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'

-------