EPA-650/4-74-023
June 1974
Environmental Monitoring  Series

                                                               \
                                                               01
                                                               O
                                            ift&iJii&i**!)^

-------
                                       EPA-650/4-74-023
COLLABORATIVE  STUDY  OF  METHOD  104  -
 REFERENCE  METHOD FOR  DETERMINATION
           OF BERYLLIUM  EMISSION
        FROM  STATIONARY  SOURCES
                         by

             Paul C. Constant, Jr. , and Michael C. Sharp
                  Midwest Research Institute
                    425 Volker Boulevard
                  Kansas City , Missouri 64110
                   Contract No. 68-02-1098
                     ROAP No. 26AAG
                  Program Element No. 1HA327
              EPA Project Officer:  M. Rodney Midgett

        Quality Assurance and Environmental Monitoring Laboratory
              National Environmental Research Center
            Research Triangle Park, North Carolina 27711
                      Prepared for

             OFFICE OF RESEARCH AND DEVELOPMENT
            U.S. ENVIRONMENTAL PROTECTION AGENCY
                  WASHINGTON, D.C.  20460

                       June 1974

-------
This report has been reviewed by the Environmental Protection Agency
and approved for publication.  Approval does not signify that the
contents necessarily reflect the views and policies of the Agency,
nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
                                  11

-------
                                 FOREWORD

          The collaborative study of "Method 104 Reference Method for De-
termination of Beryllium Emissions from Stationary Sources," was conducted
under Tasks 1 and 2 of EPA Contract No. 68-02-1098, which is Midwest Research
Institute Project No. 3814-C, entitled "Standardization of Stationary Sources
Emission Measurement Methods."  Midwest Research Institute performed an in-
house evaluation of Method 104, acquired a sampling location and field facili-
ties for the test, performed a preliminary test at this location and analyzed
its results, selected four collaborators to perform sampling according to its
plan of test, retrieved field data and analysis results from the collaborators,
statistically analyzed the results, and prepared this two volume report.

          This volume, Volume I, of the report of test, summarizes MRI's and
the collaborators' activities.  It presents the general plan of sampling and
analysis--covering the selection of the test site, the experimental design,
the selection of collaborators, the processing of samples and data, the general
approach to the analysis of results and the test schedule.  This is followed
by discussions of site modifications, preliminary sampling and analysis by
MRI, the field test, the collaborators' analyses of their samples and data,
MRI's statistical analysisof the results of the test, conclusions and recom-
mendations.  Appendices contain a copy of the write-up of Method 104, and
information from MRI field log.

          Volume II of this report of test contains a summary of velocity
profile data and velocity profiles, computer results of MRI's preliminary
test, and the collaborators' results of their analysis of beryllium test
samples and standard samples (prepared by NBS), and copies of the collabora-
tors' field data sheets for the 13 runs.

          The four organizations that participated as collaborators under
subcontract to MRI in the test of Method 104 were the Ball Brothers Research
Corporation, Boulder Colorado; the Colorado School of Mines Research Institute,
Golden, Colorado; Coors Spectro-Chemical Laboratory, Golden,Colorado; and
Entropy Environmentalists, Inc., Research Triangle Park; Hazen Research, Inc.;
and Southern Testing and Research Labs, Inc.; performed the chemical analysis
of samples for Ball Brothers Research Corporation and Entropy Environmentalists,
Inc., respectively.

                                      iii

-------
           The  following  individuals of these organizations are acknowledged
 for  their  participation  in the collaborative test:

           Ball Brothers  Research Corporation:  T. Beale, C. Dodge, S. Harmon,
 and  Dr. George E. McVehil.

           Hazen Research, Inc.:  Dr. Mark A. Peters.

         •Colorado School of Mines Research Institute:   Bob Cowan,  E.  F.  Davis,
Dr. David E. Hyatt, Carl Pearse,  and R. W.  Whitacre.

          Coors Spectro-Chemical  Laboratory:  Dan Briggs,  Glyndon Mondy,
 Fred Ranta, and Frank B. Schweitzer.

          Entropy Environmentalists, Inc.:   Anton S. Chaplin,  Roy Doster, and
Dr. James Grove.

          Southern Testing and Research Labs,  Inc.:   Mrs.  Evelyn Brady.

          Special acknowledgements are made to the Coors Porcelain  Company,
Golden, Colorado, and Mr. D.  G. Phillip of this company for providing  the
 location for the collaborative test; to the Coors Spectro-Chemical  Laboratory
 for providing field laboratory facilities for the collaborators and'MRI;  and
 to Dr. John B. Clements, Chief, Methods Standardization Branch, National
Environmental Research Center, Environmental Protection Agency, and Rodney
Midgett, Government Project Officer, Methods Standardization Branch, for
their valuable suggestions in planning and design; and  to  the  National Bureau
of Standards and Dr. John Taylor  of NBS for supplying standard beryllium
samples.

          The MRI program is  being conducted under the  management and  techni-
cal supervision of Paul C. Constant, Jr., Head, Environmental  Measurements
Section of Midwest Research Institute's Physical Sciences  Division, who is
the program manager.  Mr. William Maxwell was  MRI's  field  supervisor at
Golden, Colorado, during the  3-week collaborative test. Mr. Maxwell and
Mr. George Cobb performed MRI's preliminary sampling.  Mr.  Thurman  Oliver
performed MRI's laboratory analyses of its  preliminary  test samples and of
 its standard samples.  Mr. Michael Sharp was responsible for the  experimental
                                      iv

-------
design and statistical analyses.  Miss Christine Guenther was responsible for
MRl's data processing.
Approved for:

 EDWEST RES
TUTE
H. H. HubWard, Director
Physical Sciences Dflyision
4 October 1974

-------
                            TABLE OF CONTENTS


                                                                     Page

Summary 	    1

I.     Introduction 	    5

II.    Plan of Sampling and Analysis	    7

         A.  Selection of Test Site	    7
         B.  Experimental Design	    9
         C.  Selection of Collaborators 	   13
         D.  Sample and Data Processing	   13
         E.  Data Analysis Approach	   14
         F.  Test Schedule	   16

III.   Site Preparation	   17

IV.    Preliminary Sampling and Analysis by MRI	   19

V.     Collaborators' Field Sampling	   23

         A.  Process Sampled	   23
         B.  Orientation of Collaborators 	   24
         C.  Sampling Location	   24
         D.  Sampling Equipment Used	   25
         E.  Conduct Collaborative Test	   27

VI.    Analyses of Samples	   33

         A.  Analysis Equipment Used by the Collaborators	   33
         B.  Analysis Procedure 	   33
         C.  Problem Areas	   36
         D.  Results of Collaborators' Analyses 	   37
                                     vii

-------
                     TABLE OF CONTENTS (concluded)


                                                                     Page

VII.   Statistical Analysis of Sampling Results 	   47

         A.  Field Test Samples	   47
         B.  Standard Samples 	   54
         C.  Velocity Profiles	   58

VIII.  Conclusions	   63

IX.    Recommendations	   65

Appendix A - "Method 104--Reference Method for Determination of
               Beryllium Emissions from Stationary Sources" ....   67

Appendix B - MRl's Field Log	   73
                                     viii

-------
                            LIST OF FIGURES







No.                              Title                             Page




 1     Test Location	   10




 2     Modifications for Test Location 	   10




 3     Sampling Point Configuration	   11




 4     Photographs of the Test Site	   18




 5     Velocity Contour Profile of Run 3 of the Preliminary
6
7
8
Velocity Contour Profile of Run 4 of the Preliminary
Test (ft/min) 	
Method 104 Beryllium Sampling Train 	

21
26
28
                                    ix

-------
                           LIST OF TABLES


No.                              Title                              Page

 I      Type of Results Required from a Run	   15

 II    Results of Preliminary Sampling by MRI	   22

 III    Summary of the Results of the Collaborators' Analyses  ...   38

 IV    NBS's  Summary of Analytical Results  of  Its Standard
         Beryllium  Samples  	   44

 V      Results of MRI's Analyses of BeO Filters, Suspended BeO
         and  Be  Solutions by Atomic Absorption Spectrophotometry  .   44

 VI    Results of Analysis of NBS Beryllium Samples by
         Collaborators and MRI	   45

 VII    Beryllium Emission Rates (g/day) from Runs 1 through 13  ..   48

 VIII   Standard  Deviation Versus Level of Beryllium	   48

 IX    Beryllium Data Transformed	   49

 X      Analysis  of  Variance of Beryllium Emission Rate  	   49

 XI    Error  Components in Beryllium Emission  Rate (g/day) ....   50

 XII    Ratio  of  Mass of Beryllium Collected:   Filter/Solution.  .  .   52

 XIII   Beryllium Loading  (stack Conditions  (ug/m3) 	   53

 XIV    Analysis^of  Variance  for Beryllium Loading  (Stack
          Condition) (ug/m3)	   53

-------
                       LIST OF TABLES (Concluded)


No.                              Title                           Page

XV      Standard Sample Results,  Response = Reading - True
          Value	55

XVI     Analysis of Variance for Beryllium Standards	56

XVII    Mean Square Errors in the Collaborators' Measurement of
          Standard Samples Standards	58

XVIII   Identification of Sampling Points 	  59

XIX     Analysis of Variance for Velocities 	  60

XX      Comparison of Adjusted Versus Unadjusted Velocities Per
          Sampling Point	61
                                     xi

-------
                                  SUMMARY

           The Coors Porcelain Company Building No.  16,  Golden,  Colorado,  was
selected, based on eight criteria from 74 companies screened as the site for
the collaborative test of "Method 104--Reference Method for Determination of
Beryllium Emission from Stationary Sources," which is published in the Federal
Register. JJ&, No. 66, Friday, 6 April 1973.  The process sampled was the
manufacture of different beryllium ceramic products, such as substrates and
other items for the electronics and medical fields, chemical laboratories,
etc.  The process involves machining, grinding, blending, priming, forming
and polishing.  Air from the process is continuously exhausted through a
series of HEPA filters before entering the 3 ft x 5 ft stack from which
sampling was done simultaneously by four collaborators.

          MRI performed preliminary sampling at the test location which
verified the site and the experimental design for the collaborative test.
This experimental design was:
                         Xijk =V> +Ci +Rj +CRij


where     ^ = mean,

         Ci = ith collaborator (i = 1, . . . ,4),

         Rj = jth replicate (j = 1	14)  and

     ek(ii) = tne measurement error associated with the k^b observation in
                the ij cell.

Since k = 1, the CR interaction term (39 degrees of freedom) can be used as the
error.

          This collaborative test comprised 13 runs, each on a
different day, where four different collaborative organizations sampled
simultaneously over the same 30-point traverse, with each point being

-------
sampled 8 min by each collaborator.  The emission levels of beryllium in
the stack sampled were low, being in the neighborhood of one-tenth that
of the permissible standard emission rate.  In two cases, the collaborators
subcontracted the chemical analysis of their samples.  It is probable that
one of the subcontractors had not analyzed these types of beryllium samples
before by the procedures specified in Method 104.  Collaborators deviated
from this method with probably adverse consequences.

          The four collaborators selected—Ball Brothers Research Corporation,
Colorado School of Mines Research Institute, Coors Spectro-Chemical Laboratory,
and Entropy Environmentalists, Inc.--collaborated in the 3-week collaborative
test that took place 3-21 December 1972.

          Three types of samples were prepared by the National Bureau of
Standards specifically for this collaborative test:  filters with BeO, ampules
with suspended BeO, and ampules with soluble Be in 0.25 MHC1.  These samples
were given to the collaborators at Golden, Colorado.

          The collaborators analyzed the test samples and standard samples
at their home laboratories.  The results of the collaborators were submitted
to MRI who checked their calculations and found no significant errors.

          There were three analyses performed.  The primary one was a two-
way analysis of variance to obtain the variance of repeated observations per
collaborator, ae , and to obtain the variance between collaborators, ac2.
The analysis was done using the collaborators beryllium emission rate results.
A secondary analysis was the same except beryllium-loading results were used
in place of the emission rate results.  The third analysis, which is also a
secondary analysis, was to determine if the average velocity per sampling
point per run correctly represented the geometrical variance in velocity
throughout the test run even though they were measured at different times.

          Pertinent results from these analyses are:

          1.  From the test samples:

               a.  The collaborator-to-collaborator variability (38% coeffi-
cient of variation) and the measurement variability (44% coefficient of
variation) are approximately the same size and relatively large,

               b.  At a given level,  measurements taken by the collaborators
had a coefficient of variation of approximately 58%,

               c.  The daily fluctuation in beryllium emission rate at the
sampling location was very large (range:  0.25 to 2.7 g/day) compared to
observational errors,

-------
               d.  The collaborators did not differ significantly in the
amount of beryllium collected per run on their filters,

               e.  Almost all the differences in the collaborators' deter-
minations of the amount of beryllium collected during a run are due to the
differences in the solution (wash) portion of the samples, which included
the impinger contents,

               f.  The measurement error for the solution determinations
of beryllium is relatively larger than the measurement error in the filter
determinations,

               g.  On the average, 77% of the beryllium collected was from
the solution portion of the sample.  (There were two parts to a sample,
the solution portion and the filter catch.)*

               h.  Collaborators were more repeatable and more consistent
in determining beryllium from the filter, and

               i.  Collaborators' relative precision in the measurement
of beryllium from standard samples was considerably greater than for their
test samples, but the standard samples contained larger amounts of beryllium.

          2.  From the standard samples:

                                                 2       2
               a.  The components of variance, a   and CT  , are about the
same size (CV s 10%),

               b.  Taking into account the type of sample variance, the
CV at a given level is about 25%,

               c.  In general, the bias is proportional to the beryllium
level,

               d.  Usually a negative bias (about -20% on the average)  was
observed,

               e.  Generally, the bias was quite large compared to the
measurement variance, and

               f.  The average bias on the filter samples was essentially
zero, but only because large negative and positive biases canceled out.

          3.  From the profile analysis, it was determined that a profile
plotted from the average determinations of the four collaborators correctly
represented the geometrical variances in velocity throughout the run.

-------
          The major conclusions that can be drawn from this collaborative
test are:

          1.  Using Method 104, the bias and the measurement error will
be quite significant,  with the bias contributing more to the unreliability
of the beryllium determination.  At a given level,  measurements will have a
coefficient of variation of approximately 58%.

          2.  Method  104 is adequate as written if a coefficient  of variation
of approximately 58%  for the measurements at a given level of beryllium is
acceptable.

          3.  The  lack of precision in the results of the test samples is
exclusive of the amount of beryllium collected on the filters since these
measurements were  repeatable by a collaborator and 77% of the error was in
the beryllium measurements from the solution portion of the samples in contrast
to the  filter portion.

          4.  The  biases of the solution portion of the samples are unpredict-
able.

          5.  There is not sufficient information from this test to determine
whether  the bias problem is due to field sampling, laboratory analysis, or
both.

          Based upon  the conclusions that have been drawn from the results
of this  collaborative test, it is recommended that an investigation be under-
taken to determine the reasons for the significant biases.  If all collaborators
had measured high or  low by some significant amount, this investigation might
not be needed, but because of the amount of bias and because the bias is not
systematic—at times  it depends on level, at other times it does not, etc.--
and because it is  not known whether the bias is due principally to field
sampling or the analysis procedures, the investigation is needed.

-------
                              I.   INTRODUCTION

          The Methods and Standardization Branch, National Environmental
Research Center of the Environmental Protection Agency (EPA)  is engaged in
a program to evaluate methods, recommended and promulgated by EPA, for the
measurement of pollutant emissions from stationary sources.  Midwest Research
Institute (MRI) is working for EPA under Contract No. 68-02-1098 to provide
EPA data on the reliability and bias of the methods.

          To achieve its objective, MRI plans and executes a collaborative
test and evaluation for each method submitted to it by EPA.  Briefly, MRI
does, in the execution of a collaborative test of a method, perform an in-
house evaluation of the method (which could range from a paper evaluation to
a ruggedness test), provide sampling locations and facilities, select and
acquire collaborators, orient the collaborators\relative to the test and
analysis involved, coordinate the collaborative test, retrieve field data
and results of the collaborators' chemical analyses of their samples, statisti-
cally analyze results received from the collaborator, and report  results to
EPA.

          The work activities described above were performed by MRI on its
first test undertaken on the contract.  The method under investigation was
"Method 104--Reference Method for Determination of Beryllium Emissions from
Stationary Sources," which is given on pages 8846 through 8850 of the Federal
Register. ^8,, No. 66, Friday, 6 April 1973.  (A copy of Method 104 is given
in Appendix A.)

          This report covers the collaborative test of Method 104 in the
following order:  Section II discusses the general plan for sampling and
analysis--covering the selection of test site, the experimental design, the
selection of collaborators, the processing of samples and data, the general
approach to the analysis of test results, and the test schedule; Section III
summarizes the site modifications requirements; Section IV covers preliminary
sampling performed at the test site by MRI and the subsequent analysis of the
field samples taken; Section V covers the collaborative field test; Section VI
covers the collaborators' analysis of their field samples and standard samples
prepared by the National Bureau of Standards (NBS) presenting their results;
Section VII presents a discussion of the statistical analyses of test

-------
results; Section VIII presents conclusions MRI drew from its work; and
Section IX presents recommendations that are based upon the conclusions
drawn.  Appendices contain detailed information such as a copy of the write-
up of Method 104 and MRl's field log.

-------
                    II.  PLAN OF SAMPLING AND ANALYSIS

          The general approach used to prepare for the collaborative test of
Method 104 is presented in this section.  This plan comprises the selection
of a test site, preparation of an experimental design, selection of col-
laborators, processing of field samples and data,  analysis of results of the
test, and the test schedule.  Any deviations that  were made from the planned
activities are either discussed in this section or reference is made to other
sections of the report.

A.  Selection of Test Site

          A search was made for companies involved with the mining or pro-
cessing of beryllium and beryllium alloys.  Seventy-four  were identified as
being potential test sites.  These 74 companies were screened  using the
following criteria:

          1.  Continuous emissions,

          2.  Emissions significant enough to obtain adequate samples in a
reasonable time,

          3.  Test locations that would require a  minimum amount of modifica-
tions,

          4.  Companies that were desirous of their plant being a test site,
and those that had cooperative personnel with whom to work,

          5.  Location of site such that logistics involved by potential col-
laborators would be reasonable, and weather conditions during test period
that would not be objectionable, causing a postponement of testing or long
delays,

          6.  Availability of test location at a site,

          7.  Plant operational restrictions, and

          8.  Available facilities for use by test crews.

-------
           After  visiting  the  Coors Porcelain  Company, Golden, Colorado,  it
 was  selected  as  the  site  for  the  collaborative  test of Method 104.  The  princi-
 pal  reasons for  its  selection are:

           1.  The processes that  feed this stack are continuous; the exhausts
 from these processes are  fed  through a series of HEPA filters before entering
 the  stack; the stack flow rate  is approximately 20,000 scfm; and the beryllium
 concentration is in the neighorhood of 0.6 P-g/m^.

          2.  The test location (see Figure 1) was a 3 ft x 5 ft rectangular
 stack approximately 30 ft high  that is located at the northeast corner of the
 processing building  (No.  16), a part of Coors Porcelain Company, 600 Ninth
 Street, Golden,  Colorado.  The  principal modifications and equipment setups
 required of the  test locations  to ready them for collaborative testing were
 an 8-ft extension of the  stack, with 10 2-in. ports in this extension,
 scaffolding on the north  side of the stack (see Figure 2), a wooden platform
 to the south  side of the  stack  affixed to the roof and stack, and an appropri-
 ate  ladder.

          3.  The Coors Company was quite willing to provide a test location
 for  the collaborative testing of beryllium; considerable testing has been done
 at this plant; the Spectro-Chemical Laboratory is quite knowledgable of Method
 104, has the  personnel and equipment to perform source emission testing, has
 done such testing, and this laboratory, which is approximately 1-1/2 miles from
 the test site, had a laboratory area for the collaborators at which they could
 prepare their equipment and samples.

          4.  The location of Coors is quite good relative to the geographic
 locations of  potential collaborators.

          5.  There is no known limitation on availability of the test loca-
 tion.  The operation is on a 24-hr basis, with the 7:00 a.m. to 3:30 p.m.
 shift being the principal one.

          In  summary, the test  location shown in Figure 1, with modifications
 shown in Figure 2, would provide a good test site.  The extension to the stack
 allowed for sampling to be done at approximately 1.5 diameter upstream, and
 5.5 diameter  downstream from any flow disturbance--a sampling configuration
 as shown in Figure 3.  The continuous emissions of concentration 0.6 ug/m3
would allow for an adequate sample of beryllium to be subsequently analyzed
 according to  the procedures given in Method 104.

-------
B.  Experimental Design

          The goal of the collaborative test was to obtain sufficient sig-
nificant results so that the reliability and the bias  of Method 104 could be
determined.  A major element of the collaborative test was to have an experi-
mental design that would enable this goal to be met.  Considerations that
formed the bases of this design, which is given later  in this section in a
formal manner, are:

-------
^^-^^^ 4_
•

Building 'No. 16
Coors Porcelain Co.
600 9th St.
Golden, Colorado





^









J
A




~30'



i



<;««•







t inn A
n

5


i

_ A
                      Figure 1 - Test Location
Roof
                    Extension
                                              5 Ports on North
                                              and South Sides
                  Ladder
          Figure 2 - Modifications  for Test Location
                               10

-------
                                     I
                                             South
   I  Sampling  Ports,
     nos. 1 thru  10.

     Col laborators;
      a, b, c, & d.

     Sampling  Ports
        1 Thru 30
   Cross Section  of Stack
1,10 ,,9 ,,8 ,,7
6 .
5 .
4 .
3 .
2 .
I .
12 .
11 .
10 .
9 .
8 .
7 .
18 .
17 .
16 .
15 .
14 .
13 .
24 .
23 .
22 .
21 .
20 .
19 .
*-1.0'-^
,,6
30 .
29 .
28 .
27 .
26 .
25 .

0.5'
•

                            I
                            t
I
^3
"4
                                             North
                   Figure 3 - Sampling Point Configuration


          1.  There would be a minimum of four collaborators,

          2.  These collaborators would sample simultaneously,

          3.  Sampling would be done strictly according to Method 104,

          4.  The beryllium content of the effluent gas stream could vary
with location,

          5.  Concentrations of beryllium emitted from the stack could vary
from day to day,

          6.  The duration of a run would be sufficiently long to provide
an adequate sample for chemical analysis by atomic absorption techniques,

          7.  The level of emissions cannot be altered by spiking the gas
stream or by a reduction of the production,
                                     11

-------
          8.  A collaborator, as per the method, could analyze each sample
in parts—filter and solution catches separately—or combine the two for one
determination,

          9.  The field test would be accomplished in 3 weeks--15 working
days, and

         10.  Standard samples prepared by the National Bureau of Standards
(NBS) would be given each collaborator for analysis.

          It was desirable to execute the method in realistic fashion, i.e.,
to accumulate the determinations over 30 points within the stack (which are
the number required because of the geometry of the stack and port locations
in the stack), and to have a design so that occasional missing data would
not be disastrous.  On the other hand, it was also desirable to be able to
account for nonuniform beryllium concentrations due to time passage and/or
location within the stack.  The time dependence and inhomogeneous distribu-
tion of beryllium within the stack could only be examined by incorporating
these variables in an experimental design.  This meant making an observation
on less than 30 points.

          The experimental design was:

                   Xijk=* = Ci+Lj +CLij+ek(ij)>

where     p = mean,

         C. = ith collaborator (i = 1, .  . .  ,4),

         L.. = jth level (J • 1,  .... 14).

     ek/--\ = the measurement error associated with the kth observation in
        1J      the ij cell,

and, since k = 1, the CR interaction  (39 d.f.) can be used as the error.

          The data can be considered as a factorial design with factors:
collaborator (4), replicate  (up  to 14 at one 4-hr run per day), and "run"
(solution result and filter result--although  these will differ), it is
nevertheless convenient to incorporate all the data into one framework.

          It is necessary to include replicate as a factor because the beryl-
lium concentration may vary from day to day.

          The beryllium concentration could also vary with location, i.e.,  it
could make a difference to the result to change ports.  Therefore, each col-
                                      12

-------
 laborator was ultimately assigned to approximately the same ports as every
 other  collaborator.

C .  Selection of Collaborators

          A principal activity was to compile a list  of potential collaborators
and from this list select four to perform the testing according to Method 104.
Information was obtained from EPA and from MRI's files to compile a list of
over 20 potential collaborators.  These organizations were screened via tele-
phone calls to achieve a list of 18 candidate collaborators.

          A Request for Proposal (RFP)  was sent by MRI to these 18 candidate
collaborators.  This solicitation resulted in proposals being received from
eight of this group.  These proposals were evaluated  and each was rated, con-
sidering technical capability and cost.  The four selected were:

          Ball Brothers Research Corporation (formerly Sierra Research
            Corporation) ,

          Colorado School of Mines Research Institute,

          Coors Spectre-Chemical Laboratory, and

          Entropy Environmentalists, Inc.

Contract negotiations were conducted with these organizations after their
selection was approved by EPA and a task order for the test was received by
MRI.  Hereafter, these collaborators will be referred to by randomly assigned
code  descriptors—collaborator 1, collaborator 2, collaborator 3, and
collaborator 4.

D.  Sample and Data Processing

          There were five types of beryllium test samples involved that were
analyzed by the collaborators.  Two were field samples:  the filter sample
 (filter with catch and loose particulate matter from the filter holder) and
the acetone-wash sample (wash of probe and filter holder and impinger contents1)
to be taken with each 30-point run, according to Method 104.  The other three
types of samples were standard samples that were prepared by the National
Bureau of Standardsi' (NBS) specifically for this collaborative test:
_!/  "Preparation of Reference Materials for Stationary Source Emission
      Analysis:  Beryllium," by T. C. Rains, C. D. Olson, R. A. Velapoldi,
      S. A  Wicks, 0. Menis, and J. K. Taylor, NBSIR 74-439, March 1974.
                                      13

-------
                                  Type 2
                                Type 3
Filter, Blank

Filter, Level 1 of BeO


Filter, Level 2 of BeO


Filter, Level 3 of BeO
Ampule X, Level 1 of   Ampule Y, Level 1 of
  Suspended BeO          Soluble Be in 0.25 M HC1

Ampule X, Level 2 of   Ampule Y, Level 2 of
  Suspended'BeO          Soluble Be in 0.25 M HC1

Ampule X, Level 3 of   Ampule Y, Level 3 of
  Suspended BeO          Soluble Be in 0.25 M HC1
          Samples taken in the field were to be prepared after the run for
transfer to the collaborator's home laboratory for analysis.  This preparation
was to be done according to the procedure given in Method 104.

          Raw data taken by each collaborator in the field were to be recorded
in duplicate by the collaborators on forms that were supplied them by MRI.
One set of these forms was to be given to MRI, and the other set to be
kept by the collaborator.

          After the collaborators processed their field samples at their
home laboratory according to procedures of and performed the  calculations
required by Method 104, the 37 items of information identified in Table I
were to be submitted to MRI for each of the test runs.

          The NBS standard samples were to be analyzed according to Method
104 and the results presented to MRI in tabular form, accordingly:
       Sample Identification
              Number
           Type Sample
Measured Be
   in Mg
          The MRI field test coordinator was to keep a daily log which would
contain sailent observations on the operations of the collaborators as well
as weather data.
E .  Data Analysis Approach

          Each collaborator was to perform the appropriate calculations to
obtain beryllium emissions results from each run (see Table I).   Then raw data
and the results of the chemical analyses of samples were to be  provided MRI,
and it was to check the calculations of the collaborators.  This check was
to determine if there was any gross systematic calculation errors.
                                      14

-------
                                                   TABLE I
                                     TYPE OF RESULTS REQUIRED FROM A RUN
 1.   Run No .
 2.   Date
 3.   Time run began
 4.   Time run ended
 5.   Net time of test
 6.   Barometric pressure,  in. Hg  absolute
 7.   Standard pressure,  in. Hg
 8.   Standard temperature,  °F
 9.   Meter orifice  pressure drop, I^O
10.   Volume dry gas at meter conditions, ft-*
11.   Average gas meter temperature,  °F
12.   Volume dry gas at STP, ft^
13.   Total water collected, ml
14.   Volume water vapor  at  STF, ft3
15.   Stack gas moisture, % volume
16.   Assumed stack  gas moisture,  % volume
17.   Percent C02
18.   Percent 02
19.   Percent N2
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
Percent excess air
Molecular weight of stack gas, dry
Molecular weight of stack gas, stack conditions
Stack gas specific gravity, reference to air
Average
         >/velocity, head of stack gas,
Average stack gas temperature, °F
Pitot correction factor
Stack pressure, in. Hg absolute
Stack gas velocity at stack conditions, fpm
Stack area, ft2
Stack gas flow rate at STP, scfm
Sampling nozzle diameter, in.
Percent isokinetic
Beryllium collected - filter, u.g
Beryllium collected - solution, p,g
Beryllium collected - total, |ig
Beryllium collected - total, iig/m3
Beryllium emitted per 24-hr day, g

-------
          After this preliminary check, the results of the collaborators
were to be analyzed statistically.  Fart I of this analysis covered the field
sampling results.  It was a two-way analysis of variance to provide the
variance of repeated sampling by each collaborator, and the variance between
collaborators.  Fart II of the analysis covered the results of the collaborators'
chemical analyses of the standard samples.  The primary objectives of this
analysis were to provide estimates of variance of repeated observations by
a collaborator and the variance between collaborators, and to examine the
biases present in the measurement of beryllium.

          An important part of the statistical analysis was to determine
problem areas of Method 104, and to place these generally, or if possible
specifically, in one or both of the following areas:  field sampling and
analysis.
F.  Test Schedule

          There were two principal test periods--preliminary and collabora-
tive.  The preliminary test was conducted by MRI (see Section IV) to deter-
mine the readiness and suitability of the site for the collaborative test.
The schedules of these tests were:
                           Preliminary.Test by MRI
     Run 1
     Run 2
    6 November
    7 November
   1973
   1973
           Run 3 -
           Run 4 -
8 November
9 November
1973
1973
                             Collaborators* Test
     Preparations
     Run
     Run
     Run
     Run
     Run
     Run
1
2
3
4
5
6
 3 December
 4 December
 5 December
 6 December
 7 December
10 December
11 December
1973
1973
1973
1973
1973
1973
1973
Run 7
Run 8
Run 9
Run 10
Run 11
Run 12
Run 13
Run 14
12 December 1973
13 December 1973
14 December 1973
17 December 1973
18 December 1973
19 December 1973
20 December 1973
21 December 1973
                                     16

-------
                          III.  SITE PREPARATION

          A line sketch of the sampling location is shown in Figure 1.  The
location and geometry of this stack necessitated three principal modifica-
tions to prepare it for the collaborative test:   (1) extension of the stack
with 2-in. diameter sampling ports in this extension;  (2) scaffolding and
platforms; and (3) electrical service.  The Coors Spectro-Chemical Company
furnished laboratory space at its building in Golden approximately 1-1/2
miles from Building 16, and electrical service at the sampling location.
Figure 2 gives a sketch of the modified test site.  Figure 4 provides
photographs of the test site in use.

          This 3 ft x 5 ft rectangular stack gives an equivalent diameter of


                         =  2*3*5      -32      = 3 ft 9 in.
              L + W           3 + 5            8
By referring to Figure 101-3 of the write-up of Method 104, which is in
Appendix A, it is seen that an ideal port location would be 8 diameters,
or 30-ft downstream from the 90 degree bend and 2 diameters, or 7-1/2-ft
upstream from the top of the stack.  These dimensions would necessitate
the addition of an approximately 18-ft extension to the stack, but would
require a minimum of 12 sampling points.  A compromise was made by using
an 8-ft extension.  The sampling ports were located approximately 1.5
diameters, or 6 ft from the top of the extended stack, and approximately
5.5 diameters downstream from the 90 degree bend.  This location of ports,
according to Figure 101-3 would require a minimum of 24 sampling points
distributed over the 3 ft x 5 ft cross sectional area.  To provide a better
sampling scheme, 30 sampling points were selected (see Section IV) rather
than the minimum of 24.
                                     17

-------
 Photo 1: Looking generally east
   showing the 28 ft scaffold
   and test platform
 Photo 2: Looking south showing
   wooden rails and sampling
   trains in place
Photo 3:  On test platform looking
  south,  showing shelf where all
  four consoles were operated
Photo 4:  Looking at south ports
  with sampling trains in place
  and probes in the stack
                   Figure 4  -  Photographs of the  Test Site
                                     18

-------
              IV.  PRELIMINARY SAMPLING AND ANALYSIS BY MRI

          After the site was prepared for sampling the first part of November
1973, MRI conducted preliminary sampling according to Method 104.  The test
comprised four runs (see Table II, page 22) each of which was of a different
duration.  Two of the runs consisted of one 6-point traverse from one port,
and the remaining two runs comprised five different 6-point traverses—each
traverse was made through a different port.  The 6-point traverses were made
to determine if sufficient beryllium (see Section II.B) would be collected,
especially on a filter, during a 6-point traverse.

          On Runs 1, 2, and 4, the filter catch constituted a sample and the
acetone rinse of the probe and the filter holder and the impinger contents--
termed solution in Table I--constituted a second sample of a run.  On Run 3,
there were two solution samples; the one termed front was a wash of the
probe, filter holder and connecting glassware; and the other termed back
was a wash of the four glass impingers and connecting glassware that followed
the filter holder in the sampling train setup.  (See Figure 104-1 of the
write-up of Method 104 in Appendix A.)

          These samples were transported to MRI by airplane where they were
chemically analyzed by the procedures given in Method 104.  These results of
the analyses are given in Table II.  These results definitely show that to
provide a sufficient beryllium sample for analysis, the sampling time for a
30-point traverse sample should be 4 hr, or 8 min per point.

          Gas velocity profiles for Runs 3 and 4 are shown in Figures 5 and 6.
These profiles indicate that the gas flow in the stack at the sampling loca-
tion is not turbulent.  These profiles suggest that the beryllium concentra-
tion at any instant in time should be fairly constant throughout the sampling
plane.

          Based upon the results of this preliminary test, the experimental
design given on pages 7-14, which only provides for 30-point sampling, was
selected as the final experimental design for the collaborative test.  The
profiles indicated that the minimum of 24 sampling points (see Section III)
could be used,  but 30 points were selected to provide a better sampling
scheme.

          Computer results  of MRI's preliminary test are given in Volume II

                                     19

-------
NJ
O
                        10
    PORT LOCATION AND NUMBER

987
                 -1500
                                        234

                                            PORT LOCATION AND NUMBER
                                                   1550
                                                                                       1442
                  Figure 5 - Velocity Contour Profile  of  Run 3 of the Preliminary  Test (ft/min)

-------
                                           PORT LOCATION AND NUMBER
                                        987
IS)
                                        234
                                            PORT LOCATION AND NUMBER
                  Figure  6  -  Velocity Contour Profile of Run 4 of the Preliminary Test (ft/rain)

-------
ro
N>
                                                      TABLE II



                                       RESULTS OF PRELIMINARY SAMPLING BY MRI


Run Date
1 9/6

2 9/7

3 9/8


4 9/9

Test
Duration
Sample Time (min)
Filter 1600 24
Solution
Filter 1000 48
Solution
Filter 1400 120
Solution (back)
Solution (front)
Filter 0900 240
Solution

Time of Sampling Beryllium
Traverse s5/ Per Point in Minutes (u.g)
1 4 < 0.07
< 0.07
1 8 <0.07
0.20
5 4 <0.07
0.14
1.38
5 8 0.41
1.29
        a/  Six points  per  traverse.  Runs 3 and 4  had five different traverses for a total of  30  sampling

               points per run.

-------
                      V.   COLLABORATORS'  FIELD SAMPLING

          This section presents information on the process that was  sampled,
the orientation of the collaborators prior to the start of the sampling,  the
sampling locations, the type of sampling equipment that was used by  the col-
laborators and generally how the test was conducted by MRI.
A.  Process Sampled

          At Building No. 16 of the Coors Porcelain Company (see Figure 1
page 10)  where the collaborative test took place, different beryllium
ceramic products are manufactured, such as substrates and other items for
the electronics industry and items used in the medical field, chemical
laboratories, etc.  The processes involved in this manufacturing cover,
such operations as machining, grinding, blending, pressing, forming, polish-
ing, where the grinding operation, perhaps, was responsible for most of the
beryllium dust or particles.

          The exhaust from these processes is fed through a series of HEPA
filters, as shown below, before entering the stack from which the collabora-
tors sampled.
                                                             Stack
                                                      HEPA Filters
        Exhaust from
        All Processes
          The manufacturing operation is continuous, on a 24-hr basis, with
the 7:00 a.m. to 3:30 p.m. shift being the principal one, or the one with
the greatest output.  During this shift the rate of production of items is
nearly constant, with the exception of personnel break periods.
                                     23

-------
B.  Orientation of Collaborators
 i   ^^—^—•    -•

          The orientation of collaborators started with MRl's solicitation
for candidate collaborators and culminated with a meeting at the test site
at the beginning of the first day of the test period.  In each instance the
principal concern was to provide the collaborators, with proper information.
That is, the intent was that the collaborators were to know the purpose of
the test, when and where it was to take place, their responsibilities—em-
phasizing that Method 104 was to be followed explicitly—and to do this without
biasing the collaborators in any way.

          There were three modes of communication used in this orientation
as summarized below.

          1.  Telephone;  There were conversations with the collaborators
(a) during the selection process in which the purpose of the test, etc.,
was given, and at the same time, information was obtained on their qualifi-
cations, and (b) after their selection to participate, there were dis-
cussions about type of equipment the collaborators were going to use, the
test schedule, the fact that Method 104 should be strictly followed, etc.

          2.  Written;  These included (a) the Request for Proposal MRI sent
to candidate collaborators, (b) MRI's letter of acceptance to the collabora-
tors which also included information on the test site, the test schedule,
as well as general test instructions, and (c) a memorandum given to the
collaborators at an on-site orientation meeting and which covered informa-
tion on standard samples, their presentation of data and reporting require-
ment.

          3.  Meetings;   These included (a) a site visit of the Coors Porcelain
Company during the process of the selection of a test site where the purpose
of the test, etc., were discussed, (b) an orientation meeting of the collabora-
tors, which was the first activity of the test and where the test requirements,
field data forms, plant regulations, the aspect of bias, etc., were discussed,
and (c) interaction between the collaborators and MRI personnel during the
test.
C.  Sampling Location

          The location selected in the planning stage of the test was used
for the collaborative test, as well as MRI's preliminary testing.  This
location  is described in Sections II.A and III, and  is  shown pictorially
in Figures  1 and 2 on page  10, Figure 3  on page 11,  and Figure 4  on  page  18.
                                    24

-------
 D.   Sampling Equipment Used

          The sampling equipment used by the four collaborators conformed
 to the requirements of Method  104; in each case the equipment was principally
 that manufactured by the Research Appliance Corporation.  Notations made by
 MRl's field coordinator about  equipment are given in Section A of
 Appendix B.

          Figure 4 on page 18  gives some photographs of the type probes
 (Pyrex inserts), sampling trains and console units that were used.

          MRI had sufficient sampling equipment on site to equip a col-
 laborator and provide others with spare parts and materials if such became
 necessary.  Most of the collaborators had spare glassware and other mate-
 rials of their own.

          The information reported to MRI by the four collaborators
 about the sampling equipment they used in the field is given below.

          1.  Collaborator 1;  "The equipment used for collecting the samples
was an RAC Train Staksamplr, Model 2343 manufactured by Research Appliance
 Company, Route 8, Gibsonia, Pennsylvania 15044."

          2.  Collaborator 2;  "Beryllium sampling was accomplished by
 using the EPA collection train, Method 104, described in the 6 April 1973,
 Federal Register,  It is shown schematically in Figure 7.  The train con-
 sists of (1) a stainless steel nozzle connected to a 3-ft glass probe (2).
Following the probe is a Millipore type A-A filter backed by a Whatman No. 41
 filter, supported on a coarse  fritted glass disc in a glass filter holder.
Enclosing the filter assembly is a heated box (3) to maintain temperatures
 above the condensation point.  An ice bath containing four Greenburg-Smith
 impingers (5) is attached to the back end of the filter by a flexible
umbilical cord (4).  Condensible materials are collected in the impingers.
Leaving the fourth impinger, the sample stream flows through flexible tubing
 (6), a vacuum gage (7), needle valve (8), a leakless vacuum pump (9) in
parallel with a by-pass valve (10) and a dry gas meter (11).  Following this
is a calibrated orifice and inclined manometer (12).  A S-type pitot tube
and inclined manometer (13) measures velocity pressure and stack tempera-
ture is monitored by a thermocouple connected to a potentiometer (14).  Iso-
kinetic sampling rates are maintained through the use of a nomograph which
correlates the proper orifice pressure drop required for given velocity pres-
sure and stack temperature."

          3.  Collaborator 3;  "The equipment used was a glass EPA type
sampling train manufactured by Research Appliance Company as Model 2343-5
and equipped with a Pyrex glass lined 3-ft long probe.   The gas meter and

                                     25

-------
                                              14
                                                 2    :
                                         13
ro
CT"
Stack

Wall
                                              12
                                     Figure 7 - Method 104 Beryllium Sampling Train

-------
orifice meter were calibrated immediately prior to this test series.  The
pitot tube constant was assumed to be 0.85."

          4.  Collaborator 4:  "Main sampling equipment was a Research
Appliance Company Model  "Staksamplr," with a 3 ft effective length probe
(catalogue No. 2343-3).  This included the meter console, sample case,
umbilical cord, probe (Pyrex lined), glassware, and nomograph.  A Dwyer
Manufacturing Company Model 115 inclined Manometer (-0.05 to +0.25 in.
was used for velocity pressure measurements in place of the vertical in-
clined manometer attached to the meter console.  The filter paper used was
Gelman Triacetate Metricel (0.8 urn pore size, 900 mm diameter), catalogue
No. GA-4 plain.  The back-up filter was a Whatman No. 41.  Stack tempera-
tures were measured with a Rochester Model 1748 dial thermometer (25°-125°F),
Acetone used in clean-up of probe, filter holder, and impIngers was from
Fisher Scientific Company, catalogue No. A-18."
E.  Conduct Collaborative Test

          The collaborative test of Method 104 was conducted according to
its experimental design (see Section II.B) and the planned test started on
3 December 1973 and ended on 21 December.  The first activity on Monday
morning, 3 December was an orientation of the collaborators at the
Coors Spectro-Chemical Laboratory building.  After this meeting, the col-
laborators were shown their field laboratory area, and the test location.
The collaborators spent the remainder of the day preparing for the first
run that took place on Tuesday, 4 December 1973.

          1.  Overall operations;  There was one run per work day (Monday
through Friday), starting 4 December, with the exception of 19 December
when a severe snowstorm made sampling impossible.  As a result, 13 runs
of the 14-run test were achieved.

          Sampling started each day generally between 9:00 and 10:00 a.m.
On 2 days, testing started later.  In one case, the delay was due to the
lack of power, and in the other it was due to extremely high winds.  (More
information is given this subject in Section B of Appendix B.)  The four
collaborators sampled simultaneously covering the 30 sampling points (8 min
per point) according to the plan shown in Figure 8.  During each run, each
collaborator made five traverses, sampling each from a different port, as
indicated in Figure 8.  For example, during Run 1, traverse 1 was made by
collaborator C through port 2, collaborator A through port 4, collaborator
B through port 6, and collaborator D through port 8.  Ports 1 and 10 pro-
vided the same traverse (sampling 6 points of the total 30 points that con-
stituted a run), as did 2 and 9, 3 and 8, and 4 and 7, but the points
were sampled in reverse order, depending on whether the probe entered a
north port (1, 2, 3, 4 and 5) or a south port (6, 7, 8, 9 and 10).

                                     27

-------
oo

Porti'
1
2
3
4
5
6
7
B
9
10


Port*/






a
9
10



_ /
Port*'

1
2
3






10
Run 1
Traversed
c a
e a
a e
a c
c a
b d
b d
d b
d b
d b
Run B
Traversed
11245
d c
e d
c d
c d
d c
b a
b a
b a
a b
a b


Rim 11
Traverse^'
1 2 3 4 5

a e
a e
c a
e a
c a
d b
b d
b d
b d
d b
Run 2
Traversa!*'
a c
a c
e a
c a
a c
d b
d b
b d
b d
b d
Jhm B
TraveTBeS'
12345
e J
d c
d c
d c
c d
a b
a b
a b
b a
b a


Bun U
TreveraeE'
12345

d b
d b
b d
b d
b d
a c
e a
c a
c a
a c
Run 3 Rim 4 Run 5 Run 6 Run 7
TraveraaS' Traversed Traverse!!/ Traverse^' Traverse^'
bd cb be ad da
bd cb be ad di
dbbccada ad
db beebda ac
bdcb be ad da
ac da ad be eb
acda ad be cb
ca addacb b I
caaddaeb be
ca da at be ct
Sun 10
Traverse*'
12345 South
b d J" 1-
b d » »
d b
d b nlO ||9 ||8 ||7
d b 6 . 12 . IB . 24 .
c a
a e 5 . 11 . 17 „ 23 .
a c II Sampling fortl
• c nos 1 Ihni 10 4 . 10 . 16 . 22 .
c a
1 Collaborator*; 3 p a . 15 . 21 .
f a. b. e. id.
2 . 8 . 14 . 20 .
Run 13 . Sampling foinK
Traverse*/ 1 fhru 30 | 7 )3 19
12345 ....
II. llj ii. H4
a b ' *
b a Cren Section of Slack i
b ' I"
b a '
a b
d c North
i e
i. c
c d
c d
                             a/ FOTU 1 and 10, 2 and 9, 3 and 8, 4 and' 7. 5 and 6 constitute the  game traverse except Che points along  the traverse are sampled In reverse order.
                             b/ There vere five ttaveraea per run, with each of the four collaborators sampling with five traverses but  In a different order.
                             c/ a, b, c, and d represent the four collabpratores.
                                                                 Figure  8. -  Thirty-Point  Sampling  Flan

-------
          Sampling was performed according to Method 104, with the excep-
tions noted in Section V.E.2 below.  At the end of a run each collaborator
prepared his samples—washed probe, etc., and placed catches in appropriate
containers.  The nonlocal collaborators performed this work at the laboratory
the Coors Spectro-Chemical Laboratory provided.  The local collaborators
prepared their samples at their home laboratories.  Field data were recorded
in duplicate by each collaborator on forms that were provided by MRI.
When completed, one set was given to MRI, the other set was kept by the
collaborators.

          The NBS samples were shipped directly to Golden, Colorado, by
NBS, where MRI packaged them individually by type for each collaborator.
They were packaged well to insure that there would be no loss due to damage
during their transport to the collaborators' home laboratories.  Each pack-
age contained identification information, but at no time was any collabora-
tor made aware of the amount of beryllium that was in any standard sample.
In addition to the identification information, each collaborator was provided
the following information that was furnished by NBS:

          "Notes on Sampling:   Treat each filter as an individual sample
and dissolved in HN03, HC104 and 112804 as described in the Federal Register.
38, No. 66, 6 April 1973."

          "Each ampule is prescarred.  In sampling an ampule, check to see
if the top or neck is free of solution.  If liquid is present in the top or
neck of the ampule, shake or gently tap  the ampule until the top is free
of solution.  Then gently snap the top from the body of the ampule.  Transfer
sample (slurry or aqueous solution)  to beaker or volumetric flask using
disposable pasteur capillary pipet.   Rinse ampule with mineral acid and
proceed with the analysis."

          Each collaborator was given 30 (three of each of the 10 kinds
identified on page 14) of these standard samples during the second test day
in the field.  The local collaborators took their standard samples to their
home laboratory the same day they got them.   The other collaborators shipped
their standard samples by air  along with their test samples.

          The conduct of the test was coordinated by MRl's field coordinator.
He kept a log of field activities.  This log comprises Appendix B.

          Raw data taken by each collaborator in the field were recorded by
them on forms that were provided by MRI.  These data were recorded in
duplicate, with one set given to MRI; the other set was kept by the
collaborators.  The collaborators' field data are given in Volume II of
this report.
                                     29

-------
          2.  Problem areas;  The problem areas that were  indicated by the
 collaborators are given below.  These writeups include deviations  from the
 method as reported by the collaborator.

               a.  Collaborator 1; "On several occasions during the field
 testing  the vacuum in the sampling train would rise and the flow rate would
 decrease as if there were a plug somewhere in the sampling train.  This
 occurred for no apparent reason as it was observed that the millipore filter
 was not  wet.  This situation could only be remedied by replacing the milli-
 pore  filter.

               Membrane filters plug easily with only a small amount of
 moisture present.  On one particular occasion the filter had to be changed
 because  of condensation forming in the filter holder when the sampling train
 was taken from inside a building to the outdoors where the temperature was
 20°F.  This would be a very complicating problem in climates with high
 humidity.

               Also, paragraph 4.5.2 of Method 104 suggests that if the stack
 gas is in excess of 200°F that the filter holder be moved downstream of the
 first impinger.  For this particular collaborative test condensation was
 observed in the sampling train up to the fourth impinger on all occasions."

               b.  Collaborator 2;  "The EPA Method 104 called for a heated
 probe, but due to the low moisture content of the stack effluent, a heated
 probe was not used."

               c.  Collaborator 3;  "The prescribed technique appears to be
 adequate with a couple of exceptions:  (1) Using the recommended 100 ml of
 water in the first impinger resulted in a loss of enough water from evapora-
 tion to  stop the desired bubbling action.  With the dry climate and length
 of run involved, it was found necessary to use 150 ml in the second to have
 continuous bubbling.  (2) Due to the length of each test, the silica gel
 became ineffective before the end of most runs, usually resulting in a net
 loss of  water from the train.  For this reason the train water "gain" can
 only be  used to determine stack gas moisture content for relatively short
 test times.

               Due to the cold weather encountered,  it became impossible to
maintain isokinetic conditions with the 3/8-in. nozzle initially used.  For
 this reason, Run No. 9 and subsequent runs were made with an 1/4-in. nozzle."

               d.  Collaborator 4;  "The method required that if the glass-
ware is  idle for 2 days (whether 10 ft from the beryllium source or 1,000
miles away), it must be resoaked in hydrochloric acid.  This seems very
 arbitrary,  and was not followed by us (Friday afternoon to Monday morning
 is 2-1/2 days)."
                                     30

-------
          3.  Limitations;  The two major limitations of field testing
were (1) the low beryllium emissions, and (2) the adverse weather.  The
beryllium concentrations emitted were in the neighborhood of 0.6 ug/ra ,
which provided relatively low amounts of beryllium in the field samples
taken by the collaborators.  The total amount of beryllium sampled during
a 4-hr run generally was in the neighborhood of from 1-2 ug.  These values
compare with from 3-16 jig °f beryllium contained in the NBS standard samples
given to the collaborators for analysis.

          The weather during the test was generally clear, cold, and windy.
The temperature ranged from 24-63°F, and at times the winds exceeded 100
miles per hour.  The terrain was covered with snow.  At times the test loca-
tion was icy, and visibility was poor.  This weather created adverse working
conditions for the field personnel as well as an adverse operational environ-
ment for the field sampling equipment.  Field sampling equipment, such as the
console, needed to be kept indoors when not in use, to minimize an extensive
equipment warm-up period prior to a test, especially for the pump of the
console.  Also the temperature of the impingers at times may have created
some icing effects within the impinger.  Generally, however, the ambient
conditions did not apparently adversely affect the sampling equipment.
                                     31

-------
                            VI.   ANALYSES OF  SAMPLES

 A.   Analysis Equipment Used by the Collaborators

          Two  of  the  four  collaborators, Nos.  1 and 3,  analyzed  their  beryl-
 lium samples in their laboratories with their  own  instrumentation.  The  other
 two  collaborators, Nos.  2  and 4,  subcontracted the analysis work.  The in-
 strumentation  used for the analysis of the test samples and the  standard
 samples  is described below.

          1.   Collaborator 1;  A  Perkin-Elmer Model 303 atomic absorption
 spectrophotometer, manufactured by Perkin-Elmer Corporation, Norwalk,
 Connecticut.

          2.   Collaborator 2;  A  Perkin-Elmer Model 303 atomic absorption
 spectrophotometer was used in conjunction with a Perkin-Elmer DCRl digital
 readout.  The  lamp used was a Perkin-Elmer hollow  cathode Be lamp.

          3.   Collaborator 3:  A  Jarrell Ash Model 812-250 atomic absorption
 spectrometer with a Perkin Elmer  Model 403 burner  and a beryllium hollow
 cathode  lamp.

          4.   Collaborator 4;  A  Ferkin-Elmer 303  atomic absorption spectro-
 photometer equipped with N20-Acetylene burner New  Fisher Beryllium Hollow
 Cathode Tube.
B.  Analysis Procedure

          The procedures as presented in Method 104 were generally followed.
The procedures each collaborator used, as discussed by them, are given below.
Collaborators 2 and 4 had the analyses performed by other companies.

          1.  Collaborator 1;  The laboratory analyses were done in accordance
with Method 104 with the following exceptions:

          Paragraphs 4.8.2.1 and 4.8.2.2 both state ". .  . .cool to room
temperature and add 5 ml concentrated sulfuric and 5 ml concentrated per-
chloric acid.   Then proceed with step 4.8.2.4.

                                     33

-------
          This procedure was changed in that 5 ml of 3:1:1 H20, HN03, HC104
mixture was added instead of adding 5 ml of concentrated perchloric acid.
This was done at the recommendation of the head of the Analytical Department.
In his opinion, the use of 5 ml of perchloric acid with 5 ml of concentrated
sulfuric acid and fuming presented the hazard of appreciable quantities of
dehydrated perchloric acid—a sensitive explosive.  He did not consider over
1 ml of HC104 safe in these circumstances.

          The sensitivity of the Perkin-Elmer 303 for beryllium is listed as
0.3 ug/ml/1% absorption by the manufacturer.  The observed sensitivity of
the unit for this set of analyses was 0.06 ug/ml/1% absorption.  The observed
noise level was 0.8% absorption."

          2.  Collaborator 2;  "Samples and standards having the same run
number were analyzed at the same. time.

               a.  Solids (BeO suspensions, filters):  Samples were digested
in 250 ml covered beakers using 35 ml concentrated HN03, 5 ml concentrated
HC104 and 5 ml concentrated ^804.  All acids were added at the beginning
of the digestion.  Samples were heated on a hot plate at low heat until fumes
of sulfuric appeared (3-4 hr).  Fume for 15 min.  Remove watch glass and allow
samples to evaporate to dryness.  Continue heating at low heat until sulfuric
acid is removed from sides of beaker.  Cool.  Add 10.00 ml of a solution con-
taining 2.5% w/v 8-hydroxyquinoline (oxine) in 3 N HCl.  Allow to stand
several minutes to dissolve Be, swirling occasionally.  Transfer to plastic
vial and cap to prevent evaporation until ready for reading on atomic-
absorption spectrophotometer (Perkin-Elmer Model 303).

               b.  Liquids (acetone wash solution);  Sample transferred in
portions to 250 ml covered beaker and evaporated to about 50 ml using boiling
chip to prevent bumping.  (Boiling chips were amphoteric alundum granules from
Hengar Company, Philadelphia, Pennsylvania).  Proceed as in procedure for
solids beginning with addition of concentrated acids.  After solution in
HCl-oxine, samples were filtered through Whatman No. 40 paper.

               Several of our own standards and blanks were digested both
with and without boiling chips.  There was no detectable difference between
the two methods of digestion.

               c.  NBS Liquid standards:   The contents of the ampule was
transferred to a 10 ml volumetric flask containing 0.25-g oxine.  Use 2.5 ml
concentrated HCl for rinsing the ampule.   Diluted to mark with deionized
water.

               d.  NBS BeO suspensions;  The contents of the ampule were
transferred to a 250-ml beaker using 2-3 ml concentrated HCl for rinsing
the ampule.  Proceed with digestion under solids.

                                     34

-------
               e.   Preparation of standard curve;  Beryllium standard  con-
centrations used were 0, Oi200, 0.360, 0.500,  1.00, 2.00, 3.00, 4.00,  and
6.00  ppm.

               Standards were prepared in 2.57. w/v 8-hydroxyquinoline-3N HCl
solution by appropriate dilutions of a 1,000-ppm Be standard (Sargent-Welch
No. SC 16220, 1,000-ppm Be).

          3.  Collaborator 3;  "The total sample collected from final  impinger
(silica gel) to probe tip was determined by atomic absorption as specified
in Method 104 on a  Jarrell Ash Model 82-250 atomic absorption spectrometer
with  a Perkin-Elmer Model 403 burner and a Beryllium Hollow Cathode Lamp.  A
nitrous oxide-acetylene flame was used for excitation and the absorption was
measured at 234.86 run.  The samples were prepared in accordance with Method
104 with 1% potassium as the chloride, added to each sample according  to the
procedure of  Fleet et al., as referenced in Method 104, to overcome potential
interfering elements.

          The beryllium concentration of each sample was calculated as de-
scribed in paragraph 6.6 of Method 104 and the total beryllium emissions was
calculated as in 6.7.  The results obtained for both test samples and stand-
ards  are attached.  The water analysis represents a combination of the impinger
solutions and washings from the sampling train.  The filter analysis represents
the material collected on the Millipore Type AA filter.

          4.  Collaborator 4;  All glassware was cleaned and soaked in 1:1
HCl wash as per instructions.

          a.  Be standards were prepared from a 10-ppm standard supplied by
our customer by diluting with 257. HCl to cover the range from 6.002 ppm to
1.0 ppm.

          b.  A vial was selected at random (No. 616)  and diluted to 25 ml
to establish an approximate concentration range, and approximate volume
necessary to obtain averageable readings from the NULL meter.   We found
10-ml volume was minimum.

          c.  Vial samples were transferred to a graduated 10-ml glass
stoppered ampule by means of syringe with small diameter transparent plastic
tube  in place of needle.  Vials were rinsed with small amounts of 257. HCl
employing this syringe method.  The total volumes were adjusted to 10 ml.

          d.  The 18-vial samples were analyzed using standard Be condi-
tions as stated in instructions, i.e., UV 234.9, slit 4, etc.   Three NULL
meter readings were taken and averaged for each standard and sample; a curve
                                     35

-------
drawn plotting parts per million versus absorbance of standards, the concen-
tration of the samples read from the curve, dilution factors employed, and
the final parts per million of these samples recorded.

          e.  The 12 filters were digested as per instructions, and analyzed
similarly, the curve drawn, the parts per million of samples read, dilution
factor (10) employed and results reported.

          f.  The 13 sample filters were processed and analyzed the same  as  for
the 12 standard filters.

          g.  The water-acetone samples were evaporated to dryness on a hot
plate.

          Each acetone sample was added to its respective water-acetone
beaker and evaporated to dryness on a steam bath.  Then the nitric, sulfuric,
perchloric digestion and atomic absorption analyses completed same as for
the standard vial samples.
C.  Problem Areas

          The problems of analysis encountered that were indicated by collab-
orator 4 who had the analysis done by another company are.

          1.  Primary beryllium standard;  Method 104 requires use of Be
powder 98% purity.  Since our company is not licensed for handling Be con-
centrates, we requested and received a low concentrate standard solution
from the company who requested these analyses.

          This was not actually a "problem," but mentioned here to those
who may be sending out future study samples, so they may be cognizant of
the necessity of also supplying low concentrate standard solutions.

          2.  Establishing detection limit for our equipment:  Due to in-
stability of machine electronics it was found that scale expansion two (2)
was maximum readable expansion.  Going to a higher expansion and using
dampening consumed too large a quantity of sample for needle to stabilize,
and time required caused excessive heating of exhaust system.  Lacking
greater expansion, 0.05 ppm was established as the lower detection limit.
We contacted Perkin-Elmer, Norwalk, Connecticut; Perkin-Elmer, Raleigh,
North Carolina; and an atomic absorption authority with North Carolina Air
and Water Resources, all of whom confirmed this limit as most probable for
our equipment.

          Use of a recorder was tried but found NULL meter readings more
satisfactory.

                                     36

-------
          3.  Vials too small;  Had difficulty transferring liquid from
vial to ampule for dilution.  Could not rinse cap.  Too tedious and time
consuming.

          Another problem area reported was that regarding the 5 ml of
concentrated perchloric acid described by collaborator 1 in his analysis
procedure on page 30.
D.  Results of Collaborators' Analyses

          There are several different sets of results to report.  First,
there are the results of MRl's preliminary testing which are presented in
Table II, Section IV; second, there are the results of the collaborators'
analyses of their test samples; and third, there are the results of the
analyses of standard samples prepared by the National Bureau of Standards.
The latter two sets of results are presented below in the order given.

          1.  Test samples;  The results, as received by MRI, from the chemi-
cal analysis performed by the four collaborators of their test samples are
given in Volume 2.  These results were checked by MRI to determine if there
were any systematic or gross errors of calculations.  There were no signifi-
cant differences.

          Table III presents a summary of the amount of beryllium collected
on both the filter and the solution samples, as well as their total; beryl-
lium loading under both standard and stack conditions; and the beryllium
emission rates.  These data are presented by run for each of the four col-
laborators.  Also included in this table is a summary of the size probe tip
used by each collaborator and the volume of dry gas, the average stack-gas
velocity and the percent isokinetic as determined by them.  In addition,
this table presents MRl's comparative results from its check of the collabora-
tors calculated results.
                                     37

-------
                                                                                       TABLE III

                                                                  SUMMARY OF THE RESULTS OF THE COLLABORATORS' ANALYSES
          Run
          No..
Co
00
           Measurement

Probe Tip Dimeter (In.)
Volume Dry Gas - Standard Condition
Average Stack Gas Velocity (FPM)
Percent Isoklnetlc
Mass Be Collected - Filter (tig)
Mass Be Collected - Solution (ug)
Mass Be Collected - Total (ug)
Be Loading - Standard Condition
  (ug/lta3)
Be Loading - Stack Condition (lig/m3)
Be Emission Bate (g/day)

Probe Tip Dlaateter (in.)
Voliau Dry Gas - Standard Condition
Average Stack Gas Velocity (FPM)
Percent Isoklnetlc
Mass Be Collected - Filter (ug)
Mass Be Collected - Solution (ug)
Mass Be Collected - Total (ug)
Be Loading - Standard Condition
  (ug/*.3)
Be Loading - Stack Condition (|ig/m3)
Be Emission Kate (g/day)

Probe Tip Diameter (In.)
Volume Dry Gas - Standard Condition
Average Stack Gas Velocity (FPM)
Percent Isoklnetic
Mass Be Collected - Filter (ug)
Mass Be Collected - Solution (pg)
Mass Be Collected - Total (|ig)
Be Loading - Standard Condition
  (ug/lta3)
Be Loading - Stack Condition (ug/m3)
Be Emission Bate (g/day)
Collaborator \*J
0.375 0.25
38.6 89.8
1,701 1,701
76.5 100
0.0
1.3
1.3
„• «•
0.29
0.30
0.25
91.1
1,754
79.1
0.0
0.9
0.9
__ »
0.28
0.30
0.25
112.9
1,661
102.7
0.4
1.2
1.6
..
0.40
0.41
MRI
_
128.57
1,739
112.1
0.0
1.3
1.3
0.3563
0.2865
0.3048
„„
91.12
1,794
77.4
0.0
0.9
0.9
0.3481
0.2784
0.3048
„
112.98
1,725
99.0
0.40
1.20
1.60
0.4991
0.4024
0.4248
Collaborator 2 MRI Collaborator 3i
0.375
243.1
1,683
98
0.7
1.8
2.5
„ —
0.317
0.300
0.375
247.4
1,742
97.1
0.83
5.00
5.83
_—
0.706
0.705
0.375
248.7
1,722
97.6
0.62
2.00
2.62
__
0.318
0.315
^m
244.25
1,271
96.2
0.7
1.8
2.5
0.3607
0.2882
0.3024
mm
245.74
1,762
95.3
0.83
5.00
5.83
0.8360
0.6628
0.7128
— —
248.38
1,772
94.6
0.62
2.00
2.62
0.3718
0.2982
0.3240
0.377
231.87
1,717
91.2
0.3
2.0
2.3
--
0.28
0.29
0.377
229.57
1,679
91.7
0.2
2.0
2.2
__
0.27
0.28
0.377
216.32
1,618
88.8
0.7
0.2
0.9
_ —
0.12
0.12
'J MRI
— —
231.81
1,707
92.6
0.3
2.0
2.3
0.3496
0.2747
0.288
— —
230.01
1,672
93.3
0.2
2.0
2.2
0.3370
0.2664
0.2712
— —
216.47
1,612
90.4
0.7
0.2
0.9
0. 1466
0.1167
0.1152
Collaborator 4

      0.304
    160.905
  1.665
     99.5
      0.51
      0.66
      1.17

      0.257

      0.209

      0.304
    156.609
  1,636
     99.7
      0.45
      0.82
      1.27

      0.281

      0.227

      0.304
    162.531
  1.674
     98.6
      0.34
      0.40
      0.74

      0.161

      0.134
                                                                                                                                                                  161.01
                                                                                                                                                                1.661
                                                                                                                                                                  99.8
                                                                                                                                                                   0.51
                                                                                                                                                                   0.66
                                                                                                                                                                   1.17

                                                                                                                                                                   0.2561
                                                                                                                                                                   0.2049
                                                                                                                                                                   0.2088
                                                                                                                                                                 159.57
                                                                                                                                                               1.644
                                                                                                                                                                  99.3
                                                                                                                                                                   0.45
                                                                                                                                                                   0.82
                                                                                                                                                                   1.27

                                                                                                                                                                   0.2805
                                                                                                                                                                   0.2260
                                                                                                                                                                   0.228
                                                                                                                                                                 162.76
                                                                                                                                                               1,676
                                                                                                                                                                  98.5
                                                                                                                                                                   0.34
                                                                                                                                                                   0.40
                                                                                                                                                                   0.74

                                                                                                                                                                   0.1602
                                                                                                                                                                   0.1302
                                                                                                                                                                   0.1344

-------
                                                                                       TABLE III (Continued)
           Run
           No.
                         Measui
                                       Collaborator li/
                                                                            HRI
                                                                                     Collaborator 2
                                                                                                       MRI
Collaborator 3J>/     Mpj
                                                                                                                                               Collaborator 4
                                                                                                                                                   HRI
U)
VO
Probe Tip Diameter (in.)
Volume Dry Gas - Standard Condition
Average Stack Gas Velocity (FFH)
Percent Isokinetlc
Mass Be Collected - Filter (ug)
Mass Be Collected - Solution (ug)
Mass Be Collected - Total (ug)
Be Loading - Standard Condition
  (MB/NO,3)
Be Loading - Stack Condition (ug/m3)
Be Emission Rate (g/day)

Probe Tip Diameter (In.)
Volume Dry Gas - Standard Condition
Average Stack Gas Velocity (FPM)
Percent Isokinetlc
Mass Be Collected - Filter (pg)
Mass Be Collected - Solution (ug)
Mass Be Collected - Total (ug)
Be Loading - Standard Condition
  (ug/Nm3)
Be Loading - Stack Condition (|ig/m3)
Be Emission Rate (g/day)

Probe Tip Diameter (in.)
Volume Dry Gas - Standard Condition
Average Stack Cas Velocity (FPM)
Percent Isoklnetic
Mass Be Collected - Filter (ug)
Mass Be Collected - Solution (ug)
Mass Be Collected - Total (ug)
Be Loading - Standard Condition
  dig/Dm3)
Be Loading - Stack Condition (ug/m3)
Be Emission Rate (g/day)
0.25
109.2
1,608
101.8
__
_„
~
_—
__
~
0.25
93.3
1,575
88.0
0.0
1.2
1.2
— —
0.37
0.36
0.25
110.5
1,591
103.7
0.0
0.7
0.7
— —
0.18
0.18
__
109.26
1,642
99.8
..
__
"
— —
„
"
-—
96.07
1,608
88.6
0.0
1.2
1.2
0.4402
0.3611
0.3552
— —
110.62
1,627
101.5
0.0
0.7
0.7
0.2230
0.1820
0.1800
0.375
249.6
1,751
96.B
1.60
5.62
7.22
••••
0.812
0.876
0.375
237.1
1,634
97.9
0.62
4.50
5.12
«
0.635
0.614
0.375
227.2
1,639
95.1
0.66
3.72
4.38
— —
0.529
0.541
__
249.61
1,796
94.3
1.60
5.62
7.22
1.0193
0.8135
0.8928
— —
236.05
1,679
94.8
0.62
4.50
5.12
0.7644
0.6138
0.6312
— —
226.19
1,682
91.8
0.66
3.72
4.38
0.6824
0.5416
0.5568
0.377
208.39
1,674
88.5
0.5
6.2
6.7
— —
0.91
0.88
0.377
211.30
1,692
83.2
0.3
1.6
1.9
— —
0.25
0.26
0.377
205.71
1,537
88.7
0.2
1.4
1.6
_—
0.22
0.21
._
208.45
1,570
90.0
0.5
6.2
6.7
1.1327
0.8953
0.8592
__
219.07
1,691
87.6
0.3
1.6
1.9
0.3057
0.2421
0.2496
— —
205.63
1,533
90.2
0.2
1.4
1.6
0.2742
0.2186
0.2064
0.304
159.322
1,671
98.1
0.82
2.40
3.12
0.692
__
0.566
0.304
155.190
1,651
97.4
0.43
1.10
1.53
0.348
_.
0.280
0.304
154.910
1,600
97.4
0.26
2.24
2.50
0.570
._
0.457
—
159.36
1,675
97.9
0.72
2.40
3.12
0.6899
0.5529
0.5640
..
155.20
1,655
97.2
0.43
1.10
1.53
0.3474
0.2764
0.280B
__
154.88
1,600
97.3
0.26
2.24
2.50
0.5689
0.4663
0.4560

-------
                                                                                 TABLE III (Continued)
        Bun
        NO..
•P-
O
         Measurement

Probe Tip Diameter (in.)
Volume Dry Gas - Standard Condition
Average Stack Gas Velocity (FPM)
Percent Isoklnetlc
Mass Be Collected - Filter (ug)
Mass Be Collected - Solution (ug)
Mass Be Collected - Total (ug)
Be Loading - Standard Condition
  (ug/Nm3)
Be loading - Stack Condition (ug/m3)
Be Emission Bate (g/day)

Probe Tip Diameter (In.)
Volume Dry Gas - Standard Condition
Average Stack Gas Velocity (FPM)
Percent Isoklnetlc
Mass Be Collected - Filter (ug)
Haas Be Collected - Solution (ug)
Mass Be Collected - Total (ug)
Be Loading - Standard Condition
  (I»g/Nm3)
Be Loading - Stack Condition (ug/m3)
Be Emission Rate (g/day)

Probe Tip Diameter (In.)
Volume Dry Gas - Standard Condition
Average Stack Gas Velocity (FPM)
Percent Isoklnetlc
Mass Be Collected - Filter (ug)
Mass Be Collected . Solution (ug)
Mass Be Collected - Total (ug)
Be Loading - Standard Condition
  (ug/Hn.3)
Be Loading - Stack Condition (ug/ir)
Be Emission Rate (g/day)
                                      Collaborator !«/
                                                           MBI
                                                                    Collaborator 2
                                                                                       MRI
                                                                                                Collaborator
                                                                                                                     MM
                                                                                                                               Collaborator 4
                                                                                                                                                  MB1
0.25
107.8
1,592
103.9
0.0
0.9
0.9
— —
0.23
0.23
0.25
106.14
1,595
102.9
0.0
1.9
1.9
mm
0.50
0.49
0.25
105.89
1,550
105.6
0.5
1.3
1.8'
— —
0.47
0.45

107.47
1,621
101.8
0.0
0.9
0.9
0.2951
0.2343
0.2328
...
106.23
1,627
100.9
0.0
1.9
1.9
0.6303
0.4968
0.4944
— —
105.95
1,603
102.1
0.5
1.3
1.8
0.5987
0.4722
0.4632
0.375
228.5
1,617
95.1
0.84
4.50
5.34
— — .
0.706
0.647
0.375
214.8
1,667
89
0.68
8.20
8.88
— —
1.129
1.172
0.375
231.5
1,664
96
0.76
9.10
9. 8t
— —
1.271
1.205

227.44
1,658
94.1
0.84
4,50
5.34
0.8275
0.6535
0.6624
-M
214.62
1,621
91.3
0.68
8.20
8.88
1.4581
1.1452
1.1352
„
231.28
1,707
93.7
0.76
9.10
9.86
1.5024
1.1771
1.2288
0.377
201.58
1,682
80.9
0.8
0.7
1.5
— —
0.21
0.22
0.377
226.79
1,693
89.8
0.7
2.0
2.7
mm
0.33
0.35
0.255
119.85
1,714
103.4
0.2
1.1
1.3
-—
0.30
0.32

223.60
1,682
90.7
0.8
0.7
1.5
0.2364
0. 1858
0. 1896
— —
226.69
1,691
91.7
0.7
2.0
2.7
0.4197
0.3289
0.3048
__
119.59
1,719
103.7
0.2
1.1
1.3
0.3831
0.3011
0.3168
0.304
142.355
1,483
99.4
0.43
0.63
1.06
0.263
__
0.190
0.304
149.372
1,699
92.0
0.36
3.46
3.82
0.903
._
0.739
0.304
152.620
1,606
98.9
0.35
0.57
0.92
0.213
..
0.166
..
142.34
1,480
99.6
0.43
0.63
1.06
0.2625
0.2088
0.1896
._
149.52
1,638
95.5
0.36
1.46
3.82
0.9003
0.7089
0.7104
--p
152.63
1,605
98.9
0.35
0.57
0.92
0.2124
0.1683
0.1656

-------
Run
Ho.           Measurement

     Probe Tip Diameter (In.)
     Volume Dry Gas - Standard Condition
     Average Stack Gas Velocity (FFM)
     Percent Isoklnetlc
     Mass Be Collected - Filter (ug)
10   Mass Be Collected - Solution (ug)
     Mass Be Collected - Total (ug)
     Be Loading - Standard Condition
       (Ug/Nm3)
     Be Loading - Stack Condition
     Be Emission Rate (g/day)

     Probe Tip Dlamater (In.)
     Volume Dry Gas - Standard Condition
     Average Stack Gas Velocity (FFM)
     Percent Isoklnetic
     Mass Be Collected - Filter (ug)
11   Mass Be Collected - Solution (tig)
     Mass Be Collected - Total (ug)
     Be Loading - Standard Condition
       (Ug/Nm3)
     Be Loading - Stack Condition (ug/m3)
     Be Emission Rate (g/day)

     Probe Tip Diameter (in.)
     Volume Dry Gas - Standard Condition
     Average Stack Gas Velocity (FPM)
     Percent Isokinetlc
     Mass Be Collected - Filter (ug)
12   Mass Be Collected - Solution (ug)
     Mass Be Collected - Total (ug)
     Be Loading - Standard Condition
     Be Loading - Stack Condition
     Be Emission Rate (g/day)
Collaborator IS./
„
__
..
__
__
_-
—
__
..
—
— —
M.
..
_.
__
_.
—
,„
._
—
..
_.
~
..
._
..
—
— —
__
._
0.25
106.45
1,582
102.8
0.0
1.9
1.9
— —
0.50
0.49
0.25
109.90
1,600
103.2
0.7
9.6
10.3
• •.
2.68
2.62
0.25
105.40
1,611
99.8
0.6
3.3
3.9
— —
1.04
1.03
TABLE III (Continued)
MM Collaborator 2 MHI
„
106.78
1,617
101.0
0.0
1.9
1.9
0.6271
0.4997
0.4944
mm
110.02
1,669
99.1
0.7
9.6
10.3
3.2991
2.6746
2.7288
__
105.17
1,610
99.7
0.6
3.3
3.9
1.3068
1.0435
1.0272
0.375
217.8
1,678
89.0
1.20
7.67
8.87
— m
1.129
1.168
0.375
242.3
1,656
99.4
6.91
33.70
40.61
— m
5.012
4.951
0.375/0.25
132.6
1,585
99.3
0.77
4.16
4.93
—-
1.094
1.459
„
225.88
1,694
91.0
1.20
7.67
8.87
1.3838
1.0977
1.1376
_„
242.19
1,701
93.7
6.91
33.70
40.61
5.9091
4.8639
5.0592
__
132.63
1,622
121.7
0.77
4.16
4.93
1.3099
1.0725
1.0632
Collaborator 3J>/

      0.255
    116.64
  1,653
    105.4
      0.2
      0.2
      0.4
      0.10
      0.10

      0.255
    118.76
  1,684
    102.3
      1.7
      5.0
      6.7
      1.62
      1.66

      0.255
    110.22
  1,586
    102.0
      5.6
      1.4
      7.0
      1.80
      1.75
   MRI
  115.98
1,656
  104.8
    0.2
    0.2
    0.4

    0.1216
    0.0952
    0.0960
  118.65
1,681
  102.7
    1.7
    5.0
    6.7

    1.9900
    1.6024
    1.6464
  110.06
1,587
  102.1
    5.6
    1.4
    7.0

    2.2414
    1.7841
    1.7328
Collaborator 4

      0.304
    149.879
  1,621
     96.3
      0.36
      2.60
      2.96

      0.697

      0.547

      0.304
    163.660
  1,625
     99.9
      5.00
      3.25
      8.25

      1.78

      1.47

      0.304
    150.908
  1,635
     94.7
      0.63
      1.54
      2.17

      0.508

      0.408
                                                 MRI
  149.81
1,622
   96.2
    0.36
    2.60
    2.96

    0.6963
    0.5510
    0.5472
  163.65
1,628
   99.7
    S.OO
    3.25
    8.25

    1.7765
    1.4754
    0.4688
  151.01
1,633
   94.9
    0.63
    1.54
    2.17

    0.5064
    0.4068
    0.4056

-------
                                                                                  TABLE III (Concluded)
               Run
               No.
               13
10
         Measurement

Probe Tip Diameter (In.)
Volume Dry Gas - Standard Condition
Average Stack Cas Velocity (FPM)
Percent Isoklnetlc
Mass Be Collected - Filter (us)
Mass Be Collected - Solution dig)
Mass Be Collected - Total (|ig)
Be Loading - Standard Condition
  (H8/Nm3)
Be Loading - Stack Condition (ug/m3)
Be Emission Rate (g/day)
                                      Collaborator \Sj     MM       Collaborator 2     MM      Collaborator 3b/     MM
Collaborator 4
                                                                                                                                                   MRI
0.25
105.67
1,616
100.3
0.4
3.7
4.1
„
1.09
i.oa

105.78
1,653
98. 1
0.4
3.7
4.1
1.3660
1.0852
1.0992
0.25
107.4
1,643
100.9
0.50
5.06
5.56
— —
1.483
1.153

107.41
1,690
97.9
0.50
5.06
5.56
1.8242
1.4434
1.4904
0.255
112.96
1,632
101.9
0.9
2.7
3.6
— —
0.90
0.90
„
113.08
1,632
102. B
0.9
2.7
3.6
1.122
0.8859
0.8832
0.304
148.909
1,624
95.0
0.48
3.11
3.59
0.851
..
0.673
„
148.86
1,624
94.9
0.48
3.11
3.59
0.8499
0.5898
0.6696
               a/  Collaborator 1 changed probe tips  at the  completion  of  the  first traverse  (one port  of  six sampling points) of Run 1 and chose to  present two sets
                     of results from Run 1;  one set for each probe  tip  used.
               b/  Collaborator 3 changed probe tips  at the  first traverse of  Run  12 because  he was not maintaining isoklnetlc sampling, but chose to present his
                     results In a combined form for Run 12.

-------
          Sampling was performed isokinetically with each collaborator
generally being quite close to 100%.  Collaborator 3 was generally low, but
for most runs was between the 90-110% requirement.  Since the error in mass
collection is a function of the relationship
                       100% Isokinetic Sampled
                                 2
for a heterogeneous, particle-size distribution; it is believed that the
emitted beryllium particles were quite small; and the stream flow was not
turbulent, the deviations from 100% isokinetic are not significant.

          The ratio of beryllium loading (stack conditions) to beryllium
emission rate for each collaborator for each run is close to unity.  If the
loading were based on total mass collected rather than just beryllium and the
ratio were a constant, then this information could be used to show that devia-
tions between collaborators on a run could be attributable to problems of
analysis.  Since the procedures of Method 104 did not require total mass in-
formation, this check cannot be made here.

          2.  Standard samples;  A summary of the results of the National
Bureau of Standards on the standard samples it prepared for this collabora-
tive test are given in Table IV.  The column of recommended values were
those used by MRI in its statistical analyses.

          The results of MRI's analyses of some of NBS's standard samples are
given in Table V.  These samples were analyzed according to Method 104, as
described in the Federal Register. 38., No. 66, 6 April 1973, with the excep-
tion that HC104 was excluded from the digestion.  This was done purposely to
determine if there would be any significant deviation in the results from
those of NBS.  There were none.

          The results of the collaborators' analyses of the standard samples
furnished by NBS are given in Volume II.  These results are combined and
summarized in Table VI.  Also included in this table are MRI's results of
the standard samples it analyzed.  The first column of the table indicates
the type sample; Column 2 gives the sample identification number* range
assigned by NBS; Column 3 gives NBS's measured values; Columns 4-13 give the
collaborators' and MRI's values as well as the deviations of these values
from the NBS values; and Columns 14 through 18 give the deviations in per-
centages.
*  The collaborators data in  Volume II are referenced to these number.


                                     43

-------
                                  TABLE  IV
   NBS'S  SUMMARY OF ANALYTICAL  RESULTS  OF ITS  STANDARD BERYLLIUM SAMPLES
 Sample Type   Series
Filters
 Suspended
   Solids
Soluble Be
100-150
200-250
300-350

400-450
500-550
600-650

700-750
800-850
900-950
                                          Be,  ue/sample
Atomic Absorption!/
2.98 ± 0.06
7.77 ± 0.11
15.2 ± 0.17
3.06 ± 0.06
7.38 ± 0.22
15.0 ± 0.29
3.24 ± 0.03
7.87 ± 0.11
15.8 ± 0.14
Fluorescence^/
3.27 ± 0.12
8.39 ± 0.28
14.8 ± 0.38
2.96 ± 0.04
7.41 ± 0.18
14.5 ± 0.37
3.06 ± 0.12
7.59 ± 0.19
16.1 ± 0.4
Recommended Values^/
2.98 ± 0.13
7.77 ± 0.24
15.2 ± 0.4
3.06 ± 0.13
7.38 ± 0.48
15.0 ± 0.6
3.24 ± 0.07
7.87 ± 0.24
15.8 ± 0.3
a_/  Based on 10 determinations.  Uncertainty  is  standard  deviation  of a single
      measurement.
j>/  Based on six determinations.  Uncertainty is standard deviation of a single
      measurement.
cj  Uncertainty represents 957. confidence  interval.
                                   TABLE V

     RESULTS OF MRI'S ANALYSES OF BeO FILTERS. SUSPENDED  BEO AND  BE
              SOLUTIONSBYATOMIC ABSORPTION SPECTROPHOTOMETRY
Filter, Filter Blank
Filter, Filter Blank
Filter, BeO
        BeO
        BeO
Ampule, Suspended BeO
Ampule, Suspended BeO
Ampule, Suspended BeO
Ampule, Soluble Be in 0,
Ampule, Soluble Be in 0,
Filter,
Filter,
                        25 M HC1
                        25 M HC1
Ampule, Soluble Be in 0.25 M HC1
Sample No.

    010
    020
    130
    231
    331
    414
    508
    628
    717
    813
    929
                                              Amount in jj,g

                                                   0.0
                                                   0.0
                                                   3.0
                                                   7.5
                                                  13.3
                                                   2.7
                                                   6.8
                                                  14.6
                                                   3.0
                                                   7.8
                                                  14.8
                                     44

-------



Filter, Blank
Filter, Blink
Filter, Blink
Filter, Level 1 of BeO
Filter, Level 1 of BeO
Filter, Level I of BeO
Filter, Level o BeO
Filter. Level o BeO
Filter, Level o BeO
Filter, Level o BeO
Filter, Level o BeO
Filter, Level o BeO
Ampule X, Level 1 of Suspended BeO
Ampule X, Level 1 of Suspended BeO
Ampule X, Level 1 of Suspended BeO
Ampule X, Level 2 of Suspended BeO
Ampule X, Level I of Suspended BeO
Ampule X, Level 1 of Suspended BeO
Ampule X, Level 3 of Suipended BeO
Ampule X, Level 3 of Suipended BeO
Ampule X, Level 1 of Suspended BeO
Ampule Y, Level 1 of Soluble BeS/
Anpule T, Level 1 of Soluble
Ampule T, Level 1 of Soluble
Ampule T, Level 2 of Soluble
Ampule T, Level 2 of Soluble
Ampule T, Level 2 of Soluble
Ampule Y, Level 3 of Soluble
Ampule Y, Level 3 o( Soluble
Ampule Y, Level 3 of Soluble

1.0.
ea^L
030
to
030
100
to
130
200
to
230
300
to
MO
400
to
430
300
to
350
MO
to
650
700
to
730
aoo
to
sso
900
to
910

HB«y

0.2
0.3
0.2
1.6
1.6
1.6
3.0
4.0
4.1
9.0
9.0
v.O
2.9
1.3
1.4
1.3
1.3
4.6
a. 5
10.0
7.3
2.0
1.1
1.4
3.2
3.3
4.3
3.2
9.0
6.0
413
Cm)
0.2
0.3
0.2
-t.3>
-1.3»
-1.36
-2.7!
-3.7J
-3.61
-6.2
-6.2
-6.2
-0.16
.1.36
.1.66
-4.08
-1.88
-2.78
-6.30
-3.0
-1.30
.1.24
-2.14
-1.84
.2.67
-4.37
-3.37
-10.60
-6.8
-0.8
Collebontor Ho. 4
Measured
(ua>
0.03
0.05
0.03
3.63
1.64
3.93
10.2
9. SO
9.75
20.1
21.0
19.2
0.42
0.32
0.59
1.19
0.93
1.12
2.60
1.86
1.74
2.21
2.19
1.19
5.13
3.00
3.24
10.35
10.80
10.73
£14
iHBi
0.05
0.05
0.03
0.65
0.66
0.97
2.4}
2.03
1.96
4.9
5.8
4.00
-2.64
-2.34
-2.47
-6.19
-6.45
-6.26
-12.4
-13. 14
-13.26
-1.03
-1.05
-1.05
-2.72
-2.87
.2.61
.5.23
.3.0
.3.07
Mil
Mueured
to)
0.0
-.
—
3.0
..
—
7.3
..
—
13.3
..
—
2.7
_
—
6.B
-.
—
14.6
..
"
1.0
_
—
7.8
..
—
14.8
..
~~

&15
{ml
-.
..
—
0.02
..
—
-0.27
»
—
.1.9
..
—
-0.16
..
—
-o.sa
_.
—
-0.4
-.
—
-0.24
—
—
-0.07
..
—
-1.0
«
••

All
ill
—
-.
—
-9.4
17.4
-6.0
1.7
4.2
5.5
3.3
-1.3
-U3
-1.0
-2.0
-34.6
-1.1
-5.1
-5.1
-6.7
-6.7
-40.0
.7.4
-7.4
-7.4
-11.1
-11.1
.11.1
-3.1
-24.1
-36.7

£12
01
_
-.
—
10.7
7.1
8.4
13.1
4.2
.12.5
9.8
3.9
-0.66
1.1
-10.6
-6.5
-4.1
.16.1
-12.7
4.7
-6.7
-3.3
-1.2
-4.9
-0.0
2.4
0.13
-2.9
-4.4
0.0
-1.3

£13
ill
—
-•
—
-46.3
-46.3
-it. 3
-33.6
-48.3
-47.2
.40. B
-40.8
-40.8
-5.2
-50.9
-54. Z
-55.3
-52.6
-37.7
-41.3
-33.3
-30.0
-38.3
-66.0
.36.8
-33.9
-33.3
-42.8
-67.1
-43.0
-61.0

£14
ill.
—
..
—
21.8
22.1
12.6
31.1
26.1
25.9
32.2
18.2
26.3
-86.3
-83.0
.80.7
-83.9
.87.6
-84.8
-tz.i
-87.6
-S8.4
-11. a
-32.4
-12.6
-14. e
-16.3
.11.4
.33.2
.31,6
-31.0

£15
ill
—
—
—
0.6
..
—
.1.6
..
—
-14.3
..
—
.11.3
..
—
4.5
..
—
-2.7
..
—
-a.o
..
—
-O.9
»
—
-6.8
.-
™
a/  l.D. - BBS Identification numberii  Individual sample nuabars lie within the range) B'vra
b/  BBS  " National Bureau of Standard! values.
c/  £11  - Differences between HBS values end the collaborators1 values.
if  Soluble Be In 0.23 H BC1.

-------
          Collaborators 3 and 4 show significant deviations from NBS values
for all the samples they analyzed.  For all type samples,  with the exception
of the values of collaborator 4 filters, the deviations are less than NBS
values.  This information along with that of the test sample results (see
Table III) tend to indicate that the inaccuracies lay in the analysis of
samples rather than in sampling, but this is not conclusive as there is not
necessary and sufficient bases for such a conclusion.

          3.  Velocity profile data;  MRI calculated the stack velocity at
each of the 30 sampling points of a run as a part of its check of the col-
laborators' results.  These results, which are given in Volume II,  form the
basis for the velocity profile analyses given in Section VII.
                                     46

-------
                 VII.   STATISTICAL ANALYSIS OF SAMPLING RESULTS

           There were three analyses performed.   The primary one  was  a  two-way
 analysis of variance to obtain the variance of  repeated observations per  col-
 laborator, a ,  and to obtain the variance between collaborators,  a .  The
 analysis was done using the collaborators beryllium emission rate results.
 A secondary analysis  was the same except beryllium loading results were used
 in place of the emission rate results.   The third analysis, which is also a
 secondary analysis, was to determine if the average velocity per sampling point
 per run correctly represented the geometrical variance in  velocity throughout
 the test run even though they were measured at  different times.
 A.   Field Test  Samples

           1.  Analysis of beryllium emission  rate;  The primary  analyses of
 the field test  sample data (see  Table VII) was  a  (two-way) analysis of
 variance, and the  primary object of the  analysis  of variance was to estimate
 components of variance;  specifically, o| = variance of repeated  observations
 (per collaborator),  and a£ = variance between collaborators.  The response
 is  beryllium  emission rate in  grams per  day.

           Before analyzing the data, the beryllium emission rates were recal-
 culated using the  collaborators'  observed values.  No errors were found (simple
 correlation between  MRI  and each collaborators' results was > 99.9% in all
 cases), and therefore no decisions  about what to  do with arithmetic errors
 were necessary.  One data point  (test 4, collaborator 1) was missing.   It
 was replaced  in the  analysis via the usual procedure of minimizing the error
 sum of  squares.

          In studies of this type, it is often found that  the measurement
error is proporf.ional to the magnitude of the response (or some  function of
the response level), rather than constant,  as is required  by an  analysis of
variance assumption.  This was true of the beryllium data  also.   The correla-
tion between the standard deviation at a level and the level itself was +0.95
(see Table VIII).   Thus, the data were transformed  by a log transformation (see
Table IX), and then the analysis of variance performed.  The transformation z =
log (100 • beryllium emission rate (g/day)  = log (100 y) was used for
convenience.
                                     47

-------
                                TABLE VII

          BERYLLIUM EMISSION RATES  (g/DAY) FROM RUNS 1 THROUGH  13
Collaborator
Test
1
2
3
4
5
6
7
8
9
10
11
12
13
£1
0.30
0.30
0.41
0.85 a/
0.36
0.18
0.23
0.49
0.45
0.49
2.62
1.03
1.08
C2
0.30
0.705
0.315
0.876
0.614
0.541
0.647
1.172
1.205
1.168
4.951
1.037
1.459
03
0.29
0.28
0.12
0.88
0.26
0.21
0.22
0.35
0.32
0.10
1.66
1.75
0.90
<*
0.209
0.227
0.134
0.566
0.28
0.457
0.19
0.739
0.167
0.547
1.47
0.408
0.673
a/  Artificial:  Collaborator's data missing.
                               TABLE VIII

               STANDARD DEVIATION VERSUS  LEVEL OF  BERYLLIUM

   Level             Average  Beryllium Rate  (g/.day)               a

      1                           0.275                         0.0441
      2                           0.378                         0.2202
      3                           0.245                         0.1415
      4                           0.793                         0.1664
      5                           0.379                         0.1628
      6                           0.347                         0.1793
      7                           0.322                         0.2175
      8                           0.688                         0.3607
      9                           0.535                         0.4613
     10                           0.576                         0.4417
     11                           2.675                         1.5980
     12                           1.056                         0.5485
     13                           1.028                         0.3321
                                     48

-------
                                 TABLE IX

                        BERYLLIUM DATA TRANSFORMED

                                     Collaborator
Cl
3.40
3.40
3.71
4.44
3.58
2.89
3.14
3.89
3.81
3.89
5.57
4.63
4.68
C2
3.40
4.26
3.45
4.47
4.12
3.99
4.17
4.76
4.79
4.76
6.20
4.64
4.98
C3
3.37
3.33
2.48
4.48
3.26
3.04
3.09
3.56
3.47
2.30
5.11
5.16
4.50
C4
3.04
3.12
2.59
4.04
3.33
3.82
2.94
4.30
2.81
4.00
4.99
3.71
4.21
     Level

        1
        2
        3
        4
        5
        6
        7
        8
        9
      10
      11
      12
      13
          The analysis of variance results (in terms of z) are shown in
Table X.
                                 TABLE X

              ANALYSIS OF VARIANCE OF BERYLLIUM EMISSION RATE

   Source            dF_       SS          MS         F           EMS

Collaborator (C)
Level (L)
Error (E)
3
12
35
6.1752
22.0305
6.6220
2.0584
7.3435
0.1892
10.88
38.81^

o-e* + 13 CT2
af + 4 CT£
oi
a/  Statistically significant (» < 0.01).
dF = Degree of freedom; SS = Sum of the squares; MS = Mean square;
       F = Test statistic; and EMS = Expected mean square.
          Thus, we have confirmed the (obvious) result that the beryllium
emission rate varied significantly from day to day, and that there were sig-
nificant differences between collaborators in estimating the beryllium emission
rate.
                                     49

-------
                                                                   2
           The  component  of variance  estimates  from Table X  are:  CTfi  =  0.1892,
 CT£  = 0.1438, and  ^ =  1.7886.-   Converting  these components back  to the  y  scale
 (g/day)  yields the  theoretical results  displayed in Table XI.
                                   TABLE XI

             ERROR COMPONENTS IN BERYLLIUM EMISSION RATE (G/DAY)
Level         Average y          oe             £c         v  ac + ae

  1             0.275          0.120           0.104            0.159
  2             0.378          0.164           0.143            0.218
  3             0.245          0.107           0.093            0.142
  4             0.793          0.345           0.301            0.458
  5             0.379          0.165           0.144            0.219
  6             0.347          0.151           0.132            0.201
  7             0.322          0.141           0.122            0.186
  8             0.688          0.299           0.261            0.397
  9             0.535          0.233           0.203            0.309
 10             0.576          0.251           0.218            0.332
 11             2.675          1.164           1.014            1.544
 12             1.056          0.459           0.401            0.609
 13             1.028          0.447           0.390            0.593


          Thus, in theory the measurement error has a (uniform)  coefficient
of variation (CV) of 43.5%, and the collaborator CV is 37.9%,  so that the
standard deviation at a given level is theoretically 57.7% of  the "true value."
The average theoretical standard deviation v a^ + a|  is about 110% of the
average row standard deviation as computed directly from the  field results
(see Table IX).

          Since it is by definition Impossible to run a genuine replicate dur-
ing the field trial, the estimate of o2 from the analysis of  variance "con-
tains" any collaborator-level interaction that might exist.  Thus, a CV of 447.
within a laboratory should be regarded as an upper bound to the true value.

          It is possible to estimate o| directly by computing  the variance
of the differences in collaborator results at two similar levels, since in
theory, the variance of the independent difference XL - X2is  the variance of
XL + the variance of X£, so that the variance of (XL - X2) ^ two variances

lj  The variance of level averages is, of course, quite large.  The value is
      reported for completeness.
2/  In theory, the variance components are necessarily a uniform fraction of
      the level itself, since a log transformation was used.   (Just as
      theoretically the variance is independent of the level  if no data
      transformations are made.)

                                      50

-------
     if X} 2: X2, etc.  This way of estimating oj£ was employed and yielded
   = 0.1444 (z scale), compared to a| = 0.1892 from the analysis of variance.
Of course, this estimate is also imperfect since the data- do not yield ex-
actly  equal pairs of levels of Be.  Nevertheless, the two estimates should
bracket a|, and, since they are about the same size, it is probably true that
the measurement CV is about 40%.

          Summarizing:

          1.  The collaborator-to-collaborator variance (38% CV) and the
measurement variance  (44% CV) are about the same size and are sizable frac-
tions of the true value,

          2.  At a given level, measurements taken by various collaborators
will have a coefficient of variation of approximately 58%—this is the proper
number to estimate bounds about the true value _if there is no bias, and

          3.  The daily fluctuation in beryllium emissions rate at the
sampling location was very large (range 0.25 to 2.7 g/day) compared to ob-
servational errors.

          Finally, the filter only beryllium mass collected results and the
solution* only beryllium mass collected results were analyzed separately.
(One collaborator seldom found anything on the filter, so his results were
deleted from the corresponding analysis of variance.)  These analyses showed
that collaborators do not differ significantly in amount of beryllium col-
lected on the filter  (except, of course, the one collaborator who did not
find anything on the  filter).  In other words, almost all of the difference
in collaborators' determinations of the amount of beryllium are due to dif-
ference in the solution determination.  Also, the measurement error for the
solution determination is relatively (and absolutely, of course) larger than
the measurement error in the filter determination.  Since, on the average 77%
of the beryllium collected is from the solution, but collaborators are more
repeatible and more consistent in determining beryllium from the filter, it
would be better to multiply the filter results by an appropriate constant
than to analyze the solution jLf the filter amount had a consistent ratio to
the solution amount.  However, this is not true (see Table XII).
   Solution is  that part  of the  sample  that  results  from  a wash, e.g.,  the
     acetone wash of the  probe and  the  filter holder and  the  impinger
     contents (see Section IV).
                                   51

-------
                                TABLE XII

          RATIO OF MASS OF BERYLLIUM COLLECTED;  FILTER/ SOLUTION

  Level            Cl£/             C2              C3              C4
      1             0               0.39            0.15            0.77
      2             0               0.17            0.10            0.55
      3             0.33            0.31            3.50            0.85
      4              —             0.28            0.08            0.30
      5             0               0.14            0.19            0.39
      6             0               0.18            0.14            0.12
      7             0               0.19            1.14            0.68
      8             0               0.08            0.35            0.10
      9             0.38            0.08            0.18            0.61
     10             0               0.16            1.00            0.14
     11             0.07            0.21            0.34            1.54
     12             0.18            0.19            4.00            0.41
     13             0.11            0.10            0.33            0.15
  a/  Cl's filter results were deleted from the analysis of filter only
        readings.
          2.  Analysis of beryllium loading (stack conditions);  The re-
sponse, beryllium loading (stack conditions), in micrograms per cubic meter
was also considered, and the results compared to the analysis of variance
results using beryllium emission rate (g/day) as the response.  That is, the
components of variance were estimated from beryllium loading data and com-
pared to their counterparts from the analysis of beryllium emission rate.
The beryllium loading data are given in Table XIII.

          Because the standard deviation of a row is proportional to the
response level,  the data were log transformed (by z = log (100 y),  as before)
before the analysis of variance was executed.

          The analysis of variance results are shown in Table XIV.

          The components of variance (z scale) are:  a| = 0.194, a| = 0.153.
In terms of the original scale, then, there is a theoretically uniform mea-
surement error coefficient of variations (CV) of 44%, and a collaborator CV
of 35%.  These results are virtually identical with the results from the
analysis of beryllium emission rate (g/day).
                                    52

-------
                               TABLE XIII

                BERYLLIUM LOADING (STACK CONDITIONS (ug/m3)

 Level             Cl                C2             £3              _(*

    1             0.29             0.317           0.28           0.205
    2             0.28             0.706           0.27           0.226
    3             0.40             0.318           0.12           0.130
    4            (0.75)JL/          0.812           0.91           0.553
    5             0.37             0.635           0.25           0.276
    6             0.18             0.529           0.22           0.466
    7             0.23             0.706           0.21           0.209
    8             0.50             1.129           0.33           0.709
    9             0.47             1.271           0.30           0.168
   10             0.50             0.129           0.10           0.557
   11             2.68             5.012           1.62           1.475
   12             1.04             1.094           1.80           0.407
   13             1.09             1.483           0.90           0.590
a/  Artificial:  Collaborator's data missing.


                                   TABLE XIV

     ANALYSIS OF VARIANCE FOR BERYLLIUM LOADING (STACK CONDITIONS) (ue/m3)
Source
Collaborator (C)
Level (L)
Error (e)
dF
3
12
35
jSS
6.52781
21.8279
6.785481
MS
2.176
1.819
0.194
	 F_
11.22
9.38
..
 dF = Degree of freedom.
 SS = Sum of the squares.
 MS = Mean square.
 F  = Test statistics.
                                     53

-------
           In  fact, the correlation between the Be  loading and beryllium
 emission  rate readings is 0.99.  Therefore, the two' responses, beryllium
 loading and beryllium emission rate, produce identical estimates of error,
 i.e.,  there is no error  involved in estimating beryllium emission rate from
 beryllium loading data.
B.   Standard Samples

          The purposes of the standards data were:  (1) to provide lab-
oratory estimates of o| and a?, to compare with the corresponding
field results, and (2) to examine the biases present in the measurement of
beryllium.  Table XV shows the results where a response is the collabora-
tors' determination minus the true value (NBS's value).  Structurally the
data are a complete three factor analysis of variance with three observa-
tions per cell.  As before, the cell standard deviation is roughly propor-
tional to the level of beryllium, so a log transformation was applied and
the analysis of variance performed on the transformed data.  Results are
discussed in the original scale dig of beryllium).

          The analysis of variance results are shown in Table XVI.  The
collaborator type sample interaction is significant, and so is the three-
way interaction.

          It is easy to confuse the biases with the components of variances,
so they will be discussed separately.

         1.  Bias;  One collaborator (C2) exhibited essentially no bias on
any of the sample types, and one more collaborator (Cl) measured the filter
amounts without bias. All other estimates were biased, and with one excep-
tion (C4 on filter samples) the biases are negative, i.e., the collaborators
underestimate the- amount of beryllium.

          The CT (collaborator type) interaction term is significant because
the difference in bias between collaborators is sometimes negative (C3)  and
sometimes positive (C4) on filter samples, but the bias (if it exists) is
always negative on the other two sample types.

          The CTL (collaborator,  type, level) interaction term is signifi-
cant because one collaborator (Cl) only exhibits sizable biases at Level 3
on suspended and soluble samples,  whereas the other collaborators exhibit a
bias roughly proportional to the  level of beryllium (if they exhibit a bias
at all).

          The filter samples exhibit by far the smallest average bias, but
this is because the bias on filter determinations is sometimes positive  and
sometimes negative.


                                    54

-------
                       TABLE XV




STANDARD SAMPLE RESULTS.  RESPONSE = READING - TRUE VALUE




     (Raw Data:  Collaborators' Measurements in ug)




   Cl                    C2                     C3
Filter
Suspended
Soluble

Level
Level
Level
Level
Level
Level
Level
Level
Level

1
2
3
1
2
3
1
2
3

-0.28,
0.13,
0.8,
-0.06,
-0.38,
-1.0,
-0.24,
-0.87,
-0.8,

0.52, -0.18
0.33, 0.43
-0.2, -0.2
-0.06, 1.06
-0.38, -0.38
-1.0, -6.0
-0.24, -0.24
-0.87, -0.87
-3.8, -3.8
NBS
0.32
1.02
1.5,
0.04
-0.3,
0.7,
-0.04
0.19
-0.7,
VALUES
Filter

Level
Level
Level
1.
2,
3.
True
True
True

2.98
7.77
15.2
, 0.23, 0.25
, 0.33, -0.97
0.6, -0.1
, -0.33, -0.2
-1.19, -0.94
-0.1, -0.5
, -0.16, 0
, -0.23, 0.01
0, 0.2
-1.
-2.
-6.
-0.
-4.
-6.
-1.
-2.
-10.
IN MICROGRAMS (TRUE
38, -1.38,
77, -3.77,
2, -6.2,
16, -1.56,
08, -3.88,
5, -5.0,
24, -2.14,
67, -4.37,
6 , —6 • 8 ,
VALUES)
-1.38
-3.67
-6.2
-1.66
-2.78
-7.5
-1.84
-3.37
-9.8

Suspended
3.06
7.38
15.0
AVERAGE BIAS (READING

Filter
Suspended
Soluble

Level
Level
Level
Level
Level
Level
Level
Level
Level

1
2
3
1
2
3
1
2
3




£1
0.02
0.30
0.13
-0.39
-0.38
-2.67
-0.24
-0.87
-3.47




C2
0.27
0.13
0.67
-0.16
-0.81
0.03
-0.07
-0.01
-0.17
0.65, 0.66,
2.43, 2.03,
4.9, 5.8,
-2.64, -2.54,
-6.19, -6.45,
-12.4, -13.14,
-1.03, -1.05,
-2.72, -2.87,
-5.25, -5.0,
Soluble
3.24
7.87
15.8
0.97
1.98
4.0
-2.47
-6.26
-13.26
-1.05
-2.6
-5.07


TRUE)








C3
-1.38
-3.40
-6.20
-1.13
-3.58
-6.33
-1.74
-3.47
-9.07




C4
0.76
2.5
4.90
-2.55
-6.30
-12.93
-1.04
-2.74
-5.11

-------
                                TABLE XVI

               ANALYSIS OF VARIANCE FOR BERYLLIUM STANDARDS
Source
Collaborator
Type of Sample
Level of Be
CT
CL
TL
CTL
Error (e)
dF
3
2
2
6
6
4
12
72
SS_
1.5408
1.0466
0.7665
0.3922
0.0030
~ 0
1.6596
1.2312
MS
0.5136
0.5233
0.3833
0.0654
0.0005
0
0.1383
0.0171
F
30.04
30.60
22.42
3.82
< 1.0
< 1.0
~
"
dF = Degree of freedom.
SS = Sum of the squares.
MS = Mean square.
F  = Test statistic (F test).
                                    56

-------
           Since one collaborator always managed  to measure  beryllium with-
 out bias,  it is presumably not  an impossible  task.   It would  be  very bene-
 ficial to  distinguish this collaborators'  procedures from the others,  since
 the biases are by no means trivial in size.

           2.  Components  of variance! Converting back to the original scale,
 the theoretical measurement standard  error (ae)  is 0.096 y,*  i.e.,  in  theory
 the coefficient of variation is 9.6%. Since  replicates were  performed, ce
 can be computed per level directly from the original data.  The  three  coef-
 ficients of variation obtained  this way are 11.6%, 6.3%, and  9.2%,  respectively.
 Thus,  it appears possible that  3 u-g is harder to measure accurately than
 higher levels,  i.e., the  CV begins to diverge from 10% when the  level  of
 beryllium  is sufficiently low.   The coefficient  of variation  between collabora-
 tors (ac)  is 9.7%,  but the two  interaction components of variance are  also
 significant.

           The "total" coefficient  of  variation,  i.e., the coefficient  of
 variation  of readings at  a given level, including all contributors,  is 24.9%.
 The CV without  the  type of sample  component of variance is  23.8%.   Or, multiple
 collaborators reading a given level of beryllium in a given type sample will
 have a coefficient  of variation  23.8%, and if  (an equal mixture of)  the three
 sample types  are  employed,  the CV will be about 25%.

           Since the  measurement  error CV is 10%,  and since one collaborator
 measured without  bias  on  all three type samples at all three  levels, the difference
 between a  CV of 25%  and a  CV of  10% represents error sources  that could pre-
 sumably be eliminated  or  reduced.

           Since the  bias  situation is complicated,  a table of mean square
 errors may be helpful  in visualizing  results (see Table XVII). The mean square
 errors are based  on  a  sample size of one.

           Summarizing:

           1.  The components of variance,  a| and  a§, are about the  same size
 (CV e:  10%).  Taking  into account the type of sample variance,  the CV at a
given  level  is about 25%.

          2.  In general,  the bias is  proportional  to the amount  of  beryllium.
One collaborator measured beryllium without any essential bias but usually a
negative bias was observed (about -20%, average), and in  fact, the bias is
generally quite large compared to the  measurement variance.  Generally the
bias in observing beryllium far overshadows the unreliability  in  the mea-
suring process.  The average bias on the  filter samples was  essentially zero,
but only because large negative and positive biases cancelled  out.
   Where y is the true value of beryllium in micrograms (not the average bias).
                                    57

-------
                              TABLE XVII

          MEAN SQUARE ERRORS IN THE COLLABORATORS' MEASUREMENT OF
                        STANDARD SAMPLES

                             Cl         C2         C3            C4
 Filter
Level 1
Level 2
Level 3
0.1199
0.3296
2.1661
0.1924
0.2490
2.4460
2.0239
11.7921
49.4371
0.6971
4.8546
26.0071
 Suspended
 Soluble
Level 1
Level 2
Level 3
0.2781
0.3765
7.4260
0.1516
0.8882
1.9980
1.4029
13.0485
42.0660
6.6285
39.9221
169.1820
Level 1
Level 2
Level 3
0.1988
0.9890
14.0380
0.1461
0.2322
2.0260
3.1688
12.2730
84.2620
1.2228
7.7397
28. 1092
 C.  Velocity Profiles

          Every velocity reading contains the influences of the time when the
 sampling was taken (t) and the position (P) where it was taken.  The effects
 of  t and P upon velocity must be identified before an accurate velocity pro-
 file can be drawn.  Since each collaborator samples at one position at a time,
 only four (of 30) positions are sampled per time block.

          A velocity profile could be drawn by simply averaging the velocity
 readings on each position and ignoring the effect of time.  To check the
 validity of such a procedure, the t and P effects were separated for Run 1,
 and the results compared to the simple average estimates of the t and P
 effects.  To accomplish this, the 120 velocity observations (in Run 1) were
 considered as 120 of Che possible 900 (30 t's x 30 P's) treatment combinations
 of  time and point.  The analysis of variance of this incomplete design fur-
 nished estimates of the velocity at each t and P, and these estimates were
 compared to the simple average results.  Following standard statistical
 terminology, the t and P effects estimated from the incomplete analysis of
 variance model will be called "adjusted" estimates, and the simple average
 results will be referred to as "unadjusted" values.

          A complication arose in the analysis of variance because not all
points (or times) were coupled.  For instance, Points 1, 6, 7, 12, 13, 18,
 19, 24, 25, and 30 (see Table XVIII) for a block uncoupled with the other
points (times).   In any given time block, either four of the Points 1, 6, 7,
                                     58

-------
12, 13, 18, 19, 24, 25, and 30 appear or none of these points appear, i.e.,
these 10 points can be interrelated but are not related to the other points.
The data,  then, are not really an incomplete 30 x 30 factorial but is
instead a set of three (incomplete) 10 x 10 factorials.
                               TABLE XVIII

                    IDENTIFICATION OF SAMPLING POINTSi/

Port 10          Port 9          Port 8          Port 7          Port 6

  P6               P12            P18              P24             P30
  P5               Pll            P17              P23             P29
  P4               P10            P16              P22             P28
  P3               P9             P15              P21             P27
  P2               P8             P14              P20             P26
  PI               P7             P13              P19             P25

Port 1           Port 2          Port 3          Port 4          Port 5
aj  See Figure 3.
          Therefore, three analyses of variance were performed, and the re-
sults are shown in Table XIX.

          The significantly large F-values mean that time and sampling point
do have an influence on velocity. Therefore, the adjusted estimates of
velocity at each point are compared to the unadjusted (average) velocities
per point* (see Table XX).

          There is no significant difference between the adjusted and unad-
justed velocities per point.  In fact, the Spearman rank correlation between
the two sets of readings is +0.99.  In other words, even though each particular
point was observed in only four of the 30 time blocks, the average velocity
reading per point is virtually the correct estimate of velocity.
   For completeness,  the analysis comparison  of times was  also made,  although
     it is not shown  here.   The results  were  similar.


                                   59

-------
                                 TABLE XIX

                    ANALYSIS OF VARIANCE FOR VELOCITIES


A.  Block A: (Points 1, 6, 7, 12, 13, 18, 19, 24, 25, 30)

          Source       dF         SS            MS         F
Total
R (lift)
R (H»P)
R (tln,P)
R (PlH,t)
Error
40
10
10
9
9
21
106,267,087
106.075,422
105,831,460
398,463
154,488
37,171



44,274
17,165
1,770



25.01
9.70

B.  Block B:  (Points 2, 5, 8, 11, 14, 17, 20, 24, 26, 29)

          Source       dF         SS            MS         F
Total
R (li»t)
R (H,P)
R (tl(i,P)
R i P 1 1 1 t*l
Error
40
10
10
'9
9
21
116,276,613
116,050,664
116,139,593
110,370
198,044
27,278



12,263
22,005
1,299



9.44
16.94

C.  Block C:  (Points 3, 4, 9, 10, 15, 16, 21, 22, 27, 28)

          Source       dF         £S            MS         F

         Total         40     120,136,155
         R (n,t)       10     119,911,193
         R (n,P)       10     120,079,787
         R (tlp,,P)      9          18,342      2,038      1.13
         R (Pln,t)      9         186,929     20,770     11.47
         Error         21          38,030      1,811
                                     60

-------
                                TABLE XX

             COMPARISON OF ADJUSTED VS. UNADJUSTED VELOCITIES
                          PER SAMPLING POINT
   Point
                              a/
Adjusted Velocity-
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
1707
1800
1816
1755
1643
1550
1626
1666
1675
1704
1688
1590
1543
1600
1627
1653
1621
1566
1656
1721
1731
1753
1740
1655
1726
1843
1864
1792
1696
1573
Rank     Unadjusted Velocity-      Rank

 19              1727               20
 27              1811               27
 28              1822               28
 25              1750               22.5
 10              1644               11
  2              1549                1
  8              1614                7
 14              1661               14
 15              1657               13
 18              1700               18
 16              1689               16
  5              1607                5
  1              1563                3
  6              1608                6
  9              1626                9
 11              1642               10
  7              1618                8
  3              1561                2
 13              1646               12
 20              1716               19
 22              1727               21
 24              1753               25
 23              1750               24
 12              1672               15
 21              1744               22.5
 29              1838               29
 30              1853               30
 26              1783               26
 17              1690               17
  4              1571                4
a/  Computed from incomplete analysis of variance.
b/  Simple average reading per point.
                                     61

-------
                        VIII.  CONCLUSIONS

          This collaborative test comprised  13 runs, each on a
different day, where four different collaborative organizations sampled
simultaneously over the same 30-point traverse, with each point being
sampled 8 min by each collaborator.  The emission levels of beryllium in
the stack sampled were low, being in the neighborhood of one-tenth that of the
permissible standard emission rate.  In two cases, the collaborators sub-
contracted the chemical analysis of their samples.  It is probable that one
of the subcontractors had not analyzed these types of beryllium samples by
the procedure specified in Method 104.  Collaborators deviated from this
method with probable adverse consequences.

          The major conclusions that can be drawn from the results of this
collaborative test are:

          1.  Using Method 104, the bias and the measurement error
will be quite significant, with the bias contributing more to the unreli-
ability of the beryllium determination.  At a given level, measurements will
have a coefficient of variation of approximately 58%.

          2.  Method 104 is adequate as written if a coefficient of varia-
tion of approximately 58% for the field measurement results, at a given
level of beryllium, is acceptable.

          3.  The lack of precision in the results of the test samples is
exclusive of the amount of beryllium collected on the filters since these
measurements were repeatble by a collaborator and 77% of the error was in
the beryllium measurements from the solution samples.

          4.  The biases of the solution portion of the samples are unpre-
dictable.

          5.  There is not sufficient information from this test to determine
whether the bias problem is due to field sampling, laboratory analysis, or
both.
                                    63

-------
                          IX.  RECOMMENDATIONS

          Based upon the conclusions that have been drawn from the results
of this collaborative test, it is recommended that an investigation be under-
taken to determine the reasons for the significant biases.   If all collabora-
tors had measured high or low by some significant amount, this investigation
might not be needed, but because of the amount of bias and because the bias
is not systematic—at times it depends on level, at other times it does not,
etc.—and because it is not known whether or not the bias is due principally
to field sampling or the analysis of samples procedures, the investigation
is needed.
                                    65

-------
                     APPENDIX A
"METHOD 104—REFERENCE METHOD FOR DETERMINATION OF
   BERYLLIUM EMISSIONS FROM STATIONARY SOURCES"
                          67

-------
 8S4fi
      RULES AND  HGULATIONS
meHsurements. The ba^U for such estimates
shall be given La the test report.
  4-1  Preparation  of  sampling  train.—
41  1  Assemble the sampling train aa shown
In  figure 103-1. It is recommended that all
t-l.tasware be predefined by soaking In  wash
ncicl for  2 hours.
  4 4.2  Leak check the sampling train at the
sampling site. The leakage rate should cot be
In  excess of 1 percent of the desired sample
rate.
  4 5  Beryllium  train operation.—4.5.1  For
e.icii run. measure the velocity at the selected
sampling point.   Determine  the isokinetlc
sampling rate. Record the velocity head and
tiie required sampling rate.
  452   Place- the-  nozzle at  the sampling
point with the tip pointing directly into the
gas stream. Immediately start the- pump and
adjust the flow to tsoktnetic conditions. At
the conclusion of the test, record the  sam-
pling rate. Again  measure the velocity head
at the sampling point. The required Isoklnetic
rate at the end of the period should not have
.deviated more than 20 percent  from,  that
originally calculated.
  4.S.3   Sample at  a minimum rate of 0.5
If/mln  Samples  shall  be  taken over such a
period or periods as are necessary to- deter-
mine tha maximum emissions  which would
occur in a 24-hour period. la the case or
cyclic operations, sufficient  tests  shall be
mode so as to allow determination or calcu-
lation o( the emissions which would occur
over  the duration of the cycle. A minimum
sampling time of 2 hours is recommended.
  4.5.4-  AIL pertinent  data should be in-
cluded in the- test report.
  4.8  Sample recovery.—4.8.1  It Is. recom-
mended  that  all glassware be precleaned aa
In  § 4 4.1.  Sample recovery should also be
performed In. an area free of possible beryl-
lium contamination.  When the sampling
train  Is  moved,  exercise  care  to prevent
breakage and  contamination. Set aside a por-
tion  of the acetone used  in  Che sample re-
covery as a  blank; for analysis. The  total
amount  of acetous used should be measured
for accurate blank correction. Blanks can be
eliminated If  prior analysis shows negligible
amounts.
  4 6.2   Remove the niter and any loose par-
tlculate  matter from filter holder and  place
in a container.
  4.6.3   Clean the probe with acetone and a
brush or long rod and cotton balls. Wash into
the container. Wash out the  filter holder
with acetone and add to the same container.
  4.7  Analyst!.—4.7.1  Make the  necessary
preparation of samples and analyze for beryl-
lium. Any currently acceptable method such
as  atomic absorption, spectrographic, fluoro-
metric, chromatographic, or equivalent  may
be  used.
   5. Calibration   and 'standard*—5.1  Sam-
pling  train.—5.1.1  As  a  procedural check,
sampling rate regulation should be compared
with a dry gas meter, spirometer, rotameter
 (calibrated for prevailing atmospheric con-
ditions), or  equivalent, attached to nozzle
inlet of  the complete sampling train.
  5.1.2   Data from this test and calculations
should be shown in test report.
  5.2  Analysis.—5.2.1   Standardization   Is
made aa suggested  by  the manufacturer of
the instrument  or the procedures for the
nnalyttcal method.
  6. Calculations—6.1 Total beryllium emis-
sion. Calculate the total  amount of beryl-
lium emitted  from each  stack  per day by
equation 103-2. This equation is applicable
for continuous operations. For cyclic opera-
tions, use  only the  time per day each stack
 Is  in operation.  The  total beryllium >«mis-.
sinns from a  source will be the summation
 of  results from all stacks.
                     86,400 seconds/day
   K -   i"*-* v
when:
      R- Rale o( emission, g/Oav,
     »', .-- Tot . J wdKht of iMrylliam collMted, «(.
   V,.,,.|xTui;il vi'liunfof KM sajnplwl. ft'.
  I»j4.«.-Av«m(» stack nu velocity, feet per second.
     .-l. = Sljtk area, ft'.

  7.  Test  report. 7.1  A test report shall be
prepared which shall Include as a minimum:
  7.1.1  A detailed description of the sam-
pling train used and results of the proce-
dural check with all data and calculations
made.
   7.1.2  All pertinent data  taken  during
test, the basis for any  estimate* made,  cal-
culations. and result*.
  7.1.3  A description of the  test site. In-
cluding a block diagram with a brief  de-
scription of the process, location of the sam-
ple points In  the cress section,  dimensions
and distances  from any point of disturbance.

METHOD 104. BEFKItENCV MTTHOO  TOS DCTW-
  MtMAriON or  BXBTUJUM.  UUSSIO'NS FBOM
  STATIONAJIY  SOCTRC1S

  1.   Principle an* applicability — 1.1  Prin-
ciple. — Beryllium emissions- are Isoklnetieal-
ly sampled from the source, and tne collected
sample Is digested in an acid solution and
analyzed by atomic absorption spectrophc-
tometry.
  1.2  Applicability.—This method Is appli-
cable  for  the determination  of beryllium
emissions in  ducts  or  stacks  at  stationary
sources.  Unless   otherwise  specified,  this
method  is  not Intended  to  apply  to gas
streams  other than those  emitted directly
to   the   atmosphere   without   further
processing.
  2.   .Apparatus—2.1    Sampling train.—A
schematic  of the sampling train used by
EPA Is shown In figure  104-1. Commercial
models of this train are available, although
construction details are described In APTD-
0581,'  and operating and maintenance  pro-
cedures  are  described  in APTD-0578.  The
components essential to this sampling train
are the following:
  2.1.1  Nozzle.—Stainless steel or glass with
sharp, tapered leading  edge.
  2.1.2  Probe.—Sheathed  Pyre*3 glass.  A
heating  system  capable  of maintaining  a
minimum gas temperature In the range of
the stack temperature  at the probe outlet
during sampling  may  be used  to prevent
condensation from occurring.
        PROBE
 TYPE S     /
 PITOT TUBE'
                             HEATED AREA   FILTER HOLDER   THERMOMETER   CHECK.
                                                                  /       ^.VALVE
                                       IMPINGERS             ICE BATH
                                               BY-PASS VALVE
                                                                             VACUUM
                                                                              LINE
                                                               VACUUM
                                                                GAUGE
                                                        MAIN VALVE
             THERMOMETERS
                         DRY TEST METER
AIR-TIGHT
  PUMP
                          Figure 104-1.  Beryllium sampling train
  2.1.3  Pitot  tube.—Type  3 (figure 104-3),
or equivalent,  with a coefficient within 5 per-
cent  over the working  range, attached  to
probe to monitor stack gas velocity.
  2.1.4  Filter holder.—Pjn*. glass. The filter
holder must provide  a positive seal against
leakage from  outside or around the filter.
A heating system capable of maintaining the
niter at a minimum temperature in the range
of the  stack  temperature  may  be used  to
prevent condensation from occurring.
  2.1.5  Impingm.—Four  Qreenburg-Smith
implngers connected In series with glass ball
Joint  fittings.  The  first, third,  and fourth
impingers may be modified by replacing the
tip with a '/,-lnch  l.d. glass tube extending
to one-half Inch from  the  bottom of the
flask.
  2.1.6  Metering  system.—Vacuum gauge,
leakless pump,  thermometers  capable  of,
measuring temperature to  within 5* V, dry
gas meter with 2 percent accuracy, and re-
lated  equipment, described  In  AFTD-0581,
to maintain an Isoklnetlc sampling rate and
to determine sample volume.
  2.1.7  Barometer.—To  - measure   atmos-
pheric pressure to it 0.1 in Hg.
  2.2 Measurement  of  stack   conditions
(stack pressure,  temperature, moisture ana
velocity)—22.1   Pitot  tube.—Type  S,  or
equivalent, with a coefficient within 5 percent
over the working range.
  2.2.2  Differential  pressure   gauge.—In-
clined manometer, or equivalent, to measure
velocity head  to within  10  percent  of the
minimum value.
  1 These documents are available for a nom-
inal cost  from the  National  Technical In-
formation Service, U.S.  Department of Com-
merce.  5285  Port Royal  Road, Springfield,
Va. 22151.
  3 Mention of trade names on specific prod-
ucts does  not constitute endorsement by tha
Environmental Protection Agency.
                                      FEDERAL REGISTER, VOL 38, NO. 66—FRIDAY,  APRIL 6, 1973
                                                            68

-------
                                                   RULES  AND REGULATIONS
                                                                                                                               8847
 however, most  sample  sites differ to some
 degree and  temporary  alterations such  as
 stack  extensions or expulsion* often  are re-
 quired to insure the  best possible  sample
 site. Further, sloe* beryllium  Is hazardous.
 care should  be  taken to mlnlmice exposure.
 Finally, since the total quantity of beryllium
 to be  collected  is quite small, the test must
 be carefully conducted  to  prevent con tarn I -
 nation or loss of sample.
   4.3   Selection of a sampling site and mini-
 mum  number of  traverse points.
   4.2.1  Select  a suitable sampling site that
 is as close as practicable to the point of at-
 mospheric   emission.  If  possible,  stacks
  •FIstnlM*. Pliatlut*-
   2.2.3  Temperature gage.—Any tempera-
 ture measuring'device to measure stack tem-
 perature to within 6* F.
   2.2.4  Pressure  gage.—Pilot tube and In-
 clined manometer, or equivalent, to measure
 stack  pressure to within 0.1 In Hg.
   2.2.6  Moisture determination.—Wet and
 dry bulb  thermometer*, drying tubes, con-
 densers, or equivalent,  to  determine stack
 gas moisture content to within l percent.
 •'  2-S  Sample recovery—2.3.1  Probe  alean-
 -inf rod.—At least as long,as probe.
 -  2.3.2  Leofciew  ghat  sample bottle*.—600
 •ml.
 ••  3.3 J  Graduated cylinder.—260 ml.
   2.3.4  Plastic iar—Approximately 800 ml.
   3.4  Analyst*—2.1.1  Atomic    absorption
 rpectrophotometer—To measure abeorbance
 at 2343 Tim  Perkln Elmer Model 303, or
 equivalent, with N,O/aoetylene burner.
   2.4.2  Hot plate.
   2.4.3  Perchloric acid fume hood.
   3.   Reagent*—3.1  Stock   reagents.—3.1.1
 Hydrochloric acid.—Oonocntrat«d.
   3.1.2  Perchloric  acid.—Concentrated,  70
 percent.
   3.1.3  Nitric acid.—Concentrated.
   3.1.4  Sulfuric acid.—Concentrated.
   8.1.6  Distilled and deioniMd water.
   3.1.6  Beryllium powder;—08 percent mini-
 mum  purity.
 .-  32  Sampling—8.2.1   niter. — Milllpore
 AA, or  equivalent.  It  Is suggested that  a
 Whatman  -41  filter  be placed Immediately
 -against  the back side of the Milllpore filter
 as a  guard against  breaking the MUllpore
 filter.  In the analysis of the niter, the What-
 man 41 filter should be Included with tb«
' Milllpore filter.
   333  Silica gel,—Indicating type, 6 to 16
 mesh, dried at 850*  F for 2 hours.
•  323  Distilled  and deionined water.
 .  3.3  Sample recovery—33.1   Distilled and
 deioniaed  water.
   3.3.2  Acetone—Reagent grade.
   3.3.3  Wash. acid.—1.1  V/V  hydrochloric
 acid-water.
   3.4  Analysis.—3.4.1  Sulfuric  acid  solu-
 tion, 12 AT.—Dilute 333 ml  of  concentrated
 sulfurlc acid to 1 1  with distilled water.
 •  8.4.2  25 percent  V/V hydrochloric  acid-
 toater.
   3.6  Standard   beryllium  solution—3.5.1
 stock  solution.—l  ng/tnl  beryllium.  Dis-
 solve 10 mg of beryllium in 80 ml of 12 N
 Bulfuric acid solution and dilute to a volume
 of 1000 ml  with distilled water. Dilute a 10 ml
 aliquot to  100 ml with 25 percent V/V hydro-
 chloric acid,  giving  a  concentration  of  1
 dE/ml. This dilute stock solution should be
 prepared fresh dally. Equivalent strength (in
 beryllium)  stock solutions may be prepared
 from beryllium salts  as Bed, and Be(NO.).
 (98 percent minimum purity).
   4. Procedure. 4.1   Guidelines for source
 testing are detailed in the following sections.
 These  guidelines are generally applicable;
                                                                                          smaller than 1 foot In diameter should not
                                                                                          be sampled.
                                                                                            4.3.2  The «-mp"ng site should be at lesst
                                                                                          B stack or duct diameters downstream and
                                                                                          2 diameters upstream from any flow disturb-
                                                                                          ance such  as a bend,  expansion or  contrac-
                                                                                          tion. For a rectangular cross-section,  deter-
                                                                                          mine  an  equivalent  diameter  from  the
                                                                                          following equation:
                                                                                                                            eq.  104-1
                                                                                                         L+w
                                                                                          where:
                                                                                            D.=equlv»lent diameter
                                                                                             t=length
                                                                                            W=width
                                                                             IMJMKR OF DUCT DIAMETERS UfSTK AM"
                                                                                     (DISTANCE A)
               FROM POINT Or ANY TYPE 0*
               DISTURBANCE (BFXD. EXPANSION. CONTRACTION. ETC
                              •JUMKft OF OUCT DIAMETERS DOWNSTOAW"
                                     . .  (DISTANCE •»
                           Figure i01-3. Minimum nunoet ot traverse points.
   tnnm points on pwpmllcvlir dlmun.
              I
 Fljur. 104-5. Crwi Mellon of nclugulv tuck dl«M*d Into 12 «(*>
 •rau, KHA inmM poinu 11 MntftM ol tun SIM.

  4.2.3  When  the  above sampling site cri-
teria  can be met, the minimum number of
traverse points Is four  (4) for stacks 1 foot
In diameter or leas, eight (6) for stacks larger
than 1 foot but 2 feet In diameter or less, and
twelve (12)  for stacks larger than  2 feet.
  4.2.4  Some sampling situations may ren-
der the above sampling site criteria Imprac-
tical. When this is the case, choose a con-
venient  sampling  location  and  use figure
104-3 to  determine the minimum number
of traverse points. However, use figure  104-3
only for stacks 1 foot in diameter or larger.
  4.2.6  To  use figure 104-3,  first measure
the distance from  the chosen  sampling lo-
cation to the nearest upstream and down-
stream disturbances. Divide this distance by
the diameter or equivalent diameter to deter-
mine the distance in terms of pipe diameters.
Determine  the  corresponding number of
traverse points for each distance  from  fig-
ure 104-3. Select the higher of the two num-
bers of traverse points, or  a greater value,
such that for circular stacks the number Is
a multiple of four, and for rectangular stacks
the number follows the criteria  of  section
4.3.2.
  4.2.6  If a selected sampling point is closer
than 1 inch from the stack wall,  adjust the
location  of  that point  to  ensure that  the
sample is taken at least 1 inch away from the
wall.
  4.3 . Cross-sectional layout and location of
traverse points.
                                     FEDERAL  REGISTER, VOL 3«, NO.  66—FRIDAY, APRIL 6.  1973
                                                                   69

-------
           Table 104-1.  Location of traverse points in circular stacks
           (Percent of stack diameter from Inside wall to traverse  point)
Traversa
polnc
number
on a
diameter
1
2
3
4
£
6
7
8
9
10
11
12
13
: 14
15
16
17
18
19
20
21
22
23
24
Number of traverse points on a diameter
2
14.6
85.4




















4
6.7
25.0
75.0
93.3


















6
4.4
14.7
29.5
70.5
85.3
95.6











'




8
3.3
10.5
19.4
32.3
67.7
80.6
89.5
96.7






,

-





10
2.5
8.2
14.6
22.6
34.2
65.8
77.4
85.4
91.8
97.5












12
2.1
6.7
11.8
17.7
25.0
35.5
64.5
75.0
82.3
88.2
93.3
97.9



•

"




14
1.8
5.7
9.9
14.6
20.1
26.9
36.6
63.4
73.1
79.9
85.4
90.1
94.3
98.2








16
1.6
4.9
8.5
12.5
16.9
22.0
28.3
37.5
62.5
71.7
78.Q
83.1
87.5
91.5
95.1
98.4






IB
1.4
4.4
7.5
10.9
14.6
18.8
23.6
29.6
38.2
61.8
70.4
76.4
81.2
85.4
89-.1
92.5
95.6
98.6





20
1.3
3.9
6.7
9.7
12.9
16.5
20.4
25.0
30.6
38.8
61.2
69.4
75.0
79.6
83.5
87.1
90.3
93.3
96.1
98.7



22
1.1
3.5
6.0
8.7
11.6
14.6
18.0
21.8
26.1
31.5
39.3
60.7
68.5
73.9
78.2
82.0
85.4
88.4
91.3
94.0
96.5
98.9


24
1.1
*3.2
5.5
7.9
10.5
13.2
16.1
19.4
23.0
27.2
32.3
39.8
60.2
67.7'
72.8
77.0
80.6
83.9
86.8
89.5
92. -1
94.5
96.8
93.9
  43.1  For circular stacks  locate the tra-
verse points on at least two diameters accord-
ing to figure 104-4 and table 104-1. The tra-
verse axes shall divide the stack cross section
Into equal parts.
  433  For rectangular  stacks  divide the
crocs section Into as many equal rectangular
areas as traverse polntt,. such that the ratio
of the length to the  width of the elemental
areas is between  1 and i Locate the traverse
points  at the ccntrold of each  equal  area
according to figure 101-6
  4.4 Measurement  of  stack  conditions —
441  Set up  the appaiatus as shown In fig-
ure  104-2  Make sure all  connections are
tight and  leak  free  Measure tho velocity
head and temperature at the traverse points
specified by it 4 8 and 4 3.
  442  Measure the static pressure  In the
stack.
  4 4.3  Determine the atack gas moisture
  444  Determine the stack  gas molecular
weight from the measured moisture content
and knowledge  of  the expected  gas  stream
composition A standard Orsat analyzer has
been found valuable at combustion sources
In  all coses, sound engineering  judgment
should be used
  4.8  Preparation  of sampling train—451
Prior to assembly, clean all glassware (probe.
Implngers,  and  connectors) by  sqaklng  In
wash acid  for 2 hours Place 100 mil of dis-
                                           tilled water In each of the first two ImprLng-
                                           ers. leave tho ihtra Impinger empty, and place
                                           approximately 200 g of preweighted silica gel
                                           in the fourth Implnger. Save a portion of the
                                           distilled water  as a  blank In the' sample
                                           analysis. Set up the train and the probe as
                                           In figure 104-1. -
                                             462  Leak check the sampling train at the
                                           sampling site The leakage rate should not be
                                           In excess of 1 percent of the desired sampling
                                           rate  If condensation In the probe or filter is
                                           a problem, probe and  filter heaters will be
                                           required. Adjust  the  heaters  to  provide a
                                           temperature at  or above  the stack tempera-
                                           ture  However, membrane filters such as the
                                           Mllllpore AA are limited  to about 225" P. If
                                           the stack  gas Is In  excess of about 200*  F.
                                           consideration should be given to an alternate
                                           procedure  such as moving  the  filter holder
                                           downstream of  the  first  Implnger to Insure
                                           that the filter does not exceed Its tempera-
                                           ture limit. Place crushed Ice around the Im-
                                           plngers Add more ice during the test to keep
                                           the temperature of the gases leaving the labt
                                           Implnger at 70° P. or loss
                                             4 6  Beryllium tram operation —4 6 I  For
                                           each, run. record the data required on the
                                           example sheet shown Lu figure  101-6. Take
                                           readings  at each  sampling point at least
                                           every 5 minutes and when significant changes
                                           In stack conditions necessitate additional ad-
                                           justments In flow rate
                                             4.6 2  Sample at a rate of 0 5 to 1 0 ft '/rnln.
                                           Samples shall be taken over such a period or
                                           periods as are necessary to accurately deter-
                                           mine the maximum emissions which would
                                           occur In a 24-hour period. In  the case of
                                           cyclic  operations,  sufficient tests  shall  be
                                           made so as to allow accurate determination
                                           or calculation of the emissions which will
                                           occur over the duration  of the cycle. A mini-
                                           mum sample time of 2 hours Is recommended.

M1MT
MCAiur
OhXMM
Bin



WNI«n^
MUM II
•nuia
WIU.H
crtcici
IMVnBMMT












IOTAL
UhO.



UVUHC
»•*
w —•













tvlUGE
IM1IC
Fsl k.H»








.






fuel
IW-'f














AU3IEMT Ttumnnf
utauiiuc Mil um

KMUMie or suaaau union
vnocnv
HUD.
Ut).














NUM.
nriupaiM.
•ana
OMKI
mm
UNI.
h.N*l














\
cwuvu
as*















ASUWO UO
HU1UIOI
n&aiiNG
NOUUDIU
I1IUIC 1.
uniua

UTtl U_
not HUTU ""1"°
aUUUUIDHIMUK
AIORGUMIU
MUI
It.^L'r












A.1
euiai
ll«_jl.V












Ail '
»««
iMum
llWUMUff.
•1














•Mica
•»














                          Fill 1M-0.1 Field datt

  463  To begin sampling, position the noz-
zle  at the  first  traverse  point with the  tip
pointing directly Into the gas stream  Imme-
diately start  the pump and adjust the flow
to Ibaklnetlc  conditions  Sample  for at least
6 minutes at each traverse point; sampling
time must be the same for each point. Main-
tain Isoklnetlc sampling throughout the sam-
pling  period  Nomographs which  aid In the
rapid  adjustment of the sampling rate with-
out other computations  are  in  APTD-0576
and are available from commercial suppliers.
Note that standard monographs are applica-
ble only for type S pltot tubes and air or a
stack gas with an equivalent density. Con-
tact EPA or the sampling train supplier for
Instructions when the standard monograph
Is not applicable
  464  Turn off the pump at the conclusion
of each run and record the final readings.
Immediately remove  the  probe and  nozzle
                                                                                                                                                                             m
                                                                                                                                                                             O

                                                                                                                                                                             5
                                                         FEDERAL RCC'.STER, VOL  3S, 110. 6S—?K'OAY, APRIL 6,  1973

-------
                                                   RULES AND  REGULATIONS
                                                                                  8849
 from tbe stack and bandle ID accordance with
 the sample recover? process described in 14 7.
   47  Sample  recovery—171   (All   glass
 storage  bottles and the graduated cylinder
 must be precleaned as In 11.5.1.) This opera-
 tion should be performed In  an area free of
 possible beryllium contamination. When the
 sampling train is moved, care must be exer-
 cised to prevent breakage and contamination.
   4 72   Disconnect the probe from the un-
 plnger train. Remove tbe filter and any  loose
 paniculate matter from the filter bolder and
 place in a sample bottle. Place tbe contents
 (measured to ±1  ml) of the first three 1m-
 plngers Into another sample bottle Rinse the
 probe and all  glassware between It and the
 back bait  ol tbe  third Implnger with water
 and acetone, and  add this to  tbe latter sam-
 ple bottle  Clean tbe probe with a brush or a
 long Blender rod and cotton balls Use acetone
 while cleaning. Add these to the sample bot-
 tle Retain a sample of the water and acetone
 as a blank. Tbe total amount of wash water
 and acetone used  should be measured for ac-
 curate blank correction. Place tbe silica gel
 In the plastic jar. Seal and secure all sample
 oontalnen lor shipment If an additional test
 Is desired, the glassware can be carefully dou-
 ble rinsed  with distilled water and reassem-
 bled. However. If the glassware la to be out of
 use more  than  2 days,  tbe Initial  acid
 wash procedure  must  be  followed.
   48  Analysis.
   4.81   Apparatus  preparation —Clean all
 glassware according to  the procedure of sec-
 tion 4.61. Adjust tbe Instrument settings
 according  to the  Instrument manual, using
 an absorption  wavelength of 234.8 nm.
 •482   Sample preparation—The digestion
 of beryllium samples Is accomplished In part
 in concentrated • perchloric  add. Caution:
 The analyst must Injure that tbe sample is
.heated to light brown fumes after the Initial
 nitric  add addition; otherwise, dangerous
 perchlorates may result from  toe buuocquou.
 perchloric  acid digestion Perchloric acid also
 should  be  used only under a perchloric add
.hood.
   4821  Transfer the filter and  any loose
 paniculate matter from the sample container
 to a ISO ml beaker. Add 35 ml concentrated
 nitric acid. Beat on a botplate until light
 brown fumes are evident  to  destroy all or-
 ganic matter  Cool to room temperature and
 add 6 ml  concentrated sulfurio acid  and S
 ml concentrated perchloric acid. Then  pro-
 ceed with  step 442 4.
   4822  Place a portion  of  the water and
 acetone sample Into a 160 ml  beaker and put
 oa a hotplate. Add portions of tbe remainder
 aa evaporation proceeds and evaporate to dry-
 ness Cool  the residue and add 35 ml concen-
 trated nitric eeld. Beat on e> hotplate  until
 light brown fumes are evident to destroy any
 organic  matter.  Cool to room temperature
 and add 6  ml concentrated sulfurlc acid, and
5 ml concentrated perchloric add. Then pro-
ceed with step 4.B.3.4.
  4 B 2 3  Weigh tbe spent silica gel and re-
port to the nearest gram.
  4824  Samples  from 4.821  and 4.8.22
may be combined here  for ease of analysis.
Replace on a hotplate and evaporate to dry-
ness In a perchloric aeld hood. Cool and dis-
solve tbe  residue in 10.0 ml of 35 percent
V/V hydrochloric  acid. Samples  are  now
ready for  the  atomic absorption unit. The
beryllium  concentration of the sample must
be within  the calibration range oi tbe unit.
If necessary, further dilution of sample with
25 percent V/V hydrochloric aeld  must be
performed to bring the sample -within the
calibration range.
  483   Beryllium  determination. — Analyse
the  samples prepared In  482  at 234 S nm
using a nitrous oxide/acetylene flame. Alumi-
num. silicon and other  elements can later-
fere with  this  method  If present  in large
quantities  Standard methods are available,
however, to effectively eliminate these inter-
ferences (see Reference 6).
  5.  Calibration— S.I   Sampling  train.—
5.1 1 Use  standard  methods and equipment
as detailed la APTD-0576 to calibrate tbe rate
meter, pltot tube, dry gas meter and probe
beater  (If used). Recalibrate prior to each
test aeries
  62   Analytif.—621   Standardization  Is
made with the procedure as suggested by the
manufacturer with standard beryllium solu-
tion. Standard  solutions  will  be prepared
from the stock  solution by dilution with 25
percent V/V hydrochloric acid. The linearity
or working range should be established with
a series of standard solutions. If collected
samples are out of the  linear range, the
samples should  be diluted. Standards should
be interspersed with the samples since tbe
calibration can  change slightly with time.
  !  C:'svlsi'.sr.s  S I  Average dry gat meter'
temperature, stock temperature, stack pres-
sure end average orifice prettvre drop. — See
data sheet (figure 104-6).
  62 Dry gat volume. — Correct the sample
volume measured by the  dry gas meter to
stack conditions by using equation 104-2.'
            _r
        V"-   Vm
                                               63  Volume of water vapor.
                 Tm      P.
                                 eq. 104-2
where.
  K. -Volume of gas amplr through tbe dry gumetei
        (stack conditions). fl>.
  V.— Volume of gu sample through tin dry gas meter
        0)">(M figure 104-b).
       P.—Stack pressure, Pt.r±static pressure. In
            Hg.
       AC-Molecular weight ol stack gas (wet basis).
            tbe summation of (be products of the
            molecular weight of  each component
            multiplied by its volumetric proportion
            In tbe mixture, Ib/lb-mole.

FINAL
MIDI*
UOUtD COUSfttO
TCT/U. VOU*C COUfCTD
VOUMCOFUOUIO
MTOEOUICIO
NKHOn
VOUHE.




sue* co.
worn.
g



r| -
         TOfHTRTOvoLUcir diving total Might
MCMAftlrGEMTIOFMin. II (Mb           •««'S"*
                            * vou*t- **"*• •*

          Plain 104-7. AaalyUcal data,
                                     FIDEBAL UCI5IER. VOL. 38, NO. 66—MIOAr,  APRIL 6,  1973


                                                                  71

-------
8850
                    Mfllf XND  RI6UUTIONS
 PLANT.

 DATE_
 RUN NO.
 STACK DIAMETER, in..
 BAROMETRIC PRESSURE, in.
 STATIC PRESSURE IN STACK (Pg). In. Hg.
 OPERATORS.
                                SCHEMATIC OF STACK
                                   CROSS SECTION
         Traverse point
            number
Velocity head,
   in. H2O
                                 AVERAGE:
                                                                Stack Temperature
                          Figure 104-8. Velocity traverse data.
                                                            08  laoMnetic variation  (comparison of
                                                          velocity of got in probe tip to stuck velocity).

                                                                       ,   100V,.,.,
                                 eq. 104-8,
 where.
       I- Percrnt ol Igoklneuc sampling.
    Vi.t.1" Total volume of gudainub (alack conditions).
         ft'
     M.-Probe tip area, ft'.
      6-Sampbng tin*, sec.
   <».)....- Avenue stack gu velocity. [Ml per SA.OOU.

   7. Evaluation 'of remltt—7 1   Determina-
 tion of compliance —7.1 1  Each performance
 test shall  consist of three repetitions of the
 applicable test method. For the purpose of
 determining compliance with an applicable
'national emission standard, the average of
 results of all repetitions shall apply.
   7.2   Acceptable  bofcinetfc results—7.3 r-
 The following range seta the limit on accept-
 able Isoklnetlc sampling results:           '
   If 00 percent ^1^110 percent, the results
 are acceptable; otherwise, reject the test and
 repeat.
   7. Reference!.—1. Addendum  to Specifica-
 tions for Incinerator Testing at Federal Facil-
 ities, PBS. NCAPC. December a.  1967.
   2. Amos, M. D. and WUlls, J. B., "Use or
 High-Temperature  Pre-Mixed   Flames  in-
 Atomic Absorption Spectroscopy,"  Spectre—
 eblm. Acta, 23: 1325. 1066
.   3. Determining Dust  Concentration  In a
 Gaa Stream. ASMS Performance Test  Code
 No 27, New York, N.T., 1997.
   4. Devorkln. Howard et al.. Air Pollution
 Source Testing Manual. Air Pollution Control
 District, Los Angeles, Calif. November 1063,.
   5, Fleet. B., Uben> uL V.r and West, T. 3.
 "A Study of Some Mawix Effects  In the Deter-
 mination of Beryllium by Atomic Absorption.
 Spectroscopy In the Nitrous Oxide-Acetylene
 Flame," Talanta, 17. 203,1970.
   6. Mark, L.  i.   Mechanical Engineers'
 Handbook. McGraw-Hill Book Co.. Inc.. New
  Figure 104-8 shows a  sample recording
sheet for velocity traverse data. Use the aver-
ages in the last two columns of figure 104-8
to determine the  average  stack gas  velocity
from equation 104-5
  66  Beryllium  collected—Calculate   the
total weight of beryllium collected by using
equation 104-6
         tVi-V.Ci-V.C.-/.C...eq 104-6
where:
  W i = Total  weight  of  beryllium collected,
         08-
   V i=Total  volume  of hydrochloric  acid
         from step 4 8 2.4, ml.
   Ci=Concentratlon of beryllium found in
         sample, wg/ml.
  V«=Total volume  of  water used  In sam-
         pling  (Implnger  content*  plus  all
         •wash amounts), ml.
  C» = Blank concentration  of beryllium In
         water, fig/ml.
                 V.= Total volume of acetone used In sam-
                       pling (all wash amounts), ml.
                 C.= Blank  concentration of beryllium In
                       acetone, «g/mk
                 67  Total beryllium emission*.— Calculate
               the total amount of beryllium emitted from
               each  stack  per day by^equaUon 104-7. This
               equation Is applicable for continuous opera-
               tions For cyclic operations, use only the time
               per day each stack is In operation. The total
               beryllium emissions from a source will be the
               summation of results from all stacks.

                 „   TT.di.)..,  .4.^86.400 seconds/day
                                 X
                                              eq. 104-7
               where
                  « = Ratp ol emlsJon. (/day.
                  H",-Totai wtigbt of beryllium collected, nf.
                Vu,.,-Tol>U rolame ol gu itmple duck caudluou),
   7. Martin, Robert M.. Construction Detail*
 of Isoklnetlc Source- Sampling Equipment.
 Environmental  Protection Agency.  APTD-
 0681.                      '               !
   B. Methods for Determination of Velocity,'
 Volume,  Dust and Mist Content of Gases,
 Western Precipitation Division of Joy Manu-
 facturing Co., Loe Angeles, Calif.  Bulletin
 WP-50. 1968.
   9. Perkln Elmer Standard Conditions  (Rev.'
 March 1971).         "                   ."
   10. Perry, J. EL, Chemical Engineers' Hand- x
 book.  McGraw-Hill Book Co.  Inc..  Newt
 York, K.T., I860.             .           <">
   11. Bern. Jerome J. Maintenance. Callhra—.
 ttom.  and  Operation of  Isottnetlc Sourcei-
 Sampung Equipment. Environmental  Pro-j
 Uetlon Agency, APTD-4576.             .. i
   12. Shlgeban, R. T., W. F. Todd, and  W. S  •
 Smith, Significance of Errors In Stack  Sam-.
 plug Measurements, Paper presented at the
 annual meeting of the Air Pollution Control
 Association. St. Louis. Mo. June  14-19. 1970.
   13. Smith, W. S. et al. Stack Gas  Sam- .
 pllng Improved and Simplified  with  New .
 Equipment,  APCA Paper No. 67-119, 1967.
   14.  Smith, W. S. R. T.  Shlgehara. and
 W. F. Todd. A Method of Interpreting Stack
 Sampling Data, Paper presented at the 63d
 annual meeting of the Air Pollution Control
 Association. St. Louis. Mo . June  14-19. 1970
   IS. Specifications for  Incinerator Testing
 at Federal Facilities. PHS, NCAPC. 1967.
   16 Standard Method for Sampling Stacks
 for Paniculate  Matter. In:  1971  Book of
 ASTM  standards. Part 23. Philadelphia. 1971.
 ASTM  Designation D-2928-71.
   17. Veonard, J. K.  Elementary Fluid Me-
 chanics  John  Wiley and  Sons,  Inc.. New
 Tork, 1947.
                                                                                       .8.45 am)
                                    FCOMAL RK»IST«. VOL. 31. NO, 64—MIDAY, APRIL 6, 1973

                                                           72

-------
  APPENDIX B
MRI'S FIELD LOG
       73

-------
A.  General Equipment

     Collaborator 1

          RAG console

          RAG sample box

          RAG probe

          Rubber vacuum hose

          No stack temperature

          Using ~ 2-1/2-in. millipore type filter

          Using 0.25 in. ID probe tip

          Heat on filter compartment (~ 150°)

          Heat on probe (~ 50°).

     Collaborator 2

          RAG console

          RAG sample box

          RAG probe

          Stack thermocouple and pot used for stack temperature but not at
            every point

          Using ~ 90 mm millipore type filter with Whatman No.  41 backup

          Using 0.375 in.  ID probe tip

          No heat used on filter compartment

          Heat (~ 30°) used on probe.
                                     74

-------
Collaborator 3

      RAG console

      RAG sample box

      RAG probe (MRI)

      Rubber vacuum hose

      Bimetallic thermometer used for stack temperature—not at each
        sample point

      Using ~ 2-1/2 in. millipore filter

      Using 0.375 in. ID probe tip

      Using heat on filter box (~ 220°)

      Heat on probe (~ 60°).

Collaborator 4

      RAG console

      RAG sample box

      RAG probe

      Poly vacuum hose and Fitot  tubes

      0-0.25 in. inclined manometer used instead of console manometer
        for stack

      Bimetallic thermometer in ports used for stack temperature—not
        at each sampling point

      Using ~ 90 mm millipore type filter with Whatman No.  41 backup

      Using 0.304 in. ID probe tip

      No heat used for filter compartment

      No heat used for probe.
                               75

-------
B.  General Observations

4 December 1973          Run No. 1

          Test start                1125
stop
MRI on- site
Collaborator 1
Collaborator 2
Collaborator 3
Collaborator 4
1545
0800
0830
0830
1000
0830
          Test started late due to lack of power; had to rig extension cords
to interior of building.

          Collaborator 1 changed probe tips at end of first port.  Went to
1/4 in. as could not maintain isokinetic.

          Collaborator 2 lost vacuum and thought filter broke.  Releak checked.
No problem.  Lost ~ 4 rain.  Made up during first port change.

          Weather:  clear, cold, windy.  High in Denver:  37°.
                                    76

-------
5 December 1973          Run No. 2

          Test start                1000

               stop                 1415

          MRI on-site               0800

          Collaborator 1            0930

          Collaborator 2            0830

          Collaborator 3            0900

          Collaborator 4            0900

          Collaborator 1 had checked pump and filter to try to alleviate
problems.  It was drawing 15 Hg through clean filter.

          Collaborator 1, no vacuum during first port.  Releak checked.
Si gel Impinger and filter holder loose.  Fixed during port change.

          Weather:  clear, cold, windy.  High in Denver:  38°.
                                     77

-------
6 December 1973          Run No. 3

          Test start                0930

               stop                 1345

          MRI on-site               0800

          Collaborator 1            0900

          Collaborator 2            0830

          Collaborator 3            0900

          Collaborator 4            0830

          Collaborator 1, changed console pumps between Runs 2 and 3.  New
pump much better.

          There was no problem in maintaining isokinetic and much better vacuum
readings were obtained.  No problems with test. Some pumps were slow in start-
ing which was possibly due to cold.

          Weather:  clear, cold, windy.   High in Denver:  43°.
                                   78

-------
7 December 1973          Run No. 4

          Test start                0910

               stop                 1330

          MRI on-site               0800

          Collaborator 1            0845

          Collaborator 2            0845

          Collaborator 3            0830

          Collaborator 4            0830

          Collaborator 3, first impinger and U-tube broken during third port
change.  Replaced on-site.  Continued test.

          Collaborator 2, probe broken during fourth port change.  There
appeared to be no air leak so finished test.

          Weather:  cool, partly cloudy, moderate wind.  High in Denver:  ~ 50°
                                   79

-------
10 December 1973          Run No. 5

          Test start                0915

               stop                 1330

          MRI on-site               0800

          Collaborator 1            0845

          Collaborator 2            0830

          Collaborator 3            0830

          Collaborator 4            0845

          Collaborator 1's pump vanes stuck and pump inoperative at start of
test.  Started 16 min (2 points) behind others.  Made up during port change
(No. 1).

          Collaborator 1 broke probe during port change No. 2.  Rails dropped
off while extracting probe following port test.  Called office and new liner
brought immediately to site.  Ran without probe heat.  Held all testing about
30 min until probe liner was exchanged.

          Collaborator 1 reported that liquid portion of sample for Run 4 was
lost to accidental spillage.  No confirmation.

          Fixed rails to prevent slippage and probe breakage.

          Weather:  partly cloudy, moderate wind, warm.  High in Denver:  60°,
                                    80

-------
11 December 1973          Run No. 6

          Test start                0910

               stop                 1320

          MRI on-site               0800

          Collaborator 1            0830

          Collaborator 2            0830

          Collaborator 3            0830

          Collaborator 4            0830

          No major problems.  Collaborator 1 again had pump starting problems
but cleared up.

          Collaborator 2 thought to have leak but fixed prior to testing.

          Weather:  partly cloudy, moderate wind, cool.  High in Denver:  56°.
                                   81

-------
12 December 1973          Run No. 7

          Test start                1215

               stop                 1635

          MRI on-site               0810

          Collaborator 1            0830

          Collaborator 2            0900 (late due to high winds closing road)

          Collaborator 3            0800

          Collaborator 4            0830

          Weather bad in a.m.  Winds gusting to ~ 60-75 mph.  Conditions con-
sidered unsafe for personnel on stack.  Decision made to review conditions
at 1130 as to go or no-go for 12 December.  Situation improved by 1130.  Winds
to 40-50 mph but slackening.  Considered by all okay for sampling.  Some heavy
gusts but general abatement during sampling period.  No worse than usual by
1530.

          Collaborator 4 had trouble with pitot lines.  At first thought to
be result of turbulence in stack from high west and northwest winds but
did not improve after port change.  Other three teams pitot readings okay.
Collaborator 4 shutdown and replaced downstream line with Tygon.  Time made
up during port change.  Readings okay.
          Weather:  clear, cool, very windy.  High in Denver:  50
                                                                 o
                                   82

-------
13 December 1973          Run No.  8




          Test start                0915




               stop                 1325




          MRI on-site               0820




          Collaborator 1            0845




          Collaborator 2            0830




          Collaborator 3            0845




          Collaborator 4            0845




          No problems.




          Weather:  warm, partly cloudy, little wind.  High in Denver:  53°







14 December 1973          Run No.  9




          Test start                0920




               stop                 1330




          MRI on-site               0820




          Collaborator 1            0830




          Collaborator 2            0845




          Collaborator 3            0830




          Collaborator 4            0900




          No problems




          Weather:  clear, cool, moderate wind.  High in Denver:  47°.
                                    83

-------
 17 December  1973          Run No. 10

          Test start                0900

               stop                 1335

          MRI on-site               0815

          Collaborator 1            0830

          Collaborator 2            0830

          Collaborator 3            0830

          Collaborator 4            0830

          Collaborator 2's filter got wet during leak check, which created
 too great a vacuum such that isokinetic sampling could not be achieved.
 Filter replaced and three points picked up at conclusion of test.

          No other problems.

          Weather:  cloudy, warm, light wind.  High in Denver:  63".


 18 December  1973           No Test

          No test

          MRI on-site               0800

          Collaborator 1            0845

          Collaborator 2            0845

          Collaborator 3            0800

          Collaborator 4            0830

          Began snowing ~ 0715.  Decision made to start test as only snow,
no wind and 19 December to be worse.  Set up and ready to go by 1010 when
wind came up.  Scaffold became icy and visibility poor.  Decided not to
begin test.  Conditions worsened during lowering of equipment, wind increas-
 ing.
                                    84

-------
19 December 1973          Run No. 11

          Test start                0930

               stop                 1345

          MRI on-site               0830

          Collaborator 1            0900

          Collaborator 2            0900

          Collaborator 3            0830

          Collaborator 4            0845

          Collaborator 4 experienced rapid rise in vacuum during first
port.  Thought to be wet filter.  Filter replaced—not wet.  Most likely
source—water freezing in S-g nozzle or frozen check valve.  Lost 3 min.
Made up at end of port.

          No further problems.

          Weather:  cold, partly cloudy, no wind.  High in Denver:  24°.


20 December 1973          Run No. 12

          Test start                0915

               stop                 1330

          MRI on-site               0830

          Collaborator 1            0830

          Collaborator 2            0845

          Collaborator 3            0830

          Collaborator 4            0845

          Collaborator 2 changed probe tips following first port.  Could not
maintain isokinetic.

          Weather:  cold, clear, little wind.  High in Denver:  44°.
                                     85

-------
21 December 1973          Run No. 13

          Test start                0905

               stop                 1320

          MRI on-site               0830

          Collaborator 1            0845

          Collaborator 2            0830

          Collaborator 3            0815

          Collaborator 4            0830

          Collaborator 1 had large vacuum on first port.  Stopped and
reversed filter.  Okay.  Made up time at port change.

          Weather:  cool, partly cloudy, moderate wind.  High in Denver: ~ 50°
                                    86

-------
                                    TECHNICAL KCPCWT DATA
                               aie rraJ li.urn iua:t mi n.f /•< • i rtt iV/
 i R«.fOr
-------
                                                       INSTRUCTIONS

1.   REPORT NUMBER
     ln« it tlic I I'A rcpuri number a» it appear* on the covet of ihc publication,

2.   LEAVE BLANK

3.   RECIPIENTS ACCESSION NUMBER
     Res.-rveJ lor  i,«. by e.i.h repast i
4.   TITLE AND SUBTITLE
     lull- »tun.!.! i.i.!k Hi- 1 1.- iris ami hru-ilv the suited covcrarc of the rcpoii. and be di'i^'n.ia n 10 nuin ink1. Alien J ripoil ii prepared in more thJn one volume, repel! (he prinurv tilk1. a<.;J volume
     number jn J iru-iuile % itnirle lor ilie jpeciPc title

5.   REPORT DATE
     l-ach rcpori shilluriy 3 date indicating at lea*t month and yea:.  Indicjte the basis on which it was selected (eg., dale of issue. Jttc of
     appro\tl. J.ir nf r'fi r**c«i"i fit  I

6.   PER.'ORMIMG ORGANIZATION CODE
     Lea\c
7.   AUTHORISI
     Cue named) in conventional older (John R DM. J Robert Dee. etc ).  Lisl author's affiliation if H differs from the performing
     zatton.

8.   PERFORMING ORGANIZATION REPORT NUMBER
     Insert it pcnurming ur^jni/jiion uisnes to assign itm number

9.   PERFOOMIVG ORGANIZATION NAME ANO ADDRESS
     Give njnie. sircct. city. iUie. and ZIP code. List no more tlun t»o levels of an organizational .hircarchy.

10.  PROGRAM ELEMENT NUMBER
     L'sc the program element number under which tlie report »a< prepared  Subordinate number*, may be included in parentheses.

11.  CONTRACT/GRANT NUMBER
     Insert lontrju or crjni numbet under *hi«.h report »J* prepared

12.  SPONSORING AGENCV  NAME AND ADDRESS
     Include /.li' kodc
     Indiute ink-urn fmjl. etc . jnj u jppluaolc, Jjies covered.

14   SPONSORING AGENCY CODE
     Le.ive bbnk.

15.  SUPPLEMENTARY NOTES
     Lnlei inU.tin.HMti IH-I niv kidcJ L ho A IK re bul uieful. such a\  Prepared in cooperation with. Trjnsljtion of. Prcwntcd at conference of,
     To be published in. Supcr^J-.-s. bi-ppienientt, cti

16.  ABSTRACT
           j hnef {ItiO \\oriit or U^t factujl summary of llie most sicnificant information contained in the report.  If the report contains a
           jiit tnlilioiirjphy or literjtuie %urvey. mention it !ieie

17.  KEY WORDS AND  DOCUMENT ANALYSIS
     la) Dl SCKII'IOCS • Select irom the ! hoxjuriK c>f I ngmeerini1 and Scientific Terms the proper authorized terms that identify ilie nujor
     conicpt ol ilie rew jrih and arc Miliiucnil) ^potilic  jnd piiikiM' to tc open-
     ended terms written in di'^nplur luim tor lho»e subjects tor uhich no descriptor c\ists.
     (c) COS \1 1  I IfLIH.KOL'l' - I it-Id jnd croup assienmentt ate to be uken from the I96S COS \1I Sul-jecl Catci-orN  Li«l   Snue the ma-
     jority ol doiunii-nlk no mullii!i^i;'lui •!> m njliire. ttii. I'nm.vy I ivid/diuup jSMirniiK'nKsi u>ll tv (jvulii. discipline, jrej ol IHIM in
     endeavor, or ivpe 01 nhvMut ol-je^i  Ilie jpplK^liuiKst *ill be crosvreicr I icId/Otoup asMgnments tlijl will tollosv
     the
18.  DISTniflUTIC.'J STATEMENT
     IK'noii  u le is.il 3vji!j| 10
     the pul'lii. uitli jJ>'iv>s .inj |>riie.
19. & 20 SECURITY CLASSIFICATION
     DO NUI M.hiMik-J :c()i.al Infonnjtior. St-nue or Hie Government Punting Office, if known.
 roim ;??o i n

-------