EPA-650/4,75-025

June 1975
Environmental Monitoring Series
                   METHOD FOR OBTAINII
        REPLICATE PARTICULATE SAMPLES
              FROM  STATIONARY SOURCES
                                U.S. Environmental Protection Agency
                                 Office of Research and Development
                               National Environmental Research Center
                                 Research Triangle Park, N. C. 27711

-------
                                 EPA-650/4-75-025
      METHOD  FOR  OBTAINING
REPLICATE PARTICULATE SAMPLES
   FROM  STATIONARY SOURCES
                     by

         William J. Mitchell and M. Rodney Midgett

   Quality Assurance and Environmental Monitoring Laboratory
         National Environmental Research Center
       Research Triangle Park, North Carolina 27711
                ROAP No. 26AAG
             Program Element No. 1HA327
       U.S. ENVIRONMENTAL PROTECTION AGENCY
           Office of Research and Development
         National Environmental Research Center
       Research Triangle Park, North Carolina 27711

                   June 1975

-------
                    RESEARCH REPORTING SERIES
                                i
Research reports of the Office of Research and Development, Environ-
mental Protection Agency, have been grouped into five series.  The
five broad categories were established to facilitate further
development and application of environmental technology.  Elimina-
tion of traditional grouping was consciously planned to foster
technology transfer and a maximum interface in related fields.  The
seven series are:

                  1.  Environmental Health Effects Research
                  2.  Environmental Protection Technology
                  3.  Ecological Research
                  4.  Environmental Monitoring
                  5.  Socioeconomic Environmental Studies
                  6.  Scientific and Technical Assessment Reports
                  9.  Miscellaneous
This report has been assigned to the ENVIRONMENTAL MONITORING series.
This series describes research conducted to develop new or improved
methods and instrumentation for the identification and quantification
of environmental pollutants at the lowest conceivably significant
concentrations.  It also included studies to determine the ambient
concentrations of pollutants in the environment and/or the variance
of pollutants as a function of time or meteorological factors.

                        EPA REVIEW NOTICE
This report has been reviewed by the Office of Research and Development,
U.S. Environmental Protection Agency, and approved for publication.
Approval does not signify that the contents necessarily reflect the
views and policies of the Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement
or recommendation for use.

                           AVAILABILITY
This report is available to the public, for a nominal cost, through
the National Technical Information Service, 5285 Port Royal Road,
Springfield, Virginia  22161.
                  Publication No.  EPA-650/4-75-025
                                 »
                                ii

-------
                        TABLE OF CONTENTS.
                                                           PAGE

LIST OF FIGURES                                               V
LIST OF TABLES                                                v
ACKNOWLEDGMENTS                                              vl
INTRODUCTION                                                  1
EXPERIMENTAL                                                  5
   EPA Method 5 Particulate Train                             5
   Four-train Sampling Arrangement                            7
   Equipment Calibration                                     10
   Source Characteristics                                    1]
   Sampling                                                  12
   Sample Recovery                                           14
RESULTS                                                      15
DISCUSSION OF RESULTS                                        19
      Particulate Results                                    19
      Percent Stack Moisture Results                         21
CONCLUSIONS AND RECOMMENDATIONS                              23
REFERENCES                                                   25
TECHNICAL REPORT DATA AND ABSTRACT                           27
                              iii

-------
                              LIST OF FIGURES

                                                                Page
1.  An EPA Method 5 Particulate Train                             6
2.  Single-pi tot Sampling Arrangement                             8
3.  Double-pi tot Sampling Arrangement                             9
4.  Within-Run Standard Deviation Versus Appropriate             17
    Run Mean
                              LIST OF TABLES

                                                                Page
1.  Within-Run Variations in the Particulate Loading             16
2.  Within-Run Variations in the Moisture Determination          18
                                   iv

-------
                      ACKNOWLEDGMENTS

      The authors wish to thank the following for their assistance
in this study:   Scott Environmental Technology,  Inc., Plumsteadville,
Pa.; Mr. Jim Shelby, Mr.  Frank Shapiro, and their staff at the Dade
County Municipal Incinerator, Miami Beach,  Fla.; Mr.  Jim Rivard
at the Allen King Power Plant, St.  Paul, Minn.;  and Mr. Ray Mobley
of the Environmental Protection Agency, Durham,  N.C.

-------
METHOD  FOR  OBTAINING  REPLICATE  PARTICULATE
         SAMPLES  FROM  STATIONARY  SOURCES
                           INTRODUCTION
       Spatial and temporal variations i_n the velocity profile (the
   distribution of the flow across the stack) and in the parti oil ate
   profile (the distribution of the particulate  concentration across
   the stack)  are common  at stationary sources.  Since a shift in the
   velocity profile will  not always produce an identical shift in the
   particulate profile, the effects of these spatial and temporal
   variations  on the sampling results must be minimized when sampling
   for particulates.  To  minimize the occurrence of temporal variations
   when a stationary source is being sampled, the source is generally
   required to operate at maximum rated capacity.  Traversing while
   sampling the stack—a  technique in which the  stack is divided into
   equal concentric areas and each area is then  sampled for an equal
   period of time—is the usual manner by which  spatial variations are
   handled in  source testing.  In addition, any  residual spatial and
   temporal variations that might still affect the results are supposedly
   handled by  requiring that the average of a minimum number of sequential
   sampling runs constitute a source test result.  For example, the
   Environmental Protection Agency (EPA) defines a particulate test
   result as the average  of three sequential particulate sampling
   runs.
       Traditionally the precision and accuracy of test methods is
   established through a  collaborative test (round-robin test).  The
                                 -1-

-------
collaborative test is designed so that the participants (collaborators)
each make one or more measurements on identical samples using the
same test method.  Then from a statistical analysis of the results
an estimate is made of the precision associated with the test method.
If the collaborators also use the test method to measure standard
reference samples of known concentration, then an estimate can be made
of the accuracy of the test method.
     When we initiated our program to standardize the EPA promulgated
stationary source test method for particulates,  our collaborative test
design employed four sampling ports, which were located on the same
stack, and four source sampling teams (collaborators).  In the three
collaborative tests accomplished thus far, each team sampled simultane-
ously for particulates, but at any one time only one team was sampling
at each port.  Each team achieved a complete run by sampling until a
complete traverse of the stack had been accomplished.
     These collaborative tests of the EPA particulate method (designated
as Method 5)  were carried out at three different sources - a Portland
             23                              4
cement plant,  a municipal incinerator,  and a coal-fired power plant.
In each test, variations of from 37% to 58% were observed between collabo-
rator results.
     What' caused these wide differences?  Possibly it was spatial and
temporal variations in the source's particulate loading, but unfor-
tunately the test design contained no means for evaluating the
contribution from spatial and temporal variations.  This is because  it
was assumed that traversing while sampling would minimize these
                                -2-

-------
sources of variation.  Other possible causes for the large differences
between collaborators are:  undetected equipment malfunctions,
errors in operating the sampling train and recording the sampling data,
differences in equipment and between-collaborator variations
in their particulate recovery and analysis techniques.
     If spatial and temporal changes caused these variations, then
we had characterized the three sources but not the performance of
Method 5.  If the variations were actually attributable to the other
causes mentioned above, then we had to conclude that Method 5 as
written was not a rugged or reliable test method.  Thus, a means was
needed to determine the precision that could be obtained using the EPA
particulate sampling train in the absence of spatial and temporal
variations in the particulate profile.  If this could be accomplished,
then the variations measured between the trains would reflect only
collaborator differences in the assembly and operation of the sampling
train, differences due to properties inherent in the trains themselves;
and between-collaborator differences in each collaborator's sample
recovery and analysis procedure.
     This report describes two  four-train sampling arrangements
that were used to estimate the precision capabilities of the
Method 5 sampling train itself; that is, the agreement which could
be expected between two identical trains that had simultaneously
sampled the same particulate concentration in the absence of spatial
and temporal variations.  Three field tests of these sampling
arrangements were done.
                               -3-

-------
     From these field studies an estimate was also made of the
reliability of using condensation in the Greenburg-Smith impinger
to determine the moisture content of stack gases.
                              -4-

-------
                           EXPERIMENTAL
                                  !
A.  EPA Method 5 Parti oil ate  Train
    Figure 1  presents a schematic diagram of the  EPA Method 5
trains used.   The design and  construction specifications  for the
sampling train have been described by  Martin,   and  a description of
the operating principles of the train  has been  presented  by Rom.6
In its simplest form, the train can  be considered as composed of
five major components:
         1.  A nozzle,  a probe and a filter.
         2.  An "s" type pi tot tube  to measure  the  stack  gas velocity.
         3.  A condenser (generally  four Greenburg-Smith  impingers
             connected in series).
         4.  An orifice, a vacuum pump and a dry  gas meter.
         5.  A nomograph to relate stack gas velocity  to  the pressure
             drop across the  orifice.
         The train itself is  operated  as follows:   the pi tot tube
reading is used in conjunction with  the orifice reading and the
nomograph to obtain an isokinetic sampling rate in  the train in
order to sample for the particulate  at the same velocity  that it is
moving in the stack.  At the  end of  the sampling  run,  the particulate
that has been collected in the nozzle, probe and  on the filter is
recovered and weighed.   The particulate concentration  in  the stack,
                                -5-

-------
                                      HEATED AREA
                                                                     THERMOMETER
i
en
                  PROBE
            REVERSE-TYPE
              PITOTTUBE
PITOT MANOMETER

         ORIFICE
       MANOMETER
                       THERMOMETERS
IMPINGERS              ICE BATH
         BY-PASS VALVE

           
-------
which is usually referred to as the particulate loading,  is then
obtained by dividing the weight of particulate collected  in the train
by the total volume of gas measured at the dry gas meter  after
correcting this gas volume to the reference conditions of 21°C and
standard atmosphere pressure (760mm Hg).
B.  Four-train Sampling Arrangements
         Two sampling probe/pi tot tube arrangements were  developed.
In the first sampling arrangement, the four sampling nozzles were symme-
trically.located about a type s pitot tube by placing one nozzle at
each corner of a 6-cm square and placing the pitot tube at the center
of the square (Figure 2).  In the second sampling arrangement, each
nozzle was located at a corner of a 3.5-cm square, and the two pitot
tubes used were placed on either side of the square (Figure 3).  This
latter sampling arrangement was developed to see if decreasing the
distance between the sampling nozzles would improve the within-run
standard deviation.
         Except for the two different nozzle/pi tot tube arrangements
mentioned above, the equipment used in all three studies  was identical
in design and construction.  The nozzles were constructed from
9.5 mm (O.D.) by 7.2 mm (I.D.) stainless steel tubing.  The 150-cm
glass probes, which were constructed from 1.6-cm (O.D.) by 1.3-cm
(I.D.) borosilicate glass tubing, were identically wound  with 762
centimeters of 25 gauge nichrome wire.  Each glass probe  was con-
tained in a 25.4 mm (O.D.) by 22.1 mm (I.D.) stainless steel tube.
                                -7-

-------
                            -25cm-
9 cm
                                   NOZZLE
            63cm
                       1.5cm
                                                NOZZLE
              TRAIN 1
              ]  ~("~l'^ -
                  TRAINZ
-3cm-
  6cm
-3 cm-
                           "S" TYPE PITOT TUBE
                                 A
      -j -)	  TRAIN 4
             3-Q
                 TRAIN 3
                  -6cm-
d
                                                           1
                                                          2 cm

                                                           1
 Figure 2. Single-pilot sampling arrangement-  (top) side view, (bottom) upstream view
                               -8-

-------
                            -25cm-
  9cm

1
6.5
cm
v 	
r
T
1.5cm
i









                                NOZZLE
                                   (
                            NOZZLE
  3.5cm
         • 1.7 cm-
           -00-  -
              TYPE PITOT TUBE
    *-1.7 cm-»i
TRAIN
             -3.5cm-
                             TRAIN
                                               1.0cm
                                          1.0cm
                                           i
              CT  s"
           TYPE PITOT TUBE
Figurn 3 Double pilot sampling arrangement  (top) side view, (bottom) upstream view.
                             -9-

-------
The four filter holders used in each sampling run were contained
in the same heated, thermostatically controlled box.  The four
sets of impingers used (four impingers per probe) were also contained
in one box.  This combining of components from the four trains was a
necessary compromise because of physical limitations of space, size
and weight.
     During each sampling run, the relative spacing between the
nozzles and the pitot tube(s) was held constant using stainless
steel clamps mounted on the probes.  The four probe liners were
maintained at the same temperature by connecting each heating wire
to the same electrical outlet strip, whose voltage was controlled
with a Variac.  To facilitate movement of the probes into and out
of the stack, all the sampling equipment, with the exception of
the meter boxes, was mounted on one lab cart.
C.  Equipment Calibration
     The dimensionless pitot tube calibration coefficient (C ) of
each type s pitot tube was determined over the velocity range
6.4 to 26 m/sec using a standard pitot tube (C  = 0.99) as the
reference standard.  A check for interference of the nozzles on the
pitot tube velocity measurement was made in a wind tunnel by comparing
the C  obtained in the absence of the four nozzles with the C
     P                                                       P
obtained when the nozzles occupied the positions they would occupy
during sampling.  Since the C  determined in the presence of the
nozzles (when they were not sampling) was within one percent of the
                               -10-

-------
C  determined in the absence of the nozzles, there is no reason to
suspect that the nozzles interfered with the stack gas velocity
measurement.  In addition, a check for nozzle interference on the
pi tot tube reading was also made at each site by comparing the pi tot
pressure differential (Ap) immediately before sampling began with the
Ap measured immediately after sampling was initiated.  No sudden
change in the Ap was observed in any of the three tests.
     Each meter box was calibrated in the laboratory prior to
each test, but the calibration was also checked at the site by
setting the orifice manometer to the AH@ (calibration factor) of
the meter box and measuring the flow rate through the dry gas meter.
All meter boxes were still in calibration when checked on-site.
     The Variac voltage required to maintain the sampled gases
in the probe at 120°C was determined on-site by withdrawing gas
from the stack through one of the probes and determining the
temperature of the gas at the probe exit as a function of the
voltage.  All probes were identically wound and of the same length;
so it was assumed that all probes would have the same temperature
if operated at the same voltage.
D.  Source Characteristics
     The first study was done at a coal-fired power plant equipped
with an electrostatic precipitator.  When operating at its maximum
output  of 540 megawatts per hour, the plant used 240 tons
                               -11-

-------
of coal per hour.  Sampling was done in a horizontal duct 6.9 meters
high by 3.6 meters wide.  Stack temperature at the sampling point
was 150°C.
     The second and third studies were done at the same municipal
incinerator at different times.  The incinerator, which was equipped
with an electrostatic precipitator, burns 300 tons of refuse per
24-hour day.  The sampling was done about 1.5 stack diameters below
the top of the vertical stack.  Stack dimensions were 27.5 meters
high and 3.0 meters in diameter, and the stack temperature at the
sampling port was 215°C.
E.  Sampling
     The sampling procedure was similar to that specified in the
Federal Register  (pp. 28888-28890) except that:  1) fixed-point
sampling rather than multiple-point sampling (traversing) was used;
and 2) for some sampling runs, the sampling time was less than the 2
hours specified in the Federal Register.  (The sampling time was
varied to see if the within-run standard deviation was sensitive to the
total weight of participate collected.)
     At each site, a sampling point was selected at which the
velocity profile varied by less than 2 percent of the absolute
flow across the area that would be occupied by the four nozzles
and the pi tot tube(s).  During each test, all sampling runs were
done at the selected sampling point.  A total of 16 sampling runs
were accomplished in the three studies.  The first 4 sampling runs
                               -12-

-------
were done at the power plant, and the other 12 were done at the
municipal incinerator.
     As an aid to following the ensuing discussion, a brief
tabulation of the three tests is given below:

  Test     Run            Site            Probe/Pi tot Tube Arrangement
    1      1-4     Power Plant              4 probes, 1  pi tot tube
    2      5-10    Municipal Incinerator    4 probes, 1  pi tot tube
    3     11-14    Municipal Incinerator    4 probes, 2  pi tot tubes
          15-16                             4 probes, 1  pi tot tube
     In the first two tests, the pi tot tube (Figure 2) was connected
to one meter box.  The operator of this meter box determined the
velocity head (AP) at 5-minute intervals, and then each  meter box
operator used this Ap and his own nomograph to obtain an isokinetic
sampling rate in his train.
     In the first four runs of Test 3, the sampling train/pi tot tube
arrangement in Figure 3 was used.  One of the pi tot tubes was
attached to the meter box of train 1, and the other pitot tube was
attached to the meter box of train 3.  The operator of train 1
determined the Ap at 5-minute intervals, and then he and the operator
of train 2 used this Ap to obtain an isokinetic sampling rate in
their respective trains.  At the same time the operator  of train 3
also determined the Ap using the other pitot tube.   Then both he
and the operator of train 4 used this AP to obtain an isokinetic
sampling rate in their respective trains.  However, the  sampling

                              -13-

-------
arrangement in Figure 2 was used in the last two runs of this test
to see if decreasing the spacing between the nozzles would decrease
the within-run standard deviation.  In all 16 runs all the trains
were able to operate within 10 percent of the isokinetic sampling
rate—the allowable range for a Method 5 determination.

F.  Sample Recovery
      The particulate material collected in each train was recovered
at the end of each sampling run.  The material caught in the nozzle,
probe, and filter holder was removed by washing these train compo-
nents with acetone while scrubbing them with a nylon brush.  The
fiberglass filters, which were tared prior to the test, were removed
from the filter holder with a pair of tweezers, placed in a culture
dish and returned to the laboratory for a weight-gain determination.
      At the end of each test the volume of water collected in the
first three impingers of each train was measured in a graduated
cylinder.  (Impinger set integrity was maintained in all runs at
one test site).  The amount of water collected in the fourth impinger,
which contained a known weight of silica gel, was obtained by
weighing the previously tared silica gel.
                              -14-

-------
                               RESULTS

     Table I presents  the particulate  results  obtained  in each test
and also presents the  within-run  percent coefficients of variation
(%C.V.).  The percent  coefficient of variation, which is calculated
using Equation 1, is a statistical  technique that  expresses  the  standard
deviation as a percentage of the  mean.  This technique  was used  to facili:
tate comparing the sampling results for different  runs, because  the mean
particulate loading varied from run to run.
            y. r v   -   within-run standard deviation         ,nn       m
            * L-w-  "   run mean particulate concentration x  IUU       u;

Figure 4 is a plot of  each within-run  standard deviation versus  the
mean particulate concentration for that run.
     Table II compares the individual  train percent moisture results
and presents estimates of the within-run  variation based on  a statistical
analysis of these results.
                                 -15-

-------
             Table I.  Within-Run Variations in the Particulate Loading (mg/std  m  , dry basis)
Run
1
2
3
4
5
6
7
8
9
10
na
12a
13a
14a
15
16
Run
Time, min.
90
30
30
30
120
120
150
120
55
55
120
90
120
90
120
90
Train
1
205
190
207
155
60.4
57.1
62.5
54.2
51.0
56.3
140
113
114
126
153
103
Train
2
202
196
240
150
61.9
62.2
61.3
51.8
50.5
59.9
143
107
234b
123
141
106
Train
3
204
222
222
188
63.2
64.1
58.2
49.6
50.5
62.4
131
125
200b
134
161
104
Train
4
221
185
199
141
64.6
62.5
56.9
45.3
46.1
51.9
133
128
114
144
139
103
Mean
208
198
217
159
62.5
61.5
59.7
50.2
49.5
57.6
137
118
166
132
148
104
% C.V.
4.2
8.3
8.3
13.0
2.8
4.9
4.4
7.5
4.6
7.9
4.4
8.6
37.0
7.0
6.7
1.3
 Double-pi tot sampling arrangement used in this run.
'"Train  nozzle brushed the wall  of the stack as it was being inserted into the stack.

-------
    IS
    10
§    5
I
                                                                      A DOUBLE-PITOT ARRANGEMENT
                                                                      • SINGLE-PHOT ARRANGEMENT
                                                  I
               20
40
60
   100       120
RUNMEAN.ma'stdrnS
140
                                                                           160
                                                            180
                                                            200
220
                       Figure 4. Within-run standard deviation versus appropriate run mean.

-------
                Table  II.  Within-Run Variations  in  the  Moisture Determination

Run                Train 1     Train 2     Train 3      Train 4-                Mean      % C. V.
 1                    6.9         7.0         8.9         7.9                    7.7         12
 2                    7.9        10.0        14.9         8.9                   10.4         30
 3                    9.3        10.6        10.4         8.4                    9.7         10
 4                    9.9        11.3         7.4        11.2                   10.0         18
 5                  14.9        16.2        16.0         15.0                  15.5        4.2
 6                  15.2        16.2        15.8         15.5                  15.6        2.8
 7                  15.1        16.0        15.9         15.5                  15.6        2.6
 8                  14.2        15.5        13.6         13.5                  14.2        6.3
 9                  13.8        14.3        14.7         14.4                  14.3        2.5
10                  16.1        17.4        16.8         16.4                  16.7        3.4
11                  10.6        11.0         9.2          10.8                  10.4        7.7
12                  12.4        11.3        13.2          12.9                  12.5        6.7
13                  12.1        13.1        12.7          13.0                  12.7        3.6
14                  12.0        12.2        12.2          12.5                  12.2        1.9
15                  14.5        15.6        15.2          14.9                  15.1        3.1
16                  10.8        10.8        11.0          11.1                  10.9        1.5
Train 1
6.9
7.9
9.3
9.9
14.9
15.2
15.1
14.2
13.8
16.1
10.6
12.4
12.1
12.0
14.5
10.8
Train 2
7.0
10.0
10.6
11.3
16.2
16.2
16.0
15.5
14.3
17.4
11.0
11.3
13.1
12.2
15.6
10.8
Train 3
8.9
14.9
10.4
7.4
16.0
15.8
15.9
13.6
14.7
16.8
9.2
13.2
12.7
12.2
15.2
11.0
Train 4-
7.9
8.9
8.4
11.2
15.0
15.5
15.5
13.5
14.4
16.4
10.8
12.9
13.0
12.5
14.9
11.1

-------
                      DISCUSSION OF  RESULTS

Parti oil ate Results
     Our examination of the within-run  percent coefficients of
variation presented in Table I  showed that the two-pi tot sampling
arrangement (runs 11, 12 and 14)  and the  single-pitot sampling arrange-
ment (runs 1 through 10 and runs  15 and 16)  gave comparable results.
On this basis, a linear regression  analysis  was done to compare
the standard deviations from each run  (except run 13) with the
corresponding mean particulate loading  for that run.  The results
of this linear regression analysis, which are plotted in Figure 4,
determined that a linear relationship existed between the standard
deviation and the source particulate loading measured for that run,
but the reasons for this relationship are unknown.  (The data for
run 13 were not used because as the probes were being inserted into
the stack, the nozzles of trains  2  and  3  brushed the stack wall.)
     A similar statistical  analysis that  compared the standard
deviation for each sampling run with the  sampling time for that
run, showed that the variation in the within-run standard deviation
was not affected by the sampling  time.  Consequently, there is no
reason to suspect that the total  weight of particulate collected
or the total volume of gas  sampled  affected  the within-run standard
deviation.
     From the above discussion, it  can  be seen that the sampling
results have been interpreted  on  the assumption that during each
                                  -19-

-------
sampling run, each of the four trains sampled the same average
particulate concentration.  Thus, the four samples obtained in a
sampling run are considered to be replicates.  Because it is likely
that temporal variations in the particulate profile occurred between
sampling runs, samples from different runs must not be considered as
replicates.  Therefore, when we determine the standard deviation
observed in a sampling run, we are estimating the smallest variation
that might be expected between two identical Method 5 sampling trains
that had simultaneously sampled the same stack and had each sampled
the same particulate concentration in such a manner that spatial
and temporal variations in the source did not affect their sampling
results.  Because we cannot assume that the samples taken by the
same train in different runs are replicates, no estimate of the
within-train variation has been attempted.
     An examination of the particulate results in Table I shows
that the between-train differences observed were small when compared
to the differences observed between participants in the three collabora-
              234
tive studies.  ' '    However, this should be expected because:
1) as discussed in the Experimental Section, certain components of
the four trains were combined, and, to this extent, the trains were
not truly independent of each other in their operation; 2) the equip-
ment used here was of a reliable design from a reputable manufacturer,
but our observations from the collaborative test program indicate that
some of the equipment commercially available for particulate sampling
simply does not function as it was designed to function; 3) our use of
                             -20-

-------
fixed-point sampling spared us the trouble of having to change
sampling ports during a sampling run   (This  minimized the chances
of brushing the stack wall  while inserting and removing the probes--
the effect of which can be  seen in run 13 of  Table I--and also
minimized the chance that a leak would develop in the train during
a port change ); and 4) all sampling,  sample  recovery, and sample
analyses were carried out by the same  laboratory.  Thus, even though
different console operators were used, and sample recovery from the
different trains was performed by different individuals, members of
this laboratory executed these manipulations  in a similar manner—
a situation that would not  normally be found  among independent
collaborators in different  laboratories.
     Thus, our failure to observe in this study the large within-run
differences detected in the collaborative tests of Method 5 indicates
that in the collaborative tests either:  1) traversing while sampling
did not adequately handle spatial and  temporal variations; 2) the method
was poorly written and lacking in sufficient  detail on how various
procedural manipulations were to be carried out such that each collabora-
tor was using his own version of the method;  or 3) undetected equipment
malfunction and poor performance on the part of the collaborators were
common occurrences in all the collaborative tests.  Perhaps all three
of these made significant contributions to the overall differences.
B.  Percent Stack Moisture Results
     An examination of the within-run, percent coefficients of
variation for the determination of the percent moisture in the stack
gas (Table II) shows that the variations  ranged from a low of
                             -21-

-------
1.5 percent to a high of 30 percent.   For the following reasons
these within-run variations are attributed to volumetric measuring
errors and not to significant differences in impinger collection
efficiency:  1) a visual inspection of the moisture results  showed
that the high and low values in any one test were randomly distributed
between the four sets of impingers; 2) a linear regression analysis
detected no correlation between the within-run standard deviation
and the mean volume of water collected in the impingers during that
run; 3) an Analysis of Variance detected no relationship between
the individual percent moisture results and the location of  the
impingers in the impinger box; and 4) an examination of the  field data
sheets detected no relationship between the individual percent
moisture results and the temperature of the gas as it left the last
impinger.
                               -22-

-------
             CONCLUSIONS AND RECOMMENDATIONS

     From the participate sampling results reported here,  it  is
concluded that  either of the  four-train sampling arrangements  will
be a useful  means  for estimating the best agreement between sampling
trains that  could  be obtained from the test methods used  to characterize
stationary source  emissions.  In addition, the technique  might be a
non-controversial  means by which to compare the performance of different
types of trains -  for example, the ASTM, ASME, and EPA particulate
trains.
     Since the  results ootained in this study conflict with those
obtained from three classically designed collaborative tests  of
          234
Method b,  ' '   additional collaborative testing of Method 5 seems
justified.  However, these additional tests and similar collaborative
tests on other  stationary source test methods should be designed
so that an estimate can be made of the contribution that  spatial
and temporal variations make  to the total variation observed.   In
other words, the testing should be designed not only to evaluate
the test method, but also to  determine if traversing adequately
compensates  for spatial and temporal variations in the pollutant
profile.
     The design of these collaborative tests could be similar to
that used in the previous collaborative tests with the following
exceptions:   1) each train should be leak-checked before  sampling  is
                                 -23-

-------
initiated, after each port change, and at the end of the sampling run
(in previous collaborative tests the trains were only leak-checked
before the run started - a standard source testing procedure);  2) some
means should be devised to ensure that the nozzles do not touch the
stack wall when the probes are being inserted into and removed  from the
stack; 3) the trains should traverse in pairs and not individually;
and 4) the sampling method being tested should specify in detail
many of the procedural details that are now left to the judgment
of the sampling train operator.
     The use of paired trains, combined with the leak-check procedure
described in the above paragraph, should allow instances of train
malfunction to be easily identified.  If the traversing procedure
adequately compensated for spatial and temporal variations in the
particulate profile, then the mean particulate loadings for a set
of paired trains should not be significantly different from the
means obtained by the other set(s) of paired trains.  If the means
were not significantly different, then the differences between  the
trains would yield an estimate of the precision of the stationary
source test method.  If the means were significantly different, but
both trains in a set agreed closely, then it should be possible to
estimate the contribution that spatial and temporal variations  made
to the results.
                                   -24-

-------
                           REFERENCES

1.  Environmental  Protection Agency.    Standards  of Performance  for
    New Stationary Sources.    Federal  Register.   36_:   24876-24890,
    December 23, 1971.
2.  Hamil, H.  F.,  and D.  Camann.   Collaborative Study of the Method
    for the Determination of Particulate Emissions  from Stationary
    Sources (Portland Cement Plant).   Environmental Protection
    Agency Technical  Report No.  EPA-650/4-74-029. May 1974.
    (Available from National Technical  Information  Service,
    5285 Port Royal Road, Springfield,  Va.   22161.)
3.  Hamil, H.  F. and R.  E.  Thomas.  Collaborative Study of the
    Method for Determination of Particulate Emissions from
    Stationary Sources  (Municipal  Incinerators).  Environmental
    Protection Agency Technical  Report No.  EPA-650/4-74-022.
    July 1974.  (Available from National Technical  Information
    Service, 5285  Port Royal Road, Springfield, Va.  22161.)
4.  Hamil, H.  F. and R.  E.  Thomas.  Collaborative Study of the
    Method for the Determination of Particulate Emissions from
    Stationary Sources  (Fossil  Fuel-Fired Steam Generators).
    Environmental  Protection Agency Technical  Report
    No. EPA-650/4-74-021.  June 1974.   (Available from National
    Technical  Information Service, 5285 Port Royal  Road,
    Springfield, Va.   22161.)
                                -25-

-------
5.  Martin, R. M..  Conctruction Details of Isokinetic Source-
    Sampling Equipment.  Environmental Protection Agency
    Technical Report No. APTD - 0581.  (Available from Environ-
    mental Protection Agency, Air Pollution Technical  Information
    Center, Research Triangle Park, N.C.  27711.)
6.  Rom, J. J..  Maintenance, Calibration, and Operation of
    Isokinetic Source Sampling Equipment.  Environmental
    Protection Agency Technical Report No. APTD - 0576.
    (Available from Environmental Protection Agency, Air
    Pollution Technical Information Center, Research Triangle
    Park, N.C.  27711.)
7.  Miller, I., and J. E. Freund.  Probability and Statistics
    for Engineers.  Prentice Hall, Inc., Englewood Cliff,
    N.J.  1965.
                                -26-

-------
                                  TECHNICAL REPORT DATA
                           (Please read Instructions on the reverse before completing)
  1^41,0/4-75-025
3 RECIPIENT'S ACCESSIOI»NO.
4 TITLE AND SUBTITLE
   Method for Obtaining Replicate Particulate
   Samples from Stationary Sources
                                                          S REPORT DATE
                                                               June  1975
6 PERFORMING ORGANIZATION CODE
7 AUTHOR(S)
                                                          8 PERFORMING ORGANIZATION REPORT NO.
   William J.  Mitchell and M.  Rodney Midgett
9 PERFORMING ORGANIZATION NAME AND ADDRESS
   U. S. Environmental Protection Agency
   National Environmental Research Center
   Quality Assurance and Environmental Monitoring Lab
   Research Triangle  Park.  North Carolina 27711
10. PROGRAM ELEMENT NO.

  1HA327
11 CONTRACT/GRANT NO
 12. SPONSORING AGENCY NAME AND ADDRESS

   U. S. Environmental Protection Agency

   Office of Research and Development

   Washington, D.  C. 20460
13. TYPE OF REPORT AND PERIOD COVERED
  Final
14 SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
 16. ABSTRACT
        Two sampling arrangements  that allow four independent  trains  to sample
   simultaneously at the  same  point in the stack are described.  These  sampling
   arrangements have been used to  obtain replicate particulate samples  from
   stationary sources.  Sixteen particulate sampling runs that used these sampling
   arrangements with four identical  EPA Method 5 particulate trains determined that
   the magnitude of the within-run standard deviation was linearly related to the
   magnitude of the mean  particulate concentration measured for that  run.   However,
   no such relationship was  found  between the within-run standard deviation and
   either the run sampling time or the total volume of gas sampled.
        From a statistical analysis  of the percent moisture in the stack determination,
   an estimate is made of the  precision of the condensation technique used in the
   Method 5 sampling train.
17
                               KEY WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
                                             b.lDENTIFIERS/OPEN ENDED TERMS
             c.  COSATI Field/Group
   Environmental monitoring
   Sampling
   Particulates
   Air pollution
   Stack emissions
18 DISTRIBUTION STATEMENT
   Release unlimited
                                             Id SECURITY CLASS (This Report)

                                               Unclassified	
              21 NO. OF PAGES
                                             20 SECURITY CLASS (Thispage)
              22 PRICE
EPA Form 2220-1 (9-73)
                                           27

-------