United Ststes Office of Air Quality
Environrrenta! Protection Planning and Standards
Agency Researcn Triangle Park NC 27711
E.V.B Report 83-GLS-S
February 1984
Air
NESHAP - Glass
Manufacturing -
Arsenic
Emission Test
Report
Fostoria Glass
Moundsville,
West Virginia
-------
Copy No. of
NESHAP DEVELOPMENT
ARSENIC EMISSION TESTING AT
THE FOSTORIA GLASS COMPANY
LEAD CRYSTAL GLASS FURNACE
MOUNDSVILLE, WEST VIRGINIA
OCTOBER-NOVEMBER, 1983
Environmental
Consultants, Inc.
EMB Report No. 83-GLS-8
EPA Contract No. 68-02-3543 Prepared for:
Work Assignment No. 10 Dan Bivins
TRC Project 2318-E81-50A EPA/EMB
Prepared by:
John H. Powell
Environmental Scientist
Raymond F. Yarmac
Program Manager
February, 1984
800 Connecticut Blvd.
East Hartford, CT 06108
(203) 289-8631
-------
TABLE OF CONTENTS
SECTION PAGE
1.0 INTRODUCTION 1-1
1.1 Background 1-1
1.2 Summary of Process, Emissions, and Operating Condi-
tions 1-1
1.3 Applicability of EPA Reference Test Methods .... 1-4
1.3.1 EPA Method 5A 1-4
1.3.2 EPA Method 108 (Proposed) 1-5
1 .4 Measurement Program Summary 1-5
1.4.1 Preliminary Measurements 1-6
1.4.2 Clean-Up Evaluations 1-6
1.4.3 Primary Tests 1-6
1.4.4 Secondary Tests 1-6
1.5 Report Sections 1-7
2.0 SUMMARY AND DISCUSSION OF RESULTS 2-1
2.1 Background and Definitions 2-1
2.1.1 Particulate Arsenic 2-1
2.1.2 Gaseous Arsenic 2-2
2.1.3 Particulate 2-2
2.1.4 Condensible Organics 2-2
2.1.5 Visible Emissions 2-2
2.2 Primary Arsenic Emission Tests 2-3
2.2.1 ESP inlet 2-3
2.2.2 ESP Outlet 2-9
2.2.3 ESP Arsenic Collection Efficiency 2-10
2.3 Secondary Arsenic Emission Tests 2-11
2.4 Secondary Particulate Emission Tests 2-15
2.5 Visible Emissions Evaluation 2-18
2.6 Clean-Up Evaluations 2-18
2.7 Product Sample Analyses 2-20
2.8 Raw Batch Constituent Analyses 2-21
2.9 Ambient Air Monitoring 2-21
3.0 PROCESS AND DESCRIPTION AND OPERATIONS 3-1
3.1 Process Description 3-1
3.2 Emission Control System 3-3
3.3 Batch Composition 3-3
3.4 Process Controls 3-4
3.5 Monitoring Procedures 3-4
3.6 . Results of the October Tests 3-6
3.7 Results of the November Tests 3-8
4.0 SCOPE OF THE SAMPLING PROGRAM BY SITE 4-1
4.1 Glass Furnace Exhaust 4-1
4.2 ESP Exhaust 4-4
4.3 Product Samples 4-7
-ii-
-------
TABLE OF CONTENTS
SECTION
PAGE
5.0
5.1
5.
5.
5.
5.
5.
6.0
.1.1
,1.2
5.
5.
5.1.3
5.1.4
5.2
5.3
5,
5.
5.
5.
5.
5.
5.9
6.1.
6.2
6.3
5.3.3
SAMPLING AND ANALYTICAL PROCEDURES
Primary Testing
Sample Collection
Sample Recovery
Sample Analyses . .
Calculations
Secondary Testing/Arsenic ....
Sample Collection
Sample Recovery
Sample Analyses
Calculations
Secondary Testing/Particulate
Sample Collection
Particulate/Condensible Organic Compounds - Sample
Recovery and Preparation
Particulate/Condensible Organic Compounds - Sample
Analyses
CC>2 and 02 Determination
Preliminary Moisture Determination
Preliminary Velocity Determination
Visible Emissions
Product Samples
Ambient Air Samples
QUALITY ASSURANCE
Methods 1, 2, 4, 5, and 108
Method 3
Method 9
5- 1
5- 2"
5- 3
5- 5
5- 6
5- 7
5- 8
5- 8
5- 9
5- 9
5- 9
5- 9
5-10
5-10
5-11
5-12
5-13
5-13
5-13
5-13
5-14
6- 1
6- 1
6- 3
6- 3
APPENDICES
A
B
C
D
E
P
G
H
I
J
SAMPLING AND ANALYTICAL PROCEDURES
DATA SUMMARIES
FIELD DATA
SAMPLING LOGS
CALIBRATIONS
LABORATORY ANALYSIS
CLEAN-UP EVALUATION DATA
PROCESS OPERATIONS
PROJECT PARTICIPANTS
SCOPE OF WORK
-iii-
-------
LIST OP FIGURES
3-1
4-1
4-2
4-3
5-1
TABLE
2-1A
2-1B
2-2A
2-2B
2-3
2-4A
2-4B
2-5A
2-5B
2-6
PAGE
Lead Crystal Glass Furnace Fostoria Glass Company Mounds-
ville, west Virginia 1-3
Layout of Plant at Fostoria Glass Company 3-2
ESP Inlet Sampling Location .. 4-2
ESP Exhaust Sampling Location 4-5
Visible Emissions Observation Locations 4-8
Modified EPA Particulate Sampling Train August 18, 1977
Federal Register 5-4
LIST OF TABLES
PAGE
Summary of Primary Arsenic Emission Testing at the ESP Inlet
(English Units) . . . 2-4
Summary of Primary Arsenic Emission Testing at the ESP inlet
(Metric Units) 2-5
Summary of Primary Arsenic Emission Testing at the ESP Outlet
(English Units) 2-6
Summary of Primary Arsenic Emission Testing at the ESP Inlet
(Metric Units) 2-7
Measured ESP Arsenic Collection Efficiency 2-8
Summary of Secondary Arsenic Emission Testing at ESP inlet
(English Units) 2-13
Summary of Secondary Arsenic Emission Testing at ESP inlet
(Metric Units) 2-14
Summary of Particulate Emission Testing at the ESP Outlet
(English Units) 2-16
Summary of Particulate Emission Testing at the ESP Outlet
(Metric Units) 2-17
Clean-Up Evaluations Method 108 and Method 5A 2-19
-iv-
-------
LIST OF TABLES
TABLE PAGE
2-7 Arsenic Content Analyses of Crystal 2-20
2-8 Raw Batch Constituent Analysis 2-21
2-9 Ambient Concentrations of Lead and Arsenic 2-22
3-1 Typical Values for Selected Operating variables During
October Tests 3-7
3-2 Values for Selected Operating variables During November
Tests 3-9
4-1 Traverse point Locations ESP Inlet 4-3
4-2 Traverse Point Locations ESP Outlet 4-6
-v-
-------
1.0 INTRODUCTION
1.1 Background
Section 112 of the Clean Air Act of 1977 charges the Administrator of the
United States Environmental Protection Agency with the responsibility of
establishing National Emission Standards for Hazardous Air Pollutants (NESHAP)
that may significantly contribute to air pollution. Emission data collected
from this plant may provide a portion of the data base used by EPA to develop
NESHAP.
The EPA Industrial Studies Branch (ISB) selected the Fostoria Glass
Company in Moundsville, West Virginia as a site for an arsenic emission
measurement program because it uses a lead arsenic glass recipe and the
furnace emissions were controlled by an electrostatic precipitator (ESP) .
The test program was designed to determine arsenic concentrations and mass
emissions at the inlet and outlet of the electrostatic precipitator serving
the lead crystal glass furnace.
TRC Environmental Consultants, Inc. was retained by the EPA Emission
Measurement Branch (ENB) to perform emission measurements at the Fostoria
Glass Company. Testing was performed at- the inlet and outlet of the ESP
during the weeks of October 10 and October 31, 1983. This report has been
prepared in accordance with EPA Contract No. 68-02-3543 under the provisions
of Work Assignment No. 10.
The EPA, Industrial Studies Branch (ISB), was responsible for assuring
that process operations were suitable for testing. Process data were
monitored by Radian Corporation.
1.2 Summary of Process, Emissions, and Operating Conditions
The lead crystal glass furnace is a regenerative-recuperative natural gas
fired glass furnace. The furnace produces 24% lead crystal glass. A complete
1-1
-------
rebuild of the furnace was underway at the time of the pretest survey, and
completed in September, 1983. The furnace has maximum rated capacity of
approximately 16 tons per day (TPD). During the test program the furnace was
charged with raw batches containing arsenic trioxide (A 0 ), the usual
additive, and then with raw batches containing arsenic acid (H As 0 •
1/2
Lead crystal glass is produced by blending the raw materials and melting
them at approximately 2800°F. Arsenic is added in small quantities to the
raw material as a fining and clarifying agent. During the melting of the
glass batch raw materials, gaseous reaction products such as oxygen, nitrogen,
and carbon dioxide evolve and rise through the glass to form bubbles, reducing
the quality of the glass. The added inorganic arsenic causes the bubbles to
rise more rapidly to the melt surface and to dissipate. Arsenic-induced
chemical reactions may also reduce the rate of formation of nitrogen and
carbon dioxide bubbles.
The furnace is equipped with a United McGill 2-75-2 electrostatic
precipitator (ESP) rated at 95% collection efficiency.
Combustion air for the furnace is preheated in primary and secondary
checkers and a recuperator. Hot gases from the furnace are exhausted to the
atmosphere through one set of checkers, the ESP, the recuperator and the
exhaust stack while combustion air is drawn through the recuperator and the
other set of checkers.
A schematic of the crystal lead glass furnace i's presented in Figure 1-1.
The regenerative cycle is 20 minutes. The unit normally fires 24 hours per
day, seven days per week.
The acceptability of furnace operation for NESHAP performance testing was
determined by ISB and Radian on site. Emission sampling was performed under
1-2
-------
WEST
PRIMARY
CHECKER
P
LEAD CRYSTAL GLASS FURNACE
BURNER
GAS
INLET
(off)
DISTRIB-
UTION
CHAMBER
BURNER
GAS
INLET
(on)
TO PRODUCTION
EAST
PRIMARY
CHECKER
EXHAUST GASES
COMBUSTION AIR
STACK
SECONDARY
CHECKERS
Figure 1-1.
Lead Crystal Glass Furnace
Fostoria Glass Company
Moundsvllle, West Virginia
F.D.
FAN
ESP
ESP
RECUPERATOR
I.D.
FAN
-------
normal Fostoria operating conditions. Process operational data was recorded
by ISB/Radian personnel during the test program and is presented in Section
3.0 and Appendix H.
1.3 Applicability of EPA Reference Test Methods
EPA is required to publish a national reference test method for each
regulated source category and pollutant for which a National Emission Standard
for Hazardous Air Pollutants (NESHAP) is established. Reference test methods
are usually specified by a State regulatory agency during the State
Implementation Planning process and may be different from national reference
test methods.
The purpose of establishing a national reference test method is to ensure
that emission data collected from a specific source is representative of that
source and comparable to data collected' at other designated sources. The
primary purpose of this test program was to collect emission data using
standardized test methods which allow the data to be evaluated to develop a
NESHAP. Two test methods were selected by EPA to measure emissions from glass
furnaces. These methods are briefly described in the following subsections
and are described in detail in Section 5.
1.3.1 EPA Method 5A
EPA Method 5A measures particulate and condensible organic matter.
"Particulate matter" is defined as any finely divided solid or liquid
material, other than uncombined water, that condenses in the filtration
temperature range of 250° _+25°F (120° jfl4°C), and is collected by the
probe and filter (front half of the sampling train). "Condensible organic
matter" is defined as any material remaining after extraction, filtration and
1-4
-------
ambient evaporation of the ether-chloroform extract of the impinger portion of
the sampling train. Particulate matter and condensible organic matter are
quantified gravimetrically and results are expressed as the mass of collected
material.
1.3.2 EPA Method 108 (Proposed)
EPA Method 108 (proposed) was designed for the determination of inorganic
arsenic emissions from smelting processes. Particulate and gaseous emissions
are withdrawn isokinetically from the sources and collected on a glass fiber
filter and in water. The sampling train is similar to that of Method 5. The
collected arsenic is then analyzed by means of atomic absorption
spectrophotometry.
Method 108, as drafted, is intended for use at non-ferrous smelting
processes where high concentrations of SO, are encountered. Hydrogen
peroxide impinger solutions are included in the sampling train to scrub out
this SO . since only small amounts of SO are encountered in glass
furnace exhausts, no peroxide impinger solutions were used in this test
program. Impingers contained deionized-distilled water and were rinsed with a
0.1 N sodium hydroxide solution.
1.4 Measurement Program Summary
The emission measurement program was conducted at the Fostoria Glass
Company during the weeks of October 10 and October 31, 1983. All emission
testing was performed by TRC at the inlet and outlet of the ESP serving the
lead-crystal glass furnace. Radian monitored process operations and obtained
product samples. The following measurements were performed.
1-5
-------
1.4.1 Preliminary Measurements
Preliminary testing was performed at the inlet and outlet of the ESP to
determine volumetric flow rate and stack gas moisture content. The stack
diameters and sampling port configurations were each checked.
1.4.2 Clean-Up Evaluations
Prior to emissions testing during each of the two weeks, three Method 108
sampling trains were prepared and charged, ready to perform a test. In
addition, one Method 5A sampling train was prepared and charged. The
unexposed trains were then cleaned and the samples recovered in accordance
with the methods. The samples were then analyzed to establish background
and/or contamination levels of the sampling equipment,
1.4.3 Primary Tests
Three emissions tests were performed simultaneously at the inlet and
outlet of the ESP utilizing Method 108 during the week of October 10 with
As.O in the raw batch. Three more simultaneous sets of emissions tests
were performed during the week of October 31 with H AsO in. the raw batch
mixture. Visible emissions evaluations were performed concurrently with each
test.
1.4.4 Secondary Tests
Three sets of two simultaneous, single point secondary tests were
performed during the week of October 10 at the ESP Inlet utilizing Method 108
modified by altering the probe and filter temperatures. Three more sets of
secondary tests were performed during the week of October 31 at the ESP
Inlet. Concurrent with each set of secondary Method 108 tests performed at
1-6
-------
the ESP Inlet, a single particulate/condensible otganics emission test was
performed at the ESP Outlet. Visible emissions evaluations were performed
concurrently with these tests.
1.5 Report Sections
The remaining sections of this report present the Summary and Discussion
of Results (Section 2), Process Description and Operations (Section 3)/
Description of sampling Locations (Section 4), Sampling and Analytical
Procedures (Section 5), and Quality Assurance (Section 6). Descriptions of
methods and procedures, field and laboratory data, and calculations are
presented in various appendices as noted in the Table of Contents.
1-7
-------
2.0 SUMMARY AND DISCUSSION OF RESULTS
A summary of all collected emission data is presented in this section.
Section 2.1 provides a brief background discussion and definition of the
measured parameters. Section 2.2 presents Method 108 (modified) arsenic
emission results acquired during primary testing. Section 2.3 presents
arsenic emission results acquired during the secondary phase of the test
program. Method 5A particulate emission results are presented in Section 2.4
Section 2.5 summarizes the visible emission evaluations performed during the
primary test program. The results of the clean-up evaluations are presented
in Section 2.6. Section 2.7 presents the results of product sample analyses
for arsenic. Raw batch constituent arsenic analyses are in Section 2.8.
Ambient air monitoring results for each sampling location are presented in
Section 2.9.
2.1 Background and Definitions
This test program was designed to measure particulate and gaseous
emissions of arsenic from the lead-crystal glass furnace. Particulate and
condensible organic emissions were also measured. In addition, visible
emissions were evaluated.
2.1.1 Particulate Arsenic
Particulate arsenic emissions, for the purposes of this test program, are
defined as any arsenic that condenses at or above a specified temperature
o o
(250125 F or 550±25 P - See Section 5) and is collected in the probe
and filter (front half) of the Method 108 sampling train. Analysis is by
atomic absorbtion spectrophotometry.
2-1
-------
2.1.2 Gaseous Arsenic
Gaseous arsenic emissions, for the purposes of this sampling program, are
defined as any arsenic that does not condense at the temperatures specified in
Section 2.1.1 and that is captured in the back half of the filter holder, the
Teflon sample line, and the first three impingers of the Method 108 sampling
train. Analysis is by atomic absorbtion spectrophotometry.
2.1.3 Particulate
Particulate matter, for the purposes of this test program, is defined as
any finely divided solid or liquid matter, other than uncombined water, that
o
condenses at 250 jf 25 F and is collected in the probe and filter (front
half) of the Method 5 sampling train. Analyses are performed gravimetrically
after evaporation and desiccation of the samples.
2.1.4 Condensible Organics
Condensible organic matter, for the purposes of this test program, is
defined as that matter which remains in the impinger solution and the
back-half rinse of the Method 5 sampling train after extraction, filtration
and evaporation.
2.1.5 Visible Emissions
Visible emissions are evaluated according to appearance of a discharge
plume to a certified observer. The observer evaluates the plume by opacity in
accordance with EPA Method 9 as described in Section 5.
2-2
-------
2.2 Primary Arsenic Emission Tests
A summary of measured particulate arsenic and gaseous arsenic emission
data collected during primary testing at the ESP Inlet is presented in
Table 2-1A (English units) and Table 2-1B (metric units). Tables 2-2A
(English units) and 2-2B (metric units) present particulate arsenic and
gaseous arsenic emission data collected during primary testing at the ESP
Outlet. Table 2-3 presents the measured arsenic collection efficiencies of
the ESP. These tables include test dates and times; raw batch mixture
(As 03 or HJVsOJ; stack gas temperatures, flowrates, and moisture
contents; sample volumes and sample catches; as well as particulate arsenic,
gaseous arsenic, and total arsenic emission concentrations and mass emission
rates. Emission data is presented for six primary tests at the Inlet and
Outlet of the ESP.
2.2.1 ESP Inlet
Each primary arsenic emission test performed at the ESP Inlet was 128
minutes in duration. Two tests were performed on October 11 and one test on
October 12. Solid arsenic trioxide (As-0,) was the raw batch constituent
of interest during these first three tests (Tests P-l, P-2, and P-3). After
change-over to liquid arsenic acid (H3AsO. . 1/2 H_0) and adequate time
to purge the glass furnace, two primary tests were performed on November 1 and
another test on November 2 (Tests P-4, P-5, and P-6). Furnace operating
conditions were otherwise equivalent during all six primary tests performed.
During Tests P-l, P-2, and P-3, at the ESP Inlet the average particulate
arsenic (front half) emissions were 2.05 x 10~ gr/DSCP (0.431 Ibs/hr) or
46.98 mg/Nm (196 g/hr). The average gaseous arsenic (back half) emissions
were 3.33 x 10~ gr/DSCF (0.0007 Ibs/hr) or 7.65 x ID*2 mg/Nm (3.16 x
10~ g/hr). The average total arsenic emission rate was 2.06 x 10~2gr/DSCF
2-3
-------
TABLE 2-1A
SUMMARY Of PRIMARY ARSENIC EMISSION TESTING AT TBB ESP INLET (English Unite)
THE POSTORIA GLASS COMPANY, HOUHDSVILLB, WEST VIRGINIA
Stack Gi
Number Date Tine (°P)
Batch Mixture
As203
P-l-I 10/11/83 0914-1138 387
P-2-I 10/11/83 1358-1618 391
P-3-I 10/12/83 0957-1217 382
Average - - 387
Batch Mixture
H3»sO4
P-4-I 11/01/83 0900-1123 389
P-5-I 11/01/83 1415-1632 382
P-6-I 11/02/83 0902-1121 379
Average - - 383
Sanple Arsenic E
is Conditions Arsenic catch (oq) (articulate
(I) (DSCPM)* MDSCP) (Pront Bale) (Back Half) Total (gr/DSCP) (Lbe/hr)
10.1 2450 82.02 92.84 0.053 92.89 1.75 x 10~2 0.366
10.5 2540 87.18 126.3 0.204 126.5 2.24 x 10~2 0.487
11.2 2370 80.92 114.0 0.283 114.3 2.17 x 10"2 0.441
10.6 2450 83.17 111.0 0.180 111.2 2.05 x ID*2 0.431
11.0 2380 79.01 84.56 0.082 84.64 1.65 I 10'2 0.317
11.6 2310 75.28 114.44 0.112 114.77 2.35 I 10~2 0.465
11.6 2320 77.40 122.52 0.403 122.92 2.44 x 10"2 0.486
11.4 2340 77.21 107.17 0.272 107.44 2.15 X 10T2 0.429
Gaseous
(Back BalC)
(gr/DSCP) (Lbs/hr)
9.80110-' 0.0002
3.61 I 10-5 0.0008
5.40 x 10-5 0.0011
3.33 I 10-5 0.0007
1.60 I ID'5 0.0003
6.80 X ID'5 0.0014
8.03 X ID'5 0.0016
5.48 X 10T5 0.0011
Total
(gr/DSCP) (Lbs/hr)
1.75 X ID"2 0.366
2.24 x 10-2 .0.487
2.18 x ID'2 0.443
2.06 X 10~2 0.432
1.65 X ID'2 0.337
2.35 I 10-2 0.466
2.45 X ID'2 0.488
2.15 X ID'2 0.430
• Standard Conditions: 29.92 In. Hg • 68 degrees P.
All tests were 100 » 101 laoklnetlc.
-------
TABLE 2-1B
SUMMARY OF PRIMARY ARSBHIC EMISSION TESTING AT THB BSP INLET (Metric Unite)
THE roSTORIA GLASS COMPANY. KOUHDSVILLB, MEST VIRGINIA
Stack Gas Conditions
Teat Tenperature Moisture Flovrate
Nunbec Date Tine (°C) (») (Hni3/»ln(1
Batch Hllture
A0203
P-I-I 10/11/8} 0914-1118 197 10.1 69.4
P-2-I 10/11/83 1358-1616 199 10.5 71.9
P-3-I 10/12/8) 0957-1217 194 11.2 67.1
Average - - 197 10.6 69.4
N)
1 Batch Hllture
01 HjAaO«
P-4-I 11/01/8) 0900-1123 198 11.0 67.4
P-S-I 11/01/8) 1415-1(32 194 11.6 65.4
P-6-I 11/02/83 0902-1121 193 11.6 65.7
Average - - 195 11.4 6«.3
Arsenic Eniaslons
Sample Particulate Gaseous Total
Voluae Arsenic Catch (ng) (Front Balf) (Back Hair)
(Ha3} Particulate Gaaeoua Total (ng/N»>) (g/hr) (ng/Nn3) (g/hr) (Dg/NmJ> (g/hr)
(Front Bald (Back Balf)
2.32 92.84 0.053 92.89 40.02 166 2.28 « 10'2 9.51 I 10"2 40.1
2.47 126.3 0.204 126.5 51.13 221 8.26 x 10"2 3.56 x 1CT1 51.3
2.29 114.0 0.283 114.3 49.78 200 1.24 * 10"1 4.98 « JO"1 49.9
2.36 111.0 0.180 111.2 46.98 196 7.65 x 10T2 3.16 I 10'1 47.2
2.24 84.56 0.082 84.64 37.8 153 3.66 i 10*2 1.48 I 10~l 37.8
2.13 114.44 0.332 114.77 53.7 211 1.56 I 10"' 6.11 I 10'1 53. 8
2.19 122.52 0.403 122.92 55.9 221 1.84 I 10'1 7.25 I ID'1 56.1
2.19 107.17 0.272 107.44 49.2 195 1.26 x ID'1 4.95 I KT1 49.2
166. 1
221.3
200.5
196.3
153.1
211.6
221.7
195.5
• Standard Conditions: 760 nn llg * 20 decrees C.
All tests were 100 i 10% laoklnetlc.
-------
TABLE 2-2A
SUMMARY OF PRIMARY ARSBHIC EMISSION TESTING AT THE ESP OUTLBT (English Unite)
TUB rOSTORIA GLASS COMPANY, HOUHD3VILLB, NEST VIRGINIA
Stack Gas Conditions
Test
Number pate Time
Batch Mixture
P-l-0 10/11/83 0914-1116
P-2-0 10/11/83 1420-1619
P-3-0 IU/12/B3 0955-1158
Average
Batch Mixture
P-4-0 11/01/83 0905-1110
P-5-0 11/01/63 1417-1617
P-6-0 11/02/63 0910-1113
Average
Tenperature
205
205
.207
206
196
201
196
197
Moisture
(%)
7.6
8.1
9.0
6.3
7.5
8.0
7.7
7.7
Plowrate
(DSCPI*
4050
3920
3680
3880
4100
3890
3870
3950
Sample
Volume Arsenic Catch (ing)
(OSCPH) partlculate
(front Half)
82.71 2.34
78.41 1.65
73.91 0.85
78.34 1.61
87.90 0.69
83.56 1.36
84.23 1.22
85.23 1.09
Gaseous Total
(Back Half)
0.006 2.348
0.026 1.676
0.121 0.971
0.052 1.665
0.024 0.714
0.067 1.427
0.061 1.301
0.057 1.147
Partlculate
(Front Half)
(qr/DSCP) 1 Lhs/hr)
4.37 i 10'4 0.0152
3.25 I ID'4 0.0109
2. 03 x 10~4 0.0056
3.22 I 1CT* .0106
1.21 I 10'4 0.0043
2.51 I 10-4 0.0083
2.24 I 10~4 0.0074
1 .99 I I0"4 0.0067
Gaseous Total
(Back Ralf)
(qr/DSCP) (Lbe/hr) (gr/DSCP) (Lba/hr)
1.46 X 10'6 0.0001 4.38 I IO"4 0.0152
5.12 X 10-* 0.0002 3.30 x 10~4 0.0111
2.52 I JO'5 0.0008 2.28 I JO"4 0.0064
1.06 X 10~* 0.0004 3.32 X 10"4 0.0109
4.21 S 10-* 0.0001 1.25 x 10"4 0.0044
1.22 1 ID'* 0.0004 2.64 x 10'4 0.0088
1.49 I 10-* 0.0005 2.38 X 10'4 0.0079
1.04 X 1(T* O.OOOJ 2.09 * 10~4 0.0070
• Standard Conditions: 29.92 In. Hg t 68 degrees p.
• All tests were 100 4 10* laoklnetlc.
-------
TABLE 2-2B
SUMMARY OP PRIMARY ARSENIC EMISSION TESTING AT THE ESP OUTLET (Mettle Units)
THE POSTORIA GLASS COMPANY, NOUNDSVILLE, WEST VIRGINIA
Test
Hunber Date Tine
Batch Mixture
A8203
P-l-0 10/11/83 0914-1116
P-2-0 10/11/63 1420-1619
P-3-0 10/12/83 0955-1158
Average
Batch Mixture
H3s04
P-4-0 11/01/83 0905-1110
P-5-0 11/01/83 1417-1617
P-6-0 11/02/83 0910-1113
Average
Stack
Temperature
(°CI
96.1
96.1
97.2
96.7
91.1
93.9 •
91.1
92.0
Gas Conditions
Moisture
(%)
7.8
8.1
9.0
6.3
7.5
6.0
7.7
7.7
Flowrate
(Ha3/»ln)«
115
111
104
110
116
110
110
112
Volun
(KB1)
2.34
2.22
2.09
2.22
2.49
2.37
2.39
2.42
• Sample
e Arsenic Catch (no, )
Part Iculate
(Front Half)
2.34
1.65
0.85
1.61
0.69
1.36
1.22
1.09
Gaseous Total
(Back Half)
0.008 2.348
0.026 1.676
0.121 0.971
0.052 1.665
0.024 0.714
0.067 1.427
0.081 1.301
0.057 1.147
Arsenic Emissions
Part Iculate Gaseous
(Front Half) I Back Half)
(Bg/Hm3) (g/hr) (ng/Hm3) (g/hr)
1.00 6.90 3.42 X 10"3 2.36 I 10~2
0.740 4.95 1.17 x 10~2 7.80 x 10~2
0.408 2.54 5.79 X 10~2 3.61 X 10"1
0.717 4.81 2.43 x 10~2 1.54 x 10"1
0.277 1.95 9.64 I 10'3 6.71 X 10"2
0.575 3.77 2.81 X 10~2 1.85 I 10"1
0.513 3.36 3.41 I 10~2 2.25 I ID'1
0.456 3.04 2.40 x 1(T2 1.58 x 10'1
Total
(»q/H»3l (g/hr)
1.00 6.90
0.756 5.04
0.465 2.91
0.740 4.95
0.286 2.00
0.605 4.00
0.545 3.59
0.479 3.18
Standard Condittone: 760 on Hq 0 20 degrees c.
All teats were 100 4 10% tsok.netlc.
-------
MEASURED ESP ARSENIC COLLECTION EFFICIENCY
THE POSTORIA GLASS COMPANY, KOUNDSVULE, WEST VIRGINIA
Test
Number Date
Flowrate
"(DSCFM)*
Arsenic enlsalons (Lbs/hc)
Particulate Gaseous Total
Flowrate
(DSCFM)*
Arsenic Enlaalone ILbs/hrl
Partlculate Gaseous Total
ESP Arsenic Collection Efficiency**
Partlculate
Gaseous***
Total
(At>2Oj)
P-l 1U/11/83 0914-1138 2450
P-li 10/11/83 1358-1619 2540
P-J Iu/li./b3 0955-1217 237U
0.366 0.0002 0.366
0.487 0.0008 0.487
0.441 0.0011 0.443
Aveiage
2450
0.432
4050
3920
36BO
3880
0.01S2
0.0109
0.0056
0.0001 0.01S2
0.0002 0.0111
0.0008 0.0064
0.0004
0.0109
9S.8
97.7
96.7
97.5
50.0
75.0
27.2
95.8
97.7
98.6
97.5
I
00
P-< ii/Ul/t>J 0<»00-1123 238U
P-i 11/01/83 1415-1632 2310
P-b Il/02/b3 OVUi-1121 232U
Average 2340
0.337
0.465
0.486
0.0003
0.0014
0.0016
0.337
0.466
0.488
4100
3890
3870
0.429
0.0043
0.0083
0.0074
0.0001
0.0004
0.0005
0.0044
0.0088
0.0079
98.7
98.2
98.5
98.4
66.7
71.4
68.8
72.7
98.7
98.1
98.4
• Standard conditions: 29.92 in. Hg e 68 degrees p.
(Lbs/hr in) - (Lbe/hr out)
•• Percent Efficiency - ;
(Lhs/ht In)
••• hot actual etflclency - see text.
100
-------
(0.432 Ibs/hr) or 47.2 mg/Nm3 (196 g/hr). 99.8% of the collected arsenic
was in the particulate phase. The average volumetric flowrate of gases
entering the ESP was 2450 DSCFH (69.4 Nm /min) at 387°F (197°C) and
10.6% moisture (v/v).
During Tests P-4, P-5, and P-6 at the ESP Inlet, particulate arsenic
(front half) emissions averaged 2.15 x 10~ gr/DSCF (0.429 Ibs/hr) or
49.2 mg/Nm (195 g/hr). Gaseous arsenic (back half) emissions averaged 5.48
x 10~5 gr/DSCF (0.0011 Ibs/hr) or 1.26 x lO"1 mg/Nm3 (4.99 x
10 g/hr). The average total arsenic emission rate was 2.15 x
10~2 gr/DSCF (0.430 Ibs/hr) or 49.2 mg/Nm3(195 g/hr). 99.7% of the
collected arsenic was in the particulate phase. The average volumetric
flowrate of gases entering the ESP during these tests was 2340 . DSCFM
(66.3 Nm3/min) at 383°F (195 °C) and 11.4% moisture.
These results indicate that no significant differences in particulate,
gaseous, or total arsenic emissions are realized by changing raw batch
constituent mixtures from arsenic trioxide to arsenic acid.
Leak checks were performed following each test and found acceptable at
less than 0.02 cfm. Isokinesis was 100 +. 10% for each test performed.
2.2.2 ESP Outlet
Each primary arsenic emission test performed at the ESP Outlet was 120
minutes in duration. These tests were performed simultaneously with the
primary tests at the ESP Inlet.
During Tests P-l, P-2, and P-3 at the ESP Outlet (As O. in the raw
fc J
batch mixture), particulate arsenic (front half) emissions averaged 3.22 x
-4 3
10 gr/DSCF (0.0106 Ibs/hr) or 0.717 mg/Nm (4.81 g/hr). Gaseous arsenic
(back half) emissions averaged 1.06 x 10 gr/DSCF (0.0004 Ibs/hr) or
2-9
-------
2.43 x 10" mg/Nm3 (0.154 g/hr). Total arsenic emissions from the ESP
averaged 3.22 x 10"4 gr/DSCF (0.0109 Ibs/hr) or 0..740 mg/Nm3 (4.95 g/hr).
Particulate phase arsenic accounted for 96.7% of the total arsenic collected.
The average volumetric flowrate exiting the ESP was 3880 DSCFM (110 Nm /min)
at 206°F (96.7°C) and 8.3% moisture (v/v).
During Tests P-4, P-5, and P-6 (H-AsO. in the raw batch mixture) at
the ESP Outlet, particulate arsenic (front half) emissions averaged 1.99 x
-4 3
10 gr/DSCF (0.0067 Ibs/hr) or 0.456 mg/Nm (3.04 g/hr). The average
gaseous arsenic (back half) emission rate was 1.04 x 10 gr/DSCF
(0.0003 Ibs/hr) or 2.40 x 10~ mg/Nm (0.158 g/hr). The average total
-4
arsenic emission rate was 2.09 x 10 gr/DSCF (0.0070 Ibs/hr) or
0.479 mg/Nm (3.18 g/hr). Particulate phase arsenic accounted for 95% of
the total arsenic sample catch. The average volumetric flowrate of gases
exiting the ESP was 3950 DSCFM (112 Nm/min) at 196°F (92.0°C) and 7.7%
moisture (v/v).
The differences in arsenic emissions measured between tests performed with
arsenic trioxide and arsenic acid in the raw batch mixtures are probably
attributable to the increase in precipitator collection efficiencies as
discussed in Section 2.2.3.
Leak checks were performed following each test and found acceptable at
less than 0.02 cfm. Isokinesis was 100 + 10% for each test performed.
2.2.3 ESP Arsenic Collection Efficiency
The measured arsenic collection efficiencies (%) of the ESP during the
primary test program are presented in Table 2-3. Approximately 58% more gas
was measured exiting the ESP than was measured entering the ESP at standard
conditions. This was due to ambient air leaking into the system at various
2-10
-------
points (e.g., the recuperator and I.D. fan). For this reason, collection
efficiency was calculated using actual emission rates rather than emission
concentrations. In addition, the collection efficiencies for particulate
arsenic and gaseous arsenic presented do not take into account any phase
change of the arsenic as the gas temperatures decrease from approximately
o o
400 F at the ESP Inlet to approximately 200 F at the ESP Outlet.
During Tests P-l, P-2, and P-3, with As.03 in the raw batch, the
particulate arsenic collection efficiency of the ESP averagd 97.5%, while the
measured gaseous arsenic collection efficiency averaged 42.9%, probably due in
the most part to arsenic phase change rather than actual collection. The
average total arsenic collection efficiency was 97.5%.
During Tests P-4, P-5, and P-6, with H^sO. in the raw batch, the
particulate arsenic collection efficiency of the ESP averaged 98.4%, while the
measured gaseous arsenic collection efficiency averaged 72.7%, also probably
due to arsenic phase change rather than actual collection. The average total
arsenic collection efficiency for these tests was 98.4%.
2.3 Secondary Arsenic Emission Tests
During the secondary arsenic emission tests performed at the ESP Inlet,
emissions were sampled with two trains simultaneously at two single points of
approximately equal temperatures and velocities. Sampling probe and filter
outlet gas stream temperatures were varied between the two trains at 250 jf
25°F and 550 _+ 25°F in order to determine the effect of elevated
temperatures on the phase (particulate or gaseous) of the arsenic. It was
hypothesized that as temperatures were increased, more arsenic would be in the
gaseous phase and therefore captured in the back half (impingers) of the
sampling train rather than on the glass fiber filter (front half).
2-11
-------
A summary of measured particulate arsenic and gaseous arsenic emission
data collected during the secondary testing at the ESP Inlet is presented in
Tables 2-4A (English units) and 2-4B (metric units). These tables include
test dates and times, raw batch mixture, probe and filtered gas stream
temperatures, stack gas temperatures, moisture contents, and velocities;
sample volumes and catches; as well as particulate arsenic, gaseous arsenic,
and total arsenic concentrations. Data is presented for three sets of
secondary tests with As_0, in the raw batch mixture (Tests S-l, S-2, and
S-3) and three more sets of secondary tests with H..ASO. in the raw batch
mixture (Tests S-4, S-5, and S-6).
Secondary arsenic emission tests were originally planned to be 120 minutes
in duration. It became apparent, however, that the tests could not run this
long due to high particulate loading and the low vacuum capability of the
thimble-type filters used for this test program. These filter holders were
used for their high temperature capability needed to maintain filter outlet
temperatures at 550 _+ 25 F. The sampling time was shortened to 90 minutes
for Test S-l. Tests S-2 and S-3 were shortened even more as sampling train
vacuums approached the maximum rating for the thimble filters. Smaller
nozzles were used for Tests S-4, S-5, and S-6, thereby reducing the sample
volume and allowing 90 minute tests. The minimum allowable sample volume of
30 DSCF was exceeded for all tests.
During testing performed with arsenic trioxide in the raw batch,
particulate arsenic accounted for an average of 99.6% of the total sample
o o
catch with a filtered gas stream temperature of 250 +_ 25 F and 99.7% of
o
the total sample catch with a filtered gas stream temperature of 550 ±
25 F. During testing performed with arsenic acid in the raw batch,
particulate arsenic accounted for 99.4% of the total sample catch with a
2-12
-------
TABLE 2-4A
SUMMARY Of SECONDARY ARSENIC EMISSION TESTING AT ESP INLET {English Units)
TUB POSTORIA CLASS COMPANY, HOUNDSVILL8, WEST VIRGINIA
I
OJ
Teat
Number
Date
Tine
Filtered Gas
Stream Temp.
<_* 2S°P)
Stack
Gas Conditions
Temperature Moisture
Velocity*
IfpB)
VolUBe
(DSCPH)**
Sample
Arsenic Emissions
Arsenic Catch (no,)
Partlculate
Partlculate Gaseous Total
(Pront Half) (Back Half)
(Pront Half)
(gr/DSCP)
Gaseous
(Back Halt)
(gr/DSCF)
Total
(gr/DSCP)
Batch Mixture
ASjOj
S-lA-l
S-1B-I
S-2A-I
S-2B-I
S-3A-I
S-3B-I
Average
Average
10/12/83
10/12/83
10/13/83
10/13/83
10/13/83
10/13/83
A
B
1516-1646
1S15-164S
0901-1006
0901-1006
1300-1415
1300-1415
550
250
550
250
550
2SO
550
250
396
372
379
403
389
386
388
387
12.4
12.6
11.6
12.8
11.9
11.9
12.0
12.4
1560
1510
1330
1490
1480
1430
1460
14BO
57.75
60.48
35.64
41.69
45.23
46.17
46.21
49.45
71.29
98.93
59.60
70.75
61.10
77.94
64.00
82.54
0.217
0.115
0.196
0.226
0.175
0.697
0.203
0.153
71
99
59
70
61
78
64
62
.53
.07
.80
.98
.28
.64
.20
.90 •
1.90
2.52
2.58
2.62
2.09
2.61
2.19
2.58
X 10-'
I 10~2
I 10~2
I 10"2
. io-2
I 10"2
I IO-2
I 10~z
6.33 I
3.44 I
a. 47 i
8.36 I
5.95 i
2.33 I
6.52 X
1.17 i
10-5
10-5
10-5
10-5
10-5
,0-4
10-5
,0-4
1.91
2.53
2.59
2.63
2.09
2.63
2.19
2.60
x ID"2
X ID'2
, ,0-2
X IO-2
> ,o-2
> IO"2
X 10"2
X I0~2
Batch Mixture
S-4A-I
S-4B-I
S-5A-I
S-5B-I
S-6A-I
S-6B-1
Average
Average
11/02/83
11/02/83
11/03/83
11/03/83
11/03/83
11/03/83
A
B
1437-1607
1437-1607
0909-1039
0909-1039
1327-1457
1327-1457
250
550
250
550
250
550
2SO
550
395
402
385
395
381
320
387
372
11.2
11.2
12.2
12.4
11.9
12.0
11. B
11.9
1100
1550
1560
1600
1240
1310
1370
1490
35.42
41.97
43.21
42.52
34.44
37.60
37.69
41.70
60.20
58.60
54.00
50.6
33.80
40.0
49.30
49.70
0.345
0.284
0.221
0.181
0.207
0.176
0.257
0.214
60
58
54
50
34
40
49
49
.55
.92
.22
.80
.01
.18
.59
.97
2.62
2.15
1.93
1.84
1.51
1.64
2.02
1.88
I 10'2
I 10"2
I ID"2
I ID"2
x 10"2
X 10"2
, io-2
I ,0~2
1.50 »
1.04 I
7.88 I
6.55 I
9.29 >
7.22 I
1.07 x
B.06 I
,0-4
,0-4
10-5
10-5
10-5
,0-5
,0-4
,0-5
2.64
2.17
1.94
, .84
1.52
1.64
2.03
1 .88
I IO"2
I IO-2
> io-2
X 10-2
I ID"2
x 10"2
x IO"2
X ID"2
• Single point samples: only velocity reported.
•• Standard Conditions: 29.92 In. Hg ? (B degrees P.
All tests were 100 » 10% Insoklnetlc.
-------
TABLE 2-4B
SUMMARY OF SBCONDART ARSENIC EMISSION TESTING AT ESP INLET (Metric Unit a)
THE FOSTORIA GLASS COMPANY. NOUHDSVILLB. WEST VIRGINIA
Test
Nunber Date
Batch Mixture
AS203
S-1A-I 10/12/83
S-lB-I 10/12/83
S-2A-I 10/13/83
S-2B-I 10/13/83
S-3A-I 10/13/83
S-3B-I 10/13/83
Average A
N) Average B
1
I-1
Batch Mliture
S-4A-I 11/02/83
S-4B-I 11/02/83
S-5A-I 11/03/83
S-5B-I 11/03/83
S-4A-I 11/03/83
S-6B-1 11/03/83
Average A
Average B
Time
1516-1646
1515-1645
0901-1006
0901-1006
1300-1415
1300-1415
1437-1607
1437-1607
0909-1039
0909-1039
1327-1457
1127-1457
Filtered Gas
Streaa Tenp.
288
121
288
121
288
121
121
288
121
288
121
288
121
288
121
288
Stack
Gag Conditions
Temperature Holature
(°C» It)
202
189
193
206
198
197
198
197
202
206
196
202
194
160
197
189
12.4
12.6
11.6
12.8
11.9
11.9
12.0
12.4
11.2
11.2
12.2
12.4
11.9
12.0
11.8
11.9
Velocity
(H/Bln)
475
460
405
454
451
416
445
451
396
472
475
488
378
399
418
454
Volune
(N»3)
1 .64
1.71
1.01
1.18
1.2B
1.31
1.31
1.40
1.00
1.19
1.22
1.20
0.98
1.06
1 .07
1.18
Sample
Areenic Bniealone
Arsenic Catch
10-1
10-1
10-1
10-1
10-1
10-1
10-1
10-1
10-1
10-1
10-1
10-1
Total
log/NB3)
43.7
57. »
59.3
60.2
48.0
60.2
50.2
59.5
60.5
49.6
44.5
42.2
34.7
37.9
46.6
43.2
* Single point samples: only velocity reported.
•• standard Conditions: 760 mn llg t 20°C.
All tests were 100 * 10% insoklnetlc.
-------
filtered gas stream temperature of 250° +_ 25 F and 99.5% of the total
o o
sample catch with a filtered gas stream temperature of 550 +_ 25 P. As i s
demonstrated by these results, there are no significant differences in arsenic
phase effected by the alteration of filtered gas stream temperatures or raw
catch mixture constituents.
2.4 Secondary Particulate Emission Tests
Secondary testing was performed at the ESP Outlet concurrent with
secondary testing at the ESP Inlet to determine the emission rate of
particulate and condensible organic matter in accordance with EPA Method 5A.
A summary of these measured particulate emissions is presented in Table 2-5A
(English units) and 2-5B (metric units). These tables include sampling dates
and times, raw batch mixture, stack gas temperatures, moistures, and
flowrates; sample volumes and catches; and particulate, condensible organics,
and total particulate concentrations and emission rates.
Particulate (front half) emissions from the ESP during Tests S-l, S-2, and
S-3 (with As203 in the raw batch) averaged 0.211 Ibs/hr (6.86 x 10~
gr/DSCP) or 95.6 g/hr (15.7 mg/Nm ). Condensible organic (back half)
emissions averaged 0.048 Ibs/hr (1.56 x 10~ gr/DSCP) or 21.8 g/hr (3.58
3 -3
mg/Nm ). Total particulate emissions averaged 0.259 Ibs/hr (8.43 x 10
gr/DSCF) or 118 g/hr (19.3 mg/Nm ). The average volumetric flowrate
measured was 3580 DSCPM (101 Nm3/min) at 204°p (95.6°c) and 9.3%
moisture (v/v).
Particulate (front half) emissions from the ESP during Tests s-4, S-5, and
S-6 (with H3As04 in the raw batch) averaged 0.122 Ibs/hr (3.79 x
10 gr/DSCP) or 55.1 g/hr (8.68 mg/Nm ). Condensible organic (back half)
emissions averaged 0.019 Ibs/hr (0.60 x 10 gr/DSCF) or 8.81 g/hr
2-15
-------
TABLE 2-5A
SUMMARY OP PARTICIPATE EMISSION TESTING AT THE ESP OUTLET (English Units)
THE POSTORIA GLASS COMPANY, MOUNDSVILLE. WEST VIRGINIA
Test
Number Date Tine
Batch Mixture '
*°2°3
S-l-0 10/12/63 1520-1731
S-2-O 10/13/63 0900-1102
S-3-0 10/13/83 1251-1459
. Average
Batch Mixture
HjAoO,
N) J '
jl.S-4-0 11/02/83 1415-1620
S-5-O 11/03/83 0902-1104
S-6-0 11/03/63 1305-1507
Average
Sample partlculate Emissions
Stack Cas Conditions Volume Partlculate Catch {nqj Partlculate Gaseous Total
Temperature Moisture Plovrate (DSCPN) Partlculate Condenelble Total I Pront Half) (Back Half)
<°P) (tl (DSCPMI* (Pront Bait) organic (gr/DSCP) (Lbs/hr) (gr/DSCP) (Lba/hrl (gr/DSCP) (Lbs/hr
(Back Half)
209 9.1 3560 105.97 42.88 9.42 52.10 6.24 x 10° 0.191 1.17 x 10~3 0.042 7.62 I 10~3 0.211
204 9.5 1560 108.62 42.91 9.67 52.58 6.10 I 10° 0.186 1.17 X 10~3 0.042 7.47 I )0~3 0.228
199 9.1 3610 110.14 58.88 13.87 72.75 8.25 I 10~3 0.256 1.95 I 10*3 0.060 10.19 x ID*3 0.317
204 9.1 1580 108.24 48.22 10.99 59.21 6.86 X 10~3 0.211 1.56 X 10~3 0.048 8.43 X 1
-------
TABLE 2-5B
SUMMARY or PAP.TICULATB EMISSION TESTING AT THE ESP OUTLET (Metric Units)
TBB F03TOBI* GLASS COMPANY, MOUHDSVILLE, WEST VIRGINIA
Stack Gas Conditions
Test
Number
Batch
As20j
S-I-O
S-2-O
S-l-0
Date Tlae
Nliture
10/12/83 1520-1731
10/13/83 0900-1102
10/13/83 1253-1459
Teaperature Moleture
93.3 9.1
95.6 9.5
92. 8 9.3
Plowrate
100.8
100.8
102.8
Volune
3.00
3.08
3.12
Sample
Particulate Catch (ag)
Particulate
(front Half)
42.88
42.91
58.88
Condensible Total
Organic
(Back Hair)
9.42 52.30
9.67 52.58
13.87 72.75
Particulate Enlsalons
Particulate
(Pront
( ag/Na3 1
14.3
13.9
18.9
Halt)
(g/hr)
86.5
84.3
116
Gaseous Total
(Back Half)
(nig/Ha3) (g/hr) Ing/Ma3) (g/hr)
3.14
3.14
4.45
19.0 17.5 106
19.0 17.0 104
27.4 23.3 143
Average
95.6
9.3
101.3
3.07
48.22
10.99
59.21
15.7
95.6
3.58
21.8
19.3
118
Batch Nliture
H)A8O4
S-4-0 11/02/83 1415-1620
S-5-O ll/03/'83 0902-1104
S-6-0 11/03/83 1305-1507
Average
89.4
86.7
83.3
86.7
7.
9.
7.
7.
6 104.8
3 104.5
9 107.6
9 105.6
3.10 29.66
3.07 22.56
3.17 29.01
3.11 27.08
5.
3.
3.
4.
51 35.17
75 26.31
66 32.67
31 31.38
9.57
7.35
9.15
8.68
60.2
46.1
59.1
55.1
1.78
1.22
1.15
1.36
11.2
7.59
7.45
8.81
11.4 71.4
8.57 53.7
10.3 66.6
10.1 63.9
• standard Conditions: 760 ma Hg t 20 degrees c.
All testa were 100 » 10% Isoklnetlc.
-------
(1.36 mg/Nm ). Total participate emissions averaged 0.141 Ibs/hr (4.40 x
10" gr/DSCF) or 63.9 g/hr(10.1 mg/Nm ). The average volumetric flowrate
measured was 3730 DSCFM (106 Mm /min) at 188°F (86.7°C) and 7.9%
moisture (v/v).
The average total particulate emission rate measured during tests S-l,
S-2, and S-3 was almost twice the average measured during tests S-4, S-5, and
S-6. Primary testing indicated a greater particulate arsenic collection
efficiency during the week of October 31 than during the week of October 10.
If particulate collection efficiency followed the same trend later in the week
of October 31, it could be a possible explanation of the reduced particulate
emissions.
2.5 Visible Emissions Evaluation
Visible emissions from the ESP were evaluated by a certified observer
during each emission test performed. Evaluations began at least 10 minutes
prior to the start of each test and concluded at least 10 minutes after the
end of that test.
Although a plume opacity of 5% was observed occasionally, all six-minute
opacity averages were zero, with the exception of a six-minute period from
1106 to 1112 during Test P-l with a six-minute average of 3.75% was observed.
Visible emission evaluation field data is presented in Appendix C-3.
2.6 Clean-up Evaluations
A clean-up evaluation of each sampling train used was performed on-site
prior to each part of the emission measurement program. A summary of the
results of these evaluations is presented in Table 2-6. All results in
Tables 2-1 through 2-5 have been corrected for the "blank" values determined
during the clean-up evaluation.
2-18
-------
TABLE 2-6
. CLEAN-UP EVALUATIONS
METHOD 108 AND METHOD 5A
THE FOSTORIA GLASS COMPANY
MOUNDSVILLE, WEST VIRGINIA
Sample
Date
Sample Catch (ing)
Parameter
Front
Back
B-0-0
B-I-OB
B-O-P
(H3As04)
B-0-0
B-O-P
10/10
10/10
10/10
10/10
10/31
10/31
10/31
10/31
As
As
As
Particulate
As
As
As
Particulate
<0.053
<0.053
<0.270
<2.56
0.597
0.338
0.724
2.46
0.0045
<0.0017
<0.0016
<5.95
0.0013
0.0047
0.0011
4.50
2-19
-------
2.7 Product Sample Analyses
Samples of crystal were taken from the annealing lehr at regular intervals
during each emission test performed. These samples were composited,
pulverized, digested and analyzed for arsenic content by atomic absorbtion
spectrophotometry. Results are presented in Table 2-7. With As 0, in the
raw batch mixture, the finished crystal contained on average of 3000 vg As
per gram of glass and varied slightly between samples. With H-AsO. in the
raw batch, the finished crystal contained an average of 2700 ugAs per gram
of glass. Concentrations varied from 2080 WgAs per gram .to 3568 wgAs per
gram of glass.
TABLE 2-7
ARSENIC CONTENT ANALYSES OF CRYSTAL
THE FOSTORIA GLASS COMPANY
MOUNDSVILLE, WEST VIRGINIA
Test
Number
P-l
P-2
P-3
S-l
S-2
S-3
P-4
P-5
P-6
S-4
S-5
S-6
Date
10/11
10/11
10/12
10/12
10/13
10/13
11/1
11/1
11/2
11/2
11/3
11/3
Raw Batch
Mixture
As2°3
As203
As203
AS203
AS203
AS203
H3As04
HjAsC^
H 3AsO 4
n-^sO^
B^jAsO^
H 3AsO 4
Arsenic content
( vgAs/g glass )
3048
2913
2934
3181
2960
2757
3060
2080
2445
3568
2430
2695
2-20
-------
2.8 Raw Batch Constituent Analyses
Samples of raw batch materials were drawn on November 2 by Radian and
analyzed for arsenic content by TRC. Results are presented in Table 2-8.
In all cases except barium carbonate, the level of arsenic contamination
was below the detection limit of the method. Detection limits varied from
sample to- sample due to signal suppression on the spectrophotometer and the
size of the sample aliquot.
TABLE 2-8
RAW BATCH CONSTITUENT ANALYSIS
THE FOSTORIA GLASS COMPANY
MOUNDSVILLE, WEST VIRGINIA
Arsenic Content
Constituent (igAs/g)
Dolomite <0.54
Barium Carbonate 2.32
Sodium Sulfate <2.7
Soda Ash <0.55
Mississippi Lime <0.55
Pennsylvania Sand <5.4
2.9 Ambient Air Monitoring
Samples of ambient air at each sampling location were drawn during each
test performed to determine concentrations of arsenic and lead. Sampling and
analyses were in accordance with NIOSH Method P&CAM 173 for trace metals in
air, discussed in more detail in Section 5. Results are presented in
Table 2-9.
During testing performed with As O- in the raw batch mixture, ambient
levels of arsenic at the ESP Inlet sampling location were measured between
0.59wg/m and 3.85vig/m or below the detection limit. Ambient levels
2-21
-------
TABLE 2-9
AMBIENT CONCENTRATIONS OF LEAD AND ARSENIC
THE FOSTORIA GLASS COMPANY
MOUNDSVILLE, WEST VIRGINIA
Metal Concentrations (ng/m )
Test
Number
P-l
P-2
P-3
S-l
S-2
S-3
P-4
P-5
P-6
S-4
S-5
S-6
Inlet
Date
10/11
10/11
10/12
10/12
10/13
10/13
11/1
11/1
11/2
11/3
11/3
11/3
Lead
43.4
29.7
2.31
3.99
2.19
19.5
4.77
12.8
3.29
9.76
1.30
2.07
Arsenic
0.61
<0.28
0.59
<0.26
<0.60
3.85
<0.25
<0.26
<0.26
0.61
<0.40
<0.40
Outlet
Lead
37.6
8. .3 9
3.06
5.12
28.6
4.75
2.65
7.26
3.93
5.73
9.18
0.92
Arsenic
<0.65
<0.29
<0.30
<0.28
<0.30
<0.32
<0.30
<0.30
<0.42
<0.30
0.64
<0.59
2-22
-------
of arsenic at the ESP Outlet sampling location were" blow the detection limit
of approximately 0.3 ug/m .
Ambient concentrations of lead at the ESP inlet sampling location measured
3 3
during the first test series ranged from 2.19 yg/m to 43.4^g/m .
Concentrations of lead at the ESP Outlet sampling location ranged from 3.06
yg/m to 37.6pg/m .
During testing performed with H.AsO. in the raw batch, ambient
concentrations of arsenic at the ESP Inlet sampling location were below the
detection limit. Ambient levels of arsenic at the ESP Outlet sampling
location were also below the detection limit of the method.
Ambient concentrations of lead measured during the second test series at
the ESP Inlet ranged from 1.30 ug/m to 12.8 ug/m . Concentrations of
lead at the ESP Outlet sampling location ranged from 0.92 pg/m to
7.26 ug/m3.
2-23
-------
3.0 PROCESS DESCRIPTION AND OPERATIONS
3.1 Process Description
The glass melting furnace at Fostoria is a side-port regenerative design.
The dimensions of the furnace are 15 feet long, 11 feet wide, and the depth of
the melting area is 24 inches. Natural gas is fired preferentially, although
No. 2 fuel oil can be used if necessary. Five burners are located on either
side of the melter. Electric boosting is not employed. The furnace and its
associated refractory brick air preheaters (checkers) were recently rebuilt.
New burners were also installed. Prior to the test, the furnace had been
operating continuously for about three weeks. Under normal conditions the
furnace operates 24 hours/day, 365 days per year.
The "batch" mixture of raw materials are fed continuously into the furnace
from a hopper at the rear of the melter. As the batch melts, it flows into
the center of the furnace before exiting through a narrow throat at the
front. From the throat, the molten glass enters the refining section and is
then fed into one of three forehearths. The molten glass is removed from the
forehearths through a single orifice and is immediately channeled into the
molds and pressed into shape. During testing, only two of the three existing
forehearths were in use at any given time.
Incoming combustion air is first preheated by the .flue gases in a
recuperator near the stack. It is then passed through a secondary regenerator
before entering the primary regenerators located on either side of the
furnace. Hot off-gases from the furnace pass through a primary regenerator
opposite the one receiving the combustion air. The flue gases are ducted
through the secondary regenerator and then the emission control devices. The
clean gases are induced through the recuperator by a fan located at the base
3-1
-------
RjQcuparatot
* Secondary Ragmarator
Freaaea
Figure 3 .1 Layout of Plant at Fostoria Glass Company
3-2
-------
of the stack. At 20-minute intervals, the flow of gas and air through the
primary regenerators is reversed. Figure 3-1 illustrates the overall layout
of the furnace and the air/gas handling systems.
3.2 Emission Control System
Twin United McGill electrostatic precipitators with about 4,000 ft of
total plate area each are available for controlling emissions from the glass
furnace. The rapping cycle for the precipitators is every one hour on the
front end, and every two hours on the back end.
3.3 Batch Composition
Fostoria produces a lead-crystal glass. The total weight of the raw batch
is approximately 1,370 pounds. Normally., 5 pounds, 10 ounces of arsenic
trioxide (99% pure) is added to each batch. For the November tests, arsenic
acid was substituted for the arsenic trioxide. in addition to the raw batch
ingredients, about 740 pounds of cullet (which is typical for the Fostoria
operation) are included within .each batch giving a total batch weight of
approximately 2,110 pounds.
In order to test the accuracy of the scale used to weigh the arsenic
trioxide, each day during the October tests the glass technologist at Fostoria
(Carl Hoffman) prepared five bags of sand, each weighing 5 Ib, 10 oz. These
bags were subsequently weighed on the triple-beam balance used by TRC. There
was no significant error in either the scale or the operator's technique. The
arsenic acid is measured volumetrically. There was no practical way to
independently verify the accuracy of this measurement, although based on
observations of the procedure followed, any error can be expected to be quite
small. Since the cullet fraction is not weighed prior to being added to the
3-3
-------
batch materials, some variation in the amount of cullet entering the furnace
can be expected. The glass technologist does monitor the weight of product
output per raw material input. Roughly 2,000 pounds of glass should be
derived from each 2,110 pounds of raw material input (fusion factor = 0.948).
On this basis, the glass technologist estimated that the deviation in cullet
feed rate never exceeded 50 pounds (plus or minus). Since the furnace
produces only one type of glass, small variations in the amount of cullet
added should have no significant effect on the chemical composition of the
melt. The cullet feed rate will have some impact on the relative amounts of
volatile alkalais present, however.
3.4 Process Controls
The overall operation of the furnace is monitored in a control room
adjacent to the furnace area. All of the operating parameters are controlled
and monitored automatically. A digital recorder provides a continuous
read-out of furnace performance. The recorder can be programmed to either
print the absolute values of each of the parameters at one minute intervals,
or to display them graphically in a 'trend' mode. Normally, the latter
display is used. The digital unit does not monitor glass level or combustion
air flow; these parameters are recorded separately.
3.5 Monitoring Procedures
During the October emissions test, the following procedure was followed to
monitor the performance of the furnace. At the start of each run (and again
at the end), the time required for 25 gobs of molten glass to be released from
the forehearths was recorded. The digital recorder was temporarily programmed
to print the absolute values of the operating parameters. At 15 minute
3-4
-------
intervals, the glass level and combustion air flow charts were checked. Also
every 15 minutes, readings were taken from the ESP control panel (voltage
(AC); rectified voltage (DC); current; and inlet temperature). Since problems
had occurred with the raw material feeding mechanism in September, the rear of
the furnace was visually inspected for blockage every 30 minutes.
At the end of each run, five samples of the product were removed from each
lehr. Since the residence time in the lehrs was approximately 15 minutes less
than the typical duration of each run, the products removed were in a molten
state within the furnace at the start of testing. During all three days of
testing, part of the gob of glass exiting forehearth fl was not incorporated
into the final product. Therefore, five gobs were removed by the shop foreman
directly from the forehearths and placed in the lehr. The products and gobs
were subsequently weighed by TRC. These weights were multiplied by the rate
of gob release from the forehearths in order to calculate the pull rate of the
furnace.
A slightly modified procedure was followed during the tests, conducted in
November. Although the same process variables were monitored, readings were
taken every 20 minutes instead of at 15 minute intervals. Also, during the
November tests only Press $3 was in use for 5 out of the total of 6 test
runs. Therefore, the forehearth above Press #2 was opened in order to
increase the pull rate on the furnace to roughly that attained during the
October tests. The quantity of glass streaming from this forehearth was
measured by diverting the flow for 15 seconds and collecting the glass on a
spatula. The gobs of glass collected were then placed in the lehr and
subsequently weighed.
3-5
-------
3.6 Results of the October Tests
During the period over which the October tests were conducted, all systems
were properly functioning with two minor exceptions. First, the automatic
valve controlling the flow of natural gas to one side of the furnace was out
of order. It was therefore necessary for the furnace operator to manually
open and shut this valve at each flow reversal. Second, the flue gas flow
from the furnace after being rebuilt was only about half that of before. To
compensate for the decrease in flow, all of the flue gas was ducted to one,
rather than both, of the precipitators (ESP §1).
Throughout the October, test periods the glass melting furnace operated
smoothly. The greatest deviations occurred as a result of reversals of the
firing cycles. These reversals cause a change in the temperature within the
primary regenerators with a corresponding effect on the temperature of the
'flue gas at the ESP inlet. The temperature at the ESP inlet ranged from a
recorded low of 335 P to a high 440 P. At each reversal of the firing
cycle, there is also a brief fluctuation in the furnace pressure. These
deviations are normal for all furnaces of this design.
A summary of the important process parameters as monitored during the
tests is provided in Table 3-1. The major deviation is seen in the pull rate
of the furnace which was about 25 percent higher during the final day of
testing compared to the previous two days, and in the temperature profile
within the melting area of the furnace.
More detailed data on the pull rate of the furnace during the six tests
are provided in Appendix H.
The raw monitoring data sheets, as well as copies of the strip chart
recorder printouts for each separate run, are provided in Appendix H.
3-6
-------
TABLE 3-1 Typical Values for Selected Operating Variables Durinq October Tests
u>
I
-j
Rear Melter (°F)
Middle Melter <°P)
Front Melter (°P)
Refining Section°F)
Air Flow Rate (ftVhr)
ESP Inlet Teop (°F)
ESP DC Voltage (KV)
ESP DC current (DA)
Furnace Pull Rate (Ib/hr)
Oct
Run |1
2511
2537
2635
2214
83,000
402
19.3
9.0
975.0
11
Run |2
2512
2537
2635
2212
83,000
377
19.6
9.8
972.7
Oct
Run jl
2527
2557
2557
2214
83,000
391
19.9
11.0
985.8
13
Run |2
2525
2550
2551
2211
84,000
401
19.6
9.5
1004.3
Oct
Run 11
2524
2553
2553
2229
82,000
394
20.3
10.6
1262.3
13
Run |2
2524
2551
2553
2227
84,000
381
20.0
9.2
1280.8
-------
The only upset in the operation of the furnace occurred during the first
run on October 12, at twenty minutes past eleven, when the operator forgot to
reopen the gas valves during a flow reversal. The flow of gas to the furnace
remained off for approximately 7 minutes. As can be seen from the data sheets
supplied in Appendix H, the disruption in the supply of gas had only a minor
effect on the overall operation of the furnace. The temperature in the middle
and front of the furnace began to decline, but returned to normal within five
minutes of restoration of the gas supply. By eleven forty-five the operation
of the furnace had restabilized.
3.7 Results of the November Tests
During the November tests the flue gas was still being ducted through ESP
SI only, and the valve controlling the flow of gas into one side of the
furnace was still being operated manually.
The furnace operated smoothly throughout the test period. A summary of
the important process variables as monitored during the six test runs is
provided'in Table 3-2. The temperature of the furnace was slightly higher on
the final day of test'ing compared to the previous two days. There were also
slight variations between the second day, and the first and third days, in the
flow of combustion air into the furnace. A fairly uniform rate of pull on the
furnace was achieved. During the second run on November 1, the operator again
forgot to reopen the gas valve during a flow reversal. In this instance,
however, gas flow was restored within one minute and the operation of the
furnace was not impaired.
The raw data monitoring sheets are provided in appendix H. The only
problems encountered during the November tests were in monitoring the pull
rate of the furnace. First, the method employed to measure the flow of glass
3-8
-------
TABLE 3-2 Values for Selected Operating Variables During November Test
Rear Melter <°P>
Front Melter <°P)
Refining Section (°F)
Air Plow Rate (Et3/hr)
ESP Inlet Temp <°P>
ESP DC Voltage (KV)
ESP DC Current (tnA)
Furnace Pull Rate (Ih/hr)
Nov 1
Run tl
2517
2534
2217
82,500
407
20
12.5
950
Run |2
2513
2534
2218
82,000
419
21
13
1034
Nov 2
Run 11
2516
2542
2232
80,500
403
20
12
1370
Run 12
2505
2529
2222
81,500
404
20
12
1036
Nov 3
Run 11
2540
2562
2205
83.000
385
20
12
1043
Run 12
2538
2561
2217
83,000
375
20
12
1032
-------
streaming from Forehearth J2 involved more room for error than when product
samples were taken directly from the presses. As can be seen in the detailed
data sheets given in Appendix H, however, the actual variation in the weight
of the gobs obtained using this method was fairly small, especially during the
last three runs. During the first day of testing, samples were taken from
Forehearth §2 at the beginning, middle, and end of the run in order to account
for any irregularities in the rate of flow from the newly opened forehearth.
During the final two days, the rate of flow had stabilized and all of the
samples were taken at the beginning of the run. in the morning of November 2,
a utensil handle was being produced on press $2, and thus the pull rate could
be calculated from the actual product weight.
Second, during the first day of testing problems were encountered with
Porehearth S3. This forehearth was feeding large gobs of glass to a press on
which a heavy C\- 3 Ib) plate was being produced. Because of the weight, of
the gob, some difficulty was experienced in getting the gob to form and shear
properly from the base of the forehearth. In attempting to correct the
problem, the plant operators were making small adjustments to the forehearth
feeding mechanism during the middle of the first test run. Since these
adjustments could alter the pull rate of the furnace, a second group of 5 gob
samples were collected near the end of the run from Porehearth |3. As can be
seen in the detailed data sheet in appendix H, the rate of glass pull from
Porehearth 13 did decrease by about 50 Ib/hr between the start and the end of
the first run. The weights obtained from the samples collected at the start
of the run were averaged with those collected at the end of the run in
calculating the total furnace pull rate. The same problem briefly occurred
intermittently during the following two runs; samples were taken at one hour
3-10
-------
intervals in order to account for any variations in pull rate caused by this
problem. By the beginning of the afternoon run on November 2, the problem had
been completely rectified.
3-11
-------
4.0 SCOPE OF THE SAMPLING PROGRAM BY SITE
Emissions testing was performed on the exhaust of the lead crystal glass
furnace at the inlet and outlet of the ESP. Sampling was performed in
accordance with EPA Methods 1, 2, 3, 4, 5, 9 and 108 (modified) as described
in Section 5 and presented in Appendix A. This section presents descriptions
of each sampling location and a summary of the work performed.
Prior to any emissions testing the stack gas flow rate and percent
moisture of the gas stream was determined in accordance with EPA Methods 1, 2,
and 4 at each sampling location. Duct diameter and sampling port
configurations were confirmed at this time.
4.1 Glass Furnace Exhaust
The exhaust of the glass furnace was sampled at the ESP Inlet at the
location illustrated in Figure 4-1. Four 3-inch NPT nipples were located on
the vertical duct wall 53 inches downstream (2.5 equivalent diameters) and 13
inches upstream (0.6 equivalent diameters) from the nearest respective flow
disturbances in the 30 inch x 14.75 inch uninsulated steel duct. Dust flows
from the doghouse (a vent over the batch feeder) were shut off during .all
tests. The equivalent diameter of the duct is 20 inches. In accordance with
EPA Method 1, 32 traverse points were sampled on four horizontal traverses (8
points per port). Traverse point locations are presented in Table 4-1.
The 32 traverse points were sampled for 4 minutes each for a test duration
of 128 minutes for each primary test. Six primary tests were performed here.
During the secondary phase of the test program, two traverse points of
equivalent stack gas velocity and temperature on different traverses were
chosen as the sampling points for the two separate sampling trains. Three
sets of two simultaneous, single point tests were performed in accordance with
4-1
-------
FROM
DOGHOUSE
PRECIPITATOR
FROM GLASS
FURNACE
SAMPLING
PORTS
O
O
O
O
30"
I /
53'
3. GRADE
"Figure 4-1.
ESP Inlet Sampling Location
Fostoria Glass Company,
Moundsville, West Virginia
-------
TABLE 4-1
TRAVERSE POINT LOCATIONS
ESP INLET
THE FOSTORIA GLASS COMPANY
MOUNDSVILLE, WEST VIRGINIA
Distance From
Traverse Point Percent Stack Diameter - Inside Wall
Number From Inside Wall (inches)
1 NA* 0.95
2 NA* . 2.8
3 NA* 4.7
4 NA* 6.6
5 NA* 8.4
6 NA* 10.3
7 . NA* 12.2
8 NA* 14.0
*Not applicable for rectangular ducts.
4-3
-------
EPA Method 108 modified as described in Section 5.2. The probe and filter
outlet temperatures for the two separate trains were 250°F and 550°F
(+25 F), respectively. Sampling data was recorded at five-minute intervals.
During each primary and secondary test performed at this location, an
integrated gas sample was drawn for Orsat analysis following EPA Method 3, as
described in Section 5.3. Analysis for percent CO. and percent 0 was
performed following each test period.
An ambient air sample was drawn for arsenic and lead analysis at this
location during each emission test performed. A total of twelve samples were
drawn as described in Section 5.8.
4.2 ESP Exhaust
The sampling location for the ESP outlet is shown in Figure 4-2. The
location is inside the plant just behind the furnace and above the control
room. There are two 3-inch NPT ports at 90 along the circumference of the
stack. The steel stack is uninsulated and 24 inches in diameter. The
distance downstream to the nearest flow disturbance, the ID fan, is 204
inches. The distance upstream to the nearest flow disturbance, the top of the
stack, is greater than 48 inches. In accordance with EPA Method 1, a total of
8 traverse points were sampled at this location. Traverse point locations are
presented in Table 4-2.
Primary testing to determine arsenic emissions and secondary testing to
determine particulate emissions was performed at this location. Six 2-hour
tests were performed concurrently with the six primary tests at the furnace
exhaust to determine arsenic emissions. Six 2-hour particulate tests were
performed concurrently with the secondary tests performed at the ESP inlet
eight traverse points were sampled for 15 minutes each during testing per-
4-4
-------
TO
ATMOSPHERE
DOORS
BEHIND
STACK
T
ESP 7"
— 24'i-
>48"
SAMPLING
PORTS
204"
FROM
I.D.
FAN
Figure 4-2. ESP exhaust sampling location
Fostoria Glass Company, Moundsville, West Virginia
4-5
-------
TABLE 4-2
TRAVERSE POINT LOCATIONS
ESP OUTLET
THE FOSTORIA GLASS COMPANY
MOUNDSVILLE, WEST VIRGINIA
Distance From
Traverse Point Percent Stack Diameter Inside Wall
Number From Inside Wall (inches)
1 6.7 1.6
2 25.0 6.0
3 75.0 18.0
4 93.3 22.4
4-6
-------
formed in accordance with Methods 5A and 108. Data was recorded at 5-minute
intervals.
During each test performed at this location, one integrated gas sample was
drawn for Orsat analysis as described in Section 5.3. Analysis for percent
CO. and 0_ was performed following each test. A comparison with the
furnace exhaust Orsat analysis was indicative of leaks in the ESP system.
Visible emissions evaluations were performed concurrently with each test
performed at this location in accordance with Method 9. Observations began at
least ten minutes prior to and concluded at least ten minutes after each
test. Evaluations were made at 15-second intervals by a certified observer
and averaged for 24 consecutive readings (six minutes). A copy of the
observer's certification is presented in Appendix E-4. The visible emission
observation locations are shown in Figure 4-3.
One ambient air sample was drawn at this location for lead and arsenic
analysis during each emission test performed. A total of twelve tests were
performed as described in Section 5.8.
4.3 Product Samples
Samples of the finished crystal product were taken from the annealing lehr
during each emission test for arsenic analysis as described in Section 5.7.
Three samples were taken during each test - one at the beginning, one in the
middle, and one at the end. These samples were composited for each test and
analyzed for arsenic content.
4-7
-------
N
o
LEAD CRYSTAL
GLASS FURNACE
A.M. OBSERVATION
LOCATION
Figure 4-3. Visible Emissions Observation Locations
The Fostoria Glass Company,
Moundsville, West Virginia
4-8
-------
5.0 SAMPLING AND ANALYTICAL PROCEDURES
The purpose of this sampling program was to determine the difference
between arsenic emissions when arsenic trioxide and arsenic acid are used in
the raw batch process and to determine the effect of temperature on the degree
of arsenic control available for glass furnaces. Data acquired will be used
for the development of an arsenic emission factor for lead glass production.
A combination of EPA Reference Methods 1, 2, 3, 4, 5, and a modified draft
Method 108 was used to determine mass emissions of particulate matter and
inorganic arsenic from the glass furnace and ESP. With Method 108,
particulate and gaseous arsenic emissions are withdrawn isokinetically from
the source and collected on a glass fiber filter and in impinger water. The
collected arsenic is then analyzed by atomic absorption (AA) spectro-
photometry. Details of these methods are presented in Appendix A.
Primary sampling was performed using EPA Methods 1-5, 9, and 108 to
determine mass emissions of arsenic in accordance with the presently accepted
methods as described in Section 5.1 below. Three tests were performed with
only arsenic trioxide in the raw batch, followed by three more tests with only
arsenic acid in the raw batch. All six tests were performed in accordance
with Method 108 modified for glass furnaces. Exhaust plume opacity was
determined during each primary test in accordance with Method 9.
Following each set of three primary tests, secondary testing was performed
at the ESP Inlet using a modified Method 108 to determine the effect of
elevated filter temperatures on measured arsenic emissions. The modifications
consisted of varying the filter temperatures, utilizing out-of-stack heated
thimble filters, and sampling at two single points simultaneously with two
separate trains. Three sets of two simultaneous tests were performed with
only arsenic trioxide in the raw batch, followed by three additional sets of
5-1
-------
two simultaneous tests with only arsenic acid in the raw batch. During each
secondary test performed at the ESP Inlet, one particulate emission test was
performed at the ESP Outlet. These particulate tests enable correlation of
the glass furnace emissions with the particulate emission standard for glass
furnaces.
During each test, percent 0, and CO,, and percent moisture were
determined in accordance with EPA Reference Methods 3, and 4, respectively.
Product samples were drawn during each test and analyzed for arsenic
content by ASTM methods. Ambient air samples were also drawn during each test
for analysis of arsenic and lead by NIOSH methods.
5.1 Primary Testing
Primary testing was performed at the ESP Inlet and ESP Outlet to determine
emissions of inorganic arsenic in accordance with EPA Method 108 (modified).
This Method 108, as drafted, is intended for use at non-ferrous smelters where
high concentrations of SO- are encountered, and includes hydrogen peroxide
(HO) impinger solutions to scrub out the S0_. Since only small
amounts of SO, were expected in a glass furnace exhaust, no H-O2 was
used in the sampling train. Instead, the impingers contained distilled-
deionized (D-D) water to trap gaseous arsenic and moisture in the exhaust gas
stream. The impingers were rinsed with a 0.1 N sodium hydroxide (NaOH)
solution in accordance with the method at the conclusion of each test. Three
test runs were performed with arsenic trioxide in the raw batch during the
week of October 10. Three additional tests were performed with arsenic acid
in the raw batch during the week of October 31. A total of six primary tests
were performed.
5-2
-------
5.1.1 Sample Collection
The sampling train used is presented in Figure 5-1. The sampling train
consists of a stainless steel calibrated nozzle, a glass-lined probe/ a
glass-fiber filter and glass filter holder contained in a heated oven, a
flexible Teflon tube, and four Greenburg-Smith impingers. The first two
(pre-weighed) impingers each contain 150 ml D-D water, the third is empty, and
the fourth contains 200 g of silica gel. The first, third, and fourth
impingers are modified by removal of the tapered tip and impingement plate.
The impinger train was immersed in an ice bath to maintain the outlet gas
temperature of the fourth impinger less than 68 F. Sampling probe and
o o
filter temperatures were maintained in the range of 230 to 275 F. A
thermocouple probe was inserted into the probe and filter outlet gas stream to
ensure that proper temperatures are maintained.
The sampling train was leak checked prior to and following each test in
accordance with EPA Method 5. Leak rates were less than 0.02 cfm. Isokinetic
sampling was performed at rates less than 1.0 cfm. Data was recorded at
five-minute intervals.
Simultaneously with each test performed, an integrated gas sample was
drawn for Orsat analysis in accordance with EPA Method 3. Percent CO. and
0 , and the molecular weight of the stack gas was determined from this
analysis.
Visible emissions observations were made concurrently with all tests in
accordance with EPA Method 9. Observations were made at 15-second intervals
and averaged over six-minute periods. Observations began at least ten minutes
prior to the commencement of each test and concluded at least ten minutes
after the conclusion of each test.
5-3
-------
THERMOCOUPLE
STACK HALL ~
LEGEND
1 - NOZZLE
2 - PROBE
3 - FILTER HOLDER
4 - FILTER HEATER BOX
5 - IMPINGER ICE BATH
6 - UMBILICAL CORD
7 - VACUUM GAUGE
8 - MAIN VALVE TO PUMP
9 - PUMP
10 - BY-PASS VALVE
11 - DRY GAS METER
12 - ORIFICE AND MANOMETER
13 - PITOT TUBE AND MANOMETER
14 - STACK TEMPERATURE READOUT
15 - FLEXIBLE TUBING (Teflon)
16 - THERMOCOUPLE
FIGURE
MODIFIED EPA PARTICULATE SAMPLING TRAIN
AUGUST 18. 1977, FEDERAL REGISTER
-------
5.1.2 Sample Recovery
Sample recovery was performed in the vacant foreman's office adjacent to
the sampling area. After the probe was removed from the stack and allowed to
cool, particulate matter was wiped from the exterior of the nozzle and the
nozzle capped to prevent loss (or gain) of sample. The Teflon sample line was
then removed from the filter holder and any condensate drained into the
impingers. The filter outlet was wiped of any remaining silicone vacuum
grease and sealed with parafilm. The Teflon sample line (now drained) was
removed from the impinger train and sealed at both ends. The impinger outlet
vacuum line was then removed and the impinger train sealed. The three units
(probe and filter, Teflon line, and impinger train) were then transported to
the sample recovery area. This area was clean and wind-free in order to
minimize the chances of sample contamination.
The impinger train was inspected and abnormal conditions noted before
disassembly. The sampling train was then completely disassembled and the
liquid samples placed in polyethylene and polypropylene sample jars, while
filters were placed in inert petri dishes and sealed. The sample fractions
collected are as follows:
Container No. 1 - The 4-1/2 inch glass fiber filter was removed from
its holder and placed in a petri dish, sealed and
labeled.
Container No. 2 - The probe nozzle and front half of the filter
holder was brushed three times with a nylon
bristle brush and rinsed three times with 0.1N
NaOH. These washes were deposited in a 500 ml
sample jar and labeled.
Container No. 3 - The color of the silica gel in the fourth impinger
was noted and the silica gel weighed and then
transferred to its original labeled Nalgene
container.
5-5
-------
Container No. 4 - The ball joints of the first impinger were wiped
of silicone grease. The impinger was then weighed
to the nearest 0.5 g and the net weight of the
liquid recorded along with notations of color or
film in the impinger catch. The impinger was then
agitated in order to rinse the inside and the
contents emptied through the outlet arm into a
1000 ml sample jar without disassembling the
impinger. 30 ml of 0.1N NaOH was then poured into
the impinger, and the impinger agitated and
emptied in the above manner. This rinse was
performed in triplicate. The back half of the
filter holder and the Teflon sample line was
rinsed with 0.1N NaOH three times and the wash
deposited in the 1000 ml sample jar. The height
of the fluid was marked and the jar labeled.
Container No. 5 - Impingers 2 and 3 were treated as impinger 1
described above and emptied into a separate sample
jar.
Container No. 6
Container No. 7
200 ml of D-D H20 as a blank.
200 ml of 0.1N NaOH to serve as a blank.
5.1.3 Sample Analyses
All sample fractions were analyzed at the TRC environmental laboratories
in East Hartford, Connecticut. Samples were transported to the lab in the TRC
emission measurement van. All analyses were in accordance with EPA Method 108
as presented in Appendix A. The sample fractions were analyzed as follows:
Container No. 1 - The filter and loose particulate material was
placed in a 250 ml beaker along with the contents
of Container No. 2 (see below). The filter and
probe washes were digested by the addition of
concentrated HN03 and heat. The resultant
solution was filtered and diluted to 50 ml with DD
H20. The arsenic concentration was determined
by AA spectrophotometry.
Container No. 2 - The total content was placed with the filter into
the probe wash beaker used for Container No. 1 and
digested as described.
Container No. 3 - The silica gel was weighed to the nearest 0.5g to
determine weight gain.
5-6
-------
Container No. 4 - The liquid level in the sample jar was checked for
leakage prior to analysis. The contents were
diluted to 1000 ml with DD H20 and digested with
concentrated HNO^. The resultant solution
diluted and analyzed by AA spectrophotometry.
Container No. 5 - This sample was analyzed as described for
Container No. 4.
Filter Blanks - Two blank filters were treated as described for
Container No. 1.
0.1N NaOH Blank - 50 ml were treated as per Container No. 4.
DD B-f> Blank - 50 ml were treated as per Container No. 4.
5.1.4 Calculations
Calculations to determine inorganic arsenic concentrations and emissions
were in accordance with EPA Method 108 as presented in Appendix A. The amount
of arsenic (As) collected in each sample fraction was calculated as follows:
/
«n = ca?dvSOLN Eq. 108-7
where: Mn = Total mass of arsenic collected ( ug )
Ca = Concentration of As from standard curve (wg/ml)
F<3 = Dilution factor of sample
VSOLN = volume of solution (ml)
and the total arsenic collected is:
Mt = Mn (filter) + MN (probe) + Mn (impingers)
- Mn (filter blank) - Mn (NaOH) - Mn (H20) Eq. 108-8
The total arsenic concentration in the stack gas was calculated as follows :
6 Mt
C = 10
s v (std)
m
where: Cs = As concentration in stack gas
Vm = volume of dry gas meter at STP
5-7
-------
All other calculations were performed in accordance with EPA Methods 1 through
5 and 108. All calculations were performed utilizing a Texas Instruments TI
59 programmable calculator with printer.
5.2 Secondary Testing/Arsenic
Following the completion of each set of three primary tests performed in
accordance with Method 108 (modified), the second phase of the test program
began. Three sets of two simultaneous, single point, 90-minute tests were
performed at the ESP Inlet utilizing two different filtered gas stream
temperatures under each batch condition (three tests with arsenic trioxide and
three tests with arsenic acid). Sample recovery and analysis were as
described in Section 5.1 These tests were performed in order to determine the
effect of filter temperature on the degree of arsenic control feasible, since
a significant part of arsenic emissions from glass furnaces exhibit
significant vapor pressures at elevated temperatures. During each pair of
secondary Method 108 arsenic emission tests performed at the ESP Inlet, one
Method 5A - Particulate/Condensible Organic Emission test was performed at the
ESP Outlet.
5.2.1 Sample Collection
Two separate arsenic sampling trains simultaneously sampled two separate
traverse points of approximately equal flow rates and temperatures for 90
minutes. The sampling trains were basically the same as that described in
Section 5.1.1. The differences were the use of heated stainless steel thimble
filter holders with glass fiber filters (to accommodate higher temperatures),
two distinct probe and filter outlet temperatures, replacing the Teflon sample
line with a short length of stainless steel tubing, and the monitoring of the
5-8
-------
gas stream temperature at the outlet of the first impinger. The filter
temperatures were:
1. Train 1: As per Method 108 (230°-275°F)
2. Train 2: Actual stack temperature (550°+25°F)
The impinger train and sample train operational specifications are
outlined in Section 5.1.1. Data was recorded at 5-minute intervals.
5.2.2 Sample Recovery
Sample recovery was performed at the location and in the same manner as
discussed in Section 5.1.2.
5.2.3 Sample Analyses
Sample analyses was performed at the location and in the same manner as
discussed in Section 5.1.3.
5.2.4 Calculations
Calculations will be performed in the same manner as described in
Section 5.1.4.
5.3 Secondary Testing/Particulate
During the secondary phase of the test program, testing was performed at
the ESP exhaust to determine emissions of particulate and condensible organic
matter in accordance with EPA Method 5A.
5-9
-------
5.3.1 Sample Collection
The particulate/condensible organic sampling train was identical to the
one described in Section 5.1.1 used for the primary test. Sampling train
operations were identical, with probe and filter outlet temperatures at 250
+25°F.
5.3.2 Particulate/Condensible Organic Compounds - Sample Recovery and
Preparation
At the conclusion of each test run, separate sample fractions were
collected from the Method 5A sampling train by a three-person clean-up crew.
The liquid samples were placed in glass sample jars with Teflon-lined lids,
while the filters were placed in inert petri dishes and sealed. The sample
fractions collected are as follows:
Container No. 1 - 4-1/2 in. glass-fiber filter.
Container No. 2 - Acetone wash of nozzle, probe and front half of the
4-1/2 in. filter holder.
Container No. 3 - Contaminated impinger solution from impingers 1, 2 and 3
and D.D. ^0 wash of impingers, connectors, Teflon
sample line, back half of 4-1/2 in. filter holder and
front half of 2-1/2 in. filter holder.
Container No. 4 - Acetone wash of first three impingers, connectors, Teflon
sample line, back half of 4-1/2 in. filter holder, and
front half of 2-1/2 in. filter holder.
Container No. 5 - Silica Gel.
Sample recovery was according to EPA Method 5A as presented in Appendix
A. The probe and nozzle was brushed, and rinsed three times with acetone
which was deposited in Container 2. The front half of the 4-1/2 in. filter
holder was also rinsed with acetone, which was deposited in Container 2.
The Teflon sample line was drained into the impinger train. The Teflon
sample line was not brushed as the particulate catch there was considered to
5-10
-------
be insignificant. Impinger contents were weighed to determine moisture catch
and deposited in Container 3. The Teflon sample line, impingers, connectors
and the back half of the 4-1/2 in. filter holder were rinsed three times with
D.D. H_0 into container 3, and then rinsed three times with acetone into
Container 4.
The filter was removed from its holder and deposited into a petri dish
(Container No. 1). Filter residue on the filter holder was scraped and
deposited into the same acetone rinse container as the front half of the
filter holder. The stainless steel filter frits used in the filter holders
were not rinsed during sample recovery, as any organic compounds on the frits
would be insignificant.
Silica gel samples were weighed immediately upon the conclusion of each
test and the weights recorded by the clean-up crew. All Method 5A samples
were packed in shock-proof containers and driven to the TRC Environmental
laboratory in East Hartford for analysis upon the conclusion of the test
program.
Sample recovery data was recorded on the. sample recovery log, sample
handling log, chain of custody and analytical data forms as presented in
Appendix D of this report.
5.3.3 Particulate/Condensible Organic Compounds - Sample Analysis
All analyses were according to EPA Method 5A, as presented in Appendix A
and as approved by EPA-EMB. The sample fractions were analyzed as follows:
Container No. 1 - (4-1/2 in. glass-fiber filter) - dessicate and weigh after
24 hours, then weigh to constant weight.
Container No. 2 - (acetone probe rinse) - evaporate, dessicate and weigh
after 24 hours, then weigh to constant weight.
5-11
-------
Container No. 3 - (impinger water solution and D.D. f^O rinse) - extract,
dessicate, and weigh."
Container No. 4 - (impinger acetone wash) - evaporate, dessicate, and weigh.
container No. 5 - (Silica gel) - weigh.
Silica gel samples were weighed at the test site on a triple beam balance
upon the conclusion of each test by the Method 5A sample recovery crew. The
weight gain of the silica gel were determined to the nearest 0.5 g and
recorded.
5.4 CO., and P., Determination
Concentrations of CO and 0 was determined according to EPA Method 3
as presented in Appendix A to determine the molecular weight of the gas
stream. An integrated gas sample was taken simultaneously with each emission
sample through a stainless steel probe that is integral with the Method 5/108
probe. The sample was drawn through the probe, a flexible sample line and
into a Tedlar sample bag with an approximate volume of 1 ft in an
evacuated chamber by a diaphragm pump at a rate of approximately 0.5 liters
per minute. Data was recorded at the prescribed intervals.
Immediately upon the completion of the each test run, the integrated bag
sample was analyzed according to Method 3. An Orsat analyzer manufactured by
A.H. Thomas of Philadelphia was used to determine concentrations of CO, and
0. to the nearest 0.1 percent. Analysis was performed according to the
Method, using 3 passes through each absorbing bubbler to assure complete
absorption. Analyses were performed in triplicate.
5-12
-------
5.5 Preliminary Moisture Determination
Preliminary moisture tests were performed at each sampling location prior
to emissions testing. Testing was performed according to EPA Method 4 as
described in Appendix A.
5.6 Preliminary Velocity Determination
Preliminary velocity measurements were made according to EPA Methods 1
and 2 at each sampling location prior to emissions testing. Data was
recorded in accordance with the field data sheets.
5.7 Visible Emissions
Visible emissions observations were conducted concurrently with each
particulate/arsenic test to determine the relationship between measured and
visible emissions. Observations were made according to EPA Method 9 as
presented in Appendix A. Plume opacity was recorded to the nearest 5 percent
at 15-second intervals. Six-minute average opacity values were calculated
from each set of 24 consecutive observations, visible emissions evaluation
began at least ten minutes prior to the commencement of each test and
concluded at least fifteen minutes after the conclusion of that test.
Opacity readings were recorded by a certified observer on the Record of
Visible Emissions form presented in Appendix C.
5.8 Product Samples
Three samples (1 set) of the product were taken during each test
performed (one each at the beginning, middle, and end of the test). The
samples were crushed and composited for analysis. The twelve composite
5-13
-------
samples were analyzed for arsenic content in accordance with ASTM Methods
C169-75 and C196-80 as presented in Appendix A.
5.9 Ambient Air Samples
Ambient air at each sampling location was drawn through a 37 mm cellulose
acetate membrane filter with a personnel sampling pump at approximately 1.5
liters per minute during each test performed. A total of twenty-four tests
were performed. Samples were analyzed for arsenic and lead by atomic
absorption spectrophotometry in accordance with NIOSH Method P+CAM 173 as
presented in Appendix A.
5-14
-------
6.0 QUALITY ASSURANCE
The TRC quality assurance program was designed to ensure that emission
measurement work is performed by qualified people using proper equipment
following written procedures in order to provide accurate, defensible data.
This program is based upon the EPA Quality Assurance Handbook for Air
Pollution Measurement Systems, Volume III (EPA-600/4-7-027b).
At the beginning of each day, a meeting was held to orient personnel to
the activities scheduled for that day and to discuss results from the previous
day, and to determine if any special considerations were appropriate for the
day's work.
6.1 Methods 1, 2, 4, 5, and 108
TRC's measurement devices, including pitot tubes, dry gas meters,
thermocouples, probes and nozzles, are uniquely identified and calibrated with
documented procedures and acceptance criteria before and after each field
effort. Records of all calibration data are maintained in TRC files. These
calibration forms are presented in Appendix E.
All Method 5 and 108 sampling was 100 +10 percent isokinetic. Probe and
filter outlet temperatures were maintained within 25°F of the temperatures
specified. Deviations from these criteria were reported to the EPA/EMB task
manager to decide whether a test run should be repeated or continued.
Prior to the field test program, full clean-up evaluations of the sampling
equipment were performed. In addition, spiked samples were analyzed. This
procedure ensured the accuracy of the method. Audit samples were not
available.
A single clean-up evaluation test was performed on each initial set
(collector train) of glassware prior to collecting field samples. The tests
6-1
-------
were performed in the field sample recovery area and observed by the EPA task
manager. The sets of glassware/ including the probes, were prepared and
precleaned before conducting the clean-up evaluation tests. The impingers
were precharged as specified in the actual test program. Afterward the sample
collectors, including probes, were cleaned and the blank samples recovered and
analyzed as specified in the actual test program. Results are presented in
Section 2 of this report.
In summary, the evaluation tests were designed to precondition the sample
collectors, to establish blank background values, and to educate the clean-up
personnel in specific sample recovery procedures.
All reagents were analyzed by TRC prior to field use. Residue data from
this preliminary analysis was evaluated to assess suitability for use during
the test program. In addition, three blank samples of each reagent and 4-1/2
inch filters were collected for background analysis. All clean-up evaluation
and blank samples were analyzed in conjunction with the actual test samples.
All sample recovery was performed by a three-person clean-up crew.
Appropriate sample recovery data was recorded on the sample identification
log, sample handling log, chain-of-custody form, and analytical data forms as
presented in Appendix D.
Recovered samples were 'secured in shock-proof, metal containers in a
locked room prior to shipment for analysis.
All preparation and analysis of samples were performed at the TRC
environmental laboratories. Standards of quality assurance set forth in the
Quality Assurance Handbook for Air Pollution Measurement Systems, Volume III
(EPA-600/4-7-/027b) and the Handbook for Analytical Quality Control in Water
and Wastewater Laboratories (EPA-600/4-79-019, March 1979) are adhered to.
6-2
-------
6.2 Method 3
All Method 3 analyses were performed in triplicate, with three passes
being performed through each absorbing bubbler to ensure complete absorption.
The analyzer was leak-checked according to the method prior to any analysis.
Samples were analyzed immediately upon completion of the sampling.
6.3 Method 9
The TRC observer was certified within the past 6 months to perform visible
emission evaluations. Documentation verifying the observer's certification is
provided in Appendix E.
6-3
------- |