EPA
United States
Environmental Protection
Agency
Office of Air Quality EMB Report
Planning and Standards 84-GLS-10
Research Triangle Park NC 27711 October 1984
Air
Neshap — Glass
Manufacturing
Arsenic
Emission Test Report
Corning Glass Works
Martinsburg,
West Virginia
-------
EMISSION TEST REPORT
METHOD DEVELOPMENT AND TESTING FOR
ARSENIC FROM GLASS PLANTS
Corning Glass Works
Martinsburg, West Virginia
ESED PN: 83/20
EMB No. 84-GLS-10
by
PEI Associates, Inc.
(formerly PEDCo Environmental, Inc.)
11499 Chester Road
P.O. Box 46100
Cincinnati, Ohio 45246-0100
Contract No. 68-02-3849
Work Assignment No. 11
PN 3615-11
Task Manager
Mr. Daniel Bivins
Emission Standards and Engineering Division
U.S. ENVIRONMENTAL PROTECTION AGENCY
EMISSION MEASUREMENT BRANCH
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
February 1985
-------
DISCLAIMER
This report was furnished to the U.S. Environmental Protection Agency,
Emission Measurement Branch, by PEI Associates, Inc., Cincinnati, Ohio, in
fulfillment of Contract No. 68-02-3849, Work Assignment No. 11. Its contents
are reproduced herein as received from PEI. The opinions, findings, and
conclusions are those of the authors and not necessarily those of the EPA.
Mention of company or product names does not constitute endorsement or recom-
mendation for use.
-------
CONTENTS
Page
Figures iv
Tables v
Acknowledgment vi
1. Introduction 1-1
2. Process Description 2-1
2.1 Sampling and Analytical Protocol 2-1
2.2 Test Results—Quad and Reference Train 2-3
3. Project Quality Assurance 3-1
4. Sampling Location and Test Methods 4-1
4.1 Sampling Location 4-1
4.2 Sampling and Analytical Procedures 4-7
5. Process Description 5-1
Appendices
A Computer Printouts and Example Calculations A-l
B Field Data B-l
C Laboratory Data C-l
D Sampling and Analytical Procedures D-l
E Equipment Calibration Procedures and Results E-l
F Quality Assurance Element Finder F-l
G Project Participants and Sampling Log G-l
-------
FIGURES
Number Page
2-1 Quad Train System for Elevated Temperature Tests 2-2
3-1 Pre-Test Audit Report: Dry Gas Meter by Critical Orifice
(Meter Box FB-8, Train A) 3-6
3-2 Pre-Test Audit Report: Dry Gas Meter by Critical Orifice
(Meter Box FB-3, Train B) 3-7
3-3 Pre-Test Audit Report: Dry Gas Meter by Critical Orifice
(Meter Box FB-5, Train C) 3-8
3-4 Pre-Test Audit Report: Dry Gas Meter by Critical Orifice
(Meter Box FB-1, Train D) 3-9
3-5 Pre-Test Audit Report: Dry Gas Meter by Critical Orifice
(Meter Box FB-11, Reference Train) 3-10
3-6 Onsite Audit Data Sheet 3-11
3-7 Onsite Audit Data Sheet 3-12
3-8 Onsite Audit Data Sheet 3-13
3-9 Example of Onsite Calibration Data Sheet 3-14
4-1 Quad Train System for Elevated Temperature Tests 4-2
4-2 Four-Train Sampling System Showing Nozzle, Pi tot Tube,
and Thermocouple Position 4-3
4-3 Sampling Location (Plan View) 4-4
4-4 Sampling Location (Elevation) 4-5
-------
TABLES
Number Page
2-1 Sampling Matrix 2-4
2-2 Summary of Sampling Conditions 2-5
2-3 Summary of Arsenic Analytical Results - Quad and Reference
Train Runs 2-7
2-4 Statistical Data for Grouped Runs (Total Train) 2-10
2-5 Within-Run Statistical Data for Paired Quad Runs (Total
Train Basis) 2-11
2-6 Statistical Data of Filterable and Condensible Arsenic for
Grouped Quad Runs 2-12
3-1 Field Equipment Calibration 3-3
3-2 Arsenic Blank Data 3-16
3-3 Arsenic Laboratory Reagent Blank Data 3-17
3-4 Linear Regression Data (Flame) 3-18
3-5 Arsenic Standards Addition Results 3-20
3-6 Duplicate Analysis Data (Flame) 3-22
3-7 Duplicate Analysis Data (Furnace) 3-23
-------
ACKNOWLEDGMENT
This test program was conducted for the Emission Standards and Engineer-
ing Division of the EPA Office of Air Quality Planning and Standards.
Mr. Daniel Bivins, Emission Measurement Branch (EMB) Task Manager, pro-
vided overall project coordination and guidance and observed the test program.
Mr. Robert Dykes, representing Radian Corporation (an EPA contractor) moni-
tored process operation throughout the test period. Mr. Charles Bruffey was
the PEI Project Manager. Principal authors were Messrs. Charles Bruffey and
Thomas Wagner.
VI
-------
SECTION 1
INTRODUCTION
Arsenic is listed as a hazardous air pollutant under Section 112 of the
Clean Air Act (National Emission Standards for Hazardous Air Pollutants). To
protect public health from unreasonable risks associated with exposure to
airborne arsenic, the U.S. Environmental Protection Agency (EPA) has developed
standards to decrease inorganic arsenic emissions from the following source
categories: high-arsenic primary copper smelters, low-arsenic primary copper
smelters, and glass manufacturing plants.
To support the standards review process and provide additional arsenic
emissions data from glass manufacturing facilities, PEI Associates, Inc.,
(under contract to the Emission Standards and Engineering Division - Emission
Measurement Branch) performed a series of atmospheric emission tests on a
glass melting furnace at the Corning Glass Works facility in Martinsburg, West
Virginia, from October 15 through 17, 1984. These tests were conducted to
determine if the quantity of particulate arsenic as measured by EPA Reference
Method 108 varies with flue gas and sample temperatures. Reference Method
108* provides total arsenic results (particulate plus gaseous fraction).
A total of five quad train runs (see Figure 2-1) were conducted using
draft Method 108 procedures except that probe and filter temperatures were
elevated to 177° and 260°C (350° and 500°F) in order to evaluate the effects
40 CFR 61, Appendix B, Method 108, July 1984.
1-1
-------
of sample temperature on arsenic distribution in the sampling train. During
the quad runs, a single Method 108 sampling train with probe and filter tem-
perature of 121°C (250°F) was run for reference purposes.
Section 2 summarizes and discusses the test results; Section 3 addresses
quality assurance considerations specific to this project; Section 4 describes
the sampling locations and test procedures; and Section 5 describes source
operation. Appendix A presents sample calculations and computer printouts;
Appendices B and C contain the field data sheets and laboratory analytical
results, respectively; Appendix D details the sampling and analytical proce-
dures; Appendix E summarizes equipment calibration procedures and results;
Appendix F is a quality assurance element finder; and Appendix 6 is a list of
project participants and a sampling log.
1-2
-------
SECTION 2
SUMMARY AND DISCUSSION OF TEST RESULTS
2.1 SAMPLING AND ANALYTICAL PROTOCOL
A four-train (quad) sampling system was used to collect samples in the
rectangular breeching connecting the furnace to the exit stack. This system
allows four trains to sample simultaneously at essentially a single point in
the duct (see Section 4). Therefore, this sampling approach reduces the
effect of variations in the velocity and particulate profiles on the sampling
results. It also permits a statistically significant number of samples to be
taken in a short amount of time.
The quad runs conducted were designed to determine if the quantity of
filterable arsenic collected varies with sampling train temperature. For
comparative purposes, two of the trains were heated to approximately 177°C
(350°F) and two trains were heated to approximately 260°C (500°F). At each
temperature, one train possessed a backup filter heated to 121°C (250°F) prior
to the impinger section.
Figure 2-1 depicts the quad train configuration used in these tests.
Individual train components were recovered and separately analyzed for arsenic
to evaluate the distribution of arsenic in the sampling train.
During these runs, a single Method 108 sampling system (designated RT)
(121°C) was run for reference purposes. The reference train was located on
the opposite side of the breeching and as close as possible to the quad probe
system.
2-1
-------
BACKUP
METHOD 5
FILTER
(12TC)
IMPINGER
TRAINS
(204°C) (288°C)
HEAT BOXES
BACKUP
•METHOD 5
FILTER
(121*C)
Ar
B:
FRONT VIEW
B A D C
UJ UJ UJ UJ
CD CD CD CD
oooo
oc.cc a: oc.
0.0. o. o.
oD oA
oc oB
BACK VIEW
Figure 2-1. Quad train system for elevated temperature tests,
2-2
-------
In each train, the probe and filter temperatures were set at a predeter-
mined temperature and monitored using multiterminal digital indicators with
thermocouple leads located in each probe and immediately behind the Method 5
filter frits. Table 2-1 presents the sampling matrix followed in this test
program.
2.2 TEST RESULTS—QUAD AND REFERENCE TRAIN
Table 2-2 summarizes sampling conditions for the quad and reference train
test runs. Table 2-3 summarizes the arsenic analytical results by sample
fraction. Sample volumes are expressed in dry standard cubic meters (dsm3),
arsenic weights in milligrams (mg), and arsenic concentrations in milligrams
per dry standard cubic meter (mg/dsm3).
The filterable or front-half arsenic reported in Table 2-3 represents
that material collected on the front filters and in the sampling probes which
were maintained at 177° and 260°C for the quad runs and 121°C for the refer-
ence train runs. The condensible or back-half arsenic represents that mate-
rial which passed through the front filter and was collected in either the
connecting glassware, backup filter, or impinger section of the sampling
train.
Sample volumes were consistent and ranged between 1.19 and 1.58 dsm3 for
the quad runs and 1.31. to 1.62 dsm3 for the reference train runs. Isokinetic
sampling rates ranged from 93.5 to 103.1 percent, which is within the accept-
able range of 90 to 110 percent.
The desired temperature for paired Trains A and B was 177°C (350°F) and
for paired Trains C and D, 260°C (500°F). The reported probe and filter
temperatures represent average values determined from data recorded on the
2-3
-------
TABLE 2-1. SAMPLING MATRIX
Quad
Run No.
1
Sample
ID
1A
IB
1C
ID
Reference train
2
2A
2B
2C
2D
Reference train
3
3A
3B
3C
3D
Reference train
4
4A
4B
4C
4D
Reference train
5
5A
5B
5C
5D
Reference train
Method 108 sample temperatures3
177°C (350°F)
X (BU)
X
X (BU)
X
X (BU)
X
X (BU)
X
X (BU)
X
260°C (500°F)
X
X (BU)
X
X (BU)
X
X (BU)
X
X (BU)
X
X (BU)
Reference Method 108
train at 121°C (250°F)
X
X
X
X
X
The designation BU indicates a backup filter maintained at 121°C (250°F)
was located prior to the impinger section of the sampling train. Sampling
train components (i.e., probe, filter(s), impingers) were recovered and
analyzed separately.
2-4
-------
TABLE 2-2. SUMMARY OF SAMPLING CONDITIONS
Run
No.
1A
1Ba
1C3
ID
RT-1
2A
2B
2C
2D
RT-2b
3A
3B
3C
3D
RT-3
Sampling
type
Modified
Method 108
Method 108
Modified
Method 108
Method 108
Modified
Method 108
Method 108
Date (1984)
and
time (24-h)
10/15
1315-1451
10/15
1315-1445
10/16
0952-1122
10/16
0952-1122
10/16
1344-1514
10/16
1503-1633
Mete red
volume,
dsm3
1.58
1.55
1.46
1.62
1.46
1.44
1.22
1.30
1.31
1.53
1.52
1.32
1.41
1.53
Isoki-
netic, %
101.4
99.9
103.1
100.4
98.8
101.1
95.1
100.0
81.5
99.0
98.8
93.5
99.9
99.1
Mois-
ture, %
9.95
9.92
10.07
8.15
10.33
11.69
10.92
10.17
7.64
10.27
10.48
10.69
10.40
10.25
Sampling conditions
Flue gas
tempera-
ture, °C
393
390
388
392
395
395
394
395
386
380
378
379
375
369
Probe
tempera-
ture, °C
177
177
255
111
175
186
266
259
98
177
181
259
263
128
Filter
tempera-
ture, °C
183
178
266
122
177
176
262
270
129
180
178
268
275
118
Backup fil-
ter tempera-
ture, °C
128
NA
124
NA
125
NA
NA
118
NA
121
NA
NA
124
NA
ro
i
en
(continued)
-------
TABLE 2-2 (continued)
Run
No.
4A
48
4C
40
RT-4
5A
58
5C
5D
RT-5
Sampling
type
Modified
Method 108
Method 108
Modified
Method 108
Method 108
Date (1984)
and
time (24-h)
10/17
0840-1010
10/17
0840-1010
10/17
1253-1428
10/17
1243-1428
Metered
volume,
dsm3
1.43
1.41
1.19
1.27
1.61
1.57
1.55
1.36
1.43
1.57
Isoki-
netic, %
98.0
99.2
94.2
99.9
97.9
99.3
98.6
94.2
99.8
98.2
Mois-
ture, %
9.83
9.84
10.00
10.23
9.55
9.94
10.13
10.05
9.95
9.57
Sampling conditions
Flue gas
tempera-
ture, °C
375
380
376
379
382
365
370
367
370
370
Probe
tempera-
ture, °C
181
181
268
266
146
174
179
259
261
121
Filter
tempera-
ture, °C
177
175
263
270
115
181
180
269
275
119
Backup fil-
ter tempera-
ture, °C
119
NA
NA
124
NA
121
NA
NA
124
NA
ro
i
en
Run No. 1C is void due to excessive post-test leak rate.
Run No. RT-2 is void due to a nonisokinetic sample condition.
NA = Not appl icable.
-------
ro
TABLE 2-3. SUMMARY OF ARSENIC ANALYTICAL RESULTS
(QUAD AND REFERENCE TRAIN RUNS)
Run
No.
1A
IB
1C8
10
RT-1
2A
2B
2C
20
RT-2
3A
3B
3C
30
RT-3
4A
4B
4C
40
RT-4
5A
SB
5C
SO
RT-5
Sample
volume,
dim'
1.56
1.55
1.46
1.62
1.46
1.44
1.22
1.30
1.31
1.53
1.5?
1.32
1.41
1.53
1.43
1.41
1.19
1.27
1.61
1.57
1.55
1.36
1.43
1.57
Arsentc sample weights, ng
Filterable
Probe
rinse
10.9
21.7
15.5
13.0
22.9
20.4
16.9
10.3
8.2
14.1
22.0
6.1
22.5
40.3
32.3
22.0
17.7
9.3
38.4
19.4
19.4
8.4
17.1
41.6
Front
filter
33.8
33.4
32.2
28.9
30. B
29.2
24.6
25.7
17.8
35.9
33.6
33.4
31.6
35.3
31.3
30. B
26.9
28.4
37.4
34.4
32.1
30.5
29.4
33. B
Total
front
half
44.7
55.1
47.7
41.9
53.7
49.6
41.5
36.0
26.0
50.0
55.6
39.5
54.1
75.6
63.6
52.8
44.6
37.7
75.8
53.8
51.7
38.9
46.5
75.4
Back-half (cond«nsibles)
Glass
connector
56.6
NA
34.1
NA
19.5
NA
NA
8.6
NA
84.9
NA
NA
53.0
NA
62.5
NA
NA
27.5
NA
84.4
NA
NA
54.4
NA
Backup
filter
5.5
NA
26.3
NA
19.1
NA
NA
26.5
NA
30.3
NA
NA
34.9
NA
21.8
NA
NA
3.3
NA
35.1
NA
NA
42.1
NA
Impinger
1
34.3
131.4
17.4
75.9
12.5
90.4
60.1
7.9
59.4
16.6
130.3
77.9
12.7
108.5
7.8
92.7
69.9
24.2
92.1
8.5
121.0
104.5
10.1
95.5
2
O.B9
2.5
0.42
O.B7
0.30
1.8
3.9
0.32
1.61
0.31
4.5
1.60
0.60
2.1
0.20
1.08
1.18
0.86
1.92
0.14
4.65
1.74
0.40
2.55
3 and 4
0.10
0.26
0.05
0.41
0.045
0.20
0.82
0.09
0.09
0.03
0.30
1.01
0.05
0.35
0.02
0.22
1.02
0.09
0.10
0.48
0.07
0.96
0.05
0.17
TolaT
back
half
97.4
134.3
78.3
77.2
51.4
92.4
64.8
43.4
61.1
132.1
135.1
80.7
101.3
110.95
92.3
94.0
72.1
55.95
94.1
128.6
125.7
107.2
107.1
98.2
Concentration, mq/dsm'
Front
half
28.3
35.5
32.7
25.9
36.8
34.4
34.0
27.7
19.8
32.7
36.6
29.9
38.4
49.4
44.5
37.4
37.5
29.7
47.1
34.3
33.4
25.1
32.5
48.0
Back
half
61.6
86.6
53.6
47.6
35.2
64.2
53.1
33.4
46.6
66.3
88.9
61.1
71.8
72.2
64.5
66.7
60.6
44.0
58.4
81.9
81.1
69.2
74.9
62.5
Total
train
89.9
122.2
86.3
73.5
72.0
98.6
87.1
61.1
66.4
119.0
125.5
91.0
110.2
121.6
109.0
104.1
98.1
73.7
105.5
116.2
114.5
94.3
107.4
110.5
Filterable
arsenic,
S of total
31.5
29.1
37.9
35.2
51.1
34.9
39.0
45.3
29.9
27.5
29.2
32.9
34.8
40.6
40.8
36.0
38.2
40.3
44.6
29.5
29.1
26.6
30.3
43.4
Run 1C Is void due to excessive post-test leak rate.
NA = Not applicable.
-------
field data sheets. As shown, filter temperatures for Trains A and B ranged
from 175° to 183°C and the probe temperatures ranged between 174° and 186°C.
In Trains C and D, the filter temperatures ranged between 262° and 275°C and
the probe temperatures ranged between 255° and 268°C. The backup filter
temperatures for each quad run, Trains A and D, ranged between 118° and 128°C.
The reference train probe temperatures ranged between 98° and 146°C and the
filter temperatures ranged between 115° and 129°C.
The moisture content of the flue gas was generally consistent for each
run and ranged between 9 and 11 percent. Flue gas temperatures ranged between
367° and 393°C during the test program. As shown in Table 2-2, the flue gas
moisture content determined from the reference train for Tests 1 and 2 is at
least 20 percent lower than the corresponding quad train moisture data. This
was the result of a leakage problem that developed in the reference train
during these runs. This problem was not detected during the tests because the
sampling train could not be thoroughly leak checked according to the Method
108 procedure. As a result of the geometric configuration of the breeching
and the location of scaffolding near the test port, the sampling probe was
first inserted into the duct and then connected to the Method 108 sample box
containing the heated filter and impingers. Each component (probe and sample
box) was leak checked separately before and after each test. Because neither
sampling train component experienced a leakage problem during these runs, the
leak must have occurred at the probe front-filter connection. Therefore, the
reference train arsenic results for Runs 1 and 2 are biased low; the magnitude
of which is unknown.
As shown in Table 2-3, arsenic sample weights are reported separately for
each sample fraction analyzed. Sample concentrations are also reported on a
2-8
-------
filterable, condensible, and total train basis. The front filter weight
includes results for both the NaOH extract and the Parr bomb (HF/HNO^) ex-
tract. The Parr bomb extract results constituted less than 1 percent of the
total arsenic on the front filter.
Arsenic was found throughout each sample train; the filterable or front-
half arsenic constituted between 28 and 51 percent of the total arsenic col-
lected in the 177°C quad trains (A and B) and between 26 and 45 percent of the
total arsenic collected in the 260°C quad trains (C and D). In each indivi-
dual quad run, except Train 2A, more than 50 percent of the total arsenic
collected was found in the back half of the quad sampling trains.
This same trend was observed in the reference train tests, although the
leakage problems associated with Runs RT-1 and 2 tend to distort comparisons
between these data and the corresponding quad train results. The percentage
of filterable arsenic found in the reference train ranged between 41 and 45
percent for Runs 3 through 5 compared with a range of 26 to 41 percent and an
overall average of 33 percent for the corresponding quad runs. Data from Runs
3 through 5 suggest that a greater percentage of filterable arsenic is col-
lected at 121°C than at 177° or 260°C. More than 50 percent of the total
arsenic measured, however, was collected in the back half of the sampling
trains regardless of sample temperature.
Tables 2-4 through 2-6 present statistical data for the quad runs on both
a total train and filterable/condensible basis. The mean arsenic concentra-
tion and standard deviation for each set of runs are presented along with the
coefficient of variation (CV), which is the standard deviation expressed as a
percent of the group mean.
2-9
-------
TABLE 2-4. STATISTICAL DATA FOR GROUPED RUNS
(TOTAL TRAIN)
Quad Run
No.
1A
1Ba
1C3
ID
2A
2B
2C
2D
3A
3B
3C
3D
4A
4B
4C
4D
5A
5B
5C
5D
Overall
means
Individual run
value, mg/dsm3
89.9
122.2
86.3
72.0
98.6
87.1
61.1
119.0
125.5
91.0
110.2
109.0
104.1
98.1
73.7
116.2
114.5
94.3
107.4
Group mean
X, mg/dsm3
99.5
79.7
111.4
96.2
108.1
99. Oe
a,C
mg/dsm3
19.8
16.5
15.0
15.7
10.0
15. 7f
CV,d
%
19.9
20.7
13.4
16.3
9.2
15. 9g
Run 1C was voided due to excessive post-test leak rate.
Mean concentration.
Within-run standard deviation with N-l weighting for sampling data.
Within-run coefficient of variation is the standard deviation ex-
pressed as a percent of the mean concentration.
"Simple averages of tabulated data.
Pooled standard deviation;
2-10
-------
TABLE 2-5. WITHIN-RUN STATISTICAL DATA FOR PAIRED QUAD RUNS
(TOTAL TRAIN BASIS)
Run
No.
1A
IB
1C
ID
2A
2B
2C
2D
3A
3B
3C
3D
4A
4B
4C
4D
5A
5B
5C
5D
Desired
sampling tem-
perature, °C
177
177
260
260
177
177
260
260
177
177
260
260
177
177
260
260
177
177
260
260
Individual
run value,
mg/dsm3
89.9
122.2
-
86.3
72.0
98.6
87.1
61.1
119.0
125.5
91.0
110.2
109.0
104.1
98.1
74.7
116.2
114.5
94.3
107.4
Mean,
X
106.1
85.3
74.1
122.3
100.6
106.6
86.4
115.4
100.9
o,
mg/dsm3
22.8
18.8
18.4
4.6
13.6
3.5
16.5
1.2
9.3
cv,
%
22
22
25
4
13
3
19
1
9
Reference train
value, mg/dsm3
CC. A
DO .*!•
121.6
105.5
110.5
2-11
-------
TABLE 2-6. STATISTICAL DATA OF FILTERABLE AND CONDENSIBLE
ARSENIC FOR GROUPED QUAD RUNS
Quad
Run
No.
1A
IB
1C
ID
2A
2B
2C
2D
3A
3B
3C
3D
4A
4B
4C
4D
5A
5B
5C
5D
Filterable arsenic
Individual
front-half
value,
mg/dsm3
28.3
35.5
32.7
36.8
34.4
34.0
27.7
32.7
36.6
29.9
38.4
44.5
37.4
37.5
29.7
34.3
33.4
25.1
32.5
Overall
means
Group_
mean, X
32.2
33.2
34.4
37.3
31.3
33. 7a
o,
mg/dsm3
3.6
3.9
3.8
6.0
4.2
4.4b
cv,
%
11.3
11.7
11.1
16.2
13.4
13. lc
Condensible arsenic
Individual
back-half
value,
mg/dsm3
61.7
86.6
53.6
35.2
64.2
53.1
33.9
86.3
88.9
61.1
71.8
64.5
66.7
60.6
44.0
81.9
81.1
69.2
74.9
Group
mean, x
67.3
46.6
77.0
59.0
76.8
65. 3a
o,
mg/dsm3
17.2
14.6
13.0
10.3
5.9
12.8b
cv,
%
25.6
31.3
16.9
17.4
7.7
19. 6C
Simple average of tabulated data.
Pooled standard deviation;
2-12
-------
As presented in Table 2-4, the statistical data on a total train basis
showed an overall mean of 99.0 mg/dsm3 with mean arsenic concentrations of
individual quad groups ranging from 79.7 to 111.4 mg/dsm3. The standard
deviations of the quad groups ranged from 10.0 to 19.8 mg/dsm3 with a pooled
mean value of 15.7 mg/dsm3. The mean coefficient of variation for the five
runs was 15.9 percent.
Table 2-5 summarizes the within-run statistical data for paired quad runs
(either 177° or 260°C) on a total train basis. Comparison of results between
the two sample temperatures are difficult because both temperatures showed
large variations. This is evidenced by the standard deviations of paired runs
2A and B (a = 18.8 mg/dsm3) and 2C and D (a = 18.4 mg/dsm3).
In Runs 2 through 5, however, the paired means for the 177°C trains were
consistently higher than the paired means of the 260°C trains. In each quad
run, the mean arsenic concentrations determined for the 177°C trains were
between 13 and 19 percent higher than the mean concentrations for the 260°C
trains.
The statistical data for filterable and condensible arsenic presented in
Table 2-6 show a relatively consistent pattern for the filterable arsenic as
evidenced by a mean filterable arsenic concentration of 33.7 mg/dsm3 and a
pooled standard deviation of 4.4 mg/dsm3. The pooled coefficient of variation
for the filterable fraction was 13.1 percent. The individual group mean
values ranged from 32.2 to 37.4 mg/dsm3, suggesting a small difference in
filterable arsenic concentration as measured by the 177° and 260°C trains.
The condensible or back-half quad train arsenic data were characterized
by a mean concentration of 65.3 mg/dsm3 with individual group means ranging
between 46.6 and 77.0 mg/dsm3. The standard deviation of the quad groups
2-13
-------
ranged between 5.9 and 17.2 mg/dsm3 with a pooled mean standard deviation of
12.8 mg/dsm3 and a mean CV of 19.6 percent.
As presented in Tables 2-3 and 2-5, the test results for the Method 108
reference train compare to within 10 percent of the quad group means on a
total train basis. As discussed previously, leak problems with Tests RT-1 and
2 resulted in a low bias of arsenic results for these runs; thus, valid com-
parisons between the two sampling systems are limited to Runs 3 through 5.
In Run 3, the quad group mean was 111.4 mg/dsm3 compared with a reference
train value of 121.6 mg/dsm3. In Run 4, the quad group mean was 96.2 mg/dsm3
compared with a reference value of 105.5 mg/dsm3. In Run 5, the quad group
mean was 108.1 mg/dsm3 compared with a reference value of 110.5 mg/dsm3. The
reference train results averaged 2 percent lower than the 177°C results and 15
percent higher than the 260°C quad results.
In each run, the amount of filterable arsenic collected in the reference
train was greater than the corresponding quad train results. The mean filter-
able arsenic concentration in Quad Group 3 was 34.4 mg/dsm3 compared with a
reference train value of 49.4 mg/dsm3. In Quad Group 4, the mean filterable
arsenic concentration was 37.3 mg/dsm3 compared with a reference train value
of 47.1 mg/dsm3. In Quad Group 5, the mean filterable arsenic concentration
was 31.3 mg/dsm3 compared with a reference train value of 48.0 mg/dsm3.
In summary, the Method 108 reference train run at 121°C consistently
collected 20 to 30 percent more arsenic in the front half of the train than
the Method 108 trains heated to 177° and 260°C. The total train results are
comparable for the reference and 177°C trains; whereas, the 260°C results
average 15 percent lower than the reference train results.
Several factors that could have affected test results are addressed as
follows. The leak problems associated with Reference Train Tests 1 and 2
2-14
-------
resulted in a low bias of arsenic results for these runs; thus, valid compari-
sons with the corresponding quad runs are not possible.
As indicated in Tables 2-3 and 2-4, Quad Run 1C was void because of an
excessive post-test leakage rate. The calculated moisture content for this
train was approximately 45 percent lower than the within-run moisture data for
Trains 1A, B, and D; thus, this sample was discarded and not analyzed. No
leakage problems were detected in any of the reported quad train tests.
A heavy deposition of white condensate was observed in all of the back-
half glassware in the two sampling systems. This observation is consistent
with the reported arsenic results in the back half of each sampling train.
All back-half glassware were rinsed with 0.1 N NaOH, and visible material was
removed with the aid of a nylon brush. It is possible that some of the mate-
rial was not or could not be recovered, which could contribute to the reported
deviations in back-half arsenic results.
2-15
-------
SECTION 3
PROJECT QUALITY ASSURANCE
Because the desired end product of testing is to achieve representative
emission results, quality assurance is one of the main facets of stack sam-
pling. Quality assurance guidelines provide the detailed procedures and
actions necessary for defining and producing acceptable data. Four such
documents were used in this test program to ensure the collection of accepta-
ble data and to provide a definition of unacceptable data. The following
documents comprise the detailed site test plan prepared by PEI and reviewed by
the Emission Measurement Branch: the EPA Quality Assurance Handbook Volume
III, EPA-600/4-77-027; the PEI Emission Test Quality Assurance Plan; and the
PEI Laboratory Quality Assurance Plan. The last two, which are PEI's general
guideline manuals, define the company's standard operating procedures and are
followed by the emission testing and laboratory groups.
In this specific test program, the following steps were taken to ensure
that the testing and analytical procedures produced quality data:
0 Calibration of all field sampling equipment.
0 Checks on train configuration and calculations.
0 Onsite quality assurance checks (i.e., leak checks of the sampling
train, pitot tube, and Orsat line) and quality assurance checks of
all test equipment prior to use.
0 Use of designated analytical equipment and sampling reagents.
0 Internal and external audits to ensure accuracy in sampling and
analysis.
3-1
-------
Table 3-1 lists the sampling equipment used to perform the arsenic tests
and the calibration guidelines and limits. In addition to the pre- and
post-test calibrations, a field audit was performed on the metering and
temperature measurement systems used in the test runs. Critical orifices
constructed by PEI were used in the dry gas meter audits. The onsite audits
were made at the beginning of the test program. Figures 3-1 through 3-8
present the results of the onsite audits. These data were used to assess the
operational status of the sampling equipment relative to guidelines estab-
lished by the U.S. EPA. The results of the field audits indicate that the
sampling equipment was functioning properly throughout this test series.
PEI personnel calculated the sampling rates on site. The data were
rechecked and validated at the end of the test program by computer program-
ming. Some minor discrepancies between the hand calculations and computer
printouts resulted primarily because of round-off error. Overall, the data
compared favorably. Figure 3-9 presents an example calculation form PEI used
during this test program. Computerized example calculations are presented in
Appendix A.
As an additional check of the reliability of the method used to analyze
the samples, a blank train was assembled in the recovery area, capped off, and
set aside for about 2 hours. The blank train was assembled at the beginning
of the test series using clean glassware. The blank train was recovered in
the same manner as the test samples. These samples were shipped to the
laboratory and analyzed by the same procedures as those used for the actual
emission samples. In addition to the blank sampling train, aliquots of the
field reagents used in the collection and recovery of the samples were ob-
tained daily and analyzed by the same procedures as those used for the actual
3-2
-------
TABLE 3-1. FIELD EQUIPMENT CALIBRATION
Equipment
Meter box
Pi tot tube
Digital
indicator
Thermocouple
ID No.
FB-8 Train A
FB-3 Train B
FB-5 Train C
FB-1 Train D
FB-11 (Refer-
ence train)
511
517
509
124
125
221
134 - (stack)
128 - (stack)
Calibrated
against
Wet test meter
Standard pi tot
tube
Millivolt
signals
ASTM-3F
Allowable error
AH @ ±0.15
(Y ±0.05 Y post-test)
Cp ±0.01
0.5%
1.5%
(±2% saturated)
Actual
error
-0.08
0.034
-0.05
0.01
0.01
0.025
-0.02
0.007
0.0
0.0075
.
-
-
0.41%
0.14%
0.41%
+0.41%
+0.47%
Within
allowable
limits
X
X
X
X
X
X
X
X
X
X
OK
OK
X
X
X
X
X
Comments
Visually
inspected
on-site
Maximum
deviation
Maximum
deviation
CO
I
co
(continued)
-------
TABLE 3-1 (continued)
Equipment
Thermocouple
(cont'd)
Orsat
analyzer
Impinger
thermometer
ID No.
612 - Probe
632 - Filter
429 - Backup
filter
604 - Probe
634 - Filter
619 - Probe
635 - Filter
618 - Probe
631 - Filter
427 - Backup
filter
608 - Probe
615 - Probe
602 - Probe
607 - Probe
145
1-3
1-2
434
435
433
446
Calibrated
against
Standard gas
ASTM-3F
Allowable error
±0.5%
±2°F
Actual
error
+0.57%
-0.22%
-0.33
+0.57%
-0.20%
-0.63%
0.0%
0.57%
1.0%
-0.41%
+0.57%
+0.75%
-0.61%
0.2%
0.2%
0.2%
+1.0°F
+1.0°F
+0.5°F
+1.0°F
+1.5°F
+1.0°F
Within
allowable
limits
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Comments
Maximum
deviation
CO,
°2
d
CO
I
(continued)
-------
TABLE 3-1 (continued)
Equipment
Mettler
balance
Barometer
Dry gas
thermometer
Probe nozzle
ID No.
M-l
229
FB-8
FB-3
FB-5
FB-1
FB-11
1A
IB
1C
ID
2A
2B
2C
2D
RT tests
Calibrated
against
Type S weights
NBS traceable
barometer
ASTM-3F
Caliper
Allowable error
±0.5 g
+0.10 in.Hg.
(0.20 post-test)
±5°F
Dn ±0.004 in.
Actual
error
+0.1 g
0.01
in.Hg.
+4°F
+3°F
+2°F
+2°F
+1°F
+3°F
-3°F
+2°F
+2°F
+2°F
0.001 in.
0.001 in.
0.001 in.
0.000 in.
0.001 in.
0.001 in.
0.000 in.
0.001 in.
0.001 in.
Within
allowable
limits
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Comments
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
CO
tn
-------
FIELD AUDIT REPORT: DRY GAS METER
BY CRITICAL ORIFICE
DATE: /Q • /4
CLIENT:
U
BAROMETRIC PRESSURE (Pbar):£?,*/ in.Hg METER BOX NO.
ORIFICE NO. 7 PRETEST Y: rQt *tf 0 AH(a / ? / in.H20
ORIFICE K FACTOR: ^9^4 V/0" AUDITOR:
Orifice
manometer
reading
AH,
in.H20
^.,r
Dry gas
meter
reading
w
ft3
//£/**
#3,w<
Temperatures
Ambient
Tai/Taf
°F
^2-
^
Average
°F
^
Dry gas meter
Inlet
°F
^7
a *^
gf
Outlet
7f
#0
Average
V
/J
Duration
of
T '
min.
/^
Dry gas
meter
V ft3
/^.^•r
Vm
mstd'
ft3
I5J05
Vm
macf
ft3
tf.fr*7-
Audit,
Y
,Vt
Y
devia-
tion, %
-.•/
Audit
AH@,
in.H20
/.^/
AH@ Devia-
tion, in.H20
0.0
m
std
17.647(Vm)(Pbar + AH/13.6)
ft3
m
act
1203( 0 )( K
1/2
(Ta + 460) ^.,
Audit Y =
m
'act
m
y dev1ation = Audit Y -^re-test Y
x 100 = -
'std
Audit AH@ = (0.0317)(AH)(Pbar)(Tra + 460)
= /, in.H20
Audit Y must be in the range, pre-test Y ±0.05 Y.
Audit AH@ must be in the range pre-test AH@ ±0.15 inches
Figure 3-1. Field audit report: dry gas meter by
critical orifice (Meter Box FB-8, A Train).
3-6
-------
FIELD AUDIT REPORT: DRY GAS METER
BY CRITICAL ORIFICE
DATE: /Q. H
CLIENT:
U
BAROMETRIC PRESSURE (Pbar): iy.4-Qin.Hg METER BOX NO.
ORIFICE NO. J3 PRETEST Y:
ORIFICE K FACTOR? XT J?77*Ll(f* AUDITOR:
in.H0
Orifice
manometer
reading
AH,
in.H20
Z.J
Dry gas
meter
reading
ft*
fH>J*3
1(2/7 CK
Temperatures
Ambient
Tai/Taf
°F
P2-
fa
Average
V
°F
n-
Dry gas meter
Inlet
°F
Jit
#L
Outlet
VTof
°F
So
%>
Average
V
°F
B.5-
Duration
of
run
0
min.
*.*
Dry gas
meter
V ft3
/ ^^ TT
Vm
mstd«
ft3
. _ . -
'6 i" z "^
Vm
macf
ft3
•
llo.^?}
Audit,
Y
. ^
fj QJ-\
Y
devia-
tion, %
-* 9
i 1
Audit
AH@,
in.H20
1 97
A//
AH@ Devia-
tion, in.H20
.
(J • 0 1
m
std
17.647(V
m
m
m
act
Audit Y =
= /;
"std
= Audit Y
Y deviation =
x 100 -^ ,
-------
FIELD AUDIT REPORT: DRY GAS METER
BY CRITICAL ORIFICE
DATE:
/O > /4
CLIENT:
.V £ PA
BAROMETRIC PRESSURE (Pbar);Z9/T in.Hg METER BOX NO.
ORIFICE NO. I _ J_ PRETEST Y:
#* AUDITOR: /(Q.
ORIFICE K FACTOR :
AHJ9 / Zf 1n.H20
Orifice
manometer
reading
AH,
in.H20
Uof
Dry gas
meter
reading
vv
ft3
f(,&
Temperatures
Ambient
W
°F
fr-
fr
Average
-Ta.
°F
^
Dry gas meter
Inlet
w
°F
^
«fi
Outlet
To1/Tof
°F
74-
7f
Average
Tm-
°F
ii
Duration
of
run
0
min.
/{*2f
^o
Dry gas
meter
Vm« ft3
\<\.\?>*
Vm
mstd'
ft3
ivm
Vm
macf
ft3
(7,^^
Audit,
Y
/,oc4
Y
devia-
tion, %
A?3
Audit
AH@,
in.H20
/•f3
AH@ Devia-
tion, in.H20
,/y
m
std
ft3
m
act
1203(
)(Pbar)
(T + 460)
12
Audit Y =
m
deviation = Aud1t Y
std
x 100 • /
Audit AH@ = (0/0317)(AH)(P. )(T + 460)
.
Da r in
AH/13'6>
in.H20
Audit Y must be in the range, pre-test Y ±0.05 Y.
Audit AH@ must be in the range pre-test AH@ ±0.15 inches H^
Figure 3-3. Field audit report-dry gas meter by
critical orifice (Meter Box FB-5, C Train).
3-8
-------
FIELD AUDIT REPORT: DRY GAS METER
BY CRITICAL ORIFICE
DATE: / 6 <•
BAROMETRIC PRESSURE (Pbar):
ORIFICE NO. "7
CLIENT:
ORIFICE K FACTOR;
METER BOX NO.
PRETEST Y: O
AUDITOR: /
- I
AH9
11 - • - - .,.,-.. 7 ,..-.. . .
Drifice
nanometer
reading
AH,
1n.H20
Oi^
Dry gas
meter
reading
vv
ft3
?Li.^
3+l.ite
Temperatures
Ambient
°F
?z-
#>
Average
V
^
Dr
Inlet
°F
72-
fe
/gas meter
Outlet
°F
T<
7fc
Average
°F
r^
Duration
of
run
0
min.
Dry gas
meter
V ft3
^-
mstd'
ft3
I5J<
mact'
ft3
/6',PP-
Audit,
Y
t?VJ.
Y
devia-
tion, %
i 7
Audit
AH@,
in.ri20
/•7^
AH@ Devia-
tion, in.H20
.03' .
m
std
17.647(Vm)(Pbar + AH/13,6)
m
act
1203( 0 )(,K )(Pb'r)
(Tfl f 460)
\n
Audit Y =
Y deviation = Aud1t Y -
"std
100 = -, f
Audit AH@ = (0,0317) (.AH) (P, J(T t 460)
ua r m
0
Y (V(Pbar
in.H20
Audit Y must be in the range, pre-test Y ±0.05 Y.
Audit AH@ must be in the range pre-test AH@ ±0.15 inches
Figure 3-4. Field audit report: dry gas meter by
critical orifice (Meter Box FB-1, D Train).
3-9
-------
FIELD AUDIT REPORT: DRY GAS METER
BY CRITICAL ORIFICE
DATE:
f
BAROMETRIC PRESSURE
ORIFICE NO. 3
ORIFICE K FACTOR:
CLIENT:
in.Hg METER BOX NO. fl3" 11
PRETEST Y: /
AUDITOR:
AH(? /i/^r"in.H,0
Orifice
manometer
reading
AH,
in.H20
f 7 -*•"
Dry gas
meter
reading
vv
ft3
1-0 (,?(*><
-2^7 £5c
Temperatures
Ambient
°F
fa
^>
Average
V
°F
^
Dr
Inlet
°F
$
$7
y gas meter
Outlet
W
°F
IT
17
Average
V
°F
**.
Duration
of
run
0
min.
/72-I-
^
Dry gas
meter
V ft3
V Tt
-o.uS-
Vm
mstd*
ft3
/f.3Y-
v
act*
ft3
^o.tv
Audit,
Y
/,«5f
Y
devia-
tion, %
^/>.».
Audit
AH@,
in.H20
/,»
-------
Audit Name:
ON-SITE AUDIT DATA SHEET
Date: 10
Auditor:
Equipment
Meter box
inlet thermo.
Meter box
outlet thermo.
Impinger
thermometer
thermometer
or
Thermocouple
Orsat
analyzer
Trip
balance
Barometer
Reference
ASTM-3F at
ambient temp.
ASTM-3F at
ambient temp.
ASTM-3F at
ambient temp.
ASTM-3F at
ambient temp.
ASTM-3F at
stack temp.
% 02 in
ambient air
IOLM std.
weight
Corrected*
NWS value
Reference
Value
•ff
f^J
.
f$
rt
20.8%
fa
$(n
Value
Determined
y*
**
4t
n
yT/T
<*#•/ *r)«
^0.1
Deviation
?3
^
'*3
fa)
^
**
0
fiJ
Max. Allowable
Deviation
5°F
5°F
2°F
7°F
See table
0.7%
0.5 grams
0.20 in. Hg
Reference temp. °F
Max. deviation °F
32-140
7
141-273
9
274-406
11
407-540
13
541-673
15
674-760
17
* Correction factor:
NWS value (in. Hg) - [Altitude (ft)/1000(ft/in. Hg)] + 0.74 in. Hg**
** 0.74 in. Hg is the nominal correction factor for the reference barometer
against which the field barometer was calibrated.
If it is not feasible to perform the audit on any piece of equipment, record
"N/A" in the space provided for the data.
Figure 3-6. Onsite audit data sheet.
3-11
-------
Audit Name:
ON-SITE AUDIT DATA SHEET
Date:
Auditor
: /ff,
Equipment
Meter box
inlet thermo.
Meter box
outlet thermo.
Impinger
thermometer
Stack
thermometer
or
Thermocouple
Orsat
analyzer
Trip
balance
Barometer
Reference
ASTM-3F at
ambient temp.
ASTM-3F at
ambient temp.
ASTM-3F at
ambient temp.
ASTM-3F at
ambient temp.
ASTM-3F at
stack temp.
% Q£ in
ambient air
IOLM std.
weight
Corrected*
NWS value
Reference
Value
*f
y/
tt
20.8%
Value
Determined
KB-U
^
W
#(M
f*>
Deviation
~3
*3
+t*
Max. Allowable
Deviation
5°F
5°F
2°F
7°F
See table
0.7%
0.5 grams
0.20 in. Hg
Reference temp. °F
Max. deviation °F
32-140
7
141-273
9
274-406
11
407-540
13
541-673
15
674-760
17
* Correction factor:
NWS value (in. Hg) - [Altitude (ft)/1000(ft/in. Hg)] + 0.74 in. Hg**
** 0.74 in. Hg is the nominal correction factor for the reference barometer
against which the field barometer was calibrated.
If it is not feasible to perform the audit on any piece of equipment, record
"N/A" in the space provided for the data.
Figure 3-7. Onsite audit data sheet.
3-12
-------
Audit Name:
ON-SITE AUDIT DATA SHEET
Date:
Auditor:
Equipment
Meter box
inlet thermo.
Meter box
outlet thermo.
Impip^er
thermometer
Stack
thermometer
or
Thermocouple
Orsat
analyzer
Trip
balance
Barometer
Reference
ASTM-3F at
ambient temp.
ASTM-3F at
ambient temp.
ASTM-3F at
ambient temp.
ASTM-3F at
ambient temp.
ASTM-3F at
stack temp.
% 02 in
ambient air
IOLM std.
weight
Corrected*
NWS value
Reference
Value
tf
•If
20.8%
Value
Determined
flS
46
1*
W'\
•n
40
Deviation
/*•>'
+ \
~\
n>\
-/
-s
Max. Allowable
Deviation
5°F
5°F
2°F
7°F
See table
0.7%
0.5 grams
0.20 in. Hg
Reference temp. °F
Max. deviation °F
32-140
7
141-273
9
274-406
11
407-540
13
541-673
15
674-760
17
* Correction factor:
NWS value (in. Hg) - [Altitude (ft)/1000(ft/in. Hg)] + 0.74 in. Hg**
** 0.74 in. Hg is the nominal correction factor for the reference barometer
against which the field barometer was calibrated.
If it is not feasible to perform the audit on any piece of equipment, record
"N/A" in the space provided for the data.
Figure 3-8. Onsite audit data sheet.
3-13
-------
ISOKINETIC CALCULATION
SITE L^JiA/'^ei — /n«x/»*
Mtf . O.«40 (I COj) • 0.120 (S Oj)
• 0.2M (t «j » f CO) •
S. Molecular Might of tuck gat.
*, - ii^ ('-',») * l* •«» *
6. Stack velocity at tuck condition.
fpt. . ,
. \ j T.
7. Itokliwtlc variation
t. T.
f 1 • "ttd B * i 17. »
~k
. uj. ^ . Ttsr NO. / r/«//i/rv;
V.. ft3
Y
Pbtp. In.Hg
AH. 1n.H20
\> *R
V_ . dsef
•$td
V1c. 9
vw ,ft3
std
Bw
l'Bws
^ % C02
S^jfd^
1 N2 * % CO
Md. Ib/lb-nole
Ms. 1b/lb-m>le
P$t*t1e* 1n'H2°
P$. In.Hg
T~ *B
I$t R
VAP-
Cp
*,. fps
On, 1n.
6. «1n.
S I
RUN It
£7.319
£>.t?
W.&
f,*f
S&
&#*
ill.>
b.M
wS
0,10
^
/y
91^
21 3
Al/t
-/.^
yl.47
im
.Ml
(,y
&.(.
• W
4e
/•/./
RUN T&-
4.11
/.o/t,
v.zz
/.3t,
&
W<
1***
b.*X
.off
^ct
j».l(
//9f
.&\
tot
.nt
, f f? '
RUN 4-
fS^
.fli'
Af.sr
1.02,
^
.i«i
—
RUNJ&
5K5&
?i»
?•»
.03
/^/
//..4
*e*
^
/•I
,*?1
-?
j>
•*?
>0
^
»
X/f3-
.*/y
/3,3
.^*r
••••1
•S
&m
«+
r.X
.6*7
&
4*
*n
(#
oil
ff^
M^
-/.°
#•<$
in&
.n*
^^
^
w
.3*0
1
^
Figure 3-9.
-f - t*UMXKt/-f*s<°t
Example of onsite calibration data sheet.
3-14
-------
samples. Table 3-2 presents the results of the blank sampling train and field
blank analyses. The results are very low and indicate that background arsenic
contamination was not a problem in the sample recovery area.
Laboratory reagent blank analyses were performed during the analysis of
the field samples. The results of these analyses are presented in Table 3-3.
The average value for four filter blanks was 0.026 mg out of a range of 0.021
to 0.031; because this value is insignificant compared with the measured
values, no blank correction was made. All of the blank values for the rinse
and impinger samples were below the analytical detection limit of 0.002 to
0.006 mg.
Each sample was first analyzed by the flame technique. Sample concentra-
tions below 30 mg/liter were also analyzed using the graphite furnace. The
30-mg/liter limit was based on previous experience with Method 108, which
indicated good agreement above this level. As the analyses were completed and
the data were reduced by the laboratory, the results were reviewed by the
Quality Assurance Officer (QAO). The QAO reviewed instrument calibration, the
analysis of the standard reference solution (SRS), agreement between flame and
furnace results, and general consistency of the data. He then prepared a list
of samples for reanalysis.
The flame analysis was performed on six days. Twenty-eight sets of
standards (0, 10, 30, 50, 80, 100 ppm) were analyzed with the samples. Table
3-4 presents linear regression data on all the standards analyzed for the 11
analysis runs. The average correlation coefficient is 0.9988, out of a range
of 0.9994 to 0.9980. The average detection limit is 2.3 ppm. A value of
twice the range of the 0-ppm standard above the Y-intercept was used to
calculate the detection limit. A standard reference solution independently
3-15
-------
TABLE 3-2. ARSENIC BLANK DATA
Blank sampling train arsenic values'
Train No.
1
Filter, mg
0.021
NaOH probe
rinse, mg
0.030
Impinger
section, mg
<0.010
Total train
blank, mg
0.051
Field blank arsenic values
Date samples
taken
10/15
10/16
10/17
Corresponding
Run No.
1
2 + 3
4 + 5
Average blank values
Filter
total , mg
0.027
0.031
0.024
0.027
NaOH,b
mg/ liter
<0.013
<0.013
<0.013
<0.013
H20,C
mg/ liter
<0.013
<0.013
<0.013
<0.013
Sampling train was fully assembled in recovery area and then recovered and
analyzed as a sample.
""Between 150 and 493 ml of NaOH was used to rinse the probe. Between 36 and
167 ml of the NaOH was used to rinse Impingers 1 and 2. Between 53 and 126 ml
of the NaOH was used to rinse Impingers 3 and 4. Between 206 and 302 ml of the
NaOH was used to rinse the connector. The maximum blank for the NaOH corre-
sponds to 0.006 mg for the probe rinse, 0.003 mg for the impinger samples, and
0.004 mg for the connector samples.
"On all days, 150 ml of water was added to arsenic Impingers 1, 2, and 3. The
maximum blank for the water corresponds to 0.002 mg for Impingers 1, 2, and 3.
3-16
-------
TABLE 3-3. ARSENIC LABORATORY REAGENT BLANK DATA
Date
(1984)
11/15
Filter
total , mg
0.001
Rinse,3
mg/ liter
<0.013
Impingers,
mg/ liter
<0.013
Connector,0
mg/ liter
<0.013
Between 150 and 493 ml of samples were received as the rinse fraction.
The maximum laboratory reagent blank corresponds to 0.006 mg for this
fraction.
^Between 188 and 400 ml of samples were received as the Impingers 1 and
2 fractions and between 205 and 280 ml as the Impingers 3 and 4 frac-
tions. These correspond to maximum laboratory reagent blanks of 0.005
mg and 0.004 mg, respectively.
"Between 206 and 302 ml of samples were received as the connector frac-
tion. The maximum laboratory reagent blank corresponds to 0.004 mg for
this fraction.
3-17
-------
TABLE 3-4. LINEAR REGRESSION DATA (FLAME)
Date
(1984)
10/31
10/31
11/1
11/1
11/5
11/5
11/7
11/7
11/8
11/8
11/16
No. of
standard
curves
3
2
3
2
3
2
3
2
4
2
2
Y-intercept
+0.0026
-0.0003
+0.0077
+0.0035
+0.0075
+0.0076
+0.0052
+0.0032
+0.0067
+0.0075
+0.0005
Slope
0.00498
0.00485
0.00470
0.00484
0.00464
0.00447
0.00446
0.00441
0.00462
0.00468
0.003656
Correlation
coefficient
0.9987
0.9990
0.9980
0.9991
0.9986
0.9992
0.9982
0.9994
0.9989
0.9989
0.9988
Detection
limit, ppm
1.2
1.2
1.7
2.5
2.2
1.8
3.6
2.7
3.5
0.8
4.4
3-18
-------
prepared from As^O., with a nominal value of 150 ppm was analyzed (1-2 dilu-
tion) with each set of standards. (Standards were prepared from a commer-
cially available 1000-ppm standard solution.) The average value obtained in
the 28 analyses of this standard reference solution (SRS) was 157.9 ppm, with
a standard deviation (SD) of 10.6 ppm [6.7 percent relative standard deviation
(RSD)]. Only 1 of the 28 determinations made fell outside the range of the
mean ±2 SD (one was 136 ppm).
These data indicate that the precision and accuracy of the flame atomic
absorption analyses are well within acceptable limits. The percent difference
of the average measured value of the SRS and its predicted value is 5.3
percent; the RSD of the measured value is 6.7 percent.
Table 3-5 presents the results of four samples checked by the standard
addition method. The slopes of all the standard addition analyses are between
0.9 and 1.1. The results of standard addition show no consistent bias attrib-
utable to the sample matrices.
All samples below 30 ppm were also analyzed by furnace techniques.
Values obtained from flame and furnace techniques cannot be accurately com-
pared below 10 ppm because this value is too close to the flame detection
limit. Nine sets of standards (0, 0.01, 0.05, 0.10, and 0.15 mg/liter) were
analyzed with the furnace samples on a single analysis day. All the data were
reduced by linear regression analysis. The correlation coefficient for the
linear regression analysis was 0.9930. The detection limit for the graphite
furnace was 0.0064 ppm. A value of twice the range of the 0-ppm standard
above the Y-intercept was used to calculate the detection limit.
A standard reference solution independently prepared from As203 with a
nominal value of 0.0750 ppm was analyzed with each set of standards. (Stan-
dards were prepared from a commercially available 1000-ppm standard solution.)
3-19
-------
TABLE 3-5. ARSENIC STANDARD ADDITION RESULTS
Lab No.
DW185 filter
(1-10 dilution)
DW216 probe
DW240 impinger
DW182 bomb
Spike,
ppm
0
9.09
18.18
27.27
0
9.09
18.18
27.27
0
9.09
18.18
27.27
0
9.09
18.18
27.27
Previously
determined
flame, ppm
37.3
40.8
35.6
63.1
Measured,
ppm
34.88
41.44
52.66
59.77
38.98
47.46
54.85
63.87
34.06
44.73
55.12
62.50
62.23
72.62
80.56
Lost
Linear
regression analysis
Slope = 0.945
Y intercept = 34.30
Corr. = 0.9947
X intercept = 36.30
Slope = 0.903
Y intercept = 38.98
Corr. = 0.9993
X intercept = 43.18
Slope = 1.053
Y intercept = 34.75
Corr. = 0.9967
X intercept = 33.00
Slope = 1.008
Y intercept = 62.64
Corr. = 0.9970
X intercept = 62.13
3-20
-------
The average value obtained for the nine analyses of this SRS was 0.0774 ppm
with a standard deviation of 0.0047 (6.0 percent relative standard deviation).
Historically, the mean value for this SRS is 0.0762, with a standard deviation
of 0.0027. The values obtained for the SRS solution during this project are
in good agreement with our historical data. These data indicate that the
precision and accuracy of the furnace atomic absorption analyses are well
within acceptable limits. The difference in the average measured value of the
SRS and its predicted value is 3.2 percent; the SRD of the measured value is
6.0 percent.
The results of duplicate analyses are presented in Tables 3-6 and 3-7.
The absolute value of the percent difference was calculated according to the
following equation.
X, - h~
% Difference = — — x 100
X
where X, and X~ are the individual values
X is the average of X, and Xp
Duplicate analyses by flame atomic absorption above 15 ppm yields very
good results. The maximum percent difference is 6,3 percent. Duplicate
analyses by furnace atomic absorption yield generally good results (less than
10 percent difference) except for Samples DW258 and DW325. Sample DW325,
although a 23 percent difference, contains less than 0.2 mg of arsenic.
Sample DW258 gives a larger percent difference; one of the aliquots may have
been slightly contaminated. At less than 2 mg of arsenic, this is not a
significant problem considering 100 mg of arsenic was measured in each train.
3-21
-------
TABLE 3-6. DUPLICATE ANALYSIS DATA (FLAME)
Sample fraction (Lab No.)
Filter3 (DW177)
(DW188)
(DW201)
Backup filter9 (DW192)
Bomb (DW182B)
(DW187B)
(DW196B)
(DW200B)
Probe rinsec (DW221)
(DW273)
Impingerc (DW231)
(DW248)
(DW274)
(DW296)
(DW223)
(DW258)
(DW281)
Probe and connector rinse (DW313)
(DW305)
Impingerd (DW314)
(DW325)
Connector6 (DW283)
Impinger6 (DW288)
(DW301)
Arsenic, mg
33.6, 33.9
25.6, 26.4
26.8, 27.5
29.9, 29.9
3.16., 3.15 .
0.45°, 0.49°
8.97,. 8.46 .
0.40°, 0.31°
21.7, 20.4
22.5, 21.3
17.4, 17.7
60.1, 61.4
53.0, 54.3
27. 5W 27.3 .
2.73° 2.84°
2.38° 1.40P
<0.5D, <0.8D
7.73, 8.08
84.4, 83.0
104.5, 98.5.
<0.8D, <0.8D
62.5, 62.0
92.7, 93.9
92.1, 94.1
% Difference
0.8
2.9
2.7
0.2
0.1.
8.5b
5.8,
26. 8b
6.3
5.5
1.8
2.2
2.5
0.8.
3.8b
52. lb
b
4.4
1.6
5.9
b
0.9
1.3
2.2
Same aliquot analyzed on different days.
DFlame analysis below 12 ppm; which is 5 times the average flame detection
limit.
'Sample aliquots prepared and analyzed on different days.
Sample aliquots prepared and analyzed on the same day.
^Different dilutions of same aliquot analyzed the same day.
3-22
-------
TABLE 3-7. DUPLICATE ANALYSIS DATA (FURNACE)'
Sample fraction (Lab No.)
Filter bomb (DW187B)?
(DW200B)0
Probe rinse (DW313)C
Impinger (DW223)jJ
(DW258)2
IDW281):
(DW325)C
Arsenic, mg
0.31, 0.33
0.17, 0.18
8.39, 7.70
2.61, 2.67
1.61, 0.78
0.35, 0.36
0.17, 0.14
% Difference
8.8
8.3
8.6
2.0
70
3.6
23
All furnace analyses performed on the same day.
Different aliquots of same subsample diluted for furnace analysis.
cSample aliquots prepared on same day, a week prior to analysis.
Sample aliquots prepared on different days, 7 to 14 days prior to analysis,
Sample DW258 exhibited a laboratory contamination problem as evidenced by
the large percentage difference.
3-23
-------
SECTION 4
SAMPLING LOCATION AND TEST METHODS
A four-train (quad) sampling system was used to collect samples in the
breeching connecting the glass melting furnace to the exit stack. This system
allows four trains to sample simultaneously at essentially a single point in
the stack (see Figures 4-1 and 4-2). Therefore, this system reduces the
effect of variations in the velocity and particulate profiles on the sampling
results. It also permits a statistically significant number of samples to be
taken in a short amount of time. Further, since all five trains are identical
for every run, the within-train precision can be determined at the same time
as the relationship of the different trains is being compared. This methodol-
ogy for determining method precision was developed and validated in a previous
EPA study.* A total of five quad-train runs representing 20 individual sam-
ples were collected. During these runs, a single Method 108 train was run
with the sample nozzle positioned as close as possible to the quad nozzle
unit.
4.1 SAMPLING LOCATION
All samples were extracted from a rectangular brick breeching connecting
the furnace and stack. Figures 4-3 and 4-4 depict the sampling location.
*
Mitchell, W. J., and M. R. Midgett. A Means to Evaluate the Performance of
Stationary Source Test Methods. ES and T, 10:85-88, 1976.
4-1
-------
BACKUP
METHOD 5
FILTER
(121*C)
1MPINGER
TRAINS
BACKUP
METHOD 5
FILTER
(121-C)
FRONT VIEW
CSJ •—
LULU UJUJ
CO CD CO CD
oo oo
or or or en
0.0. O. O.
OD OA
OC OB
BACK VIEW
Figure 4-1. Quad train system for elevated temperature tests.
4-2
-------
6 1n.
3 In.
h
NOZZLE
D
V2 1n.
,NOZZLE
THERMOCOUPLE
U 2 1n.
2 In.
V
1 1n.
^-fc
1 1n.
>- A
1!
.
3/4 1n
S" TYPE PITOT TUBE
1n. ij
Figure 4-2. Four-train sampling system showing
nozzle, pitot tube, and thermocouple position.
4-3
-------
FLOW
r
i
i
BREECHING
«3.4 m
T
I
I
FURNACE BUILDING
JTEST
(EXISTING TEST PORTS) LLOCATION
.(REFERENCE
I TRAIN)
I \ \ \ \ \ \
I
PRESSURE CONTROL DAMPER
SAMPLE ACCESS PORT
(QUAD TRAIN LOCATION)
I \ \ \ \
PROPOSED
(REFERENCE TRAIN
LOCATION
Figure 4-3. Sampling location (plan view)
4-4
-------
2.4 m
I.D.
I !
EXISTING SAMPLE PORTS
I I
GRADE
1 i
1 i
1 1
1 1
i 1
1 I
1 [
1
1
1 |
1
I
1
1
= 20.5 m
(67'-6")
i
i
PRESSURE
CONTROL
DAMPERS -v
~4 m
r * "a.
14 f \
1
FURNACE BUILDING
EXISTING
j- SAMPLE POR'
1 r >
Y- i
s O
',', o -^ FLOW
^ 0
'
FLUE FLOOR
.61 m.
V
QUAD TRAIN LOCATION
Figure 4-4. Sampling location (elevation).
4-5
-------
Two sampling ports are located approximately 23.8 meters (78 feet) above
grade in the tapered brick-lined stack. Based on the pre-test site survey,
the sampling platform was determined to be too small to accommodate the quad
train sampling system to be used in these tests. As a result of the short
lead time needed to conduct the tests and the expense involved in modifying
the stack platform, an alternate location was selected for sample collection.
As depicted in Figures 4-3 and 4-4, a 35 x 46 cm (14 x 18 in.) access
port was available on the south side of the breeching for the quad system.
The opening was approximately 1.8 m (6 feet) downstream from a pressure con-
trol damper, and the distance from the top of the access port to the floor of
the breeching was 61 m (24 in.). A visual inspection of the duct cross sec-
tion showed no significant deposition of material on the floor of the breech-
ing. The quad train probe system was inserted near the top of the access port
so that the minimum distance between the quad probes and the duct floor was
approximately 51 cm (20 in.). The quad nozzles were positioned at least 76 cm
(30 in.) inside the duct for each test. The single Method 108 train was
inserted on the opposite side of the breeching at approximately the same level
as the quad probes. By locating the reference train as close as possible to
the quad probe system, a direct comparison can be made between arsenic dis-
tribution and sample temperature. In Quad Runs 1 and 2, a 2.4-m (8-ft) glass-
lined probe was used in the reference train tests so that the reference train
sample nozzle was positioned approximately 30.5 cm (12 in.) from the quad
nozzles. In Quad Runs 3 through 5, a 1.5-m (5-ft) glass-lined probe was used
in the reference train so that the distance between the reference and quad
train nozzles was approximately 122 m (48 in).
4-6
-------
Single-point, isokinetic sampling techniques were employed in each quad
and reference train test. The sampling time for all tests was 90 minutes, and
readings of stack flue gas and sampling train data were recorded at 5-minute
intervals for each quad train and at 10-minute intervals for the reference
train. A pitot tube and thermocouple attached to the quad and reference train
probes were used to set isokinetic sampling rates for each train. Sampling
rates were determined using programmable calculators. Prior to sampling,
velocity and temperature measurements were made to define sampling rates and
nozzle sizes.
In each train, the probe and filter temperatures were set at the pre-
determined temperature and monitored using multiterminal digital indicators
with thermocouple leads located in each probe and immediately behind the
Method 5 filter frits.
4.2 SAMPLING AND ANALYTICAL PROCEDURES
The sampling and analytical procedures used in this test program followed
those described in EPA Reference Methods 1 through 4* and proposed Method 108
as detailed in the site test plan prepared by PEI and reviewed and approved by
EMB. The procedures, which are described briefly here, are detailed in Appen-
dix D.
4.2.1 Velocity and Gas Temperature
A Type-S pitot tube and an inclined draft gauge manometer were used to
measure gas velocity pressures at the test site. Temperature was measured
with a thermocouple and digital readout.
40 CFR 60, Appendix A, Reference Methods 1 through 4, July 1984.
4-7
-------
4.2.2 Molecular Height
Flue gas composition was determined in accordance with the basic proce-
dures described in Reference Method 3.* Grab samples were collected before
any sampling began in order to establish baseline contents of oxygen, carbon
dioxide, and carbon monoxide. Bag samples were collected periodically during
sampling and analyzed with an Orsat gas analyzer.
Method 108* was used to measure arsenic concentration except that the
impingers containing hydrogen peroxide (H^O^) for S02 determination were
replaced with distilled HpO because of low (less than 30 ppm) concentrations
of SOp. All tests were conducted isokinetically by regulating the sampling
flow rate relative to the gas velocity in the stack as measured by the pitot
tube and thermocouple attached to the quad probe arrangement (see Figure 4-2).
Each individual sampling train consisted of a heated glass-lined probe, a
heated 7.6-cm (3-in.) diameter glass fiber filter (Whatman Reeve Angel 934AH),
and a series of five Greenburg-Smith impingers followed by a vacuum line,
vacuum gauge, leak-free vacuum pump, dry gas meter, thermometers, and a cali-
brated orifice. In each train, probe and filter temperatures were monitored
using digital indicators and thermocouple leads located in each probe and
immediately behind the Method 108 filter frit. In the quad runs, a 53-cm
(21-in.) glass connector was used to attach the front filter to a backup
filter maintained at approximately 121°C. The impingers followed the backup
filter for these trains.
The amount of water collected in the impinger section of the sampling
train was measured gravimetrically at the end of each sample run to determine
Method 108 is proposed. 40 CFR 61, Appendix B, Method 108, July 1983.
4-8
-------
the moisture content of the flue gas. The contents of the first three imping-
ers, each of which had been charged initially with 150 ml of distilled water,
were transferred to separate polyethylene containers. These impingers and all
associated connecting glassware were rinsed with 0.1 N NaOH; the rinses were
then added to the appropriate container(s). The contents of the fourth
impinger and 0.1 N NaOH rinse were placed in the container for the third
impinger.
All sample fractions were prepared using procedures described in EPA
Method 108 and analyzed by atomic absorption (AA) spectroscopy.
4-9
-------
SECTION 5
PROCESS DESCRIPTION
The off-gases from a glass melting furnace (designated Tank No. 161)
were tested. All samples were collected in the rectangular breeching con-
necting the furnace to the exit stack.
Personnel from Radian Corporation (an EPA Contractor) monitored the
process operation during the test series. A description of the process and
the operating parameters monitored during the test period is considered
confidential by Corning Glass Works and will be treated as such, pending
determination by the EPA.
5-1
------- |