EPA-650/4-75-011
February 1975
Environmental Monitoring Series
m
o
\
*L PRO
\
HI
i?^^^:!^§§^-
^^^^^|::^|^^
ll^^S^^Jt^S^!::
-------
EPA-650/4-75-011
COLLABORATIVE TEST
OF THE CONTINUOUS COLORIMETRIC
METHOD FOR MEASUREMENT
OF NITROGEN DIOXIDE IN AMBIENT AIR
by
Paul C. Constant Jr. . Michael C. Sharp
and George W. Scheil
Midwest Research Institute
425 Volker Boulevard
Kansas City, Missouri 64110
Contract No. 68-02-1363
ROAPNo. 26AAF
Program Element No. 1HA327
EPA Project Office: John H. Margeson
Quality Assurance and Environmental Monitoring Laboratory
National Environmental Research Center
Research Tiangle Park, North Carolina 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
WASHINGTON, D. C. 20460
February 1975
-------
EPA REVIEW NOTICE
This report has been reviewed by the National Environmental Research
Center - Research Triangle Park, Office of Research and Development,
EPA, and approved for publication. Approval does not signify tMt the
contents necessarily reflect the views and policies of the Environmental
Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environ-
mental Protection Agency, have been grouped into series. These broad
categories were established to facilitate further development and applica-
tion of environmental technology. Elimination of traditional grouping was
consciously planned to foster technology transfer and maximum interface
in related fields. These series are:
1. ENVIRONMENTAL HEALTH EFFECTS RESEARCH
2. ENVIRONMENTAL PROTECTION TECHNOLOGY
3. ECOLOGICAL RESEARCH
4. ENVIRONMENTAL MONITORING
5. SOCIOECONOMIC ENVIRONMENTAL STUDIES
6. SCIENTIFIC AND TECHNICAL ASSESSMENT REPORTS
9. MISCELLANEOUS
This report has been assigned to the ENVIRONMENTAL MONITORING
series. This series describes research conducted to develop new or
improved methods and instrumentation for the identification and quanti-
fication of environmental pollutants at the lowest conceivably significant
concentrations. It also includes studies to determine the ambient concen-
trations of pollutants in the environment and/or the variance of pollutants
as a function of time or meteorological factors.
This document is available to the public for sale through the National
Technical Information Service, Springfield, Virginia 22161.
Publication No. EPA-650/4-75-011
11
-------
FOREWORD
This program, "Collaborative Testing of Methods for Measurement of
N02 in Ambient Air," is being conducted under the Environmental Protection
Agency (EPA)Contract No. 68-02-1363, which is Midwest Research Institute's
(MRl's) Project No. 3823-C. The program is concerned with the evaluation
of the following four methods with regard to their precision and accuracy:
1. Sodium-Arsenite,
2. TGS-ANSA,
3. Continuous-Colorimetric, and
4. Chemiluminescence.
The collaborative study covered by this report is of the continuous-
colorimetrie procedure, which is a tentative instrumental method. In
summary, MRl's responsibility was to develop an NO2 ambient-air sampling
system for use with the four methods, provide the test site and facilities
thereon where the collaborative tests would be conducted, select the
collaborators with regard to the program, prepare a plan of test for the
collaborative test, schedule testing, coordinate the test, retrieve field
data and results from the collaborators, statistically analyze their
results, and report its findings to EPA. The 10 collaborators who par-
ticipated in the continuous-colorimetric collaborative test are:
Mr. Ken Smith
Michigan Department of
Natural Resources
Stevens T. Mason Building
Lansing, Michigan 48926
Mr. Lynn Hutchinson
Kennecott Copper Corporation
Post Office Box 11299
Salt Lake City, Utah 84111
Mr. Harold Davis
Air Pollution Control District
of Jefferson County
400 Reynolds Building
2500 South Third Street
Louisville, Kentucky 40208
Mr. Glenn Smith
Kansas City Air Pollution
Control Laboratory
Two Northeast 32nd Street
Kansas City, Missouri 64116
iii
-------
Mr. John Higuchi
Air Pollution Control District
County of Los Angeles
434 South San Pedro Street
Los Angeles, California 90013
Mr. Norman J. Lewis
New Jersey Department
of Environmental Protection
Division of Environmental Quality
John Fitch Plaza
Post Office Box 2807
Trenton, New Jersey 08625
Mr. Cleveland Dodge
Nassau County Department of Health
Division of Laboratories and
Research
209 Main Street
Hemstead, New York 11550
Mr. Rolf E. Doebbeling
State of Utah
Department of Social Services
Division of Health
44 Medical Drive
Salt Lake City, Utah 84113
Mr. Cole McKinney
Air Pollution Control District
of Jefferson County
400 Reynolds Building
2500 South Third Street
Louisville, Kentucky 40208
Mr. Larry Saad
Wayne County Department of Health
Air Pollution Control Division
1311 East Jefferson
Detroit, Michigan 48207
This report of test summarizes MRl's and the collaborators' activities.
It describes the development of the N02, ambient-air sampling system, which
covers the general concept of the system, design considerations, the design
of the system and the system checkout. Following this, there are discuss-
ions on the test site, the selection of collaborators, the formal statis-
tical design including the presentation of factors and parameters that
were considered, the collaborators' field sampling at the test site, the
collaborators' analysis results, MRl's statistical analyses of the col-
laborators' results, conclusions and recommendations. Appendices contain
a copy of the tentative, continuous-colorimetric method, information on
the permeation tubes prepared for this program by the National Bureau of
Standards, written communiques with collaborators, instructions for
collaborators, and MRl's field, operational, and data-log sheets.
These individuals named above with the collaborating organizations
are acknowledged for their excellent work in the continuous-colorimetric
collaborative test.
Special acknowledgements are made to the National Bureau of Standards
and to Mr. Ernest E. Hughes and Dr. John K. Taylor of NBS who provided the
N02 permeation tubes for this collaborative test; and to Dr. John B. Clements,
Chief, Methods Standardization and Performance Evaluation Branch, National
Environmental Research Center, Environmental Protection Agency, and
IV
-------
Mr. John H. Margeson, Government Project Officer, Methods Standardization
and Performance Evaluation Branch, for their valuable suggestions in plan-
ning and design.
This MRI collaborative program is being conducted under the manage-
ment and technical supervision of Mr. Paul C. Constant, Jr., Head,
Environmental Measurements Section of MRl's Physical Sciences Division,
who is the program manager. Those who contributed to this test are:
development of the N(>2 , ambient-air sampling system - Dr. Chatten Cowherd,
Jr., Mr. Fred Bergman, Mr. Emile Baladi, and Mr. Wallace Yocum; experimental
design and statistical analysis - Mr. Michael C. Sharp; and preparation
and operation of test facilities - Dr. George W. Scheil, Mr. John LaShelle,
Mr. Donald Gushing, and Mr. Edward Cartwright, Jr.
Approved for:
DWEST RESEARCH
H. Mr* HubbardT Director
Physical Sciences Divisiot
14 May 1975
-------
CONTENTS
Page
Foreword iii
List of Figures viii
List of Tables x
Summary L
Introduction 3
NO-, Ambient-Air Sampling System 5
General Concept 5
Design Factors 6
System Design 8
System Checkout 19
Ambient Levels of NO and N02 22
Subsystems and Units 22
System Operation 23
Test Site 25
Selection of Collaborators 31
Statistical Design 33
General Considerations and Comments 33
The Design 35
vi
-------
CONTENTS (Concluded)
Page
Collaborators' Field Sampling 37
Collaborators' Sampling Results 41
Statistical Analysis of Collaborators' Results 41
Analysis of All Spiked Readings (Precision Estimates) . . 46
Analysis of Spiked-Ambient Readings (Bias Estimates) . . 50
Summary Discussion of Statistical Analysis 51
Lower Detectable Limit (LDL) 54
Conclusions 55
Recommendations 57
Appendix A - Tentative Method for the Determination of Nitrogen
Dioxide in the Atmosphere (Continuous-Colorimetrie
Procedure) 59
Appendix B - Data on the Permeation Tubes Used as the Source of
the Spiked Levels of N02 77
Appendix C - Calibration of the Venturi and Dry-Gas Meter .... 79
Appendix D - Written Communications with Potential Collaborators . 83
Appendix E - General Analysis of Variance 87
Appendix F - Instructions for Collaborators N(>2 Collaborative Test:
Continuous-Colorimetric Procedure 95
Appendix G - N(>2 Ambient-Air Sampling System Operation Data: Test
Log Sheets with Field Operational Data .... 101
Appendix H - Collaborators Comments Ill
Appendix I - Field Data 115
vii
-------
FIGURES
No. Page
1 NO , Ambient-Air Sampling System Concept 7
2 Final Design of the NO., Ambient-Air Sampling System . . 9
3 Annotated Photographs of the NO., Ambient-Air Sampling
System in Operation 11
4 Ambient-Air Stream Splitter 14
5 Photographs of the N02 Bleed-In-Unit—Assembled and
Disassembled 16
6 Schematic Drawing of the N0« Permeation Tube Assembly . . 18
7 Schematic Drawing and Photographs of the Diffuser ... 20
8 Schematic Drawing and Photographs of the Sampling
Manifold 21
9 Collaborative Test Site: MRl's Field Station .... 26
10 Test Facilities and Collaborators Instruments .... 27
11 Collaborators' Sampling Areas at the Test Site .... 28
12 Photograph of Field Personnel of the N02 Collaborative
Test of the Continuous-Colorimetric Procedure, MRI
Field Station, 29 July to 2 August 1974 38
viii
-------
FIGURES (Concluded)
No. page
13 Collaborator-Level Interaction (w/o collaborator G) .... 49
14 Collaborator-Level Interaction in % Bias (w/o collaborator G) . 53
IX
-------
TABLES
No. Sfie
1 Collaborative-Test Schedule ......... 39
2 Hourly Average Results of Collaborators from their
Sampling N02 at Level 1 (102 ug/m3) ...... 42
3 Hourly Average Results of Collaborators from their
Sampling N02 at Level 2 (228 ug/m3) ...... 43
9
4 Hourly Average Results of Collaborators from their
Sampling N02 at Level 3 (187 ug/m3) ...... 44
5 Hourly Average Results of Collaborators from their
Sampling N02 at Level 4 (47.1 ug/m ) ...... 45
6 Analysis of Variance Spiked Readings ......
7 Collaborator Average Differences Spiked Readings
(Ug/m3)
3
8 Components of Variance Spiked Readings (ug/m ) . . . 48
Analysis of Variance Biasses (ug/m3) 52
o
10 Collaborator (Average) Biasses (ug/m ) Versus Level . 52
1-1 Level 1 Test Data ............. 117
1-2 Level 2 Test Data ............. 118
1-3 Level 3 Test Data .............
1 70
1-4 Level 4 Test Data .............
-------
SUMMARY
•A collaborative .test was conducted -by MRI in .the -Greater Kansas City
Area during the week of 29 July to. 2.August -1:974. Ten .collaborators
participated in this-test of the "Tentative'Method :for.the Determination
of Nitrogen Dioxide in the Atmosphere (Continuous-.Colorimetric 'Procedure).'
All collaborators sampled from the NO^t -'ambient-air-sampling -system that
was developed by Midwest Research Institute-specifically ..for-this .col-
laborative test program. For each of the ".four test .'days, a different
average NC>2 challenge (spike) level was used: -4"7.'l, .102, 187, and
288 ug/m . These levels were obtained .from .-permeation tubes-that were
developed by the National Bureau of Standards.
The collaborators-sampled from both .the'spiked-and unspiked (ambient)
lines of the 'NC^,^ambient-air sampling system,;providing three • sets of
collaborators' results. The first .set :of data .-per 24-hr period (a test
day) comprised results where all 10 collaborators'sampled-.from the spiked
line for approximately 14 hr (1800 to 0800). The second set of data per
test day comprised results collected for approximately 3 hr and 20 min
(0930 to 1250) by the collaborators when they-were.divided into-two groups
of five collaborators each, with one rgroup sampling -from the spiked line
while the other group sampled from the unspiked line. .The third set of
data per test -day would be collected for approximately 3 hr and 40 min
(1250 to 1630) with the two groups interchanging sampling lines. 'These
12 sets of results were used for determining ..the-bias .and-precision of
the method.
In-general, the relative measurement errors are stable over the
range of N02 measured (approximately 50 to-400 .ug/m3) and are not very
large (approximately 6% true value). The..collaborator-rcollaborator
relative standard error is also fairly .stable but larger (CTC.^ 12% true
value and Io| + o| ~ 13% true value).
-------
However, the bias is not stable with respect to N02 level, and is
not consistent within collaborators either. Although the overall average
bias is only about +10%, individual collaborators produced biasses as
great as +80% (at some levels). Thus, it is fair to say that the
continuous-colorimetric method may produce extremely inaccurate read-
ings in an unpredictable fashion (even though the overall average
results are fairly accurate).
About half of the collaborators did achieve fairly stable results
throughout the experiment. A subjective interpretation of this fact
is that the continuous-colorimetric method is difficult to use, but
will produce reliable results in some hands.
Two methods of estimating the lower detectable limit (LDL) were
used. From the results of these calculations, it is reasonable to
state that the LDL within a collaborator is probably s 13 ug/m , and
the LDL from a set of collaborators £ 19 ug/m .
-------
INTRODUCTION
The Methods Standardization and Performance Evaluation Branch,
National Environmental Research Center of the Environmental Protection
Agency (EPA) is engaged in a program to evaluate four methods for
measuring NO2 in ambient air. Midwest Research Institute (MRI) is work-
ing for EPA under Contract No. 68-02-1363 to provide EPA data on the
precision and bias of the following four methods: sodium-arsenite and
TGS-ANSA, which are manual methods, and continuous-colorimetric and
chemiluminescent, which are instrumental methods.
To achieve this objective, a collaborative testing program is
being conducted that will assess interlaboratory as well as intralab-
oratory variation. In summary, MRI in the execution of this program,
selects the collaborators, provides sampling locations and facilities
thereon, orients the collaborators relative to the program, prepares
a plan of test for each method tested, schedules testing, coordinates
the collaborative tests, retrieves field data and results of the col-
laborators' chemical analyses of their field samples, statistically
analyzes results received from the collaborators, and reports results
of the program to EPA.
These activities were performed by MRI on its third test under-
taken on the contract. The method investigated was the "Tentative
Method for the Determination of Nitrogen Dioxide in the Atmosphere
(Continuous-Colorimetric Procedure)," dated June 1974. A copy of the
write-up of this method is given in Appendix A.
The program was initiated on 30 June 1973, and this collaborative
test took place at MRl's field station in Kansas City, Missouri, during
29 July to 2 August 1974, with 10 different collaborators. The interim
period was devoted to the preparation for this test and conduction of
the first two collaborative tests, which covered the sodium-arsenite
and TGS-ANSA procedures. A major task of the preparation activity was
the development of a precision NC>2 , ambient-air sampling system that
could be housed indoors and be suitable for all four methods.
-------
The two major phases of the test program were sampling and analysis.
The sampling phase covered the field test where the collaborators ob-
tained continuous analog NC>2 readings from the ambient-air sampling
system. The analysis phase covered the calculations of average hourly
N02 levels from the collaborators' recorder charts and the statistical
analyses of their results by MRI. After the field test, the collabor-
ators returned to their home laboratories where they analyzed their
recorder charts and reported their results to MRI. Then MRI performed
its statistical analysis and prepared this report of the continuous-
colorimetric collaborative test.
This report covers the collaborative test of the tentative contin-
uous -colorimetric method in the following order: the second section
discusses the NC^, ambient-air sampling system MRI developed for this
program, covering the general concept of the system, the design con-
siderations, the system design, and the system checkout. The third
section describes the test site and the facilities that were used at
this site. The fourth discusses how the collaborators were selected
and who they are. The fifth section presents the factors and param-
eters that were considered in the formal experimental design as well as
the formal design. The sixth section summarizes the test activities
during the collaborative test. The seventh section discusses the
analyses that were performed by the collaborators. The collaborators'
results are presented in this section on MRI's test data. The eighth
section discusses the statistical analysis of the collaborators'
results and presents the results from this analysis, which includes
biasses and components of variance. The ninth and 10th sections present
conclusions and recommendations, respectively. The appendices contain
a copy of the tentative continuous-colorimetric method, data on the
permeation tubes that were used as the source of N02 in the spiked
section of the sampling system, information concerning the calibration
of the venturi and dry-gas meter, copies of written communiques MRI had
with the collaborators, a copy of the test instructions that were given
to the collaborators, the N02, ambient-air sampling system's operational
data, collaborators' comments, results of MRl's analyses, and additional
statistical-analysis information.
-------
N02, AMBIENT-AIR SAMPLING SYSTEM
GENERAL CONCEPT
Primary requirements for the evaluation of an ambient-air method
by on-site collaborative testing are: (a) that all collaborators
sample the same air, (b) that the samples be representative of ambient
air, and (c) that the concentration of N02 in the samples be accurately
known and controllable over the region of interest. The first require-
ment can be met by using a manifold system with each collaborator taking
samples from a common stream of air. The second and third requirements
are somewhat antagonistic to one another and not as easily solved.
Ideally, these requirements can be met by obtaining actual ambient
samples over a wide range of concentration. However, this approach
would require that each level be obtained at a different location with
the additional requirement of fortuitous weather conditions, since
weather conditions have a strong effect on ambient N02 concentrations.
An additional problem with this approach is that no accepted primary
reference method exists for the analysis of N02 in ambient air.
However, gravimetrically calibrated N02 permeation tubes are avail-
able which generate a stable, precise rate of release of high purity NO 2
over a period of a few years. By using a set of these tubes, different
levels of N02 can be generated by adding the N02 from the permeation
tubes to a stream of air with a known flow rate. Since the test condi-
tions must relate to actual ambient-air conditions, the N02 from the
permeation tube can be added as a known addition or spike to the ambient
air stream. The method under test should show a difference in concentra-
tion between samples of ambient and spiked air equal to the spike level.
To ensure that the N02 concentration of the spiked sample does not exceed
the maximum level of interest--350 ug/m3--and to allow control of the
spiked air N02 level over a reasonably broad range, the average ambient
levels must be well below the lowest N02 concentration to be tested, in
this case 50 ug/m .
-------
To achieve this, the following system is used: outdoor ambient air
is drawn into the sampling system through a single tube, as shown in
Figure 1. The air is divided downstream into two sections--spiked and
unspiked. A controlled flow of ambient air at a specific value exists
in the spiked section. A comparable ambient-air flow exists in the un-
spiked section, but the latter is uncontrolled. Temperature-controlled
permeation tubes provide the source of NC>2 which is injected into the
spiked section at a desired level. The NC>2 is then thoroughly mixed with
the ambient air in a mixing unit—a diffuser. The mixture is then equi-
librated before it reaches the sampling station where the collaborators
sample from identical ports—subjected to the same gas flow (spiked plus
ambient). A continuous monitor is attached to the spiked and unspiked
sampling manifolds to monitor the integrity of the spike. The collabor-
ators sample ambient air simultaneously at an identical sampling manifold
that is at a similar location in the unspiked section. The gas in both
sections is then exhausted to the outdoors.
DESIGN FACTORS
The design of the NC>2» ambient-air sampling system was based on the
following factors:
1. The flow rate of each of the four methods to be tested is approx-
imately 0.2 liter/min, with a maximum of 1 liter/min for some of the
instruments that would be used in the instrumental methods.
2. The sampling period of each instrumental method is preferably
24 hr, but could be less.
3. N02 permeation tubes whose rates are approximately 1 ug/min,
which are furnished by the government, are the source for the spiked
levels of N02- These tubes are to be operated at 25.1°C + 0.2°C.
4. The number of collaborators for each collaborative test is to be 10.
o
5. The N02 range of concern is 50 to 350 ug/m , which is representa-
tive of ambient conditions.
6. There are four different N02 spiked levels: high, low, and two
medium. Each level is maintained throughout the run's period, within the
accuracy of the system.
7. The test period is to be no more than 6 days, which is based
upon the concensus of potential collaborators surveyed.
8. The overall N02 sampling system accuracy is to be 5% or better.
9. The flow control in the spiked section is to be 2% or better.
-------
AMBIENT AIR INTAKE
CENTRIFUGAL
BLOWER
ROOF
LLJ
PURGE LINE
SAMPLING
MANIFOLD
SAMPLE
DRAW-OFF
CONTROL
VALVE
EXHAUST
EQUILIBRATION SAMPLE
•>
NO/NO2
MONITOR
RECORDER
VENTURI SECTION^ DRAW-OFF / VALVE
SAMPLING
MANIFOLD
PERMEATION
TUBES
CARRIER
GAS
EXHAUST
PURGE LINE
Figure 1. N02, ambient-air sampling system concept.
-------
10. Flow parameters of the spiked section are to be measured.
11. One N02/NO chemiluminescent device, switched between spiked and
unspiked sampling manifolds (or stations), is to be used as a monitoring
instrument.
12. Only one person from each collaborators' organization will be
needed in the field for each method.
13. There is turbulent flow in the spiked section between the
point of injection of the spiked levels of N02 and the diffuser to pro-
vide mixing of the spiked N02 with the ambient air. The diffuser insures
proper mixing. Up to 20% of the stream in each section—spiked and
ambient air--can be sampled to (a) insure that there is capacity in the
main stream to provide each collaborator with his needs in case there is
a problem with one or more collaborators drawing an excess amount, and
(b) allow the quantity of spiked flow to be drawn from the center of the
spiked line where there is assurance of equilibration. There is to be
a minimum amount of adsorption of the spiked N02 on surfaces, from its
source to and including the sampling manifold. By using Teflon or glass
as the material in which the gases come in contact and by maintaining a
high gas flow rate,which allows for extremely short residence times,
adsorptivity of N02 on surfaces and reaction with water vapor and other
losses are insignificant.
14. Each section—spiked and unspiked--is to be similar, including
material and geometric aspects.
15. Each section is to be under positive pressure so that no un-
wanted air will be pulled into the system in case there was a leak.
16. Collaborator's equipment size, configuration and power require-
ments must be met.
17. Environmental effects on operation of sampling system must be
considered.
SYSTEM DESIGN
The final design of the N02, ambient-air system is shown in a gen-
eral schematic form in Figure 2. Annotated photographs of this opera-
tional system are given in Figure 3.
The input to the system is located outdoors about 2 m above ground
level and approximately 30 m from the building. A valve at the intake
of the 2-in. aluminum tubing provides resistance to the flow of ambient
-------
Equilibration 4 5-Port
Section Sampling
Manifold
NO2 Permeation
Tube System
Monitoring Points:
1. Flow Temperature
2. Flow Pressure at Input
to Flow Meter
3. Ambient Air Flow
Flow Tempcra'ure
6. Carrier Gas Flow
7. NO2 Flow Temperature
8. Port Pressure
9. Port Pressure
10. NO2 & NO
5. Pressure Drop of Venturi & Temperature 11. NC>2 & NO
of Pressure Transducer
Notes:
1. Component within DcsKed Area
Made of Teflon
2. Piping Out Sice Dos'ied Area made
of Aluminum
3. Venturis made of Stainless Steel
4. Spiked & Unspiked Lines Symmetric with
Respect to Geometry & Material
Figure 2. Final design of the N02, ambient-air sampling system.
-------
PAGE NOT
AVAILABLE
DIGITALLY
-------
air to keep the Model 8700 DMP "Tornado" blower at a stable revolutions
per minute, and to serve as a gross flow control. A Variac inside the
building serves as an operational flow control. A blower is located at
the input end of the system to provide positive pressure in the system.
It is located outdoors to keep out the intensive noise it generates and
is housed as shown in Photographs 1 and 3 of Figure 3 to protect it from
the elements.
The line from the blower to the splitter is 2-in. diameter, aluminum
pipe. It is sufficiently long to serve as a trap for any excess moisture
and to bring the ambient air to room temperature. The splitter is also
made of aluminum. This splitter, shown in Figure 4, reduces large-scale
turbulance from the blower and divides the ambient air stream between
the spiked and unspiked 1-in. diameter, aluminum lines. A controlled
flow goes to the venturi where the air flow in the spiked line is con-
tinuously measured and recorded. This flow is determined by the follow-
ing equation:
method sampling rate (number of samples x number
Flow in liters/min = of collaborators + monitor number + purge number)
percent flow drawn through sampling manifold
0.2 liters/min x (4 samples x 10 collaborators +
1 NO/NC>2 monitor + purge-line flow)
percent flow drawn through sampling manifold
= 0.2 (4 x 10 + 1 + 4) _9 = 6Q
0.15 0.15
The monitor number and the purge number are flows attributable to the
continuous monitor and the purge line of the system, respectively. The
flow on each line—the spiked and the unspiked—is turbulent—Reynolds
No. > 2,100 --with the Reynolds number being
3
= Q = Q liter/min x 1.000 cm /liter
ffVD 0.785 x 0.15 cm2/sec x D sec x 60 sec/min
= 1.000 Q = 1.000 x 60 = 4,000
7.065 D 7.065 x 2.1
Since the spiked and unspiked sections are identical except that the
spiked section also contains the monitoring points 1, 2, 3, 4, and 5
identified in Figure 2 and the N02 permeation tube system, only the
spiked section will be discussed.
From the splitter, the spiked line connects to a Singer AL-175 dry-
gas meter, which is made by the American Meter Company. (See Photograph 9
13
-------
Ambient Air
Figure 4. Ambient-air stream splitter.
-------
of Figure 3.) This flow meter has a pressure drop of 10 mm of water and
is temperature-compensated. Thus, only the gas pressure is measured to
correct the flow readings to obtain the true flow rate of ambient air
delivered during a test run. This flow rate is determined hourly by measur-
ing the time required for a known quantity of air to pass through the meter.
The output of the flow meter is connected, as shown in Photographs 7
and 9 of Figure 3, to a stainless steel venturi, which was designed for
a flow of 60 liters/min. This venturi is used as a general flow control
device, and provides a continuous record of flow rate using a strain gage
pressure transducer and thermocouples--see Point 5 of Figure 3(A). Both
the pressure drop of the venturi and the temperature of the pressure
transducer are recorded on analog recorders. Control of the flow rate is
handled by monitoring the venturi pressure drop. When the value deviates
from a reference value, 60 liters/min, the flow rate can be changed ap-
propriately by making an appropriate adjustment of the Variac control to
the blower.
The flow temperature measurement (Point 4 of Figure 3(A)) is actually
the gas-flow temperature at the output of the gas meter and at the input
to the venturi, since those two units are physically close together (about
12.5 cm apart). Tests have shown that the temperature at this point is
identical with the temperature at the gas flow meter inlet. The gas tem-
perature at this point is normally within 0.5°C of room temperature. This
temperature measurement is used to obtain accurate gas-flow values.
To provide more accuracy, the thermocouples at Points 2 and 4 of
Figure 3(A) were replaced for this test by a 0 to 50°C bimetallic dial
thermometer that is located at Point 4 of Figure 3(A).
The output of the venturi is a few centimeters from the input of the
N02 bleed-in unit as shown in Photograph 9 of Figure 3. These two units
are connected by 1.0-in. diameter, aluminum tubing. From the input of
the NC>2 bleed-in unit through the sampling manifold, the system is made
of Teflon.
The N02 bleed-in unit, as shown in Figure 3(A) and Photographs 7
and 9 of Figure 3, receives ambient air from the venturi and a level of
NO2 (a spike) from the N02-permeation tube assembly (see Figures (A) and
(B), and Photographs 8 a-c of Figure 3). Detailed photographs of this
bleed-in unit are given in Figure 5. Photograph 1 of Figure 5 is a
closeup showing the assembled Teflon unit with its metal holding/mounting
plates. The gas stream, or ambient air, enters the opening to the right
and passes through the unit, mixing with the spiked level of NO2 which
exits through the tapered smaller tubing shown as concentric to the out-
put of the bleed-in unit at the left of Photograph 1 of Figure 5.
15
-------
Photo 1 - Detail of N02 Bleed-In Unit with Vertical
Tube from Permeation Assembly, Chamber with Central
Tapered Pickup Tube and Stainless Steel Mounting
Components.
Photo 2 - Close-Up Showing Machined Chamber with
Pickup for Bleed-In in Place.
Figure 5. Photographs of the N02 bleed-in unit-
assembled and disassembled.
16
-------
The vertical tube of this bleed-in unit accepts the N02 gas from the
permeation tube assembly. This spiked gas flows downward through this
tube, which is inside the unit (see Photograph 2 of Figure 5), and after
a short run, mixes with the ambient air as stated before.
The N02-permeation system is shown in Figure 6 and Photograph 8 a-c of
Figure 3. Details of the system are given in the captions of these
photographs. The nitrogen carrier gas is used to flush the N02 into the
system. It is passed through a charcoal and soda-lime scrubber before
it is delivered to the N02 permeation tubes. Also, the flow is set by
means of control values and rotameters. This flow is monitored during
system operation. The carrier gas is then fed into four separate branches
to achieve different levels of N02. (More detail on the permeation tubes
and their arrangements in the branches is given in Appendix B>) The N02
permeation tubes* are arranged in these four different branches to pro-
vide N02 spike levels of approximately 50, 100, 200, and 300 ug/m3.
Branch 1 has four permeation tubes, Branch 2 has five permeation tubes,
Branch 3 has two permeation tubes, and Branch 4 has two permeation tubes.
An ASTM calibration thermometer (0.1°C or better accuracy) is an integral
part of each permeation tube branch. Each set of permeation tubes is en-
closed in a glass tube which has an inlet for the nitrogen carrier gas and
an outlet for the nitrogen carrier gas/NO2 mixture. These N02 permeation
tube, enclosure units are immersed in a temperature-controlled, water
bath for operating at 25.1°C. If the temperature of this bath were to
vary more than 0.2°C, a correction would be made from the following re-
lationship:
Log r = 0.034857 (273.12 + T) - 10.29198
where T = temperature in °C of the permeation tube environment, and
r = the permeation rate.
Flow meters of the permeation tube assembly that measures the nitro-
gen flow were calibrated by the manufacturer to 1% accuracy. Thermometers
that were used to measure the gas temperature in the permeation tube hold-
ers are ASTM type that are accurate to within 0.1°C. The permeation tubes
used were calibrated by the National Bureau of Standards and checked by
EPA. (See Appendix B.) The entire permeation assembly from the tube
holders to the pickup fitting, where the spiked gas enters the main gas
stream, was checked for leaks with Snoop and found to be airtight.
*"Operation Characteristics of N02 Permeation Device," by Harry L. Rook,
Ernie E. Hughes of NBS, Washington, D.C., and Robert S. Fuerst and
John Margeson of EPA, Research Triangle Park, North Carolina. A
paper presented before the Division of Environmental Chemistry,
American Chemical Society, Los Angeles, California, 31 March to 5 April
1974.
17
-------
Nitrogen Gas
Shuroff Valve
Charcoal & Soda Lime Filter
Control Valves
Rotameters
NO2 Permeation
Tube Holders
Thermonit. lers
Temperature Controlled
Water Bath
LJ
Control Valves
to NC>2 Bleed In Unit on
Spiked Line (See Figure 3)
Figure 6. Schematic drawing oC the N02 pormcation tube assembly.
18
-------
The ambient air and the NC>2 spike flow from the bleed-in unit to the
diffuser where they are well mixed. The diffuser is a few centimeters
downstream from the bleed-in unit, as shown in Photograph 9 of Figure 3.
At the diffuser, shown schematically and in the photograph in Figure 7, the
gases enter the diffuser through the Teflon tubing (Section A of the
schematic drawing of Figure 6), pass into the spiraled tube and through
its angled holes into the space outside the tube. The flow continues
through the holes in the prescreen block, Item D of the schematic, and then
through a series of Teflon screens, Item E.
The homogeneous mixture passes through an equilibration section that
is Teflon tubing 1.0 m long. This section of tubing provides the final
equilibrated concentration. This tubing is connected to the input of the
sampling manifold. (See Photographs 10 and 11 of Figure 3.)
The 45-port sampling manifold is constructed of Teflon except for
its metal plates which are entirely external. Photographs 1 to 3 of
Figure 8, which show external and internal views of the sampling manifold
and a schematic drawing, describe the operation of the manifold. The
stream of the homogeneous mixture of ambient air and a spiked level of
N02 flows through the bottom portion of the manifold, into the exhaust
line. Section A of the manifold is in the pickup tube through which flows
the total volume of gas sampled by the collaborators. The inlet to this
pickup tube is located such that this volume is drawn from the central
portion of the main stream. The sampled volume flows past a mixing im-
peller (B) and then into the main chamber of (C) of the manifold. In
this chamber, the flow is spread evenly to the 45 symmetrically located
exit channels (D). The gas in the main chamber that is not drawn through
the exit channels to the collaborator ports flows out the exhaust duct
or purge line which has a control valve. Both exhaust lines from the
manifold join downstream to form a common exhaust tube, which also con-
tains a flow control valve.
One port of each sampling manifold is used to monitor the pressure
in the sampling manifold to determine if it remains positive (see schematic
drawing of Figure 3). Another port of each manifold is used to monitor the
N02 and NO levels being sampled by the collaborators and to monitor the
integrity of spike during test. A Bendix Model 8101 B chemiluminescence
NO-NC^-NQx analyzer is used for this and is switched between the spiked and
unspiked manifolds. (See Photograph 12 of Figure 3.)
SYSTEM CHECKOUT
Readying the system for the collaborative test comprised three
principal areas of activity: (a) determination of levels of NO and N02,
both ambient and inside the building; (b) checkout of the sampling system,
including monitoring devices and test instrumentation; and (c) checkout of
the sampling system as an operational system. These three areas are
discussed below.
19
-------
I*
OQ
(D
^J
CO
O
O !T
hh (0
3
rt 03
=r rr
(D H-
H. p.
i-h i-l
i-*i B»
C «
en K-
(D 3
H OQ
D)
a
O
rr
O
OQ
i-!
{U
13
CO
Photo 1 - Top View Diffuser Components: Housing, End
Sections, Spiroler Tube, Teflon Screens, Retaining
Rings.
Photo 2 - External View of Diffuser.
sun
Vn
'/A
BHll
Exploded cross section of all-Teflon diffuser with inlet (A),
end section (B), spiraler tube with angled holes (C),
prescreen block with holes (D), five sets of fine mesh
Teflon screen and retaining blocks (E), end section (F),
exit (G) and diffuser housing (H). Double cross-hatched
end plates are stainless steel.
-------
Photo 1 - Sampling Manifold External View.
Photo 2 - Internal View (Right Component is Inverted
in this Photo).
Cross section of all-TFE Teflon manifold with pickup tube (A), mixing impeller (B), main chamber flow
spreader (C), exits to collaborator ports (D), channel to exhaust manifold (E), and manifold exhaust
duct (F). Gas not captured by pickup assembly exhausts at left side of manifold base. Double cross-
hatched assembly plates at top, middle, and bottom are stainless steel.
Photo 3 - Internal View of Manifold Pickup Section
Showing Flow-Spiraling Impeller.
Figure 8. Schematic drawing and photographs of the sampling manifold.
-------
Ambient Levels of NO and N02
Ambient levels of nitrogen oxides at the test site were generally
low, but there were considerable variations at these levels. Since the
test site is located in a rural area south of Kansas City where there is
very little industry, the primary factors that influence NOX levels at
the site are wind speed and wind direction.
During tests of NOX levels using MRl's Bendix Model 8101 B chemi-
luminescence NO-N02-NOX analyzer for 24-hr monitoring, the lowest levels
were found when the wind was from the south. Both NO and NC>2 seldom
exceed 20 ug/m . Periods of more than 1-hr duration were measured when
readings were indistinguishable from the purified zero gas used to cali-
brate the analyzer.
With northerly winds, N02 levels were generally between 30 and 50
ug/nr and NO levels were approximately 10 ug/m . As expected, the
ambient levels followed an inverse relation with respect to wind speed.
The highest daily readings were coincident with the morning and evening
rush hours. These peak levels generally began at about 7:00 a.m. and
again at 5:00 p.m. and lasted between 2 to 4 hr .
The highest recorded levels of NO occurred under calm wind con-
ditions when the light vehicular traffic in the vicinity of the test
station generated levels in excess of 100 ug/m3. NO levels did not
exceed NOo levels at this site.
Over a 24-hr period, average N02 levels were 10 to 50 ug/m^, and
NO levels were of the order 10 to 20 ug/m . During any 24-hr period,
maximum N02 levels were generally several times higher than the minimum
levels. Thus, while N02 levels at the test site are lower than those
at urban, industrial locations, the N02 levels do exhibit the variability
found under normal ambient conditions. Indoor readings were similar
but did not show the sudden changes often found when monitoring outdoor
levels.
Subsystems and Units
The venturi and dry-gas meter were calibrated using a 1.0 ft3/min
wet-test meter. Information concerning the calibration is given in
Appendix C. The entire system was prepared for the test by bringing all
components to normal operating conditions several days prior to the test
and running the system continuously in this mode until the beginning of
22
-------
the test. Water addition to the constant-temperature bath was the only
maintenance required. The temperature variation of the permeation-tube
bath during this time was less than 0.1°C. A check of NOX levels in the
cylinders of prepurified nitrogen carrier gas found no N02 and 40 ug/m3
NO.
The Bendix NOX Analyzer was checked at MRI by a Bendix field repre-
sentative. The difference in spiked and unspiked readings of the Bendix
analyzer agreed within 10% of the calculated spike levels at all four
levels used for the test. The instrument was stable and reliable when
operated continuously at the levels found during normal testing. Checks
with calibration gases reveal that the catalytic converter efficiency does
fall off sharply above 400 ug/m .
The symmetry of the sampling ports was checked in two ways. The
primary way was that the pressure drop at each port was measured under
the normal load of 200 cc/min. This test showed that all ports gave a
pressure drop of 1.5 cm of water +0.5 cm. Such a pressure drop should
have no effect on normal sample flows and the flow rate from the ports
should be identical to that obtained by pulling free room air into the
sampling trains. However, since some of the analyzers used for this test
required flow rates on the order of 1.0 liter/min, 10 larger diameter
Teflon tube connections were provided on each manifold. These larger
diameter Teflon lines were capable of supplying more than 2.0 liters/min
without developing a pressure drop of 1.5 cm of water.
A second way was to connect the N02 monitor to ports of the spiked
and unspiked sampling manifolds and measure the level of N02 in micro-
grams per cubic meter. This was done in two ways: the system under a
load, e.g., a spiked level of approximately 350 ug/m ; and an unloaded
condition where just ambient air was passed through each section—spiked
and unspiked--of the N02 sampling system. In both cases, the NO, monitor
showed no variation between four ports spaced equally around the manifold.
System Operation
Identical materials and dimensions are used on the spiked and un-
spiked sections of the N02 sampling system. Handling and treatment of
all components were also identical.
Flow rates of the spiked and unspiked sides were within 10% of each
other with all dampers open. In normal operation, the exhaust dampers
are adjusted to give a positive pressure of 2 to 4 mm water at the sample
manifolds. Once set, this pressure is stable.
23
-------
The rise and fall times to equilibrium in response to changes in a
spike level were checked. Rise time was less than 15 min and fall time
was less than 5 min (when permeation tubes were disconnected). The fall
time is essentially that of the analyzer response time, allowing for the
purge time of the sample lines. The rise time is longer than the fall
time because of the increased pressure against which the carrier gas
stream must work when a set of permeation tubes are connected. Some flow
reversal in the permeation-tube holders occurs after connection.
Since the response times were essentially limited by flow rates and
instrument response, no observable adsorption effects were noted. Checks
of NOx levels found at the sampling ports agreed, within normal accuracy
limits, with those measured outside the building. At the 50 ug/m level
both readings were within 5 ug/m3 (0.5% of full scale), which is within
the accuracy of the monitoring instrument. Thus, the unspiked samples
at the sample ports accurately reflect ambient levels and the sampling
system may be considered to be inert with respect to
24
-------
TEST SITE
The general criteria one would use in selecting a site include the
ambient level of N02 and variation thereof, general meteorological and
climatological conditions, work facilities for the collaborators (ade-
quate space, facilities, housing, etc.)> cooperation of the organization
furnishing the site, logistic aspects, and local lodging accommodations.
The levels of NC>2 required are those representative of ambient NO2
conditions, which are in the range of a few micrograms per cubic meter
to 350 ug/nr*. These levels could be achieved at one site with a low
level of N(>2 by spiking the ambient air with various levels of NOo in a
manifold sampling system.
MRl's field station (see Figure 9) which is located in a rural area
south of Kansas City, meets all the criteria and was selected as the test
site. The N(>2, ambient-air sampling station is housed in Building 3 shown
in Figure 9. The input to the sampling system is located outside the
building near the roadway (see Photograph 3 of Figure 3).
These test facilities are described in conjunction with the sampling
system. Photographs of the facilities are given in Figure 10. Photo-
graph 3 shows the circular tables that house the sampling manifolds and
the collaborators' sampling trains. Each table—spiked and unspiked--
has a multiplicity of AC power receptacles, with each collaborator having
its own branch of outlets. Each branch has its own circuit breaker and
branch indicator. This arrangement is to protect other collaborators in
case one collaborator has a power failure due to faulty equipment.
Photographs 1 and 2 of Figure 10 give close-up views of some of the
collaborators' trains positioned in their table areas (see Figure 11).
Since each collaborator had only one instrument, Teflon or polypropylene
tubing was run through the 2-in. pipe, which spanned the two test tables,
to the other manifold. This allowed the instruments to remain in one
place during tests and yet sample from within the spiked or unspiked line
by singly switching lines.
25
-------
ro
a-,
DERAMUS FIELD STATION
Figure 9. Collaborative test site: MRl's field station.
-------
Photograph 1. A col-
laborator's instru-
ment in operation
Photograph 2. Col-
laborator preparing
for a test
Photograph 3. Unspiked
sampling line and area
in foreground
Figure 10. Test facilities and collaborators instruments.
27
-------
SPIKED
SAMPLING MANIFOLD
COLLABORATOR
AREAS*
* Unspiked manifold layout similar (see Appendix F).
Figure 11. Collaborators' sampling areas at the test site
28
-------
The windows on the north side of the building were boarded to keep
electromagnetic radiation from entering the building. With this blockage
and a temperature control system in the building, the 25.1'G permeation.
bath was able to be maintained at that temperature throughout the four
24-hr runs with no detectable deviation from the 25.1°C temperature,
except for a few hours when the deviation was 0.1°C.
29
-------
SELECTION OF COLLABORATORS
A principal activity was to compile a list of potential collabora-
tors and from this list select 10 to perform the testing according to
the tentative continuous-colorimetric method. Information was obtained
from EPA (names and addresses of 150 organizations) and from MRl's files
to compile a list of nearly 200 potential collaborators .
A letter was sent to 162 organizations seeking their desire to
participate as a volunteer collaborator on this test and evaluation
program. Attached to this letter was a "Collaborator Form" to be com-
pleted which surveyed their experience with the four methods, methods
they had used, equipment they could make available for the tests, ac-
ceptable length of test period, etc. A second letter was sent to those
who expressed interest in the continuous-colorimetric method after a
test date was selected. A copy of these letters and the collaborator
form are given in Appendix D.
A majority of the responses indicated the desire that a test period
for a method be no more than 6 days.
Nine organizations* were selected for the continuous-colorimetric
collaborative test from those organizations that responded in the affir-
mative to participate in the test. The selection was based upon the
following criteria:
1. Willingness to participate on a volunteer basis.
2. Technical capabilities •
3. Related past experience .
4. Availability .
One organization provided two collaborators, each of which had his
own equipment and worked entirely independent of one another.
31
-------
5. Ability to furnish sampling equipment, instruments, and materials
required to perform the test strictly according to the method ; and
6. Type of organization (industrial, educational, governmental--
local, state, federal—etc.).
The information needed to make the selection based on the above criteria
was obtained from the collaborator forms that were returned, and from
subsequent telephone conversations with the candidate collaborators.
The nine organizations selected as collaborators for the continuous-
colorimetric 'collaborative test were:
Michigan Department of
Natural Resources
Stevens T. Mason Building
Lansing, Michigan 48926
(Mr. Ken Smith)
Kennecott Copper Corporation
P.O. Box 11299
Salt Lake City, Utah 84111
(Mr. Lynn Hutchinson)
Air Pollution Control District
of Jefferson County
400 Reynolds Building
2500 South Third Street
Louisville, Kentucky 40208
(Mr. Harold Davis)
(Mr. Cole McKinney)
Kansas City Air Pollution
Control Laboratory
Two Northeast 32nd Street
Kansas City, Missouri 64116
(Mr. Glenn Smith)
Air Pollution Control District
County of Los Angeles
434 South San Pedro Street
Los Angeles, California 90013
(Mr. John Higuchi)
New Jersey Department of
Environmental Protection
Division of Environmental Quality
John Fitch Plaza
P.O. Box 2807
Trenton, New Jersey 08625
(Mr. Norman J. Lewis)
Nassau County Department of
Health
Division of Laboratories and
Research
209 Main Street
Hemstead, New York 11550
(Mr. Cleveland Dodge)
State of Utah
Department of Social Services
Division of Health
44 Medical Drive
Salt Lake City, Utah 84113
(Mr. Rolf E. Doebbeling)
Wayne County Department of
Health
Air Pollution Control Division
1311 East Jefferson
Detroit, Michigan 48207
(Mr. Larry Saad)
These organizations will be referred to as Collaborators A through
J, without specifying which is A, B, etc., to allow the organization data
to remain anonymous.
32
-------
STATISTICAL DESIGN
GENERAL CONSIDERATIONS AND COMMENTS
The purpose of this collaborative test was to determine the precision
and bias of the continuous-colorimetric method. A major element of the
collaborative test was to have an experimental design that would allow
this purpose to be met. Considerations that formed the bases of this
design, which is given later in this section in a formal manner, are:
1. Challenge (spike) levels of NC>2 ,
2. Ambient levels of N02 ,
3. True values of
4. Sampling time of a run,
5. Test period of the method,
6. Number of collaborators,
7. Number of samples per run,
8. Interferences,
9. Adsorptivity,
10. Sampling ports, and
11. Instrumentation.
Challenge spike level of N02 is an experimental design variate.
Four levels of challenge were selected, based upon the normal range of
33
-------
values found in ambient air on a 24-hr average basis: one low level on
the order of 50 ug/m3 ; two medium levels, one near 100 jag/m 3 and the
second near 200 ug/m3; and one high level of approximately 300 ug/m3.
A challenge level should be steady state, or continuous at a specific
level, plus or minus acceptable deviations—less than + 2%.
Ambient levels should be lower than the lowest challenge (spike)
level (approximately 50 ug/m3). Since the ambient levels are the actual
ambient levels of N02 at the test site, those levels present during the
time of testing may vary from this criteria. (See page 22.) The
ambient levels will be mixed with the challenge levels to provide the
spiked challenges. There will be just ambient challenges which are
identical with the ambient portion of the spiked challenges. The col-
laborators will sample both spiked and ambient challenges (not simul-
taneously—see Appendix D).
For a run, the true value of N02 sampled by the collaborators will
be taken as the N0£ spiked level generated by the permeation tube assembly
plus the average value of the ambient challenges sampled at the same time.
Since not all collaborators participate in estimating this "true" value,
a potential bias is created that adds the overall error in the analysis.
Ten collaborators were deemed to be sufficient to obtain a cross -
section of the population of the type organizations that would be in-
volved in sampling NO, , be within acceptable project costs, and provide
statistical significance with the results.
Adsorptivity is of concern because of the possibility of error in
the NQj level received by the collaborators' sampling devices in contrast
to the known level of the challenge--from both the standpoints of increas-
ing and decreasing the challenge level from run to run. Teflon material
was used from the N0~ bleed-in through the sampling manifold to minimize
if not eliminate the adsorptivity factor. For further assurance, prior
to commencing a run, the challenge could be run for a sufficiently long
period so that all surfaces exposed would have reached a state of equili-
brium with the new concentration. Both aspects were covered; Teflon was
used in the construction and sufficiently long challenges were made to the
system prior to commencing a run.
The port-to-port effect did not need to be incorporated in the ex-
perimental design because results of the evaluation of the N02, ambient-
air sampling system indicated that all ports were identical.
The major considerations with regard to instrumentation for the
continuous-colorimetrie collaborative test were: (a) MRI would only
instruct the collaborators that they are to use the sampling equipment
34
-------
and calibration specified in the method writeup, and (b) MRl's monitoring
instrumentation and test instrumentation used in the calculation of the
N02i ambient-air system was sufficiently reliable and accurate. In both
cases, all requirements were met.
THE DESIGN
Since some spiked readings were being taken throughout the test,
but ambient readings were only sometimes obtained, there were two experi-
mental designs used.
One statistical model applies to all the spiked readings, but does
not incorporate any ambient observations. In this analysis then, all 10
collaborators are used to estimate precisions.
This analysis of variance model is:
» Ci + 'j + \ +
where u = Overall mean,
C± = i* collaborator, i = 1, . . . , 10,
tj = jth hour, j = 1 20,
L = Kth NO level, K = 1 4, and
K> £.
Measurement error in j&tn reading in ijktn cell,
. . .
H = 1 for every ijk.
Since the NO^ level may change from hour to hour, there are no repli-
cates in this framework (£ = 1 always). Also, some cells are missing al-
together (because sometimes a collaborator was on the ambient line and
did not get a spiked reading). Each collaborator measured the spiked
line only (a maximum of) 17 out of the 20 experimental hours. So all
effects have to be "adjusted" for this sample inbalance.
35
-------
Therefore, the general analysis of variance was performed.* (In
practice, four such analyses were performed (one per level) because it
turned out that the repeatibility of the method depended on the N0« level.)
The second experimental design model describes the data set of 6 hr/run
when both ambient and spiked readings were taken (by different collaborators,
of course). Since the hourly variation in ambient N02 is significant, a
"true" value was constructed for each of the 6 hr (per level). That is,
for each hour, the true value was estimated as the spiked N02 amount plus
the average ambient reading in that hour. An individual response is a
bias; i.e., the collaborator's reading minus the true value. Thus, the
data framework becomes three responses per collaborator** per level. Thus,
the bias estimates are based on five collaborators on the spiked line and
the other five collaborators on ambient. Since these groups of five may
be separated in their means, a potential error is introduced into the bias
determination.
* See Appendix E for a discussion of this general analysis.
** Collaborator G not included.
36
-------
COLLABORATORS' FIELD SAMPLING
The collaborative test took place at the MRI Deramus Field Station
during 29 July to 2 August 1974. The 10 collaborators (see Figure 12),
started the test at 0830, 29 July, with an orientation. The N02>
ambient-air sampling system they used was shown and explained to them.
The written instructions that comprise Appendix F were given to and
discussed with the collaborators. After this orientation period the
collaborators set up their equipment in preparation for the first run.
The actual schedule of the four runs that took place is given in Table 1.
All 10 collaborators cleared the site by 1900, Friday evening, 2 August.
All collaborators sampled from the spiked line during the A runs.
During the B and C runs, the collaborators were divided into two groups
with one group sampling on the spiked line while the other group sampled
from the unspiked line. At 1250 each day all collaborators then switched
sample manifolds—those that had been on spiked went to unspiked and
vice versa. Also, the group that was on the spiked line for Run IB
changed to the unspiked line for 2B, so that each group began every other
B + C run set on the same manifold.
During the test, MRI personnel observed that all collaborators followed
the sampling procedures given in the method write-up. All collaborators
made a dynamic calibration and established a static span point at their
home laboratories prior to the test for reference at the site. They used
a static calibration check during the test period. The calibration checks
were made during 0800 to 0930 and 1630 to 1800 each day. Each collaborator
also supplied his own chemicals and prepared his own absorbing solution and
standards to minimize bias.
Each collaborator recorded all pertinent sampling data on his re-
corder chart. The calculations of the N02 levels from the recorder chart
readings were made after returning to their home laboratories.
37
-------
Front row: Paul Constant,!./ Ken Smith, Lynn Hutchinson, Cleveland Dodge,
Rolf Doebbeling, Norman Lewis, Harold Davis.
Back row: John Higuchi, Cole McKinney, Larry Saad, George Scheil,!/
I/
17
Fred Bergman,-' John LaShelle.i' Glenn Smith, John Margeson—
2/
I/ MRI personnel.
21 EPA Project Monitor.
Figure 12. Photograph of field personnel of the NC>2 collaborative
test of the continuous colorimetric procedure, MRI field station,
29 July to 2 August 1974.
38
-------
Table 1. COLLABORATIVE-TEST SCHEDULE
N02 Spike Level Date/Time
Level Run (yg/m3)
1 A 102
B 102
C 102
2 A 288
B 288
C 288
3 A 187 7-31-74 at 1800 8-1-74 at 0800
B 187 8-1-74 at 0930 8-1-74 at 1250
C 187 8-1-74 at 1250 8-1-74 at 1630
4 A 47.1 8-1-74 at 1800 8-2-74 at 0800
B 47.1 8-2-74 at 0930 8-2-74 at 1250
C 47.1 8-2-74 at 1250 8-2-74 at 1630
Started
7-29-74 at 1800
7-30-74 at 0930
7-30-74 at 1250
7-30-74 at 1800
7-31-74 at 0930
7-31-74 at 1250
Completed
7-30-74 at 0800
7-30-74 at 1250
7-30-74 at 1630
7-31-74 at 0800
7-31-74 at 1250
7-31-74 at 1630
39
-------
MRI had a laboratory supervisor who was in charge of the N(>2,
ambient-air system operation. He was on duty from 0800 to 1800 each
day, which was the period-of run starts and completions. He was avail-
able anytime during the 24-hr runs, if any problems arose, as was the
program manager.
There was a technician on duty throughout each run at all times
during the test. These people monitored the sampling system operation,
recording operational data and general observations. A general logbook
was kept as well as the log sheet for operational data. Copies of these
log sheets are given in Appendix G.
40
-------
COLLABORATORS' SAMPLING RESULTS
Each collaborator's sampling instrumentation included an analog
recorder on which all his sampling data was recorded. Each collaborator
calculated 1-hr averages from his analog sampling recordings by a method
of his own choosing. These results were submitted to MRI along with his
calibration data. The 1-hr averages of the collaborators are tabularized
by N02 spiked level* in Tables 2 through 5, with Table 2 comprising
Level 1(102 ug/m3 of N02) results, Table 3 comprising Level 2 (288 ug/m3
of N02) results, Table 4 comprising Level 3 (187 ug/m3 of N02) results,
and Table 5 comprising Level 4 (47.1 ug/m3 of N02) results. Each table
presents the results of the 10 collaborators for a test day. Explanatory
notes are given at the bottom of each table.
MRI checked the collaborators' results for any gross overall error,
e.g., misplacement of the decimal point. Also the data were culled for
statistical outliers. Collaborator G is an "outlier" and the data from
Collaborator G shown in Tables 2 and 4 are so noted. Minor deviations
were attributed to the reading of the analog charts.
The collaborators' comments on the test are given in Appendix H.
The N02 sampling-system data, along with calculated flow rates and
spike levels of the system and data on ambient test conditions, are given
in Appendix I.
STATISTICAL ANALYSIS OF COLLABORATORS' RESULTS
The analysis of the spiked readings and the analysis of the ambient
readings will be discussed separately. A summary discussion will follow.
The level value of N0« is that generated by the permeation tubes.
41
-------
Table 2. HOURLY AVERAGE RESULTS OF COLLABORATORS FROM THEIR SAMPLING N02 AT LEVEL 1
(102 Hg/m3)£/
Collaborator
Run
A
B
C
Date
7-29-74
7-29-74
7-29-74
7-29-74
7-29-74
7-29-74
7-30-74
7-30-74
7-30-74
7-30-74
7-30-74
7-30-74
7-30-74
7-30-74
7-30-74
7-30-74
7-30-74
7-30-74
7-30-74
7-30-74
7-30-74
7-30-74
Time
1800-1900
1900-2000
2000-2100
2100-2200
2200-2300
2300-2400
2400-0100
0100-0200
0200-0300
0300-0400
0400-0500
0500-0600
0600-0700
0700-0800
0930-1000£'
1000-1100
1100-1200
1200-1300
1300-1400
1400-1500
1500-1600
ieoo-1630^
A
113S'
116
133
178
184
188
190
178
160
154
148
137
162
128
Od/
Qd/
109
126
113
113
B
_
-
-
197
205
210
212
197
178
173
169
160
182
178
%d/
gd/
9d/
132
165
132
132
C
_
175
186
235
237
250
259
250
231
220
212
194
218
216
26*-'
15d/
IS*/
158
154
154
D
119
123
152
200
206
207
214
206
196
188
186
176
202
190
20^
IQd/
102
108
101
99
E
128
133
152
195
199
203
210
197
178
167
164
158
175
169
*
122
132
122
120
F
132
132
141
188
197
207
207
197
188
169
169
160
179
179
132
132
122
i9^y
gd_/
Gfc/
201
154
160
220
226
226
226
192
154
135
122
98
109
84
150
141
141
132
u$
6$
H
184
184
196
230
247
256
266
259
240
237
240
230
249
254
116
97
94
97
•i gd/
od/
I
116
120
133
169
177
178
184
175
156
147
143
135
150
143
122
118
113
111
"!'
1 1 *
1 1 *
J
115
124
139
143
147
152
158
143
-
-
-
-
-
-
141
137
132
135
2Ll/
£/
af This is the spiked value--the statistically determined average true value of NO- (which includes ambient
N02) is 112 ug/nr,. (In compiling biasses there is an individual true value per hour).
b_/ Data from collaborator "G" is unreliable at this level.
£/ Indicates reading is for s 1/2 hr.
d/ From unspiked samples—all other results are spiked samples.
-------
Table 3. HOURLY AVERAGE RESULTS OF COLLABORATORS FROM THEIR SAMPLING N02 AT LEVEL 2 (288 yg/m3)-/
Collaborator
Run
A
B
C
Date
7-30-74
7-30-74
7-30-74
7-30-74
7-30-74
7-30-74
7-31-74
7-31-74
7-31-74
7-31-74
7-31-74
7-31-74
7-31-74
7-31-74
7-31-74
7-31-74
7-31-74
7-31-74
7-31-74
7-31-74
7-31-74
7-31-74
Time
1800-1900
1900-2000
2000-2100
2100-2200
2200-2300
2300-2400
2400-0100
0100-0200
0200-0300
0300-0400
0400-0500
0500-0600
0500-0700
0700-0800
0930- 1000^
1000-1100
1100-1200
1200-1300
1300-1400
1400-1500
1500-1600
1600-1630^
A
301
310
320
325
344
344
357
321
316
321
320
299
297
293
—
301
301
301
2£/
2£/
l£/
2£/
B
329
338
348
357
376
382
385
348
350
352
342
320
320
353
—
310
310
310
6£/
gJ
6£/
0£/
C
.
414
432
434
451
470
481
442
427
434
430
406
395
432
—
395
397
404
22c/
15£/
13£/
2l£
D
303
316
328
348
372
387
39 2^
344^
363
365
360
342
342
375
299
309
303
306
8£'
6£/
6£>
E
318
333
340
350
363
372
382
342
338
348
338
314
316
350
_
310
306
310
gC/
9£/
&£'
11-
F
310
310
329
329
329
357
357
348
329
338
329
310
310
329
.
19c/
9£/
9£/
310
310
310
310
G
385
414
423
442
466
492
526
466
455
466
470
436
451
503
gC/
13 ^
2£/
2£/
428
428
436
442
H
288
293
310
319
358
373
370
324
327
329
332
300
305
339
5£/
10°-^
2£/
4£/
293
299
302
300
I
402
415
432
434
447
466
472
444
419
432
427
402
397
432
5l£/
38c/
36£/
541
545
553
J
378
402
408
417
423^
-
^
428
427
432
421
395
406
432
"f
13£/
15c/
370
374
385
395
a/ This is the spiked value. The statistically determined average true value of N02 challenge (which includes
ambient N02) is 302 ug/nr. (Incompiling biasses there is an individual true value per hour).
W Indicates reading is for <, 1/2 hr.
£/ From unspiked samples—all other values are from spiked samples.
-------
Table 4. HOURLY AVERAGE RESULTS OF COLLABORATORS FROM THEIR SAMPLING N02 AT LEVEL 3 (187 ug/m3)-
Collaborator
Run Date
A 8-1-74
8-1-74
8-1-74
8-1-74
8-1-74
8-1-74
8-2-74
8-2-74
8-2-74
8-2-74
8-2-74
8-2-74
8-2-74
8-2-74
B 8-2-74
8-2-74
8-2-74
8-2-74
C 8-2-74
8-2-74
8-2-74
8-2-74
Time
1800-1900
1900-2000
2000-2100
2100-2200
2200-2300
2300-2400
2400-0100
0100-0200
0200-0300
0300-0400
0400-0500
0500-0600
0600-0700
0700-0800
A
169
171
178
173
173
173
169
171
175
169
171
180
180
188
0930-1000£/ 2^/
1000-1100 T$J
1100-1200 2d/
1200-1300 2d/
1300-1400
1400-1500
1500-1600
1600-1630
a/ This is the spiked level.
ambient M>2) is
198 ug/m3
b_/ Data from Collaborator "G"
£/ Indicates reading
is for <,
d/ From unspiked samples --a 11
B
171
173
180
177
175
178
169
175
175
169
171
182
182
188
IJS'
C
267
269
278
278
272
310
258
263
267
261
258
263
263
271
2li/
188 178 231
184 171 229
. 186 171 231
- 188 169 231
The statistically determined
. (In computing
is unreliable
1/2 hr.
biasses
at this
other values are from
there
level.
spiked
D
199
201
208
210
228
235
223
232
237
232
230
238
248
255
2^
27*'
9d/
197
191
197
197
average
E
197
201
209
203
205
203
195
203
205
199
199
210
216
218
£l/
197
194
194
194
true value
F
207
207
216
216
207
207
207
207
216
207
207
207
216
216
207
197
197
19£
0<
05
95
of
is an individual true
samples.
286
297
304
291
286
259
244
226
226
192
188
192
188
184
244
244
229
24l£'
H
198
201
208
203
215
202
191
192
202
192
210
210
217
213
218
213
203
201
i/ 2d/ ll/
I/ l&! (£/
H - el/
NO 2 challenge (which
value per
hour).
I
207
203
209
205
207
203
194
197
203
197
197
203
207
210
207
201
199
J
229
241
252
248
248
248
242
256
263
252
258
267
271
272^/
248
250
242
246
ll£id/ 15d/
^d/ 9<1/
41/ 9d/
includes
-------
Table 5. HOURLY AVERAGE RESULTS OF COLLABORATORS FROM THEIR SAMPLING N02 AT LEVEL 4 (47.1 pg/m3)-
/
Collaborator
Run
A
B
C
Date
7-31-74
7-31-74
7-31-74
7-31-74
7-31-74
7-31-74
8-1-74
8-1-74
8-1-74
8-1-74
8-1-74
8-1-74
8-1-74
8-1-74
8-1-74
8-1-74
8-1-74
8-1-74
8-1-74
8-1-74
8-1-74
8-1-74
Time
1800-1900
1900-2000
2000-2100
2100-2200
2200-2300
2300-2400
2400-0100
0100-0200
0200-0300
0300-0400
0400-0500
0500-0600
0600-0700
0700-0800
0930-1000^
1000-1100
1100-1200
1200-1300
1300-1400
1400- 1500
1500-1600
1600- 1630^
A
66
75
77
100
77
79
90
86
92
73
73
73
79
86
73
62
58
56
IS/
l£/
]£•
l£/
JB
64
70
81
113
79
81
100
90
92
75
77
79
90
94
79
70
56
53
o£/
o£'
Q£'
o£
c
68
79
102
133
113
103
115
118
118
103
98
98
109
116
122
98
88
84
26£/
24£/
19£/
19£/
D
66^
-
89
122
94
99
110
108
107
89
88
89
loo,.
107^
35
91
40
-
33c/
5£/
25c/
13c/
E
66
73
86
116
90
90
100
96
92
83
81
81
90
96
73
64
58
—
s£'
££./
BE/
F
75
85
94
122 •
94
94
103
103
103
94
94
94
103
113
28c/
igc/
9£/
66
56
56
66
G
73
83
94
130
96
94
107
103
96
83
81
84
98
109
47£/
22£/
8£/
4£/
62
60
71
66
H
73
86
109
140
104
102
111
126
106
92
92
94
108
125
44£/
13c/
io£/
5£/
50
51
52
34
i
7l£/
81
92
120
94
94
103
98
96
83
81
79
90
98
45£/
2l£/
ll£/
9£/
56°/
56
60
64
J
84
94
105
141
105
107
122
120
100
96
96
90
109
113
43c/
2oC/
is£^
68
68
70
71
a/ This is the spiked value. The statistically determined average true value of challenge (which includes ambient
N02> is 60 ug/m^. (In computing Masses there is an individual true value per hour).
b/ Indicates reading is for <. 1/2 hr.
£/ From unspiked samples—all other values are from spiked samples.
-------
Analysis of All Spiked Readings (Precision Estimates)
Recall that the experimental design model for this set is:
» + C
i j
where |i = Overall mean,
CL = ith collaborator, i = 1 ..... 10,
t = j hour, j = 1, . . . , 20,
1^ = kth N02 level, k = 1, . . . ,4,
= Measurement error in 4th reading in ijkth cell,
SL = 1 for every ijk,
= ijk4tn response (spiked reading).
The results of the analysis of variance of the spiked readings are
shown in Tables 6 through 8.
Table 6 shows the basic analyses of variance themselves. Note that
a separate analysis of variance was done for each level. This was necessary
because the variance within a collaborator (described by ae) was not the
same at all NC>2 levels (see also Table 8).*
All the F-values in Table 6 are significant, i.e., at all levels of
NO 2 the collaborator averages are separated and a significant variability
in N02 exists in time.
Since the C and t effects are significant, it is desirable to
quantitatively describe the differences between collaborators. Table 7
displays these differences for each NC>2 level. The average M>2 value per
level is listed in this table in order to give a rough idea of the relative
separation of collaborators (even though the average value is not the true
value used in the analysis of variance) .
Since the collaborator (mean) differences are often quite large,
at least some of the collaborators must be biassed significantly. Also,
the order of the collaborators seem to vary quite a bit from level to level.
The Kendall concordance (a coefficient of agreement equal to 1 when order
is perfectly preserved) is only 0.52. Therefore it is reasonable to suppose
that a significant collaborator-level interaction exists (see Figure 13,
e.g., lines are not parallel).
An assumption (the homeoscedastic assumption) of all analysis of
variance models is that ae is uniform.
46
-------
Table 6. ANALYSIS OF VARIANCE SPIKED READINGS
Source df SS MS
A. Level
Total 139 4,345,388
Collaborator, adjusted 8 74,517 9,315 32.28
Time (t), adjusted 19 127,072 6,688 23.14
Error (e) 111 32,039 289
B. Level 2
Total 163 22,912,110
Collaborator, adjusted 9 416,836 46,315 143.79
Time (t), adjusted 19 71,471 3,762 11.68
Error (e) 134 43,165 322
C. Level 3s
Total 152 6,930,854
Collaborator, adjusted 8 117,350 14,669 1,552.28
Time (t), adjusted 19 7,171 377 39.94
Error (e) 124 11,070 89
D. Level 4
Total 164 1,391,497
Collaborator, adjusted 9 11,257 1,251 39.13
Time (t), adjusted 19 45,577 2,399 424.60
Error (e) 135 4,316 32
a/ Collaborator G is deleted as an outlier.
47
-------
Table 7. COLLABORATOR AVERAGE DIFFERENCES SPIKED READINGS (jig/m3)
a/
Difference
D-E
A-E
C-E
G-E
F-E
B-E
J-E
H-E
I-E
[E =
f Level 1
(average 154)
22.3
1.2
60.6
-
18.1
21.0
25.2
-6.0
64.5
(131)
Level 2
(average 371)
8.2
-21.1
88.9
110.7
-15.4
5.9
70.8
-20.3
100.9
(335)
Level 3
(average 212)
20.2
-24.7
61.9
-
1.8
5.6
-25.4
48.7
1.3
(202)
Level 4
(average 90)
6.9
-7.8
18.2
8.0
9.5
-4.2
17.7
13.7
3.4
(83) ]
a/ This difference is the estimate of five mean difference
between collaborators (see Appendix E).
Table 8. COMPONENTS OF VARIANCE SPIKED READINGS (jig/m3)
Level
Source
ae
ac
-t
2 2
LI
16.99
21.24
26.67
27.20
L2
17.95
47.95
18.55
51.20
L3
9.45
27.00
5.66
28.61
L4
5.65
7.81
15.39
9.64
48
-------
400
300
LJJ
oo
CQ
O
O
Z 200
O)
100
I
L-4 L-l L-3 L-2
(-90) (-154) (-212) (~371)
Figure 13 - Collaborator-Level Interaction (w/o collaborator G)
-------
The components of variance are shown in Table 8. Recall that:
CTe = standard deviation within a collaborator,
CTC = standard deviation of collaborator effects, and
at = standard deviation of hourly effects.
Although we do not know the exact NO 2 values per level, it is
surely true that in ascending order the levels are L4, Ll, L3, L2;
approximately. spanning the range 90 to 370 ug/m . Therefore, Table 8
indicates that crc is proportional to the N02 level. The
-------
Recall that a true value for each hour is constructed from the
average ambient observation during that hour. Since only half the col-
laborators were measuring the ambient at a given time, their average value
has this source of error in it.
The analysis of variance is shown in Table 9.
All the F-values in Table 9 are highly significant. Therefore, the
bias does differ between collaborators, does depend on the N02 level, and
the individual collaborator's bias curves are not parallel (See Figure 14).
o o
The average true values for the levels are 111.9 ug/m , 301.6 vg/m ,
197.7 ug/m3, and 60.0 ug/m3 (overall average is 167.8 ug/m3). Thus, the
overall average bias is not too great (about + 10%). On the average, the
bias is greatest at the highest N02 level (+ 157.) and least around
200 ug/m3 (+ 3%). (See Table 10.)
These average results are not sufficiently descriptive of the bias
situation, however. Only about four of the collaborators (D, A, F, and E)
exhibited even fairly consistently bias per N02 level results. Note that
the really large biasses are all positive (thus the average bias is pos-
itive), but in the more or less "normal" results almost half the biasses
are negative.
Summary Discussion of Statistical Analysis
In general, the relative measurement errors are stable over the
range of N02 measured (approximately 50 to 400 ]ig/m ) and not very large
(approximately ffl, true value). The collaborator-collaborator relative
standard error is also fairly stable but larger (approximately 12% true
value) so that the method standard deviation f~2 2 is on tlie average
about 13% of the true value. \| CTe + ac
However, the bias is not stable with respect to N02 level, and is
not consistent within collaborators, either. Although the overall average
bias is only about + 10%, individual collaborators produced biasses as
great as + 80% (at some levels). Thus, it is fair to say that the continu-
ous -colorimetric method may produce extremely inaccurate readings in an un-
predictable fashion (even though the overall average results are fairly
accurate).
It might be noted that about half of the collaborators did achieve
fairly stable results throughout the experiment. A subjective interpreta-
tion of this fact is that the continuous-colorimetric method is difficult
to use, but will produce reliable results in some hands.
51
-------
80
70
60
50
40
\
\
\
30
20
10
-10
\
A \
-20
JL
_L
L-4
(60)
L-l
(112)
-L
L-3
(198)
L-2
(302)
Figure 14 - Collaborator-Level Interaction in % Bias (w/o collaborator G)
53
-------
Lower Detectable Limit (LDL)
Two measuring of LDL are used in the following discussion: (a) the
smallest value of NO2 that can be reliably identified as existing (i.e.,
positive) when the method is used by a_ collaborator (a "pure" LDL), and
(b) the smallest reliable NC>2 estimate from a set of collaborators using
the method (a "practical" LDL).
Two methods of estimating the LDL were used. The first method uses
the ambient readings obtained during the actual experiment, while the
second method uses the collaborators' calibration curves.*
The ambient readings furnish estimates of a (standard deviation
within a collaborator) and ac (standard deviation between collaborators)
that allow estimation of the LDL's, although there is no way to incorporate
bias with these estimates. Using a = 4.48 ug/m3 and a =6.19 ug/m3
results in:
estimated pure LDL = 9 ug/m
estimated practical LDL = 15 ug/m3 .
The calibration curves do allow estimation of biasses** in addition to
components of variance. Using the average ae °f individual calibration
curves results in a pure LDL estimate of 13 ug/m , of which 2 ug/m3 is bias.
Using the whole data set, one arrives at an estimated practical LDL of
19 ug/m3, of which 3 ug/m3 is bias.
The two sets of results agree fairly well. It seems reasonable to
state that the pure LDL is probably £ 13 ug/m3, and the practical LDL
«£ 19 ug/m3.
* The only calibration data available were from collaborators A, G, B,
and E.
** Deviations from the correct values, e.g., a nonzero intercept for a
blank value.
54
-------
CONCLUSIONS
The major conclusions that can be drawn from the results of this
collaborative test are:
1. The N02> ambient-air sampling system developed by MRI is an
effective system for use in collaborative testing of methods such as the
continuous-colorimetric procedure.
2. If the tentative continuous-colorimetric procedure as given in
Appendix A of this report is followed by people knowledgeable of the
sampling and analysis techniques given therein, then such persons will
obtain results with an average bias of + 16.1 ug/m3 (67, true value) over
the range 90 to 370 ug/m3. On the average, the within laboratory standard
deviation (ae) is 13.5 ug/m3, and the collaborator standard deviation
Jac + ae is 32.7 (13% true value). These components are dependent, however,
upon the NO2 level.
3. The bias of the method is collaborator dependent, although four
of the collaborators produced fairly stable results in this regard.
55
-------
RECOMMENDATIONS
Based upon the conclusions that have been drawn from the results
of this collaborative test, it is recommended that:
1. The same N02 sampling system be used in the evaluation of the
chemiluminescent method to be tested .
2. The data sets to be obtained from the subsequent method to be
evaluated be based on experimental designs, test procedures and sampling
system operational procedures as similar as possible to those of the
continuous-colorimetric collaborative test so that comparisons of the
methods are based on similar criteria .
3. No further analysis be made of the results from the continuous-
colorimetric method until the results from the other method are obtained.
57
-------
APPENDIX A
TENTATIVE METHOD FOR THE DETERMINATION OF NITROGEN DIOXIDE
IN THE ATMOSPHERE (CONTINUOUS-COLORIMETRIC PROCEDURE)
59
-------
TENTATIVE :-ETK-D FOR THL QZTEPi'.IilATIG:,1 OF NITROGEN .DIOXIDE
i:; I:-:E ATLOS^JFRC (CONTIIIIT.US COLORI'.£TRIC
JUNE 1974
r.E1HGDS STANDARDIZATION BRANCH
QUALITY ASSJ?.;.','CE AM!) EiJVIROM":::TAL KOIilTOP.IfJG LABORATORY
OFFICt OF RESEARCH AND DEVELOK.-.EHT
U.S. ENVIRONMENTAL RESEARCH CHIITER
RISL'RCH TRIANGLE PAR!;, IIORIM CAROLINA 27711
aA tentative -:-J:»vJ is one v:hich has been caivfu-lly draftee! from
evuilEbU- :..;-_••-. snthl information, ro viewed cc'i tori ally '.,'ithin
the .".it'novs £ :?r.i.TJi7ation Branch ?ncl has invJorc-Dnc extrvoivc
laboratory s.. .-".ynioi. ' The nsthod is still u:i(!ir investigation
and therefore is s-jbjjc-L to revision,
61
-------
CONTII.'UCJS COLORIir.TRIC METHOD FOR MEASUREMENT OF
NITROGEN DIOXIDE IN AKBIEHT AIR
1. Principle ar.d I" pile ability
1.1 The rsthod is based on the reaction of N02 in acid media
to'produce nitrons acid (HOMO) with subsequent diezotization and
coupling. N02 in arr,bient air is continuously absorbed in a solution
of diazotizing-coupling reagents to form an azo-dye which absorbs
light, with a naxirrum absorbance at approximately 540 nm. The
transniittance, i:hich is a function of the N02 concentration, is
measured continuously in a colorimeter and the output read on a
recorder or a digital voltmeter.
1.2 The r.ethod is applicable to the continuous determina-
tion Of iutrOy=fi u~iOXldc "ill aifiblsnt all*.
2, Range and Sensitivity
2.1 Typical ranges are 0 to 470 ng/m3 (0 to 0.25 ppm);
0 to 940 ng/in3 (0 to 0.50 ppm); and 0 to 1880 ug/m3 (0 to 1.0 ppm);
Beer's lav/ is obeyed throughout this range.
2.2 For optimum sensitivity, the wavelength specification
of the filter in the colorimeter should correspond to the wave-
length of maxifiun absorbance of the dye. This may not be the
case in some instruments. Therefore, the dye should be scanned
and the wavelength of maximum absorbance determined. If the filler
is not within .•. 10 nm of the wavelength maximum obtained by scanning
the dye, the filter should be replaced by one thaL meets this
specification.
62
-------
o
"I. •
.
3.1 rjc~.nt stuJiCj neve shown th.it ozonfi can pro !uco
c r?i;.tv':' i':'_rfercnce, Liic magnitude of iv!iicii depends on the 0^
tc ;., ro: ,c. In th'; study cited, n^/f!09 ratios of 1:1, 2:1 r.nd
£_ \) f.
3.5:1 vvc. '..:3J interferences of 5, 10 and 3JT, rospocLivcly.
3.2 /.l.-.yl nitrites are positive intcrfercnts. The
:"i?c.iiit^2 cf L'TE; interference depends on the structure of the
(?\
?lkyl nitrrci.^''
3.3 A 30/1 ratio of S0? to f'Qg slo.'/ly bleo.ches the color
f3)
of the ezo-djpc in the manual procedure, ' and this effect may
be c-poli cable to the continuous procedure.
4 . Prscisio-i, .'.ecu racy and Stability
';.i !!<•• rist,-* a*-? a^pn.aMc4 cr. prcclcicr, end uccu'racj.
4.2 Air bubbles can accumulate in trie optical cell and
u'ill cause t'-n erratic response. This instability can he mini-
ni.'.cd by ifict'c:si:ig the air and solution flov; rates. Thn ratio
of the air \o ssjljtion NO'-J rate should be maintained a I the
Vrlui race -, ''jo by the K-anufacturer (sc.e Section 7).
4.3 T'i2 rod i Moil Snltzn-?.i absorbing roflgont (Section 6.8.1)
is sisble i:r o-iS ronth under laboratory conditions, 22 C. -
ev«cd to lie1. 1. The Lysiikovi solution (Section 6.8.2) develops
an abso'h-.K' cf approximately 0.0? aiter <>...• i-ranth under Irbovaiory
cc"!ditio-.s. Tho net eijsorhance (ahsorbanco drvclcped by adding a
1,0" sol-jcici - blank is unchanged afU-r one t.'o.Uh).
63
-------
-3-
The stability of both solutions is unchanged after
tcr-pemturs cycling, to simulate iirr.bient conditions, up to 30 C
for four hours per day for seven days.
5. /.pearat us
5.1 Continuci.5 \&2 analyzer. Sample air is drawn through
c gas/liquid contact coli,~n at an accurately determined flow rate
concurrent to a controlH-d flow of absorbing reagent. The sample
inlet line prior to t->s absorber column should be constructed of
eitner glass or Teflon. The absorber column must be carefully
designed end properly sized because MOp is somewhat difficult to
absorb. The colored solution is passed through a colorimeter
I'.Yisra the trsrssni traces is measured continuous!v.
5.1.1 Probe. Glass or Teflon, with inverted poly-
propylene or g"!?.ss fur.r.el at the end.
5.1.2 Installation. Instruments should be installed on
location and d2r.3pstrr.t2d, oreferably by the manufacturer, to
i.,oet or exceed the specifications described in the addendum.
5.2 Calibration. The calibration aoparatus and its use
is described in Section 3. Additional components follow:
5.2.1 Dilution Air and Flushing Air (or M2). This can be
co::pr-?ssecl (h'v.jse) air or cylinder air. It should be purified by
passing through silica gel for drying, and through activated
ch.ircoal (i'-l-"r resh), and molecular sieve (6-16 mesh, type 4A)
to re-rove &ny l^ anc^ hydrocarbons.
64
-------
-4-
:>.2.1.1 Puri'.y. Test the purity of the: dilution and flushing
air ' ..• :: .:i<:iir; L'i? instrii.riin!: in the zero mode until a stable
tesc-h1: is cbiaii -:d. Connect the dilution or flushing air to the
air -,:•-.•';•> of i'1 . ras/liquid contact column and ooerate the
instr. • 'nt in t'n? c..'bient mode. If the response changss by nore
than I,. -Ice tha n'jis: level, the air is impure. Correct before
3
5.2.2 Flc.v1 r=ters. One each with ranoes of 0-100 cm /min.,
0 to 1 ;/niin. and 0 to 20 fc/nrin. is required.
5.2.2.1 Calibration. This can be accomplished with a bubble
flo.v r.-.-.er or a '.;ot test meter. Hith a stopv/atch, determine the
o
r«:.T. of t-.ir f ic ; COT /min,} throuqh the flow meter at a minimum
of four different ball positions. Plolb ball positions versus
flow rat2S.
5.2.3 Ther.-.or-oter. Graduated in 0.1° intervals over the
rang-- 20 to 30° C.
6. Pc-.'or.t
G.I Sulfc:u1a'nide [A-tHgNjCgl^SOgf.'Mg]. MeHincj point
1G5-167°C.
6.2 Sulfisiilic Acid Mcnohydrate, [4-(NH2)Cgll4S03H-H20],
ACS r:-vr':cnt grac!2. Either the ir.onohydrato or anhydrous form
can b.? used, prcvic'cd the degree of hydration is knui.-n. If the
dccr^o of h.ydratic--1! is not known, rccryilallize from uvter and
dry ever nic.ht at lJ:00C.'r' This will give the anhydro-is
65
-------
-5-
6.3 '-'-(l-MaplitV/D-otliylcnediarsiins dihydrochlorii'a (flEDA).
D«st grade avail able.
6.4 Tartaric Acid. ACS Reagent grade.
6.5 Glacial Acetic Acid. ACS Reagent crscle.
6.6 Z-iip-piithol-S-S-disulfonic acid disodium salt.
[HQC^H-fSO^laJp] Technical Grade. This compound is also known
by its trivial na^e, R-salt.
6.7 Ilitrite-frse distilled water. Mix the water with
absorbing solution. Absence of any visible pink coloration indicates
that the water is of acceptable quality. If the solution turns pink,
redistill the v;ater in an all-glass s-till after adding a crystal
of potassium permanganate and Barium hydroxide.
fr\
5,9 .^tCCrti"" ?01llt''.0n.c'. E^t^r t*10 -frirH-Pi or! ^al-t-Tm^n^-1'
sol-Jticn or the Lyshkovr ' ' modification of the Saltzman solution
can be used.
6.8.1 f'.odified Saltzman absorbing solution. 0.5% sulfanilic
acid, 5.0% acetic acid, 0.005% NEDA. for 1 liter of solution
nrepsre as follo"s: Dissolve 5.52 g of sulfanilic acid monohydrate
(or 5.00 g of tha anhydrous material} in hot distilled water and
allow to cool to room temperature. Add 50 ml of glacial acetic
acid follo-./ed by 0.050 g of NEDA. Dilute to 1 liter with distilled
water.
6.8.2 Lyshkow soluLion. 0.153 Sulfanilanride, 1.5X Tartan'c
acid, 0.005:: ilEDA and 'i.00fj;i 2-Naphthol-3,6-disulfonic acid
diccdiuTi salt. For 1 liter of solution, dissolve 15.0 g of
66
-------
-6-
tertaric ecicl, 1.50 g sulf anil amide, 0.050 g of 2-naohthol-3,6
disulfcnic scid discdi-in salt, and 0.050 g NEDA in 500 nil of
distilled water. Dilute to 1 liter with distilled water.
7 .
Allc,/ the instrument to warm-up in accordance with
the manufacturer's instructions and until a stable baseline is
obtained. Turn p-j.-.ps on and adjust the air and absorbing reagent
flow rates and their ratio to the recommended values. Verify the
air flow rate by rr.easureir.ants with the 1 i/min. flow meter.
Calibrate the instrument as described in Section 8.
8. Calibration
8.1 General Inscription. A dynamic calibration is carried
out by generating synthetic atmospheres from the output of a reliable
I.'C^-perr.eation device and determining the instrument response. Instru-
rsnt response is then plotted against M02 concentration to obtain
a calibration curve.
8.2 ;;Q2-Per-eation Device. Obtain or prepare a reliable
NC^-perireation device with a permeation rate of approximately
1.0 pg'i'Op/.n'n. The following precautions must be observed in pre-
paring ?;92-2?:-iT.3ation devices:
1. The lu^ used to fill the device must be dry. '
2. The filling operation must be carried out in a
dry atr,jiph:-rc' to insure that water is not introduced while filling
the tube.
67
-------
-7-
3. Ths !,'92 siiould be purs, essay 93 ^ or greater.
4. All so?.ls in the device should bo leak free.
5. The Derroation rate should be chocked gravi-
retrically as follows:
a. Allc1./ the device to reach temperature equilibrium
in the N02-atrosoli2re gsnerftion system (S-jctlon 8.3). This will
be attained over-nic'it, in r:,ost cases.
b. Weigh the device periodically and record the time.
(Transport the device from the atmosphere generation system to
the balance area in a dessicstor.)
c. All weighings should be carried out at the same
relative humidity ± 10r-. The time of exposure of the device to
f'oa r^nocnhorn rJiivi" r n i.ra-j nh j nn chr»t|1 r\ hn rnnc4-^p1- ( ^ *3fi Qor \
" • " - - •- ^ -^ j- -- -.-*-___ _ % .--».. /
fro^ v/eiching to v/eic'ninq. This technique cancels any weight gain,
due to moisture - ii02 reactions at the effusing surface, and gives
a reliable measure of the fJOp-weight loss.
d. The tiir3 interval between weighings will depend
on balance sensitivity. With a sensitivity (standard deviation
at the ir?.ss being 1:2ished) or 40 pg, weighing at 24-hour intervals
will produce reliable woicjht losses.
e. Plot device weight (in micronr?.,ns) on the y--axis
versus cunulative tire (in Minutes) on the x-axis. Obtain sufficient
data (at least five well-sp.icnd points) to establish the slope
of the line, winch is the permeation rntc in ug/inin. Determir.e the
slcpo alcibraically or by rooression analysis.
f. The pi.Ti:.vMt ion r«ils should he- constant: duel in
rer.r.Mi.'il'lo a~r> " '\nt vnJh ii;p sunpl icjrr. or other nr:vious valve.
68
-------
-8-
5.3 ;,'09-,'--3sph5ra-Generatiori Systen. This consists of
an ii^-r'-jrv.oatio:*. cievice contained in a \.atsr-jacketed condenser
L.
vnic/i is cor,.nscr::d to a constant-temperature bath. A homogenous
iiO? in sir arrosii.ere is produced by flushing the MO?J effusing
frcr: t;ve calibrated liQp-perrveation device, into a mixing bulb where
it is further diluted with dilution air. Figure 1 sho//s a diagram
of this system \.itr. suggested specifications for the component
p?.r;s. The follcvinj key specifications nvjst be rr.et to insure
th»- ra-eration of raliable calibration atmospheres:
8.3.1 Temperature control must be maintained to within ± 0.1° C.
of a fixed value.
f;.?. ? rlii^Pr-'rci nn,-] Hiliitinn air T:ip^e- ir>ii«;t he Hr\/ pnH
** • ~ w
iVc.i of I'iOg (see section 5.2.1).
3.3.3 A Kjeldchl connecting bulb with a volume of at least
150 or. is required to obtain adequate mixing of K02 and dilution
air.
ci.3.4 Ccn.'-.ections must be of glass or Teflon when contacting
.'rQ.,. Rubber tubirg nay be used for flushing and dilution air
com-cr'ons. Tygon tubing should not be used. Systems for pre-
pars.icn of calibration atir.osphsres have been described in detail
by G'Keeffe end Ortr.s.-,/9' Scaringelli, et al.,^10^ and Scaringelli,
/«\
Rov.:'!r;--rrj and Pel"5. ''' Cc,r.nercial calibration systems using the
pe;-,."---ion tube technique are now available.
69
-------
-9-
8.4 fi02 Atircs1:1 r»s. /.How the N00 atmosphere ocnercition system
to equilibrate for at l(.-~it one hour with flushing and c!Million air flow-
in;;. Generate a calibration gas equal to 80 * 5£ of full scale by r.djust-
i-;• the dilution flo.v r:.te. Calculate the exact concenLrntion from the
following relationship: _
D + F
w'':3re:
3
C = NOp concentration,
P = NOp penrsaticr. rate, ug/nrin.
F = Flushing air flow rate, £/niin.
D = Dilution air flew rate, A/min.
10 = Factor to convert liters to cubic motors.
Sai..ple liir! aLiiiOs J; ^."C: until a SUtblC resCOr.SP. TS Olilninorl and r-vrnr.-!
the response. Generate four additional concentrations of approximately
10, *3' 20, 40 and 60"i of full scale and determine the response.
8.5 Other relifr.le dynamic py^ocedurcs for generating fJOp can b^
u:ed. For example, gas h''i?.sc titration of excess NO v/ith 03/ ' and
f"!2l
enslyzed cylinders of flOp in i'L that are stable. '
8.6 Calibration C'.'rve. Plot the concentration of N02 in micnicjrrns/
c-^bic r.eter (x - axis) cvvinst instriim?nt response (y - axis), and dra1-; tho
IUD of best fit. Scf'e ir.strurents are designed to give a linear and
?. non-linear response.
,,
"' Tne f;0? perreation »v.t2 e.nd highest workchle dilution air flow rate r.ny
necessTtate a hicJic-1 v
70
-------
-10-
8,7 •Freqi.:'..1cy of Calibration. Tho calibration should be;
ci'eci'Jid c'j-ily !;y - • viij thu Ciilibratlon curvj at SO!' of full
scale. Spr.rr.i-g Ly c,:.-.;:ration or a dynamic slrndard is preferred.
However, if field •.:: c," the in:, trunent i.^.'cer. this innraccical,
a static-ctlibrauicn c:\--ck can be carried cut by adding a solution
of nitri:? ic.n, ,'.'l', , (?>s .'IcN'O-) to the aL-sorbing solution to
generate tr-:j dye. {, ..st instruments have ei static calibration
ipode thrcuoh ^hich solutions din be introducc-d.) CAUTICM: Static
and dynamic calibr?.Li';..s i^ay not agree. Therefore, if static
spanning is to be us'-d, a static reference point should bo
established et t.ie ti;. E of calibration,.
9. Calculations
9.1 N02 Concentration. This is read directly -from the
calibrsLion curve. /• ore-hour oi- longer av?>v«fjG concentration is
reported. El3ctvc-'-ic or electro-ii^ch?,n.ical intogration, eciual
area aver'cj-irg, a -j'lc^-i.'oter, paper v/eighinrj techm'cjiic-s, or the
averarjs of = digi ".1 c-itout can ba used to ol.-iain the average
concentrrticr. .
follc./s.:
The i';02 concentration can> be converted to o^ni as
p-a f;02 = Pg N02/m3'' X 5.32 X' TO"4
71
-------
-11-
9. 3 Air Volume. The volume of air sampled is not
corrected to S.T.P., because of the uncertainty associated with
-a temperature and pressure valves.
10. References
1. Clark, T. A. , et a]_. ,. Environmental Protection Agency,
Research Triangle Park, N.C. 27711. "Instrumentation for
the Measurement of Nitrogen Dioxide." Presented at the
ASTM-EPA Symposium1 on "Instrumentation for Monitoring Air
Quality, 8-14 to 8/16/73, in Boulder, Colorado.
2. Thomas, M. D., _et a±. , "Automatic Apparatus for Determination
of Nitric Oxide and Nitrogen Dioxide in the Atmosphere,"
Anal. Chc-n.. 28, 1810-1816 (iy56).
3. Saltzman, B. E., "Colorimetric Mi crodeterini nation of Nitrogen
Dioxide in the Atmosphere," Anal. Ch.rni., 26_, 1949-195!? (1954).
4. Scaringelli, F. P., Rosenberg, E., and Refine, K. A.,
"Comparison of Permeation Devices and Nitrite Ion as
Standards for the Colorimstric Determination of Nitrogen
Dioxide," Environ. Sci . Tech., 4_, 924-929 (1970).
5. Saltznan, B. E., "Modified Nitrogen Dioxide Reagent for
Recording Air Analyzers.," Anal. Chem. , 32_, 135-136 (1250).
6. Lyshkov/, II. A., "A Rapid and Sensitive Colorimatric Reagent
for Nitrogen Dioxide in Air," J.A.P.C.A.. 1_5, 481-484 (1955).
7. U.S. Patent 3, 375, 079.
•72
-------
-12-
8. i,:-.-_.-;l F/:r-:.u of Star.ih.vh. Techniccl ;';oLo -','0. 5.f',5. This
c;- \-. cbi2lr,:c.i f.-;n the Su;:c;riiitcnc!-.i!t of Coci.1!:-nls,
U.S. C-?vr--.--:rt h-inting Office, Washmgbon, D.C. 2CKO?.
Prici - 70 certs.
9. O'rCssffs, A. E., and Ortnun, G. C., "Primary Standards for
Trsci- C:s Analysis," Anal. C-i^u , 33, 760 (1956).
10. Scar:r.rsi!i, F. P., O'Keeffe, A. E., Rosenberg, E., and Bell,
J. P., "Preparation of Known Concentrations of Gases and
Vapors .iith Permeation Devices Calibrated Gravirnetrically."
Ans'. r-s-i.. £2, 071 (1970).
11. Fe: = rc1 ^ecisur, 35., 22392-22395 (Ilov^ber 25, 1971).
12. "i'-'c1--:, J. E. e^nl_., "A Strair'itfor'acd ryne^ic Calibration
Prcc.^ra for 'Jsi l.'ith NOV Iiibiru.rsnii." Presenter! at th» A P r A
««
Convertion, Denver, Colorado, June 9-13, 1974. Preprint No. 74-13.
73
-------
-13-
M. P^rfor. ,i!-:o Specif icilior.s For Continuous Coloriivetric
f'.rr/jC Multiple
.'.V.ise 0.005 ppm
Lc.:rr C-jtect£blo Lir.it 0.01 ppm
Zero D.-ifc
12 Hour ± 0.02 ppm-
24 Hour ± 0.02 ppni
Sp-ri Drift - 24 hour 0.02 ppm
Ley Tine 20 minutes
Rise TiP'.e, 95fi 15 minutes
Fell Tii.iti, S5',J 15 minutes
B. Definiticns of Perforrance Specifications
R-nge •- Miniir.-..^ and m?>:ii!ium concentrations v:hich the system
shall be capable or rreasurii-.g.
ii'o'iss - Spontaneous, short durotion deviations in the instru-
r:«t cutpjt tiLout the nsan output, which are not caused by input
ccv-t;iiration ciiiir/jes.
L 0'. -e r D c \ - r [ /• j 1 ?. [. i - i r - The minimum polluted concenir.ihon
v.'ln'ch produces a sirnal of twice the noise level.
Zc-fP_Dri ft - The chanco in instruirsent output ovar a st.sl.ed
tir.-2 period c-T unadjusted continuous operation, when I'he input
ctiicci'.tration of pollutriiL is zero.
-------
11.001.0.
GLASS CONDENSER,
FLC.M'ETER. OT015 l/min.
DILUTION AIR
v—»-~
EScni
RUKBER TUBING
MYGON TUBIHG-
THERMOMETER
PERMEATION DEVICE
sUJ
RP
FER CIRCULATING PUf.lP
CO:;;;ECTCR IFOR SAMPLING
VENT TO HOOD
GLASS JOINT-
KJELDAHLHIXING-x
BULB
co;;sr;>;.T-TLp,.p. EATH
±0.1r'C
TEFLON STOPCOCKS. 6mm
GLASS r.lA,§lirO!.DJ
Figuie 1. Typical N02Atmosphere generation system.
75
-------
APPENDIX B
DATA ON THE PERMEATION TUBES USED AS THE
SOURCE OF THE SPIKED LEVELS OF NO?
77
-------
There were four branches to the NC>2 permeation tube assembly. Each
branch contained a set of permeation tubes as follows:
Permeation Tube
- IS Branch N02—
(ue/min) (ug/min)
0.001
0.002
0.002
0.002
5.536
0.002
0.002
0.001
0.003
0.001
6.382
0.0003
0.001
3.200
0.001
0.002
2.980
Permeation rates for the above tubes were determined by the National
Bureau of Standards and validated by the Methods Standardization Branch
(MSB) of EPA at 25.1°C before they were given to MRI for use on the
collaborative test.
The combinations of branches used for the four runs of the continuous-
color imetrie collaborative test are:
Level Date Branches Used
1 July 29-30 2
2 July 30-31 1, 2, 3, and 4
3 July 1-August 1 1, 3, and 4
4 August 1-2 4
Branch
1
1
1
1
1
2
2
2
2
2
2
3
3
3
4
4
4
Number
35-8
35-16
29-3
28-10
34-3
34-13
34-6
34-1
34-10
35-13
29-4
29-2
34-12
Rate of N02
(us/min)
1.434
1.597
1.345
1.160
1.195
1.275
1.548
1.226
1.138
1.990
1.210
1.210
1.770
a/ The sum of the> N02 generated by each permeation tube in the branch.
78
-------
APPENDIX C
CALIBRATION OF THE VENTURI AND DRY-GAS METER
79
-------
The venturi and ry-gas meter were calibrated using a 1.0-ft /rev.,
wet-test meter, as shown in Figure 9 of the text. The wet-test meter
is connected between the splitter and the dry-gas meter. A bubbler is
used before the wet-test meter to saturate the air with water. The
air flow then proceeds through the venturi to the N0£ bleed-in as it
does in normal operation (see Figure 3 of the text).
Since the saturated air coming from the wet-test meter is not
dried before going into the dry-gas meter, no connection for water vapor
pressure is necessary and only the normal corrections for temperature
and pressure are used. The flowrate of the wet-test meter (to stp) is:
P 294
Flowstp = Flow (meter reading) x — x —
where T = temperature of wet-test meter + 273, and
P - Pat_ + pressure of test meter manometer.
The venturi flowrate is dependent on both temperature and pressure.
Therefore Flowef. is corrected to venturi conditions
sup
T
Flowventuri = Flowstp x
where To = temperature of gas stream + 273, and
P2 = Patra + P(gas stream) •
The dry-gas meter is temperature compensated, so only pressure
corrections are made for its readings and a temperature base of 21°C
is used for calibration. Thus the true flowrate of the dry-gas meter
(Fn) is
Fm = FlowstD x
m stp
where P3 = Patm + P^gag stream)
80
-------
The correction factor f to convert £„,> measured dry-gas meter
flowrate, to true flowrate is then
The venturi and dry-gas meter were calibrated at three f lowrates;
50, 55, and 60 jg/min. Normal system f lowrates are 55 to 60 jfc/min. The
calibration factor for the dry-gas meter is constant at the calibration
flows (* 0.2%). The average value of flow from seven determinations is
used in calculating true f lowrates of the system. The plot of venturi
AP versus flowrate follows a straight line over the range used in calibra-
tion. From the slope and intercept of the line f lowrates were calculated.
81
-------
APPENDIX D
WRITTEN COMMUNICATIONS WITH POTENTIAL COLLABORATORS
83
-------
1 July 1974
Dear :
This confirms our selection of your organization as one of the 10 collabor-
ators for the EPA-Sponsored Nitrogen Dioxide Collaborative Test Using The
Continuous Colorimetric Method, and presents information about this test.
A copy of the "Tentative Method for the Determination of Nitrogen Dioxide
In The Atmosphere (Continuous Colorimetric Procedure)," is enclosed for
your study and retention. This write up is the same as the one I sent you
10 days ago with the exception of the revision (see Section 8.5 of the en-
closed write up) that pertains to the use of reliable dynamic procedures
other than the nitrogen dioxide permeation device for generating nitrogen
dioxide can be used.
The test schedule is given in Table I. The starting date is Monday, 29 July
1974. We will meet in the Lobby of the Ramada Inn (see Figure 1, upper
right-hand portion) which is located at 87th Street and Highway 1-435. From
there we will go to the field site which is several miles from the motel.
At the site there will be an orientation program of the facilities prior to
your preparing your instruments for sampling.
The schedule of Table I is based on the assumptions that all equipment of
the collaborators will be on-site early Monday morning, and that all goes
well during the week.
The sampling by each collaborator must be performed according to the attached
write-up. You should calibrate your instrument both dynamically and stat-
ically (as a reference) at your home laboratory, and then use static cali-
brations in the field. Of course, if you choose to bring your dynamic
system with you, then a static calibration at home is not required.
The sampling system that will be used in this collaborative test is shown
in diagram form in Figure 2, an enclosure to this letter. Each collaborator
will attach his instrument to ports of the spiked and unspiked manifolds
according to a specific experimental design pattern, which is given in
generality on the following page:
84
-------
MIDWEST RESEARCH INSTITUTE
XxxxxxXXxxXxXxX
1 July 1974
1600
2400
0800 0800
1000
1200 1200
1400
1600
Spiked
Unspiked
*
cl-5
r *
C6-10
1
1
1
1
I
1
1
1
1
|
C6-10
Cl-5
Cl-5
C6-10
1
1
1
1
1
1
1
J
C6-10
Cl-5
Cl-5
C6-10
1
1
V
1
1
I
1
1
1
I
j
C6-10
Cl-5
The particular partition of collaborators into two groups of five will
be done randomly; for convenience, the groups are always labeled
Cl-5 an(* ^6-10 *n tne diagram.
Enclosed is a Teflon adaptor. One of these adaptors constitutes a port of
the sampling manifolds. It is the type of port to which you will attach
your instrument. The larger-diameter end is the one to which an instrument
is to be attached. This adaptor is being sent to you, as an enclosure, so
that you will know what size tubing will be required for the connection of
your instrument to the sampling manifolds. You will need at least 30 feet
of this tubing.
The switching of your instrument between the spiked and unspiked lines will
be done manually. During the night portion (1600-0800), MRI man on duty
will do the switching. During the day, each collaborator will switch his
device.
Each collaborator will be reimbursed for travel, subsistence, and lodging
for the employee it sends to Kansas City to perform the field work as a
collaborator, as well as miscellaneous expenses such as costs for shipping
the field equipment you will use on site for the sampling. Please keep
receipts such as airline tickets and equipment shipment invoices. Mr. Fred
Damon, MRI Administrative Officer, will be contacting you to make contrac-
tural arrangements. If you have any questions you may contact him at (816)
561-0202.
Reservations will be made by MRI for each collaborator at the Ramada Inn.
These lodging charges will be billed directly to MRI, therefore, you will
need not consider this expense. Also, if you are not driving to Kansas
City, but rather arrive by plane, MRI will provide local transportation.
85
-------
MIDWEST RESEARCH INSTITUTE
XxxxxXxxXxxxxXx 3 1 July 1974
Upon your arrival at the Kansas City International Airport call Econo-Car-
Rent-A-Car Company at 464-5656 and you will be provided a car. As is the
case of the motel, these changes will be billed directly to MRI.
If you are not driving and bringing your equipment with you, I suggest that
you ship it a couple of weeks before the 29th of July so that it will be at
MRI before 27 July. For your convenience, you could send it to yourself,
% Mr. Paul C. Constant, Jr., Midwest Research Institute, 425 Volker Boule-
vard, Kansas City, Missouri 64110, COD or prepaid.
To help us with preparations at this end, I would like the following infor-
mation if you have not already given it to us:
1. The type and size (I.D.) of the connecting tubing you will bring
with you to connect your instrument to the sampling manifolds.
2. The name of the person(s) from your organization who will be
coming to Kansas City.
3. The mode of transportation the person will use to come to Kansas
City (in case of airline, the airline and flight number), and the time of
his arrival.
4. How your equipment will be sent to Kansas City and when it is
expected to arrive.
Very truly yours,
Paul C. Constant, Jr., Head
Environmental Measurements Section
PCC:cdn
Enclosures:
1. "Tentative Method for the Determination of Nitogen Dioxide
In the Atmosphere."
2. Table I — Test Schedule
3. Figure 1 — Map: Ramada Inn to Field Station
4. Map — Kansas City International Airport to Ramada Inn
5. Map — Deramus Field Station
6. Figure 2 — Nitrogen Dioxide Ambient-Air Sampling System Concept
7. Teflon Adaptor
86
-------
APPENDIX E
GENERAL ANALYSIS OF VARIANCE
87
-------
This appendix presents a brief description of the analysis of vari-
ance of the general linear statistical model. Due to the missing values
inherent in the execution of the N02 collaborative Tests 3 and 4, this
general analysis of variance was necessary in order to make the F-tests
and estimate the components of variance. The first two W)2 collabora-
tive tests produced data sets that were standard balanced frameworks
(factorials). Of course, the factorials for Methods 1 and 2 were special
cases of the general linear model, and all NO2 data sets could have been
analyzed by the general analysis of variance. However, computing algorithms
for particular cases exist that greatly reduce the labor involved, i.e.,
for many special cases of the general linear model it is not necessary
to perform the analysis of variance "the long way."
Factorial experimental designs are, in fact, so convenient to analyze
that the analysis of variance associated with them is often presented
without reference to the general analysis of variance. In other words,
certain particular cases of the general analysis of variance are so com-
mon that they have their own nomenclature. This means that the appearance
of the statistical analyses for the N02 collaborative tests changes some-
what between Methods 1, 2 and Methods 3, 4. Therefore this appendix is
offered to help the reader understand the statistical analyses performed
for all NC>2 collaborative tests.
The general analysis of variance rationale will be presented first,*
and the "special case" effects pointed out second.
The general linear statistical model supposes that a "response"
(y) is predictable from knowledge of the levels of some "treatments"
(T,B)** but that a random error (e) is associated with observing the
response. That is, the model is of the form:
ijk = H + Ti + BJ + ek(ij)<
* This subject is treated more compactly with the aid of matrix
notation. However, the presentation here will avoid the use
of matrices for the sake of simplicity.
** Two treatments are sufficient to illustrate the concepts involved;
obviously, the discussion could be generalized to K treatments.
*** Including interaction terms would only complicate the discussion.
88
-------
where ji = overall mean;
Ti = 1th level of T; i = 1, . . .,1;
B. = jth level of B; j - 1 ..... J;
ek(ii) = "measureraent" error associated with ijk response;
k = 1 . . K for every i j ; and
Yjik = ijk response.
It is assumed that the e's have independent normal distributions
with mean zero and variance ae2 , i.e., the "measurement" error is un-
biassed and of the same size everywhere. Note that since e is a random
variable, Y is also a random variable. Therefore functions of the sam-
ple Yjk's have probability distributions.
Suppose that the object of collecting data is to make "proper"
estimates of the I + J + 1 parameters u , JTiJ, JBjj. By proper we
mean estimates that satisfy some sensible criteria. The criteria used
in this case is that of least squares, i.e., the parameters are esti-
mated such that ZEE e2. .. is minimized. In other words, the parameters
ijk LJk
are assigned so that the measurement error (residual variation) is minimized.
Formally minimizing £Z£ e2. ... (via setting the partial derivative equal
ijk 1Jk
to zero) produces a set of I + J + 1 simultaneous equations (in u, T]_. • •
T!> 01 • • • Pj)- The solution to this set of "normal equations" there-
fore furnishes the prediction equation for YJJ^ . Also, the size of
the error variability can be estimated now, i.e., the (sample) variance
of the Y-Mk's is now partially "explained" by u, j^}, (3}. Whatever
is left over is "unexplained" variability. So setting up and solving
the normal equations furnishes: (1) an estimate of the model parameter
and (2) an estimate of unexplained variability, Og2 .
Now suppose that some hypothesis is of interest, e.g., HQ! TI =
T2 = • • • Tx . This hypothesis, in effect, dictates a "reduced" model
Yi'k = V + T + Pj = e(ij)k i-e-» under the null hypothesis there are
not I individual T parameters but instead only one common T parameter.
Obviously, the reduced model has to explain less of the variability in Y
than the other ("alternative") model did. So going through the same
procedure of constructing and solving a set of normal equations for the
reduced model procedures another estimate of ae2 (o^2' , say), and
o- 2<> a 2 . Now if a 2'is =s a 2 , then the hypothesis is reasonable
e - e e e
because the model works about as well when the T.J/S are constrained
to equality as when they are not. (The actual test statistic involved
89
-------
is an F-ratio but it is not pertinent to consider how to derive the
F-ratio here.) Now, suppose HQ: TI = . . . TI had been tested and
it is now desirable to test H<,: PI = • • • 3j5 this is possible of
course but in general it requires another set of normal equations and
their solutions; namely, the ae2 estimate (o"e2 say) from the model
Yijk = u + Ti + 0 + e (ij)k. Usually these two hypotheses are the interest-
ing ones but an infinitude of others (e.g., p3 = n, PL = 2.6 pz, even T3 =
P2~Pl » etc-) could be tested.
Summing up, then, in the general lineal model a set of normal equa-
tions is generated (according to the least squares criteria) and solved,
and the solution "explains" a proportion of the variability in the
response. " The unexplained proportion is represented by o_
(the "residual" variance). A hypotheses, usually of equality, dictates
a reduced model and thus another ore2 , and the comparison of OgZ1
and ae2' decides whether or not the hypothesis is reasonable.
Except there is a catch. The system of normal equations is over
determined, i.e., the system of I + J + 1 normal equations discussed
previously contains only (1-1) + (J-l) + 1 = I + J - 1 independent equa-
tions. Thus there are infinite number of solutions to any set of normal
equations. All solutions, of course, necessarily explain the same frac-
tion of the data, i.e., result in the same value of estimated Og2 .
Therefore, any solution suffices to test hypotheses of the form HQ: TI =
. . . + Tj , etc.
In other words, any solution to the normal equations allows esti-
mation of Og2 and the decision HJ,: Tj^ = . . . = TX? But suppose
HO is rejected, i.e., not all T's are equal. Now it is certainly
desirable to have the T estimates "mean something" physically, e.g.,
although setting -Tg = any value would test HQ , it would make the
parameter estimates in the prediction equation weird.
Now consider the situation from a slightly different viewpoint. The
normal equation are of dimension I + J + 1 in the parameters u {T^J,
{$*}, but only I + J - 1 independent estimates can be extracted from them.
In statistical jargon, there are only I + J - 1 independent "estimable"
functions. A statistical theorem says that the only linear combinations
of the Ti that are estimable are contrasts of the TI'S .* Using
this theorem it can be shown that, for example, setting Tj = 0 and
solving the normal equations for TI , . . . Tj^ results in the
Ti i < I solutions being estimates of T^ - Tj, i.e., after setting
TT'= 0 in the normal equations and solving for T3 (say), the numeri-
cal value of T3 is a proper estimate of the quantity T3 - Tj. . In
fact, setting Tj = 0 (and Pj = 0, I am ignoring the other treatment
Other functions of the parameters are also estimable, but in general to
have physical meaning one wants to stick to one variable (the T's or
fj's in our example) at a time.
90
-------
for simplicity) is a convenient constraint and is the usual procedure
for solving normal equations. When the general analysis of variance is
necessary as in the 3rd and 4th NO 2 collaborative tests, then, the re-
sults are estimates of from T^ - Tj- , for example, the C^ - C^Q esti-
mates for collaborator differences, etc.
So far we have considered the original statistical model to generate
an I + J + 1 system of normal equations that require two constraints,
i.e., have only I + J - 1 independent equations. Another way to approach
the situation is to change the model itself so that the normal equations
have a unique solution; i.e., put the constraints in beforehand, so to
speak. This approach is called "reparametrization." There are, as one
would expect, an infinite number of ways to reparametrize a model, but
one method is standard because it is convenient and does not confuse the
physical interpretation of the new model. Consider the original model
u + T£ + 0j + efc(ij) > but rewrite the equation as:
Yijk = fa + T_ + p.) + (Ti - T; ) + (pj - p.) + ek(ij) , or
Yijk = u* + T±* + Pj* + ek(ij) ,
where now E T.* = 0 and L 04* = 0.
1 l J
This is a valid reparametrization, i.e., the normal equations have a
unique solution for u*, JT^*}, |Pj*}> and the new parameters are mean-
ingful since Tj.* estimates the difference between T^ and its mean.
The reparametrization, in effect, uses Z T.^* = 0 and S Pj* = 0 as
the two constraints. Unfortunately, where this reparametrization is
used the results, by custom, are not presented as T^* but merely as
T-'s and the reader is supposed to remember that "£ T^" = 0. Therefore
it is natural to suspect that something has changed when the results of
the reparametrized model are presented as compared to the results from
the original model using Tj = 0 and p. = 0 .
It is easier to solve normal equations by setting Tj = 0 (and
0^=0) than it is to reparametize, i.e., when executing the general
analysis of variance the easiest estimates to get are the T^A— T-j-'s.
But in special cases like balanced factorials the normal equations are
trivially easy to solve under the above reparametization. In fact,
the solution is to trivial that the normal equations are not even
written down, but the reparametization is in effect, i.e., the constraints
Z Tj* = 0, £ PJ* = 0 are used.
Perhaps a numerical example will clarify the situation. Consider
the following data set:
91
-------
Pi
YHI = i
Y112 = 2
2
3
1
3
4
6
5
6
3
5
8
10
9
9
8
11
31
34
31
12 29 55 96
This data set is balanced (a 3 x 3 factorial with 2 observations
per cell), but we will not take advantage of any shortcuts in order to
illustrate all the concepts involved. The complete set of normal equa-
tions is:
18 u + 6 TI + 6 T2 + 6 T3 + 6 P! + 6 02 + 6
96.
6 u + 6 TL + 2 P! + 2 02 + 2 03 = 12.
6 u 6 T2 + 2 BI + 2 p2 + 2 03 = 29.
6 u 6 T3 + 2 0! + 2 02 + 2 03 = 55.
6 u + 2 TL + 2 T2 + 2 T3 + 6 0!
+ 2
+ 2
T2 + 2
+ 6
= 31.
34.
6 u + 2 TI + 2 T2 + 2 T3
+ 6 03 = 31.
We need to solve these normal equations, and then we need the new
set under HQ , where HQ is (say) HQ: T^ - T2 = T3 . Looking at the
normal equations, though, we see that if £ TI = 0 amd S Pi =0» the solu-
tions are obvious. For example, the first equation yields ji = 96/18,
the second equation yields £ + Tx = 12/6]^ i.e., T^ = 12/6 - 96/18, etc.
Thus, for this balanced data set, the normal equations can be solved by
inspection under the constraints £ Tj_ = 0, £ p. = 0, i.e., with re-
parametization. Also, under the null hypothesis HQ: TI = T2 = T3 (say),
the reduced normal equations are also solvable by inspection, and in fact
yield the same Bj estimates. In statistical jargon, the "adjusted" 0
reduction equals the "unadjusted" 3 reduction, consequently in balanced
designs like this the AOV table shows a row for the gss, i.e., since gss
is the same adjusted and unadjusted, it is just called the pss.
92
-------
Tf
So we see that the reparametization (i.e., the constraints T^ = 0,
|3j* = 0) allows an easy solution to all pertinent normal equations. (In
fact, it is so easy that methods books for such analyses omit the subject
of normal equations altogether.)
But now suppose the observation Y^2 was missing.* Then the normal
equations would be: (1) u + 5 TI + 6 T2 + 6 T3 + 5 0i + 6 P2 + 6 03 = 9
(2) 5 u + 5 TI + P]_ + 2 P2 + 2 03 = 10» etc- Obviously, the constraints
E TI = 0, £ Pj =0 are not particularly helpful in solving these normal
equations. The constraints 13 = 0, ^3 = 0 are handier. So these con-
straints would be used, and as a result the numerical solution for 1^
(say) would be an estimator of T^ - 1$, etc. Also, the reduced set of
normal equations (under HQ: T^ = T2 = 13) remains to be solved, i.e.,
there are adjusted and unadjusted sums of squares to compute.
In summary any statistical model of the form Y^^ = u + TI + P +
ek(ij)' * = lj ' ' ' ' Ij » 3 = lj ' ' ' ' J Produces: (!) a residual
variation ae^ for the model and a residual variation Og2 for the
model constrained by hypotheses of the form 1^: T^ = . . . = Tj, (2) a
statistical test for HQ , and (3) I - 1 (independent) estimates of dif-
ferences among the {TjJ , J - 1 independent estimates of differences
among the jp.) . In particular, which I - 1 estimates are produced depends
upon whether or not the design is balanced. In general, the I - 1
Ti - TT estimates are the easiest to obtain, but in a balanced design
I *
the T^* s subject to the constraint^ S^TJI = 0 are the easiest to obtain,
i.e., the new T-j^'s (T^* = T^ - T) are easier to estimate than dif-
ferences among the original T^'s.
Unfortunately for the sake of clarity, balanced designs are such an
important special class of linear models that, to some extent, they have
their own nomenclature.
In particular, in balanced designs the parameter estimates are esti-
mates of differences between treatment levels and their mean, but this
is not shown in the AOV table. Also, adjusted and unadjusted estimates
are equal so the words adjusted and unadjusted are not used.
This appendix intends only to resolve the apparent differences in
the form of the results for the first two collaborative tests. Actually,
even in the general case the analysis of variance does not require solu-
tion of both the null and alternative normal equations, i.e., by subtrac-
tion among sums of squares one can test HQ: Tj = . . . = Tj by solving
only one system of normal equations. Also, the analysis of variance
produces components of variance estimates that are not discussed here,
interaction effects can be included, etc.
"Fudging" rules of thumb exist for such slight distortions of balance,
but this is not germane to the discussion.
93
-------
APPENDIX F
INSTRUCTIONS FOR COLLABORATORS N02 COLLABORATIVE TEST:
CONTINUOUS-COLORIMETRIC PROCEDURE
95
-------
General Information
1. Calibration, sampling, analysis, etc., should be done explicitly
as stated in the June 1974 write-up furnished you on the "Tentative
Method for the Determination of Nitrogen Dioxide in the Atmosphere,"
(Continuous-Colorimetric Method).
2. Each collaborator will have an area for his analyzer on one or
the other table. Use of connecting tubing to the opposite table will
enable alternate sampling of the spiked and unspiked manifolds by each
collaborator.
3. For each run, each collaborator will connect his analyzer to
either the unspiked manifold or the spiked manifold, according to the
instructions given before the run. Heavier tubing, for those collaborators
whose instruments require higher flow rates is available in several areas
of both the spiked and unspiked side. Please ask the MRI person in
charge concerning use of the heavier tubing.
4. Collaborators are urged to make ample explanatory notes on their
analyzer charts to coordinate information and aid in data reduction. All
pertinent data should appear on the charts.
5. Each collaborator should work independently of other collabora-
tors.
6. On the spiked table, a separate power circuit (120 v 60-cycle,
four-outlet strip) is to be used by each collaborator for his analyzer.
On the unspiked side one strip will be shared by two collaborators where
necessary. These strips are under the table tops near the periphery of
the tops.
Test Instructions
1. Preparation by collaborators will precede sampling runs.
2. Analyzer calibration will be included in preparation time, not
to be done during the run.
3. Upon notification of "start testing" from the MRI person on
site, collaborators will mark the appropriate place on their analyzer
charts. It is expected that the analyzer will be operating and the
connecting tubing in proper hook-up arrangement prior to the start
signal.
96
-------
4. Note all pertinent data on the1 analyzer charts as the run
progresses.
5. Upon notification by the MRI person in charge, indicate the
end of,the run by marking on- the analyzer chart.
General Schedule
There will be two runs per each of four 24-hr periods. The dura-
tion of one. run will be approximately 14 hr; theother will be 7 hr. For
the 14-hr runs, all collaborators will sample from the spiked manifold.
For the 7-hr runs, each collaborator will sample from each manifold--
spiked and unspiked—for approximately 3-1/2 hr.
The hours 0800-1630 on Monday, 29 July are for collaborator prepara-
tion. From then on testing will be according to the following schedule:
Time Activity Spiked Line Unspiked Line
1630-1800 Calibrate
1800-0800 Sample 1 through 10
0800-0930 Calibrate
0930-1250 Sample 4, 6, 7, 8, 10 1, 2, 3, 5, 9
1250--1310 Switch lines
1310-1630 Sample 1, 2, 3, 5, 9 4, 6, 7, 8, 10
The numbers given in the spiked and unspiked columns are the collaborators'
I.D. numbers. (See attached table for your number.)
97
-------
COLLABORATOR ANALYZER LOCATION AEEAS
AND PRINCIPLE PORT ASSIGNMENTS
Collaborator
Spiked Table
Unspiked Table
I.D. No.
1
2
3
4
5
6
7
8
9
10
Name Area Ports Area Ports
Ken Smith
Lynn Hutchinson 't
Harold Davis
^^
Glenn Smith L
John Higuchi
]
> 1-6
i
^ 7-12
f^m
I 32-37
-
J 26-31
.
FT] 7-12
Norman Lewis |6 ] 13-18
Cleveland Dodge
•M
7 1-6
^^
Rolf Doebbeling QJ 26-31
Cole McKinney
Larry Saad 1
IT] 38-43
0 38-43
.
NOTE: Assignments of secondary ports (on the opposite table from the analyzer
location) will be made prior to the start of each test.
98
-------
SPIKED
SAMPLING MANIFOLD
COLLABORATOR
AREAS*
99
-------
UNSPIKED
SAMPLING MANIFOLD
COLLABORATOR
AREAS*
100
-------
APPENDIX G
N02 AMBIENT-AIR SAMPLING SYSTEM OPERATION DATA: TEST LOG SHEETS WITH
FIELD OPERATIONAL DATA
101
-------
LOG SHEET
Side 2
10
12
13
14
15
17
18
19
20
21 22
23
24
25
TIME
ttr
/?oo
NO2 ANALYZER - PAPER
OPERATION
VAC PRESSURE
f."
If
If
O2 PRESSURE
UNSPIKED - NO
NO2
JL.f
SPIKED - NO
NO2
/ 7
'7
17
2.1
2 1,
/f
ll
/
VENTURI MANOMETER - HI
r",
i'l?'/
5*9
LO
M'
•2*7
VENTURI AP
,.'7'
:jl_
INDOOR TEMP.
I -2
13
j •>
it
tit
GAS FLOW TEMP.
U.5
' I :
J.I
GAS FLOW MANOMETER - HI
5.
LO
W£
Ifo
GAS FLOW PRESSURE
EXHAUST - U MANIFOLD
S MANIFOLD
CHECK SAMPLE TRAINS
INITIALS
-/.- ',-£
> 1 •
i f i'« o
4*6
fir'.
JBt
rt.f
- \J7.a
.o S*/
r*.?
-------
LOG SHEET
Sidel
10
12
13
14
15
16
17
18
19
20
21
22 23 24 25
DATE
X-
7/
y*,
TIME
l?"i
-*•''
fee
//OO
Joo
'Voo
fibd
70O
OUTDOOR TEMP.
T1*-.
7"
(.i
o
fo
WET BULB
, i
f 7
Tl
n
A?
P
DRY BULB
•*
-------
LOG SHEET
Side 1
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
DATE
z
&i
TIME
JLOOO 4/«
-------
LOG SHEET
Side 2
10
12
13
14
15
16
17
18 ,
19
20
21
22
23
24
25
TIME
ISOO
(190
tioi
6 •/•;*.
cCcc
//o«
Woo
NO2 ANALYZER - PAPER
OPERATION
VAC PRESSURE
i*
L
O2 PRESSURE
9-0
2.0
2°
3LO
I"
UNSPIKEO - NO
NO2
n
SPIKED- NO
A*
o
Ul
NO2
3?
•XX
13
-It*
V
Iff
79
VENTURI MANOMETER - Ml
ft',
£1
Soo
LO
117
uv.
nS
ii-g
IV I
«*•/£>
VENTURI &p
INDOOR TEMP.
13.
13
a*
l-i
GAS FLOW TEMP.
>«•*
IT.
IT.
L'i-
aa-J
73.5
ay
GAS FLOW MANOMETER - HI
ft.
fit
fl'
f/t
&L
5i>7
fofc
LO
311
Jit
GAS FLOW PRESSURE
EXHAUST - U MANIFOLD
s
S MANIFOLD
CHECK SAMPLE TRAINS
ft?
TJf
a.i
INITIALS
tv*-
•5-. i
i /?£* "»«•>
51.3
-------
LOG SHEET
/-3-r
Side 2
10
11
12
13
14
15
16
17
18
19
20
21
23
24 25
TIME
,1"
,.*
oi'-
Ojtc
/eoo
/¥••
NO2 ANALYZER - PAPER
OPERATION
VAC PRESSURE
2 >
2?
To
i
V*
O2 PRESSURE
-JO
4.0
7-
XO
io
UNSPIKED - NO
NO2
SPIKED - NO
NO2
if
If
T-6
zc
if
VENTURI MANOMETER - HI
£»
LO
03
»U>15
21)!
n1
VENTURI AP
INDOOR TEMP.
*.*(
*•/
77 T
n*
GAS FLOW TEMP.
I? »
27,5-
GAS FLOW MANOMETER - HI
'L'l
LO
if*
i't
Iff
it*
5 4 '?
.
GAS FLOW PRESSURE
EXHAUST - U MANIFOLD
S MANIFOLD
CHECK SAMPLE TRAINS
INITIALS
a*
-------
LOG SHEET
Side 1
10
II
12
13
14
15
16
17
18
19
20
21
22
23
24
25
DATE
Ki
x.
*AL
&.
'X
K
.
v/
V.
TIME
/ft*
4/00
JLIflB
c(.ct
£./.-
0?..
|eO«»
lift
/foo
If, <."
OUTDOOR TEMP.
tr
IS
yd
ftf
il
11
11
D-
D-
90
i:
83
WET BULB
(,*
67
tf
t1
cc
DRY BULB
90
17
If.
77
11
22.
22.
n
REL. HUMIDITY
V/&
£7*
5?
BAR. PRESSURE
737
7V
737
HI
m.
7JZ
WIND SPEED
t.a
J-**
1-7
C-.5L
WIND DIRECTION
e
-------
LOG SHEET
Side 1
2 3 4 5 6 7 8 9 10 II 12 13
14
15
16
17
IB
19
20
21
22
23
24
25
DATE
V,
s/
i/i
2A
/*
TIME
/«W/1»
c/OD
oy*
'7-
OUTDOOR TEMP.
77
7V
7*
u
•77
7J
WET BULB
11
6*
L -
L/
DRY BULB
f6
71
7-
CC
REL. HUMIDITY
n
66*
77*
77
BAR. PRESSURE
734;*
736
Til,
7/6
O
03
WIND SPEED
*.?
0.6
— B-
/.c-
WIND DIRECTION
/Vu/
POWER LIGHTS
LEAKS PERMEATION
ASSEMBLY
BATH TEMP.
91.1
z-;, '
25
ly,
15,1
25V
ROTAMETERS - I
/a/
•O
H <•
IOC
/Lt.
1 0?
1 -'0
-4
N2 PRESSURE
tit
Vf
Uc
"7)0
(V.
RECORDER I - PAPER
OPERATION
RECORDER 2 - PAPER
OPERATION
.X
-------
LOG SHEET
Side 2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24 25
TIME
*0»|1«*
clK.
tit,
n.
NO2 ANALYZER - PAPER
OPERATION
VAC PRESSURE
IS
15
7s-
O2 PRESSURE
10
If-
UNSPIKED - NO
ft?
NO2
SPIKED -NO
o
\o
N02
/i
/JL
/J
II
VENTURI MANOMETER - HI
f»7
Ol
LO
*>»
J-ll-
7Z?
227
7?7
? /£
VENTURI
INDOOR TEMP.
93
93
JLl.?
11-
.'1
2?
'x*
GAS FLOW TEMP.
93
7.T-
2t
GAS FLOW MANOMETER - HI
LO
M
GAS FLOW PRESSURE
EXHAUST - U MANIFOLD
L--
S MANIFOLD
CHECK SAMPLE TRAINS
INITIALS
w
SI
JX
44 i
-------
APPENDIX H
COLLABORATORS COMMENTS
111
-------
Collaborator A
1. Solution flow fell off towards the end, resulting in several
no record areas. Bubbles were entering the flow cell causing fake high
or low readings.
2. Recording of known permeation rate remained constant throughout
the test, as did wet standards.
3. Base-line drift was erratic changing at random. This changing
did not seem to affect response or calibration while in an operational
mode.
4. When I returned, re-calibration of the instrument indicated
drastic changes in instrument response. I will not report this re-
calibration since something, as yet not defined, has definitely altered.
Collaborator J
Our instrument was dynamically calibrated prior to the tests. The
sensitivity of the instrument was set for 1% full scale responsed equivalent
to 0.0025 ppm N02- Static calibrations were run during the testing for
indicating references only. No adjustments were made to change the span
sensitivity during the test period.
Base-line calibrations were run during the testing, and reset when
needed. The standard value used for the base line was 5% of full scale.
No correction was incorporated in the data for apparent drift during a
test run.
Collaborator I
The value for each run was obtained by taking readings every 15 min
and averaging the result. The flowrate was varied in order to accommodate
the higher concentrations.
Collaborator E
The use of a planimeter for the number of measurements required
was impractical. However, the reeuction to hourly averages allowed a
fairly accurate determination of mean responses by estimating the square
wave which would give the same area as the actual response. The square
wave was determined by the horizontal line drawn through the actual
response curve at the point where it was estimated that the area below
the line and above the actual curve was equal to the area above the
line and below the actual curve.
112
-------
Collaborator G
Because a N02 permeation tube was not available at the time, the
Sectarian (K-1008) was calibrated indirectly using a Bendix calibrator
(8851) in conjunction with a Bendix NO - N02 - NOX (8101B) Chemiluminescent
alalyzer.
Our first step consisted of calibrating the NO and NOx mode of the
Bendix analyzer with the Bendix calibrator and a 99.0 ppm NO gas as a
source.
Next, some of the NO gas was converted to N02 by mixing it with ozone
generated by the Bendix calibrator.
The amount of N02 produced was determined by subtracting the final
NO concentration from the initial NO concentration; assuming that the
difference has been converted to N02- The sample stream was then in-
troduced into the Beckman acralyzer and an attempt to adjust the infinity
control to the concentration indicated by the Bendix analyzer was made.
There are some potential inaccuracies in this method of calibration;
it does not allow for error within the equipment itself nor does it take
into account N02 contamination of the calibration gas. (Because the gas
is certified as to purify, this error should be insignificant.)
Initially it was discovered that the Beckman acralyzer was reading
20.5% higher than the Bendix analyzer; and all attempts to adjust the
Beckman machine to a value that corresponded with the Bendix were futile.
At the time it was believed that there was not enough electronic ad-
justment on the Beckman acralyzer to set an N02 value corresponding to
the Bendix analyzer; however, at the MRI testing site, it was discovered
that the solution pump had developed a leak, which resulted in displacing
solution with air. In addition, it was determined that, the lowering
of the solution flow from 20 ml/20 min to 15 ml/20 min resulted in a
sharp increase of color development in the reagent and in an apparent
high N02 concentration.
The Beckman acralyzer has since been recalibrated using the corrected
solution flow (20 ml/20 min) and this time the readout of N02 concentrations
from the Beckman and Bendix machines corresponded closely. This can be
verified from the attached calibration curve.
113
-------
The data from 6:00 p.m. Monday to 8:00 a.m. Tuesday and from
6:00 p.m. Wednesday to 8:00 a.m. Thursday should be deleted because of
the excessive optical drift experienced by the equipment during these
periods. All other data up until Thursday morning (at which time the
solution pump was repaired) should be either deleted or reduced by 20.5%.
Data collected from Thursday morning through Friday afternoon should be
correct and no correction factor needs to be applied.
All data has been averaged in 1/2-hr periods. No correction factor
has been applied to the 1/2-hr averages. The 7 hr runs have been broken
down into 1/2 hr, 3-1/2 hr and 7 hr averages. The 20.5% correction factor
has been applied to the 3-1/2 hr and 7 hr averages.
The 14 hr runs have been broken down into 1/2 hr and 14 hr averages
and the 20.5% correction factor has been applied to the 14 hr averages.
114
-------
APPENDIX I
FIELD DATA
115
-------
The first nine columns of Tables 1-1 through 1-4 list various
readings used in calculating flowrates and spike levels, which are
given in Columns 10-14. The last six columns list various ambient air
conditions at the test site. The venturi and meter flowrates (Columns
10-11) are calculated from the calibration equations in Appendix C.
Due to the temperature compensations of the dry-gas meter and the above-
ambient pressure of the gas stream at these instruments, the flowrates
are calculated at 21°C and 760 mm Hg. The readings of the two devices
are then averaged (Column 12) and the average flowrate is corrected to
the temperature and pressure at the sampling ports (Column 13). Some of
the methods being evaluated with this sytem are not corrected for tem-
perature and pressure. However, if the spike levels are not calculated
at the temperature and pressure existing at the manifold ports, a sig-
nificant degree of uncertainty enters into any subsequent use of the
spike level. The spike level (Column 14) is determined from the permea-
tion rates of the permeation tubes used in each test.
116
-------
Table 1-1 L£VEL 1 TEST DATA
Calculated Floy Ratea and Spike Uvala
W>2 Sampling Syitem Data
pate
7-29-74
1800
1900
2000
2100
2200
2300
2400
7-30-74
0100
0200
0300
0400
0500
0600
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
1700
Room
Temp.
22.0
23.0
23.0
23.0
23.0
23.0
23.0
23.0
23.0
22.0
22.0
22.0
22.0
22.5
22.5
24.5
24.0
2S.O
25.0
25.0
24.0
24.5
24.5
24 5
Flow
Bar. Plan Temp. -
Preai. Preis. Meter
(in HR) tm Hal 1?CL
738
739
738
738
738
738
738
739
739
739
739
739
739
739
740
740
740
740
740
741
741
739
739
738
i 22.5
22.5
22.5
22 5
22.5
22.5
22 5
22.5
22.0
21.5
21.5
21.0
21.0
21.5
23.5
23.5
23.5
23.5
23.5
23.5
24.0
24 2
24.2
i 24.2
Flow
Rote
Meter
59.4
60.2
60.1
59.1
59.8
60 1
60.5
61.0
61.8
61.8
61.8
61.6
61.8
61.8
60.9
60.7
60.7
59.7
59.4
59.
58.
58.
58.
57.
Venturl
Pre«aure
Reading
(mm H20)
267
277
275
273
271
279
281
285
290
290
290
290
290
290
279
284
280
271
271
267
272
264
262
259
M2
Carrier
Flovrate
fee/mini
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
Permeation
Tube Temp.
25.1
25.1
25.1
25.1
25.1
25.1
25.1
25.1
25.1
25.1
25.1
25.1
25.1
25.1
25.1
25.0
25.0
25.0
25.0
25.0
25.0
25.2
25.2
25.2
Flovrate
Venturl
to 21'
+ 760 m Hg
U/mln)
59.0
60 3
60.1
59 8
59 6
60.6
60.8
61.4
62.1
62.2
62.2
62.3
62.3
62.2
60.5
61.1
60.6
59.3
59.5
59.1
59.6
58 4
58.2
57.7
Meter
to 21*
+ 760 OB Hg
M/mlnl
59.6
60 4
60.3
59.3
60.0
60.3
60.7
61.3
62.1
62.1
62.1
61.9
62.1
62.1
61.3
61.1
61.1
60.0
59.7
59.8
59.0
58.7
58.5
58.0
Averaee Flotrratc
to 21*
+ 760mmHg
59 3
60 4
60 2
59.6
59.8
60.4
60.8
61.3
62.1
6?. 2
62.2
62.1
62.2
62.2
60.9
61.1
60.9
59.7
59.6
59.4
59.3
58.6
58.3
57.9
Ambient!/
a/mln)
61 4
62 5
62 3
61.7
61 9
62 6
62 9
63 4
64 1
64 0
64 0
63.9
64.0
64 0
63 1
63 3
63.0
61 9
61 7
61 5
61 4
60.9
60.7
60 3
jplko
Level
Ambient*/
(u»/m3)
104
102
102
103
103
102
101
100
99
99
99
100
99
99
101
101
101
103
103
103
104
104
105
106
NO
Back-
ground
(u»/m3)
0
0
0
0
0
0
0
10
0
0
0
0
0
10
20
10
0
0
0
0
0
0
0
0
N02
Beck-
ground
(u«/m3l
20
20
30
60
90
90
110
110
80
60
60
50
40
80
90
80
30
10
10
10
10
15
15
15
Ambient Conditions
Out-
door
Temp
1^1
29
29
26
23
21
21
21
20
19
19
17
17
17
17
19
24
28
29
31
32
32
34
34
32
Wind wind
Speed Dlrec-
fm/acc) tion
5 HU
6 Ml
2 W
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
5 «w
5 NW
4 Ml
Relative
Humidity"
_(SJ
26
26
34
36
47
44
47
50
58
58
60
59
64
64
62
41
27
26
24
20
20
ia
18
20
Avarage
a/ Tenperaturo ind presiure it sampling porta
-------
Table 1-2 LEVEL 2 TEST MTA
CO
loom
Pate Temp
Time CO
7-30-74
1800 23 5
1900 24 0
2000 24 0
2100 23 0
2200 24 0
2300 23.0
2400 24 0
7-31-74
0100 23 0
0200 23.0
0300 22 5
0400 22 0
0500 22 0
0600 22 0
0700 22 0
0800 22 5
0900 23 0
1000 23 0
1100 23 5
1200 23.5
1100 24 0
1400 24.5
1500 24.5
1600 24 0
1700 24 0
NOj
Flow
Bar Flow Temp -
Preaa Preaa Meter
(mm Ha) (in Hit) CO
738 8 23 5
738 23 5
738 23 5
738 23 0
739 23 0
739 23 0
739 22 5
739 22 5
739 22.0
739 22 0
739 22 0
739 22 0
739 22 0
739 22 0
739 22.5
739 23 0
739 23 2
739 23 5
739 23.8
739 23 5
739 24 0
739 24 0
738 23 5
738 23 5
Sampling Syatea Data
Flow-
rate
Meter
U/mlnl
58 3
58 8
59 0
59 7
59 2
59.8
60.4
60 5
60.7
61 2
60.9
60 5
61 3
61.0
60 7
60 2
60.4
59.6
59.3
58. B
58.5
58 2
57.8
58 1
Venturl M2
Preaaure Carrier
Reading Flowrate
(an H.O) (ce/mln)
258 800
264 800
268 800
270 800
270 800
270 BOO
278 800
282 800
284 800
286 BOO
286 800
288 800
2B8 800
286 800
281 800
278 800
278 BOO
270 800
266 800
262 800
260 800
259 800
255 800
259 800
Permeation
Tube Temp
t*Q
25 1
25 1
25 1
25 1
25 1
25 1
25 1
25 1
25 1
25 1
25 1
25 1
25 1
25 1
2S.1
25 1
25 2
25 2
25.2
25 2
25 2
25 2
25 1
25 1
FlovratoB
Venturi
to 21*
+ 760 mm Hg +
(i/min) _
57
58
58
59
59.
59 4
60 5
61 0
61 3
61 6
61 6
61 8
61.8
61.6
60 9
60.4
60 4
59.3
58 7
58.3
58 0
57.8
57 4
57 9
Meter
to 21'
760 mm Hg
(//din)
58 5
59 0
59 2
59 9
59 4
60.0
60 7
60 8
61 0
61 J
61 2
60 8
61.6
61 3
61.0
60.5
60 7
59.8
59.5
59.0
58 7
58 4
57.9
58 3
Average Flowrate
to 21'
* 760 tun Hg Ambient!'
(f/min) fl/min)
58 1 60 3
58 7 61.0
59 0 61 3
59 6 61 8
59.4 61 5
59.7 61 8
60.6 62 7
60 63 0
61. 63 1
61. 63 i
61 63 3
61 63 3
61 63 7
61 63.4
60 63.0
60. 62 6
60. 62 7
59 61 8
59 61 4
58 60 8
58. 60 6
58. 60 4
57. 59 9
58. 60 3
Spike
Level
Amblent-
fua/m )
296
293
291
289
291
289
285
234
283
281
282
282
281
282
284
285
285
289
291
294
295
296
298
296
1.0 N02
Back- Back-
ground ground
(ug/m ) (ug/m )
0 20
0 20
0 40
0 40
0 50
0 90
30 110
0 60
0 40
0 50
0 50
0 40
0 30
0 40
0 40
0 35
0 30
0 20
0 20
0 20
0 10
0 10
0 20
0 20
1 38
Ambient Conditions
Out-
door Wind
Temp Speed
CO (m/a«c)
32 3
32 0
32 0
27 0
26 0
24 0
23 0
22 0
22 1
22 3
22 0
21 0
20 3
20 4
21 4
25 3
27 2
29 3
32 5
34 4
35 4
33 5
34 2
32 2
wind Relative
Direc- Humidity
tlon (t>
CW 22
20
19
30
46
37
39
42
42
SE 42
45
44
SE 58
SE 58
SE SB
SE 48
SE 42
S 40
SE 32
SE 29
SU 29
SW 32
KE 26
KE 36
•/ Temperature «i"» pre«ure «t •oaplLng ports.
b/ Initial ipike level incorrect—changed to p
proper level at 1820
-------
Table 1-3 LEVEL 3 TEST DATA
Room
Date Tenp
Time CC1
7-31-74
1800 24 0
1900 24 0
2000 24 0
2100 24 0
2200 24 0
2300 24 0
2400 24 0
»-'•>*-
0100 23 0
0200 23 0
0300 24 0
0400 23 0
0300 23 0
0600 23 0
0700 23 0
0800 23 0
0900 23 5
1000 22 5
1100 23 5
1200 24 0
1300 24 0
1400 24 0
1500 24.0
1600 24 0
1700 24 0
Ht>2 Sampling
Flov
Bar Plow Temp -
Press Press Meter
(mm to) fam UK) CO
737 8 23 5
737 8 23 5
737 23 5
737 23 5
737 23 0
738 23 0
738 22 5
738 22 5
738 22 5
738 22 5
748 22 5
738 22 5
738 22 5
738 22 5
738 22 5
738 22 5
738 23 0
738 23 0
738 23 0
737 23 2
737 23 5
737 23 S
737 23 5
737 23 5
System Dita
Flow- Venturl
race Pressure
Meter Reading
W-iitn) (us K20)
58 2 260
59 1 262
59 1 270
59 2 270
59 8 272
59 4 270
39 6 274
39 8 274
60 1 277
60 5 277
60 1 277
60 7 277
60 2 277
60 7 277
59 6 274
59 5 280
59 7 277
59 8 274
60 8 281
59 9 273
60 0 273
59 5 273
59 1 270
59 3 274
"2
Carrier
Plowrate
(cc/mln)
600
600
600
600
600
600
600
600
600
600
600
600
600
600
600
600
600
600
600
600
600
600
600
600
Permeation
Tube Temp
25 1
25 1
25 1
25 1
25 1
25 1
25 1
25 1
25 1
25 1
25 1
23 1
25 1
25 1
25 0
25 1
25 1
25 1
25 1
25 1
25 1
25 1
25 1
25 1
Flowratc
Venturl Meter
co 21* co 21*
+ 760 mm Kg + 760 en Kg
(*/mln) (./mln)
57 58 3
S8 59 2
59 39 2
59 59 3
59 60 0
59 59 6
59 59 B
59 60 0
60 60 3
60 60 7
60 60 3
60 60 9
60 60 4
60 60 9
59 59 8
60 59 7
60 59 9
59 60 0
60 61 0
59 60 1
59 60 2
59 59 6
59 1 59 2
59 6 59 4
AveraKc Plot
to 21*
+ 760 ma Hg
58 1
38 7
39 1
59 2
39 7
59 5
59 9
60 0
60 3
60 5
60 3
60 6
60 4
60 6
59 8
60 2
60 0
59 9
60 9
60 0
60 0
59 5
59 1
39 5
frate
Ambient!/
d/mln)
60 4
61 0
61 5
61 6
52 0
61 7
62 0
62 1
62 4
62 6
62 4
62 8
62 5
62 8
61 9
62 3
62 2
62 1
63 1
62 3
62 4
61 9
61 S
61.9
i Spike
Level
Ambient-'
(UK/B3)
192
190
189
188
187
188
187
187
186
183
186
183
186
185
187
186
187
187
184
186
186
187
189
188
HO N02
Back- Back-
ground ground
(uc/m3) (m/m3)
0 20
0 20
0 30
0 30
0 20
0 20
0 10
0 20
0 20
0 30
0 20
0 20
0 30
0 30
0 30
0 30
0 30
0 25
0 25
0 30
0 10
0 10
0 10
0 20
Ambient Conditions
Out-
door Hind
Temp Speed
CO (m/.ec)
12 8
31 6
29 S
28 5
27 6
27 6
25 6
26 7
23 5
25 6
24 4
23 0
23 3
23 3
23 6
23 3
25 0
27 3
28 2
30 4
30 S
31 2
33 1
28 0
wind Relative
Dlrec- Humldlt)
C 36
E 35
E 41
E 40
E 44
E 44
E 67
E 57
E 52
E 59
65
C SB
E 65
E 65
S 59
59
SE 54
SE 51
SE 44
SE 41
SE 42
SE 32
44
Average
a/ Temperature and pressure at gampIing porca
-------
Table 1-6 LEVELS TEST DATA
Calculated Flouratea and Spike Level a
Room
Time CC\
8-1-74
1800 23 0
1900 23 0
2000 23 0
2100 23 0
2200 23 0
2300 21 0
2400 21 0
8-2-74
0100 23 0
0200 2] 0
0100 23 0
0400 22 5
OSOO 22 0
0600 21 0
0700 22 0
0800 22 0
0900 22 0
1000 23 0
1 100 24 0
1200 23
1100 23
1400 23
1500 21
1600 23
HO? Sampling SyB
Flow
Bar Flow remp -
(m Hill (m. Hal CO
736 9 23 0
716 9 21 0
735 9 23 0
736 9 21 0
716 9 23 0
736 9 21 5
716 9 22 0
716 9 22 0
716 9 22 0
716 9 22 0
736 9 22 0
716 9 22 0
716
736
736
716
736
716
715
735
715
715
735
22 0
22 0
22 5
22 5
22 5
23 0
23 5
23 5
23 5
23 5
23 5
cem Data
Flow- Vencurl
('/mini (mm H-01
lO 0 275
60 275
60 275
60 275
60 275
60 281
61 279
61 285
61 291
61 291
61 291
61 291
61 289
61 8 289
61 4 282
61 2 279
60 7 278
61 285
60 279
60 276
59 271
59 271
60 2 276
"2
tcc/nln) CC1
200 25 1
200 25 1
200 25 1
200 25 1
200 25 1
200 25 1
200 25 1
200 25 1
200 25 1
200 25 1
200 23 1
200 25 1
200 25 1
200 75 1
200 25 0
200 25 1
200 25 1
200 25 1
200 25 1
200 25 1
200 25 1
200 25 1
200 25 1
Flowrace
Vencurl Mecer
Co 21' Co 21*
(I/nlnl (X/mlnl
59 8 60 1
59 8 60 2
59 7 60 4
59 6 60 6
59 8 60 8
60 6 61 0
60 5 61 3
61 2 61 9
61 61 7
61 61 4
61 61 9
61 61 9
61 61 7
61 61 9
60 61 5
60 61 3
60 60 8
61 61 4
60 60 S
59 60 1
59 59 7
59 59 9
59 60 2
Average Flovrace
Co 21
(t/aln
59 9
60 0
60 1
60 2
60 3
60 8
60 9
61 5
61 8
61 7
61 9
61 9
61 7
61 8
61 1
60 8
60 S
61 2
60 3
59 9
59 5
59 S
60 0
ml
1 ('/mini
62 3
62 4
62 5
62 6
62 7
63 1
61 1
61 8
64 0
61 9
64 1
64 1
61 9
64 0
63 4
63 1
62 8
63 6
62 9
62 5
62 1
62 1
62 5
Spike
Level
Ambient*-'
47 7
47 6
47 5
47 5
47 4
47 1
47 1
46 6
46 4
46 5
46 1
46 1
46 5
46 4
46 8
47 0
47 3
46 7
47 2
47 5
47 8
47 9
47 5
SO S02
Back- Back-
ground ground
(11 g/iiil) (u,/,3)
0 20
0 30
0 40
0 40
0 40
0 40
0 40
0 60
0 (0
0 50
0 40
0 30
0 40
0 40
0 50
10 60
0 30
0 25
0 15
0 15
0 15
0 10
0 20
Out-
door Ulnd
Temp Speed
( *C) (n/secl
24 7
26 0
24 0
23 0
22 2
22 0
19 0
21 0
21 0
20 0
20 0
20 0
20 0
21 0
21 0
22 0
23 4
28 4
28
29
29
28
28
Uind Relative
llrec- Mu-nldltJ
ESE 66
57
66
74
SE 73
72
90
77
81
85
85
80
80
85
83
77
NU 69
SU 45
SU 45
SU 45
SU 45
SU 100
SU 52
£/ Temperature and pressure at • a tiling port!
-------
.1 Pc*iro>-Vi..u "•"' " AN !.:.«• 1C' : NAMe A.'.'O AU.*iiLS5
Midwest Research Institute
425 Volker Boulevard
Kansas City, Missouri 64110
. :.(,• ;c,v_ r;r ccj^r DA' .\
I/1, i. r, • ! li.i.- • \ ' • . I- >' • '•<
EPA-650/4-75-011
I TL^ -^'.'J SOPTITLE
"COLLABORATIVE TEST OF THE CONTINUOUS COLORIMETRIC
METHOD FOR MEASUREMENT OF NITROGEN DIOXIDE IN
_ AIRi'.
Paul C. Constant, Jr., Michael C. Sharp,
and George W. Scheil
~12 IPONSORir.'S ACiENCY NAML AND ADDrLSS
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
;i in cini NI > Ai.cir.s,iov. 0
3 RCl'ORT DATt
14 May 1975
8 PERFORMING ORGANIrflAI ION RCI'GHT f.i'
10 PROGRAMtLfrVbNTNO
1HA327
II. CONTRACT. GRAM NO
68-02-1363
13. TYPE OF REPORT AND Pl-PlOD COV r.r1.- D
14. SPONSORING AGfcNCv CODE
15 SUPPLL.'T f»T.\RY NOTES
1C ASS1 RACr
A report on the collaborative test, by 10 collaborators, of the
"Tentative Method for the Determination of Nitrogen Dioxide in the
Atmosphere (Continuous-colorimetric Method)" to determine the
precision and bias of the method. The report covers the N02 , ambient-
air sampling system, test site, selection of collaborators, statistical
design, collaborators field sampling, statistical analysis of
collaborators' results, conclusions and recommendations.
17.
KEY V.ORDS AND mCLlMENT ANAL\S>S
DESCRIPTORS
Air Pollution
Nitrogen dioxide
Design
Data
Statistical analysis
l> IDENTIFlcFiS/Or-EN ENDED TERMS
Open-Ended
N02» ambient-air sampling
system
N02-atmospherfc generatior
system
Collaborative test
Continuous-colorimetric me
COSA D 1 uli! On-lip
13B
14B
7C
thod
it
IB bLCUHITY CI.AiS i ...n
Unclassified
21 NO A ( f. in 2^'u
73'
------- |