±2.5 mm (0.1 In.) Hg.
2.2 Sample Recovery.
2.2.1 Sample container—SOO-ml glass-stop-
pered iodine flask.
2.2.3 Pipette—50-ml volumetric type.
2.2.3 Beakers—250 ml.
2.2.4 Wash bottle—Glass.
2.3 Analysis.
2^3.1 Flask—500-ml glass-stoppered iodine
flask.
1 Mention of trade names or specific prod-
ucts does not constitute endorsement by the
Environmental Protection Agency.
2.3.2 Burette—One 50 ml.
3.3.2 FWsfc—125-ml conical.
3. Reagents.
3.1 Sampling.
3.1.1 Absorbing solution—Cadmium hy-
droxide (Cd(OH),)— Mix 4.3 g cadmium nil-
fate hydrate (3 CdSO,.8H,O) and 0.3 g of
sodium hydroxide (NaOH) in 1 liter of dis-
tilled water (H,O). Mix well.
Note: The cadmium hydroxide formed In
tills mixture will precipitate as a white sus-
pension. Therefore, this solution must be
thoroughly mixed before using to ensure an
even distribution of the cadmium hydroxide.
3.1.2 Hydrogen peroxide, 3 percent—Dilute
30 percent hydrogen peroxide to 3 percent
as needed. Prepare fresh dally;
3.2 Sample recovery.
3.2.1 Hydrochloric acid solution (HCl). 10
percent by weight—Mix 230 ml of concen-
trated HCl (specific gravity 1.19) and 770 ml
of distilled H..O.
3.2.2 Iodine solution, 0.1 AT—Dissolve 24 g
potassium iodide (KI) in 30 ml of distilled
HaO in a 1-liter graduated cylinder. Weigh
12.7 g of resublimed iodine (I,) into a weigh-
ing bottle and add to the potassium iodide
solution. Shake the mixture until the iodine
is completely dissolved. Slowly dilute the so-
lution to 1 liter with distilled H..O, with
swirling. Filter the solution, if cloudy, and
store In a brown glass-stoppered bottle.
3.2.3 Standard iodine solution, 0.01 N—Di-
lute 100 ml of the 0.1 W iodine solution in a
volumetric flask to 1 liter with distilled
water.
Standardize dally as follows: Pipette 25 ml
of the 0.01 N Iodine solution into a 125-ml
conical flask. Titrate with standard 0.01 N
thiosulfate solution (see paragraph 3.3.2) un-
til the solution is a light yellow. Add a few
drops of the starch solution and continue
titrating until the blue color just disap-
pears. From the results of this tltratlon, cal-
culate the exact normality of the iodine
solution (see paragraph 5.1).
3.2.4 Distilled, deionized water.
3.3 Analysis.
3.3.1 Sodium thiosulfate. solution, standard
0.1 N—For each liter of solution, dissolve
24.8 g of sodium thiosulfate (NA,j3,O, • 5HL.O)
In distilled water and add 0.01 g of anhydrous
sodium carbonate (Na.CO,) and 0.4 ml of
chloroform (CHC1.) to" stabilize. Mix thor-
oughly by shaking or by aerating with nitro-
gen for approximately 15 minutes, and store
in a glass-stoppered glass bottle.
Standardize frequently as follows: Weigh
Into a 500-ml volumetric flask about 2 g of
potassium dlchromate (K,Cr,O,) weighed
to the nearest milligram and dilute to the
500-ml mark with distilled Hp. Use di-
chromate which has been crystallized from
distilled water and oven-dried at 182'C to
190'C (360°F to 390-F). Dissolve approxi-
mately 3 g of potassium Iodide (KI) In 50 ml
of distilled water In a glass-stoppered, 600-ml
conical flask, then add 5 ml of 20-percent
hydrochloric acid solution. Pipette 60 ml of
the dlchromate solution into this mixture.
Gently swirl the solution once and allow it
to stand In the dark for 5 minutes. Dilute
the solution with 100 to 200 ml of distilled
water, washing down the sides of the flask
with part of the water. Swirl the solution
slowly and titrate with the thoisulfate solu-
tion until the solution is light yellow. Add
4 ml of starch solution and continue with a
slow tltratlon with the thiosulfate until the
bright blue color has disappeared and only
the pale green color of the chromic Ion re-
mains. From this tltration, calculate the ex-
act normality of the sodium thiosulfate solu-
tion (see paragraph 5.2).
3.3.2 Sodium thiosulfate solution, standard
0.01 N—Pipette 100 ml of the standard 0.1 N
thiosulfate solution into a volumetric flask
and dilute to one liter with distilled water.
FEDERAL REGISTER, VOL. 39, NO. 47—FRIDAY, MARCH 8, 1974
164
-------
9.122
RULES AND REGULATIONS
3.3.3 Starch indicator solution—Suspend
10 g of soluble starch In 100 ml of distilled
water and add 15 g of potassium hydroxide
pellets. Stir until dissolved, dilute with 900
ml of distilled water, and let stand 1 hour.
Neutralize the alkali with concentrated hy-
drochloric acid, using an Indicator paper
similar to Alkacld test ribbon, then add 2 ml
of glacial acetic acid as 'a preservative.
Test for decomposition by titrating 4 ml of
starch solution In 200 ml of distilled water
with 0.01 N Iodine solution. If more than 4
drops of the 0.01 N Iodine solution are re-
quired to obtain the blue color, make up a
fresh starch solution.
4. Procedure.
4.1 Sampling.
4.1.1 Assemble the sampling train as shown
In Figure 11-1, connecting the five midget
Implngers in series. Place 15 ml of 3 percent
hydrogen peroxide In the first Implnger. Place
15 ml of the absorbing solution In each of
the next three Implngers, leaving the fifth
dry. Place crushed Ice around the Implngers.
Add more Ice during the run to keep the
temperature of the gases leaving the last
Implnger at about 20*C (70'F), or less.
4.1.2 Purge the connecting line between
the sampling valve and the first Implnger.
Connect the sample line to the train. Record
the Initial reading on the dry gas meter as
shown in Table 11-1. ^
Flgu'elM. H;S sani.liiq train.
TABLE 11-1.—Field data
Test
Date
Barometric pressure..
Gas volume Rotamctcr
('lock through wiling, Lpm
time meter (V'n,), (cubic feet
liters (cubic jmr minulc)
feet)
Meter
leiDIHiiatUlC,
4.1.3 Open the flow control valve and ad-
just the sampling rate to 1.13 liters per
minute (0.04 cfm). Read the meter temper-
ature and record on Table 11-1.
4.1.4 Continue sampling a minimum of 10
minutes. If the yellow color of cadmium sul-
fide is visible in the third Implnger, analysis
should confirm that the applicable standard
has been exceeded. At the end of the sample
time, close the flow control 'valve and read
the nnal meter volume and temperature.
4.1.5 Disconnect the impinger train from
the sampling line. Purge the train with clean
ambient air for 15 minutes to ensure that all
HjS Is removed from the hydrogen peroxide.
Cap the open ends and move to the sample
clean-up area.
4.2 Sample recovery.
4.2.1 Pipette 50 ml of 0.01 N Iodine solution
Into a 250-ml beaker. Add 50 m) of 10 percent
HC1 to the solution. Mix well.
4.2.2 Discard the contents of the hydrogen
peroxide Implnger. Carefully transfer the con-
tents of the remaining four implngers to a
500-ml iodine flask.
4.2.3 Rinse the four absorbing Implngers
and connecting glassware with three portions
of the acidified iodine solution. Use the en-
tire 100 ml of acidified iodine for this pur-
pose. Immediately after pouring the acidified
iodine Into an Implnger, stopper It and shake
for a few moments before transferring the
rinse to the iodine fiask. Do not transfer any
rinse portion from one Implnger to another;
transfer it directly to the Iodine flask. Onc«
acidified iodine solution has been poured Into
any glassware containing cadmium sulfide
sample, the container must be tightly stop-
pered at all times except when adding more
solution, and this must be done as quickly
and carefully as possible. After adding any
acidified iodine solution to the Iodine flask,
allow a few minutes for absorption of the H,s
into the iodine before adding any further
rinses.
5. Calculation*.
5.1 Normality oj the *laiular
-------
RULES AND REGULATIONS 9323
where (English units) I
g 02e3_".0(lS.43gT/g)
*-°-263 (1,000 I/mi
V«.,d=scf.
CHls=gr/dscf.
6. References.
6.1 Determination of Hydrogen Sulflde, Ammoniacal Cadmium Chloride Method,
API Method 772-54. In: Manual on Disposal of Refinery Wastes, Vol. V: Sampling
and Analysis of Waste Gases and Particulate Matter, American Petroleum Institute,
Washington, D.C., 1954.
6.2 Tentative Method for Determination of Hydrogen Sulfide and Mercaptan Sulfur
In Natural Gas, Natural Gas Processors Association, Tulsa, Oklahoma, NGPA Publi-
cation No. 2265-65, 1965.
' [PR Doc.74-4784 Filed 3-7-74;8:46 am]
,..__._ . FIOSlAt REOISTEI, VOL 39, NO. 47—FIIIOAr, MARCH 8. 1974
WO. 47—rv. II • "9
166
-------
APPENDIX E
MISCELLANEOUS TESTS, CALCULATIONS. AND RESULTS
167
-------
CONTENTS
Page
1.0 Direct Measurement of SCL Lost in Condensed Water 169
2.0 Analytical Derivation of SO- Losses into Condensate .... 170
3.0 Method 6 Analysis of Sample Delivered by Refrigerated
Condenser and Permeation Dryer Systems 172
4.0 Independent Verification of Linearity of Several
Monitors 174
5.0 Disc Diluter System Calibration and Testing ........ 175
6.0 Preliminary Method 6 Correlation Test and Results , . . , 179
7.0 Computer Programs Used in Reduction of Data 179
168
-------
1.0 DIRECT MEASUREMENT OF S02 LOST IN CONDENSED WATER
During the work at the power plant site, a few drops of condensate normally
accumulated in the interconnection between the heated probe used for col-
lecting Method 6 samples and the cooled sampling impinger train. Several
tests were performed to directly determine the maximum error that loss of
S02 into this.condensate might produce. First, a blank Method 6 analysis
was performed using ambient air. The result indicated that the ambient
air S02 concentration was negligible. Specifically, less than one drop of
titrant was required versus a normal range of 6 to 10 ml of titrant for
the stack gas (about 200 ppm S02) analysis. The second test involved col-
lection of the condensed water at the end of one sample collection run
prior to connecting a second absorber train for a second run. The conden-
sate was collected in a spare impinger by tilting and tapping the probe.
Each of three samples collected in this manner was titrated. The results
were calculated on the basis of equivalent ppm S02 in 0.028 m^ (1 cu ft)
of sample gas, which was the normal sample size for Method 6 analyses.
The results of these three titrations were 5, 4, and 3 ppm S02 equivalent.
The corresponding Method 6 results for the samples collected as the con-
densate accumulated were 80 (erroneous due to leaking stopcock), 180, and
219 ppm S02^ The Method 6 results would have been about 2% higher if the
probe condensate had been included in the collected samples. As performed,
the titration included l^SO^ as well as S02 in the condensate. Therefore,
direct inclusion of the condensate in the sample could lead to high S02
measurements (due to H2S04), while noninclusion of the S02 dissolved in
the condensate would cause low indications of S02- If the condensate had
been delivered to the first impinger before air purging, the t^SO^ would
have been trapped while the S02 would have been carried over into the
169
-------
absorbing impinger during the air purge step, resulting in a correct anal-
ysis. In any event, these results indicate that the error due to neglecting
the possible loss of S02 in the condensate could not have been greater than
2% of the amount reported.
2.0 ANALYTICAL DERIVATION OF S02 LOSSES INTO CONDENSATE
In support of the experimental measurement of S02 lost in condensed water
in the Method 6 sampling probe reported in Paragraph 1.0 of this appendix,
a generalized equation was derived for estimating the maximum error in
measured concentration of any gaseous sample constituent from this source,
applicable for any condenser method of drying. The loss of S02 predicted
for the prevailing power plant water loading was consistent with the ex-
perimental results reported in Paragraph 1.0. The derivation and an
example calculation for S02 are given in the following paragraphs.
The amount of any gas that can be dissolved in a given amount of water is
simply proportional to the solubility constant of the gas in question (by
definition of solubility constant). The solubility constant for all gases
is temperature dependent, and for acid (and basic) gases it is also pH de-
pendent. When more than one acid gas is involved, the pH change due to
one acid (or basic) gas affects the solubility of the others. The gas that
is the anhydride of the strongest acid will dominate for roughly equal con-
centrations. For example, the formation of H2S03 due to 500 ppm of S02 may
be expected to lower the pH of the condensate enough to significantly re-
duce the solubility of C02 in the condensate. For much larger concentra-
tions of C02 than S02> the solubility of both gases will be lower than
that for either gas alone. While a general case, multiple acid gas con-
stituent derivation is possible, it is sufficient for a worst-case estimate
of loss of any one acid-gas constituent to assume that it is the only acid
gas present—which maximizes the solubility in water. For this reason, the
following derivation applies to any gas (such as oxygen) as well as to the
maximum loss for gases that react with water (such as S02, C02, and
170
-------
The amount of water condensed from a given volume of sample is given by
(Fwg-Fwc) 18 ml H20 Ps(atm) 273(K)
Vw = 1-(FWS-FWC) X Vs (liters> x 22.4 liters vapor X . 1 atm X T (K)
S
where
V = volume of condensed water in milliliters
w
Vs = volume of "dry" sample collected in liters
FWS = mole fraction of water in sample
Fwc = mole fraction of water in sample leaving condensor
Ts = temperature of sample entering probe (and condensor)
Ps = total pressure of sample entering probe (and condensor)
The volume of acid gas lost in this condensate (worst case) is as follows:
Vgl(ml) = Vw(ml) x K(T)
where
Vgi = volume of gas dissolved in condensate (STP)
K(T) = solubility of gas in water in ml (STP) per ml water at
one atmosphere partial pressure of the gas in question,
and at the temperature, T, of the condensor (condensate)
F = mole fraction of gas = concentration (ppm) x 10
PS is assumed to be ambient atmospheric pressure and equal for sample
entering the probe, condensor, and flowmeter. The total volume of gas
that the original sample contains is as follows:
v /i- ^ 273K 10 ml
vgt(»D = x v* (llters) x x ~
ws we
171
-------
The fractional loss of gas in the condensate is given by the ratio Vgi/Vgt
which, by combination of the above equations, yields the following:
Qgl i 8 x 10-* (FWS-FWC)
where Qgl is the fractional quantity of gas lost in the condensate.
It follows, therefore, that the relative error in determining gas concen-
trations due to loss by solution in condensed water depends only upon the
initial and final water mole fractions and upon the solubility of the gas
of interest. The relative error is independent of the sample size,
temperature, pressure, and the mole fraction of the gas of interest
(under the assumed sampling conditions).
The solubility of SC>2 being given in handbook tables as about 80 ml (STP)
per (ml water x atmosphere), the above equation predicts that the maximum
error in SC>2 determinations for 13% water vapor loading (the value pre-
vailing at the power plant site) is 0.83%. This prediction is consistent
with the experimental results reported in Paragraph 1.0.
3,0 METHOD 6 ANALYSIS OF SAMPLE DELIVERED BY REFRIGERATED CONDENSOR
AND PERMEATION DRYER SYSTEMS
A secondary objective of the Field Test Work Plan was to acquire as much
supportive data as possible. In particular, it was desirable that Method 6
analyses of the samples delivered by the two dryrsample sampling interface
systems under evaluation be directly compared to Method 6 analyses of
samples obtained through the manual probe which was installed specifically
for collecting Method 6 samples. Unfortunately, only one Method 6 im-
pinger train was available, which made it impossible to obtain samples
that correlated in time. However, the power plant stack gas S02 content
was quite stable with time during long periods of each day, justifying
serial collection of samples for this purpose. Consequently, when the
172
-------
work load permitted, the test described in the following paragraphs was
performed.
At the end of the power plant site evaluation Method 6 samples were
collected through the refrigerated condenser system and the reflux-probe-
permeation dryer sampling system (one each). The purpose of this test
was to acquire direct comparative data of the two instrument sampling sys-
tems against the manual Method 6 probe, which was almost entirely glass.
Unfortunately, a glass part of the Method 6 sampling train was accidentally
broken before the manual sample was acquired, making it impossible to
achieve the major experimental objective.
The Method 6 analysis of the samples collected indicated 157 ppm SC>2 for
each sampling system. Only one sample for each system was taken, giving
the results questionable significance; however, the Method 6 sample col-
lection technique and, therefore, the Method 6 test results, had become
reasonably reliable, improving the confidence level for single determi-
nationa. The lack of a large difference between the results for the two
radically different sampling systems was consistent with the generally
good agreement between the SC>2 values determined by the two (different
type) monitors throughout the test period. This result was also consis-
tent with the observation that both monitors gave similar calibration
results regardless of whether the zero and calibration gases were intro-
duced immediately before the analyzers, or within the heated sampling-
interface housings mounted on the stack.
It is concluded that no significant difference in S02 concentration exists
for samples obtained through refrigerated condensers and through permeation
dryers.
173
-------
4.0 INDEPENDENT VERIFICATION OF LINEARITY OF SEVERAL MONITORS
During the laboratory familiarization phase, a gas blending system was
used. This system is a duplicate of a system fabricated for the EPA under
separate contract. It was possible to make known blends of nitrogen (or
zero air) with tank gas to verify linearity of response (within several
percent, at least) of the various analyzers. Once linearity of any
analyzer was established, it was possible to determine the relative con-
centrations of all tank gases (within the validated range) with respect
to an arbitrarily chosen mixture. This technique provided an independent
means of assessing the quality of Method 6 determinations being performed
on tank gases as part of the training and familiarization work.
Specifically, leaks in the Method 6 sampling train produced some pre-
liminary analyses of tanks indicating concentration ratios that were
inconsistent with the ratios determined by linear instruments. Elim-
ination of the inconsistency by finding and correcting the leaks pro-
vided a high level of confidence in the absolute accuracy of the
Method 6 determinations by which the tank mixtures were eventually
calibrated.
Troubleshooting of the Meloy Analyzer was also greatly facilitated by use
of the gas-blending system for verification of linearity. The Meloy cir-
cuitry provided a logarithmic output covering many decades of sulfur con-
centration, followed by a linearized output of 0 to 1 ppm for the diluted
sample, corresponding to several hundred ppm S02 input to the Dyfusatron.
When intermittent failures plus abnormally large thermal errors developed,
it was possible to establish that the linearized output was still basic-
ally linear in spite of the instabilities of the output. The Meloy out-
put slope changed intermittently, but not the intercept or linearity.
Unfortunately, problems developed with the blender when the Houston-Atlas
System was being tested. This fact, coupled with the inability to employ
174
-------
Method 11 for absolute determinations of the concentrations of H^S in the
mixed tanks, resulted in serious compromises in the quality of the evalu-
ation testing of the Houston-Atlas System.
The gas blender was employed in calibration of the Dynasciences Model SS-
310 for use with the Disc Diluter. While the Disc Diluter Model SS-310
System could have been calibrated empirically (only), an absolute cali-
bration of the Model SS-310 coupled with the established dilution ratio
of the Disc Diluter made it possible to perform a Disc Diluter leakage
error test by comparing actual output versus theoretical output for a
known sample entering the Disc Diluter. (The results of this test are
given in Paragraph 5.0 of this Appendix).
Finally, the Beckman NDIR Analyzer had a linearized output, and it would
have been possible to adjust the linearizing board to provide exact agree-
ment with the three gases used for calibration and linearity testing.
However, this would have compromised the objectives of the evaluation,
since the normal method of adjustment utilizes the original nonlinear cal-
ibration curve included in the instruction manual of each NDIR analyzer.
Accordingly, the NDIR was linearized as it would normally be for a typical
customer. Any inaccuracy in the gases used by the Beckman Laboratory
would, therefore, cause nonlinearity in the output of the NDIR analyzer
evaluated. This factor may account for the failure of the NDIR to meet
the EPA-650/2-74-013 requirements in the pre-Claus Plant installation test,
but it is also possible that the accuracy with which the standard gases
were determined was marginal. This explanation is suggested by the failure
of all three monitors (for SOo) to pass this particular test, but insta-
bilities of the Meloy Defusatron (which affected two of the three
analyzers) clouded the issue.
175
-------
5.0 DISC DILUTEE SYSTEM CALIBRATION AND TESTING
Several significant tests and the results obtained are documented here
since they are believed to be more indicative of the performance of the
Disc Diluter than were those obtained in the field tests, during which
repeated mechanical drive failures occurred, resulting in termination of
the effort.
As discussed in Paragraph 4.0 of this Appendix, a gas blending technique
was employed to calibrate the 0-10 ppm range of the Dynascience Model SS-
310 on a calibration gas which was to be employed in field testing
(approximately 450 ppm S02 in N2). The absolute error of this calibration
was estimated to be less than 5%. With the aid of a stopwatch for timing
a number of revolutions, the rotational speed of the Disc Diluter was ad-
justed to the desired value. To avoid a time consuming, precise, mea-
surement of the volume of the holes in the disc, the dilution ratio
employed was based upon that given in the Monsanto Research Corporation
Instructions. The flow rate of carrier gas was verified for one setting
of the flowmeter by using the dry gas meter that was used for Method 6
testing. Assigning another probable error of about 5% in the dilution
ratio seems reasonable, resulting in a probable error in the Model SS-310
theoretical output of about 10%.
The calibration gas was then delivered to the Disc Diluter and the thrust
loading on the disc was increased in steps until further increases made
little difference in the output of the Model SS-310. It was found that
the speed of rotation decreased noticeably as the analyzer output reached
the zero-slope condition, but the speed control was reset to maintain a
constant dilution ratio. The Model SS-310 output increased with disc
tension and became less noisy until it reached a plateau.
176
-------
The result is interpreted to mean that the leakage of calibration gas out
of, and/or air into, the diluted sample became approximately zero as the
disc thrust loading was increased. It is considered significant that the
Model SS-310 output was approximately 10% below the theoretical value
under these conditions, which is probably within the uncertainty of the
theoretical value. Furthermore, the same result was obtained with the
probe out of the stack at the power plant site, the only additional cor-
rection required being for the effect of barometric pressure (altitude)
upon the system calibration.
When the thrust loading appeared to be correctly set, it was found that
the disc friction produced heating that resulted in excessive thrust
loading after operating for about fifteen minutes. At this time, the
disc tension was backed off to obtain smooth operation, without a sig-
nificant increase in leakage. This suggested that it might be necessary
to set the disc tension on the low side when out of the stack to achieve
proper tension when it reached stack gas temperature. It was believed
that a reasonable amount of trial-and-error adjustment at the test site
would suffice.
The calibration and linearity test was performed with the Disc Diluter
adjusted as described above, but with the Model SS-310 span readjusted
to provide a 0-500 ppm range for the system. Under these conditions
the system passed the test.
The next test was intended to verify that the effect of temperature upon
dilution ratio was that predictable from the gas law when leakage was
eliminated. A rather peculiar thermal effect was shown by curves in-
cluded in the Monsanto Research Instructions. The assumption here was,
of course, that leakage accounted for the effect observed in the tests
conducted by Monsanto, and that leakage effects would be minimized in
the current test.
177
-------
With the Disc Diluter operating in a stable manner, an ordinary laboratory
"heat gun" was used to slowly elevate the temperature of the entire Disc
Diluter probe, while the carrier air flow rate (measured at room tempera-
ture) was maintained constant. After heating, the probe temperature was
allowed to stabilize for a few minutes to eliminate thermal gradients.
The output decreased in accordance with the gas law prediction within the
limits of uncertainty of the actual probe temperature, which could not be
precisely determined with available equipment. The law predicts that the
effective dilution ratio decreases with increasing sample (or calibration
gas) temperature when the carrier flow rate (measured at ambient tempera-
ture) remains constant. The relationship is as follows:
T
D(T) = D(A) x ~
P
where D(T) = dilution ratio for sample at temperature Tp
D(A) = dilution ratio for sample at ambient temperature
Ta = ambient temperature (at which probe dilution factor is
known)
T = temperature of probe and of sample or calibration gas
Finally, it should be noted that this change in dilution ratio applies
when the probe and monitor are calibrated with the probe at ambient
temperature and then used at a different temperature. Provided the cal-
ibration gas flow rate and the probe heat transfer characteristics are
such that the calibration gas reaches stack (and probe) temperature be-
fore it enters the disc, the same dilution ratio change will occur for
both. For in-situ calibration, therefore, the monitor span adjustment
that calibrates the system correctly on calibration gas will provide
correct analysis of the sample with no further adjustment or correction.
For such in situ calibration, therefore, the above temperature-dilution
178
-------
ratio function simply predicts the difference in sensitivity for stack
gas temperature variations after calibration at any particular stack
temperature, Ta. This entire discussion assumes that there are no
additional errors due to leakage of the disc, or due to changes in
leakage with temperature.
6.0 PRELIMINARY METHOD 6 CORRELATION TEST AND RESULTS
This section contains the correlation test results obtained with the Disc
Diluter and two other systems in three preliminary tests performed at the
Power Plant site on 26 November 1974. The digital data acquisition system
was down at the time due to a failure of the E-A D-2020 Data Logger. One
of two dual Texas Instruments, Inc., analog recorders was also down due to
a stripped chart drive gear train. The Meloy analyzer was also down.
The Disc Diluter drive train had just been repaired, and it lasted long
enough to permit performance of three Method 6 correlation tests. The
raw data are presented in Table 18. These results were not included in
the body of the Final Report because extensive changes and repairs were
made before formal testing was initiated.
It is apparent that the Disc Diluter performance was well within the re-
quirements of EPA-650/2-74^013, even if the data from the third run is
included in forming an "average results." The downware trend of the Disc
Diluter Model SS-310 analog record compared to the records for the other
two systems indicated that failure began shortly before the third Method
6 sample was collected.
7.0 COMPUTER PROGRAMS USED IN REDUCTION OF DATA
The original Work Plan included generation of a program for the reduction
of all data from punched paper tape records. The effort was abandoned
when it became apparent that equipment failures were compromising the
cost-effectiveness of this approach. The preliminary effort did not
179
-------
Table 18. RESULTS OF PRELIMINARY METHOD 6 CORRELATION TESTING AT
POWER PLANT. These data for the Disc Diluter are the
only results obtained.
Correlation
Test Run
Number
1
11:30 to
12:00
2
17:45 to
16:15
3
19:08 to
19:38
Refrigerated
Condenser,
Dynasciences
Model SS-330
ppm S02
226
221
218
Beckman
Reflux Probe,
Permeation Dryer,
Model 865 NDIR
ppm S02
228
238
230
Permeation
Diluter,
Flame Photometric
(Meloy)
ppm S02
Down
Down
Down
Disc Diluter ,
Dynasciences
Model SS-310
Wet Basis
183
180
158*
Dry Basis
205
202
177*
Method 6
Result
ppm SOo
(Dry Basis)
204
212
204'
co
o
*The disc appeared to be leaking during this run, after approximately 8 hours of continuous
operation, but subsequent examination revealed that the drive train had fractured on an
angle, resulting in "slip-clutch"-type operation.
-------
result in a program worthy of reporting. However, several computer pro-
grams that were generated for use in data reduction utilizing manual entry
techniques may be of general interest. The following paragraphs briefly
describe the application of each of the three programs that were used
with the Beckman time-shared system. The following pages are copies of
teletype program listings. It should be noted that these programs cal-
culate results using the summation of absolute values. Refer to Appendix
F for a detailed discussion of this interpretation of the instructions of
EPA-650/2-74-013.
181
-------
METH4, METH6, AND METH11
These are used for calculating concentrations of water, S02, and H2S,
respectively, from data obtained from manual sampling and volumetric
and/or titrimetric determinations, using EPA compliance test Methods
A, 6, and 11, respectively.
METH4 14:44 04/30/75 WEDNESDAY
100
110
120
130
140
150
160
170
180
190
200
210
220
230
240
250
260
270
280
290
300
310
320
330
340
350
360
370
380
390
DIM X(25),Y<25),P(25>,T<25>,F(25>>S(25>
DIM G(25),W(25),B(25)
PRINT "H0W MANY DETERKINATI0NS ?"
INPUT N
F0R 1 « 1 T0 N
PRINT "ENTER GAS METER DATA"
INPUT X(i>«YCX>«PjT(I>
PRINT "ENTER IMPINGER DATA"
INPUT FCX)«S(I)
NEXT I
F0R I = 1 T0 N
T(I) * TCX>+460
U(I) = 17.71*(X(I)-Y(I))*P(I)/T(I>
W = .0474*C(I)
PRINT
NEXT I
:TEST N0. §*
: V GAS STD = »».»»»
i V H20 STD = .#****
: FRACT H20 = .**#*#
: C H2B = ###### P.p
END
FT3
.M.
182
-------
NETK6
14:46 04/30/75 WEDNESDAY
100
110
120
130
140
150
160
170
180
190
200
210
220
230
240
250
260
270
280
290
300
310
320
330
340
350
360
370
380
390
400
410
DIM XC25
DIM A
(25
DIM D(25
DIM H(25
PRINT
INPUT
FCR I
PRINT
INPUT
PRINT
INPUT
NEXT
FKR I
T d )
Gd )
C ( I )
D(I )
Ed)
PRINT
PRINT
PRINT
PRINT
PRINT
PRINT
PRINT
NEXT
JTEST
8 V
C
: C
: C
END
)
)
)
)
»
t
t
t
"HfcW
N
=
1
Y<25)
C(25)
E(25)
J(25)
MANY
Tfc N
t
t
P
F
(25
(25
>«T(25>«G(25)jS<25>«B<25>»MC25>jQ<25>
)
DETERMINATIONS ?"
"ENTER GAS
X(
I
)
>Y(I )
>
P
METER DATA"
(
I
"ENTER TITRAT
S (
I
=
= T
3 |
= (
= C
= D
US
US
US
US
us
I
Ne
I
1
<
7
7
<
<
>
1
>K( I >
T0 N
)+46C
t
•71*(X<
•05E-5
I
I
)
)*1 .60
)*3.7
ING 360
ING 370
ING 380
ING 390
ING 400
•
GAS
S02
»«
STD -
B
I
*
1
<
)
I
)
,T(1 )
I0N DATA"
)
*iJ(I).»M(I)sQ(I>.»A(I)
-YU»*P(I>/T(I>
(S
(
I>-Kd)-B(I)+H(I»*M(I>*QU)/A(I)/G(I>
86E10
7047E-4
/
t
»
t
j
I
G
C
(
(
D<
E
*
(
t
I
I
I
I
.
>
)
)
)
i
= *§»»»«»
502 = «»#»»»»»
S02
- tiiit
P
•
»* FT3
LB/FT3
# UG/M3
P.M.
183
-------
METH11 14:49 04/30/75 WEDNESDAY
100 DIM X(25)..YC25),P(25>.,TC25),G(25>,S<25>,B(25)>C<25>
110 DIM D<25>,E<25>,H(25>,J(25>
120 PRINT 'ENTER STANDARD DATA1
130 INPUT A*F,M,Q
UO U = 2-04*A/F
150 0 = »1*U
160 R = 0*Q/M
170 PRINT
180 PRINT USING 190,U,0
190 :THI0SULFATES ARE *.***»* N, ******* N
200 PRINT
210 PRINT USING 220*R
220 :I0DINE IS ******* N
230 PRINT
240 PRINT 'H0W MANY DETERMINATI0NS ?'
250 INPUT N
260 F0R I = 1 T0 N
270 PRINT "ENTER GAS METER DATA"
280 INPUT X«YCl')*P*T
290 PRINT "ENTER TITRATI0N DATA"
300 INPUT SU>,HU),B,J
310 NEXT I
320 F0R I = 1 T0 N
330 T = TCD+460
340 G(I) = 17.71*(X
350 CCI) = (3.75661E-5)*(H(I)*R-S(I)*0-JG(I)
410 PRINT USING 480^C
430 PRINT USING 500*E(I)
440 PRINT
450 NEXT I
460 :TEST N0. ft
470 : V GAS STD = ##.### FT3
480 : C H2S = .####f## LB/FT3
490 : C H2S = ********** UG/M3
500 : C K2S = «HHi P.P.M.
510 END
184
-------
EPA DATA
This is used for entering data from correlation tests and zero and calibration
drift tests.
EPADATA 7:46 05/14/75 WEDNESDAY
100 U1KENSI0K R(50),FI1(5,50)
110 100 F£Rt-iAT(33H N£. 0F R/S, N0» CF It^STRUMENTS . ? )
120 101 F0RKATUOK ENTER R'S)
130 102 F0RMAT(5(F5.0,3X))
140 103 F0RhAT(10H EixTLR U'S)
150 104 FERNAK12H IlvSTRUMEM ,11)
160 WRITE (6,100)
170 READ <5/*> tv/JJ
180 WRITE (6,101)
190 READ (5**)
-------
EPASTAT "
This was designed for calculating average deviations, 95% Confidence
Intervals, and Relative Accuracy for correlation tests. This program
was also used for calculating average drifts and 95% Confidence Inter-
vals for zero and calibration drift data. The latter calculation was
accomplished by entering a series of zeros in EPA DATA parallel to
individual drift values.
EPASTAT 7:39 05/14/75 WEDNESDAY
100 DIMEhSIBK R (50 ) ,'r i-i(5, 50 ) ,ELEM(5, 50 ) ,LLEH2 (5, 50 )
11 0 DIHENS I fcK TERM1 (5 ) * TERh2 (5 ) , TERW3 (5 ) , T 1 SUM (5 ) » T2SUM< 5 )
120 DINElMSIfcu FUWCT(S)
130 100 F0RMAT(5(F5.0,3X))
140 101 F6RMAT
190 CALL tJPEiv (l^'CHfcLLA'j'INPUT* )
200 READ (1) h*JJ*«FK(<;jI>/jBl jJ(j>jR(I>,l8i ,K>
210 CALL CL0SE (1)
260 D0 20 1=1,K
270 D0 20 J«1/JJ
280 ELEM(I,J) =0.
290 20 CBKTINUE
300 D0 30 1 = 1 ,<^
310 D0 30 J=l>oJ
320 ELEM = ELEM(J,I)*ELEM
-------
EPASTAT 7:39 05/14/75 WEDNESDAY
(Continued)
450
470
460
490
500
510
520
530
540
550
560
570
580
590
610
36 CiilVriKUE
D0 50 1=
D0 40 o
TISUH(J)
T2SUM(o)
40 CONTINUE
II- IRST/ILAST
= 1/JJ
= TISUi',(0)+ELEM(0/ I)
= T25UK(J)+ELEM2(J/I)
RSUM = RSUK+RU)
50 C0KTINUE
Dfc 60 J
TEP.N1 (J)
TERK2 ( J )
TERK3 ( J )
FUKCT(J)
60 C0KTII\iUE
WRITE (6
= l/oo'
= TlSUl-i(o)/FNUH
= PREFIX*5QRT(F{vUK*T2SUK(J)-(TlSUM(0)*TlSUK(
= RSUN/FNUM
= 100.*(TERM1 (o')+TERh2(0))/TERK3(o)
(
,201) IFIR5T/ILAST
))
615 WRITE (6/200) TERH3U)
620 VR1TE (6/202) (TERK1 (c :) *TERh2 (0 ) /FUKCT ( J )/ u= 1 / Oo )
630 ufci T£ 35
640 STC-P
650 EKD
187
-------
APPENDIX F
COMMENTS ON INTERPRETATION OF CALCULATIONS AND
REPORTING INSTRUCTIONS OF EPA-650/2-74-013
188
-------
APPENDIX F
In reviewing all data calculations and reported results just prior to
publication of this Final Report, it was discovered that a potential
ambiguity in interpretation of the guidelines exists. Pages 56 through
60 of Report EPA-650/2-74-013 provide guidelines for calculations, data
analysis, and reporting.
Equations (E-l) and (E-2), as defined, apply to general statistical anal-
ysis of a number of individual values. Results reported as the sum of
the absolute mean value and the 95% confidence interval as required in
the guidelines are not ambiguous if all of the individual values are of
the same algebraic sign. However, a question arises when these equations
are.applied to a number of individual values having both positive and
negative signs, as may be the case for accuracy, calibration error, and
both two- and twenty-four-hour zero and calibration drifts, all of which
are calculated using deviation values in accordance with the instructions.
In these cases, two interpretations are possible, depending upon whether
absolute values are to be summed or whether the absolute value of a sum
is to be used.
Interpretation of Equations (E-l) in accordance with normal mathematical
conventions would result in a mean deviation, X, formed by taking the
algebraic sum of all deviations and dividing by the number of data points.
With this interpretation X may be either positive or negative. Similarly,
in Equation (E-2), used for calculating the 95% confidence interval, the
summation of the deviation (individual values) would normally be alge-
braic, with retention of sign (until the summation is squared). The
summation in both equations could be zero if cancelling deviations,
either large or small, are involved.
189
-------
It is believed that this interpretation is mathematically correct, and
was probably intended in the original draft. The final draft, however,
supports a different interpretation as is further explained in the
following paragraphs:
This interpretation of Equations (E-l) and (E-2) as applied to the cal-
culation of the mean value and 95% confidence interval of a group of
deviation values results in smaller means and larger confidence inter-
vals than does the other possible interpretation, which is discussed
further below. However, the reported values, which are the sura of the
absolute values of the mean and the 95% confidence interval, are not sig-
nificantly different when calculated by the two different interpretations.
The second possible interpretation is that which was employed in all cal-
culations reported in the body of the Final Report, and the data and cal-
culations shown in Appendices D and E. This interpretation of the guide-
lines is suggested by the reason for revision of reporting requirements
stated on pages 6 and 7 of EPA-650/2-74-013. The revision eliminated the
separate reporting of the mean value (with sign) and the 95% confidence
interval and requires reporting of the sum-of^the-mean and the 95%
confidence interval. This revision is said to allow greater flexibility
in the application of the guidelines to specific test results without
increasing the absolute uncertainty (accuracy) allowed. Furthermore,
the reporting requirements of paragraphs 6.2.1 through 6.2.6 of EPA-650/2-
74-013 all read: "report the sum of the absolute mean value and the 95%
confidence interval as a percentage of ". The interpretation given
in this instruction is to form the summations of the individual values
using their absolute values, in both Equation (E-l) and (E-2). The value
reported is, then, the sum of this mean and the 95% confidence interval,
both of which are positive. In the reported results, all of which were
calculated using this interpretation, the mean values are larger and the
190
-------
confidence intervals are smaller than they would be if the first interpre-
tation had been used. The sum of the two, however, which is the only
value reported in the body of the Final Report,.does not differ sig-
nificantly from that which would have been calculated under the other
interpretation of the reporting instructions. In particular, the results
are mathematically identical whenever all deviations are of the same sign
(either all positive or all negative). As discussed above, the other in-
terpretation of the instructions is to form the algebraic sum of the devi-
ations in both equations, then to report the absolute value of the mean
deviation plus the absolute value of the 95% confidence interval (which
is inherently a (±) interval).
In conclusion, it seems appropriate to present a tabulation of some of
the results obtained when calculated by the two different methods. Since
most Power Plant data happened to have deviations of only one sign, none
of the results for that test are affected significantly, and more than
80% are not affected at all. At the Glaus Plant, however, erratic be-
havior of several systems resulted in large deviations of both signs.
All such results were recalculated to verify that no gross discrepancies
would occur if the other interpretation of the instructions had been used.
The results, given in Table 19. are typical of those found for all four
relative accuracy data groups for the three SC>2 monitors, and clearly
indicate that the implications of the test results are not altered by
the choice of interpretation of the calculation and reporting instruc-
tions. Similar differences were found for the first data set for the
twenty-four-hour zero drift results, which changed by less than 6% of
the value reported except for tbe Houston Atlas calibration drift value,
which would have been 47% instead of 61% if calculated by the other
approach. For such large errors, this shift is also considered to be
insignificant.
191
-------
Table 19. EXAMPLE RESULTS FOR GLAUS PLANT TESTS. Examples of mean values, 95% Confidence
intervals, and accuracies (relative) calculated using two different interpre-
tations of the instructions given in EPA-650/2-74-013.
Data
Group
No.
Calculated
Results
Values Calculated for Report (%)
Values Calculated
by other Interpretation (%)
Mean
Accuracy
(Deviation)
Beckman
Reflux Probe,
Permeation Dryer,
Model 865 NDIR
Permeation
Diluter,
Dynasciences
Model SS-310
Permeation
Diluter
Flame Photometric
(Meloy)
Houston Atlas
Diluter and
H2S Monitor
Vs. PGC (only)
2.6
-1.4
10.9
-10.8
16.2
-10.3
35.8
-26.8
95%
Confidence
Interval
1.3
5.8
11.8
+2.2
+6.0
8.3
+12.2
±21.1
Reported
Accuracy
(Sum of
Absolutes)
3.9
16.7
27.1
3.6
16.8
44
23.5
48
Mean
Accuracy
(Deviation)
5.8
11.0
16.6
-1.3
+8.9
26
+16.3
-14.8
95%
Confidence
Interval
3.9
4.3
10.2
±6.0
±6.7
12
±10.7
±21.4
Reported
Accuracy
(Sum of
Absolutes)
9.7
15.3
26.8
7.3
15.6
38
27.0
36.2
-------
As a final comment, there is considerable interpretive advantage in
reporting the mean values (with sign) and the 95% confidence interval
separately. In particular, it is apparent from Table H-l that inter-
mittent shifts in the permeation diluter temperature control caused the
mean calibration of both the Dynasciences Model SS-310 and the Meloy Flame
Photometer to shift from negative to positive between data groups 1 and 2.
The 95% confidence intervals were reasonable for both monitors in spite of
the abnormal effect of ambient temperature upon the Meloy permeation
diluter. In addition, both the mean deviation and the 95% confidence
interval were about twice as large for the Meloy analyzer as for the
Model SS-310, reflecting the fact that loss of PMT temperature control for
the Meloy Flame Photometric Analyzer introduced a second, abnormal, de-
pendence upon temperature. Similarly, the large negative mean deviations
of the Houston Atlas compared to the PGC simply reflect the fact that the
two were not standardized on the same gas, while the relatively constant
95% confidence interval for the two data groups (+21%) indicates that
repeatable results were being obtained, even though the randon variations
of the Houston Atlas compared to the PGC were rather large. It should be
noted that there is no valid basis for assigning these accuracy figures
to the Houston Atlas alone, since the performance of the PGC was not
independently assessed.
193
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-76-17 ]_
2.
3. RECIPIENT'S ACCESSION'NO.
4. TITLE AND SUBTITLE EVALUATION OF MONITORING SYSTEMS FOR
POWER PLANT AND
SULFUR RECOVERY PLANT EMISSIONS
5. REPORT DATE
June 1976
6. PERFORMING ORGANIZATION CODE
7.AUTH0R(S) Maibone w> Greene
Robert L. Chapman
Samuel f!. Creasnn
R. Neal Harvey
Glen A. Heyman
Wi111 am R Ppgygrm
8. PERFORMING ORGANIZATION REPORT NO.
FR-2678-102
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Beckman Instruments, Inc.
Advanced Technology Operations
1630 S. State College Blvd.
Anaheim, CA 92806
10. PROGRAM ELEMENT NO.
1AA010
11. CONTRACT/GRANT NO.
68-02-1743
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory
Office of Research $ Development
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final 6/74 - 6/75
14. SPONSORING AGENCY CODE
EPA-ORD
15. SUPPLEMENTARY NOTES
is. ABSTRACT This project was conducted to evaluate a number of commercially available
extractive-type sampling and monitoring systems for monitoring sulfur dioxide and
hydrogen sulfide source emissions. Evaluation testing was performed at a Fossil-
Fuel-Fired Power Plant and at a Claus Sulfur Recovery Plant to obtain representative
ranges of stack gas temperature, water and particulate loading, and concentrations
of S02 and I^S. Tests were performed to determine Error, Relative Accuracy, Two-
and Twenty-Four-Hour Zero and Calibration Drifts, Response Time, and Operational
Period in accordance with published EPA guidelines. The performance in each test
was judged against the published EPA performance criteria (EPA-650/2-74-013).
The detailed field test results, the complete Work Plan, sampling interface drawings,
the results of evaluations of Compliance Test Methods 6 and 11, miscellaneous obser-
vations and results, and a discussion of the calculation and reporting instructions
of the EPA Guidelines are given in the appendices. A detailed description of the
systems evaluated, summaries of the field test results, and relevant comments con-
cerning the results are given in the body of the report.
Because the sites chosen for evaluation testing provided wide ranges of sample temper-
ature, solids and water loading, and S02 concentration, the results are considered
to be relevant to most stack-gas monitoring problems. Results at the Claus Plant
site were somewhat compromised by equipment
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
*Air pollution
*Flue gases
*Gas sampling
*Sulfur dioxide
*Hydrogen sulfide
*Evaluation
b.IDENTIFIERS/OPEN ENDED TERMS C. COSATI Held/Group
13B
21B
14B
07B
3. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
1S. SECURITY CLASS (This Report/
UNCLASSIFIED
21. NO. OF PAGES
202
20. SECURITY CLASS (This page)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
194
------- |