July 1999

Environmental Technology
Verification Report

TSI COMBUCHECK
Single Gas Monitor

Prepared by

Bairelle

. . . Putting Technology To Worl

Battelle Memorial Institute

Under a cooperative agreement with

mi EPA U.S. Environmental Protection Agency


-------
July 1999

Environmental Technology Verification

Report

Advanced Monitoring Systems

TSI COMBUCHECK
Single Gas Monitor

By

Thomas Kelly
Ying-Liang Chou
Susan J. Abbgy
Paul I. Feder
James J. Reuther
Karen Riggs

Battelle
Columbus, Ohio 43201


-------
Notice

The U.S. Environmental Protection Agency (EPA), through its Office of Research and Develop-
ment has financially supported and collaborated in the extramural program described here. This
document has been peer reviewed by the Agency and recommended for public release. 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 (ORD) 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. At present, there are twelve environmental technology areas
covered by ETV. Information about each of the environmental technology areas covered by ETV
can be found on the Internet at http://www.epa.gov/etv.htm.

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. In 1997, through a competitive cooperative agreement, Battelle Memorial Institute
was awarded EPA funding and support 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/07/07_main.htm.

Vll


-------
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. In particular we recognize Joseph
Tabor, Steve Speakman, and Joshua Finegold of Battelle, and William Buttner and Mel Findlay of
TSI, Inc.

Vlll


-------
Contents

Notice	ii

Foreword 	vii

Acknowledgments	 viii

List of Abbreviations	 xiii

1.	Background	1

2.	Technology Description 	2

3.	Test Design and Procedures 	4

3.1	Introduction 	4

3.2	Laboratory Tests 	5

3.2.1	Linearity 	7

3.2.2	Detection Limit 	8

3.2.3	Response Time 	8

3.2.4	Interrupted Sampling 	8

3.2.5	Interferences 	8

3.2.6	Pressure Sensitivity 	9

3.2.7	Ambient Temperature 	10

3.3	Combustion Source Tests 	11

3.3.1	Combustion Sources 	11

3.3.2	Test Procedures 	12

4.	Quality Assurance/Quality Control 	16

4.1	Data Review and Validation 	16

4.2	Deviations from the Test/QA Plan 	16

4.3	Calibration of Laboratory Equipment 	18

4.4	Standard Certifications 	18

4.5	Performance System Audits 	19

4.5.1	Internal Audits 	19

4.5.2	External Audit 	22

ix


-------
5.	Statistical Methods 	23

5.1	Laboratory Tests	23

5.1.1	Linearity	23

5.1.2	Detection Limit	24

5.1.3	Response Time	25

5.1.4	Interrupted Sampling 	25

5.1.5	Interferences	26

5.1.6	Pressure Sensitivity	26

5.1.7	Ambient Temperature	26

5.2	Combustion Source Tests 	27

5.2.1	Accuracy 	27

5.2.2	Zero/Span Drift 	27

5.2.3	Measurement Stability 	27

5.2.4	Inter-Unit Repeatability 	28

5.2.5	Data Completeness	28

6.	Statistical Results	29

6.1	Laboratory Tests 	29

6.1.1	Linearity 	29

6.1.2	Detection Limit 	31

6.1.3	Response Time 	32

6.1.4	Interrupted Sampling 	34

6.1.5	Interferences 	35

6.1.6	Pressure Sensitivity 	36

6.1.7	Ambient Temperature 	38

6.1.8	Zero and Span Drift 	39

6.2	Combustion Source Tests 	40

6.2.1	Relative Accuracy 	41

6.2.2	Zero and Span Drift 	45

6.2.3	Measurement Stability 	46

6.2.4	Inter-Unit Repeatability 	49

6.3	Other Factors 	50

6.3.1	Cost 	51

6.3.2	Data Completeness 	51

6.3.3	Maintenance/Operational Factors 	51

x


-------
7.	Performance Summary	52

8.	References	54

Appendix A: Data Recording Sheets 	A-l

Appendix B: External Technical Systems Audit Report 	B-l

Figures

Figure 2-1. TSI COMBUCHECK	2

Figure 3-1. Manifold Test Setup 	7

Tables

Table 3-1. Identity and Schedule of Verification Tests Conducted on

TSI COMBUCHECK Single Gas Montiors 	4

Table 3-2. Summary of Interference Tests Performed 	9

Table 3-3. Span Concentrations Provided Before and After Each Combustion Source .... 14

Table 4-1. Results of QC Procedures for Reference NOx Analyzers for Testing

TSI COMBUCHECK Monitors	17

Table 4-2.	Equipment Type and Calibration Date 	18

Table 4-3.	Observations and Findings from the Internal Technical Systems Audit 	20

Table 4-4.	Performance Evaluation Results	21

Table 6-la.	Data from NO Linearity Test of TSI COMBUCHECK Monitors 	29

Table 6-lb.	Data from N02 Linearity Test of TSI COMBUCHECK Monitors	30

Table 6-2.	Statistical Results for Test of Linearity	30

Table 6-3.	Estimated Detection Limits for TSI COMBUCHECK Monitors	31

Table 6-4.	Response Time Data for TSI COMBUCHECK Monitors	33

Table 6-5.	Response Time Results for TSI COMBUCHECK Monitors	34

xi


-------
Table 6-6. Data from Interrupted Sampling Test with TSI COMBUCHECK Monitors ... 34

Table 6-7. Pre- to Post-Test Differences as a Result of Interruption of Operation

of TSI COMBUCHECK Monitors	34

Table 6-8.	Data from Interference Tests on TSI COMBUCHECK Monitors	35

Table 6-9.	Results of Interference Tests of TSI COMBUCHECK Monitors	35

Table 6-10.	Data from Pressure Sensitivity Test for TSI COMBUCHECK Monitors	37

Table 6-11.	Pressure Sensitivity Results for TSI COMBUCHECK Monitors	37

Table 6-12.	Data from Ambient Temperature Test of TSI COMBUCHECK Monitors	38

Table 6-13.	Ambient Temperature Effects on TSI COMBUCHECK Monitors 	39

Table 6-14. Data from Linearity and Ambient Temperature Tests Used to Assess

Zero and Span Drift of the TSI COMBUCHECK Monitors 	40

Table 6-15. Zero and Span Drift Results for the TSI COMBUCHECK Monitors	40

Table 6-16a. Data from the Gas Rangetop in Verification Testing of TSI COMBUCHECK

Monitors 	42

Table 6-16b. Data from Gas Water Heater in Verification Testing of

TSI COMBUCHECK Monitors	42

Table 6-16c. Data from the Diesel Generator at High RPM in Verification Testing

of TSI COMBUCHECK Monitors	43

Table 6-17. Relative Accuracy of TSI COMBUCHECK Monitors	44

Table 6-18. Data Used to Assess Zero and Span Drift for TSI COMBUCHECK

Monitors on Combustion Sources	45

Table 6-19. Results of Zero and Span Drift Evaluation for

TSI COMBUCHECK Monitors	46

Table 6-20. Data from Extended Sampling Test with Diesel Generator at Idle,

Using TSI COMBUCHECK Monitors	47

Table 6-21. Results of Evaluation of Measurement Stability for

TSI COMBUCHECK Monitors	50

Table 6-22. Summary of Repeatability 	50

xii


-------
List of Abbreviations

AC

alternating current

AMS

Advanced Monitoring Systems

ANSI

American National Standards Institute

Btu/hr

British thermal unit per hour

ccm

cubic centimeter per minute

CEMS

continuous emission monitoring system

CO

carbon monoxide

co2

carbon dioxide

DC

direct current

EPA

U.S. Environmental Protection Agency

ETV

Environmental Technology Verification

FID

flame ionization detector

ft3

cubic feet

gpm

gallons per minute

kW

kilowatt

LOD

limit of detection

1pm

liters per minute

m3

cubic meters

nh3

anhydrous ammonia

NIST

National Institute of Standards and Technology

NO

nitric oxide

NOx

nitrogen oxides

no2

nitrogen dioxide

02

oxygen

PE

performance evaluation

ppm

parts per million, volume

ppmC

parts per million carbon

QA

quality assurance

QC

quality control

QMP

Quality Management Plan

rms

root-mean-square

RPM

revolutions per minute

SAS

Statistical Analysis System

Xlll


-------
SCAQMD	SCAQMD Air Quality Management District

SCR	selective catalytic reduction

S02	sulfur dioxide

UHP	ultra-high purity

XIV


-------
Chapter 1
Background

The U.S. Environmental Protection Agency (EPA) has created the Environmental Technology
Verification Program (ETV) 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 substantially 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, permitting, purchase and use of environmental technologies.

ETV works in partnership with recognized testing organizations, stakeholder groups consisting of
regulators, buyers and vendor organizations, 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 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 Memorial Institute, operate the Advanced Monitoring Systems (AMS) program under
ETV. The AMS program has recently evaluated the performance of portable nitrogen oxides
monitors used to determine emissions from combustion sources. This verification statement
provides a summary of the test results for the TSI COMBUCHECK Model 8750 Single Gas
monitor.

1


-------
Chapter 2
Technology Description

The objective of the ETV Advanced Monitoring Systems (AMS) pilot is to verify the
performance characteristics of environmental monitoring technologies for air, water, and soil. This
verification report provides results for verification testing of COMBUCHECK Model 8750
electrochemical NO and N02 single gas monitors, manufactured by TSI, Inc., St. Paul, Minn. The
following is a description of the TSI single gas monitors based on information provided by the
vendor.

The TSI COMBUCHECK is a hand-held single gas monitor
with interchangeable electrochemical sensors to measure
carbon monoxide (CO), oxygen (02), nitric oxide (NO),
nitrogen dioxide (N02), or sulfur dioxide (S02). The range
of the COMBUCHECK is 0 to 2000 ppm for carbon
monoxide and nitric oxide, 0 to 100 ppm for nitrogen
dioxide, 0 to 1000 ppm for sulfur dioxide, and 0 to 30
percent for oxygen. Only the capabilities for NO and N02
measurement were evaluated in this test.

The COMBUCHECK monitor can be used in a variety of
environments and for many applications, including measuring
flue gas concentrations to optimize combustion efficiency,
sampling gas levels near combustion appliances, and
monitoring ambient concentrations. A backlit display is pro-
vided for use in poor lighting areas. The COMBUCHECK
gas sensors are interchangeable with a simple field
calibration. The Model 8750 COMBUCHECK monitor has a
built-in pump to provide a fast response to changes in gas
concentrations. The flexible stainless steel probe has a liquid/
particulate filter. The COMBUCHECK will operate under
conditions from 0 to 50°C (32 to 122°F), while sources up
to 540°C (1,000°F) can be sampled. An optional portable
printer is also available to provide hard copy documentation
of readings while in the field.

Figure 2-1. COMBUCHECK

2


-------
The COMBUCHECK weighs 0.84 pound and measures 4.0" x 6.6" x 1.5". It can be operated for
over 24 hours on four AA alkaline batteries. An optional AC adapter is also available.

As the product name implies, the COMBUCHECK is intended primarily for rapid inspection and
maintenance checks of heaters, furnaces, and boilers. It is not intended for long-term or
continuous monitoring. This instrument is a new product for TSI, and its verification testing was
intended partly to determine how well a low-end single gas monitor would stand up to emission
analyzer conditions.

In this verification test, four COMBUCHECK monitors were used, two for NO and two for N02.
In all testing, the two NO (or N02) analyzers were operated simultaneously to assess unit-to-unit
repeatability. The four units were operated on AC power throughout verification testing.

3


-------
Chapter 3
Test Design and Procedures

3.1 Introduction

The verification test described in this report was conducted in January 1999 on commercial
portable nitrogen oxide monitors. The tests were conducted at Battelle in Columbus, Ohio,
according to procedures specified in the Test/QA Plan for Verification of Portable NO/NO 2
Emission Analyzers.{l) Verification testing of the monitors involved the following tests:

1.	A series of laboratory tests in which certified NO and N02 standards were used to
challenge the monitors over a wide concentration range under a variety of conditions.

2.	Tests using three realistic combustion sources, in which data from the monitors
undergoing testing were compared to chemiluminescent NO and NOx measurements made
following the guidelines of EPA Method 7E.(2)

These tests are listed in Table 3-1.

Table 3-1. Identity and Schedule of Verification Tests Conducted on TSI COMBUCHECK
Single Gas Monitors

Test Activity

Date Conducted

Laboratory Tests



Linearity

January 15, 1999, p.m.

Interrupted Sampling

January 15, p.m. - January 16, a.m.

Interferences

January 16, a.m.

Pressure Sensitivity

January 16, a.m.

Ambient Temperature

January 16, p.m.

Source Tests



Gas Rangetop

January 17, a.m.

Gas Water Heater

January 17, a.m.

Diesel Generator High RPM

January 17, p.m.

To assess inter-unit variability, four identical TSI COMBUCHECK monitors, two with NO
sensors and two with N02 sensors, were tested simultaneously. These four monitors were

4


-------
designated as Unit A NO, Unit A N02, Unit B NO, and Unit B N02 throughout all testing. The
commercial monitors were operated at all times by a representative of TSI so that each monitor's
performance could be assessed without concern about the familiarity of Battelle staff with the
monitors. At all times, however, the TSI representative was supervised by Battelle staff.

Displayed NO and N02 readings from the monitors (in ppm) were manually entered onto data
sheets prepared before the test by Battelle. Battelle staff filled out corresponding data sheets,
recording, for example, the challenge concentrations or reference analyzer readings, at the same
time that the TSI operator recorded data. This approach was taken because visual display of
measured NO and N02 (or NOx) concentrations was the "least common denominator" of data
transfer among several N0/N02 monitors tested. Copies of the blank data recording sheets used
by Battelle and vendor staff are included as Appendix A of this report.

Verification testing began with TSI staff setting up and checking out the four TSI monitors in the
laboratory at Battelle. Once vendor staff were satisfied with the operation of the monitors, the
laboratory tests were begun. These tests were carried out in the order specified in the test/QA
plan.(1) Upon completion of laboratory tests, the monitors were moved to a nearby building where
the combustion sources described below were set up, along with two chemiluminescent nitrogen
oxides monitors which served as the reference analyzers. The combustion source tests were
conducted indoors, with the combustion source exhausts vented through the roof of the test
facility. This arrangement assured that testing was not interrupted and that no bias in testing was
introduced as a result of the weather. Sampling of source emissions began with the combustion
source emitting the lowest NOx concentration and proceeded to sources emitting progressively
more NOx. In all source sampling, the monitors being tested sampled the same exhaust gas as did
the reference analyzers. This was accomplished by inserting the TSI monitors' gas sampling
probes into the same location in the exhaust duct as the reference analyzers' probe.

3.2 Laboratory Tests

The laboratory tests were designed to challenge the monitors over their full nominal response
ranges, which for the TSI COMBUCHECK monitors were 0 to 2,000 ppm for NO and 0 to
100 ppm for N02. The lab tests were aimed at quantifying the full range of performance of the
monitors.

Laboratory tests were conducted using certified standard gases for NO and N02, and a gas
dilution system with flow calibrations traceable to the National Institute of Standards and
Technology (NIST). The NO and N02 standards were diluted in high purity gases to produce a
range of accurately known concentrations. The NO and N02 standards were EPA Protocol 1
gases, obtained from Scott Specialty Gases, of Troy, Michigan. As required by the EPA
Protocol(3) the concentration of these gas standards was established by the manufacturer within
1 percent accuracy using two independent analytical methods. The concentration of the NO
standard (Scott Cylinder Number ALM 057210) was 3,925 ppm, and that of the N02 standard
(Scott Cylinder Number ALM 019660) was 493.2 ppm. These concentrations were confirmed in a
performance evaluation audit performed near the end of the verification tests, by comparison with
independent standards obtained from other suppliers.

5


-------
The gas dilution system used was an Environics Model 2020 mass flow controlled diluter (Serial
Number 2108). This diluter incorporated four separate mass flow controllers, having respective
ranges of 10, 10, 1, and 0.11pm. This set of flow controllers allowed accurate dilution of gas
standards over a very wide range of dilution ratios, by selection of the appropriate flow con-
trollers. The mass flow calibrations of the controllers were checked against a NIST standard by
the manufacturer prior to the verification test, and were programmed into the memory of the
diluter. In verification testing, the Protocol Gas concentration, inlet port, desired output concen-
tration, and desired output flow rate were entered by means of the front panel keypad of the 2020
diluter, and the diluter then set the required standard and diluent flow rates to produce the desired
mixture. The 2020 diluter indicated on its front panel display the actual concentration being
produced, which in some cases differed very slightly from the nominal concentration requested. In
all cases the actual concentration produced was recorded as the concentration provided to the
monitors undergoing testing. The 2020 diluter also provided warnings if a flow controller was
being operated at less than 10 percent of its working range, i.e., in a flow region where flow
control errors might be enhanced. Switching to another flow controller then minimized the
uncertainties in the preparation of the standard dilutions.

Dilution gases used in the laboratory tests were Acid Rain CEM Zero Air and Zero Nitrogen from
Scott Specialty Gases. These gases were certified to be of 99.9995 percent purity, and to have the
following maximum content of specific impurities: S02 <0.1 ppm, NOx <0.1 ppm, CO < 0.5
ppm, C02 < 1 ppm, total hydrocarbons <0.1 ppm, and water < 5 ppm. In addition the nitrogen
was certified to contain less than 0.5 ppm of oxygen, while the air was certified to contain 20 to
21 percent oxygen.

Laboratory testing was conducted primarily by supplying known gas mixtures to the monitors
from the Environics 2020, using a simple manifold that allowed the monitors to sample the same
gas. The experimental setup is shown schematically in Figure 3-1. The manifold itself consisted of
a 9.5-inch length of thin-walled 1-inch diameter 316 stainless steel tubing, with 1/4-inch tubing
connections on each end. The manifold had three 1/4-inch diameter tubing side arms extending
from it: two closely spaced tubes are the sampling points from which sample gas was withdrawn
by the monitors, and the third provided a connection for a Magnehelic differential pressure gauge
(±15 inches of water range) that indicated the manifold pressure relative to the atmospheric
pressure in the laboratory. Gas supplied to the manifold from the Environics 2020 always
exceeded by at least 0.5 1pm the total sample flow withdrawn by the monitors. The excess vented
through a "T" connection on the exit of the manifold, and two coarse needle valves were
connected to this "T," as shown in Figure 3-1. One valve controlled the flow of gas out the
normal exit of the manifold, and the other was connected to a small vacuum pump. Closing the
former valve elevated the pressure in the manifold, and opening the latter valve reduced the
pressure in the manifold. Adjustment of these two valves allowed close control of the manifold
pressure within a target range of ±10 inches of water, while maintaining excess flow of the gas
mixtures to the manifold. The arrangement shown in Figure 3-1 was used in all laboratory tests,
with the exception of interference testing. For most interference testing, gas standards of the
appropriate concentrations were supplied directly to the manifold, without use of the Environics
2020 diluter.

6


-------
Figure 3-1. Manifold Test Setup

Laboratory testing consisted of a series of separate tests evaluating different aspects of monitor
behavior. The procedures for those tests are described below, in the order in which the tests were
actually conducted. The statistical procedures that were applied to the data from each test are
presented in Chapter 5 of this report.

3.2.1 Linearity

At the request of the TSI representative, before starting the linearity test the TSI NO monitors
were calibrated with a 200 ppm NO concentration, and the TSI N02 monitors were calibrated
with a 25 ppm N02 concentration. Both of these calibration mixtures were prepared using the
EPA Protocol Gases and Environics calibrator.

The linearity of monitor response was then tested by wide-range multipoint calibrations with NO
and N02. Linearity testing consisted of a 21-point response check for NO, and for N02. At the
start of this check, the TSI monitors were provided with the appropriate zero gas, and then with
an NO or N02 span gas concentration at the respective nominal full scale of the monitors (i.e.,
2,000 ppm NO or 100 ppm N02). No adjustment was made to the monitors to match that span
value, and the 21-point check proceeded without further adjustments. The 21 points consisted of
three replicates each at 10, 20, 40, 70, and 100 percent of the nominal range, in randomized
order, and interspersed with six replicates of zero gas.(1) Following completion of all 21 points,
the zero and 100 percent spans were repeated, also without adjustment of the monitors. This
entire procedure was performed for NO and then for N02.

7


-------
3.2.2	Detection Limit

Data from zero gas and from 10 percent of full-scale points in the linearity test were used to
establish the NO and N02 detection limits of the monitors, using a statistical procedure defined in
the test/QA plan.(1)

3.2.3	Response Time

During the NO and N02 linearity tests, upon switching from zero gas to an NO or N02
concentration of 70 percent of the respective full scale (i.e., about 1,400 ppm NO or 70 ppm
N02), the monitors' responses were recorded at 10-second intervals until fully stabilized. These
data were used to determine the response times for NO and for N02, defined as the time to reach
95 percent of final response after switching from zero gas to the calibration gas.

3.2.4	Interrupted Sampling

After the zero and span checks that completed the linearity test, the TSI monitors were shut down
(i.e., their electrical power was turned off overnight), ending the first day of laboratory testing.
The next morning the monitors were powered up, and the same zero gas and span concentrations
were run without adjustment of the monitors. Comparison of the NO and N02 zero and span
values before and after shutdown indicated the extent of zero and span drift resulting from the
shutdown. Full-scale NO and N02 levels (i.e., 2,000 ppm NO and 100 ppm N02) were used as the
span values in this test.

3.2.5	Interferences

Following monitor startup and completion of the interrupted sampling test, the second day of
laboratory testing continued with interference testing. This test evaluated the response of the TSI
monitors to species other than NO and N02. The potential interferants listed in Table 3-2 were
supplied to the monitors one at a time, and the NO and N02 readings of the monitors were
recorded. The potential interferants were single components, except for a mixture of S02 and NO,
which was intended to assess whether S02 in combination with NO produced a bias in NO
response.

8


-------
Table 3-2. Summary of Interference Tests Performed

Interferant

Interferant	Concentration

CO	496 ppm

C02	5.03%

S02	501 ppm

NH3	494 ppm

Hydrocarbon Mixture*	485 ppm Cl3 98 ppm C2,

48 ppm C3 + C4

S02 and NO	451 ppm S02 + 383 ppm NO

*C = methane; C2 = ethane; and C3 + C4 = 24 ppm propane + 24 ppm n-butane.

The CO, C02, S02, and NH3 mixtures used in the interference test were all obtained as Certified
Master Class Calibration Standards from Scott Technical Gases, at the concentrations indicated in
Table 3-2. The indicated concentrations were certified by the manufacturer to be accurate within
± 2 percent, based on analysis. The CO, C02, and NH3 mixtures were all in Ultra-High Purity
(UHP) air, and the S02 mixture was in UHP nitrogen. The S02/N0 mix listed in Table 3-2 was
prepared by diluting the S02 standard with the NO Protocol Gas using the Environics 2020.

The hydrocarbon mixture listed in Table 3-2 was prepared at Battelle in UHP hydrocarbon-free
air, starting from the pure compounds. Small quantities of methane, ethane, propane, and n-
butane were injected into a cylinder that was then pressurized with UHP air. The required
hydrocarbon concentrations were approximated by the preparation process, and then quantified by
comparison with aNIST standard containing 8.61 ppm carbon (ppmC) in the form of propane.
Using a gas chromatograph with a flame ionization detector (FID) the NIST standard was first
analyzed twice, producing peak areas of 18,627 and 18,791 area units per 8.61 ppmC of propane.
The average FID response factor (18,709 units (±116 units)/8.61 ppmC) was then used to
determine the concentrations of the components of the prepared hydrocarbon mixture. Two
analyses of that mixture both gave a result of 485 ppm methane; the corresponding results for
ethane were 97 and 98 ppm; for propane 23 and 24 ppm; and for n-butane 24 and 25 ppm.

In the interference test, each interferant in Table 3-2 was provided individually to the sampling
manifold shown in Figure 3-1, at a flow in excess of that required by the TSI monitors. Each
period of sampling an interferant was preceded by a period of sampling the appropriate zero gas.

3.2.6 Pressure Sensitivity

The pressure sensitivity test was designed to quantify the dependence of monitor response on the
pressure in the sample gas source. By means of two valves at the downstream end of the sample

9


-------
manifold (Figure 3-1), the pressure in the manifold could be adjusted above or below the ambient
room pressure, while supplying the manifold with a constant ppm level of NO or N02 from the
Environics dilution system. This capability was used to determine the effect of the sample gas
pressure on the sample gas flow rate drawn by the monitors, and on the NO and N02 response.

The dependence of sample flow rate on pressure was determined using an electronically timed
bubble flow meter (Ultra Flow Primary Gas Flow Calibrator, Model 709, Serial No. 010928;
SKC, Inc.). This flow meter was connected in line (i.e., inserted) into the sample flow path from
the manifold to one of the commercial monitors. Zero gas was supplied to the manifold at ambient
pressure, and the monitor's sample flow rate was measured with the bubble meter. The manifold
pressure was then adjusted to -10 inches of water relative to the room, and the monitor's flow
rate was measured again. The manifold pressure was adjusted to +10 inches of water relative to
the room, and the flow rate was measured again. The bubble meter was then moved to the sample
inlet of the other commercial monitor, and the flow measurements were repeated.

The dependence of NO and N02 response on pressure was determined by sampling the
appropriate zero gas, and an NO or N02 span gas equivalent to 70 percent of the respective full
scale, at each of the same manifold pressures (room pressure, -10 inches, and +10 inches). This
procedure was conducted simultaneously on both monitors, first for NO at all three pressures, and
then for N02 at all three pressures. The data at different pressures were used to assess zero and
span drift resulting from the sample pressure differences.

3.2.7 Ambient Temperature

The purpose of the ambient temperature test was to quantify zero and span drift that may occur as
the monitors are subjected to different temperatures during operation. This test involved
providing the monitors with zero and span gases for NO and N02 (at the same 70 percent of
nominal range values used in the pressure test) at room, elevated, and reduced temperatures. A
temperature range of 7 to 41 °C (45 to 105°F) was targeted in this test. The elevated temperature
condition was achieved using a 1.43 m3 steel and glass laboratory chamber, thermostated at 41 °C
(105 °F) using external heat lamps. The reduced temperature condition was achieved using a
conventional domestic refrigerator (Crosley Model CT19A5W) with a refrigerator volume of 13.1
ft3.

The general procedure was to provide zero and span gas for NO, and then for N02, to the
monitors at room temperature, and then to place the monitors and the sampling manifold into the
heated chamber. Electrical and tubing connections were made through a small port in the lower
wall of the chamber. A thermocouple readout was used to monitor the chamber temperature and
room temperature. After 1 hour or more of stabilization in the heated chamber, the zero and span
tests were repeated. The monitors, manifold, and other connections were then transferred to the
refrigerator. After a 1-hour stabilization period of 1 hour or more, the zero and span checks were
repeated at the reduced temperature. The monitors were returned to the laboratory bench; and,
after a 1-hour stabilization period, the zero and span checks were repeated a final time.

10


-------
3.3 Combustion Source Tests
3.3.1 Combustion Sources

Three combustion sources (a gas rangetop, a gas residential water heater, and a diesel engine)
were used to generate NOx emissions from less than 10 ppm to over 150 ppm. Emissions
databases for two of these sources (rangetop and water heater) exist as a result of prior
measurements, both of which have been published.(4'5)

3.3.1.1	Rangetop

The low-NOx source was a residential natural gas fired rangetop (KitchenAid Model 1340),
equipped with four cast-iron burners, each with its own onboard natural gas and combustion air
control systems. The burner used (front-left) had a fixed maximum firing rate of about 8 KBtu/hr.

The rangetop generated NO in the range of about 6 to 7 ppm, and N02 in the range of about 2 to
3 ppm. The database on this particular appliance was generated in an international study in which
15 different laboratories, including Battelle, measured its NO and N02 emissions.(4)

Rangetop NOx emissions were diluted prior to measurement using a stainless-steel collection
dome, fabricated according to specifications of the American National Standards Institute (ANSI
Z21.1).(6) For all tests, this dome was elevated to a fixed position 2 inches above the rangetop
surface. Moreover, for each test, a standard "load" (pot) was positioned on the grate of the
rangetop burner. This load was also designed according to ANSI Z21.1 specifications regarding
size and material of construction (stainless steel). For each test, the load contained 5 pounds of
room-temperature water.

The exit of the ANSI collection dome was modified to include seven horizontal sample-probe
couplers. One of these couplers was 1/4-inch in size, three were 3/8-inch in size, and three were
1/2-inch in size. Their purpose was to accommodate various sizes of vendor probes and one
reference probe simultaneously during combustion-source sampling.

This low-NOx combustion source was fired using "standard" natural gas, obtained from Praxair,
Inc., which was certified to contain 90 percent methane, 3 percent ethane, and the balance
nitrogen. This gaseous fuel contained no sulfur.

3.3.1.2	Water Heater

The medium-NOx source was a residential natural gas-fired water heater (Ruud Model P40-7) of
40-gallon capacity. This water heater was equipped with one stamped-aluminum burner with its
own onboard natural gas and combustion air control systems, which were operated according to
manufacturer's specifications. The burner had a fixed maximum firing rate of about 40 KBtu/hr.
Gas flow to the water heater was monitored using a calibrated dry-gas meter.

The water heater generated NO emissions in the range of 80 to 120 ppm, and N02 in the range of
5 to 15 ppm. NOx emissions dropped as the water temperature rose after ignition, stabilizing at

11


-------
the lower end of the ranges noted above. To assure constant operation of the water heater, a
continuous draw of 3 gpm was maintained during all verification testing. The database on this
particular appliance was generated in a national study in which six different laboratories measured
its emissions, including Battelle.(5)

Water heater NOx emissions were not diluted prior to measurement. The draft hood, integral to
the appliance, was replaced with a 3-inch diameter, 7-inch long stainless-steel collar. The exit of
this collar was modified to include five horizontal sample-probe couplers. One coupler was
1/4-inch in size, whereas the two other pairs were either 3/8- or 1/2-inch in size. Their purpose
was to accommodate various sizes of vendor probes and one reference probe simultaneously
during sampling.

This medium-NOx combustion source was fired on house natural gas, which contained odorant-
level sulfur (4 ppm mercaptan). The composition of this natural gas is essentially constant, as
monitored by a dedicated gas chromatograph in Battelle's laboratories.

3.3.1.3 Diesel Engine

The high-NOx source was an industrial diesel 8 kW electric generator (Miller Bobcat 225D Plus),
which had a Deutz Type ND-151 two-cylinder engine generating 41 KBtu/hr (16 horsepower). In
testing of the TSI COMBUCHECK monitors, this device generated NOx emissions up to about
150 ppm, during operation at high load (3,500 RPM). About 60 to 70 ppm of the NOx was N02.
The database on the diesel generator emissions was generated in tests conducted in the 2 weeks
prior to the start of the verification tests.

NOx emissions from this engine were not diluted prior to measurement. The 1-inch exhaust outlet
of the engine, which is normally merely vented to the atmosphere, was fitted with a stack designed
to meet the requirements of the U.S. EPA (Method 5).(9) The outlet was first expanded to 2 inches
of 1.5-inch diameter copper tubing, then to 15 inches of 2-inch diameter copper tubing, and finally
to 2 inches of 3-inch diameter copper tubing. The 3-inch diameter tubing was modified to include
five horizontal sample-probe couplers. One of these couplers was 1/4-inch in size, two were 3/8-
inch in size, and two were 1/2-inch in size. These couplers held the sample probes in place. The
3-inch tube was connected to a 3-inch stack extending through the roof of the test laboratory.

This high-NOx combustion source was fired on commercial diesel fuel, which, by specification,
contains only 0.03 to 0.05 weight percent sulfur.

3.3.2 Test Procedures

The procedures followed during combustion source testing consisted of those involved with the
sampling systems, reference method, calibration gas supply, and the sources, as follows.

12


-------
3.3.2.1	Sampling Systems

Prior to sampling, the TSI representative inserted the four monitors' probes into the exhaust duct
of the rangetop, water heater, or diesel engine. The TSI probes were located closely together in
pairs, sampling from a point within about 1/4 inch of the inlet of the reference analyzers' probe.

The reference analyzer probe consisted of a 26-inch long, 1/4-inch diameter stainless-steel tube,
the upstream 2 inches of which were bent at a right angle for passage into the center of the source
exhaust duct. Each combustion source had a dedicated sampling probe, connected to the
reference analyzers with 1/4-inch tubing. Because of the small size of the TSI monitors, the
lengths of sample-transfer tubing required to connect the TSI instruments to the rangetop, water
heater, and diesel engine were all less than 4 feet. The lengths of sample-transfer tubing required
to connect reference instruments to the rangetop, water heater, and diesel engine were about
7 feet, 9 feet, and 4 feet, respectively.

The TSI monitors used unheated 1/8" sample probes. Neither the reference sampling probe nor
the reference sample-transfer lines were heated. Visible condensation of combustion-generated
water did not occur. The reference analyzer moisture-removal system consisted of a simple ice
bath (0°C, 32 °F). The reference particulate-removal system consisted of a 47-millimeter in-line
quartz fiber filter.

3.3.2.2	Reference Method

The reference method of NO determination against which the commercial monitors were
compared was the ozone chemiluminescence method for NO that forms the basis of EPA Method
7E.(2) The reference measurements were made using a Model 10 and a Model 14A source-level
NOx monitor (both from Thermo Environmental Instruments), located side-by-side near the
combustion sources. These monitors sampled from a common intake line and operated on
identical ranges of 100 ppm or 1,000 ppm full scale, depending on the source. Both instruments
use stainless steel catalytic converters maintained at 650°C (1,202°F) for reduction of N02 to NO
for detection. Digital electronic voltmeters were connected directly to the amplifier output of the
monitors, to provide direct digital display of the data. The Model 10 and 14A monitors provide
sequential, rather than simultaneous, measurement of NO andNOx, so display of both readings
required manual switching of sampling modes on both instruments. This requirement resulted in
the NO and NOx readings from the reference analyzers being separated in time by about
15 seconds, due to the stabilization needed after switching. This effect is believed to have
negligible impact on the verification results due to the stability of source emissions.

The chemiluminescence analyzers were calibrated using the Environics Series 100 and the EPA
Protocol 1 gases. The calibration procedure was specified in the test/QA plan, and required
calibration at zero, 30 percent, 60 percent, and 100 percent of the applicable range value (i.e., 100
or 1,000 ppm). Calibration results closest in time to the verification source test were used to
establish scale factors applicable to the source test data. The conversion efficiency of the stainless
steel converters was determined by calibrating with both NO and N02 on the applicable ranges,
using the EPA Protocol 1 gases. The ratio of the linear regression slope of the N02 calibration to

13


-------
that of the NO calibration determined the N02 conversion efficiency. For the TSI source tests,
which took place on January 17, 1999, calibration data from January 16 were applied. Conversion
efficiency values of 91.8 percent and 90.6 percent were found for the Model 14A and Model 10
monitors, respectively.

3.3.2.3 Calibration Gas Supply

Before and after sampling of each combustion source, both the monitors undergoing testing and
the reference analyzers were supplied with zero gas and with standard NO and N02 mixtures at
levels comparable to those expected from the source. To prepare these mixtures, Protocol 1 gases
identical to those used in the laboratory testing were diluted using an Environics Series 100
Computerized Multi-Gas Calibrator (Serial Number 2416). The same Acid Rain CEM zero gases
were used for dilution and zeroing as were used in the laboratory tests. When low dilution ratios
were required for some calibration points, Tylan FC-260 (3 1pm) and FC-280 (5 1pm) mass flow
controllers were used instead of the Environics calibrator. The Tylan flow controllers were
calibrated using the same SKC electronic bubble flow meter used in the laboratory tests, and were
operated with a Tylan four-unit control and readout device. The pre- and post-test span values
used with each combustion source are given in Table 3-3.

Table 3-3. Span Concentrations Provided Before and After Each Combustion Source

Source

NO Span Level (ppm)

N02 Span Level (ppm)

Gas Rangetop

20

10

Gas Water Heater

100

15

Diesel-High RPM

200

50

The pre- and post-test zero and span values were used to assess the drift in zero and span
response of the reference and tested analyzers caused by exposure to source emissions.

3.3.2.4 Operation of Sources

Verification testing was conducted with the combustion sources at or near steady-state in terms of
NOx emission. For the rangetop, steady-state was achieved after about 15 minutes, when the
water began to boil. For the water heater, steady-state was achieved in about 15 minutes, when its
water was fully heated. Because the water heater tank had a thermostat, cycling would have
occurred had about 3 gpm of hot water not been continuously drained out of the tank.

For the diesel engine, steady-state was achieved in about 10 minutes of operation. The diesel was
operated at full speed (3,500 RPM) to achieve its lowest NOx emissions.

The order of operation of the combustion sources was (1) rangetop, (2) water heater, and
(3) diesel engine. This allowed the monitors to be exposed to continuously increasing NO and

14


-------
N02 levels, and avoided interference in low level measurements that might have resulted from
prior exposure to high levels.

Sampling of each combustion source consisted of obtaining nine separate measurements of the
source emissions. After sampling of pre-test zero and span gases provided from the calibration
source, and with both the reference and vendor monitors sampling the source emissions, the TSI
operator indicated when he was ready to take the first set of readings (a set of readings consisting
of the NO and N02 responses on the four TSI monitors). At that time the Battelle operator of the
reference analyzers also took corresponding readings. The monitors undergoing testing were then
disconnected from the source, and allowed to sample room air until readings dropped well below
the source emissions levels. The monitors were then reconnected to the source, and after
stabilizing another set of readings was taken. There was no requirement that monitor readings
drop fully to zero between source measurements. This process was repeated until a total of nine
readings had been obtained with both the vendor and reference monitors. The same zero and span
gases were then sampled again before moving to the next combustion source.

One addition to this procedure was the extended sampling test, conducted as the last operation in
the combustion source testing. This test involved continuous sampling of the diesel engine
emissions for a full hour with no intervals of room air sampling. Data were recorded for both
reference and vendor monitors at 1-minute intervals throughout that hour of measurement. This
test was conducted only after nine sequential sets of readings had been obtained from all the
combustion sources by the procedure described above. The COMBUCHECK single gas monitors
are not intended for extended continuous sampling of combustion sources, and so had not been
tested in this application. However, with the agreement of the TSI operator, the monitors were
subjected to the extended sampling test.

15


-------
Chapter 4
Quality Assurance/Quality Control

Quality control (QC) procedures were performed in accordance with the quality management plan
(QMP) for the AMS pilot(7) and the test/QA plan(1) for this verification test.

4.1	Data Review and Validation

Test data were reviewed and approved according to the AMS pilot QMP, the test/QA plan, and
Battelle's one-over-one approval policy. The Verification Testing Leader reviewed the raw data
and data sheets that were generated each day and approved them by adding his signature and date.
Laboratory record notebooks were also reviewed, signed, and dated by the Verification Testing
Leader.

Other data review focused upon the compliance of the chemiluminescent reference analyzer data
with the quality requirements of Method 7E. The purpose of validating reference data was to
ensure usability for the purposes of comparison with the demonstration technologies. The results
of the review of the reference analysis data quality are shown in Table 4-1. The data generated by
the reference analyzers were used as a baseline to assess the performance of the technologies for
N0/N02 analysis.

4.2	Deviations from the Test/QA Plan

During the physical set up of the verification test, deviations from the test/QA plan were made to
better accommodate differences in vendor equipment, availability of Battelle personnel and
equipment, and other changes or improvements. Any deviation required the approval signature of
Battelle's Verification Testing Leader and the pilot manager. A planned deviation form was used for
documentation and approval of the following changes:

1.	Dr. Agnes Kovacs did not participate in the statistical analysis of data from the verification
test.

2.	The order of testing was changed, and a span value of 70 percent of range (rather than
100 percent) was used in the pressure sensitivity test.

3.	The order of the ambient temperature test was changed.

4.	The exact concentrations used in the mixture of S02 and NO for the interference test were
changed.

5.	A different diesel generator was used than that originally planned.

16


-------
6.	An oxygen sensor was not used during source tests.

7.	Thermo Environmental Models 14A/10 NO/NOx analyzers were used for reference method.

8.	Triplicate calibration points were not run on reference method analyzers.

9.	Unheated sample line and tubing were used, based on the experience of Battelle staff in
sampling the combustion sources used in this test, and other combustion sources.

Table 4-1. Results of QC Procedures for Reference NOx Analyzers for Testing TSI
COMBUCHECK Monitors

N02 conversion
efficiency

N02 conversion
efficiency

Calibration of reference Meets criteria
method using four points (r2 > 0.9999)
at 0, 30, 60, 100% for
NO

Calibration of reference Meets criteria
method using four points (r2 > 0.9999)
at 0, 30, 60, 100% for
N02

91.8% for Model 14A in 100
ppm and 1,000 ppm ranges

90.6% for Model 10 in 100 ppm
and 1,000 ppm ranges

Calibrations
(100 ppm range)

Meets ± 2% requirement
(relative to span)

Model 10

Model 14A

NO

NO

Error, % of % of Scale

Error, % of % of Scale

Span

Span

0.5% 30%

0.9% 30%

0.4% 60%

0.4% 60%

no2

no2

Error, % of % of Scale

Error, % of % of Scale

Span

Span

0.3%
0.7%

30%
60%

0.2%
0.5%

30%
60%

Zero drift	Meets ± 3% requirement

(relative to span) on all
combustion sources

Span drift	Meets ± 3% requirement

(relative to span) on all
combustion sources

Interference check < ± 7% (No interference
	response observed)	

17


-------
There was one undocumented deviation. Due to a delay in the arrival of the protocol gases used
in the verification test, Battelle was not able to run one instrument through the entire test
sequence prior to verification testing. This deviation has no impact on the final data, for the
reasons described in the Performance System Audits section of this report.

4.3 Calibration of Laboratory Equipment

Equipment used in the verification test required calibration before use. Equipment types and
calibration dates are listed in Table 4-2. Documentation for calibration of the following equipment
was required before use in the verification test, and was maintained in the test file.

Table 4-2. Equipment Type and Calibration Date

Equipment Type

Calibration Date/



Temperature Check

Flow Controllers (Gas Dilution System) Environics Series 100

6/11/98

Flow Controllers (Gas Dilution System) Environics Model 2020

12/16/98

Digital Temperature Indicator Model 402A

1/7/99

Dwyer Magnahelic Pressure Gauge

1/11/99

Model R-275 In-line Dry Gas Meter

1/11/99

Doric Trendicator 400A Thermocouple Temperature Sensor

1/18/99

Model DTM-115 Reference Dry Gas Meter

9/22/98

4.4 Standard Certifications

Standard or certified gases were used in all verification tests, and certifications or analytical data
were kept on file to document the traceability of the following standards:

#	EPA Protocol Gas Nitrogen Dioxide

#	EPA Protocol Gas Nitric Oxide

#	Certified Master Class Calibration Standard Sulfur Dioxide

#	Certified Master Class Calibration Standard Carbon Dioxide

#	Certified Master Class Calibration Standard Ammonia

#	Certified Master Class Calibration Standard Carbon Monoxide

#	Nitrogen Acid Rain CEM Zero

#	Acid Rain CEM Zero Air

#	Battelle-Prepared Organics Mixture.

All other QC documentation and raw data for the verification test are located in the test file at
Battelle, to be retained for 7 years and made available for review if requested.

18


-------
4.5 Performance System Audits

Internal and external performance system audits were conducted and the results are summarized
in the following sections.

4.5.1 Internal Audits

Three internal audits were conducted during verification testing. A technical systems audit was
conducted to assess the physical setup of the test, a performance evaluation audit was conducted
to evaluate the accuracy of the measurement system, and an audit of data quality was conducted
on 10 percent of all data generated during the verification test. A summary of the results of these
audits is provided below.

4.5.1.1	Techni cal Sy stem s Audit

A technical systems audit is a qualitative onsite audit of the physical setup of the test. The
auditors determine the compliance of testing personnel with the test/QA plan. A self-assessment is
required for each test as outlined in the AMS pilot QMP. The QA/QC Reviewer for the verifi-
cation test conducted the internal technical systems audit on January 18, 1999. Observations and
findings from this audit are listed in Table 4-3.

4.5.1.2	Performance Evaluation Audit

The performance evaluation audit was a quantitative audit in which measurement standards were
independently obtained and compared with those used in the verification test to evaluate the
accuracy of the measurement system. One such assessment was conducted by Battelle QA staff on
February 4, 1999. No independent assessments of this type were conducted by EPA staff.

The performance evaluation (PE) samples were NO and N02 calibration gases independent of the
test calibration standards containing certified concentrations of NO andN02. Accuracy of the
reference analyzers was determined by comparing the measured N0/N02 concentrations using the
verification test standards with those obtained using the certified PE values. Percent difference
was used to quantify the accuracy of the results. The PE sample for NO was an EPA protocol gas
having nearly the same concentration as the NO standard used in verification testing, but
purchased from a different commercial supplier. The PE standard for N02 was a commercial
standard of 50.5 ppm N02 in air, whose concentration had been confirmed by comparison with a
50 ppm standard reference material of NO in nitrogen, obtained from the National Institute of
Standards and Technology. Table 4-4 summarizes the reference method performance evaluation
results. Included in this table are the performance acceptance ranges and the certified gas
concentration values. The acceptance ranges are guidelines established by the provider of the PE
materials to gauge acceptable analytical results. As shown in Table 4-4, all of the observed
concentrations were well within the acceptance ranges.

19


-------
Table 4-3. Observations and Findings from the Internal Technical Systems Audit

Observation/Finding

Corrective Action/Impact on Final Data

Method 7E calibration was not completed prior to
verification testing. Analyzers gave unreliable results
during first test, which prompted a calibration on
1/13/99. Full four-point calibration was not
performed until 1/15/99 on the 0-100 range and
1/16/99 on the 0-1000 range for both Models 14A
and 10. All criteria meet stated objectives in Method
7E for the calibration (linearity, calibration error)
performed on the 15th and 16th.

From Pressure Sensitivity Test conducted 1/12/99 an
explanation is needed of correction factor to be
applied to data.

Start and stop time for instruments to equilibrate at
each temperature is not noted on data sheets.

Calibration drift for all data reviewed is less than
± 3% relative to the span except Model 10 span post-
test on 1/14/99 for diesel engine test which = 3.6%.
Data for test should be flagged at minimum.

Data and calculations for calibration drift test not
found on test data sheets. Recommend a better system
be implemented for assessing quality of the
calibration drift for reference analyzers immediately
following collection of test data so decision whether
or not to proceed is clear to all participants.

Zero/span values are documented on diesel engine
test data sheets for all tests except on 01-13-98 post-
test blank with no explanation.

Vendor source testing that was conducted prior to the
first full four-point Method 7E calibration was
repeated at a later date. Thus all vendor testing was
conducted with fully calibrated reference analyzers.
There is no impact on verification data because the
first vendor test was repeated after Method 7E
calibration was implemented.

The 02 sensors of the vendor's monitors showed the
presence of 02 in the sample gas at a time when only
NO in pure N2 was being provided to the analyzers.
This indicated a leakage of air into the sample
manifold (which was at reduced pressure relative to
the room). The amount of dilution caused by the
leakage of air was calculated from the 02 level
observed, and exactly accounted for an apparently
low NO response from the vendor's monitors. That
is, the 02 data were used to correct the observed NO
responses to what they would have been with no air
leakage. The leakage was corrected, and did not
occur in testing of the TSI monitors.

Added start and stop time to data sheets as a method
to document equilibration.

All source tests with the TSI monitors met the 3%
requirement (see Table 4-1). No impact on final data.

Comparison of drift is easily made visually; written
comments will be added if termination of a test is
called for.

This test was terminated. Notes were added as
suggested and the test was later repeated in its
entirety. No impact on TSI test.

20


-------
Table 4-4. Performance Evaluation Results





Reading (V)

Zero (V)

Zero
Corrected

Apparent
Concentration*

Percent
Difference**

Limits

Unit 14A
Test Std

NO in N2
(ppm)
3,925

9.92

0.01

9.91

3905.3

0.5%

±2%

PE Std

3,988

10.13

0.01

10.12







Unit 10
Test Std

NO in N2
(ppm)
3,925

1.01

-0.01

1.03

3895.7

0.7%

±2%

PE Std

3,988

1.04

-0.01

1.05







Unit 14A
Test Std

N02 in
Air (ppm)
50.0***

4.40

0.01

4.39

48.7

2.5%

±5%

PE Std

50.5

4.56

0.02

4.54







Unit 10
Test Std

N02 in
Air (ppm)
50.0***

0.44

-0.01

0.45

50.0

0.1%

±5%

PE Std

50.5

0.44

-0.01

0.45







* Concentration of Test Standard indicated by comparison to the Performance Evaluation Standard.
** Percent difference of apparent concentration Relative to Test Standard concentration.
*** Prepared by dilution of 493.2 PPM N02 protocol gas.

4.5.1.3 Audit of Data Quality

The audit of data quality is a qualitative and quantitative audit in which data and data handling are
reviewed and data quality and data usability are assessed. Audits of data quality are used to
validate data at the frequency of 10 percent and are documented in the data audit report. The goal
of an audit of data quality is to determine the usability of test results for reporting technology
performance, as defined during the design process. Validated data are reported in the ETV
verification reports and ETV verification statement along with any limitations on the data and
recommendations for limitations on data usability.

The QA/QC Reviewer for the verification test audited 10 percent of the raw data. Test data sheets
and laboratory record books were reviewed, and calculations and other algorithms were verified.
Calibration drift test results were calculated and compared to the Method 7E criteria. Calculations
that were used to assess the four-point calibration of the reference method were also verified to be
correct.

21


-------
4.5.2 External Audit

EPA conducted an on-site technical systems audit during the verification testing. This audit was
conducted to observe and evaluate whether the verification team followed the test/QA plan. The
external technical systems audit report is attached in Appendix B and the assessment is
summarized below.

The auditors assessed the verification test procedures and personnel against the Quality Manage-
ment Plan for the ETV Advanced Monitoring Systems Pilot(7), the Test/QA Plan for Verification
of Portable NO/NO 2 Emission Analyzers,(1) and U.S. EPA Method 7E Determination of Nitrogen
Oxides Emissions from Stationary Sources (Instrumental Analyzer Procedure) {2) The auditors
were on site from January 20, 1999, through January 21, 1999. The technical systems audit was
performed on the flow rate and ambient temperature laboratory tests and the relative accuracy
tests with the gas rangetop, water heater, and a portion of the high RPM emissions of the diesel
generator. No performance evaluations were conducted as a part of this audit.

This external technical systems audit showed that the verification test was well managed,
particularly considering its complexity. All personnel appeared to be well trained for their
particular duties. All involved showed enthusiasm and ingenuity during the verification testing.
Personnel were very familiar with the test/QA plan. With one exception, differences for this
verification test from the original test/QA plan were well documented by deviation reports on file
at Battelle. The deviation report format includes a date, cites the deviation, provides an explana-
tion of the deviation, and requires a Battelle approval signature. It was impressive that the
deviation reports were present and were completed up front.

Two major findings resulted from this external technical systems audit. First, as a result of a delay
in the arrival of the protocol gases used in the verification test, Battelle was not able to run one of
the instruments through the test sequence prior to the first test. This undocumented deviation was
from Section 5.6 of the test/QA plan, Test Schedule, and stated "To avoid bias in testing of the
first analyzers through the sequence, Battelle's personnel will first conduct the entire test
sequence using an analyzer already on hand at Battelle. Testing will then continue with analyzers
named in Section 2.4." Second, the test/QA plan states that "The chemiluminescent monitors to
be used for Method 7E reference measurements will be subjected to a four-point calibration with
NO prior to the start of verification testing, on each measurement range to be used for
verification." The combustion source tests were started on January 13, 1999. No four-point
calibration with NO was recorded in the combustion source testing laboratory notebook prior to
January 13. This finding is also a finding in Battelle's internal audit conducted during the first
week of the verification test.

The impact of these two findings on the data presented in this report is as follows. Although
Battelle did not run an instrument through the entire test sequence prior to initiating testing, each
component of the test system was checked independently. Therefore, the absence of this pre-test
check will not impact the final data. The lack of initial calibration of the reference analyzer does
not affect any of the verification test data since initial tests were repeated after full calibrations
had been completed. There is no impact on the TSI test data from either of these factors.

22


-------
Chapter 5
Statistical Methods

5.1 Laboratory Tests

The monitor performance characteristics were quantified on the basis of statistical comparisons of
the test data. This process began by converting the spreadsheet files that resulted from the data
acquisition process into data files suitable for evaluation with Statistical Analysis System (SAS)
software. The following statistical procedures were used to make those comparisons.

5.1.1 Linearity

Linearity was assessed by linear regression with the calibration concentration as the independent
variable and the analyzer response as the dependent variable. Separate assessments were carried
out for each TSI monitor. The calibration model used was

where Yc is the monitor's response to a challenge concentration c, h(c) is a linear calibration
curve, and the error term was assumed to be normally distributed. (If the variability is not
constant throughout the range of concentrations, then weighting in the linear regression is
appropriate. It is often the case that the variability increases as the true concentration increases.)
The variability (oc) of the measured concentration values (c) was modeled by the following
relationship,

where a, k, and P are constants to be estimated from the data. After determining the relationship
between the mean and variability, appropriate weighting was determined as the reciprocal of the
variance.

Yc = h(c) + error

(1)

o2c = a + kc^

(2)

weight = w c =

(3)

23


-------
The form of the linear regression model fitted was h(c) ao a,c. In the concentration sub-
region where the linear calibration model provides a valid representation of the concentration-
response relation, concentration values were calculated from the estimated calibration curve using
the relation

, ,	Y - a

c = h \Yc) =	(4)

a j

A test for departure from linearity was carried out by comparing the residual mean square

1 6 ~

t£ (Yc - ao - aiC)2ncwc.	(5)

4 ,-=i '	' '

to an F-distribution with 6-2 = 4 numerator degrees of freedom.

Yci is the average of the nci analyzer responses at the ith calibration concentration, cr The
regression relation was fitted to the individual responses; however, only the deviation about the
sample mean analyzer responses at each calibration concentration provide information about
goodness-of-fit.

E E

j-i

Vc

i=l

aj

-a,

~a\c)2

w . =

E E

j-i

(Y.-Y )2

^ CI	Cl'

W

i=l

E (r,,

-a,

~a\c)7

n w

i = 1

The first summation on the right side of the equation provides information only about response
variability. The second summation provides all the information about goodness-of-fit to the
straight-line calibration model. This is the statistic that is used for the goodness-of-fit test.

5.1.2 Detection Limit

Limit of detection (LOD) is defined as the smallest true concentration at which a monitor's
expected response exceeds the calibration curve at zero concentration by three times the standard
deviation of the monitor's zero reading, i.e., a0 + 3 a0, if the linear relation is valid down to zero.
The LOD may then be determined by

LOD =

(a + 3o ) - a

^ o	o>	0

a,

3a

a,

(6)

24


-------
where a0 is the estimated standard deviation at zero concentration. The LOD is estimated as
lod = 3
-------
5.1.5 Interferences

Interference is reported as both the absolute response (in ppm) to an interferant level, and as the
sensitivity of the monitor to the interferant species, relative to its sensitivity to NO or N02. The
relative sensitivity is defined as the ratio of the observed N0/N02/N0x response of the monitor to
the actual concentration of the interferant. For example, a monitor that measures NO is challenged
with 500 ppm of CO, resulting in an absolute difference in reading of 1 ppm (as NO). The relative
sensitivity of the monitor is thus 1 ppm/500 ppm = 0.2 percent. The precision of the interference
results was not estimated from the data obtained, since only one measurement was made for each
interferant.

5.1.6	Pressure Sensitivity

At each of ambient pressure, reduced pressure (-10 inches of water), and increased pressure
(+10 inches of water), the sample flow rate, the response on zero gas, and the response on span
gas were measured for each monitor. Variability in zero and span responses for reduced and
increased pressures was assumed to be the same as the variability at ambient pressure. The
variability determined in the linearity test was used for this analysis. The duct pressure effects on
monitor flow rates and response were assessed by separate linear regression trend analyses for
flow rate and for response. The precision of the pressure effects on zero concentration response
and on span gas response was estimated based on the variability observed in the linearity test.
Statistical significance of the trends across duct pressures was determined by comparing the
estimated trends to their estimated standard errors, based on two-tailed t-tests:

t = p / (0.040825ct(c)) for the zero concentration test

t = p / (0.0707 1
-------
5.2 Combustion Source Tests
5.2.1 Accuracy

The relative accuracy (RA) of the monitors with respect to the reference method is expressed as:

_	v

I ,71 ^ d
\d\ + fn-1 	

Jn	(7)

RA = 	=	s— x 100%	v '

x

where d refers to the difference between the average of the two reference units and one of the
tested units and x corresponds to the average of the two reference unit values. Sd denotes the
sample standard deviation of the differences, based on n = 9 samples, while tan_i is the t value for
the 100(1 - a)th percentile of the distribution with n - 1 degrees of freedom. The relative accuracy
was determined for an a value of 0.025 (i.e., 97.5 percent confidence level, one-tailed). The RA
calculated in this way can be determined as an upper confidence bound for the relative bias of the
monitor |d|/x, where the bar indicates the average value of the differences or of the reference

values.

Assuming that the reference method variation is due only to the variation in the output source and
the true bias between the test and reference methods is close to zero, an approximate standard
error for RA is

SE *

S;

yfnx ^

0.3634 + t

n-1

2(«-l)

X 100%

(8)

5.2.2	Zero/Span Drift

Statistical procedures for assessing zero and span drift were similar to those used to assess
interrupted sampling. Zero (span) drift was calculated as the arithmetic difference between zero
(span) values obtained before and after sampling of each combustion source. The same calculation
was also made using zero and span values obtained before and after the linearity and ambient
temperature tests. No estimate was made of the precision of the zero and span drift values.

5.2.3	Measurement Stability

The temporal stability of monitor response in extended sampling from a combustion source was
assessed by means of a trend analysis on 60 minutes of data obtained continuously using the diesel
generator as the source. The existence of a difference in trend between the test unit and the
average of the reference units was assessed by fitting a linear regression line with the difference
between the measured concentration for a test unit and the average of the reference units as the
dependent variable, and time as the independent variable. Subtracting the average reference unit

27


-------
values adjusts for variation in the source output. The slope and the standard error of the slope are
reported. The null hypothesis that the slope of the trend line on the difference is zero was tested
using a one-sample two-tailed t-test with n - 2 = 58 degrees of freedom.

5.2.4	Inter- Unit Repeatability

The purpose of this comparison was to determine if any significant differences in performance
exist between two identical monitors operating side by side. In tests in which monitor per-
formance was verified by comparison with data from the reference method, the two identical units
of each type of monitor were compared to one another using matched pairs t-test comparisons. In
tests in which no reference method data were obtained (e.g., linearity test), the two units of each
monitor were compared using statistical tests of difference. For example, the slopes of the
calibration lines determined in the linearity test, and the detection limits determined from those
test data, were compared. Inter-unit repeatability was assessed for the linearity, detection limit,
accuracy, and measurement stability tests.

For the linearity test, the intercepts and slopes of the two units were compared to one another by
two-sample t-tests using the pooled standard error, with combined degrees of freedom the sum of
the individual degrees of freedom.

For the detection limit test, the detection limits of the two units were compared to one another by
two-sample t-tests using the pooled standard error with 10 degrees of freedom (the sum of the
individual degrees of freedom).

For the relative accuracy test, repeatability was assessed with a matched-pairs two-tailed t-test
with n - 1 = 8 degrees of freedom.

For the measurement stability test, the existence of differences in trends between the two units
was assessed by fitting a linear regression to the paired differences between the units. The null
hypothesis that the slope of the trend line on the paired differences is zero was tested using a
matched-pairs t-test with n - 2 = 58 degrees of freedom.

5.2.5	Data Completeness

Data completeness was calculated as the percentage of possible data recovered from a monitor in
a test; the ratio of the actual to the possible number of data points, converted to a percentage, i.e.,

Data Completeness = (Na)/(Np) x 100%,

where Na is the number of actual and Np the number of possible data points.

28


-------
Chapter 6
Statistical Results

6.1 Laboratory Tests
6.1.1 Linearity

Tables 6-la and b list the data obtained in the linearity tests for NO and N02, respectively.
Table 6-2 shows the results of the linear calibration curve fits for each unit, based on the data
shown in Tables 6-la and b.

Table 6-la. Data from NO Linearity Test of TSI COMBUCHECK Monitors



Actual NO

Unit A NO

Unit B NO

Reading

(PPm)

(PPm)

(PPm)

1

0.0

11

13

2

2002.0

2103

2063

3

189.2

200

243

4

493.4

505

578

5

0.0

15

33

6

1396.0

1542

1563

7

392.1

473

461

8

189.2

247

236

9

0.0

20

18

10

392.1

437

436

11

493.4

556

552

12

1399.0

1576

1555

13

0.0

28

25

14

2002.0

2259

2215

15

1399.0

1600

1573

16

493.4

585

572

17

0.0

32

28

18

388.4

454

444

19

189.2

228

220

20

2002.0

2287

2228

21

0

34

30

29


-------
Table 6-lb. Data from N02 Linearity Test of TSI COMBUCHECK Monitors

Actual N02	Unit A N02	Unit B N02

Number	(PPm)	(PPm)	(PPm)

1

0.0



0.2



0.2

2

100.1



135.4



135.4

3

10.0



15.7



16.6

4

40.1



52.7



51.9

5

0.0



0.2



0.5

6

70.1



88



86.8

7

20.0



26.2



26.8

8

10.0



13.9



14.4

9

0.0



0.7



1.6

10

20.0



24.4



23.9

11

40.1



47.4



45.9

12

70.1



84.5



82.8

13

0.0



1.3



2.4

14

100.1



121.4



119.2

15

70.1



85



84.6

16

40.1



49.9



49.7

17

0.0



1.3



2.6

18

20.0



23.7



23.5

19

10.0



13.2



13.7

20

100.1



116.7



113.8

21

0.0



1.6



2.9

Table 6-2. Statistical Results for Test of Linearity



Unit A



Unit B

Linear Regression

NO

no2



NO

no2

Intercept (ppm) (Std

22.292(3.599)

0.972 (0.236)

23.844(2.627)

1.854 (0.410)

Err)











Slope (Std Err)

1.101 (0.011)

1.218 (0.015)

1.100 (0.005)

1.184 (0.018)

R2

0.9981

0.9972



0.9996

0.9956

The results in Table 6-2 indicate that the TSI monitors responded approximately linearly to the
NO and N02 concentrations provided, but with slopes that differed substantially from 1. The R2
values for all four monitors exceed 0.99, but the slopes are 1.1 for both NO monitors and about
1.2 for the N02 monitors. This observation appears to result from error or instability in the
calibration of the TSI units. As noted in Section 3.2.1, the TSI NO and N02 monitors were

30


-------
provided before the linearity test with 200 ppm NO and 25 ppm N02, respectively, and their
responses were calibrated to those standards. Shortly thereafter, when NO and N02
concentrations were provided over the full nominal ranges of the monitors, the responses obtained
were consistently high. This is clearly evident in Tables 6-la and b, and also was apparent in the
pre- and post-test span gas responses. The 2,000 ppm NO span gas produced responses of 2,099
and 2,023 ppm on Units A and B, respectively, before the linearity test, and 2,242 and 2,182 ppm
after the test. Similarly, the 100 ppm N02 span gas produced readings of 123.4 and 122.7 ppm
before the linearity test, and 113 and 109.5 ppm after the test. Thus, the linearity test results
shown in Tables 6-la and b and 6-2 indicate that the TSI monitors may provide linear response
over their full nominal ranges for NO and N02, but that instability in calibration may be a concern.

Table 6-la also indicates that the readings of the TSI NO monitor on zero gas increased
considerably during the linearity test, to 30 ppm or more. This observation clearly indicates a
"memory" effect, in which the response of the monitor remains elevated following exposure to
high NO levels. In practice, TSI recommends that the user re-zero the monitor after a high NO
exposure, or wait a sufficient time (an hour or so) for the monitor to return to baseline while
sampling clean, dry air. Neither of those approaches was feasible in this test.

Although the TSI operator was allowed to judge when the TSI monitors had stabilized on zero
gas, and to record readings at that time, Table 6-la indicates that much longer stabilization times
would have been needed for readings to return to baseline. The conclusion is that measurement of
low NO concentrations should not be attempted following high NO exposures, unless substantial
restabilization times are feasible.

6.1.2 Detection Limit

Table 6-3 shows the estimated detection limits for each test unit and each analyte, determined
from the data obtained in the linearity test.

Table 6-3. Estimated Detection Limits for TSI COMBUCHECK Monitors



Unit A



Unit B



NO

no2

NO

no2

Estimated Detection Limit (ppm)

25.661

1.490

20.774

2.877

(Standard Error) (ppm)

(8.119)

(0.472)

(6.570)

(0.911)

Table 6-3 shows that calculated detection limits were about 1.5 and 2.9 ppm for the two N02
monitors, but about 26 and 21 ppm for the two NO analyzers. The higher detection limits for NO
result directly from the much greater variability of the NO readings on zero air in the linearity test
(Table 6-la), and are probably not a good indication of the NO detection limits achievable with
the COMBUCHECK monitors. Zero readings from the two NO analyzers increased consistently
over the duration of the linearity test, from about 12 ppm to over 30 ppm. This increase is thought

31


-------
to be due to a memory effect, i.e., the NO sensors do not fully return to zero after sampling of
relatively high NO levels. Thus, the higher detection limits for NO are probably due at least in part
to the much greater calibration range used in the linearity test for NO, relative to the range used
for N02. Clearly, when measurements are being made over the full NO range of the monitors,
care must be taken that the baseline is fully stabilized before making measurements in the low end
of the range. When fully stabilized to a near-zero baseline, the detection limits for the NO and
N02 monitors should be comparable to the measurement resolution of the monitors, i.e., about 1
ppm for NO and 0.1 to 0.2 ppm for N02.

6.1.3 Response Time

Table 6-4 lists the data obtained in the response time test of the TSI COMBUCHECK Monitors.
Table 6-5 shows the response times of the monitors to a step change in analyte concentration,
based on the data shown in Table 6-4.

Table 6-5 shows that the TSI COMBUCHECK monitors provide substantially faster responses
for NO than for N02, with some difference in time response between different monitors.

The Unit A NO monitor showed a response time of 65 seconds, whereas Unit B responded in less
than half that time. For N02, the response times of both monitors were 2 minutes or more, and
were closely similar. All the response times in Table 6-5 are within the 4-minute time response
stated in the SCAQMD test protocol,(8) and also agree well with the 90 percent response times
stated in the COMBUCHECK specifications.

32


-------
Table 6-4. Response Time Data for TSI COMBUCHECK Monitors



Unit A NO

Unit A N02

Unit B NO

Unit B N02

Time (sec)

(ppm)

(ppm)

(ppm)

(ppm)

0

14

0.2

33

0.5

10

88

57.1

1252

81.9

20

1358

110.0

1431

99.9

30

1398

81.3

1499

78.6

40

1415

81.6

1508

78.1

50

1423

82.0

1514

78.5

60

1434

82.3

1514

78.8

70

1444

82.4

1518

79.2

80

1452

82.3

1520

79.2

90

1454

82.3

1524

79.2

100

1459

82.3

1526

79.2

110

1463

82.4

1526

79.7

120

1470

83.0

1528

80.1

130

1470

83.4

1531

80.7

140

1476

83.8

1532

81.1

150

1481

84.3

1534

81.8

160

1482

84.5

1536

82.1

170

1482

85.1

1538

82.6

180

1489

85.5

1538

83.0

190

1490

86.1

1529

83.5

200

1493

86.3

1544

83.2

210

1495

86.4

1542

84.0

220

1498

86.8

1544

84.4

230

1503

86.8

1544

84.6

240

1504

87.0

1544

85.0

250

1505

87.0

1545

85.0

260

1505

87.3

1545

85.2

270

1505

87.3

1545

85.2

280

1509

87.6

1548

85.2

290

1510

87.6

1548

85.3

300

1514

87.6

1547

85.5

33


-------
Table 6-5. Response Time Results for TSI COMBUCHECK Monitors





Unit A



Unit B



NO

no2

NO

no2

Response Time* (sec)

65

126

26

142

* The analyzer's responses were recorded at 10-second intervals; therefore the point in time when the
95 percent response was achieved was determined by interpolating between recorded times to the nearest
second.

6.1.4 Interrupted Sampling

Table 6-6 shows the zero and span data resulting from the interrupted sampling test, and
Table 6-7 shows the differences (pre- minus post-) of the zero and span values. Span
concentrations of 2,000 ppm NO and 100 ppm N02 were used for this test. Zero differences were
less than 3 ppm for the two N02 monitors, and 20 to 30 ppm for the two NO monitors. For all
four monitors, the change in zero readings was downward from pre- to post-shutdown, and in
fact zero readings on all four monitors were essentially zero after the shutdown. This observation
shows that the zero readings pre-shutdown were elevated due to the NO and N02 exposures
during the linearity test, and that the monitors returned to zero after the several hours of
shutdown, i.e., the elevation of zero gas readings was reversible. The span differences in
Table 6-7 are in opposite directions for the NO and N02 analyzers. NO span readings decreased
by about 6 percent of the 2,000 ppm span value, whereas N02 span readings increased by 7.9 and
12.2 percent of the 100 ppm span value.

Table 6-6. Data from Interrupted Sampling Test with TSI COMBUCHECK Monitors



Unit A NO

Unit A N02

Unit B NO

Unit B N02

Pre-Shutdown Date:

1/15/99

Time:

17:00



Pre-Shutdown Zero (ppm):

22

1.6

30

2.9

Pre-Shutdown Span (ppm):

2242

113

2182

109.5

Post-Shutdown Date:

1/16/99

Time:

09:25



Post-Shutdown Zero (ppm):

0.5

0

0.5

0

Post-Shutdown Span (ppm):

2118

120.9

2054

121.7

Table 6-7. Pre- to Post-Test Differences as a Result of Interruption of Operation of TSI

COMBUCHECK Monitors













Unit A

Unit B

Pre-Shutdown—Post-Shutdown

NO

no2

NO N02

Zero Difference (ppm)



21.5

1.6

29.5 2.9

Span Difference (ppm)



124

-7.9

128 -12.2

34


-------
6.1.5 Interferences

Table 6-8 lists the data obtained in the interference tests. Table 6-9 summarizes the sensitivity of
the monitors to interferant species, based on the data from Table 6-8. Table 6-9 shows both the
absolute differences observed for each interferant relative to the preceding zero reading, and the
resulting relative sensitivity to each interferant, in percent of the NO sensitivity. For the S02/N0
mixture, the entries indicate the impact of the S02 on the response to NO. Thus, the ppm values
show the difference between NO response and the expected 383 ppm value, and the relative
sensitivity is based on the 451 ppm S02 concentration.

Table 6-8. Data from Interference Tests on TSI COMBUCHECK Monitors

Interferant

Interferant, Cone.



Response (ppm equivalent)



Gas

(ppm)

Unit A NO

Unit A N02

Unit B NO

Unit B NO,

Zero



3

0.3

6

0.9

CO

496 ppm

5

0.6

26

1

Zero



2

0.4

3

0.9

o
o

N)

5.03%

2

0.4

3

0.9

Zero



2

0.2

2

0.4

nh3

494 ppm

2

0.2

2

0.5

Zero



2

0.4

2

0.7

HCs

590 ppm

2

0.2

3

0.5

Zero



2

0.2

2

0.2

S02

501 ppm

3

-1.0

2

-2.2

Zero



2.5

0

3.4

0

SO, + NO

451+383 ppm

395

2.5

409

3.4

Table 6-9. Results of Interference Tests of TSI COMBUCHECK Monitors

Unit A Response ppm	Unit B Response ppm

(% relative sensitivity)	(% relative sensitivity)

Interferant	NO	N02	NO	N02

CO (496 ppm)

2 (0.49%)

0.3 (<0.1%)

20 (4%)

0.1 (<0.1%)

C02 (5.03%)

0

0

0

0

NH3 (494 ppm)

0

0

0

0.1 (<0.1%)

HCs (590 ppm)

0

-0.2 (<

1 (0.2%)

-0.2 (< 0.1%)





0.1%)





S02 (501 ppm)

1 (0.2%)

-1.2(0.2%)

0

-2.4 (0.5%)

S02 (451 ppm) +

9.5

2.5

22.6

3.4

NO (383 ppm)

(2.1%)

(0.6%)

(5.0%)

(0.8%)

35


-------
Table 6-9 shows that, with one exception, none of the individual interferants caused a response
greater than about +/- 2 ppm on any of the monitors. Thus, interference from these gases is neg-
ligible. The one exception is that on NO Unit B, CO at 496 ppm caused an increased response
above zero of 20 ppm. If real, this result implies an interference from CO equal to about 4 percent
of the sensitivity to an equal concentration of NO.

The S02/N0 mixture in Table 6-9 resulted in responses slightly higher (by 9.5 and 22.6 ppm,
respectively) than those expected for the 383 ppm NO level. These results indicate an enhance-
ment of the NO response due to the S02 present; at the levels tested the effect of the S02 equals 2
to 5 percent of the response to an equal concentration of NO. These percentages are comparable
to the span drift observed with the TSI NO monitors (see Section 6.1.8), and do not strongly
indicate an interference from S02 in the presence of NO.

The S02/N0 mixture also produced a slight increase in response (2.5 and 3.4 ppm, respectively)
on the two TSI N02 monitors. It is not known whether this represents a cross-sensitivity of the
N02 sensors to NO, or an effect of the S02 present. S02 alone produced a very slight reduction in
response of the two N02 monitors (Table 6-9). If it is assumed that only the NO in the S02/N0
mixture effects the N02 monitors, then the result places an upper bound on the cross-sensitivity to
NO of the N02 monitors. That is, the NO sensitivity of Units A and B of the TSI N02 monitors
cannot exceed 2.5/383 = 0.65 percent and 3.4/383 = 0.89 percent, respectively.

6.1.6 Pressure Sensitivity

Table 6-10 lists the data obtained in the pressure sensitivity test. Table 6-11 summarizes the find-
ings from those data in terms of the ppm differences in zero and span readings at the different
duct gas pressures, and the ccm differences in monitor flow rates at the different duct gas
pressures.

No significant effect of duct pressure on zero and span values was seen with any of the TSI
COMBUCHECK monitors. This conclusion is based on using the response variability determined
in the linearity test. Zero readings of the N02 monitors changed by less than 1 ppm, whereas those
of the NO monitors changed by about 8 to 20 ppm. Table 6-10 clearly shows that the zero
readings of the NO monitors were elevated by several ppm at any pressure, as a result of exposure
to the 1,400 ppm NO span gas used in this test. The span differences shown in Table 6-11 for the
N02 monitors are equivalent to about 7 to 14 percent of the 70 ppm N02 span gas concentration.
The span differences in Table 6-11 for the Unit A NO monitor are equivalent to about 1 percent
or less of the 1,400 ppm NO span gas used. Those for the Unit B NO monitor are 3.5 to 4 percent
of that span concentration.

Tables 6-10 and 6-11 also indicate that the sample gas flow rates drawn by the TSI monitors are
highly dependent on the duct pressure, and that this effect is highly different from one monitor to
another. Sample flow rates at +10 inches of water exceeded those at ambient pressure by 34 to
84 percent; flow rates at -10 inches of water were reduced by 14 to 47 percent. This behavior
occurs because no effort is made to regulate the sample flow of the COMBUCHECK monitors,
since the response of the NO and N02 sensors is essentially insensitive to flow rate. This lack of
flow rate dependence is confirmed by the data in Tables 6-10 and 6-11 and the discussion above.

36


-------
Table 6-10. Data from Pressure Sensitivity Test for TSI COMBUCHECK Monitors

Pressure	Unit A NO Unit A N02 Unit B NO Unit B N02

Ambient Flow rate (ccm)

658



594



638

620

Zero (ppm)

0







0



NO span (ppm)

1465







1437



Zero (ppm)

11



0.4



14

0.4

N02 span (ppm)





86.3





81.7

Zero (ppm)





1





1.1

+10 in. H20 Flow rate (ccm)

1030



1090



852

1087

Zero (ppm)

9







13



NO span (ppm)

1470







1486



Zero (ppm)

19



0.5



30

0.8

N02 span (ppm)





94.5





91.6

Zero (ppm)





1.1





1.5

-10 in. H20 Flow rate (ccm)

560



418



551

328

Zero (ppm)

16







25



NO span (ppm)

1449







1493



Zero (ppm)

29



0.9



29

1.1

N02 span (ppm)





92.8





86.8

Zero (ppm)





1.2





1.5

Table 6-11. Pressure Sensitivity Results for TSI COMBUCHECK Monitors







Unit A



Unit B





NO



no2

NO

no2

Zero High-Ambient (ppm diff*)



8.

5

0.1

14.5

0.4

Low-Ambient (ppm diff)



17



0.35

20

0.55

Significant Pressure Effect



N



N

N

N

Span High-Ambient (ppm diff)



5



8.2

49

9.9

Low-Ambient (ppm diff)



-16



6.5

56

5.1

Significant Pressure Effect



N



N

N

N

Flow High-Ambient (ccm diff*)



372



496

214

467

Rate Low-Ambient (ccm diff)



-98



-176

-87

-292

* ppm or ccm difference between high/low and ambient pressures. The differences were calculated based on
the average of the zero check responses.

37


-------
6.1.7 Ambient Temperature

Table 6-12 lists the data obtained in the ambient temperature test with the TSI COMBUCHECK
monitors.

Table 6-12. Data from Ambient Temperature Test of TSI COMBUCHECK Monitors

Unit A NO Unit A NO, Unit B NO Unit B NO,
Condition	(PPm) (PPm)	(PPm)	(PPm)

(Room Temp.)

Temp.	22.8°C (73°F)

Zero	7	6

NO span	1476	1487

Zero	1.2	1.4

N02span	87.2	81.7

(Heated)

Temp. 40°C (104°F)

Zero	10	24

NO span	1459	1467

Zero	27	1.2	39	1.5

N02 span	92.8	86.7

(Cooled)

Temp.	6.7°C (44°F)

Zero	7	6

NO span	1483	1442

Zero	16	2.9	12	3.5

N02 span	85.6	80.9

(Room Temp.)

Temp. 22.2°C (72°F)

Zero	4	6

NO span	1455	1441

Zero	1.7	1.6

	NO, span	93.4	85	

Table 6-13 summarizes the sensitivity of the monitors to changes in ambient temperature. This
table is based on the data shown in Table 6-12, where the span values are 1,400 ppm for NO and
70 ppm for N02.

38


-------
Table 6-13. Ambient Temperature Effects on TSI COMBUCHECK Monitors





Unit A



Unit B







NO

no2

NO

no2

Zero

Heat - room (ppm diff*)

4.5

-0.25

18

0



Cool - room (ppm diff)

1.5

1.45

0

2.0



Significant Temp. Effect

N

N

N

N

Span

Heat - room (ppm diff)

-6.5

2.5

3

3.35



Cool - room (ppm diff)

17.5

-4.7

-22

-2.45



Significant Temp. Effect

N

N

N

N

* ppm difference between heated/cooled and room temperatures. The differences were calculated using the
average of the pre-span zero responses recorded at room temperature.

Table 6-13 shows that no significant dependence on ambient temperature was found for the TSI
monitors. Changes in zero readings with temperature were a few ppm, except that the TSI Unit B
NO monitor showed a zero drift of 18 ppm in moving from room temperature to the heated
chamber. Span values showed some variation during the test, but no consistent pattern with
temperature. The span differences shown in Table 6-13 for the NO monitors amount to 0.2 to 1.6
percent of the 1,400 ppm span concentration. The span differences shown for the N02 monitors
amount to 3.5 to 6.7 percent of the 70 ppm N02 span.

6.1.8 Zero and Span Drift

Zero and span drift was evaluated from data taken at the start and end of the linearity and ambient
temperature tests. Those data are shown in Table 6-14, and the drift values observed are shown as
pre- minus post-test differences in ppm in Table 6-15. The NO and N02 span concentrations used
in the linearity test were 2,000 ppm and 100 ppm, respectively; those in the temperature test were
1,400 ppm and 70 ppm, respectively. Table 6-15 shows that for both the NO and N02 monitors,
zero and span drift were both much larger as a result of the linearity test than as a result of the
temperature test. This may be due to the higher NO and N02 concentrations and longer duration
of sampling in the linearity test. Note that the manufacturer recommends allowing the instrument
baseline time to return fully to its initial value after prolonged exposure (i.e., >3 to 5 minutes) to
high concentrations of analyte. However, N02 monitor zero levels dropped during the linearity
test, whereas those of the two NO monitors rose 20 and 30 ppm, respectively. Zero readings
changed by 3 ppm or less on all four TSI monitors as a result of the ambient temperature test.

The span drift observed for the NO monitors in the linearity test (Table 6-15) is equivalent to 7 to
8 percent of the 2,000 ppm NO span value. The corresponding span drift for the N02 monitors is
equivalent to 10 to 13 percent of the 100 ppm N02 span value.

39


-------
Table 6-14. Data from Linearity and Ambient Temperature Tests Used to Assess Zero and
Span Drift of the TSI COMBUCHECK Monitors





Unit A NO

Unit A N02

Unit B NO

Unit B N02

Test



(ppm)

(ppm)

(ppm)

(ppm)

Linearity

Pre-Test Zero

2

10.4

0

10.4



Pre-Test Span

2099

123.4

2023

122.7



Post-Test Zero

22

1.6

30

2.9



Post-Test Span

2242

113

2182

109.5

Ambient Temperature

Pre-Test Zero

7

1.2

6

1.4



Pre-Test Span

1476

87.2

1487

81.7



Post-Test Zero

4

1.7

6

1.6



Post-Test Span

1455

93.4

1441

85

Table 6-15. Zero and Span Drift Results for the TSI COMBUCHECK Monitors





Unit A

Unit B





NO

no2

NO

no2

Pre- and Post-Differences



(ppm)

(ppm)

(ppm)

(ppm)

Linearity Test

Zero

-20

8.8

-30

7.5



Span

-143

10.4

-159

13.2

Ambient Temperature Test

Zero*

3

-0.45

0

-0.2



Span

21

-6.2

46

-3.3

* Drift is the difference (pre-monitoring minus post-monitoring) between the first and last zero
response at room temperature.

The span drift observed for the NO monitors in the ambient temperature test (Table 6-15) is
equivalent to 1.5 to 3.3 percent of the 1,400 ppm NO span value. The corresponding span drift
for the N02 monitors is equivalent to 4.7 to 8.9 percent of the 70 ppm N02 span value.

6.2 Combustion Source Tests

The following sections describe the results of the combustion source tests with the TSI
COMBUCHECK monitors. After these tests were performed, TSI staff reported an error in
operating their N02 monitors in that the in-line water traps used in the sample intake lines had
been mislabeled and, consequently, were installed backward. This error may have had a severe
effect on TSI's N02 measurements and, in turn, on NOx levels calculated from the N02 data. The

40


-------
potential impact of this error on TSI's source testing results is noted in the following sections, but
cannot be quantified or corrected.

6.2.1 Relative Accuracy

Tables 6-16a through c list the measured NO, N02, and NOx data obtained in sampling of the four
combustion sources. Note that the TSI monitors measure either NO or N02, and the indicated
NOx totals are the sum of those data from the respective pairs of monitors designated A or B; in
contrast, the reference monitors measure NO and NOx, and N02 is determined by difference.
Tables 6-16a through c show that a wide range of NO and N02 concentrations was emitted by the
three sources.

Table 6-17 displays the relative accuracy (in percent) of the TSI monitors for NO, N02, and NOx
for each of the sources. Estimated standard errors are shown with the relative accuracy estimates.
These standard error estimates were calculated under the assumption of zero true bias between
the reference and test methods. If the bias is in fact non-zero the standard errors underestimate the
variability. In considering the relative accuracy results, it should be noted that the
COMBUCHECK specifications indicate accuracy for NO of 5 ppm or 5 percent of reading,
whichever is greater, and for N02 of 3 ppm or 5 percent of reading, whichever is greater.

At the request of the TSI representative, the TSI NO and N02 monitors were calibrated with
200 ppm NO and 25 ppm N02, respectively, prepared using the EPA Protocol Gases and
Environics dilution system, before starting the source sampling portion of the verification test.
The monitors were adjusted to match those respective standard concentrations. The span gas
concentrations listed in Table 3-3 were then provided to the TSI monitors before and after
sampling of each respective combustion source, but no adjustment of the TSI monitors was made.
As will be discussed in Section 6.2.2, the TSI readings observed in sampling of those span gases
do not show close quantitative agreement with the span values, indicating that the calibration of
the monitors was not stable after adjustment to the 200 ppm NO and 25 ppm N02 calibration
gases.

One possible explanation for this is the finding, after these tests were completed, that the in-line
water traps used with the TSI N02 monitors were labelled incorrectly, and consequently were
installed backward in the sample lines. In this backward configuration, the traps would collect
particles and not allow condensed water to reach the reservoir. The collected particles and water
might have been an effective trap for N02, contributing to the variability observed in N02 mea-
surements. The general effect of operating with the water trap reversed has been demonstrated by
TSI in the laboratory, but it is impossible to determine the quantitative effect on the test results. In
most of the combustion tests, the TSI N02 monitors read high relative to the reference analyzers,
whereas in TSI's tests reversal of the water trap was found to first increase, then sharply decrease,
the N02 response.

41


-------
Table 6-16a. Data from the Gas Rangetop in Verification Testing of TSI COMBUCHECK Monitors







TSI Monitor Data









Reference Analyzer Data







Unit A NO

Unit A N02

Unit A NOs

Unit B NO

Unit B N02

Unit B NOx

14ANO

14A N02

14A NOs

10 NO

10 no2

10 NOx



(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

1

5

3.2

8.2

6

6.3

12.3

6.1

2.2

8.3

5.8

1.9

7.7

2

7

3.7

10.7

5

6.9

11.9

6.3

2.4

8.7

6.0

2.2

8.2

3

4

4.7

8.7

5

7.7

12.7

6.4

2.5

8.9

6.0

2.3

8.3

4

4

4.7

8.7

5

7.1

12.1

6.5

2.4

00
00

6.2

2.1

8.3

5

4

5.3

9.3

5

8.5

13.5

6.5

2.4

8.9

6.2

2.1

8.3

6

4

5.7

9.7

5

8.4

13.4

6.4

2.5

8.9

6.1

2.4

8.5

7

4

6

10

5

9.2

14.2

6.5

2.5

9.0

6.2

2.3

8.5

8

5

6.5

11.5

5

9.4

14.4

6.5

2.4

8.9

6.2

2.2

8.4

9

4

5.9

9.9

5

8.7

13.7

6.5

2.3

00
00

6.2

2.2

8.4

NJ

Table 6-16b. Data from Gas Water Heater in Verification Testing of TSI COMBUCHECK Monitors







TSI Monitor Data









Reference Analyzer Data







Unit A NO

Unit A N02

Unit A NOs

Unit B NO

Unit B N02

Unit B NOx

14ANO

14A NO,

14A NOs

10 NO

10 no2

10 NOx



(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

1

85

16.7

101.7

89

13.6

102.6

122.9

2.5

125.4

112.5

7.4

119.9

2

86

16.1

102.1

89

14.8

103.8

120.9

4.6

125.5

112.5

7.9

120.4

3

88

15.3

103.3

91

16.8

107.8

118.9

6.7

125.6

112.2

7.5

119.7

4

79

14.9

93.9

84

18.8

102.8

103.2

15.4

118.6

96.6

14.8

111.4

5

67

13.2

80.2

72

23.3

95.3

93.0

7.2

100.2

87.2

7.8

95.0

6

57

13.8

70.8

64

15.4

79.4

83.8

5.5

89.3

78.4

6.9

85.4

7

61

12.8

73.8

63

17.1

80.1

82.5

5.0

87.5

77.7

6.3

84.0

8

58

12.7

70.7

62

17.4

79.4

81.4

6.6

87.9

76.7

7.0

83.8

9

57

12.7

69.7

63

18.3

81.3

81.3

5.8

87.1

76.3

7.2

83.5


-------
Table 6-16c. Data from the Diesel Generator at High RPM in Verification Testing of TSI COMBUCHECK Monitors



Unit A NO

(ppm)

Unit A N02
(ppm)

TSI Monitor Data
Unit A NOs Unit B NO
(ppm) (ppm)

Unit B N02
(ppm)

Unit B NOx
(ppm)

14ANO
(ppm)

14A N02
(ppm)

Reference Analyzer Data
14A NOs 10 NO
(ppm) (ppm)

iono2

(ppm)

10 NOx
(ppm)

1

84

150.9

234.9

100

256.3

356.3

87.9

65.8

153.7

91.7

67.1

158.8

2

88

73

161

101

95.7

196.7

80.6

66.0

146.6

86.7

69.3

156.0

3

86

44.1

130.1

94

41

135

83.3

65.5

148.8

90.7

68.2

158.9

4

77

20.7

97.7

63

41.2

104.2

78.6

62.7

141.3

86.7

67.1

153.8

5

74

146.5

220.5

69

216.7

285.7

78.8

62.2

141.0

84.7

68.2

152.9

6

72

144.9

216.9

73

209

282

77.5

62.5

140.0

82.7

64.8

147.5

7

76

145.1

221.1

73

203.6

276.6

80.3

61.9

142.2

85.7

66.0

151.6

8

76

95.2

171.2

71

144.1

215.1

77.5

61.1

138.6

82.7

64.8

147.5

9

80

57.3

137.3

75

69.4

144.4

78.4

61.6

140.0

82.7

66.0

148.6


-------
Table 6-17. Relative Accuracy of TSI COMBUCHECK Monitors





Unit A





Unit B





NO

no2

NOx

NO

no2

NOx

Source

(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

(ppm)

Gas Rangetop

40.43*

156.62

21.31

23.85

284.68

60.94

(6 ppm NO, 2 ppm

(4.79)**

(12.49)

(3.09)

(2.00)

(12.23)

(2.61)

N02)***













Gas Water Heater

29.52

128.62

19.53

25.39

169.13

14.90

(90 ppm NO, 7 ppm N02)

(1.37)

(12.50)

(0.85)

(1.52)

(12.21)

(1.77)

Diesel Generator-High

8.56

110.50

44.60

15.11

216.23

93.02

RPM

(1.38)

(21.91)

(9.17)

(4.04)

(35.48)

(15.69)

(80 ppm NO, 65 ppm NO?)	

*Relative accuracy, percent relative to mean of two reference analyzers.

** Standard error of the relative accuracy value.

***Approximate NO and N02 levels from each source are shown; see Tables 6-16a through c.

Consistent with the hypothesis that the improperly functioning water traps affected the validity of
the N02 measurements, Table 6-17 shows that the relative accuracy of the TSI monitors in sampl-
ing the gas rangetop emissions was better for NO than for N02. For NO, TSI Unit A read about
2 ppm low, and Unit B over 1 ppm low, relative to the reference analyzers (Table 6-16a). These
results are within the 5 ppm accuracy specification for NO noted above. At the low NO levels
present, these readings result in relative accuracies of about 40 percent and 24 percent,
respectively. For N02, both TSI units read considerably higher than the 2 to 2.5 ppm levels
recorded by the reference analyzers, and the two TSI units did not agree closely with one another.
Although usually within the 3 ppm N02 accuracy specification, relative accuracies of over
150 percent and nearly 300 percent resulted for N02 Units A and B, respectively. Relative
accuracies of 21 and 61 percent for NOx resulted from the combined relative accuracies of the
pairs of units designated A and B, respectively.

Tables 6-17 and 6-16b show a similar pattern in sampling of the gas water heater. The two TSI
NO monitors read consistently low relative to the reference analyzers (Table 6-16b), resulting in
relative accuracies of about 30 and 25 percent, respectively (Table 6-17). The two TSI N02
monitors read consistently high relative to the reference analyzers, resulting in relative accuracies
of about 130 and 170 percent for N02 Units A and B. Neither the NO nor the N02 monitors
showed accuracies consistent with their specifications, which were noted above. The overall
relative accuracies for NOx were about 20 and 15 percent, respectively.

In sampling of the highest NOx levels, from the diesel generator (Table 6-16c), the TSI monitors
showed good relative accuracy for NO. As Table 6-17 shows, relative accuracy for the NO
Units A and B was 8.6 and 15.1 percent, respectively. Both these values are well within the
20 percent relative accuracy value stated in the SCAQMD protocol.(8) However, for N02, the TSI
monitors exhibited readings generally much higher than those from the reference analyzers, and
also much more variable (Table 6-16c). The two TSI N02 analyzers also did not agree closely
with one another in this test. As a result, relative accuracy for the two N02 analyzers was about

44


-------
110 and 215 percent, respectively. These N02 results are thought to be caused by improper N02
sampling, due to the reversal of the water traps used with the monitors.

6.2.2 Zero and Span Drift

Table 6-18 shows all the data used to evaluate zero and span drift of the TSI COMBUCHECK
monitors from the combustion source tests. These values are from the zero and span gases
provided to the TSI monitors before and after sampling of the indicated combustion source. As
noted above, prior to the combustion source sampling the TSI monitors were calibrated with 200
ppm NO and 25 ppm N02. Thus, the zero and span gas results shown in Table 6-18 were
obtained without adjustment of the monitors.

Table 6-18. Data Used to Assess Zero and Span Drift for TSI COMBUCHECK Monitors
on Combustion Sources





Unit A NO

Unit A N02

Unit B NO

Unit B N02

Source



(ppm)

(ppm)

(ppm)

(ppm)

Gas Rangetop*

Pre-Test Zero

0

0

0

-0.3



Pre-Test Span

10

12.4

10

12.6



Post-Test Zero

0

0

0

0



Post-Test Span

15

12.7

12

12.9

Gas Water Heater**

Pre-Test Zero

0

0

0

0



Pre-Test Span

95

18.4

96

18.5



Post-Test Zero

0

0.2

0

1.2



Post-Test Span

90

22.9

92

23.7

Diesel-High RPM***

Pre-Test Zero

-3

0

0

0



Pre-Test Span

187

74.8

189

80



Post-Test Zero

0

0.4

3

0.7



Post-Test Span

176

67.2

171

79.1

* Span values 20 ppm NO and 10 ppm N02.
** Span values 100 ppm NO and 15 ppmN02.
*** Span values 200 ppm NO and 50 ppm N02.

Table 6-19 summarizes the zero and span drift results from the combustion source tests. This
table shows that zero drift for the N02 monitors was always within 1.2 ppm over the combustion
source sampling, and that zero drift for the NO monitors was always within 3 ppm. Span drift for
the N02 monitors was usually smaller in an absolute sense (i.e., in ppm) than that of the NO
monitors, perhaps because of the lower span gas concentrations for N02. With the gas rangetop
the N02 span drift values are equivalent to 3 percent of the 10 ppm N02 span level; the NO span
drift values equal 10 and 25 percent, respectively, of the 20 ppm NO span level.

45


-------
Table 6-19. Results of Zero and Span Drift Evaluation for TSI COMBUCHECK Monitors





Unit A

Unit B

Pre-Test—



NO

no2

NO

no2

Post-Test



(ppm)

(ppm)

(ppm)

(ppm)

Gas Rangetop*

Zero

0

0

0

0



Span

-5

-0.3

-2

-0.3

Gas Water Heater**

Zero

0

-0.2

0

-1.2



Span

5

-4.5

4

-5.2

Diesel Generator-High RPM***

Zero

-3

-0.4

-3

-0.7



Span

11

7.6

18

0.9

* Span values 20 ppm NO and 10 ppm N02.
** Span values 100 ppm NO and 15 ppmN02.
*** Span values 200 ppm NO and 50 ppm N02.

With the gas water heater, N02 span drift equaled 30 to 35 percent of the 15 ppm span value, and
NO span drift equaled 4 to 5 percent of the 100 ppm NO span value. With the diesel generator,
N02 span drift was 1.8 to 15 percent of the 50 ppm span value, and NO span drift was 5.5 to
9 percent of the 200 ppm span value.

Note that most of the span readings shown in Table 6-18 do not agree closely with the known
span values provided. This behavior parallels that found in the source test results discussed under
Relative Accuracy (Section 6.2.1). For example, the TSI responses to span gases in the gas range-
top testing are low for NO and high for N02 (Table 6-18). This observation suggests that calibra-
tion drift in the TSI monitors, following adjustment to the 200 ppm NO and 25 ppm N02 calibra-
tion gases, was a factor in the results obtained during the combustion source tests. As noted
above, improper installation of the water traps was likely one contributor to span drift for N02.

6.2.3 Measurement Stability

Table 6-20 shows the data obtained in the extended sampling test, in which the TSI
COMBUCHECK and reference monitors sampled diesel emissions for a full hour without
interruption or sampling of ambient air. It must be stressed that the COMBUCHECK monitors
are not designed or intended to sample a combustion source for more than a few minutes at a
time. Thus the extended sampling test was not appropriate for these monitors. Nevertheless, the
TSI operator agreed to participate, and the TSI data were compared to the average of the
reference analyzer data to assess whether a different trend in the emission data was observed for
the TSI monitors relative to the reference analyzers. Table 6-21 shows the results of this
evaluation, in terms of the slopes and standard errors of the NO, N02, and NOx data with time.
Also shown in Table 6-21 is an indication of whether the slopes indicated by the TSI monitors
differed from those observed by the reference analyzers.

46


-------
Table 6-20. Data from Extended Sampling Test with Diesel Generator at Idle, Using TSI COMBUCHECK Monitors



Unit A NO

(ppm)

Unit A N02
(ppm)

TSI Analyzer Data
Unit A NOx Unit B NO
(ppm) (ppm)

Unit B N02
(ppm)

Unit B NOx
(ppm)

14ANO
(ppm)

14A N02
(ppm)

Reference Analyzer Data
14A NOs 10 NO
(ppm) (ppm)

10 no2

(ppm)

10 NOx
(ppm)

1

80

57.3

137.3

75

69.4

144.4

79.0

62.0

141.0

84.7

69.3

154.0

2

85

36.7

121.7

78

45.4

123.4

75.2

65.8

141.1

85.7

63.7

149.4

3

83

25.7

108.7

78

35.5

113.5

76.6

60.4

137.0

81.6

66.0

147.6

4

87

21.5

108.5

83

27.7

110.7

79.1

61.8

140.9

83.7

67.1

150.7

5

83

16.3

99.3

81

25.2

106.2

74.5

63.8

138.2

80.6

68.2

148.8

6

88

15

103

87

21.8

108.8

78.9

62.6

141.5

83.7

68.2

151.8

7

86

7.5

93.5

86

19

105

76.2

61.4

137.6

81.6

68.2

149.8

8

88

5.2

93.2

89

17.2

106.2

75.6

66.5

142.1

79.6

70.4

150.0

9

87

4.4

91.4

89

16.1

105.1

78.0

63.0

141.0

83.7

64.8

148.5

10

87

6.9

93.9

88

14.8

102.8

76.1

61.4

137.5

82.7

62.6

145.3

11

88

58.7

146.7

90

16

106

78.6

63.5

142.0

83.7

64.8

148.5

12

86

9.1

95.1

89

13

102

78.9

60.0

138.9

83.7

62.6

146.3

13

87

3.7

90.7

90

15.1

105.1

76.2

61.5

137.7

80.6

64.8

145.5

14

85

3.2

88.2

90

15.3

105.3

79.0

61.2

140.1

81.6

69.3

150.9

15

87

3.3

90.3

91

13.7

104.7

76.5

64.8

141.3

81.6

66.0

147.6

16

87

3.1

90.1

92

13.1

105.1

78.6

62.4

141.0

82.7

66.0

148.6

17

86

3.1

89.1

90

12.4

102.4

78.1

60.0

138.1

81.6

66.0

147.6

18

87

7.9

94.9

91

14.8

105.8

79.1

62.9

142.1

82.7

67.1

149.7

19

87

87.3

174.3

89

8

97

78.8

60.1

138.9

81.6

68.2

149.8

20

88

68.8

156.8

93

12.6

105.6

79.6

62.0

141.6

81.6

69.3

150.9

21

85

10.3

95.3

94

15.5

109.5

78.7

61.9

140.6

84.7

63.7

148.4

22

86

3.1

89.1

91

12.4

103.4

77.0

61.0

138.0

83.7

62.6

146.3

23

86

2.4

88.4

91

10.8

101.8

79.0

62.8

141.8

82.7

64.8

147.5

24

86

2.3

88.3

90

13.4

103.4

78.0

61.0

139.0

83.7

62.6

146.3

25

87

2.3

89.3

93

15.8

108.8

79.5

60.6

140.1

82.7

61.5

144.2

26

84

2.7

86.7

90

13.8

103.8

79.7

60.1

139.9

81.6

64.8

146.5

27

85

3.1

88.1

93

13.4

106.4

79.1

60.3

139.5

82.7

61.5

144.2

28

84

3.4

87.4

92

14.6

106.6

79.0

63.3

142.2

81.6

64.8

146.5

29

85

3.4

88.4

90

12.4

102.4

79.3

58.6

137.9

82.7

62.6

145.3

30

84

12.9

96.9

92

12.8

104.8

79.2

60.2

139.5

82.7

63.7

146.4


-------
Table 6-20. Data from Extended Sampling Test with Diesel Generator at Idle, Using TSI COMBUCHECK (continued)



Unit A NO

(ppm)

Unit A N02
(ppm)

TSI Analyzer Data
Unit A NOx Unit B NO
(ppm) (ppm)

Unit B N02
(ppm)

Unit B NOx
(ppm)

14ANO
(ppm)

14A N02
(ppm)

Reference Analyzer Data
14A NOs 10 NO
(ppm) (ppm)

10 no2

(ppm)

10 NOx
(ppm)

31

85

37.2

122.2

92

9

101

78.7

63.5

142.1

83.7

63.7

147.4

32

86

29.4

115.4

93

11.5

104.5

78.0

63.9

141.9

82.7

64.8

147.5

33

85

53.1

138.1

91

10.7

101.7

78.5

65.0

143.5

83.7

64.8

148.5

34

85

13.5

98.5

93

4.2

97.2

78.0

60.9

138.9

82.7

63.7

146.4

35

85

3.4

88.4

93

47.4

140.4

79.0

62.2

141.3

84.7

61.5

146.2

36

86

2.7

88.7

93

15.4

108.4

78.2

60.8

138.9

83.7

61.5

145.2

37

85

2.7

87.7

93

5.3

98.3

79.0

62.0

141.0

84.7

61.5

146.2

38

84

2.8

86.8

93

31.1

124.1

79.3

60.4

139.8

83.7

63.7

147.4

39

85

4.2

89.2

93

13.9

106.9

78.8

60.2

139.0

82.7

62.6

145.3

40

82

4.3

86.3

89

4.7

93.7

79.4

63.2

142.6

83.7

64.8

148.5

41

86

3.4

89.4

93

31.3

124.3

78.2

62.6

140.8

82.7

62.6

145.3

42

82

3.3

85.3

93

15

108

79.3

63.4

142.7

83.7

63.7

147.4

43

86

3.3

89.3

93

4.7

97.7

80.0

63.0

143.0

84.7

63.7

148.4

44

85

5.1

90.1

94

8.7

102.7

79.0

62.6

141.7

83.7

63.7

147.4

45

87

3.8

90.8

93

12.2

105.2

81.0

61.5

142.5

83.7

64.8

148.5

46

83

3

86

94

4.5

98.5

80.4

59.1

139.5

84.7

62.6

147.3

47

84

3

87

91

4.4

95.4

81.0

61.8

142.8

84.7

63.7

148.4

48

87

2.8

89.8

95

45.9

140.9

80.6

62.4

143.0

84.7

64.8

149.5

49

84

3.4

87.4

93

9.2

102.2

80.0

61.8

141.8

84.7

63.7

148.4

50

84

5.2

89.2

93

4.4

97.4

81.8

60.6

142.4

85.7

63.7

149.4

51

85

2.4

87.4

94

4.7

98.7

82.1

61.2

143.3

85.7

62.6

148.3

52

85

2.3

87.3

89

60.4

149.4

81.4

61.4

142.8

84.7

62.6

147.3

53

85

2.1

87.1

93

13.5

106.5

81.8

60.4

142.2

83.7

64.8

148.5

54

84

9.5

93.5

91

5.1

96.1

80.4

61.5

141.9

82.7

66.0

148.6

55

83

4.8

87.8

91

4.5

95.5

81.8

60.4

142.2

85.7

63.7

149.4

56

82

3.7

85.7

93

11.4

104.4

82.3

60.6

142.9

85.7

63.7

149.4

57

82

2.3

84.3

92

29.2

121.2

81.7

60.0

141.8

84.7

63.7

148.4

58

83

2.1

85.1

92

6.9

98.9

81.8

61.0

142.8

83.7

64.8

148.5

59

83

4

87

93

45

138

81.0

62.5

143.5

84.7

63.7

148.4

60

87

140.8

227.8

99

210

309

81.7

60.9

142.6

86.7

61.5

148.2


-------
Inspection of Table 6-20 shows that while the TSI NO monitors gave stable performance, the TSI
N02 monitors showed erratic performance during the extended sampling test. N02 readings from
both TSI units were often low relative to the reference analyzers, but were highly variable, and
excursions to high readings on one unit were not usually coincident with excursions on the other
unit. This behavior has been explained by the finding that the in-line water traps used in the TSI
monitors were labeled incorrectly during production, and consequently were installed backwards
in the sample lines to the monitors. In this reversed configuration, the traps would block both
water and soot particles, creating an efficient system for removing N02. Thus the N02 results in
Table 6-20 are not considered a fair example of the performance of the TSI N02 monitors. The
NOx results from the TSI monitors are subject to the same conclusion.

However, despite not being designed for extended sampling, the TSI NO monitors showed good
performance, consistent with the relative accuracy values found for NO on the diesel source
(Table 6-17). As shown in Table 6-21, Unit A showed a slight decrease in NO over the 1-hour
sampling period, and Unit B a slight increase, the latter being consistent with the indication of the
reference analyzers. The trends in NO were significantly different for both Units A and B from
that of the reference analyzers. However, the actual differences in NO trends were fairly small.
For exmple, Unit A showed an NO slope of -0.038 ppm/min (-2.3 ppm/hr), and Unit B a slope of
0.163 ppm/min (+9.8 ppm/hr), compared to the reference analyzer slope of 0.064 ppm/min (+3.8
ppm/hr). Thus, over the one-hour sampling period, the different trends in NO amounted to about
a 6 ppm deviation from the trend of the reference analyzers, or roughly 7 percent of the NO level
in the diesel exhaust. This degree of performance is good considering that the NO monitors are
not intended for this application.

6.2.4 Inter- Unit Repeatability

The repeatability of test results between the TSI monitors was assessed in those cases where the
data lent themselves to application of a t-test. The resulting t-statistics and associated p-values are
listed in Table 6-22. Highlighted in bold are those p-values less than 0.05, which indicate a
statistically significant difference between the two TSI COMBUCHECK NO or N02 units at the
95 percent confidence level. As Table 6-22 shows, significant differences between Units A and B
were found only in relative accuracy and in measurement stability.

The unit-to-unit differences found in the source combustion tests are related to two factors. One
is the failure of the monitors to provide accurate readings even on NO and N02 span gases,
despite having been calibrated with similar mixtures prepared from the identical sources. This
observation is described in Sections 6.2.1 and 6.2.2. The second factor is the widely variable
readings obtained from the two N02 monitors in source sampling, which is discussed in Sections
6.2.1 through 6.2.3, and which was likely caused by the reversed installation of the in-line water
traps. This factor invalidates the N02 and NOx data from the combustion source tests, and may
have been a cause of the unstable span responses.

49


-------
Table 6-21. Results of Evaluation of Measurement Stability for TSI COMBUCHECK
Monitors

Unit A	Unit B	Reference Units Average



NO

no2

NOx

NO

no2

NOx

NO

no2

NOx

Slope

-0.038

-0

.168

-0

.206

0.163

0.171

0.334

0.064

-0.045

0.019

(Std Err)

(0.012)

(0.

184)

(0.

187)

(0.023)

(0.213)

(0.212)

(0.007)

(0.010)

(0.011)

Difference in Slopes

-0.102

-0

1?3

-0

??5

0.099

0.216

0.315







from the Average of
the Reference Units

(0.015)

*

(0.

183)

(0.

184)

(0.025)

*

(0.213)

(0.211)

—

—

—

(Std Err)























* Statistically significant different slopes between test unit and the average of the reference units at the 5%
significance level.

Table 6-22. Summary of Repeatability



Unit A vs. Unit B



NO

no2

NOx

Linear Regression

Intercept

t-statistic

-0.348

-1.863

—





p-value*

0.734

0.092

—



Slope

t-statistic

0.168

1.456

—





p-value

0.870

0.176

—

Detection Limit



t-statistic

0.468

-1.352

—





p-value

0.644

0.186

...

Relative Accuracy

Gas Rangetop

t-statistic

1.644

32.195

10.932





p-value

0.139

<0.001

<0.001



Gas Water Heater

t-statistic

8.222

2.326

4.842





p-value

<0.001

<0.048

<0.001



Generator-High

t-statistic

0.204

3.896

3.583



RPM

p-value

0.843

<0.005

<0.007

Measurement

Slope

t-statistic

-11.638

-2.038

3.266

Stability



p-value

<0.001

0.046

0.002

* p-value <0.05 indicates that two test units are statistically different at the 5 percent significance level.

6.3 Other Factors

In addition to the performance characteristics evaluated in the laboratory and source tests, three
additional factors were recorded: monitor cost, data completeness, and maintenance/operational
factors.

50


-------
6.3.1 Cost

The cost of each single gas monitor as tested in this verification test was approximately $700.

6.3.2	Data Completeness

The data completeness in the verification tests was 100 percent for all four units of the TSI
COMBUCHECK.

6.3.3	Maintenance/Operational Factors

The short duration of the verification tests prevented assessment of long-term maintenance costs,
durability, etc. The TSI monitors are hand-held devices, readily portable and easy to calibrate and
operate. They require minimal setup or take-down time. Their main operational limitation is that
they are single-gas analyzers; thus, multiple units would be needed to determine multiple gases.

51


-------
Chapter 7
Performance Summary

As discussed elsewhere in the report, the TSI COMBUCHECK is a low-cost, versatile single-gas
detector designed for spot-checking, short-measurement applications. It is not intended to be used
for long-term measurements of combustion. The primary analytes are CO and oxygen; those two
versions of the COMBUCHECK were not tested in this study. Performance of the NO and N02
monitors tested herein may not be representative of the product line in general. The testing
performed in this study indicates that the manufacturer's cautions against prolonged exposure to
high concentrations certainly hold true for the NO and N02 versions of the instrument.

The TSI COMBUCHECK monitors are capable of providing linear response over their full ranges
of 0 to 2,000 ppm NO and 0 to 100 ppm N02, provided that stable calibration is achieved. High
correlation of TSI response to calibration concentration was observed, but slopes of about 1.1 to

1.2	were found for NO and N02, respectively. This finding is related to the failure of the monitors
to provide accurate readings on span gases after calibration at 200 ppm NO and 25 ppm N02.
Detection limits determined from the calibration data were about 20 to 25 ppm for NO and 1.5 to
3 ppm for N02. These values are undoubtedly inflated by the monitors' slow rate of return to a
baseline reading, after exposure to a high NO or N02 level. Care must be taken after conducting
high level measurements to allow for an extended stabilization period on clean, dry air, before
attempting relatively low measurements.

No substantial interference was observed with either the NO or N02 monitors from 496 ppm CO,

5.03	percent C02, 494 ppm NH3, 590 ppm of total hydrocarbons, 501 ppm of S02, or 451 ppm
S02 in the presence of 383 ppm NO.

Response times for NO for two units tested were 65 and 26 seconds, respectively; response times
for N02 for two units tested were 126 and 142 seconds, respectively. Zero drift was less than
3 ppm for all four TSI monitors in sampling combustion sources and in most laboratory testing.
Zero drift in the NO monitors was 20 to 30 ppm, in laboratory testing of linearity over the full
2,000 ppm measurement range of the monitors. Span drift varied widely depending on the type of
test being conducted and the levels of span gases provided. In laboratory testing with span gases
of 1,400 to 2,000 ppm, NO span drift was 2 to 8 percent of the span level. Corresponding N02
span drift, with span gases of 70 to 100 ppm, was 5 to 13 percent of the span level. In source
testing, with span gases of 20 to 200 ppm, NO span drift was 4 to 25 percent of the span level; for
N02 with span gases of 10 to 50 ppm, span drift was 2 to 35 percent of the span value. No
significant additional drift occurred when the TSI monitors were shut down completely overnight.

52


-------
Over a range of+10 to -10 inches of water (relative to ambient pressure) the sample gas pressure
had no significant effect on the zero or span readings of the TSI monitors. The ambient tempera-
ture over the range of 7 to 41 °C (45 to 105°F) also had no significant effect on the zero and span
readings.

Relative accuracy for NO for the two tested monitors was within the manufacturer's stated 5 ppm
tolerance with NO levels below 10 ppm, but ranged from 8.6 to 40.4 percent, over three com-
bustion sources having NO emissions from about 6 to 100 ppm. Highly variable and inconsistent
readings were observed from the N02 monitors in some tests, apparently as a result of the
incorrect installation of the in-line water traps in the monitors. As a result, relative accuracy
results obtained for N02 and NOx are not considered a valid assessment of analyzer performance.
The variability of the monitors was such that the two TSI NO monitors, and the two TSI N02
monitors, performed significantly differently from one another in all the tests conducted with
combustion sources.

53


-------
Chapter 8
References

1.	Test/QA Plan for Verification of Portable NO/NO 2 Emission Analyzers, Battelle, Columbus,
Ohio, December 1998.

2.	U.S. EPA Method 7E, Determination of Nitrogen Oxides Emissions from Stationary Sources
(Instrumental Analyzer Procedure) Code of Federal Regulations, 40 CFR, Ch. 1, Part 60,
Appendix A (1991).

3.	Traceability Protocol for Establishing True Concentrations of Gases Usedfor Calibrations
and Audits of Continuous Source Emission Monitors: Protocol Number 7, Research
Triangle Park, NC: U.S. Environmental Protection Agency, Quality Assurance Division, June
1978.

4.	Interlaboratory Program to Validate a Protocol for the Measurement of NO 2 Emissions
from Rangetop Burners, GRI-94/0458, Gas Research Institute, Chicago, Illinois, December
1994.

5.	Interlaboratory Study to Determine the Precision of an Emission Measurement Protocol for
Residential Gas Water Heaters, GRI-96-0021, Gas Research Institute, Chicago, Illinois,
March 1996.

6.	American National Standard (ANSI Z21.1) "Household Cooking Gas Appliances," American
National Standards Institute, 24th Edition, American Gas Association, 1990.

7.	Quality Management Plan (QMP) for the ETV Advanced Monitoring Systems Pilot, U. S.
EPA Environmental Technology Verification Program, Battelle, Columbus, Ohio,

September 1998.

8.	Portable NOx Analyzer Evaluation for Alternative Nitrogen Oxide Emission Rate
Determination at Process Units, Source Testing and Engineering Branch, SCAQMD Air
Quality Management District, Los Angeles, CA, September 21, 1994.

9.	U.S. EPA Method 5, Determination of Particulate Emissions from Stationary Sources, Code
of Federal Regulations, 40 CFR, Ch. 1, Part 60, Appendix A (1991).

54


-------
Appendix A
Data Recording Sheets

A-l


-------
Date:

Linearity Test Data Sheet

_ Vendor/Analyzer: _

Form Filled Out By:

Pre-Test Z/Span: Unit A: Zero (NO/NO2)
Unit B: Zero (NO/NO2)
NO Test

Unit A
(NO/NO2)

UnitB
(NO/NO2)

1.

2.

3.

4.

5.

Time Response 6.

7.

Span (NO/NO2) 	L

Span (NO/NO2) 	L

NO? Test

Unit A
(NO/NO2)

UnitB
(NO/NO2)

L_

L.
7.

9.

10..

11-.

12..
13..
14..
15..
16..
17..
18..
19..
20..
21.

9.

10.

11.

12.

13.

14.

15.

16.

17.

iiL

19.

20.

21.

Post-Test Z/Span: Unit A: Zero (NO/NO2) /	Span (N0/N02) 	/

UnitB: Zero(NO/NC>2) /	Span(N0/N02) 	/


-------
Interrupted Sampling Data Sheet

Date: 	 Vendor/Analyzer: 	

Form Filled Out By: 	

Pre-Shut Down Z/Span:

Date: 	 Time:

Unit A (NO/NO2) Zero /	Span

Unit B (NO/NO2) Zero /	Span

Post-Shut Down Z/Span:

Date: 	 Time:

Unit A (NO/NO2) Zero /	Span

Unit B (NO/NO2) Zero /	Span


-------
Date:

Interference Test Data Sheet

	 Vendor/Analyzer: 	

Form Filled Out By: 		

Interference Gas Concentration
Zero

CO	496 ppm

Zero

C02	5.03%

Zero

NH3	494 ppm

Zero

Hydrocarbons	590 ppm

Zero

S02	501 ppm

Zero

SO2 + NO	451 ppm + 393 ppm

Response fNO/NO-A
Unit A	UnitB


-------
Flow Rate Sensitivity Data Sheet

Date: 	-	Vendor/Analyzer: 	

Form Filled Out By: 		

Flow Rate Data: Unit A	Unit B

Tccml	(ccrn)

Ambient P				

+10inH2O				

-lOinEbb				

Response Data:

Ambient P

Unit A
(NO/NCM

UnitB
(NO/NO

Zero

NO Span
Zero

NO2 Span
Zero

+10inH2O Zero

NO Span
Zero

NO2 Span
Zero

-lOinE^O Zero

/NO Span
Zero

NO2 Span
Zero


-------
Ambient Temperature Test Data Sheet

Date: 		 Vendor/Analyzer:

Form Filled Out By: 	

Room Temperature: 		Response fNO/NCM

Unit A UnitB
Zero	/	/

NO Span	/	/

NO2 Span	/	/

Zero	/	/

Cold Chamber Temperature: 	

Zero	/		/

NO Span	/		/

NO2 Span	/		!_

Zero	/		!_

Heated Chamber Temperature: 	

Zero	/ 	L

NO Span	/	/

NO2 Span	/ 	L

Zero	/ 	!_

Room Temperature:
Zero
NO Span
NO2 Span
Zero


-------
Accuracy Test Data Sheet: Rangetop Combustion

Date		Vendor Analyzer:	

Form Filled Out By:	

Pre-Test Zero/Span

Calibration Gas & Concentration:	 Instrument Range:	

Calibration Gas & Concentration:	 Instrument Range:	

Unit 14A: Zero (NO/NCtyNOx) / /	Span (N0/N02/N0x)

Unit 10: Zero (N0/N02/N0x) / /	Span (NO/NOz/NOx) / I

Unit 14A	Unit 10

(N0/N02/N0x)	(N0/N02/N0x)

1.	/	/		_/

2.	/	/		/

3.	/	/		/

4.	/	/		/

5.	/	/		/

6.	/	/		/

7.	/	I		/

8.	/	/		/

9.	/	/		/

Post-Test Zero/Span

Calibration Gas & Concentration:	 Instrument Range:	

Calibration Gas & Concentration:	 Instrument Range:	

Unit 14A: Zero (N0/N02/N0x) / /	Span (N0/N02/N0x)

Unit 10: Zero (NO/NOz/NOx) / /	Span (N0/N02/N0x) _/_/

Mod-1:01/17/99


-------
Accuracy Test Data Sheet: Water Heater Combustion

Date		Vendor Analyzer:	

Form Filled Out By:_		

Pre-Test Zero/Span

Calibration Gas & Concentration:	 Instrument Range:	

Calibration Gas & Concentration:	 Instrument Range:	

Unit A: Zero (NO/N02/NOx) / /	Span (N0/N02/N0x) / /

Unit B: Zero (NO/N02/NOx) / /	Span (N0/N02/N0x) / /

Unit A	Unit B

(NO/N02/NOx)	(N0/N02/N0x)

1.		/

2.		/

3.		/

4.		/

5.		/

6.		/

7.		/

8.		/

9.		I

Post-Test Zero/Span

Calibration Gas & Concentration:	 Instrument Range:	

Calibration Gas & Concentration:	 Instrument Range:	

Unit A: Zero (NO/N02/NOx) / /	Span (NO/N02/NOx)

Unit B: Zero (NO/N02/NOx) / /	Span (N0/N02/N0x)

Mod-1:01/17/99


-------
Accuracy Test Data Sheet: Diesel-Engine Combustion

Date		Vendor Analyzer 		

Form Filled Out By:			__	

Pre-Test Zero/Span

Calibration Gas & Concentration:	 Instrument Range:	

Calibration Gas & Concentration:	 Instrument Range:	

Unit 14A: Zero (N0/N02/N0x) / /	Span (N0/N02/N0x) 	/

Unit 10: Zero (N0/N02/N0x) / /	Span (N0/N02/N0x) 	/

Unit 14A	Unit 10

(N0/N02/N0x)	(N0/N02/N0x)

1.	/	/		/

2.	/	/		/

3.	/	/		/

4.	/	/		/

5.	/	/		/.

6.			L

7.	/	/		./.

8.		/

9.	/	/		/.

Post-Test Zero/Span
Calibration Gas & Concentration:	 Instrument Range:	

Calibration Gas & Concentration:	 Instrument Range:	

Unit 14A: Zero (N0/N02/N0x) / /	Span (N0/N02/N0x) 	/

Unit 10: Zero (N0/N02/N0x) / /	Span (N0/N02/N0x) 	/.

Mo-1: 01/17/99


-------
Measurement-Stability Test Data Sheet: Diesel-Engine Combustion

Date	 Vendor Analyzer	

Form Filled Out By:			

Diesel-Engine Load:	

Time	Unit A

(t+min#)	(N0/N02/N0x)

I-	/	/

2.	/	/

3.		

4.	/	/

5.	/	/

6.	/	/

7.	/	/

8.	/	/
9-	/	/
10.	/	/

II-	/	/

12.	/	/

13.	/	/

14.	/	/

15.	/	/

16.	/	/

17.	/	/

18.	/	/
19- 	/	/

20.	/	/

21.	/	/
22	/	/

23.	/	/

24.	I	/

25.	/	/

26.	/	/

27.		

28.	/	/

29.	/	/

Unit B

(N0/N02/N0x)

30.


-------
Measurement-Stability Test Data Sheet: Diesel-Engine Combustion

Date		Vendor Analyzer:	

Form Filled Out Bv: 	

Diesel-Engine Load:,

Time
(t + min#)

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52

53.

54.

55.

56.

57.

58.

59.

Unit A

(N0/N02/N0x)

Unit B

(N0/N02/N0x)

60.


-------
Appendix B
External Technical Systems Audit Report

B-l


-------
Environmental Technology Verification Program
Advanced Monitoring Systems Pilot
Air Monitoring Systems

N0/N02 Monitors Verification Test
January 20-21,1999 Audit

Audit Report: ETVAMS001
Revision 1

d Elizabeth A. Betz r _

Elizabeth T. Hunike	-


-------
ETV AMS Air Audit
Page 1

1.0	Audit Information

1.1	Auditors:

Elizabeth A. Betz

Human Exposure & Atmospheric Sciences Division
U. S. EPA, NERL (MD-77)

Research Triangle Park, NC 27711
(919)541-1535

Elizabeth T. Hunike

Atmospheric Methods & Monitoring Branch
Human Exposure & Atmospheric Sciences Division
U. S. EPA, NERL (MD-46)

Research Triangle Park, NC 27711
(919) 541-3737

1.2	Dates of Audit: January 20-21, 1999

1.3	Location of Audit: Battelle Memorial Institute, Columbus, Ohio

1.4	Battelle Staff Interviewed and/or Observed:

Karen Riggs	ETV AMS Pilot Manager

Susan Abbgy	QA/QC Reviewer

Sandy Anderson	QA Manager
Verification Test Team:

Tom Kelly	Verification Test Leader

Joe Tabor	Laboratory Verification Testing

Jim Reuther	Emission Source Verification Testing

Steve Speakman	Operator, Emission Sources/Reference Method

2.0 Background

Throughout its history, the U.S. EPA has evaluated technologies to determine their effectiveness in
preventing, controlling, and cleaning up pollution. EPA has expanded these efforts by instituting the
Environmental Technology Verification Program (ETV) to verify the performance of a larger number of
innovative technical solutions to problems that threaten human health or the environment. The goal of
ETV is to verify the environmental performance characteristics of commercial-ready technology through
the evaluation of objective and quality assured data, so that potential purchasers and permitters are
provided with an independent and credible assessment of what they are buying and permitting. The
ETV Program Verification Strategy outlines the goals, operating principles, pilot selection criteria, and
implementation activities. ETV includes twelve pilot projects. In these pilots, EPA is using the
expertise of partner verification organizations to design efficient processes for conducting performance
tests of innovative technologies. The implementation activities involve forming stakeholder groups who
identify technologies needing verification, designing a generic verification protocol and then Test/QA


-------
ETV AMS Air Audit
Page 2

Plans for the specific technology to be verified. The verification tests are run on the identified
technologies wishing to participate and verification statements based on the test results are generated.

One pilot, entitled Advanced Monitoring Systems (AMS), is to verify the performance of commercially
available technologies used to monitor for environmental quality in air, water and soil. This pilot is
managed by EPA's National Exposure Research Laboratory in Research Triangle Park, North Carolina
and their verification partner for the AMS pilot project is Battelle Memorial Institute, Columbus, Ohio.
This pilot has been divided into three sub-pilots, each looking at monitoring systems for a specific
media, air, water and, eventually, soil. The Air AMS portion has evolved to the point of actually
running verification tests on available air monitoring instrumentation.

3.0	Scope of Audit

3.1	Audit Preparation. The auditors reviewed the following documents pertinent to the ETV AMS
Pilot:

a.	Environmental Technology Verification Program Quality and Management Plan for the
Pilot Period (1995-2000), May 1998

b.	Environmental Technology Verification Program Quality Management Plan for the
ETV Advanced Monitoring Systems Pilot, September 1998

c.	Test/QA Plan for Verification of Portable NO/NO 2 Emission Analyzers, December 4,
1998

d.	U. S. EPA Method 6C, Determination of Sulfur Dioxide Emissions from Stationary
Sources (Instrumental Analyzer Procedure)

e.	U. S. EPA Method 7E, Determination of Nitrogen Oxides Emissions from Stationary
Sources (Instrumental Analyzer Procedure)

Based on the above material, a checklist was prepared. The U. S. EPA ETV AMS Pilot Manager,
Robert G. Fuerst, was provided the checklist prior to the audit. The completed checklist for this audit is
attached.

3.2	Audit Scope.

The audit encompassed a technical systems audit of a verification test (VT) on nitrogen oxides monitors
at Battelle. A technical systems audit is a qualitative onsite audit of the physical setup of the test. The
auditors determine the compliance of testing personnel with the test/QA plan. The auditors were on site
from Wednesday afternoon through Thursday afternoon. The technical systems audit was performed on
the flow rate and ambient temperature of the laboratory portion of the VT and the relative accuracy
tests with the gas cooktop, water heater and a portion of the lower range emissions of a diesel
generator. No performance evaluations were conducted as a part of this audit.

4.0	Executive Summary

4.1	The VT is well-managed, particularly considering its complexity. All personnel appeared to be
well-trained for their particular duties. All involved showed enthusiasm and ingenuity during the VT.


-------
ETV AMS Air Audit
Page 3

4.2	The significant findings of this audit, cited in paragraph 5.0 below, had also been found by
Battelle's QA staff during their audit earlier in the VT.

4.3	The technical systems audit showed that the VT personnel were very familiar with the Test/QA
Plan. With one exception, differences for this VT from the original Test/QA Plan were well
documented by deviation reports on file at Battelle. The deviation report format includes a date, cites
the deviation, provides an explanation of the deviation and requires an approving Battelle signature. It
was impressive that the deviation reports were present and were completed up front. The one
difference from the VT that was not cited in a deviation report was that Battelle had intended to run an
analyzer already on hand completely through the VT before the first vendor's analyzer. This was not
done nor was a deviation report generated. The remaining differences were cited in the deviation
reports.

5.0	Major Findings

5.1	Undocumented Deviation from the Test/QA Plan. The undocumented deviation was from
section 5.6, Test Schedule, and stated "to avoid bias in testing of the first analyzers through the
sequence, Battelle's personnel will first conduct the entire test sequence using an analyzer already on
hand at Battelle. Testing will then continue with analyzers named in section 2.4." Due to a delay in the
arrival of the protocol gases used in the VT, Battelle did not run one of their instruments through the
test sequence. As a result a leak in the gas supply system in the laboratory test portion was not detected
before the first vendor started the VT sequence.

5.2	Initial Calibration of Instruments for Emission Source Testing. The Test/QA Plan states that
"the chemiluminescent monitors to be used for Method 7E reference measurements will be subjected to
a 4-point calibration with NO prior to the start of verification testing, on each measurement range to be
used for verification." The initial Emission's portion of the VT was started on January 13, 1999. There
was no 4-point calibration with NO recorded in the Emission's VT laboratory notebook prior to the
January 13th testing. This finding is also a finding in Battelle's Internal Audit conducted during the first
week of the VT.

6.0	Results of Technical Systems Audit

6.1	Organization. The Battelle ETV AMS VT team consisted of four members. All team members
were very knowledgeable of the procedures and helpful to the auditors. There are also two Battelle
Quality Assurance staff members that are members of the ETV AMS team. Both were available and
very helpful to the auditors. These Battelle QA staff members are responsible for running the internal
audits required by the ETV related QMPs. One such audit was conducted the week prior to this EPA
audit.

6.1.1 The Test/QA Plan stated that a Dr. Agnes Kovacs would be providing statistics and data
analysis for this VT. One of the documented deviations was that Dr. Kovacs would not be participating
in the VT as she has left Battelle. Although the deviation report stated that someone in the Statistics
and Data Analysis Department would be taking her place, there was no indication in the deviation report
as to who it would be.


-------
ETV AMS Air Audit
Page 4

6.2	Gas Cylinder Certifications. A review of the gas cylinder certifications uncovered some minor
discrepancies. The expiration date on two of the cylinder certifications did not match the expiration
date on the cylinders. The discrepancy was corrected by the gas manufacturer on the day of inspection.
Battelle did not initially have certifications for the gas cylinders used in the source test. The gas
manufacturer was contacted by phone and faxed in certifications for 3 of the 4 cylinders. The original
certificates were later located on one of the team member's desk. The gas cylinder for one of the
certificates reviewed was not found among the ETV VT equipment.

6.3	Temperature Sensor Certification. The certificate in the notebook maintained for the Laboratory
Test Portion was for Model 402A, Serial # 40215 Temperature Indicator. This indicator was not seen
by the Auditors. The Temperature Indicator used in the Laboratory Test portion to read the
temperature of the monitors during the Ambient Test was Model 400A, decal # LN-560558. The
certificate was not in the notebook, however, the indicator did have a label on it that stated that it was
certified 1-7-99. Discussion with Susan Abbgy, after the audit, clarified that LN-560558 was an internal
Battelle laboratory number and that the manufacturer's serial number on LN-560558 was 40215.
However, the certificate did reflect an incorrect model number for Temperature Indicator Serial #

40215.

6.4	Deviation Reports. The dated reports cited the deviation, provided an explanation/justification of
the deviation and required an approving Battelle signature. It was impressive that the deviations reports
were present and were completed up front.

6.4.1	The Flow Rate Sensitivity Test procedure had three deviation reports. The Test/QA Plan
called for the use of 60% span value during the test. A deviation report cited that this was changed to
70% span value to correlate to the Linearity Test. The two other reports related to the Flow Rate
Sensitivity Test were very similar and called for a change in the order of the procedure to reduce the
amount of plumbing changes required.

6.4.2	The Ambient Temperature Test had one deviation report. The order of the test was
changed. The procedure called for doing a cooled chamber test first and then hot. The deviation report
stated that all VTs will be done in the reverse order. The reason for the deviation was based on
discussions with the vendors that indicated the rise in temperature after exposure to NO may cause
more drift. The order was reversed to more clearly observe any drift.

6.4.2.1 During the Ambient Temperature Test observed, slight changes were made to
accommodate the mass of the monitors. The vendor's monitors were larger than previous monitors and
generated and held heat longer. The door to the heated chamber, once the monitors reached its
temperature, had to remain slightly ajar to hold the chamber temperature at a constant value. The
heated monitors were then placed in the cold chamber (a standard household refrigerator). The heat
given off by the monitors raised the temperature in the refrigerator over 100°F. To obtain a cooled
chamber reading the team members relocated the monitors to the outdoors which produced a cooled
ambient temperature within the 45°F ±5°F for the one hour required for temperature equilibration and
the additional time required to perform the zero and span check. This was a fine example of the
ingenuity the VT team members showed to accommodate differences in monitors.

6.4.3	Interference Test. The mixture of S02 and NO for the Interference Test was changed
from interferant levels of 250 ppm each of S02 and NO to interferant levels of 451 ppm S02 and 393


-------
ETV AMS Air Audit
Page 5

ppm NO. According to the deviation report, this change was made because the NO standard available
wasn't at the anticipated concentration when the Test/QA Plan was written.

6.4.4 Source Testing.

6.4.4.1	The Test/QA Plan cited the use of two diesel generators for the Source Test.
The selection of these generators was made based on studies that Battelle had used in the past that
provided a database of emission levels generated by these sources. However, these generators were
property of the Air Force and were unavailable at the time of the VT due to military activities in the
Middle East. Battelle substituted one generator they had on site and collected emission data at two
speeds to provide two higher emission levels than previously provided by the cooktop or water heater.
This substituted generator produced two levels of emissions; however, neither level was over 500 ppm
of NO. The database that Battelle had on the originally planned generators showed that one model
would produce ranges between 100-1000 ppm NOx and the second model would produce ranges
between 600-2300 ppm NOx. The impact of this change is that there will be no verification for higher
ranges.

6.4.4.2	The oxygen sensor was not used during the source test. This VT's focus was
the verification of NO/N02 levels and not to compare oxygen data. Source stability will be documented
by NOx measurements instead of oxygen measurements. The source stability for the water heater and
the cooktop is also documented in two Battelle reports on data from these specific sources used in
interlaboratory comparisons from 1994 through 1998. The initial generators planned for the VT also
had similar data bases. The source stability of the generator actually used was verified by data collected
in December and January prior to the VT. The actual data collected by the reference monitors during
the VT also verified the source stability.

6.4.4.3	ThermoEnvironmental Models 14A and 10 NO/N02 analyzers were used for the
reference method. The Test/QA Plan called for identical Beckman Model 955 monitors. The reason
stated in the deviation report for the substitution was that the Thermo Instruments are newer and are in
more current use.

6.4.4.4	Triplicate readings of calibration points were not run in the calibration of the
reference method analyzers. Method 7E does not require triplicate readings of calibration points.

6.4.4.5	One deviation report addressed the use of unheated sample lines and poly tubing.
The Test/QA Plan is based on EPA Method 7E but based on Battelle's own experience with the sources
in the laboratory environment an unheated inlet was used. Additionally it should be noted that the VT is
conducted inside in a laboratory setting with controlled temperature and humidity and Method 7E is for
stack sampling. The only comment on this deviation report is that the originator of the deviation signed
the report instead of obtaining an independent approval signature.

6.5 Leak Detected in the System in the Laboratory Test Portion. During the first vendors's
laboratory test portion, a leak was detected in the system. The data sheets for the laboratory test
portion of the first vendor's VT showed a note that a leak was detected and the vendor recorded
oxygen levels. Also noted on the data sheet was a correction factor that would be used on the vendor's
data that was made based on the vendor's oxygen readings. The correction factor notes were brought
to the auditor's attention by Battelle's QA staff. Because the VT did no verification of oxygen levels,


-------
ETV AMS Air Audit
Page 6

the correction factor may be inaccurate. As part of the documentation for that VT, the accuracy of the
oxygen readings by the vendor needs to be addressed.

6.6	Initial Calibrations and Tests in the Source Laboratory. As stated under major findings,
paragraph 5.2 above, the initial calibrations of the chemiluminescent monitors used as the Method 7E
references were not done before the first VT. In addition no interference test was conducted prior to
1-18-99 which was after the second VT. However, all subsequent VTs had the required initial
calibration and interference tests. This was also a finding in Battelle's internal audit conducted a week
earlier. Battelle will need to address this in the VT report.

6.7	Corrections of Data Sheets. In most instances, corrections made on the data sheets followed
Good Laboratory Practices; however, some did not (i.e., one line was not drawn through the incorrect
entry and the correction was not dated and initialed).

6.8	Source Laboratory Notebook Entries. The initial entries were difficult to follow because the
writing was almost illegible and there were missing entries. However, with the exception of the first
VT, the four-point initial calibrations are recorded and the time and dates of the VTs are also shown.
The actual source test data are recorded on data sheets. The notebook is only used to record the
calibration and interference test data on the reference monitors and to record the times, dates and
comments on the VTs.


-------
Checklist for Verification Test (VT) of Portable N0/N02 Emission Analyzers
Date(s): January 20-21. 1999	Location: Battelle. Columbus. Ohio

Personnel Involved in the Audit:



Titles

Names

EPA Auditor(s):

| Elizabeth Betz

| Elizabeth Hunike

Battelle QA Rep present:

QA/QC Reviewer

Susan Abbgy

QA Manager

Sandy Anderson

Battelle Auditees:

ETV AMS Pilot Manager

Karen Riggs

Verification Test Leader

Tom Kelly

Laboratory Verification Testing

Joe Tabor

Emission Source Verification
Testing

Jim Reuther

Operator, Emission
Sources/Reference Method

Steve Speakman

Vendor(s) Present:

Horiba

J. David Vojtko

General

Comments

Are the Testers familiar with:

ETV QMP

All staff seem familiar with the
documents and there are copies of each
in the ETV reference notebooks
maintained in the Laboratory and
Source Testing areas

Verification Protocol

Test/QA Plan

QA Manager

Generic Verification Protocol:

Finalized?

The Protocol has been finalized and is
in the process of being placed on the
web.

Test Plan:

Approved and Signed?

The test plan has been reviewed by the
vendors. Approval signatures have
been received as vendors have arrived
to participate in the verification test


-------
ETV AMS Pilot
N0/N02 Checklist
Page 2

Technologies:

-Electrochemical (EC) sensors

Testo's Model 350 electrochemical NO and N02 analyzer

Also by direct measurement: 02, CO, S02, Stack Temperature, Stack Pressure
By calculation: C02

Energy Efficiency System's ENERAC 3000SEM electrochemical NO & N02 analyzer

Also by direct measurement: 02, CO, S02, C02, Stack Temperature
TSI's COMBUCHECK electrochemical NO or N02 analyzer
ECOM's A-Plus electrochemical NO and N02 analyzer

Also by direct measurement: 02, CO, S02, Stack Temperature, Stack Pressure
By calculation: C02

-Chemiluminescence emitted from the reaction of NO with 03 produced within the analyzer
Horiba's Model PG-250 portable gas analyzer

Also by direct measurement: 02, CO, S02, C02

The audit was run during the second week of the Test Plan and the 4th vendor was being verified. The

vendor was Horiba.	

Pre-Test Requirements:

Dry Gas Meter:	Initial Calibration Date: See Below

Accurate within 1% and measured in ft3

Calibrated against a volumetric standard within 6 months preceding VT
During VT, checked at least once, against reference meter

In-Line Meter. Serial # 1036707. Rockwell R-275. certified 1/18/99	

Reference Meter model DTM 115 certified 9/22/98	

Temperature Sensor/Thermometers:	Initial Calibration Date: See Below	

Calibrated against a certified temp, measurement standard within 6 months preceding VT
During VT, checked at least once, against an ASTM mercury-in-glass reference thermometer at
ambient temperature and be within 2%.

Temperature Indicator. Serial #40215. Model 402A. certified 1/7/99. certificate available but didn't
locate this indicator. Temp indicator in Lab. LN-560558. Model 400A. certified 1/7/99.	

Oxygen Monitor:	Initial Calibration Date:	

Calibrated within the last six months

During VT, checked once every test day by sampling of ambient air
During operation of one combustion source, assessed for accuracy

Did not use as cited in a documented deviation report.	


-------
ETV AMS Pilot
N0/N02 Checklist
Page 3

Chemiluminescent Monitors to be used for Method 7E

Initial Interference Response conducted prior to VT	Date: See Below	

Measurement System Preparation prior to VT	Date: See Below	

Analyzer Calibration Error prior to VT	Date: See Below	

Sampling System Bias Check prior to VT	Date: See Below	

N02 to NO Conversion Efficiency	Date: See Below	

Calibrations	Initial Calibration Date: See Below

4-point calibration with NO & N02 prior to VT, on each measurement range

For Horiba's VT both were run 1/20/99. however neither were done before first VT. Interference
response was conducted prior to Horiba's VT but not prior to the first VT.	

Each point shall be prepared in triplicate - cited in a documented deviation report	

Calibration error requirement: <±2% of span for the zero, midrange and high-range calibration

gases.

Zero and Span checks done daily AM and PM during the VT

Observed AM checks before source test, not present for PM.	

Gas Dilution System	Initial Calibration Date: 12/16/98	

Flow measurement/control devices calibrated prior to VT by soap bubble flow meter.

Calibration Standards:

EPA Protocol 1 Gases (Calibration paperwork available):

NO in N2, High Range: 80-100% of span
Mid-Range: 40-60% of span
Zero: Concentration <0.25 % of span, ambient air

Protocol Cylinder # ALM057210 expiration date on certificate and cylinder tag did not match.	

Cylinder # ALM017108 expiration date on certificate and cylinder tag did not match.	

Certificate available for Cylinder # ALM036273 but could not locate cylinder.	

Certificates for Source Lab cylinders (AAL14789. ALM014050. AAL17452. ALM015489N) could not
be initially located.	

Sample Location:

Minimum of 8 duct diameters downstream and 2 duct diameters upstream of flow disturbances
and center point of the flue vent
The minimal distances from flow disturbances cited in the Reference Method relate to particulate and
are not critical for gases and were not used. Vendor's instrument sampling tubes were placed beside
those for the reference instruments.


-------
ETV AMS Pilot
N0/N02 Checklist
Page 4

Day One - Laboratory Tests:

Linearity: (response over the full measuring range) - Not Observed
21 measurements for each analyte (NO, N02 or NOx)

Zero six times, each other three times
Calibration points used: 0, 10, 20, 40, 70 and 100% of the analyzer's measuring range
Horiba: 0-25, 0-50, 0-100, 0-250, 0-500, 0-1000, 0-2500
0, 250, 500, 1000, 1750 for 0-2500

Initial Zero and Span check?

After every three points, pure dilution gas provided and the analyzers' readings recorded?

Is the order of concentration points followed?

Final Zero and Span Check?

Linearity test was not observed; however, data sheets were examined. The 100% span used for the
Horiba was 500 ppm. The laboratory log sheets verified that 21 measurements were made, the order of
concentration points cited was used, and that initial and final Zero and Span checks were done.	

Response Time Determinations - Not Observed

Analyzer's response recorded at 10 second intervals during Response Time check (estimated to be 30
readings)

Detection Limit - Not Observed

Detection limit is based on data from zero and 10% readings during Linearity test (9 readings)

Interrupted Sampling (four readings total) - Not Observed
Zero and Span recorded at end of Linearity Test on Day One


-------
ETV AMS Pilot
N0/N02 Checklist
Page 5

Day Two - Laboratory Tests

Interrupted Sampling continued - Not Observed

Zero and Span are recorded after analyzer has been powered up before any adjustments
Same Span from previous day is used
Interference Tests: - Not Observed

Actual concentrations were obtained from the data sheets. A documented deviation cited the change in
the SOt and NO interferant concentrations.

Interferant

Interferant Concentration

Target Analyte

CO

500 ppm - Actual concentration used - 496 ppm

NO, N02, NOx

o
o

N)

5% - Actual percentage used - 5.03%

NO, N02, NOx

so2

500 ppm - Actual concentration used - 501 ppm

NO, N02, NOx

nh3

500 ppm - Actual concentration used - 494 ppm

NO, N02, NOx

Hydrocarbon Mixture

~ 500 ppm Cl3 -100 ppm C2, ~ 50 ppm C3 and C4
Hydrocarbon concentration used - 590 ppm

NO, N02, NOx

S02 and NO

250 ppm each - Actual concentration used -
451 ppm S02 & 393 ppm NO

NO, N02, NOx

Analyzer zeroed first and recorded

Interferant gas supplied, analyzer stabilized and analyte concentrations recorded (6 readings)

Flow Rate Sensitivity (9 readings) - Not Observed

Type of flow measuring device: automated bubble flowmeter, rotameter, or other
Ambient atmosphere and ambient flow rate recorded

Zero gas provided and recorded, span gas provided and recorded, zero provided again and recorded
Adjust pressure in system to +10" of water, record flow rate, repeat zero, span and zero
Adjust pressure in system to -10" of water, record, flow rate, repeat zero, span and zero

A leak was detected during the running of the flow rate test for the first vendor. The data sheets reflect
this and also indicate a correction factor would be used in the calculations. The correction factor was
based on the O^ value recorded on the vendor's monitors.	


-------
ETV AMS Pilot
N0/N02 Checklist
Page 6

Ambient Temperature (12 readings)

Room Temperature recorded (assumed to be above 45°F and below 105°F)

Zero and Span and Zero done at each temperature

Instrument allowed to equilibrate to chamber temperature for 1 hour

The ambient temperature test was observed. Room temperature readings were done first. Then the
monitors were placed in a heated chamber at 105°F at 13:24 and first readings were at 14:45. The
chamber door had to remain slightly aiar to keep the temperature constant. Next the monitors were

placed in the cooling chamber which was a household refrigerator. The heated monitors kept	

overheating the refrigerator. After the initial hour to equilibrate the monitors, the refrigerator	

temperature was at 110° F. To obtain the cooled ambient temperature needed for the test, the monitors
were placed out the laboratory window onto the adjacent roof for an hour and were brought to 47°F.

The cooling chamber test readings were taken from 6:38 pm to 6:43 pm.	

This showed great ingenuity of the laboratory test staff to obtain the required ambient conditions for the
test.	


-------
ETV AMS Pilot
N0/N02 Checklist
Page 7

Day Three and Four - Source Tests
Method 7E

Measurement System Performance - Chemiluminescent Monitors

Thermo Environmental Instruments Model 10 and Model 14 A. Data were recorded off a voltage
meter attached to each instrument and voltage readings were then converted to concentrations.
The Fluke voltage meter attached to Model 14A was calibrated 11/2/98 and the one attached to
Model 10 was calibrated 11/3/98.	

Zero Drift: <± 3% of the span over the period of each run
Calibration Drift: <± 3% of the span over the period of each run
Interference Check: <±7%

Measurement System Specifications:

A documented deviation cited changes to the sample probe and lines initially indicated to be allowed by
EPA Method CTM-022 but later revised per July 16. 1999 letter from Battelle indicating the changed
was based on Battelle's own experience with the sources used in the laboratory environment.

Sample Probe - Glass, stainless steel, or equivalent
Sample Line - Heated stainless steel or Teflon tubing
Sample Transport Lines - Stainless Steel or Teflon tubing
Calibration Valve Assembly - 3-way valve assembly or equivalent

Moisture Removal System - refrigerator-type condenser or similar device(?) - Ice Chest was used.
Particulate Filter - borosilicate or quartz glass wool or glass fiber mat, non-reactive with NOx,

in-stack or heated out-of-stack
Sample Pump - Leak free pump of any non-reactive material
Sample Flow Rate Control - control valve and rotameter or equivalent
Sample Gas Manifold - any non-reactive material

Data Recorder - strip chart recorder, analog computer or digital recorder;
resolution shall be 0.5% of span

A data recorder was not used. The test data was recorded on log sheets, one filled out by the

vendor on his monitors and one filled out by source laboratory operator for the reference	

monitors. Calibrations prior to VT are recorded in a bound notebook. Entries are also made to
indicate the date and times the VTs in the source laboratory were run.	

Sampling:

Measurements obtained only after twice the response time has elapsed

Zero and Calibration Drift tests performed immediately preceding and following every run

Adding zero gas & calibration gas (closely approximates the source) at calibration valve
Sampling continues only when zero and calibration drift are within specifications

Emission Calculations: - No calculations were observed

Concentrations are: avg readings (initial & final sampling system bias checks are averaged)
adjusted for the zero and upscale sampling system bias checks


-------
ETV AMS Pilot
N0/N02 Checklist
Page 8

Relative Accuracy Tests
Low NOx Sources

Gas Cooktop: NO and N02 ranges 1-9 ppm

Must operate continuously during test (can't cycle off)
Must operate at steady-state (See Page 8 or 9)

Condition/Specification

Comments

Analyzers (two each) zeroed and span checked
initially only

/- Span was 20 ppm NO and 10 ppm N02

Sampling probes of analyzers placed beside
reference method probe

/- Lines to instruments are then connected into a
metal tube to top of stove top.

Analyzers are allowed to stabilize

/

After initial readings, probes are switched to
ambient air and stabilized

/

Sample Probes are returned to source for a total
of nine samplings

/

Final zero and span check conducted on analyzer
after each source, using the same span as initial
check

/

The cooktop used in the VT has been used by Battelle in a previous study. The data on the source

levels generated by the cooktop are documented in a Battelle report entitled "An Interlaboratory	

Program to Validate a Protocol for the Measurement of MX Emissions from Rangetop Burners." The
data covers 1994 through 1998.	

The gas supply for the cooktop is from a certified cylinder without sulfur.


-------
ETV AMS Pilot
N0/N02 Checklist
Page 9

Water Heater: NO and N02 ranges 10-80 ppm

Must operate continuously during test (can't cycle off)
Must operate at steady-state (See Page 8 or 9)

Condition/Specification

Comments

Analyzers (two each) zeroed and span checked
initially only

/ - Span was 100 ppm NO and 15 ppm N02

Sampling probes of analyzers placed beside
reference method probe

/ - connect in a "T" together

Analyzers are allowed to stabilize

/

After initial readings, probes are switched to
ambient air and stabilized

/

Sample Probes are returned to source for a total
of nine samplings

/

Final zero and span check conducted on analyzer
after each source, using the same span as initial
check

/

The water heater used in the VT has been used by Battelle in a previous study. The data on the source
levels generated by the water heater is documented in a Battelle report entitled "An Interlaboratorv
Study to Determine the Precision of an Emission Measurement Protocol for Residential Gas Water
Heaters." The data covers 1994 through 1998.	

The gas supply for the water heater was from the city gas supply. However. Battelle has a gas	

chromatograph monitoring the concentration of the gas daily.	


-------
ETV AMS Pilot
N0/N02 Checklist
Page 10

Medium NOx Source

First Diesel Generator: NO and N02 ranges 100-1000 ppm NOx
Must operate at steady-state

Condition/Specification

Comments

Analyzers (two each) zeroed and span checked
initially only

/ - Generator was run at high RPM
Span was 200 ppm NO and 50 ppm N02

Sampling probes of analyzers placed beside
reference method probe

/

Analyzers are allowed to stabilize

/

After initial readings, probes are switched to
ambient air and stabilized

/

Sample Probes are returned to source for a total of
nine samplings

/- initial sampling observed only, auditors
departed

Analyzers are evaluated at three separate load
conditions per generator



Extended sampling interval (one hour) is
conducted during one load condition

See Note Below

Final zero and span check conducted on analyzer
after each source, using the same span as initial
check



Note: The Test/OA Plan called for two specific generators from the Air Force that were unavailable at
the time of the VT. A generator on-site was modified to be both the medium and high source. This
generator was run at a high RPM for the medium source and at idle for the high source. Because of the
noise level at the high RPM. most of the extended sampling interval (one hour) was done during the
high source test and not the medium source. One vendor chose to not submit its monitors to the high
source so its extended sampling interval was done during the medium source (high RPM).	

Steady-State:

Temperature changes in the center position of the exhaust of not more than ±10°F;

NOx changes at the center of the exhaust duct of < ±5% relative to the mean over the 15 minute
interval as determined using the EPA reference method

02 changes, at the center of the exhaust duct of < ±0.50% absolute (±5000 ppm) from the mean
sampled over the 15 minute interval.


-------
ETV AMS Pilot
N0/N02 Checklist
Page 11

High NOx Source - Not Observed

Second Diesel Generator: NO and N02 ranges 600-2300 ppm NOx
Must operate at steady-state

Condition/Specification

Comments

Analyzers (two each) zeroed and span checked
initially only



Sampling probes of analyzers placed beside
reference method probe



Analyzers are allowed to stabilize



After initial readings, probes are switched to
ambient air and stabilized



Sample Probes are returned to source for a total of
nine samplings



Analyzers are evaluated at three separate load
conditions per generator



Extended sampling interval (one hour) is
conducted during one load condition



Final zero and span check conducted on analyzer
after each source, using the same span as initial
check



Note: Instead of a second generator, the generator was run at idle to produce a span of 400 ppm NO
and 100 ppm MX.	

Steady-State:

Temperature changes in the center position of the exhaust of not more than ±10°F;

NOx changes at the center of the exhaust duct of < ±5% relative to the mean over the 15 minute
interval as determined using the EPA reference method

02 changes, at the center of the exhaust duct of < ±0.50% absolute (±5000 ppm) from the mean
sampled over the 15 minute interval.


-------