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
------- |