July 2004
Environmental Technology
Verification Report
Thermo Electron Corporation
Model 17C Ammonia Analyzer
Prepared by
Battelle
Battelle
The B usincss of Innovation
In collaboration with the
U.S. Department of Agriculture
Under a cooperative agreement with
&EPA U.S. Environmental Protection Agency
ETV ElV ElV
-------
July 2004
Environmental Technology Verification
Report
ETV Advanced Monitoring Systems Center
Thermo Electron Corporation
Model 17C Ammonia Analyzer
by
Ken Cowen
Ann Louise Sumner
Amy Dindal
Karen Riggs
Zack Willenberg
Battelle
Columbus, Ohio 43201
and
Jerry Hatfield
Richard Pfieffer
Kenwood Scoggin
USDA National Soil Tilth Laboratory
Ames, Iowa 50011
-------
Notice
The U.S. Environmental Protection Agency (EPA), through its Office of Research and
Development, has financially supported and collaborated in the extramural program described
here. This document has been peer reviewed by the Agency. Mention of trade names or
commercial products does not constitute endorsement or recommendation by the EPA for use.
11
-------
Foreword
The U.S. Environmental Protection Agency (EPA) is charged by Congress with protecting the
nation's air, water, and land resources. Under a mandate of national environmental laws, the
Agency strives to formulate and implement actions leading to a compatible balance between
human activities and the ability of natural systems to support and nurture life. To meet this
mandate, the EPA's Office of Research and Development provides data and science support that
can be used to solve environmental problems and to build the scientific knowledge base needed
to manage our ecological resources wisely, to understand how pollutants affect our health, and to
prevent or reduce environmental risks.
The Environmental Technology Verification (ETV) Program has been established by the EPA to
verify the performance characteristics of innovative environmental technology across all media
and to report this objective information to permitters, buyers, and users of the technology, thus
substantially accelerating the entrance of new environmental technologies into the marketplace.
Verification organizations oversee and report verification activities based on testing and quality
assurance protocols developed with input from major stakeholders and customer groups
associated with the technology area. ETV consists of seven environmental technology centers.
Information about each of these centers can be found on the Internet at http://www.epa.gov/etv/.
Effective verifications of monitoring technologies are needed to assess environmental quality
and to supply cost and performance data to select the most appropriate technology for that
assessment. Under a cooperative agreement, Battelle has received EPA funding to plan,
coordinate, and conduct such verification tests for "Advanced Monitoring Systems for Air,
Water, and Soil" and report the results to the community at large. Information concerning this
specific environmental technology area can be found on the Internet at
http ://www. epa. gov/etv/centers/center 1. html.
111
-------
Acknowledgments
The authors wish to acknowledge the support of all those who helped plan and conduct the
verification test, analyze the data, and prepare this report. We would like to thank
Ernie Bouffard, Connecticut Department of Environmental Protection; Rudy Eden, South Coast
Air Quality Management District; Roy Owens, Owens Corning; Jim Homolya, U.S.
Environmental Protection Agency; Bruce Harris, U.S. Environmental Protection Agency; and
Lowry A. Harper, U.S. Department of Agriculture, for their careful review of the verification
test/QA plan and this verification report. We also thank Amy Morrow and Diane Farris of the
U.S. Department of Agricuture Analytical Laboratory for their assistance in performing the
reference sample analysis.
iv
-------
Contents
Page
Notice ii
Foreword iii
Acknowledgments iv
List of Abbreviations viii
1 Background 1
2 Technology Description 2
3 Test Design and Procedures 3
3.1 Test Design 3
3.2 Site Descriptions 4
3.2.1 Site Description—Phase I 4
3.2.2 Site Description—Phase II 5
3.3 Test Procedures 5
3.3.1 Accuracy, Linearity, Precision, and Response Time 5
3.3.2 Calibration and Zero Drift 5
3.3.3 Interference Effects 6
3.3.4 Comparability 6
4 Quality Assurance/Quality Control 8
4.1 Equipment Calibrations 8
4.1.1 Reference Method Sampling Equipment 8
4.1.2 Analytical Equipment 8
4.1.3 Meteorological Equipment 9
4.1.4 Ammonia Dilution System 9
4.2 QC Samples 9
4.2.1 Field Blanks 9
4.2.2 Denuder Breakthrough Checks 11
4.2.3 Duplicate Samples 14
4.2.4 Laboratory Blanks 15
4.2.5 Calibration Checks 15
4.2.6 Gas Standard Dilution Checks 16
4.3 Audits 17
4.3.1 Performance Evaluation Audit 17
4.3.2 Technical Systems Audit 17
4.3.3 Audit of Data Quality 18
4.4 QA/QC Reporting 18
4.5 Data Review 18
v
-------
5 Statistical Methods and Reported Parameters 19
5.1 Relative Accuracy 19
5.2 Linearity 19
5.3 Precision 19
5.4 Response Time 20
5.5 Calibration and Zero Drift 20
5.6 Interference Effects 20
5.7 Comparability 20
6 Test Results 21
6.1 Relative Accuracy 23
6.2 Linearity 25
6.3 Precision 26
6.4 Response Time 26
6.5 Calibration and Zero Drift 28
6.6 Interference Effects 29
6.7 Comparability 30
6.8 Ease of Use 30
6.9 Data Completeness 34
7 Performance Summary 35
8 References 37
Appendix: Model 17C Checklist A-l
Figures
Figure 2-1. Model 17C Ammonia Analyzer 2
Figure 3-1. Phase I Test Site 4
Figure 3-2. Phase II Test Site 5
Figure 3-3. Reference Method Sampling Cartridge 6
Figure 4-1. Denuder Breakthrough During Phase I as a Function of
Integrated Ammonia Concentration 12
Figure 4-2. Denuder Breakthrough During Phase II as a Function of
Integrated Ammonia Concentration 13
Figure 4-3. Analysis of Diluted Ammonia Standards Using the
Denuder Reference Method 16
Figure 6-1 Phase I Meteorological Conditions and
Model 17C Ambient NH3 Measurements 21
Figure 6-2 Phase II Meteorological Conditions and
Model 17C Ambient NH3 Measurements 22
Figure 6-3. Phase I Accuracy Results for the Model 17C 23
Figure 6-4. Phase II Accuracy Results for the Model 17C 24
Figure 6-5. Results of Linearity Check of the Model 17C During Phase I 25
Figure 6-6. Results of Linearity Check of the Model 17C During Phase II 26
Figure 6-7. Comparison of Ambient Reference Measurements with
Averages from the Model 17C During Phase I 31
vi
-------
Figure 6-8. Comparison of Ambient Reference Measurements with
Averages from the Model 17C During Phase II 31
Figure 6-9. Scatter Plot of Model 17C Results versus Ambient Reference
Measurements During Phase I 32
Figure 6-10. Scatter Plot of Model 17C Results versus Ambient Reference
Measurements During Phase II 32
Tables
Table 4-1. Minimum Detectable Ambient NH3 Concentrations During Phase I 10
Table 4-2. Minimum Detectable Ambient NH3 Concentrations During Phase II 11
Table 4-3. Denuder Breakthrough Checks During Phase I 13
Table 4-4. Denuder Breakthrough Checks During Phase II 14
Table 4-5. Duplicate Reference Method Samples 15
Table 4-6. Data Recording Process 18
Table 6-1. Relative Accuracy Results 24
Table 6-2. Calculated Precision of the Model 17C 27
Table 6-3. Response Time Determinations 27
Table 6-4. Calibration and Zero Checks During Phase I 28
Table 6-5. Calibration and Zero Checks During Phase II 29
Table 6-6. Interference Effect Evaluation 29
Table 6-7. Activities Performed During Phase I 33
Table 6-8. Activities Performed During Phase II 34
Table 7-1 Model 17C Performance Summary 36
vii
-------
List of Abbreviations
AFO
animal feeding operation
AMS
Advanced Monitoring Systems
CI
confidence interval
DL
detection limit
EPA
U.S. Environmental Protection Agency
ETV
Environmental Technology Verification
FIA
flow injection analysis/analyzer
kg
kilogram
L
liter
Lpm
liter per minute
mL
milliliter
Hg
microgram
|j,m
micrometer
mg
milligram
mm
millimeter
NIST
National Institute of Standards and Technology
nh3
ammonia
nh4
ammonium
NO
nitric oxide
no2
nitrogen dioxide
Nt
nitric oxide, nitrogen dioxide, and ammonia
ppb
part per billion
%D
percent difference
QA
quality assurance
QC
quality control
QMP
quality management plan
RA
relative accuracy
RPD
relative percent difference
RSD
relative standard deviation
SD
standard deviation
TSA
technical systems audit
USD A
U.S. Department of Agriculture
viii
-------
Chapter 1
Background
The U.S. Environmental Protection Agency (EPA) supports the Environmental Technology
Verification (ETV) Program to facilitate the deployment of innovative environmental
technologies through performance verification and dissemination of information. The goal of the
ETV Program is to further environmental protection by accelerating the acceptance and use of
improved and cost-effective technologies. ETV seeks to achieve this goal by providing high-
quality, peer-reviewed data on technology performance to those involved in the design,
distribution, financing, permitting, purchase, and use of environmental technologies.
ETV works in partnership with recognized testing organizations; with stakeholder groups
consisting of buyers, vendor organizations, and permitters; and with the full participation of
individual technology developers. The program evaluates the performance of innovative
technologies by developing test plans that are responsive to the needs of stakeholders,
conducting field or laboratory tests (as appropriate), collecting and analyzing data, and
preparing peer-reviewed reports. All evaluations are conducted in accordance with rigorous
quality assurance (QA) protocols to ensure that data of known and adequate quality are
generated and that the results are defensible.
The EPA's National Exposure Research Laboratory and its verification organization partner,
Battelle, operate the Advanced Monitoring Systems (AMS) Center under ETV. The AMS
Center, in collaboration with the U.S. Department of Agriculture's (USD A) National Soil Tilth
Laboratory, recently evaluated the performance of the Thermo Electron Corporation Model 17C
ammonia (NH3) analyzer.
1
-------
Chapter 2
Technology Description
The objective of the ETV AMS Center is to verify the performance characteristics of
environmental monitoring technologies for air, water, and soil. This verification report provides
results for the verification testing of the Model 17C. The following is a description of the Model
17C, based on information provided by the vendor. The information provided below was not
subjected to verification in this test.
The Model 17C (Figure 2-1) is a chemiluminescence analyzer that uses the reaction of nitric
oxide (NO) with ozone (03) to measure NH3 concentrations in the atmosphere. A sample is
drawn into the Model 17C by an external pump. After the sample reaches the reaction chamber,
it mixes with 03, which is generated internally. The reaction of NO with 03 produces a
characteristic luminescence with an intensity proportional to the concentration of NO. Light is
emitted when electronically excited nitrogen dioxide (N02) molecules decay to lower energy
states. The light emission is detected by a photomultiplier tube, which in turn generates an
electronic signal. The signal is processed by the microcomputer into a NO concentration
reading.
To measure the NO, N02, and NH3 concentrations, N02 and
NH3 are transformed to NO in a stainless steel converter heated
to approximately 775°C before reaching the reaction chamber.
Upon reaching the reaction chamber, the converted molecules,
along with the original NO molecules, react with 03. The
resulting signal represents the total NO, N02, and NH3 reading
(Nt). Separately, N02 is transformed into NO in a molybdenum
converter heated to approximately 340°C. The NO, plus
converted N02 concentrations are measured as NOx. The N02
concentration is determined by subtracting the signal
obtained in the NO mode from the signal obtained in the Figure 2-1. Model 17C Ammonia
NOx mode. The NH3 concentration is determined by Analyzer
subtracting the signal obtained in the Nt mode from the signal obtained in the NOx mode.
NO, N02, and NH3 concentrations are displayed on the front panel of the Model 17C as analog
output. The Model 17C has a 1-part-per-billion (ppb) detection limit and operates manually or
automatically, with a sample flow rate of 0.6 liters per minute (Lpm). The Model 17C requires
500 Watts of power from 90 to 110, 105 to 125, or 210 to 250 volts alternating current. The
Model 17C consists of two components: the analyzer and the converter. The analyzer
dimensions are 426 millimeters (mm) by 219 mm by 584 mm, and the converter dimensions are
426 mm by 175 mm by 389 mm. The analyzer weighs 27 kilograms (kg), and the converter
weighs 9 kg. The Model 17C costs about $17,000.
2
-------
Chapter 3
Test Design and Procedures
3.1 Test Design
Livestock agriculture is thought to be the primary source of atmospheric NH3 in the United
States and accounts for approximately 70% of NH3 emissions in the United States.(1) As a result,
a means to accurately quantify these emissions is needed. The objective of this verification test
was to verify the Model 17C's performance in measuring gaseous NH3 in ambient air at animal
feeding operations (AFOs).
This verification test was conducted according to procedures specified in the Test/QA Plan
for Verification of Ambient Ammonia Monitors at Animal Feeding Operations,(2) with the
exception of six deviations that are addressed later in this report. The verification test was
conducted in two phases, each at separate AFOs. The first phase of testing was conducted
between September 8 and October 3, 2003, at a swine finishing farm near Ames, Iowa. The
second phase was conducted between October 20 and November 14, 2003, at a cattle feedlot in
Carroll, Iowa. These sites were selected to provide realistic testing conditions, which were
expected to exhibit a wide range of NH3 concentrations during the test periods.
The verification test was designed to evaluate the following performance parameters:
¦ Relative accuracy
¦ Linearity
¦ Precision
¦ Response time
¦ Calibration/zero drift
¦ Interference effects
¦ Comparability
¦ Ease of use
¦ Data completeness.
During each phase of the verification test, the Model 17C response to a series of NH3 gas
standards of known concentration was used to quantify relative accuracy (RA), linearity,
precision (repeatability), and calibration/zero drift. The Model 17C response time, the time to
reach 95% of the stable signal, was also assessed during the delivery of the NH3 standards.
During Phase n, interference effects were quantified from the Model 17C response to various
chemical species that may be present at AFOs; the potential interferent gases were delivered
both in the presence and absence of NH3. The Model 17C response to ambient air was also
evaluated during both phases as the comparability to simultaneous determinations by an ambient
NH3 reference method (acid-coated denuders). Additionally, the ease of use of the Model 17C
3
-------
was evaluated based on operator observations. Data completeness was determined based on the
amount of data collected as a percentage of the amount of data that could have been collected.
3.2 Site Descriptions
The Model 17C was installed at the Phase I and II testing locations by a vendor representative.
Battelle and USDA staff worked with the vendor representative to establish procedures for
operating the Model 17C during this verification test. The vendor representative trained Battelle
and USDA staff to check several instrument parameters to verify the operation of the Model 17C
and identify signs of malfunction, which was done on a daily basis. A checklist, provided by the
vendor representative and included as Appendix A, was completed by Battelle and USDA staff
when regular maintenance activities were performed. In the event of an instrument malfunction,
Battelle and/or USDA staff could contact the vendor representative and conduct minor trouble-
shooting procedures upon request as necessary, but were not expected to make any major
repairs. The vendor representative remained on-site until the installation was complete. All the
testing activities were conducted by Battelle and/or USDA staff. The vendor representative
returned to the test site after the completion of Phase I to install the Model 17C at the Phase II
test site.
3.2.1 Site Description—Phase I
Figure 3-1 shows a schematic diagram of the swine farm during Phase I of the verification test.
The AFO included ten animal barns arranged
in two parallel rows of five, with each barn
housing up to 2,000 swine. The urine and feces
from the swine exited the barns through metal
gratings in the floor and were deposited in two
nutrient lagoons located on the southern end of
the AFO. The perimeter of the AFO was lined
with trees, with agricultural fields surrounding
the AFO perimeter. A temperature-regulated
instrument trailer was placed on-site during the
test to house the monitoring equipment and to
provide a sheltered work space. The Model
17C was installed inside the instrument trailer,
and a Teflon inlet line was used to supply
outside air to the Model 17C. Ambient air
passed through a Teflon filter before entering
the Teflon inlet line. The inlet was mounted on
a tripod on the west side of the trailer at a
height of approximately 2 meters. The platform
shown in Figure 3-1 was installed to hold some
of the monitoring equipment.
Figure 3-1. Phase I Test Site
N
mmmm*
4
-------
3.2.2 Site Description—Phase II
Figure 3-2 shows a schematic diagram of the
cattle feedlot during Phase II of the
verification test. The instrument trailer used
in Phase I of this verification test was also
used in Phase II and was in a harvested corn
field surrounded on three sides by cow pens.
The farm was surrounded on all sides by
corn fields, most of which had been
harvested. Approximately 2,000 to 3,000
head of cattle were on the farm during the
verification test. The Model 17C was
installed in the instrument trailer as in Phase
I, with an inlet height of approximately
1.5 meters.
3.3 Test Procedures
Figure 3-2. Phase II Test Site
All tests utilized the continuous NH3 measurement data record stored by the Model 17C that
were downloaded from the instrument and transferred digitally to computer spreadsheets for
analysis.
3.3.1 Accuracy, Linearity, Precision, and Response Time
During the first week of each phase of testing, the Model 17C was independently supplied with
compressed NH3 gas standards to achieve NH3 concentrations over a range from 0 to 10,000 ppb
(Phase I) or 0 to 2,000 ppb (Phase II) to simulate the range expected in ambient air during each
phase. The gases delivered to the Model 17C were prepared by diluting higher-concentration
NH3 standard gases (i.e., 100 to 500 parts per million) in zero air using a calibrated dilution
system provided by the USD A.
The NH3 gas was supplied to the Model 17C for between 30 minutes and three hours at each
concentration level. Accuracy, linearity, and precision were established based on the continuous
digital data set recorded by the Model 17C during the periods when the NH3 gas was supplied.
Data were used for the calculations once the signal had stabilized at a constant concentration
(i.e., the signal did not appear to be increasing or decreasing with time). The time required to
reach 95% of the stable reading for each concentration was also recorded for the Model 17C.
These data were used to assess the response time of the Model 17C.
3.3.2 Calibration and Zero Drift
On Monday, Wednesday, and Friday of the first and last weeks of testing during each phase, the
Model 17C was supplied with an NH3 gas standard at nominally 1,000 ppb and zero air to check
the calibration and zero drift of the Model 17C, respectively. Zero air and the 1,000-ppb NH3
standard were each supplied to the Model 17C for approximately one hour, during which time
the measured concentrations were recorded by the Model 17C.
5
-------
3.3.3 Interference Effects
During the second phase of testing, the Model 17C was independently supplied with a series of
potential interference gases (hydrogen sulfide, nitrogen dioxide, 1,3-butadiene, and
diethylamine) to assess any impact the gases have on the Model 17C response. The interferent
gases were supplied from diffusion tubes (VICI Metronics, Poulsbo, Washington) at
concentrations of approximately 100 to 300 ppb in zero air and in a 500-ppb NH3 standard as
carrier gases.
The process for supplying the interferent gases was as follows: zero air was supplied to the
Model 17C until a stable reading was achieved. The interferent gas was then added to the zero
air flow and supplied to the Model 17C until a stable reading was observed (at least 2 minutes).
The Model 17C was then flushed for at least 2 minutes with zero air, and the next interferent gas
was delivered. This process was repeated for the four interferent gases. A 500-ppb NH3 standard
was then supplied to the Model 17C until a stable reading was achieved. The interferent gas was
then added to the NH3 standard for delivery to the Model 17C and the process outlined above
was repeated, delivering the 500-ppb NH3 standard for at least 2 minutes between each
interferent gas.
3.3.4 Comparability
The comparability of the Model 17C with a standard reference method was established by
comparing the average Model 17C readings with
time-integrated NH3 samples collected using citric-
acid-coated denuders. The reference samples were To Pump
collected based on procedures described in the EPA
Compendium Method 10-4.2, Determination of
Reactive Acidic and Basic Gases and Acidity of
Fine Particles (<2.5 fim)0>
For this test, NH3 samples were collected using a
ChemComb Model 3500 Speciation Sampling
Cartridge (Rupprecht & Patashnick Co., East
Greenbush, New York). Figure 3-3 shows a
schematic illustration of the ChemComb sampling
cartridge. Samples were collected by drawing
ambient air through an impactor at a nominal rate
of 10 Lpm to remove particulate matter with
aerodynamic diameters greater than 2.5 micro-
meters (|im). The air was passed through two or
more citric-acid-coated denuders to collect gaseous
NH3. A single Teflon filter was used to collect the
particulate matter that passed through the denuder.
For Phase I, air flow was controlled using
diaphragm pumps with needle valves. During
Phase n, automated Partisol Model 2300 speciation
samplers (Rupprecht & Patashnick Co., East
Greenbush, New York) were used. The Partisol
samplers were equipped with mass-flow controlled
Teflon filter
Denuder
Coating: 1% citric acid
Impactor
Inlet
Figure 3-3.
Cartridge
Reference Method Sampling
6
-------
sampling systems that were pressure- and temperature-corrected. This improved the accuracy of
the sampled air volume and also reduced the overall labor requirements. The samplers had not
been available during Phase I.
The procedures that were used for preparing and coating the denuders were based on the
procedures given in the ChemComb Operating Manual(4) and the test/QA plan.(2) The denuders
were coated in an NH3-free glove box at a USD A National Soil Tilth Laboratory facility in
Ames, Iowa, and stored in an NH3-free glove box until they were installed in the ChemComb
sampling cartridge and transported to the test site. Cartridges were assembled in the laboratory
and transported to the test site. All denuders were used within 72 hours of being coated and
within 24 hours of being transported to the field.
Reference samples were collected during the second and third weeks of testing during each
phase. To capture diurnal variations in NH3 concentrations, sampling was conducted on
approximately the following schedule: 8:00 a.m. to 12:00 p.m., 12:00 p.m. to 2:00 p.m.,
2:00 p.m. to 4:00 p.m., 4:00 p.m. to 8:00 p.m., and 8:00 p.m. to 8:00 a.m., so that five sets of
samples were collected in each 24-hour period. The short-term (2-hour and 4-hour) sampling
captured the midday peaks in NH3 concentrations, whereas the 12-hour sampling captured
overnight, generally low, concentrations. After sampling, the sampling media were retrieved and
transported to the USDA laboratory for extraction and analysis. During Phase I, sampling was
conducted at two locations: the instrument trailer near the Model 17C inlet and near the platform
shown in Figure 3-1. Duplicate samples were obtained at each location. Sampling was
conducted daily, Monday through Friday, during the two-week reference sampling period.
During Phase n, the reference sampling for single-point monitors was conducted at one location
near the monitor inlets at the instrument trailer. Duplicate samples were also obtained at this
site. The sampling schedule for Phase II deviated from the test/QA plan in that sampling was
conducted every other day, including weekends, during the two-week sampling period. The
schedule allowed sufficient time for sample transportation and processing between sampling
days.
Extraction and analysis of the denuders were performed as described in the test/QA plan,(2) with
one exception. The water volume used to extract the denuders was increased from 10 milliliters
(mL), as specified in the test/QA plan, to 20 mL. The volume was increased to accommodate the
sample volume requirements of the analysis method described below. A deviation was filed to
address this change, which does not impact the quality of the reference data. Samples were
extracted in an NH3-free glove box and stored in acid-washed scintillation vials to prevent
contamination. The samples were analyzed by USDA by flow injection analysis (FIA) using a
Lachat QuikChem Automated Flow Injection Ion Analyzer (Lachat Company, Loveland,
Colorado) according to QuikChem Method No. 10-107-06-2-A. This method involves heating
the NH3 sample with salicylate and hypochlorite in an alkaline phosphate buffer, which
produces an emerald green color proportional to the NH3 concentration. The color was
intensified by adding sodium nitroprusside and monitored photometrically.
When possible, samples were analyzed within 24 hours of extraction, as specified in the test/QA
plan. When analysis within 24 hours of extraction was not possible, the samples were stored
frozen until the analysis could be performed, in accordance with the test/QA plan.
7
-------
Chapter 4
Quality Assurance/Quality Control
QA/quality control (QC) procedures were performed in accordance with the quality management
plan (QMP) for the AMS Center(5) and the test/QA plan for this verification test.(2)
Six deviation reports were filed during this test and have been addressed in this report. In
summary, a change was made in the reference sampling schedule and equipment for Phase II
(Section 3.3.4), the denuder extraction volume was increased (Section 3.3.4), some percent
difference values measured for duplicate reference samples exceeded 10% (Section 4.2.3),
laboratory blank tolerances were redefined (Section 4.2.4), the order in which laboratory blanks
and calibration check standards were submitted for analysis was changed (Section 4.2.4 and
4.2.5), and not all of the test data were reviewed within two weeks of the end of the test phase
(Section 4.5). None of these deviations have impacted the quality of this verification test.
4.1 Equipment Calibrations
4.1.1 Reference Method Sampling Equipment
Reference method sampling was conducted based on the procedures described in the EPA
method(3) and the ChemComb operating manual.(4) A single-point calibration of the flow rate
through each of the sampling systems (i.e., pump, flow controller, filter pack, denuder,
impactor) was performed prior to starting each phase using a flow meter with a National
Institute of Standards and Technology (NIST)-traceable calibration. The flow rate of each
sampler was checked at the beginning and end of each sampling period using an in-line flow
meter. The flow rate was readjusted if the flow check was not within ± 5% of the nominal flow
rate of 10 Lpm (i.e., 9.5 Lpm to 10.5 Lpm). All calibration results were documented for
inclusion in the verification test data files. For Phase n, flows were controlled by the pressure-
and temperature-corrected mass flow controllers used in the USDA's Parti sol samplers. These
samplers shut off automatically if the flow deviated by ± 5% from the 10 Lpm setpoint for more
than 5 minutes, and the data were flagged. Actual sample volumes were recorded by the
samplers.
4.1.2 Analytical Equipment
The reference samples were analyzed in the USD A laboratory using FLA. A five-point
calibration was measured on the FLA for the reference sample analysis prior to each analytical
session by the USD A staff performing the analysis. The calibration was conducted according to
the manufacturer's recommendations and included concentrations of NH3 standard solutions
throughout the operating range of the FLA. The calibration was acceptable if the coefficient of
8
-------
determination (r2) of the calibration curve was greater than 0.99. The FIA detection limit (DL)
was 0.03 milligrams per liter (mg/L) and was determined as three times the standard deviation of
repeated measurements of a low-level NH3 standard. Any analytical results that fell below the
FIA DL were used without any further adjustment.
Calibration check standards were analyzed after every fifteenth sample in the batch. These
calibration checks were considered acceptable if the measured concentration agreed within 10%
of the standard solution concentration. If a calibration check failed to agree within 10% of the
standard concentration, the FIA was recalibrated, and all analyses since the last acceptable
calibration check were repeated. All calibration results were documented for inclusion in the
verification test data files.
4.1.3 Meteorological Equipment
The sensors used for meteorological monitoring had been calibrated by the manufacturer (Met
One Instruments, Inc., Grants Pass, Oregon) within one year of their use in this verification test.
The calibration results were included in the verification test data files.
4.1.4 Ammonia Dilution System
The USDANH3 dilution system (Environics, Tolland, Connecticut) employs three heated mass
flow controllers and valves dedicated for the dilution of compressed NH3 mixtures. The output
flow rates were verified using an independent, NIST-traceable flow meter and agreed to within
10%.
4.2 QC Samples
4.2.1 Field Blanks
At least 10%) of all reference samples collected were field blanks. The field blanks were collected
by installing the sampling media (i.e., denuder and filters) in the sampling train without drawing
any air through the train. The media were recovered and handled as normal samples. Field
blanks were collected at each of the sampling locations and during each of the sampling periods
(e.g., 8:00 a.m. to 12:00 p.m.). Field blank results were used to detect potential sample
contamination (as defined in the test/QA plan as field blank values greater than 5% of any
reference samples for that day) and also to determine the reference method DL.
The reference method DL was determined from the field blank results and reported in terms of
an NH3 mass corresponding to three times the standard deviation of the NH3 mass collected on
the field blanks. Reference method DLs were determined for each phase and were more than six
times higher than the equivalent FIA DL (0.6 microgram [|ig] NH3 per 20-mL sample).
The reference method DLs, reported as NH3 masses, were used to determine the minimum
detectable NH3 concentration for each phase. Since the mass of NH3 collected by the reference
method is a function of the sampling time, flow rate, and the ambient NH3 concentration, the
minimum time-integrated ambient NH3 concentration detectable by the reference method varies
depending on the sample period duration. (This assumes a constant flow rate.) For example, to
collect 100 |ig NH3, the time-integrated ambient NH3 concentration must be 20 ppb for a 12-
9
-------
hour sample and 120 ppb for a 2-hour sample. Accordingly, the minimum ambient NH3
concentrations that could be detected from the collection of 2-, 4-, and 12-hour samples at a
nominal flow rate of 10 Lpm were calculated from the reference method DL for each phase.
4.2.1.1 Phase I
During Phase I of testing, a total of 11 field blanks were collected (10% of reference samples).
The sample cartridges were exposed to ambient air (caps removed) for approximately the time it
would take to connect the cartridges to the pump tubing. The caps were then replaced and the
cartridges handled in the same way as regular reference samples. The average NH3 mass
collected on the field blanks was 5.3 |ig, with a range of 1.5 to 7.0 |ig. This range of collected
NH3 corresponded to 0.5% to 6.5% of the NH3 mass collected on any of the reference samples
on the corresponding days during which the field blanks were collected. Two of the Phase I field
blanks were above 5% of the minimum reference sample mass for that corresponding day. These
field blanks collected 5.6 |ig NH3, which was slightly above the average field blank NH3 mass
during Phase I; however, the field blanks were collected on days that exhibited lower ambient
NH3 levels, resulting in a relatively large percentage of the reference mass (6.5% and 5.9%).
These field blanks did not show unusually high levels of contamination, and it does not appear
that they had a significant impact on the Phase I reference method results. The standard
deviation of the NH3 collected on field blanks for Phase I was 1.6 |ig and the Phase I reference
method DL was 10.1 |ig NH3. The minimum detectable ambient NH3 concentrations are shown
in Table 4-1 for 2-, 4-, and 12-hour samples. During Phase I, all measured NH3 levels were
greater than these minimum NH3 concentrations, with a minimum measured value of 107 ppb
for a 2-hour sample.
Table 4-1. Minimum Detectable Ambient NH3 Concentrations During Phase I
2-Hour
Sample
4-Hour
Sample
12-Hour
Sample
Minimum detectable NH3
concentration
12.1 ppb
6.0 ppb
2.0 ppb
Number of reference samples collected
46
45
19
Number less than the minimum
detectable NH3 concentration
0
0
0
4.2.1.2 Phase II
During Phase II of testing, the reference sampling was conducted somewhat differently than in
Phase I, in that all the reference sampling cartridges and field blanks were installed in the
sampler prior to the first sampling period on a given day. The reference sample and field blank
cartridges were thus exposed to the ambient environment for a period of approximately
24 hours. Nonetheless, the average measured NH3 mass in the field blanks for Phase II was
somewhat lower than in Phase I. A total of 14 field blanks was collected in Phase n. The average
NH3 mass collected on these blanks was 2.5 |ig NH3, and the range was 0.5 to 4.6 |ig NH3. The
mass collected on the field blanks ranged from 1.2% to 55.0% of the smallest reference sample
mass collected on the same day, with an average of 19.2%. These percentages are not indicative
10
-------
of unusually high levels of contamination, but rather are a result of relatively low ambient NH3
levels at the AFO. The impact of these blank levels on the results of this verification test may be
manifested as a small positive bias of the reference method results relative to the readings of the
technologies being verified. This bias would be most pronounced on days with low ambient NH3
concentrations. The highest field blank percentages were measured on days when the integrated
ambient NH3 levels were as low as 6 ppb, which is approaching the 4.9-ppb minimum detectable
ambient NH3 concentration for a 2-hour sample. Assuming an ambient air sample volume of
1.2 cubic meters, the smallest volume collected during Phase n, the maximum field blank value
corresponds to an ambient concentration of 5.5 ppb. Thus, the sample handling may account for
up to 5.5 ppb of the measured values.
The standard deviation of the NH3 collected from field blanks for Phase II was 1.4 jug, which
resulted in a 6.6 |j,g NH3 Phase II reference method DL. The minimum detectable ambient NH3
concentrations for 2-, 4-, and 12-hour samples (at a nominal flow rate of 10 Lpm) are shown in
Table 4-2. During Phase n, one measured NH3 concentration in ambient air fell below the
minimum detectable NH3 concentration, as summarized in Table 4-2.
Table 4-2. Minimum Detectable Ambient NH3 Concentrations During Phase II
2-Hour
4-Hour
12-Hour
Sample
Sample
Sample
Minimum detectable NH3 concentration
7.9 ppb
4.0 ppb
1.3 ppb
Number of reference samples collected
56
56
29
Number less than minimum detectable NH3
2
0
0
concentration
4.2.2 Denuder Breakthrough Checks
4.2.2.1 Phase I
Use of backup denuders is called for in the test/QA plan during periods when breakthrough
greater than 10% of the front denuder is observed or expected. Owing to the high NH3 levels
observed during Phase I, all reference samples collected during Phase I included at least one
backup denuder, and most samples (>70%) included two backup denuders. These backup
denuders were used to check the degree of NH3 breakthrough. The breakthrough checks were
conducted at both of the sampling locations and included checks during each of the five
sampling periods (i.e., 8:00 p.m to 8:00 a.m., 8:00 a.m. to 12:00 p.m., etc.). Figure 4-1 shows
the percentage of NH3 collected on the backup denuders relative to the front denuder (i.e.,
breakthrough) as a function of the average NH3 concentration for each of the sampling period
lengths (combined data from both sampling locations). The solid symbols in this figure represent
the first backup denuder (identified as Denuder 2 in the legend), and the open symbols represent
the second backup denuder (identified as Denuder 3 in the legend). This figure illustrates that
the first backup denuder captured a significant fraction of NH3 relative to the front denuder
during many of the sampling periods (up to 200% of the front denuder). The second backup
denuder captured more than 10% of the NH3 on the front denuder in only three cases. It is
unlikely that NH3 was lost due to breakthrough of the second backup denuder for these or any of
11
-------
250%
200%
150%
m
£ 100%
m
50%
0%
~ 2 Hour Denuder 2
o 2 Hour Denuder 3
a 4 Hour Denuder 2
a 4 Hour Denuder 3
¦ 12 Hour Denuder 2
~ 12 Hour Denuder 3
9 ~
p s <> \
A
O 0-
200 400 600 800 1000 1200 1400
Integrated Ammonia Concentration (ppb)
1600 1800
Figure 4-1. Denuder Breakthrough During Phase I as a Function of Integrated
Ammonia Concentration
the reference samples. Therefore, these samples were not eliminated from the reference data. The
relatively high collection of NH3 on the first backup denuder may have been caused by
displacement by species with a higher affinity for the citric acid coating. Presumably these
species would remain on the front denuder, so it is unlikely that NH3 was lost as a result. Table
4-3 summarizes the results of the breakthrough checks for Phase I.
4.2.2.2 Phase II
The NH3 levels measured during Phase II were significantly lower than observed during Phase I.
Thus, the sampling approach was changed such that all samples still included one backup
denuder, but only 19% of the samples collected during Phase II included two backup denuders.
Figure 4-2 shows the percentage of NH3 collected on the backup denuders relative to the front
denuder as a function of the average NH3 concentration during the corresponding sampling
period, using the same symbols as in Figure 4-1. Data for all three Phase II sampling locations
are included here. As shown in the figure, any high breakthrough values observed on the second
backup denuder (Denuder 3 in the legend) occurred at very low NH3 concentrations where the
mass of NH3 collected was similar to that collected for field blanks. The high values do not
indicate that breakthrough occurred, but rather that the measurements were near the DL of the
overall reference method. In general, breakthrough onto the first backup denuder (Denuder 2 in
the figure legend) was low. With the exception of one sample that occurred at a low ambient
NH3 concentration, breakthrough observed on the second backup denuder was always less than
10% of the amount collected on the front denuder. Table 4-4 summarizes the results of the
breakthrough checks for Phase n.
12
-------
Table 4-3. Denuder Breakthrough Checks During Phase I
2-Hour Samples
4-Hour Samples
12-Hour Samples
1st Backup
Denuder
(%)
2nd Backup
Denuder
(%)
1st Backup
Denuder
(%)
2nd Backup
Denuder
(%)
1st Backup
Denuder
(%)
2nd Backup
Denuder
(%)
Percent of reference
samples with denuder
100
72
100
80
100
74
Average
concentration as % of
concentration on
front denuder
19.4
1.2
42.4
2.5
82.5
6.5
Maximum
concentration as % of
concentration on front
denuder
111.0
3.6
199.3
41.7
159.2
28.8
Percent of samples
with breakthrough
greater than 10% of
front denuder
57
0
82
3
100
14
240
220
^ 60
vO
0s
-C
O)
3
o
£ 40
03
CD
i_
CO
20
0
0 50 100 150 200 250 300 350
Integrated Ammonia Concentration (ppb)
Figure 4-2. Denuder Breakthrough During Phase II as a Function of Integrated
Ammonia Concentration
*
~ 2 HourDenuder2
O 2 HourDenuder3
A 4 HourDenuder2
A 4 HourDenuder3
~
¦
¦ 12 HourDenuder2
~ 12 HourDenuder3
\ 1
~ m ^
13
-------
Table 4-4. Denuder Breakthrough Checks During Phase II
2-Hour Samples
4-Hour Samples
12-Hour Samples
1st Backup
Denuder
(%)
2nd Backup
Denuder
(%)
1st Backup
Denuder
(%)
2nd Backup
Denuder
(%)
1st Backup
Denuder
(%)
2nd Backup
Denuder
(%)
Percent of reference
samples with denuder
100
18
100
18
100
24
Average concentration as
% of concentration on
front denuder
8.6
4.1
4.4
2.8
5.2
1.1
Maximum concentration
as % of concentration on
front denuder
[233.3](a)
53.8
11.3
17.2
7.5
45.9
2.5
Percent of samples with
breakthrough greater than
10% of front denuder
29
10
10.7
0
17.2
0
{!i> Suspect value rejected based on Q-test and not included in other calculations. This value corresponded to an NH3
concentration that was less than the minimum detectable NH3 concentration.
4.2.3 Duplicate Samples
For at least 10% of the reference samples, duplicates were collected using a collocated sampling
train (within 1 meter). These duplicate samples were collected at both of the sampling locations
during Phase I, and only at the trailer location during Phase n, and were collected during each of
the sampling periods. The relative percent difference (RPD) between the duplicate samples was
calculated by dividing the absolute difference of the sample concentrations by the average of the
sample concentrations.
Table 4-5 summarizes the results of the duplicate sampling for both Phases I and n. During
Phase I, a total of 18 sets of duplicate samples were collected. Eight of the duplicate samples
were collected at the sampling location next to the trailer, and the other 11 duplicate samples
were collected at the sampling location next to the platform. For Phase I, the duplicate samples
showed absolute RPD values between 0.6% and 22%, and the average RPD was 9%. During
Phase n, duplicate samples were collected during every sampling period at the sampling location
next to the trailer, resulting in a total of 35 duplicate measurements. The absolute RPD varied
between 0.7% and 32%, with an average of 7%. Although the average RPD values are
comparable in Phases I and n, the absolute differences were significantly smaller during
Phase n. For both phases combined, the absolute RPD for 13 of the duplicate samples exceeded
the QA limit of 10% specified in the test/QA plan. To verify the quality of the reference method,
NH3 gas standards were delivered to the reference method. Repeated delivery of the same
concentration standard gave an average RPD of 1.3%. Thus, it is probable that the exceedences
were caused by non-uniformity in the air sampled and did not impact the quality of the reference
method itself. However, some contributions may result from small variations in sampling flow
rates and analytical uncertainties.
14
-------
Table 4-5. Duplicate Reference Method Samples
Phase I
Phase II
Absolute
Absolute
RPD Difference
RPD Difference
(%) (ppb)
(%) (ppb)
Average 9 28
Maximum 22 109
Minimum 0.6 1
Number of duplicate samples 18
Number with RPD > 10% 6
7 5
32 18
0.7 0.6
35
7
4.2.4 Laboratory Blanks
Laboratory blank solutions were prepared for the FIA using distilled, deionized water. In each
analytical batch, at least 10% of the number of reference samples analyzed were laboratory
blanks, and were submitted to the laboratory as blind samples. The analysis of the laboratory
blanks deviated from the test/QA plan in that, rather than submitting the blanks routinely (e.g.,
every tenth sample), the blanks were interspersed among the other samples and submitted as
blind samples.
During Phase I, a total of 31 laboratory blank samples were analyzed. The analytical results from
the laboratory blanks indicated no apparent drift in the calibration of the FIA, and none of the
blank values were greater than 5% of the lowest measured reference sample on that day. (Note:
The test/QA plan indicates that laboratory blanks should not exceed 5% of any concentration
measured on that day. As written, this threshold includes field blanks and backup denuder
samples. A deviation report has been filed to change this threshold so that it applies only to
composite reference samples and does not include samples that would be expected to have low
concentrations, such as field blanks.) During Phase n, a total of 27 laboratory blank samples
were analyzed. Similarly, the analytical results from the laboratory blanks indicated no apparent
drift in the baseline of the FIA, and none of the blank values was greater than 5% of the lowest
measured reference sample on that day.
4.2.5 Calibration Checks
In addition to analyzing every 15th calibration check samples, as described in Section 4.1.2, at
least 10% of the samples were submitted to the laboratory as blind calibration check samples.
These blind calibration check samples were prepared by diluting NIST-traceable NH4+ standard
stock solution.
During Phase I, 38 NH4+ blind calibration check samples were prepared from 15 different
standard solutions, ranging in concentration from 0.4 to 8 mg/L NH3. Measured concentrations
for 10 of these calibration check samples differed from the delivered standard concentration by
more than 10%, and the full set of measured values was on average 1.9% lower than the
delivered concentration. It should be noted that the calibration check samples were prepared
from NH4+ standards that were diluted from a 1,000-mg/L stock solution and that errors may
15
-------
have occurred during the dilution process. For example, nine of the 10 calibration check samples
that failed were prepared from four different standard solutions. Of these four standard solutions,
a total of 10 samples were submitted to the laboratory for analysis, and 9 of the samples fell
outside the 10% acceptance criterion. Of the 28 additional samples submitted to the laboratory
from the 11 other prepared standard solutions, only one fell outside the 10% acceptance
criterion, and the concentration of that standard solution was near the quantitation limit of the
FIA. As such, it is likely that the preparation of the standard solutions contributed to the failure
of the calibration check samples, rather than the calibration of the FIA.
During Phase n, 24 calibration check samples were prepared from four different standard
solutions. Measured concentrations for six of these calibration check samples differed from the
delivered standard concentration by more than 10%, and the full set of measured values was on
average 4.4% lower than the delivered concentration. Of the six calibration check samples that
failed, five were prepared from two of the four standard solutions. It is possible that the failures
may be attributable to inadvertent dilution or degradation of the standard solutions used, since
these standards were prepared prior to submission of the first samples and failed consistently
only near the end of the analysis period. The sixth calibration check sample that failed may be
associated with a transcription error in the submission log.
4.2.6 Gas Standard Dilution Checks
At each of the nominal NH3 levels to be used for the accuracy and linearity checks, at least one
sample of the dilution of the NH3 gas standard was collected using the reference method. These
samples were analyzed as regular samples and used to check the accuracy of the dilution system.
Figure 4-3 shows the measured NH3 captured by the sampling cartridges versus the NH3
delivered during the dilution checks.
Standard Ammonia Concentration (ppm)
Figure 4-3. Analysis of Diluted Ammonia Standards Using the Denuder
Reference Method
16
-------
A dilution check was conducted before Week 2 of Phase I. However, the sampling line was
thought to have not been flushed with the diluted NH3 sample prior to collecting the check
samples, and the measured concentrations did not agree within 10% of the expected concentra-
tion. Consequently, the dilution check was repeated prior to Phase n, and the results are shown in
Figure 4-3. The average RA of the measured concentrations was 4% and indicates that the NH3
gas standards as delivered by the dilution system were accurate with respect to the reference
method.
4.3 Audits
4.3.1 Performance Evaluation Audit
A performance evaluation audit was conducted to assess the quality of the measurements made in
this verification test. This audit addressed only those measurements that factor into the data used
for verification, i.e., the sample flow rate and the analytical laboratory measurements. This audit
was performed once during the verification test by analyzing a standard or comparing a reading to
a reference that was independent of standards used during the testing.
The flow rates of the reference method sampling assemblies were audited once during each phase
of testing using a flow meter independent of the meter used to calibrate the flow rate. During
Phase I, agreement between the audit flow rate and the nominal flow rate indicated a bias in the
calibrated flow rates. The flow rates were recalibrated. The bias was later attributed to a faulty
audit flow meter, and the original flow calibrations were verified against a second audit flow
meter.
The performance of the FIA was audited by analyzing an NH4+ standard independent of those
used for the calibration, but were the same as those used for the calibration checks described in
Section 4.2.5. These samples were provided as blind audit samples, and the operator of the FIA
was not aware of the concentrations of the samples. In several cases, agreement between the
measured concentration and the standard concentration was not within ±10% (ranged from -43%
to 64%>). The cause of the discrepancy was investigated but could not be identified. It is possible
that some of the discrepancy is attributable to uncertainties associated with dilution of the stock
1,000 mg/L NH4+ standard solution. Multiple solutions were prepared, and only some of those
solutions showed discrepancies with the analytical results. The relative agreement between the
reference samples collected during the gas standard dilution check (performed between Phases I
and II) and their expected values provide additional verification of the accuracy of the FIA.
4.3.2 Technical Systems Audit
Battelle's ETV Quality Manager performed a technical systems audit (TSA) of the performance of
this verification test during each phase of the test. The purpose of this TSA was to ensure that the
verification test was being performed in accordance with the test/QA plan(2) and that all QA/QC
procedures were implemented. As part of the audit, Battelle's ETV Quality Manager reviewed the
reference sampling and analysis methods used, compared actual test procedures to those specified
in the test/QA plan, and reviewed data acquisition and handling procedures. Observations and
findings from this audit were documented and submitted to the Battelle Verification Test
Coordinator for response. The records concerning the TSA are permanently stored with the
Battelle Quality Manager.
17
-------
4.3.3 Audit of Data Quality
At least 10% of the data acquired during the verification test was audited. Battelle's Quality
Manager traced the data from the initial acquisition, through reduction and statistical analysis, to
final reporting, to ensure the integrity of the reported results. All calculations performed on the
data undergoing the audit were checked during the technical review process.
4.4 QA/QC Reporting
Each audit was documented in accordance with Sections 3.3.4 and 3.3.5 of the QMP for the ETV
AMS Center.(5) Once the audit report was prepared, the Battelle Verification Test Coordinator
ensured that a response was provided for each adverse finding or potential problem and imple-
mented any necessary follow-up corrective action. The Battelle Quality Manager ensured that
follow-up corrective action was taken. The results of the TSA were sent to the EPA.
4.5 Data Review
Records generated in the verification test were reviewed before these records were used to
calculate, evaluate, or report verification results. Table 4-6 summarizes the types of data recorded.
The review was performed by a technical staff member involved in the verification test, but not
the staff member who originally generated the record. The person performing the review added
his/her initials and the date to a hard copy of the record being reviewed. In some cases, entries in
the laboratory record books or on field data sheets were not reviewed within two weeks after
completion of each phase. A deviation report was filed to address this.
Table 4-6. Data Recording Process
Responsible
Data to be Recorded
Party
Where Recorded
How Often Recorded
Disposition of Data(a)
Dates, times of test events
USDA/
Laboratory record
Start/end of test, and
Used to organize/check
(site activities, etc.)
Battelle staff
books/field data sheet.
at each test activity.
test results; manually
incorporated in data
spreadsheets as
necessary.
Reference method
USDA/
Laboratory record
At least at start/end of
Used to organize/check
sampling data
Battelle staff
books, chain-of-custody
reference sample, and
test results; manually
forms, or file data sheets
at each change of a
incorporated in data
as appropriate.
test parameter.
spreadsheets as
necessary.
Meteorological conditions
Battelle
Meteorological station
data logger.
Continuously.
Used to assess
meteorological
conditions during
testing as necessary.
Ammonia analyzer
Vendor or
Data acquisition system
Continuously at
Electronically
readings
designee
(data logger, personal
specified acquisition
transferred to
computer, laptop, etc.).
rate throughout
analyzer operation.
spreadsheets.
Reference sample analysis
USDA/
Laboratory record
Throughout sample
Transferred to
and results
Battelle staff
books, data sheets, or
data acquisition system,
as appropriate.
handling and analysis
process.
spreadsheets.
w All activities subsequent to data recording were carried out by Battelle.
18
-------
Chapter 5
Statistical Methods and Reported Parameters
The statistical methods presented in this chapter were used to verify the performance parameters
listed in Section 3.1.
5.1 Relative Accuracy
The percent difference (%D) of the average Model 17C response to each NH3 gas standard was
calculated according to Equation 1
x- x„
%D = -x 100 C1)
Xn
where x is the average Model 17C response to an NH3 gas standard of nominal concentration xn.
For each phase of testing, the RA with respect to all of the gas standards (n) delivered to the
Model 17C was calculated using Equation 2:
Average RA = — (ll'/oOil) x 100 (2)
5.2 Linearity
Linearity was assessed by a linear regression analysis using the compressed gas standard
concentrations as the independent variable and results from the Model 17C as the dependent
variable. Linearity was expressed in terms of slope, intercept, and r2 and was calculated
independently for each phase of the verification test. The 95% confidence interval (CI) for the
slope and intercept was also calculated.
5.3 Precision
Precision was calculated in terms of the percent relative standard deviation (RSD) of Model 17C
measurements of several NH3 gas standards. The mean and standard deviations of those readings
were calculated. The RSD was then determined as:
sn
RSD = 4^x100 (3)
x v '
19
-------
where SD is the standard deviation of the Model 17C readings and x is the mean of the Model
17C readings. Precision was calculated independently for each phase of testing.
5.4 Response Time
Response time was assessed in terms of both the rise and fall times of the Model 17C when
sampling NH3 gas standards or zero air. Rise time (i.e., 0% to 95% response time for the change
in NH3 concentration) was determined from the Model 17C response to a rapid increase in the
delivered NH3 concentration. Once a stable response was achieved with the gas standard, the fall
time (i.e., the 100% to 5% response time) was determined in a similar way, switching from the
NH3 standard back to zero air or a lower concentration NH3 gas standard. Rise and fall times were
determined for the Model 17C during each phase of testing. Response times are reported in terms
of seconds (s). It should be noted that response times include the time associated with
equilibration of NH3 on the tubing and inlet surfaces during delivery of the gas standards.
5.5 Calibration and Zero Drift
Calibration and zero drift were reported in terms of the mean, RSD, and range (maximum and
minimum) of the readings obtained from the Model 17C in the repeated sampling of the same
NH3 standard gas and of zero air. For zero drift, the SD is reported instead of the RSD since
dividing the SD by a value approximately equal to zero is not meaningful. The calibration and
zero drift were calculated independently during each phase of testing so that up to six NH3
standard and zero readings (Monday, Wednesday, and Friday for two weeks) were used for this
calculation in each phase. The results of these checks indicate the day-to-day variation in zero and
standard readings.
5.6 Interference Effects
The extent of interference was calculated in terms of the ratio of the response of the Model 17C to
the interfering species, relative to the actual concentration of the interfering species. For example,
if 100 ppb of an interfering species resulted in a 1-ppb increase in the NH3 reading of the Model
17C, the interference effect was reported as 1% (i.e., 1 ppb/100 ppb). The interference effect was
reported separately for each interferent, both in the absence and in the presence of NH3.
5.7 Comparability
Comparability between the Model 17C results and the reference method results with respect to
ambient air was assessed by linear regression using the reference method NH3 concentrations as
the independent variable and results from the Model 17C as the dependent variable. Compara-
bility was expressed in terms of slope, intercept, and r2 and was calculated independently for each
phase of the verification test. If the measured concentration of NH3 did not vary by at least a
factor of five during each phase of testing, then comparability for that phase was calculated using
Equation 1 and reported as a percent difference rather than in terms of the linear regression
results.
20
-------
Chapter 6
Test Results
The results of the verification test of the Model 17C are presented in this section. The Model 17C
stored NH3 measurement data as both 1- and 5-minute averages. When available, 1-minute
averaged data were used for the relative accuracy, linearity, precision, calibration/zero drift,
response time, and interference tests. The comparability tests utilized 5-minute averaged data.
Meteorological conditions collected using the meteorological monitoring station during Phase I
are shown in Figure 6-1. The ambient data set collected by the Model 17C is shown in the bottom
panel, along with the wind direction, wind speed, and ambient temperature data. The shaded
regions indicate the ammonia reference method sampling periods. The average ambient NH3
concentration measured by the Model 17C was 515 ppb, with a range of 20 ppb to 3,564 ppb. The
3000-
-Q
Q.
Q.
¦% 2000
E 1000
Reference
Measurement
Period
Reference
Measurement
Period
T
A
T
9/11/2003 9/16/2003 9/21/2003 9/26/2003 10/1/2003
Date
Figure 6-1. Phase I Meteorological Conditions and Model 17C Ambient NH3
Measurements
21
-------
large gaps in the Model 17C ambient NH3 data set are discussed in Section 6.8, although some
gaps appear because data points were not plotted for periods when Battelle or USDA staff were
performing testing activities on the Model 17C. The meteorological conditions, which were
recorded as 1-hour averages, varied widely over the duration of Phase I. The average ambient
temperature during Phase I of the test was 14°C, with a range of -4 to 29°C. The average relative
humidity was 66%. Winds were predominantly from the southeast and northwest, with wind
speeds up to 17 miles per hour (6 miles per hour average). When winds were observed from the
southeast, the monitors were exposed to emissions from the nutrient lagoons, whereas the
monitors sampled barn emissions during periods of northerly winds.
The ambient temperatures during Phase n, presented in Figure 6-2 along with other meteoro-
logical conditions, were significantly cooler than during Phase I, with an average ambient
temperature of 4.5°C (range -10 to 29°C) and an average relative humidity of 75%. Winds were
predominantly from the northwest and quite variable in speed, averaging 7 miles per hour
(30 miles per hour maximum). Figure 6-2 shows the Phase II wind direction, wind speed, and
ambient temperature data and the ambient NH3 data set collected by the Model 17C (bottom
panel). The shaded region shows the period during which NH3 reference measurements were
conducted. The Model 17C NH3 measurements ranged from 5 ppb to 667 ppb during Phase II and
averaged 114 ppb.
C
o
t3
e s
Q o)
300-
200-
100-
-------
6.1 Relative Accuracy
During the first week of each phase of the verification test, the Model 17C was supplied with
compressed NH3 gas standards at several concentrations. The NH3 gas standards were diluted in
zero air and delivered to the inlet of the Model 17C at a flow rate of 5 Lpm.
Figures 6-3 and 6-4 present the NH3 concentrations recorded by the Model 17C during the RA
checks, along with the nominal NH3 concentration levels supplied to the Model 17C for Phase I
and Phase n, respectively. The averages of the measurements at each nominal NH3 concentration
are presented in Table 6-1 along with the calculated %D, the number of data points, and the
average RA for each phase.
As shown in Table 6-1, during Phase I, the %Ds of the Model 17C ranged from -10.2% to 2.8%
over the seven concentration levels measured, and the average RA over all the measurements was
3.7%. During Phase n, the %D of the Model 17C ranged from -11.5% to -8.9%, and the average
RA was 10.5%. It should be noted that, although the Model 17C was calibrated prior to Phase I,
it was not recalibrated prior to Phase II after being transferred to the cattle feedlot.
12000
i Gas Standard Concentration
10,000 ppb
10000 -
17C-NH3 Continuous Measurements
g- 8000 -
o
6000 -
5000 ppb
0
o
o
O
£ 4000
Time
Figure 6-3. Phase I Accuracy Results for the Model 17C
23
-------
Time
Figure 6-4. Phase II Accuracy Results for the Model 17C
Table 6-1. Relative Accuracy Results
Phase I
Phase II
NH3 Gas
Average
Average
Standard
Measured
Number
Measured
Number
Concentration
Concentration
of Data
Concentration
of Data
(PPb)
(PPb)
Points
%D
(PPb)
Points
%D
0
5.7
52
NA
3
6
NA
300
269
51
-10.2
273
4
-8.9
600
576
46
-4.1
535
11
-10.9
1,000
984
123
-1.6
886
10
-11.4
1,500
1,527
31
1.8
1,328
11
-11.5
2,000
2,051
12
2.6
1,802
9
-9.9
5,000
5,140
5
2.8
(a)
(a)
(a)
10,000
10,270
8
2.7
(a)
(a)
(a)
Average RA
3.7%
10.5%
(a) The concentration levels and sequence of NH3 concentrations supplied to the Model 17C were changed for the RA
checks conducted during each phase. Consequently, not all concentration levels were measured during both RA
checks.
NA = not applicable.
24
-------
6.2 Linearity
Figures 6-5 and 6-6 show the results of the linearity check for Phase I and Phase n, respectively.
During Phase I, a linear regression of the Model 17C response versus the gas standard concentra-
tion, over the range from 0 to 10,000 ppb, showed a slope of 1.03 (± 0.01), an intercept of -24
(± 23) ppb, and a coefficient of determination (r2) of 1.000, where the numbers in parentheses
represent the 95% CI. During Phase n, the Model 17C showed a linear response, over the range
from 0 to 2,000 ppb, with a slope of 0.90 (± 0.02), an intercept of -0.6 ppb (± 20.3), and an r2 of
1.000.
12,000
10,000 -
_Q
Q_
Q_
C
o
c
CD
O
c
o
O
"a
CD
=3
(/)
03
CD
,000 -
6,000 -
4,000 -
2,000 -
+ Average Response
1:1 Line
0 2,000 4,000 6,000 8,000 10,000 12,000
Gas Standard Concentration (ppb)
Figure 6-5. Results of Linearity Check of the Model 17C
During Phase I
25
-------
2000
_Q
Q.
Q.
^ 1500
o
H—I
2
H—<
c
CD
O
o 1000
o
C/)
03
| 500
0
Figure 6-6. Results of Linearity Check of the Model 17C
During Phase II
6.3 Precision
Table 6-2 presents the calculated precision of the Model 17C measured during the accuracy and
linearity checks. During Phase I, the precision of the Model 17C readings varied from 0.2% to
0.5% RSD, with an average precision of 0.3%. During Phase n, the precision of the Model 17C
readings ranged from 0.2% to 0.6% RSD at the five concentration levels measured in the
accuracy/linearity checks, also with an average of 0.3%.
6.4 Response Time
Response time was determined during each phase from the amount of time required for the Model
17C to reach 95% of the change in the stable concentrations during the accuracy/linearity checks
calculated from the change in NH3 concentration. Table 6-3 presents a summary of the response
time determinations for the Model 17C. Rise times ranged from 180 to 4,560 seconds, with fall
times between 120 and 180 seconds during Phase I. The 300-ppb, 600-ppb, and 1,000-ppb
nominal NH3 standards were each delivered for 3 hours, during which time the signal rose slowly
at a rate of approximately 7 to 28 ppb per hour. Thus, the "stable reading" for each standard
began approximately 2 hours after the start of the standard delivery. This had a significant
influence on the rise time calculations. During Phase n, rise times ranged from 600 to 900
seconds (calculated from 5-minute-averaged data), and the fall time from 1,000 ppb to zero was
1,200 seconds. The steady rise discussed above was not apparent during the Phase II checks.
4 Average Response
1:1 Line
y = 0.90x - 0.60
r2 = 1.000
500 1000 1500 2000
Gas Standard Concentration (ppb)
26
-------
Table 6-2. Calculated Precision of the Model 17C
Phase I
Phase II
NH3 Gas Standard
Average Measured
Average
Measured
Concentration
Concentration
RSD
Concentration
RSD
(PPb)
(PPb)
(%)
(PPb)
(%)
300
269
0.5
273
0.6
600
576
0.2
535
0.4
1,000
984
0.3
886
0.2
1,500
1,527
0.2
1,328
0.3
2,000
2,051
0.5
1,802
0.4
5,000
5,141
0.2
(a)
(a)
10,000
10,269
0.2
(a)
(a)
Average RSD
0.3
0.3
(^ The concentration levels and sequence of NH3 concentrations supplied to the Model 17C were changed for the RA
checks conducted during each phase; not all concentration changes were measured during both RA checks.
Table 6-3. Response Time Determinations
Phase I(a)
Phase II(b)
Rise Time
Fall Time
Rise Time
Fall Time
Change
(seconds)
(seconds)
(seconds)
(seconds)
0 - 300 ppb
4,560
—
900
—
300 - 600 ppb
2,520
—
600
—
600 - 1,000 ppb
420
—
600
—
1,000 - 1,500 ppb
(c)
—
600
—
1,500 - 2,000 ppb
(c)
—
600
—
1,000- 10,000 ppb
180
—
(<0
—
10,000- 5,000 ppb
—
180
—
(c)
5,000 - 2,000 ppb
—
120
—
(c)
2,000 - 1,500 ppb
—
120
—
(c)
1,500 - 0 ppb
—
180
—
(c)
1,000 - 0 ppb
—
(c)
—
1,200
(^ Only 1 -minute averaged data were available for this check.
(b:i Only 5-minute averaged data were available for this check.
(c:i The concentration levels and sequence of NH3 concentrations supplied to the Model 17C were changed for the RA
checks conducted during each phase; not all concentration levels were measured during both RA checks.
27
-------
However, the temporal resolution of the response time calculation was limited since the data used
for the calculation were 5-minute average values. It should be noted that the minimum response
time that can be calculated from 5-minute average values is 10 minutes. The measured response
times include the time associated with the equilibration of NH3 on the tubing and inlet surfaces
during delivery of the gas standard.
6.5 Calibration and Zero Drift
The calibration/drift checks were conducted by supplying a 1,000-ppb NH3 gas standard and zero
air to the Model 17C for a period of one hour each on Monday, Wednesday, and Friday during the
first and last week of each phase. The values reported in Tables 6-4 and 6-5 are based on the
average readings during the calibration and zero checks when the readings of the Model 17C had
stabilized. The results from the Phase I calibration and zero drift checks are presented in
Table 6-4. Unfortunately, some data from the last week of Phase I were not recovered from the
Model 17C and were not available for interpretation. No drift was observed in response to
1,000 ppb NH3 or zero air during Phase n. The results of the Phase II calibration and drift checks
are summarized in Table 6-5. No clear trend was observed for either the calibration or zero drift;
however, the response to 1,000 ppb NH3 increased by 34 ppb and 17 ppb during Weeks 1 and 4,
respectively.
Table 6-4. Calibration and Zero Checks During Phase I
Zero Check (ppb)
Calibration Check(a) (ppb)
Max- Min- Number
Check Mean SD(b) imum imum of Data
Number (ppb) (ppb) (ppb) (ppb) Points
Max- Min- Number
Mean RSD imum imum of Data
(ppb) (%) (ppb) (ppb) Points
Week 1 (C)
Monday
wW,eek! 8 0.5 9 8 11
Wednesday
™ef 1 6 0.5 6 5 47
rnday
(c)
(c)
1,015 0.6 1,024 1,008 24
Week4 4 o.4 4 3 13
Monday
w^eek^ 11 03 12 11 12
Wednesday
Week 4 $
Friday
3 554d)
0.2 3,567 3,547 21
l,016(e)
1,016 0.4 1,023 1,010 20
(f)
(a) 1,000-ppb NH3 nominal concentration.
(b) SD reported for zero drift check since the RSD is not meaningful for near-zero values.
(c) Gas delivery too short to reach stable value.
(d) 3,500-ppb NH3 nominal concentration.
(e) Equivalent response to a 1,000-ppb NH3 nominal standard.
(f) Data for this check were lost due to download failure.
28
-------
Table 6-5. Calibration and Zero Checks During Phase II
Zero Check (ppb)
Calibration Check(a) (ppb)
Max-
Min-
Number
Max-
Min-
Number
Check
Mean
SD(b)
imum
imum
of Data
Mean
RSD
imum
imum
of Data
Number
(ppb)
(ppb)
(ppb)
(ppb)
Points
(ppb)
(%)
(ppb)
(ppb)
Points
Week 1
Monday
4
0
4
4
22
830
1.4
848
817
16
Week 1
(<0
(<0
Wednesday
Week 1
Friday
3
0.8
4
2
7
864
0.1
866
863
5
Week 4
Monday
7
2.2
12
5
24
859
0.3
862
855
8
Week 4
Wednesday
3
0.8
5
4
5
886
0.2
888
883
10
Week 4
Friday
(d)
876
0.2
880
872
35
(a) 1,000-ppb NH3 nominal concentration.
(b) SD reported for zero drift check since the RSD is not meaningful for near-zero value.
(c) Data for this check were lost due to download failure.
(d) Zero air was not supplied for a sufficient amount of time to reach a stable reading.
6.6 Interference Effects
The effect of potential interferent gases on the response of the Model 17C was assessed by
supplying the Model 17C with a series of four gases (hydrogen sulfide, nitrogen dioxide, 1,3-
butadiene, diethylamine) in zero air and in a 500-ppb NH3 standard. The response of the Model
17C during the introduction of these gases is summarized in Table 6-6. The interference gas
concentrations carry an uncertainty of approximately ± 15%.
Table 6-6. Interference Effect Evaluation
Gas
Interferent Gas
Concentration
(PPb)
Interference Effect (%)
Zero-Air Matrix
500-ppb NH3 Matrix
Hydrogen sulfide
285
0.5(a)
-0.2(a)
Nitrogen dioxide
95
-2.6
-5.9
1,3-Butadiene
95
0(a)
3 2(a)
Diethylamine
96
51.8
50.6
{!i> Signal not significantly different from baseline without interferent gas.
29
-------
The response of the Model 17C to hydrogen sulfide and 1,3-butadiene was negligible. The
Model 17C showed small negative response to N02 in both zero air and a 500-ppb NH3 standard.
Diethylamine showed an interference effect of 51.8% in zero air and 50.6% in a 500-ppb NH3
standard.
6.7 Comparability
Figures 6-7 and 6-8 show the NH3 concentrations measured using the reference method, along
with the corresponding average readings of the Model 17C for the reference sampling periods,
during Phase I and Phase n, respectively. In general, the Model 17C appears to track changes in
NH3 concentrations measured with the reference method. These data also are presented in
Figures 6-9 and 6-10 as scatter plots to illustrate the correlation between the reference and Model
17C data.
A linear regression of the Model 17C responses during the reference sampling periods versus the
NH3 determined from the reference method was calculated for each phase. For Phase I, the linear
regression results showed a slope of 1.20 (± 0.05), an intercept of 16 ppb (± 29), and an r2 value
of 0.984, where the numbers in parentheses represent the 95% CI. For Phase n, the linear
regression results showed a slope of 0.86 (± 0.03), an intercept of -0.5 ppb (± 3.8), and an r2 value
of 0.990.
6.8 Ease of Use
The Model 17C was installed at the Phase I testing locations by two vendor representatives and at
the Phase II testing site by one vendor representative. The Model 17C could be installed and
operated by a user with minimal experience using instructions in the Model 17C manual. The
Phase I installation involved an on-site calibration of the Model 17C with both NO and NH3
standards, which were provided by the USDA and were independent of any gas standards used in
the verification test. The installation took approximately one day, including time for the Model
17C to warm up before the calibration was performed. The vendor representative trained Battelle
and USDA staff to perform two regular maintenance activities, which were conducted during both
Phase I and Phase n. The Teflon filter at the end of the ambient inlet was changed once per week,
and the desiccant was changed as needed, approximately every two weeks. Otherwise, no
maintenance was performed on the Model 17C. A checklist was prepared by Battelle staff from
information provided by the vendor representatives to establish whether the Model 17C was in
proper working order. The checklist, shown in Appendix A, was completed when regular
maintenance activities were performed, and the status of the instrument was checked daily by
verifying that no alarms were showing on the Model 17C display.
The Model 17C stores a finite amount of data before they are overwritten. For the 5-minute data
set, approximately 10 days of data were stored. Less than 24 hours of 1-minute averages were
stored. As a result of these limits, the data were downloaded at least every 10 days and following
testing activities whenever possible. Data were downloaded via a serial port connection between
the Model 17C and a laptop computer provided by USDA with the software provided by the
vendor. On several occasions, the data download was incomplete or failed due to software
30
-------
§: 1400
Date
Figure 6-7. Comparison of Ambient Reference Measurements with
Averages from the Model 17C During Phase I
CO
CO
CO
CO
CO
CO
CO
CO
CO
o
o
o
o
o
o
o
o
o
CD
00
o5
o
T—
T—
CNi
CO
CNj
CNj
CNj
CNj
CO
CO
o
o
o
O
O
O
o
o
o
T—
T—
T—
T—
T—
T—
T—
T—
Date
o o o o
LO CD
Figure 6-8. Comparison of Ambient Reference Measurements with
Averages from the Model 17C During Phase II
31
-------
Reference Ammonia Concentration (ppb)
Figure 6-9. Scatter Plot of Model 17C Results versus
Ambient Reference Measurements During Phase I
Reference Ammonia Concentration (ppb)
Figure 6-10. Scatter Plot of Model 17C Results versus
Ambient Reference Measurements During Phase II
32
-------
failure. This problem is recognized by Thermo Electron and is attributed to an incompatibity
between the Model 17C and computers with clock speeds greater than 266 MHz. Representatives
of Thermo Electron have indicated that an effort is underway to resolve this issue. During both
phases of testing combined, 42 download attempts were documented. Of these attempts, 19 were
successful, 17 were incomplete, and 6 failed; as a result, some data were not recovered. The time
spent attempting data downloads was not documented, but is estimated to be approximately three
hours for each phase. A summary of these and other activities involving the Model 17C during
Phase I and Phase II are presented in Tables 6-7 and 6-8, respectively.
Table 6-7. Activities Performed During Phase I
Time
Service
Offline (a)
Down Time (b)
Time (c)
Date
(minutes)
(minutes)
(minutes)
Activity
9/08/03
30
Supplied zero air and NH3 standard®
9/10/03
1,595
Supplied zero air and NH3 standard®
9/11/03
1,040
Supplied zero air and NH3 standards®
9/12/03
120
Alarm: Cooler temperature high, caused by
increased temperature of instrument trailer;
reduction in trailer temperature eliminated error;
no associated data loss
9/12/03
5
Changed Teflon inlet filter
9/13/03
5
Changed desiccant
9/17/03
120
Power loss: Instrument recovered without user
intervention
9/25/03
5
Changed desiccant
9/26/03
5
Changed Teflon inlet filter
9/29/03
990
Supplied zero air and NH3 standards; room air
sampled for 12 hours overnight®
10/01/03
960
Supplied zero air and NH3 standard®
10/03/03
120
Supplied zero air and NH3 standard®
10/03/03
(12,300)(e)
Data download failure. 205 hours of data lost
Totals
4,735
240
(12,300)(e)
20
99% (66%)l-e-1 data completeness® and
20-minute service time.
(a) Time Offline = time that the Model 17C was taken offline for zero or standard gas measurements. The period over which time
offline was evaluated began at 8:00 a.m. on 9/8/03 and ended at the conclusion of testing at 5:00 p.m. on 10/3/03. The amount
of time was rounded to the nearest 5 minutes.
(b) Down Time = time that the Model 17C was not operating or was operating but not reporting reliable measurements. The period
over which down time was evaluated began at 8:00 a.m. on 9/8/03 and ended at the conclusion of testing at 5:00 p.m. on
10/3/03. The amount of time was rounded to the nearest 5 minutes. Down time that did not result in loss of data is not included
in the availability determination.
(c) Service Time = time spent conducting routine operation and maintenance activities and troubleshooting problems. The period
over which service time was evaluated began at 8:00 a.m. on 9/8/03 and ended at the conclusion of testing at 5:00 p.m. on
10/3/03. The amount of time was rounded to the nearest 5 minutes.
(d) Testing activity performed by Battelle/USDA operator.
(e) Data downloads were incomplete or failed, probably as a result of a software failure observed regularly during the verification
test.
(f) Data completeness = the ratio of time that the Model 17C was not experiencing down time to the total time available for
monitoring ambient NH3 mixing ratios from the start of testing on 9/8/03 to the end of testing on 10/3/03. The total time that
was available for monitoring was 36,540 minutes or 609 hours.
33
-------
Table 6-8. Activities Performed During Phase II
Date
Time
Offline (a)
(minutes)
Down Time (b)
(minutes)
Service
Time (c)
(minutes)
Activity
10/20/03
10/20/03
10/22/03
10/20-
10/23/03
170
120
60
3,300
Alarm: Cooler temperature high, caused by
increased temperature of instrument trailer;
reduction in trailer temperature eliminated error; no
associated data loss
Delivered zero air and NH3 standards1-"1-1
Delivered zero air and NH3 standard®
Data lost; download failure1-"1-1
10/23/03
10/24/03
10/26/03
10/31/03
11/7/03
11/9/03
11/10/03
11/10/03
11/12/03
11/12/03
11/13/03
11/14/03
150
120
75
120
360
480
84
1,440
5
5
5
Delivered zero air and NH3 standards1-"1-1
Delivered zero air and NH3 standard1-"1-1
Instrument time does not change for daylight savings
Changed Teflon inlet filter
Changed dessicant
Data lost; download failure1-6-1
Delivered zero air and NH3 standard1-"1-1
Changed Teflon inlet filter
Delivered zero air and NH3 standard1-"1-1
Delivered zero air and NH3 standards'-"1-1
Performed interference tests'-"1-1
Delivered zero air and NH3 standard1-"1-1
Totals
1,679
60 (4,740)(e)
15
99% (86%)(e) data completeness® and 15-minute
service time.
(a) Time Offline = time that the Model 17C was taken offline for zero or standard gas measurements. The period over which time
offline was evaluated began at 8:00 a.m. on 10/20/03 and ended at the conclusion of testing at 5:00 p.m. on 11/14/03. The
amount of time was rounded to the nearest 5 minutes.
(b) Down Time = time that the Model 17C was not operating or was operating but not reporting reliable measurements. The period
over which down time was evaluated began at 8:00 a.m. on 10/20/03 and ended at the conclusion of testing at 5:00 p.m. on
11/14/03. The amount of time was rounded to the nearest 5 minutes.
(c) Service Time = time spent conducting routine operation and maintenance activities and troubleshooting problems. The period
over which service time was evaluated began at 8:00 a.m. on 10/20/03 and ended at the conclusion of testing at 5:00 p.m. on
11/14/03. The amount of time was rounded to the nearest 5 minutes.
(d) Testing activity performed by Battelle/USDA operator.
(e) Data downloads were incomplete or failed, probably as a result of a software failure observed regularly during the verification
test.
(f) Data completeness = the ratio of time that the Model 17C was not experiencing down time to the total time available for
monitoring ambient NH3 mixing ratios from the start of testing on 10/20/03 to the end of testing on 11/14/03. The total time that
was available for monitoring during Phase 2 was 35,280 minutes or 588 hours.
6.9 Data Completeness
During Phase I, the Model 17C was operating and collecting data for more than 99% of the
available time. However, because of difficulties associated with the data downloading procedure
(described in Section 6.8), only 66% of the data were recovered. Similarly, in Phase n, the Model
17C was operating and collecting data for more than 99% of the available time, but only 86% of
the data were recovered.
34
-------
Chapter 7
Performance Summary
The performance of the Model 17C was evaluated in two phases in this verification test. Table 7-1
presents a summary of the performance of the Model 17C NH3 analyzer during this verification
test.
35
-------
Table 7-1. Model 17C Performance Summary
Parameter
Results
Phase I
Phase II
Relative
accuracy1
(a)
Average RA = 3.7%
%D Range = - 10.2% to 2.
Average RA = 10.5%
%D Range = - 11.5% to -8.9%
Linearity
Range = 0 to 10,000 ppb
Slope = 1.03 (±0.01)
Intercept = -24 ppb (± 23)
r2 = 1.000
Range = 0 to 2,000 ppb
Slope = 0.90 (± 0.02)
Intercept = -0.6 ppb (±20.3)
r2 = 1.000
Precision
Average RSD = 0.3%
Range = 0.2% to 0.5%
Average RSD = 0.3%
Range = 0.2% to 0.6%
Response time
Rise time = 180 to 4,560 seconds®
Fall time = 120 to 180 seconds®
Rise time = 600 to 900 seconds®
Fall time = 1,200 seconds®
Calibration/
zero drift
No apparent drift in response to zero air or a
nominal 1,000-ppb NH3 gas standard during
Week 1 or Week 4.
No apparent drift in response to zero air or a
nominal 1,000-ppb NH3 gas standard during Week 1
or Week 4.
Interference
effects®
Interference check conducted
during Phase II.
Hydrogen sulfide (285 ppb): no apparent effect
Nitrogen dioxide (95 ppb): a small negative
response in zero air and 500 ppb NH3
1,3-butadiene (95 ppb): no apparent effect
Diethylamine (96 ppb): —50% response in both
zero air and 500 ppb NH3
Comparability
Slope = 1.20 (±0.05)
Intercept = 16 ppb (± 29)
r2 = 0.984
Slope = 0.86 (± 0.03)
Intercept = -0.5 ppb (± 3.
r2 = 0.990
Ease of use
Daily checks were simple and quick
Little skill required to operate
Minimal maintenance required
Regular data download necessary
Data download software unreliable
Daily checks were simple and quick
Little skill required to operate
Minimal maintenance required
Regular data download necessary
Data download software unreliable
Data
completeness
99% data collected, 66% data recovered®
99% data collected, 86% data recovered®
Relative accuracy is expressed as an average absolute value of the percent difference from NH3 gas standards.
Only 1-minute averaged data available for this test. Standards for rise time calculation were delivered for three hours.
Only 5-minute averaged data available for this test.
Calculated as the change in signal divided by the interferent gas concentration, expressed as a percentage.
Data loss due to incomplete/failed download attempts.
36
-------
Chapter 8
References
1. National Air Pollutant Trends, 1900-1998. EPA-454/R-00-02, U.S. Environmental
Protection Agency, Office of Air Quality Planning and Standards, Research Triangle Park,
North Carolina, 27711.
2. Test/QA Plan for Verification of Ambient Ammonia Monitors at Animal Feeding
Operations, Battelle, Columbus, Ohio, September 2003.
3. Determination of Reactive Acidic and Basic Gases and Acidity of Fine Particles (<2.5 /dm),
Environmental Protection Agency Compendium Method 10-4.2, EPA/625/R-96/01 OA, U.S.
Environmental Protection Agency, Office of Research and Development, Cincinnati, Ohio,
45268.
4. Operating Manual, ChemComb Model 3500 Speciation Sampling Cartridge, Revision A,
January 2000, Rupprecht & Patashnick Co., Inc. East Greenbush, New York, 12061.
5. Quality Management Plan (QMP) for the ETV Advanced Monitoring Systems Center, U. S.
EPA Environmental Technology Verification Program, prepared by Battelle, Columbus,
Ohio, Version 4.0, December 2002.
37
-------
Appendix A
Model 17C Checklist
A-l
-------
ETV Verification of Ambient Ammonia Monitors
Thermo Electron 17C
Checklist
Vendor Representative Contact Information
Change Teflon filter weekly Date changed:
Data download
Short records weekly Date:
Long records after calibration tests Date:
Check 17C and 17C Converter power is onCheck instrument display for time or alarm:
(a) If alarm:
1. Press menu
2. Press down arrow to alarm
3. Press enter
4. Press down arrow and record alarms
5. Notify vendor representative
Check desiccant: Change when 3/4 of column turns pink. Date Changed:
Signature:
Date:
Comments:
A-2
------- |