BEPA
United States
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park NC 27711
EMB Report 80-ELC-7
March 1981
Air
Arc Furnace -
Revision
Argon Oxygen
Deca rburiza tion
Emission Test Report
Altech Specialty Steel
Corporation
Albany, New York
-------
EMISSION TEST REPORT
AL TECH SPECIALTY STEEL CORPORATION
WATERVLIET, NEW YORK
ESED NO. 79/9
EMB NO. 80-ELC-7
by
PEDCo Environmental, Inc.
11499 Chester Road
Cincinnati, Ohio 45246
Contract No. 68-02-3546
Work Assignment No. 2
PN 3530-2
Task Manager
Dennis Holzschuh
Emission Measurement Branch, MD-13
Emission Standards and Engineering Division
OFFICE OF AIR QUALITY PLANNING AND STANDARDS
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
July 1981
-------
CONTENTS
Page
Figures iv
Tables vi
Quality Assurance Element Finder vii
Acknowledgment viii
1. Introduction 1-1
2. Process Operation 2-1
3. Summary of Results 3-1
Particulate matter 3-3
Particle size 3-21
Visible and fugitive emissions 3-33
Fabric filter dust samples 3-40
Fluoride, chromium, lead, and nickel 3-45
4. Sampling Sites and Test Methods 4-1
Site l--Uncontrolled north EAF and north AOD 4-1
Site 2—Uncontrolled South EAF and south AOD 4-7
Site 3—Fabric filter outlet 4-7
Velocity and gas temperature 4-12
Molecular weight 4-13
Particulate matter 4-13
Particle size distribution 4-15
Visible and fugitive emissions 4-17
Fabric filter dust samples 4-18
5. Quality Assurance 5-1
6. Standard Sampling and Analytical Procedures 6-1
Determination of particulate emissions 6-1
Determination of particle size distribution 6-8
References R-l
Appendix A Computer printouts and example calculations A-l
Appendix B Field data B-l
11
-------
Appendix C
Appendix D
Appendix E
Appendix F
Appendix G
CONTENTS (continued)
Sample recovery and analytical data
MRI process summary
Calibration procedures and results
Quality assurance summary
Project participants and activity log
Page
C-l
D-l
E-l
F-l
G-l
ill
-------
FIGURES
Number Page
3-1 Average Particle Size Results for Uncontrolled
Emissions, Site No. 1 3-25
3-2 Particle Size Results for Uncontrolled Emissions
During Various Furnace Operations, Site No. 1 3-26
3-3 Average Particle Size Results for Uncontrolled
Emissions, Site No. 3 3-28
3-4 Average Particle Size Distribution of Fabric
Filter Dust Samples 3-43
4-1 Control System Schematic, Top View 4-2
4-2 Control System Schematic and Location of Sam-
pling Sites, Elevation View 4-3
4-3 Control System Configuration and Location of
Sampling Sites 4-4
4-4 Inlet Sampling Locations 4-6
4-5 Sampling Site No. 3, the Fabric Filter Outlet 4-9
4-6 Sampling Location at Site No. 3, the Fabric
Filter Outlet 4-10
4-7 Location of Sampling Points at Site No. 3, the
Fabric Filter Outlet 4-11
5-1 Audit Report Sample Meter Box 5-7
5-2 Audit Report Sample Meter Box 5-8
5-3 Audit Report Sample Meter Box 5-9
IV
-------
FIGURES (continued)
Number Page
1 IL-.-TL _ - -
5-4 EPA Method 5 Dry Gas Meter Performance Test
Data Sheet 5-10
5-5 EPA Method 5 Dry Gas Meter Performance Test
Data Sheet 5-11
5-6 EPA Method 5 Dry Gas Meter Performance Test
Data Sheet 5-12
6-1 Particulate Sampling Train Schematic 6-4
6-2 Particle Size Distribution Sampling Train at
Site 1 6-11
6-3 Particle Size Distribution Sampling Train at
Site 3 6-12
-------
TABLES
Number Page
2-1 EAF and AOD Production Summary 2-4
3-1 Samples Collected at Al Tech Specialty Steel 3-2
3-2 Summary of Gas Stream Characteristics 3-6
3-3 Filterable Particulate Emission Summary 3-7
3-4 Filterable Particulate Collection Efficiency 3-10
3-5 Particulate Emission Factors 3-12
3-6 Particulate Emission Factors Based on Production 3-13
3-7 Summary of Particle Size Distribution and
Fractional Efficiency 3-29
3-8 Summary of Visible and Fugitive Emissions 3-35
3-9 Comparison of Melt Shop Visible Emissions to
Process Operation 3-37
3-10 Summary of Trace Element Analyses on Fabric
Filter Dust Samples 3-41
3-11 Summary of Supplemental Analyses for Fluoride,
Chromium, Lead, and Nickel 3-47
5-1 Field Equipment Calibration 5-3
5-2 Dry Gas Meter Audit Results 5-6
5-3 Filter Blank Analysis 5-13
5-4 Reagent Blank Analysis 5-15
5-5 Trace Element Audit Results 5-16
VI
-------
QUALITY ASSURANCE ELEMENT FINDER
(1) Title page
(2) Table of contents
(3) Project description
(4) Project organization and responsi-
bilities
(5) QA objective for measurement data
in terms of precision, accuracy, com-
pleteness, representativeness, and
comparability
(6) Sampling procedures
(7) Sample custody
(8) Calibration procedures and frequency
(9) Analytical procedures
(10) Data reduction, validation, and
reporting
(11) Internal quality control checks and
frequency
(12) Performance and system audits and
frequency
(13) Preventive maintenance procedures and
schedules
(14) Specific routine procedures used
. to assess data precision, accuracy and
completeness of specific measurement
parameters involved
(15) Corrective action
(16) Quality assurance reports to management
Location
Section Page
11
1 1-1
Appendix F F-2
Appendix F F-2
Section 6 6-1
Appendix C C-l
Appendix E E-.1
Section 6 6-1
Appendix F F-3
Appendix F F-4
Appendix F F-3
Appendix F F-5
Appendix F F-4
Appendix F F-5
Appendix F F-6
vn
-------
ACKNOWLEDGMENT
Messrs. William Terry and Lalit Banker of Midwest Research
Institute, the New Source Performance Standards contractor,
monitored the process operation during the test series, assisted
in the coordination of tests to process conditions, and provided
most of the information contained in Section 2 of this report.
Mr. Art Stienstra, Sr., of Al Tech Specialty Steel Corporation
was available to coordinate plant activities, and Mr. Dennis
Holzschuh, Task Manager for the U.S. Environmental Protection
Agency, was on site:to monitor the test series.
-------
SECTION 1
INTRODUCTION
During the week of April 6, 1981, PEDCo Environmental per-
sonnel conducted an emission sampling program at the steel melt
shop operated by Al Tech Specialty Steel Corporation (Al Tech) in
Watervliet, New York. The purpose of this test program was to
provide data for assessing the need for revising present New
Source Performance Standards (NSPS) for electric arc furnaces
(EAF) to include argon-oxygen decarburization (AOD) furnaces.
This plant was selected for source testing for the following
reasons:
1) It exhibits best available control -technology.
2) The emissions capture and control equipment is repre-
sentative of the industry.
3) Both AOD and EAF furnaces are controlled by the same
device.
4) Emission data could be obtained by nonstandard sampling
techniques on the typical positive-pressure fabric
filter.
Particulate matter concentrations and mass emission rates
were measured at two inlet sites and one outlet site. Tests at
the two inlet sites were conducted according to U.S. Environ-
mental Protection Agency (EPA) Reference Method 5.* The sampling
*
40 CFR 60, Appendix A, July 1, 1980.
1-1
-------
train used for outlet tests was as described in Method 5, but
procedures were adapted to the large exhaust area and correspond-
ing low gas velocity.
Inlet and outlet tests for particulate matter were performed
simultaneously to enable determination of control efficiency as
well as values for controlled and uncontrolled emissions. Flue
gas flow rates, temperature, and composition were measured in
conjunction with these tests. In addition, particle size distri-
bution samples were collected simultaneously at one inlet site
and at the fabric filter outlet. Method 9* procedures were used
to evaluate visible emissions (VE) from the melt shop and fabric
filter outlet throughout the test. Visual determinations of
fugitive emissions (FE) from the fabric filter dust handling
system were recorded according to the proposed Method 22.**
Samples of dust collected by the fabric filter were obtained for
analysis of particle size distribution and elemental composition.
Tests took place simultaneously at all sites, including visible
and fugitive emission sites. A representative from the NSPS
contractor assisted in coordinating the tests with process opera-
tions. Subsequent analyses were performed on two outlet particle
size samples and two fabric filter dust samples to determine the
concentration of fluoride, chromium, lead, and nickel.
This report documents the activities and results of the test
program. Section 2 describes the processes that were tested and
the operating conditions during the sampling period. Section 3
40 CFR 60, Appendix A, July 1, 1980.
**
Federal Register, Vol. 45, No. 224, November 19, 1980.
1-2
-------
presents the results and discusses them, whereas Section 4
describes the sampling sites and general test procedures used.
Section 5 briefly outlines quality assurance measures and audit
results. Section 6 details the particulate matter and particle
size distribution sampling and analytical procedures. The
appendices contain computer output and example calculations
(Appendix A), field data (Appendix B), sample recovery and
analytical data (Appendix C), MRI process summary (Appendix D),
calibration procedures and results (Appendix E), a quality
assurance summary (Appendix F), and a list of project partici-
pants (Appendix G).
1-3
-------
SECTION 2
PROCESS OPERATION
Al Tech operates a total of two EAF's and two AOD's in the
production of many different grades of steel, including stainless
steel and high-temperature alloys. Each EAF and AOD has a rated
capacity of 27.3 Mg (30 tons) and produces an average heat of 29
Mg (32 tons). The facility was operating 24 hours per day, 5
days per week, during the test series, but this schedule normally
fluctuates between 3 and 5 days per week, depending on product
demand.
The north AOD vessel and EAF No. 89 are located in the north
end of the melt shop, and EAF No. 90 and the south AOD vessel are
located in the south end of the shop. Typically, EAF No. 89
feeds the north AOD, and EAF No. 90 feeds the south AOD. If
necessary, however, the north EAF can feed the south AOD vessel,
and vice versa. Normal melt shop operation consists of charging
and backcharging the EAF with cold scrap and fluxes; meltdown;
tapping the EAF; charging molten metal, fluxes, and alloys into
the AOD vessel; refining in the AOD vessel; and tapping the
refined metal into a ladle, which is then transferred to the
teeming aisle.
2-1
-------
The EAF charge materials consists of 80 percent scrap and 20
percent additives. The scrap used at this plant is relatively
clean, contains a high percentage of stainless steel, and is not
pretreated. The initial additives are generally lime, charge
chrome, ferronickel, and iron ore. Other materials may be added
during the melting phase. A minimum amount of oxygen, if any, is
blown into the molten metal while it is in the EAF, because
refining is accomplished primarily in the ADD. Hot metal is
tapped into a ladle, weighed, and transferred to the AOD vessel.
The average EAF heat time (tap to tap) is 3.7 hours.
Hot metal from the EAF is charged into the AOD vessel,
where major alloy additions are made. The additions include high-
carbon magnesium, high-carbon chrome, nickel, silicon, molyb-
denum, and aluminum. The argon-oxygen-nitrogen gas mixture is
blown into the molten bath through tuyeres located in the bottom
of the vessel until the carbon in the metal has been oxidized to
specification. Reduction mixes are added to remove sulfur and
other impurities. The refined molten metal is tapped from the
AOD into a ladle and transferred to the teeming area where the
hot metal is poured into ingot molds. After it is tapped, the
AOD remains idle until another hot metal charge from the EAF is
ready. The AOD heat (charge through tap) takes about 1.5 to 2
hours, but can last longer if more refining is needed.
In the emission capture system, two hoods are provided above
each electric furnace and each AOD vessel. The electric furnace
hoods are situated so that one hood captures most of the
2-2
-------
emissions from charging and melting, whereas the other captures
most of the tapping emissions. The ADD hoods are situated
so that each one captures half the refining emissions. A divert-
er stack directs the AOD refining emissions toward the hoods to
avoid excessive drift of the emissions as a result of cross
drafts. Each hood has a set of automatic dampers, which can be
closed to direct more suction to another hood. The dampers on
the hoods on the charging side of each EAF are always open. The
dampers on the tapping side open automatically during tapping and
remain closed the rest of the time. The dampers on the hood of
the AOD vessels are always open because the automatic mechanism
no longer functions.
Combined exhaust gases from all four furnaces are ducted to
a Wheelabrator-Frye, positive-pressure fabric filter to remove
particulate matter. The gas flows through the fabric filter at a
rate of approximately 280 m /s (600,000 acfm) and exits via a
monovent type of exhaust. Mechanical shakers clean the bags in
each compartment at periodic intervals. Al Tech has assigned two
people to the regular maintenance of the fabric filter. A
periodic visual inspection is made, and broken bags are changed
when needed.
Technical data on the process and control system are in-
cluded in Table 1 of Appendix D.
The operations of the four furnaces and the fabric filter
were monitored by Lalit Banker and William Terry of Midwest
Research Institute (MRI). Table 2-1 presents a summary of
2-3
-------
TABLE 2-1. EAF AND AOD PRODUCTION SUMMARY3
PEDCo
run No. ,
date
1
4/7/81
2
4/8/81
3
4/9/81
Furnace
North AOD
89 EAF
90 EAF
South AOD
North AOD
89 EAF
90 EAF
South AOD
North AOD
89 EAF
90 EAF
South AOD
Partial or completed
heats tested
Heat No.
98033
08440
98047
98068
08440
08444
08445
98047
08444
98045
98064
98064
98039
08464
08446
08447
08464
08446
98090
98091
98091
98044
08451
08465
08452
08451
Heat
tlme.b
minutes
85
100
190
225
440
240
235
no
155
90
65
210
299
185
225
300
150
230
225
225
340
280
350
225
e
160
Metal
produced
Mg tons
28.0 30.9
27.9 30.8
27.7 30.5
28.1 31.0
25.4 28.0
30.2 33.2
27.2 30.0
28.9 31.8
33.9 37.4
Total
27.9 30.8
30.8 34.0
29.9 33.0
29.9 33.0
28.6 31.5
29.5 32.5
28.8 31.8
32.7 36.1
32.2 35.5
Total
29.1 32.1
30.4 33.5
28.1 31.0
24.9 27.5
30.4 33.5
32.5 35.8
.
35.0 38.6
Total
Tested production
rated
Mg/h
8.7
8.5
6.5
5.2
28.9
8.5
7.1
7.7
6.9
30.2
4.9
7.8
9.8
5.7
28.2
tons/h
9.6
9.4
7.2
5.7
31.9
9.4
7.8
8.5
7.6
33.3
5.4
8.6
10.8
6.3
31.1
aCompiled from process data in Appendix D and field data (test times) in
Appendix 8.
b Charge-to-tap time for AOD's; tap-to-tap time for EAF's.
c EAF production is the weight of metal transferred to an AOD; AOD production
1s the final tap weight.
dTested production rates were determined by dividing the total weight of
metal produced during a test by the sampling time. The weight of metal
produced by a given furnace during a test was calculated by first
dividing the minutes of normal operation actually sampled (from charging
through tapping, not Including delays or patching) by the total minutes
of normal operation 1n the heat, and then multiplying by the weight of
metal produced during the entire heat.
eThe only normal operation of this heat sampled was the tap.
2-4
-------
production data for the test series. The EAF heat times varied
from 3.1 to 7.3 hours (tap to tap), and the weight of metal
produced per heat ranged from 24.9 to 32.5 Mg (27.5 to 35.8
tons). The AOD heat times varied between 1.1 and 3.8 hours
(charge through tap), and the weight of metal produced per heat
varied between 27.9 and 35.0 Mg (30.8 and 38.6 tons). The
variations in heat times were related in part to delays caused by
one or more cranes that did not operate properly.
Production rates were calculated to determine if the tests
were conducted during representative operating conditions. The
tested production rate for each furnace was determined by divid-
ing the total weight of metal produced during a test by the
sampling time. The metal produced by a given furnace during a
test was calculated by first dividing the minutes of normal
operation actually sampled (from charging through tapping, not.
including delays or patching) by the total minutes of normal
operation in the heat, and then multiplying by the weight of
metal produced during the entire heat. For comparison, the
normal production rate for all four furnaces operating at once
was calculated at 31.4 Mg/h (34.6 tons/h), based on an average
heat time of 3.7 hours and an average metal production of 29 Mg
(32 tons) per furnace. The same heat time was used for the EAF's
and AOD's, because one AOD heat normally occurs for each EAF
heat. The tested production rate and the normal production rate
both include the effects of delays and intervals between heats
and should therefore be comparable. The average equivalent
2-5
-------
production rate for the test series was 29.1 Mg/h (32.1 tons/h),
which indicated that the tests were conducted while the process
was operating at 93 percent of normal production rate.
Tests were delayed at various times during the series to
avoid sampling when emissions were not representative of normal
operation. On April 7, the test was stopped for 10 minutes
because EAF No. 89 was idle for 33 minutes waiting for the pit
crane to clean the tap runner. At 3:00 p.m. on April 8, the test
was stopped for approximately 15 minutes because two of the
furnaces were not operating. On April 8, the test was stopped
before the end of the heat in EAF No. 89 because the furnace bay
crane lost power. On April 9, the start of the test was delayed
for about 30 minutes because the furnace bay crane temporarily
lost power again.
The fabric filter was operating normally during the test
period. The fabric filter control panels were monitored hourly
for the duration of the tests. Test personnel periodically
closed off a baghouse compartment for a short period to move
their sampling equipment. Because one compartment is always in
the cleaning mode, this meant two compartments were not in
operation during that short period of time. The pressure drop in
each compartment, the inlet gas temperature, the fan amperage for
each of the three fans, and the number of compartments cleaning
or closed off were observed to assure normal fabric filter opera-
tion. The pressure drop varied between 1.0 and 1.5 kilopascals
(4 and 6 in. H2O) throughout the tests. All three fans were
2-6
-------
operating at full power during the tests. The flue gas inlet
temperature fluctuated from 43° to 54°C (100° to 130°F) during
the test period.
The test conditions were representative of normal plant
operation, and the data should be useful.
During the tests, emissions periodically filled the melt
shop. Some of the emissions were deflected when cranes passed
above the furnaces and vessels and when the cranes were in
position for a charge or tap. The cross drafts that developed
from open doors in the scrap bay and tapping pit also sometimes
deflected emissions from the hood. The inoperable automatic
dampers on the AOD hoods were always open and prevented their
optimal use. Although these dampers could be operated manually,
they were left open throughout the tests in line with normal
operating procedure. Although the emissions were not always
completely captured, the shop always cleared in 5 to 10 minutes.
Plant workers indicated that emission capture was better when the
capture and control system was new and the automatic AOD dampers
were operating properly.
Emissions generated by the EAF's were greatest during melt-
down, but emissions were also significant during tapping. Emis-
sions from the AOD were greatest during the initial stages of the
heat when the oxygen concentration in the blowing gases was the
highest. Emissions from the AOD during blowing appeared to be
equal to or greater than EAF meltdown emissions. For long periods
of time, however, the AOD vessel is in a nonblowing position
2-7
-------
while operators wait for sample results to determine what alloys
are needed and how much more blowing is necessary to meet final
specifications.
2-8
-------
SECTION 3
SUMMARY OF RESULTS
This section details results obtained from the emission test
program. Particulate matter tests were run simultaneously at two
inlet sites and the fabric filter outlet, and particle size
distribution tests were run at one inlet site and at the outlet.
Visible emissions from the melt shop and fabric filter outlet
were evaluated concurrently with particulate test runs, as were
fugitive emissions from the fabric filter. Fabric filter dust
samples were collected and analyzed for trace elements and
particle size distribution. Table 3-1 summarizes the type and
number of samples that were collected.
In brief, uncontrolled particulate matter concentrations
averaged 245 milligrams per dry normal cubic meter (mg/dNm ) at
20°C and 101 kilopascals (kPa), or 0.1076 grains per dry standard
cubic foot (gr/dscf) at 68°F and 29.92 in.Hg. At the outlet,
particulate concentration averaged 3.46 mg/d$m (0.0015 gr/dscf),
to yield a 98.6 percent control efficiency. Both levels of
concentration were in the range of expected values, which were
1 2
based on previously reported data on EAF's. '
Individual 6-minute set averages of visible emissions from
the melt shop ranged from 0 to 15 percent opacity during charging
3-1
-------
TABLE 3-1. SAMPLES COLLECTED AT AL TECH SPECIALTY STEEL
Sampling site
No. 1 - North
No. 2 - South
No. 3 -
Fabric fil-
ter outlet
Shop exit
Fabric filter
dust handling
system
Sample type
Particulate
Particle
size
Particulate
Particulate
Particle
size
VE
VE
FE
Dust
Sampling
method
EPA 5
High-capacity
impactor
Impactor
EPA 5
Modified EPA 5
Impactor
EPA 9
EPA 9
EPA 22
Grab
Number3
of
samples
3
3
3
3
3
3
2C
2C
2c
3
....
Time for
each sample
~ 6 h
~ 3 h
-20 min
- 6 h
~ 5-1/2 h
~ 5 h
- 7 h
~ 7 h
~ 5 h
1 per
day
Additional analysis
Type
Condensibles
b
Condensibles
Organic and
inorqanic
Condensibles
Trace metals,
particle
size^
No.
3
3
3
3
Method
Gravimetric
Gravimetric
Back-half E/C
extract
SSMS,d Coulter
I
ro
Does not include preliminary, blank, or duplicate runs.
Two samples were analyzed later for fluoride by EPA Method 13B and for chromium, lead, and nickel by atomic
absorption.
GThe third run could not be performed because of unfavorable weather conditions.
Spark source mass spectroscopy.
-------
and tapping operations. Average opacities of individual sets
ranged from 0 to 22 percent during refining and other process
operations. Visible emissions from the fabric filter outlet were
zero percent opacity, even after compartment cleaning cycles.
These and other results are presented and discussed in
detail in the following sections. Results are grouped by emis-
sion type. The sections in each group describe the sampling
scheme used at each site, summarize data and results, and discuss
the results.
3.1 PARTICULATE MATTER
Two inlet sites and the fabric filter outlet were tested
simultaneously. Site 1 represented emissions from the north EAF
(designated by the plant as No. 89) and the north AOD; Site 2
represented emissions from the south pair of furnaces (including
EAF No. 90); and Site 3 represented fabric filter outlet emis-
sions.
3.1.1 Sampling Scheme
Tests at all sites commenced simultaneously and ran for
approximately the same time until respective traverses were
completed, about 5 to 6 hours. This procedure enabled calcula-
tion of control efficiencies and emission factors. Each test run
at Site 1 was to have included two integral EAF heats (from
initial charge through the final tap) of approximately 3 to 3.5
hours each and two complete AOD heats of approximately 1.5 to 2
hours each. Because a malfunctioning crane caused shifts in the
3-3
-------
normal sequence of operation, the second and third tests were
stopped prior to completion of the second EAF heat, and the third
test included only portions of the AOD heats. Nevertheless, the
A
actual tests were generally representative of integral EAF heats.
So that tests at Site 2 could be conducted simultaneously with
other tests, the sample times were not coordinated to include an
integral number of heats, as the operating schedule of the south
furnaces was staggered from that of the north furnaces. Outlet
tests at Site 3 also could not be coordinated to represent an
integral number of heats (because of the different furnace
schedules), but they were conducted in conjunction with tests at
other sites. Fabric filter cleaning cycles were sampled in the
normal fashion as they occurred.
The NSPS contractor representative, who was on site to
monitor process operations, assisted in the coordination of tests
with process conditions. Based on his observations, tests were
interrupted if one of the EAF's experienced an operational delay
of longer than 20 minutes. Delays from 1 to 1.5 hours could be
tolerated for the AOD's without interrupting tests, because that
amount of AOD downtime would normally occur within the time frame
of an EAF heat.
Particulate matter sampling and analytical procedures at
both inlet sites followed those described in EPA Methods 1, 2, 3,
and 5 of the Federal Register.* A Method 5 sampling train and
40 CFR 60, Appendix A, July 1, 1980.
3-4
-------
analytical procedures were used on the outlet tests, but the
sampling procedures were modified by sampling at a constant rate,
at a site that did not meet minimum Method 1 criteria, and at
fewer points than specified by Method 1. The sampling rate was
based on the estimated average velocity of the gas stream at the
sampling location, which was calculated by dividing the total
inlet flow rate measured by Method 2 by the total exhaust area
represented by the sampling cross section. Three tests were run
at each site. Integrated gas samples were collected once at each
site (according to Method 3) to verify that the gas streams were
essentially air. Additional molecular weight determinations were
not made.
3.1.2 Gas Conditions and Particulate Emissions
Summaries of the measured stack gas and particulate emission
data are presented in Tables 3-2 and 3-3. Volumetric flow rates
are expressed in actual cubic meters per second (m /s) and actual
cubic feet per minute (acfm) at stack conditions. Flow rates
corrected to zero percent moisture and standard conditions [20°C
and 101 kPa (68°F and 29.92 in.Hg)] are expressed as dry normal
cubic meters per second (dNm /s) and dry standard cubic feet per
minute (dscfm). Average stack gas velocities are expressed in
actual meters per second (m/s) and actual feet per second (ft/s)
at stack conditions. Particulate concentrations in Table 3-3
are reported in milligrams per dry normal cubic meter and
grains per dry standard cubic foot. Emission rates are expressed
in kilograms per hour and pounds per hour. The product of the
3-5
-------
TABLE 3-2. SUMMARY OF GAS STREAM CHARACTERISTICS*
Run No.
AIP-1
AIP-2
AIP-3
Date .
(1981)
4/7
4/8
4/9
Average
A2P-1
A2P-2
A2P-3
4/7
4/8
4/9
Average
A3P-1
A3P-2
A3P-3
4/7
4/8
4/9
Average
Flow rate
dNm3/s
130.0
121.8
127.2
126.3
129.9
129.3
130.4
129.9
259.9
251.1
257.6
256.2
dscfm
275,500
258,200
269,600
267,800
275,200
273,900
276,300
275,100
550,700e
532,100
545,900
542,900
Temperature
°C
43
47
42
44
46
48
45
46
42
48
46
45
°F
110
117
107
111
116
118
114
116
108
119
114
114
Moisture,
%
0.81
0.89
1.25
0.98
0.65
0.69
1.29
0.88
0.41
0.68
1.17
0.75
Velocity0
m/s
19.1
18.2
19.0
18.8
19.3
19.3
19.7
19.4
2.4
2.4
2.5
2.4
ft/s
62.8
59.8
62.2
61.6
63.3
63.4
64.7
63.8
3. Of
7.9
8.2
8.0
Flow rate
m3/s
139.6
133.0
138.4
137.0
140.7
141.1
143.9
141.9
275.7
273.0
282.0
276.9
acfm
295,800
281 ,900
293,300
290,300
298,100
298,900
304,800
306,600
584,100f
578,600
597,600
586,800
Average C0£ <0.2%, 02 = 20.3%.
Dry normal cubic meters per second at 20°C and 101 kPa, and dry standard
cubic feet per minute at 68°F and 29.92 in.Hg.
cVelocity at stack conditions.
Flow rate at stack conditions.
g
The outlet standard flow rates were determined by summing flows at both
inlet sites.
Outlet gas velocities and actual flow rates are based on measured inlet flow
rates converted to outlet conditions.
3-6
-------
TABLE 3-3. SUMMARY OF FILTERABLE PARTICULATE EMISSIONS
Site
Inlet No. 1
north
furnaces
Run
No.
'A1P-1
A1P-2
A1P-3
Average
Inlet No. 2
south
furnaces
A2P-1
A2P-2
A2P-3
Average
Combined
inletsb
1
2
3
Average
Outlet No. 3C
A3P-1
A3P-2
A3P-3
Average
Concentration9
mg/dNm3
217
182
523
307
198
240
111
183
208
212
315
245
3.27
2.88
4.24
3.46
gr/dscf
0.0948
0.0794
0.2288
0.1343
0.0867
0.1047
0.0485
0.0800
0.0908
0.0924
0.1397
0.1076
0.00143
0.00126
0.00185
0.00151
Mass emission rate
kg/h
102
80
240
140
93
111
52
86
194
191
292
226
3.06
2.61
3.94
3.20
Ib/h
224
176
529
309
205
246
115
188
428
421
643
498
6.75
5.75
8.68
7.06
Isokinetic
rate, %
108
102
102
104
104
105
!07d
106
102
Milligrams per dry normal cubic meter at 20°C and 101 kPa, and grains per
dry standard cubic foot at 68°F and 29.92 in.Hq.
L J
Combined inlet mass emission rates represent the sum of results for Sites 1
and 2; concentrations are weighted averages based on standard flow rates.
C0utlet mass emission rates are based on measured outlet concentrations and
total inlet standard flow rates.
Isokinetic sampling rates for outlet runs were calculated by using average
gas velocities that are based on inlet flow rates converted to outlet
conditions.
3-7
-------
concentration and the volumetric flow rate is the mass emission
rate. The particulate data represent filterable material col-
lected in the sample probe and on the filter, both of which were
heated to approximately 121°C (250°F). The isokinetic rate is
the ratio of the velocity of the sample gas stream entering the
nozzle to the local stack gas velocity, expressed as a percent-
age.
The volumetric flow rates at inlet Sites 1 and 2 averaged
126 dNm3/s (268,000 dscfm) and 130 dHm3/s (275,000 dscfm).
The outlet flow rate averaged 256 dNm /s (543,000 dscfm). The
flow for each outlet run represents the sum of the standard flow
rates measured at the two inlet sites.
The actual flow rate at inlet Site 1 averaged 137 m /s
(290,000 acfm) at 44°C (111°F) and less than 1 percent moisture,
and was equivalent to a gas velocity of 19 m/s (62 ft/s). The
actual flow rate at inlet Site 2 averaged 142 m /s (301,000 acfm)
at 46°C (116°F) and less than 1 percent moisture, which equalled
a gas velocity of 19 m/s (64 ft/s). The outlet flow rate aver-
aged 277 m3/s (587,000 acfm) at 45°C (114°F) and less than 1
percent moisture, which represented an average gas velocity of
2.4 m/s (8.0 ft/s) at the sampling cross section. Outlet veloci-
ties and flow rates are based on inlet standard flows and mea-
sured outlet stack conditions of temperature, pressure, and
moisture.
For calculation purposes the stack gases were essentially
air. For one run at each site, the carbon dioxide (C02)
3-8
-------
concentration averaged less than 0.2 percent, and the oxygen con-
centration was 20.3 percent by volume.
The combined inlet concentration of filterable particulate
matter averaged 245 mg/dNm (0.1076 gr/dscf). This is the
average of concentrations measured at both inlet sites, weighted
according to measured standard flow rates. At inlet Site 1 the
average particulate concentration was 307 mg/dNm (0.1343
gr/dscf), and at Site 2 the average particulate concentration was
183 mg/dNm (0.0800 gr/dscf). These concentrations correspond to
mass emission rates of 140 kg/h (309 Ib/h) at Site 1, 86 kg/h
(188 Ib/h) at Site 2, and 226 kg/h (498 Ib/h) total uncontrolled
emissions.
The outlet particulate concentration averaged 3.46 mg/dNm
(0.00151 gr/dscf), with a corresponding mass emission rate of
3.20 kg/h (7.06 Ib/h). The concentration results for each run
are actually representative of the exhaust stream from four of
the eight fabric filter compartments, but emission rates are
based on the total system flow.
All isokinetic sampling rates were between 102 and 108
percent. The outlet values are based on average gas velocities
at the sampling cross section, which were calculated from mea-
sured inlet flow rates and outlet stack conditions.
3.1.3 Control Efficiencies and Emission Factors
Control efficiencies were calculated by dividing the dif-
ference between the outlet and weighted inlet particulate con-
centrations by the weighted inlet value. Table 3-4 presents a
3-9
-------
TABLE 3-4. FILTERABLE PARTICULATE COLLECTION EFFICIENCY
Run
1
2
3
Average
Inlet concentration3'
mg/dNirr
208
212
315
245
gr/dscf
0.0908
0.0924
0.1397
0.1076
Outlet concentration
mg/ dNnP
3.27
2.88
4.24
3.46
gr/dscf
0.00143
0.00126
0.00185
0.00151
% efficiency
98.4
98.6
98.7
98.6
U)
M
O
Weighted average from Sites 1 and 2.
Milligrams per dry normal cubic meter at 20°C and 101 kPa, and grains per dry standard cubic foot at
68°F and 29.92 in.Hg.
cPercent efficiency = Cin1et - Coutlet
'inlet
x 100.
-------
summary of filterable particulate concentrations and indicates
the fabric filter collection efficiency for each run. Control
efficiencies were 98.4, 98.6, and 98.7 percent on the three test
days.
Table 3-5 presents particulate emission factors for uncon-
trolled and controlled emissions, which were calculated by
dividing the appropriate hourly mass emission rate by the corre-
sponding furnace metal capacity. Results are reported in kilo-
grams per hour per megagram of furnace capacity (kg/h per Mg)
and in pounds per hour per ton (Ib/h per ton). Based on a total
capacity of 109 Mg (120 tons) for the four furnaces, the emission
factor for uncontrolled emissions averaged 2.1 kg/h per Mg, or
4.2 Ib/h per ton. It was possible for two runs at Site 1 to be
conducted over an integral number of heats. Each run consisted
of two EAF heats and two AOD heats. Emission factors for these
tests should be more representative of uncontrolled emissions on
a per-heat basis. Results for the two runs averaged 1.67 kg/h
per Mg (3.33 Ib/h per ton) based on a metal capacity of 54.5 Mg
(60 tons ). The average controlled emission factor for all four
furnaces was 0.03 kg/h per Mg (0.06 Ib/h per ton).
Emission factors shown in Table 3-6 are based on actual
production data. Results were calculated by dividing the fil-
terable mass emission rate by the corresponding equivalent
tested production rate. Emission factors are reported in kilo-
grams per megagram (pounds per ton) of metal produced. The
average uncontrolled emission factor was 7.8 kg/Mg (15.6 Ib/ton),
3-11
-------
TABLE 3-5. PARTICULATE EMISSION FACTORS'
Run No.
AlP-lb
1
AlP-2b
2
3
Average
Uncontrolled
kg/h per Mg
1.88
1.78
1.47
1.75
2.08
2.07
(1.67)
Ib/h per ton
3.73
3.57
2.93
3.51
5.36
4.15
(3.33)
Run No.
1
2
3
Average
Controlled
kg/h per Mg
0.028
0.024
0.036
0.029
Ib/h per ton
0.056
0.048
0.072
0.059
Factors are based on emissions per unit of furnace metal capacity in kilograms
per hour per megagram (pounds per hour per ton).
Tests were conducted for an integral number of heats and represent emissions
from the north furnaces at a metal capacity of 54.4 Mg (60 tons). All other
runs represent total emissions from the four furnaces at a metal capacity of
109 Mg (120 tons).
'Values in parentheses ( ) are for the integral heat tests on the north
furnaces.
3-12
-------
TABLE 3-6. PARTICULATE EMISSION FACTORS BASED ON PRODUCTION1
Run No.
1
2
3
Average
Metal production
rateb
Mg/h
28.9
30.2
28.2
29.1
tons/h
31.9
33.3
31.1
32.1
Emission factor0
Uncontrolled
kg/Mg
6.7
6.3
10.4
7.8
Ib/ton
13.4
12.6
20.7
15.6
Controlled
kg/Mg
0.11
0.09
0.14
0.11
Ib/ton
0.21
0.17
0.28
0.22
Calculated by dividing the filterable mass emission rate by the corresponding
average metal production rate.
bFrom Table 2-1.
cKilograms per megagram (pounds per ton) of metal produced.
3-13
-------
based on a production rate of 29.1 Mg/h (32.1 ton/h). At the
same production rate, controlled emissions averaged 0.11 kg/Mg
(0.22 Ib/ton).
3.1.4 Discussion
In general, the particulate tests were conducted according
to schedule. No problems were encountered with the sampling
equipment, and the few problems associated with the process
operation were considered minor. The report does not include the
results of preliminary tests that were conducted at each site to
compare particulate loadings with estimated sampling times and to
eliminate any problems associated with test coordination or
physical sampling maneuvers. This section discusses validity of
results, deviations in test methods and calculations caused by
the fabric filter site configuration, and effects of process
operations.
The primary purpose of the long sampling time was twofold:
(1) to collect approximately 25 to 50 mg in the front-half of the
outlet sampling train so as to minimize handling and weighing
errors; and (2) to satisfy NSPS minimum requirements for sample
time and volume. The actual catch weights were between 17 and 25
mg, which were considered satisfactory. The actual minimum
sampling time and volume were 5.3 hours and 5.9 dNm (208 dscf).
These met the minimum criteria of 4 hours and 4.5 dNm (160 dscf)
set forth in Subpart AA of the Federal Register.*
*
40 CFR 60, Subpart AA, July 1, 1980.
3-14
-------
The back-halves of each inlet and outlet run were analyzed
for condensible matter. These results are included in the com-
puter printouts in Appendix A, but are not summarized here
because they are considered to be biased. The results do not
agree well with expected values based on previously reported
tests at similar installations. Probable cause of the biases is
believed to be the long sample line used between the heated
filter and first impinger, even though it was Teflon-lined.
The inlet flow rates measured at Sites 1 and 2 were within
6 percent of each other, which indicated that the three induced-
draft (I.D.) fans were operating effectively to promote the
equal capture of emissions from each half of the melt shop. The
sum of the inlet flows compared very well with the system design
flow rate of 280 m /s (600,000 acfm), as all runs were within 4
percent of this .value.
The outlet volumetric flow rates, dry and at standard con-
ditions, were assumed to be equal to the sums of the inlet
standard flow rates. This assumption was necessary because the
site configuration made it impossible to obtain accurate velocity
data at the outlet. Flows at stack conditions were calculated
from the standard flow rates by use of measured values of tem-
perature, pressure, and moisture at the outlet.
The procedure just described is typical for tests at posi-
tive-pressure fabric filters without stacks, and assumes no air
leakage. Air inleakage at the I.D. fans was probably no more
than 10 percent, as evidenced by the agreement between measured
3-15
-------
and design flows. In addition, air leakage through the open
grating at the bottom level of the fabric filter was minimized by
covering the openings during the test period. At Al Tech these
open gratings represented sources of dilution air. Comparison of
inlet and outlet moisture contents seemed to indicate some
dilution, but the limits of accuracy for the moisture determina-
tions at the 1 to 2 percent level are probably greater than the
reported differences. The close agreement between inlet and
outlet gas temperatures indicated that any entry of dilution air
was negligible. A significant inflow of ambient air would have
caused a temperature decrease, which was not evident. For these
reasons, the amount of dilution air that entered the system was
considered to be minimal, and results reported for the outlet
flow rates should be representative of actual conditions.
The fabric filter site configuration required that several
modifications be made to EPA reference methods for their use in
the outlet tests. Although they could not be analyzed precisely,
the effects of these deviations on outlet particulate concentra-
tion results were considered to be relatively minimal. The two
deviations from Method 1 were the use of a sampling location less
than two equivalent duct diameters downstream from the nearest
disturbance and sampling at fewer than the minimum number of
points. The three deviations from Method 5 were the lack of
velocity monitoring at individual sampling points, use of a
constant sampling rate at all points during a given test, and
traversing only half of the large exhaust area for each test.
3-16
-------
These deviations, which generally would apply when testing
positive-pressure fabric filters without stacks, are discussed
in the following paragraphs.
The outlet site configuration did not provide any sampling
location capable of meeting minimum Method 1 criteria. The
throat of the monovent was selected as the sampling location,
primarily because it was the smallest cross-sectional area avail-
able. This location would be less prone to concentration biases
caused by faulty bags, and would provide the highest gas veloci-
ties.
The reason for sampling four rather than eight points per
compartment was to lessen the possibility of biasing results.
Because moving from one traverse point to another during sampling
required test personnel to enter a compartment while the fabric
filter was operating, extraneous dust could have been stirred up
by bumping the probe against nearby beams or by personnel activ-
ity, and results could have been biased if any such dust had
entered the nozzle during sampling. By sampling only four com-
partments and using extreme care during point changes, we were
able to avoid these potential problems in all of the tests.
It should be noted that a better sampling approach for this
type of fabric filter configuration would be to sample from out-
side on the baghouse roof and to use ports located in the mono-
vent throat. This would reduce the possibility of sampling
extraneous dust and shorten the time required to change traverse
locations. More points and compartments per run could then be
3-17
-------
sampled. Because safe access to the roof was not readily avail-
able, ports were not installed at this plant.
A constant sampling rate was used for the outlet tests
because individual point velocities could not be accurately
measured to make isokinetic sampling rate adjustments. The
preliminary traverse of three compartments verified the inac-
curacy of velocity measurement attempts. Most of the very low
velocity heads were readable on an expanded scale manometer;
however, turbulence at the low flows caused fluctuations and
frequent negative readings. Thus, an average gas velocity at the
outlet sampling location was calculated from previous data
obtained at the two inlet sites and used to set a constant sam-
pling rate. The two inlet flow rates were calculated and
totaled; then the sum was converted from standard conditions of
temperature and pressure to outlet conditions. This flow was
then divided by the total cross-sectional area of all eight
compartments represented by the sampling plane. The resultant
average gas velocity was assumed to represent each traverse
point. For each outlet test, inlet data obtained from the pre-
vious day's activities were used to estimate the average veloc-
ity. The constant isokinetic sampling rate to be used that day
was then calculated from this estimated velocity. For data
recording and computer calculation purposes, an equivalent
velocity head was calculated and entered on the field data sheets
for each traverse point as if it had actually been measured.
3-18
-------
Other parameters entered on the data sheets were measured accord-
ing to normal procedures.
All outlet computer calculations in Appendix A for flow
rate, emission rate, velocity, and isokinetics are based on these
estimated velocities; however, results reported in tables and
text have been adjusted to reflect measured rather than estimated
values. This was accomplished by dividing the measured total
inlet flow rate by the flow rate used initially to estimate the
outlet velocity. Results were adjusted by applying this ratio,
either directly or inversely as appropriate, to computer outputs
that were based on estimated values. These calculations are
shown in Appendix A.
The average isokinetics indicated for the outlet tests were
all within the acceptable range (100 + 10 percent). These iso-
kinetic calculations included the assumption that the average
velocity for the four compartments tested per run was approxi-
mately equal to the average velocity for all eight compartments.
We believe this is a reasonable assumption (+_ 10 percent),
although the middle compartments may have had slightly higher
velocities than the end ones. Another factor affecting iso-
kinetic and velocity results was the fabric filter cleaning cycle
time. Reported results do not account for the time when only
seven compartments were operating instead of eight, but the
greatest effect would occur if one compartment were always off
line. This would reduce reported outlet isokinetics by a factor
of 0.875 and increase gas velocities by the inverse of this
3-19
-------
factor. Isokinetics would be between 89 and 94 percent; there-
fore, emission results should not be affected significantly by
this consideration.
Although average outlet isokinetics were acceptable, values
at individual traverse points could be much different, depending
on the local gas velocities. The range of actual isokinetic
variation was difficult to gauge without valid point velocity
data, but an estimate of plus or minus 50 percent seems realis-
tic. The overall isokinetic rates, however, were within speci-
fied limits, and any point-specific biases should tend to be
averaged. Even so, it is expected that the precision of the
method used (constant sampling rate) is less than that of Method
5, which requires isokinetic sampling at each point. Results
should be viewed in this light, even though they were fairly
consistent and appear to be representative.
Evaluation of the process data furnished by MRI and actual
sampling times indicated that operation conditions were repre-
sentative during the tests. The equivalent tested production
rate averaged 93 percent of the maximum production rate of the
four furnaces. Although the metal production of individual
furnaces varied considerably, the overall production rates for
each test were within 10 percent of each other. A periodically
malfunctioning crane caused a shift in the normal sequence of
furnace operations and several extended furnace delays. Tests
were interrupted when the emissions were considered to be sig-
nificantly affected. The results of Tests 1 and 2 did not seem
3-20
-------
to be affected by the delays or variation in furnace operations,
but Test 3 results may have been. The concentration at Site 1
increased and the concentration at Site 2 decreased, but these
changes could not be related to specific process activities.
Indeed, the equivalent production rates for Test 3 showed the
reverse: a decrease for Site 1 furnaces and an increase for Site
2 furnaces. The overall increase in uncontrolled emissions was
verified by the corresponding increase in outlet emissions, but
the overall production rate for Test 3 was slightly lower than
the previous tests. The increase in Test 3 emissions caused a 12
percent increase in the average outlet concentration, which was
considered relatively insignificant. Overall results were
therefore taken as being representative of normal operating
conditions.
3.2 PARTICLE SIZE
Tests for particle size distribution were conducted at Site
1 (the north furnaces) to represent uncontrolled emissions, and
at Site 3 to represent controlled emissions. These tests were
performed in conjunction with particulate matter tests.
3.2.1 Sampling Scheme
Inlet particle size tests were conducted over an entire heat
cycle to represent average emissions, and during shorter inter-
vals, to represent different process modes. Tests during inte-
gral heats were initiated at the beginning of a charge for
either the North AOD or EAF No. 89, whichever occurred first.
3-21
-------
Tests continued for one heat and concluded at the end of tapping
operations at EAF No. 89, which yielded a sampling time of
approximately 3 to 3.5 hours. This time should have included an
entire ADD heat of between 1.5 and 2 hours. Because a malfunc-
tioning crane caused as shift in the sequence of normal opera-
tions, the first test included almost two ADD heats, and the
third test included portions of two different AOD heats. The
shorter particle size runs were performed at various times during
the second half of each particulate test. The sampling times for
these shorter tests were approximately 20 minutes, adjusted as
necessary to obtain proper loadings.
Outlet particle size distribution samples were collected
simultaneously with each particulate test, which yielded a 5-
hour sampling time for each run. Each test was conducted in the
same four fabric filter compartments as the coinciding particu-
late tests, but at different times to minimize interferences.
Because of the overlapping schedules between north and south
furnaces, no attempt was made to represent integral heats.
Fabric filter cleaning cycles were sampled as they occurred.
The integral heat runs at the inlet were coordinated with
process operations ^.with the assistance of the NSPS contractor
representative. Based on his observations, tests were inter-
rupted as necessary to avoid sampling during unrepresentative
conditions.
Andersen Mark III Cascade Impactors were used to collect the
shorter inlet samples and all of the outlet samples, and an
3-22
-------
Andersen Heavy Grain Loading Impactor was used to obtain integral
heat samples. All inlet samples were collected at an average
velocity point in the north furnace duct. Velocity data were
obtained periodically during each run by the use of Method 2
equipment. Outlet samples were run in duplicate, with each
impactor positioned at one of the four sampling points per
compartment. The sample time for each run was divided equally
among the same four compartments in which the particulate tests
were being conducted. Because accurate velocity data could not
be obtained at the outlet site, estimated velocities calculated
from previous inlet data were used to set approximate isokinetic
sampling rates. The results of three runs for each type of
sample are included in the report.
3.2.2 Particle Size Distributions and Fractional Efficiencies
Cumulative distribution curves represent the total weight
percent of particulate matter smaller than the indicated aerody-
namic particle diameter in micrometers. Each distribution curve
was plotted manually and represents the best-fit curve through
individual and average test data points. Each data point was
plotted manually and indicates both the 50 percent effective cut-
size of each impactor stage and the cumulative weight percent of
material collected in subsequent stages.
The three cut-points for each Andersen Heavy Grain Loading
Impactor test at Site 1 were determined graphically from informa-
tion supplied by the manufacturer. Cut-points for the eight Mark
III Impactor stages were calculated by computer programs
3-23
-------
contained in "A Computer-Based Cascade Impactor Data Reduction
System," (CIDRS) developed for EPA by Southern Research Institute
(SRI). All particle size results are based on a particle
density of one gram per cubic centimeter. Data reduction and
intermediate result calculations for both types of impactors were
performed by the CIDRS programs with moisture contents obtained
from simultaneous particulate tests. All calculations and
results are included in Appendix A.
Figure 3-1 shows the average cumulative distribution curve
for uncontrolled emissions over an entire heat cycle. Actual
results were limited to three cut-points between 1.6 and 13 ym,
but the curve was extrapolated down to a diameter of 0.5 ym for
comparison purposes. The average distribution indicated that 50
percent by weight of uncontrolled particulate emissions consisted
of particles with aerodynamic diameters of 0.55 ym or less.
Eighty percent by weight had diameters less than or equal to
10 ym.
Figure 3-2 shows the results of three runs conducted at Site
1 during various times of process operation. The three cumula-
tive distribution curves distinctly indicate a short-term varia-
tion of emissions. Run 1 represented EAF charging emissions, Run
2 represented emissions during the EAF melting phase, and Run 3
represented emissions from both an AOD and an EAF. The percent
by weight of emissions that had diameters equal to or smaller
than 10 ym varied among the three runs from 54 to 29 to 74 per-
cent, respectively. Only the run representing AOD and EAF
emissions was similar to average integral heat results.
3-24
-------
to
I
to
01
WJ
M.t
<-> t
1.1
irt
rtt
i iQ.EppHaai; JUCEJU.^
itt
ttr
—i
1.0 10.0
AERODYNAMIC PARTICLE SIZE, micrometers A1PS-1 O
A1PS-2 A
, A1PS-3 0
EXTRAPOLATED
Figure 3-1. Average particle size results for uncontrolled
emissions, Site No. 1.
100
-------
Ul
I
ro
r EAF MELT AND ADD
AERODYNAMIC PARTICLE SIZE, micrometers
A1PM-1A O
A1PM-2B A
A1PM-3B D
Figure 3-2. Particle size results for uncontrolled emissions
during various furnace operations, Site No. 1.
-------
Figure 3-3 shows the average distribution curve for the
outlet samples. Results indicated that 50 percent of the mass
emissions consisted of particles having aerodynamic diameters of
5 ym- or less. Sixty-six percent by weight had diameters smaller
than or equal to 10 ym.
Table 3-7 presents the fractional collection efficiencies
for various size ranges. Weight percents in each size range were
determined from the average inlet and outlet cumulative distri-
bution curves plotted in Figures 3-1 and 3-3. These percentages
were multiplied by the average inlet and outlet particulate
concentrations shown in Table 3-3 to calculate controlled and
uncontrolled mass loadings in the respective size ranges.
Fractional efficiencies were calculated for each size range by
dividing the difference between inlet and outlet concentrations
by the inlet values. Efficiency ranged from a high of 99.7 per-
cent for particles smaller than 0.5 ym in diameter to a minimum
of 96.2 percent for particles between 5 and 10 ym in diameter;
the overall efficiency was 98.6 percent.
3.2.3 Discussion of Results
Results of all the tests are not reported. Some runs were
not included because of poor isokinetics, undesirable stage
loadings, or problems related to impactor assembly.
When evaluating these results, one should remember that
particle sizes are in terms of aerodynamic diameters based on a
particle density of 1 g/cm . Cumulative distribution curves
based on physical diameters and actual density would be shifted
3-27
-------
•*.•
•*.•
tt.l
M
M
Ul
ISl
M Oft
«/> "
o
UJ
5"
*— 1
o
ae
*""" M
1 »
>fi
53 »
IE M
to
£ «•
W _ M
1 1°
fO t-
8
CUMULATIVE PERCEN
— M «i • 8
M
t.t
•.1
=
q
— J
3
J
L I.
t:.:::., il
L I
^Tt7 I'll JT3
iTF-i.: ^f-f?4- ?*f
—*— t :•• *- *JJ-
-*...».-«.- ..*. « , — ^,
— * ' • T * t • — *
-J! I-*.. — ~ — —
S^-
&d^^
-r^f-r^-
_*_...x.. :;
c
^
.ill..
iiin
1+- t^+ri
l—. — L L* — j-.
= ?i^ ^
+ £ 5 ffl
n 'i , min*
; "'t
Jggr^b; .= = g.
. i 1 4l J>Mt •
: rrj — -r: ~4- -f
» -*_^ ^_ -^ — — [7
r — Er = = =
;SE=^-=3
5t553"-^
•* 7**; ,X — T- * ^
. , • r
ilSSIl!!;
* ;' I1; '•• "•
Trttrssr^g
-. *^M f4. .*. -*. — .
^ „, ^ jit — — i
1.K llil ..4. — — I
' • ! ' '
null!
r r ~
. t •
_J...J..J
iL. _
IT* T
S rrr-
lj k« —
*• *»4- —
ifW
JT
ibi
...
*
U
!
,ii!
1
...
I.1
ill
_..!.
!
-r-r: -nft
... — u
-F
d
•H
M
r-
- -f
^4 • • .
: J±:
th^?±
=ifiS
| U -f4-J.
I I
---4-4-
i
f±t-trr
" T "^l
Ti
TJ
:m
t .lit U^r
r^Lr
•5nl
r ii
ii i|i
I L.
:$$
rl^
t-;i 2|
4- 4- • + ' '|i
::t^
rn
- *-*j* -*—
4> • . . *1»
I rTj-n;
:^rr
, i
;|l!
1
:T?
:fi
• Ii
•U-
T'
^
•Y1
-*i»
H-
^
~r*i
1 i
II'
]i
T
1_
\K
'!'
*t^
M— •
^
N
-
. 1
,
:
^r;
~.
— 14
— U
c;i
if..
-^
iiz n
I
^E =
m
1
(
1 ''!
li |(
.. *±i.
^.T
. \
-rt — ••
n •"•*:
JJ
9=^:
r*»f
t !1
ll'
***t
:
'!?- * i
-
x
-
,i
^
•-
^
^
._ I
___ I
T-- 4TI
Tt-
. i
fel:i±I
;--{•! -*t*f
rK^T^
ifH:il
— r-t-^ ' i i
III 3.T7T
: 1 1 1
-till -4,
: , t • U
--j-pTtir
..|. n UIJT
ijllti^lZ
ntrcrln:^:
— rtr— -n-~
r^1 f " if T"
; : i ii ' • ,;,
H^H^
4..I -£. ' '
-rtf 7— -r1 —
^2n;?tE=
^nr-jr-^.
— ^-* J^*- — —
( . .. 1.
*-4i-*tr ^T-
1 ' j t i ' [ .
mtc^i
rttr^T
-U^. m zti zn
*0.<-t.f«J — -.—I.,..
ill I -! it
in i ^^
,11'
•r~
=1
"
L.,.,iIl£ElLlIi
' illl
1
~W
ST'Tf
" |r
-LUii
-#
-.-v
>- r^i
Sis
,f 1
11 ill1
p^
ill
1 II
u IF
_.£ in
1 ii
r-4. ,*.
I'! :"•
[•i M
I
...
.J
-
Lili^L
-
-*t
_ * *
- t
_•
-
— s
— •
1.0 10.0 100
AERODYNAMIC PARTICLE SIZE, micrometers A3PS-1AO
A3PS-ZAA
A3PS-3AD
Figure 3-3. Average particle size results for controlled emissions, Site No. 3.
-------
TABLE 3-7. SUMMARY OF PARTICLE SIZE DISTRIBUTION AND FRACTIONAL EFFICIENCY
I
to
vo
Cumulative weight percent8 less Inlet
than larger stated size Outlet
Weight percent In stated size Inlet
range Outlet
Partlculate concentration , Inlet
In stated size range, mg/dNm Outlet
gr/dscf Inlet
Outlet
Fractional collection effi-
ciency0 In stated size range
Aerodynamic particle size range, micrometers
D<0.5
49
10
49
10
120
0.346
0.0527
0.0002
99.7
0.510 urn
-
20
34
49.0
1.18
0.0215
0.0005
97.6
Total
100
100
100
100
245
3.46
0.1076
0.0015
98.6
Weight percents are taken from plots of average distributions.
Partlculate concentration = total emissions x weight % in stated size range; concentrations given in terms
of milligrams per dry standard cubic meter at 20°C and 101 kPa and grains per dry standard cubic foot at
68°F and 29.92 in. Hg. Total concentrations were taken from Method 5 test results.
Collection efficiency = Inlet concentration - outlet concentration x 100
inlet concentration
-------
to the left toward smaller sizes if the actual density were
greater than 1 g/cm , and vice versa. A quick approximation of
the physical diameter can be obtained by dividing the reported
aerodynamic diameter by the square root of the actual particle
density. For example, the specific gravity of the fabric filter
dust samples was about 3.3 g/cm . Using this particle density
would increase the amount of controlled emissions smaller than
1 ym from roughly 20 to 30 cumulative weight percent.
As expected, results of the Mark III blank run at the outlet
indicated that stack gases did not react with the glass fiber
filter media to create false weight changes.
Results representing average uncontrolled emissions over an
entire heat cycle generally agreed with expected distributions
based on previous EAF data. Particulate concentrations indicated
by the first two distribution tests compared favorably with the
results of particulate tests conducted over similar time frames.
Results of the third distribution test did not agree with the
third particulate test, but this lack of agreement was believed
to be representative of the short-term variation in emissions.
The isokinetic sampling rates for the three runs (104, 100,
and 112 percent, respectively) were considered acceptable. The
Heavy Grain Loading Impactor sampling rates were all within the
limits suggested by the manufacturer, and results are believed to
be within generally expected limits of accuracy.
The particle size tests performed during various times of
process operation show that uncontrolled emissions are not
3-30
-------
consistent over the short term. Although these results gen-
erally can be related to process operations, one should remember
that the actual sampling times do not precisely coincide with
only one specific process mode. For example, Run 1 was conducted
during an EAF charge, but the time required to obtain an ade-
quate sample was longer than the charging operation. As a
result, Run 1 overlapped into the melting phase of EAF operation.
Also, the mode of ADD operation (and relative emission genera-
tion) during Run 3 was not clear. Evaluation of the appropriate
heat sheet indicated that the AOD may not have been generating
any emissions during the actual sampling period.
Comparison of particulate concentrations among the Mark III
runs showed disagreement with the relative observation that
emissions during EAF meltdown were greater than emissions during
an EAF charge. For this reason, and because of the dilution
effects when only one furnace was operating, particulate concen-
trations were not reported. The average isokinetic sampling
rates (107, 110, and 106 percent) were all considered acceptable,
and impactor sampling rates were within suggested operating
limits. In addition, the distributions are not considered to be
biased toward larger sizes by the use of glass fiber collection
4
media. This possibility was discounted, as the filter material
did not seem to increase the collection efficiencies of upper
impactor stages because most of the captured material was either
in the cyclone precutter or on the lower stages. Results,
3-31
-------
therefore, are taken to be representative of the variation in
uncontrolled emissions.
Outlet particle size distributions showed a higher number of
large particles than expected. It was expected that 80 to 90
cumulative weight percent of emissions would be of particles with
aerodynamic diameters of approximately 2.5 pm or less. One of
the several plausible explanations for the apparent discrepancy
is the possible bias caused by increased efficiency of the upper
4
impactor stages when glass fiber filters are used. Examination
of analytical results indicated that this may have occurred,
but it could not be verified. The adjustment to outlet results
for this type of bias would reduce the aerodynamic diameter from
5 to 3.5 ym at a cumulative weight of 50 percent; however, this
adjustment would not completely account for the difference
between actual and expected results.
A second explanation could be related to inaccuracies of the
method at very low sample weights; however, results are con-
sidered to be acceptable, based on several observations. The
particulate concentrations indicated by the particle size runs
were between 65 and 85 percent of results from simultaneous par-
ticulate tests, which is good agreement for the two different
methods. The increase in emissions indicated by the third par-
ticulate test is verified by a corresponding increase in particle
size emission results. The close agreement among the three
cumulative distribution curves is further evidence that particle
size results are representative. If the stage sample weights had
3-32
-------
been too low for accurate determinations, the three curves
probably would have shown more variability and flatter slopes.
All of these observations suggest that results were within
expected limits of accuracy for particle size distributions
tests.
The most likely explanation for the higher than expected
number of large particles is related to electrostatic charge.
Small charged particles could have agglomerated to form particles
of larger diameter. Evidence of a high electrostatic charge
inside the fabric filter was discovered while the open gratings
at the bottom level of the filter were being covered. For this
reason, results of particle size tests are considered to be
representative of actual conditions during the tests and reflect
possible particle agglomeration due to electrostatic charges.
The fractional efficiencies reported in Table 3-7 also could
be affected by the possible particle agglomeration just described.
Particle agglomeration would explain the relatively low control
efficiencies indicated for the larger particle sizes compared
with the abnormally higher efficiencies for smaller sizes. Large
particles formed after filtration would indicate false filtering
efficiencies for the different size classifications; however, the
overall efficiency would be unaffected.
3.3 VISIBLE AND FUGITIVE EMISSIONS
Visible emissions from the melt shop and fabric filter
monovent were evaluated simultaneously with particulate matter
3-33
-------
tests. In accordance with Method 9* procedures, emissions were
observed in 6-minute sets, and individual opacity readings were
recorded at 15-second intervals. Fugitive emissions from the
fabric filter were evaluated periodically throughout the test
series according to procedures outlined in the proposed Method
22.** Fugitive emissions were recorded as the cumulative minutes
of any emissions visually detectable during 20-minute observation
periods.
3.3.1 Results
Table 3-8 summarizes the visible emissions detected during
charging, tapping, and other furnace operations. The melt shop
was divided into a north and a south segment for evaluation
purposes. Visible emissions from the north segment were attrib-
uted to the north furnaces, and vice versa. When one of the
north furnaces was charging, only the emissions from the north
segment of the melt shop were considered. Tabulated emission
data for each furnace type and process mode include the total
number of data sets for both melt shop segments; for example, the
five data sets during Run 1 for EAF charging represented three
sets from the north furnaces and two from the south. During Run
1, individual 6-minute set averages ranged from 0 to 4 percent
opacity for charging and tapping operations. Average opacities
of individual sets ranged from 0 to 13 percent for refining and
other operations. During Run 2, individual set averages during
40 CFR 60, Appendix A, July 1, 1980.
**
Federal Register, Vol. 45, No. 224, November 18, 1980.
3-34
-------
TABLE 3-8. SUMMARY OF VISIBLE AND FUGITIVE EMISSIONS0
Melt shopb
Date
(1981)
4/7
4/8
Total
Run
No.
1
2
Furnace
type
EAF
ADD
EAF/AOD
EAF/AOD
EAF
ADD
EAF/AOD
EAF/AOD
Process
mode
Charge
Tap
Charge
Tap
Tap/Charge
Charge/Tap
Refining
and other
Charge
Tap
Charge
Tap
Refining
and other
Charging
and
tapping
Refining
and other
Number of
sets
5
5
4
2
1
1
60
8
5
4
3
74
38
134
Range of
readings,
% opacity
0
0
0-5
0
0-10
0
0-25
0-20
0-25
0-20
0-10
0-25
0-25
0-25
Range of
set averages,
% opacity
0
0
0-3
0
4
0
0-13
0-12
0-15
0-8
0-5
0-22
0-15
0-22
Fabric filter outlet
Number of
of sets
73
Range of readings,
% opacity
0
Range of set
averages, % opacity
0
Fugitive emissions from fabric filter
Accumulated observation
period, minutes
595
Minutes
0
Accumulated emission time,
% of observation period
0
Data were collected during 7 hours of process operation on April 7, 7
hours on April 8, and 20 minutes on April 9. Unfavorable weather conditions
prevented additional readings on April 9.
Each set of readings represents emissions from either the north or south
segment of the melt shop. Data for Run No. 3 could not be obtained because
of unfavorable weather conditions on April 9.
3-35
-------
charging and tapping ranged from 0 to 15 percent opacity.
Average opacities of individual sets ranged from 0 to 22 percent
during refining and other operations. No data were obtained
during Run 3 because of adverse weather conditions.
Table 3-9 lists the time and average opacity for each 6-
minute set of visible emissions data obtained at the melt shop.
The heat sheets in Appendix D were used to determine the times of
various process operations for comparison to emissions. None of
the emissions seemed to be caused by abnormal operations.
No visible emissions from the fabric filter monovent exhaust
were detected at any time during the test series, even after
compartment cleaning cycles. A total of 73 six-minute sets of
data were collected.
The fabric filter structure was observed for a total of 595
minutes. No fugitive emissions were detected at any time.
3.3.2 Discussion
The higher periods of melt shop emissions could be related
to daily activities that normally occur during the first part of
the day shift. No significant emissions were detected after 1130
hours on April 7 or 1230 hours on April 8, 1981. Examination of
the available process information did not yield support for this
hypothesis, however.
The low opacity data for the fabric filter outlet were
supported by the low particulate concentration results.
3-36
-------
TABLE 3-9. COMPARISON OF MELT SHOP VISIBLE EMISSIONS TO PROCESS OPERATION
Process mode3
EAF
N-T
N-C
NC, STi
S-CTL
S-T2B
K
s-cb
N-T
N-AE
N-C
N-RT-
N-RT
N-RT
L.
ADD
N-C
N-C
N-Tb
N-C
S-Cb
N-T
S-Tb
NC,STl&.SBB
S-T2b
j-i,
S-C
Run number, date,
set
Run No. 1
10:15 -
10:27 -
10:33 -
10:43 -
10:49 -
10:55 -
11:01 -
11:07 -
11:13 -
11:25 -
11:37 -
11:49 -
12:01 -
12:13 -
12:25 -
12:37 -
12:49 -
1:01 -
1:13 -
1:25 -
1:37 -
1:49 -
2:01 -
2:13 -
2:25 -
2:37 -
2:49 -
3:01 -
3:13 -
3:25 -
3:37 -
3:49 -
4:01 -
4:13 -
4:25 -
4:37 -
4:49 -
5:01 -
5:13 -
time
, 4/7/81
10:20 a.m.
10:32
10:38
10:48
10:54
11:00
11:06
11:12
11:18
11:30
11:42
11:54
12:06 p.m.
12:18
12:30
12:42
12:54
1:06
1:18
1:30
1:42
1:54
2:06
2:18
2:30
2:42
2:54
3:06
3:18
3:30
3:42
3:54
4:06
4:18
4:30
4:42
4:54
5:06
5:18
Average opacity,
%
North
0
4
3
0
0
13
9
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
South
0
4
3
12
11
13
0
2
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
(continued)
3-37
-------
TABLE 3-9 (continued)
Process modea
EAF
N-C
N-C
N-AE.S-TI
N-AE.S-BB
S-BB
S-BB
S-T2
N-RWP
N-RWP,S-C
N-RWP, S-C
N-C
N-T
N-C
S-Ti
S-BB
S-BB
S-T2
s-cb
S-AE
ADD
N-Cb
S-C
N-T
N-C
S-T
,i-,
S-C
Run number, date,
set
Run No. 2
9:45 -
9:51 -
9:57 -
10:03 -
10:09 -
10:15 -
10:21 -
10:27 -
10:33 -
10:39 -
10:45 -
10:51 -
10:57 -
11:09 -
11:15 -
11:20 -
11:27 -
11:33 -
11:39 -
11:45 -
11:51 -
11:57 -
12:03 -
12:09 -
12:19 -
12:25 -
12:37 -
12:49 -
1:01 -
1:13 -
1:25 -
1:37 -
1:49 -
2:01 -
2:13 -
2:25 -
2:37 -
2:49 -
3:01 -
3:15 -
3:27 -
3:39 -
time
, 4/8/81
9:50 a.m.
9:56
10:02
10:08
10:14
10:20
10:26
10:32
10:38
10:44
10:50
10:56
11:02
11:14
11:20
11:26
11:32
11:38
11:44
11:50
11:56
12:02 P.m.
12:08
12:14
12:24
12:30
12:42
12:54
1:06
1:18
1:30
1:42
1:54
2:06
2:18
2:30
2:42
2:54
3:06
3:20
3:32
3:44
Average opacity,
%
North
8
10
5
12
22
15
5
8
10
6
2
1
2
5
4
6
10
12
8
0
13
20
12
8
12
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
South
0
10
5
12
22
15
5
8
10
6
2
0
0
0
0
0
10
12
8
8
13
20
12
8
12
2
0
0
0
0
0
0
0
0
0
0
0
0
0
2
0
0
(continued)
3-30
-------
TABLE 3-9 (continued)
Process mode3
EAF
N-Cb
ADD
S-PS
S-PS
Run number, date,
set time
Run No. 2, 4/8/81
3:51 - 3:56 p.m.
4:03 - 4:08
4:15 - 4:20
4:27 - 4:32
4:39 - 4:44
Average opacity,
%
North
0
0
0
0
0
South
0
0
0
0
0
N = North furnace, S = South furnace, C = Charge, T = Tap, CTL = Clean tap
ladle, AE = Add or adjust electrode, RT = Repair tap spout, BB = Burn bottom,
RWP = Repair water pipe, PS = Patch seam.
^Process mode actually began 3 to 6 minutes prior to indicated set time.
3-39
-------
3.4 FABRIC FILTER DUST SAMPLES
Samples of dust collected by the fabric filter were obtained
daily from the dust-handling system just below the central junc-
tion of the screw conveyors. Samples were collected in a manner
that did not interfere with other ongoing tests. The laboratory
split each sample into two fractions for separate analyses of
trace elements by spark source mass spectroscopy (SSMS) and for
particle size distribution by Coulter Counter.
3.4.1 Trace Elements
Table 3-10 summarizes the results of SSMS analyses on the
three dust samples. Concentrations are given in micrograms of
element per gram of sample. Less than (<) and greater than (>)
marks are used to denote concentrations outside the quantifica-
tion limits for the particular element and sample analysis. The
minimum detection limits for the majority of elements ranged from
0.1 to 0.4 yg/g; major constituents are listed as >1000 yg/g.
Results for several elements are not reported, and indium was
added to each sample as an internal standard. Elements are
listed alphabetically for convenience. The analytical results
included in Appendix C are listed in order of decreasing atomic
number.
3.4.2 Particle Size Distribution
Figure 3-4 shows the best-fit cumulative distribution curve
for the three dust samples. This curve represents the weight
percent of particulate matter smaller than the indicated physical
particle diameter (in micrometers). Each data point was
3-40
-------
TABLE 3-10. SUMMARY OF TRACE ELEMENT ANALYSES ON
FABRIC FILTER DUST SAMPLES
Element
Aluminum
Antimony
Arsenic
Barium
Beryllium
Bismuth
Boron
Bromine
Cadmiun
Calcium
Carbon
Cerium
Cesium
Chlorine
Chromium
Cobalt
Copper
Dysprosium
Erbium
Europium
Fluorine
Gadolinium
Gallium
Germanium
Gold
Hafnium
Hoi mi urn
Hydrogen
Indium
Iodine
Iridium
Iron
Lanthanum
Lead
Lithium
Lutetium
Magnesium
Manganese
Mercury
Molybdenum
Neodymium
Concentration, yg/g (ppm weight)
Sample 1
>1000
160
420
>1000
0.3
350
42
170
150
>1000
NRa
21
7
>1000
>1000
340
>1000
<0.4
<0.4
<0.4
>1000
<0.4
660
100
O.4
3
<0.4
NR
STDb
2
<0.4
>IOOO
21
>IOOO
160
<0.4
>1000
>1000
NR
>1000
5
Sample 2
>1000
190
410
>1000
0.1
420
23
94
180
>1000
NR
19
14
>1000
>1000
190
>1000
<0.1
<0.1
0.4
>1000
<0.1
810
55
<0.1
1
<0.1
NR
STD
3
•*0.1
>1000
25
>IOOO
9
<0.1
>1000
>1000
NR
>1000
3
Sample 3
>1000
44
120
>1000
0.2
44
53
43
132
>1000
NR
6
3
>1000
>IOOO
430
>1000
<0.2
<0.2
<0.2
>1000
<0.2
330
41
*0.2
1
«0.2
NR
STD
3
•P0.2
>1000
6
>1000
79
1000
>1000
NR
>1000
1
(continued)
3-41
-------
TABLE 3-10 (continued)
El ement
Nickel
Niobium
Nitrogen
Osmium
Oyxgen
Palladium
Phosphorus
Platinum
Potassium
Praseodymium
Rhenium
Rhodium
Rubidium
Ruthenium
Samarium
Scandium
Selenium
Silicon
Silver
Sodium
Strontium
Sulfur
Tantalum
Tellurium
Terbium
Thallium
Thorium
Thulium
Tin
Titanium
Tungsten
Uranium
Vanadium
Ytterbium
Yttrium
Zinc
Zirconium
Concentration, yg/g (ppm weight)
Sample 1
>1000
190
NR
<0.4
NR
<0.4
>1000
<0.4
>1000
2
<0.4
<0.4
170
<0.4
4
1
120
>1000
<4
>1000
>1000
>1000
3
3
<0.4
12
8
<0.4
410
>1000
100
6
850
<0.4
15
>1000
130
Sample 2
>1000
56
NR
<0.1
NR
<0.1
840
<0.1
>1000
1
<0.1
<0.1
530
<0.1
2
0.1
66
>1000
66
>1000
>1000
>1000
<0.9
5
<0.1
<0.1
5
<0.1
220
710
54
3
r >1000
<0.1
8
>1000
39
Sample 3
>1000
52
NR
<0.2
NR
<0.2
970
<0.2
>1000
1
<0.2
<0.2
200
<0.2
2
0.5
28
>.1000
220
>1000
800
>1000
1
2
*0.2
2
2
<0.2
82
660
50
2
>1000
<0.2
16
>1000
36
Not reported.
""internal standard.
3-42
-------
t-*i
U)
«.i
PARTICLE SIZE, micrometers
Figure 3-4. Average particle size distribution of
fabric filter dust samples.
-------
plotted manually from differential distribution data reported by
the laboratory. The average curve indicated that 50 percent by
weight of collected dust consisted of particles with physical
diameters of 2.2 ym or less. Ninety-three percent by weight had
diameters of less than 10 ym.
3.4.3 Discussion
The concentrations of several trace elements seem to vary
considerably, and this could be related to different specifica-
tions of the metal in the furnaces. It should be noted, however,
that SSMS is more of a qualitative than quantitative analytical
technique. [The results of an audit sample (Section 5) bear this
out; they indicate that reported element concentrations are only
accurate within a factor of +3.]
When evaluating the particle size results, one should note
that the cumulative distribution curves are based on physical
diameters rather than aerodynamic diameters as reported for
emission tests. If desired, an approximation of the aerodynamic
diameters can be made by multiplying the reported physical
diameters by the square root of the actual particle density.
Using the specific gravity analysis results of 3.3 g/cm would
decrease the amount of dust smaller than 3 ym from approximately
70 to 35 cumulative weight percent.
The reported cumulative weight distributions also assume
that all particles have the same density. This assumption was
necessary to convert particle volume data measured by the Coulter
Counter to a weight basis.
3-44
-------
The Coulter Counter results were compared with a theoretical
size distribution based on emission test results at the inlet and
outlet sites. This theoretical cumulative weight curve was esti-
mated from the average size distributions, fractional efficien-
cies, and mass loadings listed in Table 3-7. The Coulter Counter
and theoretical curves both indicated that 75 cumulative weight
percent of the collected dust consisted of particles with aero-
dynamic diameters of approximately 6 jam or less. The curves
differed considerably at smaller sizes, but this may have re-
sulted from the agglomeration of particles in the outlet gas
stream.
The particle size distribution samples were originally sub-
jected to Banco analysis, but agglomeration of the particles
during analysis prevented an accurate determination.
3.5 SUPPLEMENTAL ANALYSES FOR FLUORIDE, CHROMIUM, LEAD, AND
NICKEL
Several outlet samples and fabric filter dust samples were
analyzed for particulate fluoride content by procedures described
in EPA Method 13B*, and for chromium, lead, and nickel content by
Atomic Absorption Spectrophotometry. These analyses were per-
formed subsequent to the completion of originally scheduled lab-
oratory work to better quantify emission levels indicated as
greater than 1000 yg/g by the SSMS analyses on the fabric filter
dust samples.
40 CFR 60, Appendix A, July 1, 1980.
3-45
-------
Separate fluoride analyses were performed on two acetone
rinses and one set of filters from the outlet particle size
samples, two of the fabric filter dust samples, and appropriate
blanks. Metal analyses were performed on two sets of outlet
particle size samples (acetone rinse and filters combined), two
dust samples, and appropriate blanks. The fourth outlet sample
was obtained from a duplicate test run.
Laboratory results of fluoride and metal analyses on the
outlet samples were reported as milligrams of pollutant. The
species concentration (in micrograms per gram) was calculated by
dividing the mass of pollutant by the mass of particulate matter
reported in earlier gravimetric analyses. The laboratory results
for dust samples were reported in concentrations of milligrams
per gram, which were easily converted to micrograms per gram.
The two concentration results of each species and sample type
were averaged. Average concentrations were multiplied by average
filterable particulate emission results to determine pollutant
gas stream concentrations, mass emission rates, and emission
factors. For this purpose, results of dust sample analyses were
assumed to be representative of uncontrolled emissions. Results
are summarized in Table 3-11.
The outlet fluoride results are based on acetone rinse
analyses only because the filter analysis had a high blank value
of fluoride. This was caused by the filter material, which was
glass fiber instead of paper as specified by Method 13B. The
total filter blank value was 2.3 times larger than the net
3-46
-------
TABLE 3-11. SUMMARY OF SUPPLEMENTAL ANALYSES FOR
FLUORIDE, CHROMIUM, LEAD, AND NICKEL
Uncontrolled emissions0
Pollutant
species
Fluoride
Chromium
Lead
Nickel
Concentration _
ug/g of solid
47,200
39,200
8,400
16,300
mg/dNnr3
12
9.6
2.1
4.0
gr/dscf
0.0051
0.0042
0.0009
0.0018
Emission rate
kg/h
11
8.9
1.9
3.7
Ib/h
24
20
4.2
8.1
Emission factors
kg/h/Mg
0.098
0.081
0.017
0.034
Ib/h/ton
0.20
0.16
0.035
0.068
kg/Mg
0.37
0.31
0.066
0.13
Ib/ton
0.74
0.61
0.13
0.25
Controlled emissionsb
Fluoride0
Chromium
Lead
Nickel
31,600
17,400
5,800
7,600
0.11
0.060
0.020
0.026
0.00005
0.00003
0.000009
0.00001
0.10
0.056
0.019
0.024
0.22
0.12
0.041
0.054
0.0009
0.0005
0.0002
0.0002
0.0019
0.0010
0.0003
0.0004
0.0035
0.0019
0.0006
0.0008
0.0070
0.0038
0.0013
0.0017
aBased on average uncontrolled particulate emissions and average of analyses on two dust samples, assuming
that the concentration in the uncontrolled gas stream is the same as in the collected dust.
Based on average controlled particulate emissions and an average of analyses on two outlet samples.
C8ased on analyses of acetone rinses only; the glass fiber filter analysis had a high blank weiaht
of fluoride and was not used.
-------
fluoride on the filter, which increased the possibility of error
in the results. Because the amount of fluoride in the acetone
rinse was much smaller than that indicated by the filter analy-
sis, a small error in the filter results would have a significant
impact on total fluoride. For these reasons, and because the
filter result indicated a much higher concentration of fluoride
than did the acetone rinse results, the filter analysis was
disregarded. If the filter result is correct, the outlet
fluoride concentration would be 56,700 yg/g instead of the 31,600
yg/g indicated in the table. This would not compare favorably
with the 47,200 yg/g of fluoride measured in the dust samples
because it would contradict the trend of the other results,
which indicate lower pollutant concentrations in the outlet
samples than in the dust samples.
3-48
-------
SECTION 4
SAMPLING SITES AND TEST METHODS
This section describes the sampling sites and outlines
the various test methods that were used to characterize par-
ticulate matter emissions, particle size distributions, visible
and fugitive emissions, and fabric filter dust samples. The
schematics of the air pollution control system presented in
Figures 4-1 and 4-2 identify the relative locations of each
sampling site. Figure 4-3 presents several photographs of the
control system configuration and sampling sites.
4.1 SITE 1—UNCONTROLLED NORTH EAF AND NORTH AOD
Uncontrolled emissions from the north furnaces were sampled
for particulate matter and particle size distribution. Site 1
was located in the 3.0-m (10-ft) diameter duct between the point
where the north EAF and AOD ducts meet and the junction of north
and south ducts. Two sampling ports, 90 degrees apart, were
located 2.9 diameters downstream and 1.3 diameters upstream of
45-degree bends, as shown in Figure 4-4. Forty-four traverse
points were used to sample the cross-sectional area of the duct
for particulate matter, with 22 points on each traverse diameter.
Each particulate run covered two consecutive EAF heats. Tests
were started at the beginning of charging operations and
4-1
-------
NORTH ADD
SITE
NO. 1
EXISTING CATWALK
SITE
NO. 2
FABRIC FILTER
(8 COMPARTMENTS)
SOUTH ADD
Figure 4-1. Control system schematic, top view.
4-2 .
-------
CURED nor
I
U)
Figure 4-2. Control system schematic and location of sampling sites, elevation view.
-------
NORTH AOD (LEFT) AND
NORTH EAF HOOD DUCTS
SOUTH EAF (LEFT) AND SOUTH
AOD HOOD DUCTS
SAMPLING LOCATIONS AT THE NORTH EAF/AOD DUCT
(LEFT, SITE NO. 1) AND THE SOUTH EAF/AOD DUCT
(SITE NO. 2)
Fiqure 4-3. Control system configuration and location of sampling sites.
4-4
-------
01
FABRIC FILTER AND MONOVENT. SAMPLING
LOCATIONS ARE INSIDE EACH COMPARTMENT
NEAR THE BASE OF THE MONOVENT.
FANS AND COMBINED INLET DUCT
Figure 4-3. Control system configuration and location of sampling sites (continued)
-------
TRAVERSE
POINT NO.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
DISTANCES
cm
15.0
22.4
30.0
38.1
46.7
55.9
66.3
77.7
91.4
107.6
131.1
196.6
220.5
236.2
249.9
261.4
271.8
280.9
289.6
297.9
305.6
312.7
(in.)
( 5.9)
( 8.8)
( 11.8)
( 15.0)
( 18.4)
( 22.0)
( 26.1)
( 30.6)
( 36.0)
( 42.3)
( 51.6)
( 77.4)
( 86.8)
( 93.0)
( 98.4)
(102.9)
(107.0)
(110.6)
(114.0)
(117.3)
(120.3)
(123.1)
CROSS SECTION
PARTICULATE
-TRAVERSE DIAMETERS
10.2 cm (4 1n.)
PORT
304.8 cm (120 in.)
STACK I.D.
NORTH
PORT0
PARTICLE,-.
SIZE PORT0
SOUTH
PORT°
4 m (13 ft)
GAS FLOW
FROM NORTH
—
AOD AND EAF
SITE NO. 1
45 deg. BEND
NEAR JUNCTION
TOP VIEW OF
INLET SITES
8.8 m (29 ft)
45 deg. BEND AT
CREST OF ROOF
4m (13 ft)
8.8 m (29 ft)
-*-!
(
\
(
}
NORTH
PORT0
EXTRA o
PORT w
SOUTH -
PORT °
U-»-J
GAS FLOW
^ FROM SOUTH
AOD AND EAF
f
)
SITE NO. 2
(
)
61 cm (2 ft)
OFF AXIS
Figure 4-4. Inlet sampling locations.
4-6
-------
continued through tapping. Initial sampling of each traverse
point lasted 4 minutes. At the completion of a full traverse,
the nozzle was positioned at a point of average velocity and
sampling continued until the end of the first heat. A new
traverse was then initiated for the second heat, and each point
was sampled for 4 minutes. By the end of the second EAF heat, a
minimum of two complete traverses had been conducted. Actual
sampling time, which depended on heat times, ranged from 352 to
399 minutes.
Particle size distribution samples were collected at a point
of average velocity near the centroid of the duct. A separate;
port was used to minimize interferences with the particulate
matter tests. Sampling times ranged from 17.5 minutes for the
Andersen Mark III samples to 207 minutes for the Andersen Heavy
Grain Loading Impactor samples, which covered an integral heat.
4.2 SITE 2—UNCONTROLLED SOUTH EAF AND SOUTH ADD
Site 2 was similar to Site 1, but only particulate matter
tests were conducted at this site. Forty-four traverse points
were sampled for 8 minutes each, to yield a total sampling time
of 352 minutes per run. These tests were coordinated with tests
at Site 1, not with process conditions of the south furnaces.
4.3 SITE 3--FABRIC FILTER OUTLET
Controlled emissions from the fabric filter serving all four
furnaces were sampled for particulate matter and particle size
distribution. The cleaned gases from each compartment exit
4-.7
-------
through the common monovent located atop the fabric filter in a
configuration typical of positive-pressure fabric filters.
Method 1* criteria could not be met at this site; therefore, the
throat of the monovent was chosen as an optimum sampling location
because it represented the smallest cross-sectional area. This
small area would not only provide the highest gas velocities,
but also offer less chance for bias due to faulty bags.
Figure 4-5, a top view of the fabric filter arrangement,
shows the general location of sampling points. Figure 4-6, an
end view of the fabric filter, shows the location of the sampling
plane used in each compartment with respect to the site configu-
ration. Figure 4-7 gives specific dimensions for the sampling
plane cross-sectional area and the location of sampling points
in a typical compartment. It should be noted that the dimensions
of the sampling plane were slightly different from those of the
actual monovent throat because the tests were conducted just
below the throat. Other dimensions are given for the compartment
widths and panel offsets created by structural I beams. As the
figure indicates, dimensions for the sampling cross section and
the exhaust opening in the two end compartments were different
from those in the six middle compartments. Based on these
measurements, the total cross-sectional area of the sampling
2
plane in all eight compartments was calculated to be 113.2 m
(1218.8 ft2).
*
40 CFR 60, Appendix A, July 1, 1980.
4-8
-------
I
vo
_ _J
8
7
-35 4 m (116 ft)
TNI FT Dl CMIIM
1 1
1 1
| MONOVENTv
1 i V«
1 \
1 1 ' * — 1
j 1 !
OUTLINE OF EXHAUST ARE
AT TOP OF COMPARTMENTS
1 i
I— 4- — I J
1 ^
1 1
| '
i i
6 j 5 | 4
1 !
COVERED WALKWAY
\/^
L ^^ _
1. -«. -
3
SAMPLE
POINTS ~
'
2
V
4 1
-*
3 2
CMPT.
NO. 1
"J
k
"^-*
^~-
A
SEE SECTION K-K
IN FIGURE 4.3-2
A
^_J
TOP VIEW OF FABRIC FILTER
1 ACCESS DOOR PER
COMPARTMENT
RUN 1 - COMPARTMENT NOS. 1 THROUGH 4
RUN 2 - COMPARTMENT NOS..3 THROUGH 6
RUN 3 - COMPARTMENT NOS. 5 THROUGH 8
Figure 4-5. Sampling Site No. 3, the fabric filter outlet.
-------
SEE SECTION B-B
IN FIGURE 4.3-3
WALKWAY
WITH
ACCESS
DOOR
METHOD 5
PROBE AND HEATED
FILTER BOX
SUSPENDED FROM
CENTER OF RIDGE BEAM
SAMPLE
LINE
^/IMPINGtRS AND METERING EQUIPMENT
RAISED
CENTER
WALKWAY
MONOVENT
GAS FLOW
GRATING LEVEL ABOVE BAGS
13.6 m (44.7 ft)
H = 3.4 m (11 ft)
W = WIDTH OF EXHAUST OPENING AT SAMPLING PLANE = 3.51 m (11.5 ft)
^ = 87.8^cm (2.88ft)
SECTION A-A, FABRIC FILTER END VIEW FACING SOUTH
Figure 4-6. Sampling location at Site No. 3, the fabric filter outlet.
-------
EDGE OF EXHAUST OPENING AT SAMPLING PLANE
MONOVENT PANEL
SEPARATES
EXHAUSTS
BETWEEN
COMPARTMENTS
3.51 m
(11.5 ft)
4 SAMPLING POINTS
PER COMPARTMENT
END PANELS SEPARATING COMPARTMENTS
4.42 m (14.5 ft)
•DIMENSIONS IN BRACKET [ ] ARE FUR THE TWO END COMPARTMENTS. IF DIFFERENT.
NOTE: BAG ROWS ARE IN-OPEN AREAS BETWEEN CATWALKS.
TOP VIEW OF FABRIC FILTER EXHAUST AREA
WITH MONOVENT NOT SHOWN. ONLY ONE OF
EIGHT COMPARTMENTS IS SHOWN.
Figure 4-7. Location of sampling points at Site No. 3, the
fabric filter outlet.
-------
Each particulate test consisted of sampling four points in
each of four compartments (for a total of 16 points); at 20
minutes per point, this yielded 320 minutes of sampling time.
The tests began at the same time as the inlet tests and concluded
at the end of the 16-point traverse. Figure 4-5 shows the
specific compartments tested during each run. At the end of
three runs, each compartment had been tested at least once.
Each particle size distribution sample was collected at one
sampling point in each of four compartments, which yielded a
total sampling time of 300 minutes. These samples were collected
simultaneously with particulate matter tests and in the same
compartments; however, both probes were not in the same compart-
ment at the same time.
During each test, the entry of dilution air through the open
grating at the bottom level of the fabric filter was minimized by
covering the gratings with kraft paper and boards. The gratings
in all four compartments to be tested on a given day were covered
during the preceding day to allow fabric filter conditions to
(
equilibrate.
4.4 VELOCITY AND GAS TEMPERATURE
A type S pitot tube and an inclined draft gauge manometer
were used to measure the gas velocity pressures at the two in-
lets. Velocity pressures were measured at each sampling point
across the duct to determine an average value. Measurements
were taken in accordance with procedures outlined in Method 2 of
4-12
-------
the Federal Register.* Velocities at the outlet site were
calculated from inlet flow rates and the size of the outlet area.
The temperature at each sampling point was measured by using a
thermocouple and potentiometer.
4.5 MOLECULAR WEIGHT
Flue gas composition was determined in accordance with
procedures described in Method 3.* An integrated bag sample was
collected at each site during the preliminary runs on Monday, and
an Orsat Gas Analyzer was used to analyze the bag contents for
oxygen and carbon dioxide. Since these results verified that the
gas streams were essentially air, additional samples were not
collected.
4.6 PARTICULATE MATTER
Method 5* was used to measure particulate concentrations at
the two inlet sites. All tests were conducted isokinetically by
traversing the cross-sectional area of the stack and regulating
the sample flow rate relative to the gas velocity in the duct as
measured by the pitot tube and thermocouple attached to the
sample probe. Each sampling train consisted of a heated, 316
stainless steel-lined probe, a heated 87-mm (3-in.) diameter
glass fiber filter (Gelman Type AE), a Teflon sample line, and a
series of Greenburg-Smith impingers followed by an umbilical line
and metering equipment. At the end of each test, the nozzle,
probe, and filter holder portions of the sample train were
40 CFR 60, Appendix A, July 1, 1980.
4-13
-------
acetone-rinsed. The acetone rinse and filter media were dried at
room temperature, desiccated to a constant weight, and weighed on
an analytical balance. Total filterable particulate matter was
determined by adding the net weights of the two sample fractions.
Any condensate in the sample line was drained into the impinger
section of the sampling train. After the amount of water col-
lected in the impingers was measured, the contents were recovered
and gravimetrically analyzed for condensible matter by evaporating
the solutions in an oven at 105°C.
Method 5 equipment and modified sampling procedures were
used for particulate tests at the fabric filter outlet. The :
sampling train was similar to those used at the inlet sites
except for the lack of a pitot tube and the use of a glass-lined
probe. Tests were conducted at a constant sampling rate based on
the estimated average velocity of the entire sampling area. This
average velocity was calculated by first converting the total
flow rate measured at the two inlet sites to outlet conditions of
temperature, pressure, and moisture, and then dividing by the
total outlet sampling area. The resultant average velocity was
assumed to represent each sampling point and was used to calcu-
late an average isokinetic sampling rate. The heated probe and
filter assembly was suspended from the center of the ridge beam
in a fabric filter compartment, as shown in Figure 4-6. The
nozzle was positioned at each of the four sampling points in a
compartment by rotating the probe and filter assembly. When the
sampling points were changed, care was taken to avoid stirring up
4-14
-------
any dust or bumping the probe against any structural members.
Compartment cleaning cycles were sampled as they occurred. Com-
partment gas flows were interrupted while test equipment was
moved from one compartment to another, but conditions were
allowed to equilibrate for several minutes before sampling
resumed. Outlet samples were recovered and analyzed in a manner
similar to inlet particulate samples, except that the impinger
solutions were analyzed for organic and inorganic matter by
ether-chloroform extraction.
Sampling times and volumes for the outlet particulate tests
exceeded the respective minimum requirements of 4 hours and 4.5
dNm (160 dscf) specified in Subpart AA of the Federal Register.*
4.7 PARTICLE SIZE DISTRIBUTION
Particle size samples at the inlet site were collected with
an Andersen Mark III Cascade Impactor and an Andersen Heavy Grain
Loading Impactor (HGLI). The Mark III is an in-stack, multistage
cascade impactor that yields a total of eight particle cut-sis:es
ranging, nominally, from 0.5 to 15 ym. Substrates for this
impactor were 64-mm diameter glass fiber filters. The Mark III
was used to collect samples over time intervals of approximately
20 minutes. The HGLI is an in-stack multistage impactor designed
specifically to allow longer sampling times at high grain load-
ings. The three nominal cut-points are 2, 5, and 10 ym. The
only filter in the HGLI is a glass fiber thimble used as the
backup stage. This impactor was used to collect samples over an
40 CFR 60, Subpart AA, July 1, 1980.
4-15
-------
entire EAF heat, which was approximately 3.5 hours. A cyclone
precutter was attached to the front of each type of impactor to
remove larger particles and to avoid the use of buttonhook
nozzles. Because the sampling rate could not be adjusted to
obtain the 15-ym cut-point of the cyclone precutter, the weight
of particulate collected by the cyclone was added to the weight
in the first stage of the respective impactor.
All inlet samples were collected at a point of average
velocity near the centroid of the duct. The isokinetic sampling
rate was based on initial measurements of velocity pressure and
temperature. Constant cut-point characteristics were maintained
during sampling, but velocity pressures and temperatures were
measured periodically at the sampling point to evaluate the
actual variation in isokinetics. Nozzles were selected to keep
sampling rates in the recommended range of 8.5 to 21 liters per
minute (0.3 to 0.75 acfm). Each filter was recovered, desic-
cated, and weighed on an analytical balance. Acetone rinses of
appropriate stages were evaporated, desiccated, and weighed.
Particle size samples at the outlet were collected by using
a Mark III impactor fitted with a straight nozzle. The impactor
and probe were suspended in a fabric filter compartment from the
center of the ridge beam, in a manner similar to that used on
particulate matter testing equipment. Each sample was collected
for an equal amount of time at one point in four different com-
partments. The initial isokinetic sampling rate was based on the
calculated average velocity of the entire sampling area.
4-16
-------
Constant cut-point characteristics were maintained throughout
each test, and gas temperatures were measured with a thermocouple
attached to the impactor probe. Each filter was recovered,
desiccated, and weighed on an analytical balance. The inlet
chamber and nozzle were brushed and rinsed with acetone and the
rinse was evaporated, desiccated, and weighed.
4.8 VISIBLE AND FUGITIVE EMISSIONS
Certified observers recorded visible emissions from the melt
shop and fabric filter monovent according to procedures described
in EPA Method 9.* Data were taken in 6-minute sets (simultane-
ously with particulate tests), and individual readings were re-
corded in percent opacity at 15-second intervals. Intermittent
rest periods were taken to prevent eye fatigue; however, as long
as emissions were visually detectable, readings were continued
until a break was absolutely necessary. The emission points were
casually monitored during break periods, and readings were
resumed if emissions greater than zero opacity were noticed.
Fugitive emissions from the fabric filter dust-handling
system were observed according to the proposed Method 22.**
Emissions were recorded as the cumulative amount of time that
any fugitive emissions were visually detectable during a 20-
minute observation period. Several observation periods were
recorded during the test series.
Observers were positioned on the side of a hill, approxi-
mately 75 meters (250 feet) southwest of the baghouse.
40 CFR 60, Appendix A, July 1, 1980.
Federal Register, Vol. 45, No. 224, November 14, 1980.
4-17
-------
Adverse weather conditions prevented visual emission observations
during the third test.
4.9 FABRIC FILTER DUST SAMPLES
Samples from the dust-handling system were obtained just
below the central junction of the screw conveyors that connect
individual hoppers. Because the hoppers were emptied only once
a day, a single grab sample was taken on each test day. For each
sample, approximately 1 liter of dust was collected in a glass
jar that had been rinsed with dilute nitric acid in the labora-
tory. Upon return to the laboratory, each sample was split into
two fractions; one for trace element analysis, and one for
particle size distribution analysis.
Spark Source Mass Spectroscopy was the analytical technique
used for qualitative examination of the presence of approximately
70 elements. A known concentration on indium was added to each
sample prior to ionization. All elements were ionized with
approximately equal sensitivity. A photographic plate was used
to record the mass spectra. The plate was examined and the
response of each element was related to that of indium. Relative
sensitivity factors based on previous analyses of standards were
used to compensate for the variation in response of the photo-
plate for different elements.
The Coulter Counter technique was used to determine particle
size distributions after problems associated with particle
agglomeration prevented the initial attempts by Bacho analysis.
For the Coulter analysis, particles in each sample were suspended
4-18
-------
in a sodium chloride electrolytic solution. Electrical current
passed from one immersed electrode through a small aperture to
another electrode. As a particle passed through the aperture, it
displaced a volume of electrolyte and changed the electrical
current by an amount proportional to the size of the particle.
The volume and number of particles were used to establish a
differential distribution by volume. Assuming all particles were
of equal density, the volume distribution also represented a
weight distribution.
4-19
-------
SECTION 5
QUALITY ASSURANCE
Quality assurance is one of the main facets of stack sam-
pling because the end product of testing is to produce repre-
sentative emission results. Quality assurance guidelines provide
detailed procedures and actions necessary for defining and pro-
ducing acceptable data. Four documents were used in this test
program to provide the required guidance to help ensure the
collection of acceptable data and determine when data quality is
unacceptable. These documents are the source-specific test plan
prepared by PEDCo and reviewed by the Emissions Measurement
Branch; the EPA Quality Assurance Handbook Volume III, EPA-600/4-
77-027b; the draft PEDCo Environmental Emission Test Quality
Assurance Plan; and the PEDCo Environmental Laboratory Quality
Assurance Plan. The last two quality assurance plans are PEDCo's
general guideline manuals, which define the standard operating
procedures followed by the company's emission testing and the
laboratory groups.
Appendix F provides more detail on the Quality Assurance
procedures, including QA objective; data reduction; quality
control checks; performance and system audits; preventive main-
tenance; precision, accuracy, and completeness; corrective
action; and quality assurance reports to management..
5-1
-------
Relative to this specific test program, the following are
the steps that were taken to ensure that quality data were ob-
tained by the testing and analytical procedures.
0 Calibration of field sampling equipment. (Calibration
guidelines are described in more detail in Appendix E.)
0 Train configuration and calculation checks.
0 Onsite quality assurance checks, such as sample train,
pitot tube, and Orsat line leak checks.
0 Use of designated analytical equipment and sampling
reagents.
Table 5-1 lists the sampling equipment used to conduct
particulate and particle sizing tests, along with calibration
guidelines and limits. In addition to the pre- and post-test
calibration, a field audit was performed on the dry gas meters by
the use of critical orifices calibrated and supplied by the EPA.
The audit results in Table 5-2 show that all dry gas meters used
for this test series were within limitations stipulated in EPA
Method 5. Dry gas meter performance test procedures and field
audit sheets are shown in Figures 5-1 through 5-6.
Between runs, onsite preliminary calculation checks were
performed to verify isokinetic sampling rates and to compare
moisture contents, flow rates, and other parameters with expected
values. These checks indicated that the tests were being con-
ducted properly.
As a check of the reliability of the method used to analyze
the particulate matter and particle size filters, sets of blank
filters that had been preweighed in the laboratory were resub-
mitted for replicate analysis. Table 5-3 summarizes the results
5-2
-------
TABLE 5-1. FIELD EQUIPMENT CALIBRATION
tn
u>
Equipment
Meter box
Meter box
Meter box
Meter box
Meter box
Meter box
P1tot tube
PI tot tube
P1tot tube
1.0.
No.
FB-3
FB-5
FB-2
FB-7
FB-6
FB-8
251
252
253
Calibrated
against
Wet test meter
Standard pi tot tube
Allowable
deviation
AY prea + 0.020
AH@ + 0.15
AY pt>stbrjL 0.05
A Cp + 0.01
Actual
deviation
-0.003
-0.09
+0.012
-0.008
-0.05
+0.019
-0.007
-0.08
-0.02
-0.009
-0.06
+0.009
+0.001
-0.06
+0.002
-0.010
+0.06
+0.018
0.003
0.003
0.001
Within
allowable
limits
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
Comments
Y pre = 1.008
7 post = 1.020
T pre = 1.056
7 post - 1.075
7 pre = 1.007
7 post = 0.987
7 pre = 1.004
7 post = 1.013
7 pre = 0.973
7 post = 0.975
7 pre = 0.982
7 post = 0.999
Cp = 0.81
Cp = 0.81
Cp = 0.80
(continued)
"Allowable deviation AY pretest = +0.02 7 pretest.
Allowable deviation AY post-test = +0.05 7 pretest.
-------
TABLE 5-1 (continued)
Ul
I
Equipment
Thermocouple
Thermocouple
Thermocouple
Thermocouple
Thermocouple
Digital
Indicator
Or sat
analyzer
Trip balance
Barometer
1.0.
No.
149
138
129
128
254
219
222
208
207
142
198
227
Calibrated
against
ASTM reference
thermometer
Millivolt signals
Standard gas
Type S weights
NBS-traceable
barometer
Allowable
deviation
+ 1.5*
0.5*
+0.5*
+ 0.5 g
0.20 1n. Hg
post-test
Actual
deviation
-0.68
-0.54
+0.69
-0.35
-0.61
Avg. 0.1*
Avg. 0.16*
Avg. -0.10*
2.5°F
-0.2*
0.0 g
0.00 1n.
Hg
Within
allowable
limits
/
/
/
/
/
/
/
/
/
/
/
/
Comments
Actual deviation 1s
an average of eight
temperature points;
No. 207 tested and
calibrated by manu-
facturer
CO 1s highest
deviation
(continued)
-------
TABLE 5-1 (continued)
Equipment
Dry gas
thermometer
Probe nozzle
I.D.
No.
FB-2
FB-3
FB-5
FB-6
KB-7
FB-8
3-111
2-101
A2PS-P
A1PS-2
A1PM-2B
A1PM-1B
A1PM-2A
A3PS-P
A3PS-1B
Calibrated
against
Reference thermom-
eter type ASTM 2F
or 3F
Call per
Allowable
deviation
+5°F
On + 0.004 in.
Actual
deviation
1 1.5°F
0 1.7°F
I 1.2°F
0 1.9°F
I 2.7°F
0 1.6°F
1 0.6°F
0 1.6°F
I 3.5°F
0 1.4°F
I 1.0°F
0 2.3°F
0.001
0.002
0.001
0.001
0.003
0.003
0.002
0.001
o.obi
Within
allowable
limits
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
Comments
1 = inlet thermom-
eter
0 = outlet thermom-
eter
Nozzles for particle-
size tests were
labeled according to
run numbers
-------
TABLE 5-2. DRY GAS METER AUDIT RESULTS
Meter box No.
FB-2
FB-3
FB-5
FB-6
FB-7
FB-8
Calibrated against
Critical orifice No. 2
Critical orifice No. 2
Critical orifice No. 2
Critical orifice No. 1
Critical orifice No. 1
Critical orifice No. 1
Deviation, %
+ 3.3
+ 2.0
+ 0.4
- 0.3
+ 2.7
- 0.2
5-6
-------
M.'DIT REPORT GAVPt.R MPTEr!
Dace
Baromcteric pressure ( Phal.» In Hg ) 3O
bar'
Crifice number
Trlfice K factor
6",
Client
0 5
Meter box number f~&
Pretest Y /.OO~f
Auditor .£>
Ul
I
-J
Crlticc
manometer
reading
AH
In H20
Z.4
ury gas
meter
reading
'$.
O(e?g>.ooo
O83..000
rr>' eas
meter
volume
V
r?
/if .000
Temperatures
Ambient
T /T
ai af
°F
5-y
^
Avcracc
a
°F
^V
Inlet
Tmi
°F
^
^
Outlet
nio
°F
sy
55-
Average
T
m
°F
St. -2-Z
Samp linf;
time
0
min
/&.V
V
"std
ft3
/4.i*t>
V
m
act
ft3
/v-*/
Percent
error
Jt-Z.Z'/c
= ( 17.647 )( V )( Y )( P. -I All/13.6 )/(
A60 )
"act = ( 1203 )( 0 )( K )( Pbar )/( Ta + 46° >'
z.V
Vmstd=( 17.647 )( /V,*~>)
"act"( 12°3
error = ( V - V
mstd "act
100
act
100
Figure 5-1. Audit report sample meter box.
-------
A'.!DIT REPORT 3AKPI.F. METEI!
Date
V- 7 -
Barometerlc pressure ( P. , in Hg )
Orifice number 2
rrifice K factor
Client U^>
Meter box number /*jfl ^3
Pretest Y
Aud i to r £> ^
(Jl
I
00
Oriticc
manomcier
reading
AH
in H20
**«•
ury gas
meter
reading
ft3
JV7^
&t.~c
I'ry gas
meter
volume
V
t?3
•/-f.ooo
Temperatures
Ambient
T /T
ai af
°F
£4
<^
Average
T
a
°F
**
Inlet
°Fl
fl
W
Outlet
T
nio
°F
^6
^
Average
m
°F
**
Sampling
time
0
min
Z
V
-std
ft3
/^
V
mact
ft3
,735,
Percent
error
^
-------
AUDIT REPCIFIT
^:^:T!^r; r/\x
Date
Barometerlc pressure ( Pfe f. In Hg ) ^O • 3
Orifice number -2.
Crlflce K factor
Client
Meter box number
Pretest Y ^
s~\ /
Auditor
- 7-
ft^-awes.
Ul
I
VO
entice
manoraeter
reading
AH
In H20
z>i*
ury gas
meter
reading
'£•
TS'S-Ot
-770.000
rry gas
meter
volume
V
m
ft3
//2 . t?067
Temperatures
Ambient
T /T
al af
°F
&>
bo
Average
a
°F
t?o
Inlet
Tn,l
°F
^6
V
Outlet
T
mo
°F
fa£~
k
-------
CM MTNOO S BUT CAS M1EH MM
:( TtSI MU SMIt
(Jl
I
tat. Y/7/J7
iaraaatric
NKar In N
Pratatt V
Pollutant Coda
Prattora. fc^,. 3O • Vfc -MQ 77J> Survav Nuabar
•. Pfe A^Oae>
"735.1 Oi
Cat voluaa
dry gat
•atar
W..
Art) lent
•c
60 F /^.<
Teaperitu
0
Inlet
•"'
9y
8«V
ra
•y git aattr
Outlet'
lao
•F
'7 »"
Ty
'Average
la
•F
'---
VI « Initial raadlng af dry gat aatar.
Vf • final rtadtng af dry gat aatar.
to • ¥f - »l « iraluaa af gat patting through dry gat aatar
»irs
f T£ • tUfl)
Tha «a1«a af 0.4M7 It abtainad fn» ^ '
•feara: •.«•? It tka camrartton farior fro
Tttd • *»••
0.02B117
• ft1 te •'
Saapling
tiaa
0. Bin.
,7^a-
Vacwai
tatting
In. Hg
20
Came tad
gat voluoa
«Mttd)
• *^oVO
~~) C 3 e<
^ OCA-t^o-va^
SlgMtura
Figure 5-4. EPA Method 5 dry gas meter performance test data sheet.
-------
tat.*
7/*'
fM NHHOO S Ml CAS NUER PERFORMANCE TEST 0*1» SHCEI
rle Prettwe. 1^ JO. Vt» "Ma 775
fc. Pfe 7
»t«UUT /.OOV
Pollutant Codt_
ar
Participant! ID
Critic* Mo
Ul
Tnt
1
1
Orifice
ojanowter
reading
AM.
In. H,0
/ . 7
Cat voluw
dry gat
wtar
V»f
ft»
767.700
ISO. JUDO
Cat volia>t
dry gat
wtar
va>
/ 9-. 5
^ ... _
r • • ' ~
AaDitnt
la
'JU7 ^.
55
Teiyeratu
o
Inlat
•"
-------
•'•
IM NUHOO 4 Ot» C*S WTEI Hit
IfSI Ml* SHUT
cn
I
(-•
N)
»u V/7/S-,
^ft-f. r ,_.
••iMatrlc PrefMir*. tV JO. Vjfc, •«
mutt r
,. Pfr 9
. 7*35-
•
T**t
Ho.
1
f
1
Orlflc*
MMMttr
reading
AH.
In. hjO
5.3
Cat voluM
dry gat
Mttr
V,M
ft1
feblO.^bo
£J3- Voc?
•ollutant Codt
Ig 773 Survt
Parti
v Number
cipantt 10
OH f let Ho. /&>(•*, J-
Cat voluM
dry gat
Mttr
-1
wo lent
la
•c
^^.^
Taaptratu
r.:.-. o
Inltt
•r
~?y
7 •
r«
ry gat Mttr
Outlet"
1*0
rf£
.'r*
•"
,»
Sailing
tiM
0. «tn.
27 /J*0"
'•^*o
VacuM
tatting
in. Hg
/9
C*rr*ct*d,
Mt Ml«M
.Z-^l .
x * is r
VI • initial reading of dry gat Mttr.
Vf • final reading of dry gat Mttr.
Va • Vf - VI * MluM of gat patting through dry gat Mttr
AH
v._f.g).|t • (*«499/)CMi)(V)(>^kK_ * 13.6)
^^ __?y
«• -I- .f •Wu .^Md fro. ^ • ° 0»J»
0.0717 It the CMMrtlM (actor fro* ft-1 ta •'
T«U'
•ttd
Figure 5-6. EPA Method 5 dry gas meter performance test data sheet.
-------
TABLE 5-3. FILTER BLANK ANALYSIS
Type of filter
Participate: 87-mm
Gelman A/Ea
Andersen Mark,
III Impactor0
Andersen Mark
III Impactor .
Blank test run0
Andersen Heavy.
Grain Loading
Impactor, (HGLI)
Thimble0
Filter No.
2165
Y94
Z51
Y66
Y75
Y88
Z39
Y76
Z53
B179
X31
X32
X33
X34
X35
X36
X37
W60
B222
39
61
Tare
weight,
mg
362.3
127.8
136.0
126.8
134.9
127.2
136.0
127.4
136.6
180.5
148.9
137.6
145.6
138.6
151.1
138,9
150.2
135.2
188.0
2470.6
2176.6
Blank
weight,
mg
362.6
127.8
136.3
126.8
135.0
127.4
136.3
127.4
136.8
180.6
149.1
137.8
146.0
139.0
151.3
139.2
150.4
135.4
188.7
2480.4
2178.6
Net
weight,
mg
+ 0.3
0.0
+ 0.3
0.0
+ 0.1
+ 0.2
+ 0.3
0.0
+ 0.2
+ 0.1
+ 0.2
+ 0.2
+ 0.4
+ 0.4
+ 0.2
+ 0.3
+ 0.2
+ 0.2
+ 0.7
+ 9.8
+ 2.0
Comments
c
Expected deviation, +0.5 mg.
Expected deviation, +0.3 mq.
•s ' """'
'The high net weight is probably a result of very small particles leaking
around the prefilter.
Expected deviation, +0.5 mg.
5-13
-------
of the blank filter analyses. These results show good data
reproducibility from an analytical standpoint.
A blank run was performed at the fabric filter outlet to
determine whether stack gases reacted with the filter media to
produce erroneous results. This was accomplished by placing a
backup filter in front of a normally prepared impactor and then
sampling in the usual manner. Table 5-3 lists results of the
blank run, which shows that stack gases did not significantly
affect filter media.
In addition, blanks were taken to check the quality of
reagents used to recover and analyze particulate and particle
size samples. Table 5-4 summarizes the results of these blank
analyses, which show that most reagents met designated specifica-
tions for quality.
A trace element audit sample was analyzed, along with the
fabric filter dust samples, to check the accuracy of the SSMS
analytical procedures. The audit sample was taken from Standard
Reference Material No. 1633, "Trace Elements in Coal Fly Ash,"
which was obtained from the National Bureau of Standards. The
results (shown in Table 5-5) indicate that the analyses were
within a factor of three of true values, which is the expected
limit of SSMS accuracy.
Sampling equipment, reagents, and analytical procedures for
this test series followed and met all necessary guidelines set
forth for accurate test results in Volume III of the Quality
5-14
-------
TABLE 5-4. REAGENT BLANK ANALYSIS
Type of blank
Participate blanks:
Acetone
Acetone
Water
Particle size blanks:
Acetone
Acetone
Acetone
Analytical blanks:
Ether/chloroform
Water
Container
No.
3982A
3989A
3983A
2526A
2531A
2549A
BT459 (org)
BT459 (aq)
Volume of
blank, ml
237
227
378
382
190
518
150
250
Weight after
evaporation and
desiccation,3
mg/g
+ 0.01
+ 0.007
+ 0.004
+ 0.004
+ 0.02
+ 0.005
+ 0.005
+ 0.002
Comments
0.01 used in
calculations
Tolerance: +0.01 mg/g.
5-15
-------
TABLE 5-5. TRACE ELEMENT AUDIT RESULTS
Element
Arsenic
Cadmium
Chromium
Copper
Lead
Manganese
Nickel
Selenium
Uranium
Vanadium
Zinc
Concentration, yg/g
NBS-certifieda
61+6
1.45 + 0.06
131 + 2
128 + 5
70 + 4
493 + 7
98 + 3
9.4 + 0.5
11.6 + 0.2
214 + 8
210 + 20
Measured
120
1
150
71
110
170
54
16
18
99
220
Percent b
difference
+100
- 30
+ 10
- 40
+ 60
- 70
- 40
+ 70
+ 60
- 50
0
SRM No. 1633, "Trace Elements in Coal Fly Ash".
Percent difference =
^"
actual
deviation is +200%, -70% (+ factor of 3).
x 100, to the nearest 10%. Expected
5-16
-------
Assurance Handbook.5 Therefore, test results reported in this
document should be within the expected accuracies of the method
used.
5-17
-------
SECTION 6
STANDARD SAMPLING AND ANALYTICAL PROCEDURES
This section describes the test methods, sampling equipment,
and analytical techniques that were used in this test program for
determination of particulate matter and particle size distribu-
tion.
6.1 DETERMINATION OF PARTICULATE EMISSIONS
In this test program, the sampling and analytical procedures
used to determine particulate emissions at Sites 1 and 2 were
those described in Method 5 of the Federal Register.*
6.1.1 Sampling Apparatus
The particulate sampling train used in these tests met
design specifications established by the EPA. The sampling
apparatus, which was assembled by PEDCo personnel, consisted of
the following:
Nozzle - Stainless steel (316) with sharp, tapered leading
edge and accurately measured round opening.
Probe - Stainless steel (316) with a heating system capable
of maintaining a minimum gas temperature of 121°C (250°F) at
the exit end during sampling.
Pitot Tube - A type S pitot tube that met all geometric stan-
dards was attached to a probe to monitor stack gas velocity
pressure.
40 CFR 60, Appendix A, July 1, 1980,
6-1
-------
Temperature Gauge - A Chromel/Alumel type-K thermocouple (or
equivalent)was attached to the pitot tube, in an inter-
ference-free arrangement, to monitor stack gas temperature
within 1.5°C (5°F) using a digital readout.
Filter Holder - The filter holder was made of Pyrex glass,
with heating system capable of maintaining a filter tem-
perature of approximately 121°C (250°F).
Filter - An 87-mm (3-in.) diameter glass fiber filter
(Gelman A/E) was used.
Draft Gauge - The draft was measured with an inclined manom-
eter (made by Dwyer) with a readability of 0.25 mm (0.01
in.) H20 in the 0 to 25 mm (0 to 1 in.) H-O range.
Impingers - Four Greenburg-Smith design impingers were con-
nected in series with glass ball joints. The first, third,
and fourth impingers were modified by removing the tip and
extending the tube to within 1.3 cm (0.5 in.) of the bottom
of the flask.
Metering System - The metering system consisted of a vacuum
gauge, a leak-free pump, thermometers capable of measuring
temperature to within 1.5°C (5°F), a calibrated dry gas
meter, and related equipment, to maintain an isokinetic
sampling rate and to determine sample volume. The dry gas
meter was made by Rockwell, and the fiber vane pump was made
by Cast.
Barometer - An aneroid type barometer was used to measure
atmospheric pressures to 0.3 kPa (+0.1 in.Hg).
6.1.2 Sampling Procedure
After the sampling site and minimum number of traverse
points were selected, the stack pressure, temperature, moisture,
and range of velocity head were measured according to procedures
described in the Federal Register.*
Approximately 200 grams of silica gel were weighed and
placed in a sealed impinger prior to each test. Glass fiber
filters were desiccated for at least 24 hours to a constant
40 CFR 60, Appendix A, Methods 1, 2, 3, or 4, July 1, 1980
6-2
-------
weight and weighed to the nearest 0.1 mg on an analytical balance.
One hundred milliliters of distilled water was placed in each of
the first two impingers; the third impinger was initially empty;
and the impinger containing the silica gel was placed next in
series. The train was set up as shown in Figure 6-1. The sam-
pling train was leak-checked at the sampling site prior to each
test run by plugging the inlet to the nozzle and pulling a 50-
kPa (15-in.Hg) vacuum, and at the conclusion of the test by
plugging the inlet to the nozzle and pulling a vacuum equal to
the highest vacuum reached during the test run.
The pitot tube and lines were leak-checked at the test site
prior to each test run and at the conclusion of each test run.
The check was made by blowing into the impact opening of the
pitot tube until 7.6 cm (3 in.) or more of water were recorded on
the manometer and then capping the impact opening and holding it
for 15 seconds to assure it was leak-free. The same procedure
was used to leak-check the static pressure side of the pitot
tube, except suction was used to obtain the 7.6-cm (3-in.) H20
manometer reading. Crushed ice was placed around the impingers
to keep the temperature of the gases leaving the last impinger at
20°C (68°F) or less.
During sampling, stack gas and sampling train data were
recorded at each sampling point and whenever significant changes
in stack flow conditions occurred. Isokinetic sampling rates
were set throughout the sampling period with the aid of a nomo-
graph or calculator. All sampling data were recorded on the
Emission Testing Field Data Sheet.
6-3
-------
1.9-2.5 cm
(0.75-1 irO
1.8 cm (0.75-1 in.)
"H
— v
cd — ^^.
"*
PROBE
PITOT TUBE
THERMOMETER
NOZZLE^
STACK WALL
LI PROBE
"S" TYPE
PITOT
TUBE
THERMOCOUPLE
HEATED >
FILTER^
r ----- -n
|
THERMOMETER
100 ml. OF WATER
TEMPERATURE
INDICATOR THERMOMETERS
o
CALIBRATED
ORIFICE
-^.CONTROL
| VALVES
hJU
MANOMETER
VACUUM\LINE
GAUGE
Figure 6-1. Schematic of participate samplinq train.
-------
6.1.3 Sample Recovery Procedure
The sampling train was carefully moved from the test site to
the cleanup area. The volume of water from the first three
impingers was measured, and the silica gel from the fourth
impinger was weighed to the nearest 0.1 gram. Sample fractions
were recovered as follows:
Container No. 1 - The filter was removed from its holder,
placed in a petri dish, and sealed.
Container No. 2 - Loose particulate and acetone washings
from all sample-exposed surfaces prior to the filter were
placed in a polyethylene jar, sealed, and labeled. Par-
ticulate was removed from the probe with the aid of a brush
and acetone rinsing. The liquid level was marked after the
container was sealed.
Container No. 3 - A minimum of 200 ml of acetone was taken
for the blank analysis. The blank was obtained and treated
in a similar manner as the acetone washing.
Container No. 4 - After being measured, distilled water in
the impinger section of the sampling train was placed in a
polyethylene container. The impingers and connecting glass-
ware were rinsed with distilled H20, and this rinse was
added to the container for shipment to the laboratory.
Container No. 5 - A minimum of 200 ml of distilled water was
taken for the blank analysis. The blank was obtained and
treated in a similar manner as the water rinse.
Container No. 6 - An unused glass fiber filter was taken for
blank analysis.
All pertinent data were recorded on the Sample Recovery and
Integrity Data Sheet.
6.1.4 Analytical Procedures
The analytical procedures used were those described in the
Federal Register.*
Container No. 1 - The filter and any loose particulate
matter were desiccated in the petri dish for 24 hours to a
constant weight and then weighed to the nearest 0.1 mg.
40 CFR 60, Appendix A, July 1, 1980.
6-5
-------
Container No. 2 - The volume of acetone washings was mea-
sured and transferred to a tared beaker. The sample was
evaporated to dryness at ambient temperature and pressure,
desiccated for 24 hours to a constant weight, and weighed
to the nearest 0.1 mg.
Container No. 3 - The volume of acetone blank was measured
and transferred to a tared beaker. The blank was evaporated
to dryness at ambient temperature and pressure, desiccated
for 24 hours to a constant weight, and weighed to the near-
est 0.1 mg.
Container No. 4 - The volume of the impinger contents and
distilled water rinse was measured and transferred to a
tared beaker. The sample was evaporated to dryness at 105°C,
desiccated to a constant weight, and weighed to the nearest
0.1 mg.
Container No. 5 - The volume of distilled water blank was
measured and transferred to a tared beaker. The blank was
evaporated to dryness at 105°C, desiccated to a constant
weight, and weighed to the nearest 0.1 mg.
The term "constant weight" means a difference of no more
than 0.5 mg or 1 percent of total weight less tare weight, which-
ever is greater between two consecutive readings, with no less
than 6 hours of desiccation between weighings. All analytical
data were recorded on the Analytical Particulate Data Sheet.
Acetone and water blank data were recorded on respective blank
data sheets.
6.1.5 Modifications to Standard Procedures for Site 3 Tests
Several modifications were made to the standard sampling and
analytical procedures to conduct particulate matter tests at Site
3. Some sampling equipment and analytical procedures were also
different, but were not method deviations. These modifications
and differences are as follows:
6-6
-------
Sampling Apparatus--
Probe - The probe was of borosilicate glass with a heating
system capable of maintaining a minimum gas temperature of
121°C (250°F) at the exit end during sampling.
Pitot Tube - A pitot tube was not used.
Sampling Procedure--
The sampling site and number of sampling points did not meet
minimum Method 1* criteria, but were selected according to
practical considerations. The stack pressure, temperature, and
moisture at this site were measured according to standard pro-
cedures, but accurate velocity pressures could not be determined
at each point.
The sampling train was prepared, assembled, and leak-checked
according to standard procedures, except that a pitot tube was
not used.
During sampling, stack gas temperature and sampling train
data were recorded at each sampling point. An average isokinetic
sampling rate was set initially, based on the estimated average
velocity for the entire sampling cross-sectional area. This
estimated velocity was calculated from data that had been ob-
tained at Sites 1 and 2 and adjusted for the differences in
cross-sectional area, temperature, pressure, and moisture. The
resultant value was assumed to represent the average velocity at
each Site 3 sampling point. The average isokinetic sampling rate
was held constant throughout the sampling period. All sampling
40 CFR 60, Appendix A, July 1, 1980,
6-7
-------
data, including the estimated average velocity pressure, were
recorded on the Emission Testing Field Data Sheet.
Initial calculations of emission results were based on esti-
mated velocities and later adjusted to reflect actual velocities
measured at Sites 1 and 2 during simultaneous tests.
Sample Recovery Procedure—
The sample recovery procedures were the same as those for
Sites 1 and 2.
Analytical Procedures—
Container No. 4 - The volume of distilled water and water
rinse was measured and transferred to a separatory funnel.
The sample was extracted three times with diethyl ether,
and each time the water was drained back into the original
sample container and the ether into a clean, tared beaker.
The sample was then extracted three times with chloroform,
and each time the chloroform was drained into the beaker
with the ether. After the final extraction, the water
portion was drained into a separate tared beaker, evaporated
to dryness at 105°C, desiccated, and weighed to a constant
weight to obtain the condensible inorganic content. The
ether/chloroform portion was evaporated to dryness at
ambient temperature, desiccated, and weighed to a constant
weight to obtain the condensible organic content.
Container No. 5 - The distilled water blank was treated in
an identical manner as Container No. 4. The aqueous frac-
tion was used as a water blank, and the organic fraction was
used as an ether/chloroform blank.
All other procedures for the determination of particulate
emissions were as used in tests at Sites 1 and 2.
6.2 DETERMINATION OF PARTICLE SIZE DISTRIBUTION
Three different configurations of in-stack cascade impactors
were used to collect samples for particle size distribution
6-8
-------
measurements. The following sampling and analytical procedures
were used.
6.2.1 Sampling Apparatus
The source sampling train used in these tests met design
specifications established by the EPA. Assembled by PEDCo per-
sonnel, it consisted of:
Nozzle - Stainless steel (316) with sharp tapered leading
edge and accurately measured round opening.
Temperature Gauge - A Chromel/Alumel type-K thermocouple (or
equivalent) was attached to the probe to monitor stack gcis
(impactor) temperature to within 1.5°C (5°F) using a digital
readout.
Metering System - The metering system consisted of a vacuum
gauge, a leak-free pump, thermometers capable of measuring
temperature to within 1.5°C (5°F), a dry gas meter with 2
percent accuracy, and related equipment to maintain an
isokinetic sampling rate and to determine sample volume.
The dry gas meter was made by Rockwell, and the fiber vane
pump was made by Cast.
Condenser - The condenser consisted of a moisture-removal
device capable of maintaining a temperature less than 20°C
(68°F) and an attached thermometer to monitor temperature.
Impactor - An Andersen Mark III with eight stages and a
backup filter was used at Sites 1 and 3. An Andersen Heavy
Grain Loading Impactor with three stages and a backup filter
was used at Site 1. A cyclone precutter was attached to the
front of each impactor used at Site 1.
Barometer - An aneroid type barometer was used to measure
atmospheric pressures to 0.3 kPa (+0.1 in.Hg).
6.2.2 Sampling Procedure
The stack pressure, temperature, moisture, and velocity
pressure of the selected sampling site were measured with Method
5 equipment according to procedures described in the Federal
6-9
-------
Register.* One or more points representing average velocity were
selected as sampling points.
Each type of impactor was assembled appropriately. Assembly
of the Andersen Mark III (Mark III) involved alternating the
stage plates, collection media, flat crossbars, and Inconel
spacer rings so as to provide eight cut-sizes. The collection
substrates were Reeve Angel 934 AH glass fiber filters that had
been heated in a 204°C (400°F) oven for 1 to 2 hours, desiccated
for 24 hours to a constant weight, and weighed to the nearest 0.1
mg on an analytical balance.
Assembly of the Andersen Heavy Grain Loading Impactor (HGLI)
involved inserting a glass fiber thimble in the backup stage and
threading together the various parts of the third-stage cyclone
and first-and-second stage jet-impaction chambers. The glass
fiber thimble had been desiccated for 24 hours to a constant
weight and weighed to the nearest 0.1 mg on an analytical balance.
If used, the cyclone precutter was threaded together and
attached to the front of the impactor.
The sampling train was assembled as shown in Figure 6-2 or
6-3. It was leak-checked at the sampling site prior to each test
run by plugging the inlet to the impactor (or cyclone precutter,
if used) and pulling a 50-kPa (15-in.Hg) vacuum. Once the
desired vacuum was reached, the leakage rate was checked at the
dry gas meter for 1 minute. If the leak rate was less than 0.6
liter/min (0.02 cfm), the train was considered ready for
*
40 CFR 60, Appendix A, Methods 2, 3, or 4, July 1,. 1980.
6-10
-------
METER BOX
TEMPERATURE
INDICATOR
CYCLONE
PRECUTTER
THERMOCOUPLE
PROBE TUBE
NOZZLE
Figure 5-2. Particle size distribution sampling train at Site 1.
6-11
-------
PROBE TUBE
METER BOX
TEMPERATURE
INDICATOR
Figure 6-3. Particle size distribution sampling train at Site 3.
6-12
-------
sampling. Any excessive leaks were corrected before the train
was used. The impactor was then placed at the selected sampling
point and allowed to preheat for several minutes before sampling
began. While the impactor was preheating, the nozzle was capped
or pointed away from the gas flow. A leak-check was not per-
formed after the test run so as to avoid the possibility of
dislodging the particles on individual stages.
During sampling, stack gas and sampling train data were
recorded at regular intervals based on the length of the run.
Velocity pressure data at Site 1 were obtained periodically using
separate Method 5 equipment. Average velocities at Site 3 were
estimated from previous data measured at Sites 1 and 2. The
isokinetic sampling rate was set initially, and constant cut-
point characteristics were maintained throughout the sampling
period. Preliminary impactor runs were made at each site to
determine the Mark III sampling times required to allow uniform
loading on the backup filter and to prevent loadings of greater
than 10 mg on any one stage. All sampling data were recorded on
the Impactor Testing Field Data Sheet.
6.2.3 Sample Recovery Procedure
After the test was completed, the impactor was removed from
the probe and carefully moved to a designated cleanup area while
still in an upright position. The impactors were recovered as
follows:
6-13
-------
Mark III;
Container No. 1 - Particulate in the nozzle and inlet cham-
ber was removed by brushing and rinsing with acetone into a
polyethylene container, which was sealed and labeled.
Containers No. 2 through 10 - Each filter was removed from
its stage and carefully placed in a petri dish. Loose
particulate from the bottom side of the previous stage
plate, the Inconel spacer, flat crossbar, and the top side
of the plate directly under the filter were brushed into the
same petri dish as the respective filter. Each petri dish
was sealed and labeled.
Container No. 11 - If the cyclone precutter was used, par-
ticulate from all sample exposed surfaces except the in-
terior of the cyclone exit tube was brushed and acetone-
rinsed into a polyethylene container, which was sealed and
labeled. Particulate from the interior of the cyclone exit
tube was added to Container No. 1.
Heavy Grain Loading Impactor With Cyclone Precutter;
Containers No. 1 through 5 - Particulate from all sample-
exposed surfaces after the cut-point of the preceding stage
and prior to the cut-point of a given stage was brushed and
rinsed with acetone into a polyethylene container. After
the container was sealed and labeled, the liquid level was
marked.
Container No. 6 - The glass fiber thimble was carefully
removed from the holder and placed in a glass jar. The jar
was then sealed and labeled.
All pertinent data were recorded on Sample Recovery and
Integrity Data Sheets.
6.2.4 Analytical Procedures
Filters - Each glass fiber filter or thimble and any loose
particulate matter were desiccated in respective sample
containers for 24 hours to a constant weight and weighed to
the nearest 0.1 mg on an analytical balance.
Acetone Rinses - The volume of each acetone washing was
measured and transferred to a tared beaker. The sample was
evaporated to dryness at ambient temperature and pressure,
desiccated for 24 hours to a constant weight, and weighed to
the nearest 0.1 mg.
6-14
-------
The term "constant weight" means a difference of no more
than 0.5 mg or 1 percent of total weight less tare weight, which-
ever is greater between two consecutive weighings, with no less
than 6 hours of desiccation between weighings. All analytical
data were recorded on Andersen Impactor or HGLI Particulate
Analytical Data Sheets.
6.2.5 Blanks
Several unused glass fiber thimbles and a complete set of
unused Mark III filters were returned to the laboratory in
their respective containers. Approximately 200 ml of the acetone
used for sample recovery was taken as a blank. In addition, a
blank test run was conducted with the Mark III impactor to deter-
mine if stack gases had reacted with the filter media to cause
false weight changes. In the blank run a backup filter was
placed in front of a normally assembled impactor to filter out
all particulate matter so that only the stack gases would contact
the filter media.
All blanks were recovered and analyzed in the same manner as
the actual samples. Data were recorded on the respective blank
data sheets.
6-15
-------
REFERENCES
1. U.S. Environmental Protection Agency. A Review of Standards
of Performance for Electric Arc Furnaces in the Steel In-
dustry. EPA 450/3-70-033, October 1979.
2. U.S. Environmental Protection Agency. Background Informa-
tion for Standards of Performance: Electric Arc Furnaces in
the Steel Industry. EPA 450/274017b, October 1974.
3. Southern Research Institute. A Computer-Based Cascade
Impactor Data Reduction System. Prepared for U.S. Environ-
mental Agency under Contract No. 68022131. March 1978.
4. University of Florida. Use and Limitations of In-Stack
Impactors. Prepared by the Department of Environmental
Sciences for U.S. Environmental Protection Agency under
Grant No. R803692-02. February 1980.
5. U.S. Environmental Protection Agency. Quality Assurance
Handbook for Air Pollution Measurement Systems, Volume III.
EPA-600/4-77-027b, August 1977.
R-l
------- |