-------
Analogous to the laboratory tests with feldspar dust, no significant
penetration of the in-stack filters was found. Penetration of the Method 5
filter trains was found in three instances but this likely resulted from the
filter slipping out of position and not an actual penetration. (Filter slippage
occurred because the screen filter supports used did not contain a deep enough
recessed area to accommodate the glass fiber filter. These screens were made by
hand and were used in place of the standard fritted-glass filter supports used
in the other field tests to ensure that any material penetrating the Method 5
filter would not be trapped in the filter support itself.)
Statistical analysis of the results showed that the concentration level
increased over the four days of testing, but no correlation between concentra-
tion and between-train (Gelman versus Method 5, BGI versus Method 5) precision
or concentration and within-train (Method 5 versus Method 5, BGI versus BGI,
Gelman versus Gelman) precision was found. Further, no significant difference
was detected between Method 5 and Gelman or Method 5 and BGI, which indicates
that the two in-stack filters would give similar results if compared directly
(Table 5). (Some small amount of stratification was observed, based on a
position by position comparison of the sampling results for the assembly that
contained four Method 5 trains, and this difference in positions was handled
in the statistical analysis.)
17
-------
TABLE 5. RESULTS AT FELDSPAR PLANTS
Sampling
assembly
A
B
C
A1
B1
C1
A"
Train
position
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
Type
train
Method 5
BGI
BGI
Method 5
Method 5
Method 5
Method 5
Method 5
Method 5
Gel man
Gel man
Method 5
Method 5
Method 5
Method 5
Method 5
Method 5
BGI
BGI
Method 5
Method 5
Gel man
Method 5
Gel man
Method 5
Method 5
Method 5
Method 5
Parti cul ate cone.
measured (mg/m )
First run
72.2
88.6
94.2.
74. 4b
87.3
80.5
85.7
81.7
83.7
74.4
83.5
84.3
85. 9C
87.3
92.8
82.6
120
118
123
113
131
128
122
114
127
122
142
106
Second run
78.5
89.0
85.8
74.1
90.5
72.9
77.9
72.3
83.9
95.6
79.2
95.2
131
129
120
124
.126
124
105
113
175r
170C
139
153
135C
150
136
131
Mg in
back-up
per run
0.2
0.0
0.4
2.5
0.3
0.3
0.0
0.0
0.0
0.0
1.2
0.0
6.8
0.0
0.2
0.0
0.0
0.0
0.0
0.0
0.0
7.8
0.0
0.0
8.0
0.0
0.0
0.4
Average
back-upocatch
in mg/m /train
0.1
0.0
0.1
0.8
0.1
0.1
4 0.0
0.0
0.0
0.0
0.8
0.0
6.6
0.0
0.2
0.0
0.0
0.0
0.0
0.0
0.0
4.6
0.0
0.0
7.2
0.0
0.0
0.3
aSee Figure 1 for explanation of train position.
Filter cracked during testing.
cFilter definitely slipped out of position during testing.
18
-------
REFERENCES
1. U.S. Environmental Protection Agency. Standards of Performance for New
Stationary Sources. Federal Register 36(247):24876-24890, December 23,
1971.
2. U.S. Environmental Protection Agency. Standards of Performance for New
Stationary Sources. Federal Register, 42_( 37): 7584-7596, February 23, 1978.
3. Mitchell, W. J., and M. R. Midgett. A Means to Evaluate the Performance of
Stationary Source Test Methods. Environ. Sci. and Technol., Kh85-88, 1976.
19
-------
APPENDIX A
A FIELD EVALUATION OF EPA METHOD 17 IN A KRAFT PULP MILLa
I. INTRODUCTION
The proposed standard for particulate matter emissions from Kraft pulp mill recovery
furnaces is 100 milligrams/dry standard cubic meter (mg/dscm) as set forth in the Federal
Register, Part II, Volume 41, Number 187, Friday, September 24, 1976. The specified test
method for particulate determination is EPA Method 5. EPA Method 17 may be used as an
alternative method provided that a constant value of 9 mg/dscm is added to the Method 17
results and provided that the stack temperature is no greater than 205 °C. Water is used as a
cleanup solvent instead of acetone in the sample recovery in both methods.
A field evaluation of EPA Method 17 at a Kraft pulp mill recovery furnace to include a
comparison with EPA Method 5 was therefore considered appropriate. This report presents
the results obtained and conclusions drawn from statistical analysis of the test data.
Report prepared by Dr. H. F. Hamil and Mr. R. E. Thomas of Southwest Research
Institute, San Antonio, TX, under EPA Contract 68-02-2489 and under the direc-
tion of Mr. Michael C. Osborne, EPA, Research Triangle Park, NC.
20
-------
II. TEST DESCRIPTION
The field evaluation of Method 17 specified in the work assignment has a two-fold pur-
pose: a field comparison of EPA Methods 17 and 5 by simultaneous sampling of the same
source, and an evaluation of various commercially available Method 17 filter holders by
simultaneous sampling of the same source.
The test plan was provided by the work assignment monitor and is shown in Figure 1.
During the first week, six runs were to be accomplished with two Method 5 and two Method
17 trains sampling simultaneously on each run. The Method 17 filters were to be backed up
by a heated Method 5 filter and glass-lined probe to collect any paniculate which might
pass through the in-stack filter. Nutech filter holders were to be used in the first week's test
as representative Method 17 filter assemblies. The relative position of the trains in the
sampling assembly were to be rotated after each run to avoid any bias due to sample nozzle
location.
During the second week, eight runs were to be accomplished with one Method 5 train
and three Method 17 trains sampling simultaneously on each run. The Method 17 trains
1. All test runs will be conducted with the configuration shown in Figure 2.
2. Runs 1-6 Characteristics
a. Trains 1 and 3 will be standard Method 5 trains with heated filter boxes.
b. Trains 2 and 4 will be Method 17 trains with heated backup Method 5 filters.
c. The relative position of trains 1,2,3, and 4 within the sampling arrangement will be rotated
clockwise after each run.
d. An additional front-half cleanup will be required following the regular cleanup of runs 1,2, and 6.
e. Saving and determining the back-half catch will also be required in runs 1, 2, and 6.
f. The Nutech filter holders should be used to represent Method 17 unless it is observed during pilot
calibration that these holders cause significant turbulence that changes the pilot tube coefficient.
If such a disturbance is observed, the Misco filler holders should be used in place of the Nutech
filter holders.
g. All filter holders and train components should be thoroughly leak checked before sending them
into the field.
(continued) _. _ .
Figure 1. Test plan for Kraft mill recovery furnace.
21
-------
3. Runs 7-14 Characteristics
a. Trains 1,2,3, and 4 should use Misco, BGI, Nutech and Method 5 filter holders, respectively.
b. Trains 1,2, and 3 should be operated with filters in-stack as characterized by Method 17.
c. Train 4 should be operated as a standard Method 5 train with the filter holders out-of-stack in a
heated compartment.
d. The relative position of trains 1,2,3, and 4 within the sampling arrangement will be rotated
clockwise after each run.
e. It will not be necessary to do any additional front-half cleanups or to save the ba.ck-half catch.
Figure 1. (continued)
were to employ one each Nutech, Misco, and BGI filter holder. No backup Method 5 filters
were to be used in the Method 17 trains during the second week of testing. The relative posi-
tion of the trains within the sampling bundle were to be rotated after each run as in the first
week of testing.
The use of a four-train sampling system was required to accomplish the test plan. To
assure that no bias due to spatial variation in particulate loading was introduced into the
sampling, the probe bundle was assembled so the sample nozzles were situated at the cor-
ners of a square with sides of 6 cm. A single pilot tube was centrally located. The pilot tube-
sample nozzle spacings are shown schematically in Figure 2. A view of the assembled probe
bundle with a Method 5 probe and the three Method 17 probes is shown in Figure 3.
A common thermostatically-controlled oven was fabricated to house the four Method
5 filter holders required in the first week of tesling. A view of Ihe filler holder arrangements
is shown in Figure 4. The filler holder oven was mounled on a sland in such a manner lhai
four slandard impinger irains wilh ice balhs could be accommodaled in an assembly
mounted on a common base.
This configuration allowed Ihe enlire sampling syslem lo be assembled and leak
checked. The enlire unil could ihen be moved to place Ihe sample nozzles al Ihe desired
sample poinl in Ihe slack. Single poinl sampling wilhoul iraversing was employed
Ihroughoul Ihe tesl. The assembled sampling syslem is shown in Figure 5.
To avoid differences in filler media, all Melhod 5, Nulech Melhod 17, and Misco
Melhod 17 fillers were cut from Gelman lype A-E glass fiber high-volume fillers. BGI glass
fiber filters were used with the BGI filter holders.
The sile selecied for Ihe field evaluation was Ihe recovery furnace of a Krafl pulp mill
owned and operaied by Ihe Arkansas Kraft Corporalion, Morrillon, Arkansas. The
22
-------
-6
tea
y- ""
-------
FigureS. Four probe sample bundle.
:igure 4. Filter holder oven
-------
Ready for leak check.
During sampling.
Figure 5. Sampling system.
25
-------
East Stack
-O-
15'
West Stack
*-68"ID-
-38'-^-
,,4" Port
JL
20"
17"
t
25"
-18'
Roof on ESP
' r^l/
Figure 6. Arkansas Kraft Corporation precipitator exitsfront view.
of week one, problems were encountered with leaks in the Method 5 filter holders on trains
C and D, which were positioned nearest to the oven heating element. Presumably uneven
heating caused the metal clamps on the Method 5 filter holders to loosen. Repositioning of
the filter holders as far from the heating element as possible in conjunction with checking
the tightness of the filter holder clamps after oven heatup corrected the problem on sub-
sequent runs. Even though leaks occurred in only one of the four trains in each of the first
two runs, the data from these runs are considered invalid for the purpose of the test and are
not included in this report. However, two additional runs were made the first week, giving
the total of six valid runs required by the test plan. A summary of the run parameters for
the fourteen valid runs is presented in Appendix A.
26
-------
15'-
38"
Ladder
Figure 7. Arkansas Kraft Corporation precipitator exitstop view.
The paniculate loadings for each train within each run are presented in Table I for the
first week's test and in Table II for the second week's test. The particulate loading Cs for
the Method 17 determinations was calculated three ways for the first week of testing (Table
I). The Cs values presented under the heading "Method 17" were calculated using the
actual particulate catch in the Method 17 filter and filter holder front half. The Cs values
presented under the heading "Method 17 Adjusted" were calculated as just described
above but with a constant correction of 9 mg/dscm added according to the instructions in
the Federal Register pertaining to use of Method 17 in Kraft pulp mills. The Cs values
presented under the heading "Method 17 Plus Backup" were calculated using the par-
ticulate catch from the Method 17 filter and filter holder front half, the backup probe, and
the backup Method 5 filter and filter holder front half.
The particulate loading Cs for the Method 17 determinations for the second week of
testing is presented in Table II. Also presented in Table II are the differences, in mg/m3,
between the Method 5 train and the individual Method 17 trains.
The actual weight of particulate caught and the location within the sample train are
given in Table III for week one and Table IV for week two. Also included in Table III is the
back-half catch for each train obtained by workup of the impinger catch. These back-half
catch values are substantial but probably do not reflect filter inefficiency. The residue
which was obtained upon evaporating the impinger water to dryness was not a particulate
solid but was an oily semiliquid material. It probably consists of materials volatile at the
filter oven temperature which are condensed in the impingers.
27
-------
TABLE I. PART1CULATE LOADING, Cs
mg/dscm
WEEK 1
Method 17a Method 17b
Run Number Method 17 Adjusted Plus Backup Methods
3A
B
C
D
4A
B
C
D
5A
B
C
D
6A
B
C
D
7A
B
C
D
8A
B
C
D
Mean
18.8
20.7
40.3
46.3
46.4
47.0
44.4
37.0
60.9
53.5
61.2
54.1
44.22
27.8
29.7
49.3
55.3
55.4
56.0
53.4
46.0
69.9
62.5
70.2
63.1
53.22
34.4
34.9
53.5
55.6
60.7
62.5
54.4
46.7
70.4
62.3
73.7
63.5
56.05
35.4
34.4
49.2
57.4
58.2
52.9
51.6
56.4
71.5
70.1
63.1
73.5
56.12
a. Actual Method 17 Cs value plus 9 mg/dscm per Kraft pulp mill
regulations.
b. Actual Method 17 Cs value plus the material collected in the
glass-lined probe and the back-up Method 5 filter and filter holder.
28
-------
TABLE II. PARTICULATE LOADING, Cs
mg/dscm (Sampling Train)
WEEK 2
Run
No. Methods Nutech
53.0(B) 35.2(A)
Armal Method 17a
11SCO
BGI
44.7(C) 45.4(D)
33.l(C) 20.3(D) 28.0(B) 29.l(A)
68.9(8) 5l.8(A) 54.3(C) 25.7(D)
49.4(8) 54.l(Q
68.6(A) 50.0(D)
39.8(8) 35.0(A) 30.0(C) 29.8(D)
3l.3(A) 27.I(D) 29.7(8) 26.6(C)
15 162.5(8) 137.7(A) 164.9(Q 138.3(D)
16 162.7(Q 196.7(D) 155.8(8) 127.5(A) 34.0
Mean 8.3
Differences
(Method S-Method 17)
Nutech Misco BGI
17.8 8.3b 7.6
12.8 5.1 4.0
17.1 14.6 43.2C
18.6 19.2 14.5
4.8 9.8 10.0
4.2 1.6 4.7
24.8 2.4 24.2
6.9 35.2
7.9 17.9
a. Does not include the 9 mg/dscm conversion factor. In the statistical analysis,
9 mg/dscm was added to these values per Kraft pulp mill regulations.
b. Leak rate at conclusion of sampling four times the allowance leak rate.
c. Large paniculate deposit observed in probe indicating leakage around Filter.
Results not deleted from analysis.
29
-------
TABLE III. PARTICULATE CATCH, mg
WEEK !
Run
3
4
5
6
7
8
Train
A
B
C
D
A
B
C
D
A
B
C
D
A
B
C
D
A
B
C
D
A
B
C
D
Method 5
Filter
12.8
29.1
13.0
31.5
52.8
10.1
59.3
6.8
8.3
55.5
11.2
52.9
53.2
7.1
58.6
7.2
9.6
79.2
6.3
75.2
71.3
11.4
84.4
6.1
Method 17
Filter
17.5
14.1
._.
47.3
.
50.5
46.9
52.0
-__
46.1
...
41.0
78.3
60.9
74.7
65.0
Method 17 Nozzle
and Filter Holder
2.8
8.3
...
,
1.0
5.0
9.1
...
3.6
...
6.1
...
2.9
3.8
8.3
...
3.3
2.6
a. Oily residue obtained upon evaporation of water.
Method 5 Probe and Impinger Train and Back3
Front of Filter Holder of Filter Holder
4.1 46.5
9.8 40.0
2.5 43.1
6.0 32.4
6.9 30.8
5.7 36.9
9.9 30.3
4.4 34.8
8.4 34.6
15.7 34.2
7.1 34.2
9.7 34.0
8.1 61.6
4.7 68.3
8.0 60.4
4.4 69.9
3.2 42.3
17.0 35.4
5.1 43.4
13.4 38.4
10.0 . 100.9
4.5 106.0
9.0 100.8
5.7 106.1
30
-------
TABLE IV. PART1CULATE CATCH, mg
WEEK 2
Methods Method 17 Nozzle, Probe, Nozzle, M-17 Method 17 Filter
Run Train Filter Filter M-5 Front Half Filter Holder Holder Type
9 A 43.4 3.2 Nutech
B 59.2 8.5
C 53.0 5.0 Misco
D 54.5 4.3 BG1
10 A 30.2 3.9 BGI
B 29.2 3.3 Misco
C 29.9 9.1
D 18.2 5.8 Nutech
11 A 68.1 1.6 Nutech
B 68.9 22.4
C 63.5 9.0 Misco
D 28.0 6.8 BGI
12 A 72.4 18.6
B 61.8 2.4 Misco
C 66.8 4.3 BGI
D 62.3 3.2 Nutech
13 A 38.5 3.8 Nutech
B 43.5 6.0
C 28.9 8.6 Misco
D 33.3 3.3 BGI
14 A 29.0 8.1
B 28.9 6.4 Misco
C 26.5 5.2 BGI
D 26.7 5.2 Nutech
15 A 146.9 4.0 Nutech
B 171.8 10.9
C 167.5 7.2 Misco
D 144.9 5.9 BGI
16 A 113.7 8.7 BGI
B 102.4 48.5 Misco
C 142.3 17.1
D 181.4 8.2 Nutech
The particulate material recovered in the three special cleanup procedures is tabulated
in Table V. As can be seen from the values presented, the residual particulate left in a
sampling train after a properly conducted sample recovery is quite small and should not
affect the accuracy of a determination.
Since the second week of the test was intended to compare the performance of various
commercially available Method 17 filter holders, a discussion of their relative merits will be
presented here. The statistical evaluation of their performance with regard to particulate
determination is included in the next section.
31
-------
TABLE V. PARTICULATE RECOVERED POSTCLEANUP/CLEANUP
mg
1A IB 1C ID 2A 2B 2C 2D 8A 8B 8C 8D
Nozzle, probe, M-5
filler front-half 1.1 4.4 2.8 0.4 1.9 2.1
Nozzle, M-17
filter front 1.7 1.4 1.3 2.0 1.9 3.0
M-17 back, probe,
M-5 front 1.9 1.4 2.3 1.5 3.1 1.3
Three Method 17 filter holders were evaluated: Nutech, Misco and BGI. The Nutech
filter holder was quite heavy and somewhat cumbersome to assemble. The substitution of
hex-head bolts for the slotted-head bolts used in assembling the filter holder permitted the
use of a wrench instead of a screwdriver. This substitution speeded up the assembly and
provided a means of more even tightening of the assembly to prevent leaks. Sample
recovery with the Nutech is relatively straightforward, but care must be exercised in
disassembly for sample recovery, since that portion of the filter in contact with the gasket
adheres to the gasket quite strongly. The portion of the filter supported by the sintered
metal support plate breaks free, and the balance of the filter must be scraped from the
gasket with a spatula. With careful recovery techniques, this filter holder appears to give
good results.
The Misco filter holder is more compact than the Nutech and is easier to assemble.
However, two shortcomings were reported by the field crew. Due to the relatively small
filtration area available with the Misco, the pressure drop across the filter increases more
rapidly during a run than with the other two filter holders. During the two final runs of the
test, the ESP was operating in an upset condition. The particulate loading was high (ca.
125-200 mg/dscm), and the Misco filters blinded off to such an extent that isokinetic sampl-
ing could not be maintained for one hour. Both these runs were terminated at the point that
isokinesis could not be maintained. The second problem involves the Teflon gaskets used in
the Misco filter holder. The three wing nuts used to secure the assembly can be tightened
finger-tight before a run, and the gaskets will provide a leak-free seal. However, due to the
high coefficient of thermal expansion and plastic flow characteristics of Teflon, the gaskets
deform upon heating to about 180°C (350 °F). As a result of this deformation, the gaskets
tend to leak upon cooling. More important than these leaks on cooling is the fact that the
filter is subjected to fairly high pressure by the gasket due to the expansion of the gasket on
heating. The glass fiber filter material is pressed into the surface of the gasket and is
extremely difficult to remove during sample recovery. Care must be exercised to avoid
scraping Teflon off the face of the gasket while trying to remove the filter material with a
spatula. This problem made sample recovery from the Misco filter holder more difficult
than for either of the other two filter holders evaluated. Use of alternative gasket materials
should be considered when using the Misco unit at elevated temperatures.
32
-------
The BGI thimble filter holder was the lightest, easiest to clean, and easiest to assemble
of. the three units evaluated. Since it is only slightly larger in diameter than a standard
Method 5 probe, it is easy to insert and withdraw through a standard four-inch sample
port. The major problem encountered with the BGI involves the thimble filter rather than
the filter holder itself. The inside diameters of the lot of thimble filters used were not
uniform. As a result, the fit of the thimble onto the thimble mounting tube of the filter
holder ranged from snug to loose. The possibility of particulate matter bypassing the filter
exists for those thimbles which do not fit tightly. During the portion of the test involved
with filter holder evaluation, no backup filters were used on the Method 17 trains.
However, during sample recovery, the glass probe liners used behind the Method 17 filters
were examined visually for particulate matter. At the end of Run 11, Train D, equipped
with the BGI thimble, was observed to have considerable particulate material in the probe
liner. The calculated Method 17 Cs value for this run was approximately 50 percent of the
Cs values for the other two Method 17 determinations and the Method 5 determination in
Run 11. Either some modification of the filter holder clamp or closer quality control during
filter manufacture to assure a leak-free fit of the thimble into the holder should be
considered.
33
-------
III. RESULTS
The data from the test runs were submitted to statistical evaluation. For Method 17,
9 mg/dscm were added to the observed paniculate loading as specified in the Test Methods
and Procedures section of the Federal Register pertaining to Kraft mills to obtain the
adjusted particulate concentration. The results from each week of the test are treated
separately below.
A. WEEK 1
The particulate loadings from Runs 3 through 8 were used to evaluate Method 17
(adjusted) and Method 17 plus back-half catch in comparison with Method 5 for particulate
determination. The first two runs were not considered representative due to the problems
encountered and are not included in these analyses. The two evaluations were treated
separately, and the results are summarized below.
7. Method 5 versus Method 17 (Adjusted)
The sampling trains, A, B, C, and D, retained their integrity throughout the test. The
only change was made at the probe, with either only a nozzle or a combination of nozzle
and Method 17 filter attached. As such, the possibility exists that differences might exist
between trains which could influence the results. This possibility of train-to-train dif-
ferences is investigated using an analysis of variance (ANOVA) with factors of runs and
trains. The run-by-train interaction is used as the error term for this ANOVA in the absence
of replicate samples. The results are summarized in Table VI.
The between-trains factor is determined not to be significant at the 5-percent level, and
the conclusion is that there is no need to account for the sample train in future analyses. In
addition to this, a Friedman test,* a nonparametric equivalent to the two-way ANOVA,
was run on these data to confirm the result. The test statistic was calculated at 7.04 with 3
degrees of freedom, which is not significant when compared to a table of the chi-squared
distribution at a significance level of 5 percent.
Since the train factor was not significant, the mean square for error from the ANOVA
was used as an estimate of sampling variability and the differences between the Method 5
and Method 17 results investigated. A t-test was used for this comparison, using the average
Snedecor, G.W. and Cochran, W.G., Statistical Methods, 6th Edition Iowa State University Press, Ames., 1967.
34
-------
TABLE VI. SUMMARY OF ANALYSES COMPAR-
ING METHOD 17 (ADJUSTED) AND METHOD 5
ANOVA Table
Source df SS MS F
Between Runs
Between Trains
Run x Train
(error)
Total
5
3
15
23
3553.97
52.78
244.50
3851.25
710.79
17.59 1.08*
16.30
y = MSerror
Friedman Test
Train
B C
Rank sum 9.5 20 18 12.5
Test Statistic = 7.04, 3 df*
Means (mg/dscm)
Method 5 Method 17 (adjusted)
56.15 53.22
t statistic =1.78, 15df*
Not significant at 5-percent level.
FQ 05(3,15) = 3.29,X^05(3) =7.81,1005(15) =2.13
values for the 12 determinations by each method. A t-statistic is calculated according to the
formula
where *s average Method 5 results
Xi 7 average Method 17 results
&2 mean square for error from the ANOVA.
The calculated t was 1.78, which is not significant at the 5-percent level. The con-
clusion is, then, that the two methods gave equivalent results with the correction factor of 9
mg/dscm added to the result obtained from the particulate catch.
To estimate the variability associated with each method, a percent coefficient of var-
iation was calculated for each run using the ratio of the standard deviation between the two
results to the average value. The individual values obtained and the averages are shown in
35
-------
3
4
5
6
7
8
Avg
12.97
8.07
0.76
10.60
1.96
7.57
7.99
Table VII. The average for Method 5 was 6.34 percent, while for Method 17, it was 7.99
percent.
TABLE VII. COMPARATIVE
COEFFICIENTS OF VARIATION,
METHOD 17 (ADJUSTED)
VERSUS METHOD 5 (Percent)
Run Method 17 (Adjusted) Method 5
2.00
10.93
6.66
6.25
1.41
10.79
6.34
2. Method 5 versus Method 1 7 plus Back-Half Catch
The paniculate loadings for Method 17 were added to the particulate catch from the
probe wash and Method 5 backup filter to determine total particulate matter for each run.
The concentrations calculated in this manner were analyzed in the same manner as the
Method 17 data, and the results are presented below.
The ANOVA for investigation of a train effect is shown in Table VIII. As before, the
F-ratio for between-train variability is insignificant, and no allowance for train differences
is indicated. The Friedman test gave a test statistic of 6.40, which also is not significant, and
it is concluded that no bias existed among the four trains.
The mean for the twelve Method 17 determinations with the back-half workup includ-
ed was 56.05, compared to the Method 5 mean of 56.15 mg/dscm. The resultant t-statistic
was calculated to be 0.05 and is clearly nonsignificant.
The percent coefficients of variation for these runs are shown in Table IX, along with
their average value. The mean of 5.97 percent compares favorably to the Method 5 value of
6.34 percent.
B. WEEK 2
Using the eight runs from the second week of testing, the three types of Method 17
filters were compared both against each other and against the Method 5 data. The first
analysis, as with the previous data, was to determine if any train-to-train bias existed which
could influence the test results.
36
-------
TABLE VIII. SUMMARY OF ANALYSES
COMPARING METHOD 17 PLUS BACKUP CATCH
AND METHOD 5
ANOVA Table
Source df SS MS F
Between Runs 5 3145.92 629.18
Between Trains 3 73.27 24.42 1.60*
Run x Train (error) 15 229.33 15.29
Total 23 3448.52
52 = MSerror = 15.29
Friedman Test
Train
A B C D
Rank Sum 11 19 19 11
Test Statistic = 6.40, 3 df*
Means (mg/dscm)
Method 17 +
Method 5 back-half
56.15 56.05
t-statistic = 0.05, 15df*
Not significant at 5-percent level.
F0.05(3,15) = 3.29, X2} 05(3) = 7.81, to05 (15)= 2.13
TABLE IX. COMPARATIVE COEFFICIENTS
OF VARIATION, METHOD 17 PLUS
BACKUP CATCH VERSUS METHOD 5
(Percent)
Method 17 +
Run
3
4
5
6
7
8
Avg
Back-Half
1.14
2.75
2.02
10.81
8.61
10.49
5.97
Method5
2.00
10.93
6.66
6.25
1.41
10.79
6.34
37
-------
The ANOVA is summarized in Table X. The F-ratio for the between-trains component
is less than one and clearly nonsignificant. The Friedman test for these data gave a test
statistic of 1.35, which also is not significant. As a result, no train bias can be detected., and
no allowance is made in the subsequent analyses for the sampling train used to collect the
sample.
TABLE X. SUMMARY OF ANALYSES ON SECOND
WEEK'S DATA
ANOVA Table
Source df SS MS F
Between Runs 7 83,275.0811,896.44
Between Trains 3 239.98 79.99 <1*
Run x Train (error) 21 3,647.56 173.69
Total 31 87,162.62
Friedman Test (train bias)
Train
A B C D
Rank Sum 18 23 21 18
Test Statistic = 1.35, 3 df*
Friedman Test (filter type)
Nutech Misco BG1
Rank Sum 16 16 16
Test Statistic = 0.00*
Means (mg/dscm)
Methods Nutech Misco BGI
77.49 78.23 78.60 68.56
LSD 0513.71
Ho: AN + MM + MB ~ 3/*5 = °
Test Statistic = 1.57, 21 df*
*Not significant at 5-percent level.
F0 05 (3,21) = 3.07,\^M(3) = 7.81, t0 05 (21) = 2.08
The first examination of the Method 17 trains was made using the Friedman test. The
factors in this analysis were the runs and the type of filter. The values for the three Method
17 trains were ranked across each run and the rank sums computed. The rank sunn was
calculated to be 16 for each of the three trains, and no difference can be detected.
38
-------
A parametric approach is also used. A Least Significant Difference (LSD) is calculated
according to the formula
where
a2 mean square for error from the ANOVA
and n =8, the number of runs used to obtain the
mean for each type of filter.
Substituting gave an LSD of 13.71 for these data. The averages were 78.23, 78.60, and
68.56 mg/dscm for the Nutech, Misco, and BGI filters, respectively. Since the range of
these means, 10.04, was less than the LSD, no difference among the filter types can be
detected.
The Method 5 mean for these runs was 77.49 and shows good agreement with both the
Nutech and Misco average values. To test for consistency between the two methods, a con-
trast among the means is tested. The hypothesis is
HO:MN +MM +MB ~3MS =0
where
MN mean Method 17 result using Nutech filters
MM mean Method 17 result using Misco filters
MB mean Method 17 result using BGI filters
and MS mean Method 5 result.
The test statistic is calculated as
Xia ' Xvi ~T~ Xp 3Xe
A. _ II m. P 3
where
XN xs sample means corresponding to the /i's
above
Cj coefficient of the means in the null
hypothesis
n number of runs
and a1 mean square for error from the ANOVA.
Substituting gives a calculated t-statistic of -1.57. This does not exceed the tabled t-value
for 21 degrees of freedom at the 5-percent level, and the results may be considered
equivalent.
39
-------
IV. SUMMARY AND CONCLUSIONS
The results of the statistical analysis of data from the first week of testing indicate that
Method 5 and Method 17 gave equivalent particulate results at a Kraft pulp mill recovery
furnace at the 95-percent confidence level when the correction factor of 9 mg/dscm
specified in the Federal Register was added to the Method 17 determination. Additionally,
the Method 5 versus Method 17 plus backup filter data indicated that the particulate
loadings obtained on all four trains were equivalent throughout the first week of testing.
The variability of Method 5 and Method 17 (Adjusted) was good, being 6.34 percent and
7.99 percent, respectively.
The results of the statistical analysis of data from the second week of testing indicate
that the particulate loadings determined with all three Method 17 filter assemblies were
equivalent at the 95-percent confidence level. The Method 17 (Adjusted) particulate
loadings determined during the second week of testing were also equivalent to the Method 5
result at the 95-percent confidence level.
40
-------
APPENDIX A
RUN PARAMETERS SUMMARY
41
-------
<
at
H
Z Z
z z z z
Z Z Z Z
o
I?
o
u
a a
Q r~~ y? ^4 *"** 4 ^ 2 2 **"" S ^ ""^ ^2 *"~ 2 2 ^ ^2 *"^
dddd dddd 6006 dddd dodo
r- r- r- r-
l
u a:
O o\
" « »N
; IN oo
' "
_. I _. t ci r-r-r-oo ^r-i^- «^io>^>«^ OOOO^OCTV
2 R R ?5 S RRRR R R R S aSaS S?3?5?5
0j (dddd dddd dddd dddd dddd
^f I-«O (Ni^-^C
r-»ooi--^> rn IN r-l r-
odadodod
2 £ S S S$2= 2^*2 SSS8
r4«Nr4(N
& & & O> fNtNfSfN r<(NtN(N «NfNfS*N OOOOOOOO
<& <& \Q & ^o^O^O^ ooaoooao OOOQOOQO rnm«*im
* * ^ ^ viw^irtw^ irt w^ w^ irt vjW)«/i«r> r~- r- t~- t^-
oddd dodo dddd ddod ddod
-------
jj O
<
cc
I-
z z Z 2= «2 z z
8
3*
o
g
u
_i
N
N
r^ ^ f- Q O r-- »r> f"J r-l */"> r- Q O r-- >/% *N «N r- v» o
oooo o o o o° oooo oooo oooo
s
aoacaomrn«*i
r^ «_ OOOO
_ __ rr, _ oo oo (N O OOr-
%O ^) O "^
-------
_
" *
S si z z 5 « « S Z
8
O
O
ct
O
dodo odod odod dodo
ssss
oooooooo r-r-r-
CO
3 Of!
.S «
«*1 [ ON >n «N I
o 8 o? o o & a? ON & §^ 5! & *^^o
A 6
H >
x <
=3 °-
S z
a D
>| w-iOmrn r-Or-O SSiNS f^rNriS
"'^a'^S aS'S'S odoo odoo
11
oo ^ r-t oo i
*i r- r- ^
. Q \& oo ^
' 3 «2 r- o o 5^ U
OOOO r-r-r-r-
r- i^ r- r~ CNOXO
0-
<
(*i ^ r*l Ov
r*i O i-
i «*1 * <
OOOO *eyj>\C v^ >r> in >o vO >C vO S S ^ S S
o'do'd do'od do'dd odod
c E
c c s e
3 =1 O v>
«r "r m TT
s s 5 S
bo bfi O O
S = -a -a
S g-S-S
a -a S E
-S-gss
§ S S S
44
-------
APPENDIX B
A COMPARATIVE STUDY OF METHOD 17 WITH
METHOD 5 AT A GRAIN ELEVATOR3
H. F. Hamil
I. INTRODUCTION
Prior to proposing new source performance standards for
grain elevators, the Emissions Standards and Engineering Division
of the Office of Air Quality Planning and Standards obtained source
emission data on approximately twelve grain elevators. This work
was performed to characterize the emissions from grain elevators
and to determine an emission rate based upon best available con-
trol technology. Most, if not all, of these source tests were
conducted using EPA Method 5. Based upon the results of this
testing, a particulate emission standard of 0.023 grams per dry
standard cubic meter was set. This level of particulate emissions
was achievable by use of baghouse collectors.
When the particulate emission standard for grain elevators
was proposed in the Federal Register, the reference method listed
for determination of particulate emissions was not Method 5, but
Method 17, an in-stack filtration method.
The Quality Assurance Branch, Environmental Monitoring and
Support Laboratory (QAB/EMSL) initiated a comparative study of
EPA Methods 5 and 17 in a grain elevator to determine their
relative performance in collecting samples simultaneously. This
study was conducted at a baghouse collector-equipped grain ele-
vator. This facility operated with particulate emissions which
were near the proposed standard. The results of the QAB compara-
tive testing indicated that Method 5 collected about 20 percent
more particulates than did Method 17 when equivalent volumes of
stack gas were sampled. It should be noted that this relatively
large percentage difference amounted to an average difference of
only 6 milligrams between the two methods. Such small variations
in mass of particulate collected were relatively unimportant in
the past for sources with higher standards. With the lower
proposed standards, these differences should be resolved, since1
a review of the data collected by the Emission Measurement Branch
for the purpose of proposing the grain elevator standard indicated
that the average particulate catch by Method 5 in these tests
was 50 milligrams. Additional comparative testing at a grain
elevator was considered warranted.
Report prepared by Dr. H. F. Hamil and Mr. R. E. Thomas of Southwest Research
Institute, San Antonio, TX, under EPA Contract 68-02-2^89 and under the direc-
tion of Mr. Michael C. Osborne, EPA, Research Triangle Park, NC.
45
-------
II. TEST PLAN AND SITE
A. Test Plan
Week 1. Sampling was accomplished with a four-probe/one-
pitot tube assembly. The sample nozzles were at the corners of
a square array 6 centimeters on a side. The one pitot tube was
centrally located between the four sample nozzles. Two Method 5
trains were operated diagonally opposite two Method 17 trains,
as shown in Figure 1. Glass-lined probes with new glass liners
were used on all four trains, with Method 5 filters installed
as backups on the Method 17 probe-filter assemblies to collect
any particulate not collected by the Method 17 filter. BGI
thimbles and holders were used as representative Method 17 filters,
A complete cleanup of all trains was performed prior to
the first run. Upon completion of the sample recovery, including
retention of the impinger catch and train back-half cleanup, an
additional complete train cleanup was performed to insure no
particulate remained in the sample trains. In this fashion, a
complete train cleanup, separate from sample recovery, preceded
each sampling run, and an additional train cleanup was performed
after the last run. The wash solutions from these additional
cleanups were worked up to determine residual particulate matter
in the trains after normal sample recovery. Six runs were made
during week 1 by this sampling procedure.
Week 2. The test plan for the second week incorporated
one major modification. This modification consisted of moving
the Method 17 filter out-of-stack to the same location in the
trains as a Method 5 filter in order to evaluate the performance
of both Method 5 and 17 filters under equivalent conditions
(Figure 2). No backup filters were used. All other aspects
of the tests were identical to the first week's tests described
above. Five runs were made the second week.
During both weeks of the test, the positions of the sam-
pling probes were rotated clockwise in order to avoid any spatial
bias.
B. Test Site
The site selected for this test was a grain elevator
operated by the Cargill Company at Reserve, Louisiana. This
facility was equipped with baghouse collectors and operated at
very low particulate emission levels. Pretest sampling to evalu-
ate^the site indicated that emissions were in the range of 1.27 x
10~ to 1.71 x 10 grams per dry standard cubic meter, as deter-
mined by use of EPA Method 5. The baghouse collector selected
for the test was located on the conveyer used in unloading grain
barges. A stack extension was installed on the baghouse o\itlet
to bring the exhaust gas to ground level. This extension fed into
46
-------
a test section from which the samples were taken. A sketch of
the site complete with stack extension is shown in Figure 3.
III. FIELD TESTING
Pretest preparations required by the methods were carried
out and included all specified equipment calibrations. The
pitot tube calibration was conducted with the full four-probe
assembly in order to assure that no bias to the velocity head
measurements by the four probes would be introduced. The test
was conducted the first two weeks of May 1977, commencing on
May 2 and being completed on May 13. The test plan originally
called for five runs each week. The initial run on May 2 was
terminated prematurely by a plant shutdown caused by equipment
malfunction. An additional five valid runs were conducted the
first week, and five valid runs the second week. Since the par-
ticulate loading at the site was quite low, it was considered
necessary to sample for three hours in order to obtain ca 50
milligrams of particulate. The first run was considered invalid
due to a run time of about 1.5 hours. Other criteria for validity
were passage of initial and final leak checks and isokinetic
sampling within allowable limits of variation. During the course
of the two-week test period, the facility was unloading wheat,
corn, and soybeans, which resulted in considerable variation in
the particulate loading at the baghouse outlet. Specific atten-
tion by test personnel to plant operation was required, since the
unloading operations were intermittent, even though the conveyor
ran continuously. To assure that sufficient particulate matter
was collected, sampling was conducted only when the conveyor was
loaded and was interrupted when the conveyor was running empty.
Only one problem of significance occurred with the test
equipment. One control console developed a leak in the pump
oiler. This console was replaced with a backup unit. During
the sample workup in the laboratory, two other problems more
serious in nature were encountered.
The first involved the Method 17 filters (BGI thimbles).
Prior to the test, the filters were desiccated to constant weig-ht
according to the method (less than 0.5 mg change on two subsequent
weighings). Since the particulate matter is collected on the
outside of the BGI filter, some means of protecting the used
filter to prevent loss of particulate during desiccation and
weighing is required. Each thimble was placed in a glassine
envelope for this purpose. The thimble and envelope was desiccated
to constant weight as a unit. Some difficulties were encountered
in the pretest drying to constant weight. The filter-envelope
combination appeared to pick up moisture fairly rapidly on the
balance pan, even though the weighings were conducted in a con-
stant temperature-constant humidity room. However, after several
47
-------
days the filters were brought to constant weight. When the
filters used in field sampling were desiccated to constant weight,
some were found to weigh less than the pretest tare weights.
Only one unused Method 17 filter was available. This filter wais
desiccated to constant weight over a period of ten days and was;
found to weigh 37.2 mg (0.847 wt %) less than the original tare
weight. It appears that although in the pretest weighing of the
Method 17 filters constant weight was achieved, the filters were
not completely dry. Subsequent investigations showed that the
problem was not due to the filter thimbles but due to the glass-
ine envelopes. These envelopes, intended for photographic negative
storage, have incorporated in them a certain amount of a desiccant
and will dry to a constant weight (less than 0.5 mg change on
two subsequent weights) while still retaining considerable moisture
Continued desiccation will eventually remove this water.
Based upon the weight change of the unused Method 17 fil-
ter, a correction factor (0.9916 x original tare weight) was used
to correct the tare weights of the filters. It is realized that
this is not a particularly satisfactory solution to the problem,
but it is the only means available to try to salvage the data.
The Method 17 particulate catch weights are uncertain, and
quantitative comparisons of particulate loadings by the two
methods are not possible, as will be discussed in the next section.
The second problem concerned the amount of material which
passed through the filtration system and was collected in the
back-half (impinger section) of both the Method 5 and Method 17
determinations. The particulate which passed the filter was
determined by placing the impinger water in a sample bottle along
with an acetone wash of the filter holder back-half, the fritted
glass filter support, the impingers and connecting glassware.
Particulate mass was determined by evaporation of this combined
acetone-water mixture to dryness. After evaporation to dryness,
small pieces of glass from the fritted glass filter support were
observed in most of the samples. As a result, the impinger resi-
due weights are somewhat inflated.
IV. RESULTS
The run parameters for the eleven runs are presented in
Table I. The first run is of questionable value due to the short
sampling period. For Runs 2 through 11, the Method 17 runs are
designated with an asterisk. Runs 2 through 6 were made the first
week. In Table I, two particulate loadings (Cs) were calculated
for the Method 17 runs; under the column headed Cs is the value
calculated on the total particulate catch (Method 17 filter
assembly plus backup filter assembly), while the column headed
Method 17 Cs is calculated using the particulate catch from the
Method 17 filter assembly. The Cs values for Method 5 are calcu-
lated in the normal manner. The within-run agreement between the
48
-------
Method 5 determinations is quite good over the range of particu-
late loadings studied (9-45 mg/m ). Unfortunately, the Method 17
results cannot be compared quantitatively to the Method 5 results
due to the uncertainty in the Method 17 results introduced by
the problem with the thimble tare weights. However, a comparison
of the two particulate loadings calculated for each Method 17
run provides some useful information. The difference in these
two Cs values gives a measure of the amount of particulate which
passed through the Method 17 filter and was collected in the back-
up Method 5 probe, filter and filter holder. An examination of
the data revealed that a fairly constant amount of particulate
was collected in the backup train over the range of particulate
loading studied. The mean.value for particulate collected in the _
backup train was 3.73 mg/m with a standard deviation of 0.56 mg/m ~
The lowest value was 2.52 mg/m and the highest value was 4.17 mg/m
Since the amount of particulate remained relatively constant over
the range of particulate loading studied, the percent error intro-
duced in a Method 17 determination would be greatest for low
particulate loading and decrease as the loading increased. This
apparent increase in filtration efficiency with increased particu-
late loading may reflect the effect of a filter cake buildup on
the filter surface which results in more efficient particulate
retention as the run progresses.
The results of the second week's runs to assess the rela-
tive performance of Method 5 and Method 17 filters located in the
same position in the train {out-of-stack) are inconclusive due
to the uncertainty in the Method 17 results.
The results of the impinger train workup to determine the
filtration efficiency of the entire filter assembly are inconclu-
sive due to contamination of the back-half catch with glass
particles from the Method 5 filter supports.
The additional cleanup procedures before the first run,
between each pair of runs and after the last run were performed
to assess the amount of particulate which remained in the train
after a normal sample recovery. A total of 48 individual train
cleanups were performed during the course of the two-week test.-
The mean value of particulate recovered from the sample nozzle,
probe and filter holder on these cleanups was 4.01 mg with a
standard deviation of 1.9 mg, indicating that very little material
is left in the sample train if good sample recovery procedures
are followed.
The weights of particulate caught in each run and the
location of the catch in the train components are presented in
Table II. The values for impinger train residue for each run
are presented in Table III, and the values for particulate recovered
in the additional train cleanup procedures are presented in Table TV.
49
-------
V. CONCLUSIONS
Despite the fact that a quantitative comparison of
Methods 5 and 17 could not be made, the data from the first week
of the test indicate that a significant amount of particulate
(3.75 mg/m ) passes through the Method 17 filter. The amount of
particulate passing the filter was relatively constant for
particulate loadings in the range of 10-45 mg/m . No informa-
tion on the amount of particulate passing the filter at higher
loadings is available. It is possible that the amount of parti-
culate passing the filter may be site specific in that it could
be affected not only by particulate loading but also by the
physical and chemical properties of the particulate (size distri-
bution, composition, etc.).
The data from the second week of the test did not allow
an assessment of the relative filtration efficiency of the two
filter assemblies operated in the same mode (out-of-stack).
However, review of the impinger train residue values for Runs 7
through 11 does not show gross discrepancies in the Method 17
versus Method 5 results. It appears possible that the Method 5
probe acts as a precollector and has an effect on filter efficiency.
These conclusions are necessarily speculative due to the
uncertainty in the Method 17 data. Further investigations are
needed to confirm or refute these tentative findings.
One observation of importance in the use of the BGI
thimble filters should be noted. The thimbles were seated in the
filter holder merely by friction fit. No securing device was
used to hold the filters in place. There was sufficient varia-
tion in the inside diameter of the thimbles in the lot used
that the fit of the filter in the filter holder varied from snug
to loose. There is firm evidence on one run (7C) that consider-
able particulate bypassed the filter and was collected in the
impinger train, which was very milky in appearance at the end
of the run. In this run, 103 mg of particulate was collected
by the probe and filter and 48.2 mg by the impingers, compared
to 174.6 mg by the probe and filter and only 8.1 mg by the impingers
on Run 7A, the other Method 17 determination in that run. Some
means of assuring that particulate blow-by does not occur should
be incorporated in the use of BGI filters.
50
-------
Method
C
s
W
M
O
4J
O U
H
Nozzle
No.
m
"M
$
0
CO
s
.2
>e
1
1
0)
1
<
CJ^
(^
ft
§ o*
PJ S
co VD r** ^*
r» r- CM CM
r^ en ^* oo
CM i- r- T-
vo in ro m
r^ r- i- O
T * r** vo ^*
in o ^> f
OOOO
CM CO r- ^
VO VO VO VD
O O O O
0 0 0 0
ro ro ro ro
vo Is- r> CM
if ID *f \Q
CM CM CM CM
OOOO
oooo
in in in in
o> en en en
VO VO T- r-
^f ^f ro ro
oooo
CN CM CM CM
T^ in ro en
0 00 VD Cn
T- CM vo en
CM 00 CT> 00*
oo r^ r** r**
vo o ro in
O O r- «-
in in in in
^J1 ^* vo vD
0 0 0 0
in m in m
ri- ro o vo
T- i- T- ro
in m ^r CM
CM CM CM CM
T- T- en en
ro ro CM CM
oooo
< « u a
ro <*
CO vO
r- ro
^j» en co vo
ro r** ^d1 T
en r- T- vo
IT- T
in ^* ^D ^*
T- T- r- O
en CM en in in
oooo
* *
*«! m o a
CM CN CM CM
en
1 CN
vo m r-
cn r« in
1 . . .
1 en vo en
in en oo r-
I- «- T- O
^- ^ vD o
^d1 00 ^J* ro
o en o o
r- CM CO f
vo vo vo vo
oooo
co ro ro ro
*r oooo
CM CM CM CM CM
o oooo
0 OOOO
oooo
oo oo co oo
CM CN CM CM
r- r- r- r-
CN CM CM CM
00 CM 00 m
VO VO CM CM
vo m m o
en ro en r^
in in in m
T- rj- in 0
CM ro CM ro
\r r- r- r~
m in in m
oooo
00 03 00° 00
m m m m
** v- CN en
in ro CM r-
r- 03 co in
CM CM CM CM
VD VD VD VD
OOOO
* *
< m o a
ro ro ro n
O CM
t T-
T- r~ oo vo
o ro o ^i1
CT> <* CO vO
sp o o o
i- CM CM T-
r~ in 03 03
ro en CN ro
o en o o
T- CM ro «*
co oo oo oo
oooo
ro co ro ro
O 00 VD O CM
CM T- T- I- Y-
o oooo
o oooo
oooo
oo oo oo oo
en en en en
*
CM CM CM CM
M1 03 VD VO
in vo CM T-
r* CM VD ro
VD ro vO in
in in in in
CM o r» vD
in in in in
ro ro ro co
oooo
in m in in
in in m m
in r- r- o
OD O f» O
vo co vo m
CN CM CM CM
CM CN CM CM
OOOO
* *
< CP U O
00000300
oooo
oooo cncncT*c?>
rOCOrOfO (NCMCMCN
oooo oooo
00 CD CO CO CO CO CO CO
00000000 CMCMCMCM
CMCMCMCM r-r-^-T-
ro p* o ro 00 ro *~ vo
voorocn vOCMr~oo
CN«-CNVD oorocom
oo^t^m roinocn
tnminin *f ^ *f r*>
co CM vo CM
m in
inmmin
inmmin
ininmin t-^-r-r-
cncncncn r-r~r~r«
^ji^^^ mininin
mininm mmmm
vo in CM n
r~ oo r- m
m vo o CM
r~ ro en <*
CM <* r- o
CM CM CM CM
CM CM CM CM
CNCMCMCM OOOO
O O O O
OOOO
-------
d
o to
to
rj
rj
U U
H
0)
N 0
r
to
ft
II
,q
ft
CD
to
s
in
"»>
1
1
CO
w
a
PU
c
3 0
Qi K
en t** r*~ i^
en vo co *-
ro ro in 1/1
r- ** co en
o t^ cr> vo
^* ro *\l* ro
o o o o
T CN ro ^l1
en en en en
en
o o o o
O 0 O O
ro T- ro (N
in CN r~ ro
CN CN en CN
r-- 'd' in in
^ co en m
CN VO CN CC
*d* vO vO vc
in in in in
en en en en
m in tn in
co en CN CN
T- in en CN
ro ro CN r-
CM CN CN CN
CN CM CN OJ
o o o a
* *
< pa o c
en en en cr
vo vo co m
co vo r^ vo
^f co in vo
CM T- CN t-
r~ en 'S1 vo
CN -31 CO r-
m »- in ^"
O 0 0 O
r CN CO ^3*
VD VO VO VO
t V- T F-
o o o o
ro ro ro ro
r-. t- CN vD r-
CN o en en en
o o o o o
o o o o o
0000
CO CO CO CO
o o o o
*
o a o o
00 ^* ^ l-f">
in r~ o ro
en T- ^ vo
0 CO 0 CO
m
0 00
o o o
o o
CO CO
CM CM
en en
in M1
in ro
«- o
T- f.
*3- VO
^r vo
in in
o o
.
r^ r^
in in
vo vo
vo en
CN CN
in m
r- T-
O O
*
1 < CQ
) r- *
CO *sf
co m
r- ro
ro ^
"d1 vD
ro ^*
ro ro
o o
ro ^f
o o
CN CN
O O
ro ro
vO CO
0 0
o o
o o
CO CO
CN CM
en en
en co
m CN
vo vo
o o
in ro
T-^ VO
vo vo
in m
0 0
in m
CN O
CO CO
o en
CN t-
m in
0 0
*
O Q
in
ro
co
vO
II
CNJ
4J
in
in
ro
co
vO
II
05
C
3
S-l
t)
en
en
vo
II
CM
4-1
M-l
CO
VO
II
CM
CO
o
II
O
rj>
o
ff
C
-rH
C
-H
<£>
m
£
to
n:
o
CN
X
c
H
SB
cu
52
-------
H EH
M.5
I
8
C M
H C1J
BJ -P
M
0) 4-1
tP O
n
14
0)
O
(3 rH
ffl X
o fc
h CD
p. -p
m -H
a
o
r^ JH
E T- a)
n
0)
CO
00 VD CO T-
oi TJ r~ r-
T- *t
mvooom
tj- o o oo
^ f CN ,-
cs 01 VD ro
CM r- T- T-
CT> vo in
r- t- r- CN
in
cn i-
o o
CN ^ CN en
n
n
I I
i m i o
i i
I & i
i i
en o n en
. i . . i . i
oo t ro to i m i
T- I T- T- I T- I
I I
| O | CM
I CN I CN
ro r>
i I
i o i m
i CM i
CM
m
cr>
en
i ID i CN
|CN|M>
ro
t£>
VD
cn
cn
cn i CM
oo i o
00
nmom
CD rs VD 10
10 CM en
CM CN
ro
CM
O CM
I CM
CN
m
ocncs
ro i ro
in o in o
o i o
« U P
CM
rn
in
vo
53
-------
^l
e
-*
*"» W
*O O
** E~*
SS
o
HS
n3
HB
m H
-t EH
.Q PH
P O C
Q) -H
S r* ft
0 g
flj H
W
(3
a> o
N 4-) 0)
N c -d
O O rH
S3 H 0
CM K
in
13 V)
>d -P
rC ^5 rH
^ 0 -H
S P^
13
0) !H
_Q d)
r~ >^ rH
T- P-l O
ffi
>d us
O M
,C <1) (U
P H -P
O -P
JZ rH
*J -H
Q) CM
2
C
(0
EH
c
s
t i CM i iinit ^ i r* i imico
. | . | | . | . . | . | | . | .
coiooi icoim r^ir^i I^IVD
^
^r^i~*co CMCOOCO
I . i . i . i i . i . . i . i
I r- i o coicni loico OICMI
\ r- \ *~ | | |T-| T-lr-l
mrovor- rovomvo
1 . 1 . 1 . 1 1 . 1 . . 1 . 1
ICf>|r- r- I T- I IP»I» OlrOi
ICMICO roiroi ICMiro CMICMI
vor~ r~ocMCM mr-
|.| |.|. !! 1*1*
inioi ivoir-- P» i CM i imivo
^* 1 CO | ICOIr- CMICMI I t- I *-
O ro O O ^3*
0 00 *
cnioji imim coi-i i i
cMir^i imicri mi^i ICMICO
T-I i i i imi ivoim
4J
cn
Q)
EH
CO
cMCMcnr- t~-incocM
t^j .... ....
Q) IV0I10 ^" 1 CT> 1 ICMICTl in I VO I
a) i «* i ^r r-ioi looir- coicoi
S|r-|«-t-|T-| 1 1 1 1
d
C
O
u
W CD
P-H '
54
VO 1 CO 1
* 1 * 1
r- i co i
cn *
I I
1 CM I CO
1 T- 1 t-
in CM
i i
1 ro 1 *-
1 ro 1 CM
ro O
1 * 1
r- 1 r- 1
CM I T- I
o r~
i i
CM i ro i
T- 1 ro |
T T*~
T- CM
.
1 VO | CM
1 ro I m
1 *- 1 t-
ril CO U C3
s
-------
It
c
^1
t
^J
PC
*J
<
*t
m
3
» n
T~
*
»*
)
VO
*
r~
cn
*~
*
n.
O4
OJ
Q
ro
U
n
pj n
i~> n
H
H
K
<
H S ro
rH (4*
*
^
GO
r-
*
n
o
CM
*
CM
O
a*
*
T"
^f
M1
0) E-i
H
.Q K
38
g CM
M
a
s
n
04
rt
04
*
T*
r«
T
«
CO
r^
T"~
vo
v
OJ
«
CO
O4
CO
Q
«
CJ
00
<
t
*
*
0
«~
*
GO
VO
*
t
O
*T
*
ro
vo*
T~
D
CO
U
co
m
CO
<
CO
V
in
*
n
cn
in
CO
*
r^
CO
Q
r*
°u
t^
CQ
r-
<
r>-
*
r-
*
o
^«
CM
00
^*
*
t '
t
Q
VO
O
VO
S
S
*
<*
*
OJ
t"
*
m
f«
T
K
VO
*
OJ
^
*-^
S
o
m
m
in
<
in
*
m
in
04
*
CO
cn
4c
VO
VO
^~
*
in
,
C
id
g
C
H
f*,
CQ
M id
cu S
4J
rH CO
H ti
*4 0
in co
o
d id
3,8
CO +J
So
(U 4->
P CO
CQ
d m
C rH
-r| CJ1
f*
co VM
A 0
P CO
-rl +J
rl C
HH cu
CO §)
n id
Id rl
rH
C CU
rl +J
A 3
S3
*£
«4-l C
O 0
0
P
rH O
g*
co -d
M CU
>
id tj
CO
CU CO
*§
Ss
id CQ
& id
S
CO
O -P
3 -H
rH rl
rH
o
5
5
CO
,c
+j
m
O
d
C
*^
^>
4J
H
MH
rl
O
o
a
id
5
(0
3
>^
rH
^Q
s
S
0,
cu
3
S
CQ
CO
VI
CU
rH
A
rl
CQ
-rl
>
id
1
U
r^-
C
3
&
W
S
o
c
H
id
*
c
H
rl
CO
T(
rH
£
M
CU
+J
rH
H
iw
in
d
o
5
0)
S
O
K
1
1
-rl
tn
e
o
c
id
,c
jj
+^
CQ
CO
0) CO
rH rH
CU Id
a g
co a.
M
Q) >,
d rH
rH .C
o m
*3
n
CO CO
P -rl
rH
rl *J
B
§
(H
M
I)
3
U
g
U
55
-------
ft
£3
§i
H
a
> u
H
*
in «
r-
CM
r- <
r-
CM
VD
M*
**
CM
vO
ro
r--
*
T
Q
CM
O
CM
pq
CM
ft.
CM
CO P
VO
tf U
VO
o pq
. vD
m
in <
VO
O
c\
in
00
VO
ro
CM
cn
,_
P
U
pq
ri
T~
00 C
IT
<*
<* C.
if
r~-
*- p
u
ro
0 *
LI
ro
CM
a1
) en
i
in
3 o
T .
ro
H CM
n .
m
P
o
O
o
m
o
<
0
O us
-
* "H
CO -P
<\
(0
0)
r- u
r*
>H
m (0
4J
(Q
c
"^1
* SK
P~l
VD C
m
£i
C ^5
VO HO)
' g
00 i>
to 0)
0) C
4J O
rH #
H 0)
P
r-
U
r
*~
PQ
r
<
T
*"
VO -i G
5-H *W O
i i T i y
2 P4 S
rfj W Q1
W W tO (Ti
3 13 n3 S-i
P
o
T
U
o
^
m
o
<
o
VO U 0 P
. £ »-
CM 1 -P
nj ^
a)
Q) W
rQ ^)
Q
a
en
C
c
p
a
e
C
P
o
ro
) CO U
\ o
'
3 CO 03
> o
^f
< CO <
n o
i
r- >,
rt w
^ rrt
^ S
CO
0) 4J
^ ?
r- r-l to
(S M-l
CN >
CO
^ CO
tJ1 n3
H i-H
^ ffi cr
CO
w
EH
O
«- 2
cn *
56
-------
a
o
A
4J
o
s
M
a
+j
Q,
D
.*
O
(3
PQ
in
V)
'O O
O -P
X .H
-p -H
0) PL,
3
01
57
-------
§
C/3
M
o
u
A.
H
£3
Oi
w
CO
8
tN
0)
M
3
Cn
H
58
-------
0)
§
*
w
10
w
B
o
H
M
CQ
CO
I!
59
-------
TECHNICAL REPORT DATA
ead htumctions on llie reverse bciorf comnletins)
"EPAT600/4-80-022
2.
3. RECIPIENT'S ACCESSION-NO.
4 TITLE A\O SLSTiTLc
COMPARATIVE TESTING OF EPA METHODS 5 AND 17
AT NONMETALLIC MINERAL PLANTS
5. REPORT DATE
April 1980
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
William J. Mitchell, M. Rodney Midgett and C. Bruffey
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Quality Assurance Division
Environmental Monitoring Systems Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
10. PROGRAM ELEMENT NO.
11. CONTRWCT/cHANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Monitoring Systems Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA 600/08
15. SUPPLEMENTARY NOTES
To be published as Environmental Monitoring Series report.
16. ABSTRACT
Comparative tests between EPA Reference Method 5 and commercially-available
in-stack filters (EPA Reference Method 17) showed that in-stack filters can be
acceptable substitutes for the Reference Method at nonmetallic mineral plants.
These comparative tests were conducted in the exhaust' stacks from a clay calciner,
a feldspar grinding mill and a limestone crusher. Particulate concentrations
encountered ranged from 30 to more than 300 mg/m3. Laboratory studies using
feldspar dust gave results identical to the field study results. Insignificant
amounts of material were found downstream of the in-stack filters. These results
are compared to similar results obtained at a grain elevator and a kraft pulp mill.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
COSATI Field/Group
Particulate
In-stack filter, out-of-stack filter
EPA Method 17
EPA Method 5
43F
68A
'.3. jljTRlsUTION STATEMENT
Release to Public
19 SECURITY CLASS f Tins Report)
Unclassified
21. NO. OF PAGES
fin
20. SECURITY CLASS (Tins page I
Unr.1 as^i f "ipH
22. PRICE
EPA Form 2220-1 (9-73)
-------