&EPA
United States
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park NC 27711
EMB Report 80-ELC-10
April 1981
Air
Electric Arc Furnace
Revision
Argon Oxygen
Decarburization
Emission Test Report
Carpenter Technology
Corporation
Reading, Pennsylvania
-------
EMISSION TEST REPORT
Carpenter Technology Corporation
Reading, Pennsylvania
ESED No. 79/9
EMB No. 80-ELC-10
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
April 1981
-------
CONTENTS
Figures
Tables
Quality Assurance Element Finder
Acknowledgment
1. Introduction
2. Process Operation
3. Summary of Results 3~1
Particulate matter 3-3
Particle size 3-22
Visible and fugitive emissions 3-31
Fabric filter dust samples 3-34
Supplemental analyses for fluoride, chromium,
lead, and nickel 3-39
4. Sampling Sites and Test Methods 4-1
Site 1—Inlet ADD 4-1
Site 2—Fabric filter outlet 4-5
Site 3—Scavenger duct 4-8
Site 4—Continuous casting torch cutter 4-9
Velocity and gas temperature 4-9
Molecular weight 4-12
Particulate matter 4-12
Particle size distribution 4-13
Visible and fugitive emissions 4-15
Fabric filter dust samples 4-16
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
11
-------
Appendix A
Appendix B
Appendix C
Appendix D
Appendix E
Appendix F
Appendix G
CONTENTS (continued)
Page
Computer printouts and example calculations A-l
Field data
• Sample recovery and analytical data
MRI process summary
Calibration procedures and results
Quality assurance summary
Project participants and activity log
B-l
C-l
D-l
E-l
F-l
G-l
111
-------
FIGURES
Number Page
2-1 Process and Control System Schematic at Cartech 2-4
3-1 Average Particle Size Results for Uncontrolled
Emissions, Site 1 3-25
3-2 Average Particle Size Results for Controlled
Emissions, Site 2 3-26
3-3 Average Particle Size Distribution of Fabric
Filter Dust Samples 3-38
4-1 Sampling Sites for Uncontrolled Emissions 4-2
4-2 Sampling Location for Uncontrolled AOD Emis-
sions, Site 1 4-3
4-3 Location of Sampling Points at Site 1 4-4
4-4 Fabric Filter, Site 2 4-6
4-5 Location of Sampling Points at the Fabric
Filter Outlet, Site 2 4-7
4-6 Location of Velocity Traverse Points at
Site 3 . 4-10
4-7 Location of Traverse Points at Site 4 4-11
5-1 Dry Gas Meter Audit 5-7
5-2 Dry Gas Meter Audit 5-8
5-3 Dry Gas Meter Audit 5-9
5-4 Dry Gas Meter Audit 5-10
5-5 Dry Gas Meter Audit 5-11
6-1 Particulate Sampling Train Used at Site 1 6-4
IV
-------
FIGURES (continued)
Number Page
6-2 Particulate Sampling Train Used at Site 2 6-5
6-3 Particle Size Distribution Sampling Train
Used at Site 1 6-10
6-4 Particle Size Distribution Sampling Train
Used at Site 2 6-11
v
-------
TABLES
Number Page
2-1 No. 2 AOD Production Summary 2-6
3-1 Samples Collected at Cartech 3-2
3-2 Summary of Gas Stream Characteristics at Sites
1 and 2 3-6
3-3 Summary of Gas Stream Characteristics at Sites
3 and 4 3-7
3-4 Summary of Filterable Particulate Emissions Data
at the Inlet, Site No. 1 3-8
3-5 Summary of Particulate Emissions Data at the
Outlet, Site No. 2 3-9
3-6 Filterable Particulate Collection Efficiency 3-13
3-7 Particulate Emission Factors Based on Furnace
Capacity 3-14
3-8 Particulate Emission Factors Based on Production 3-15
3-9 Summary of Particle Size Distribution and
Factional Efficiency 3-27
3-10 Summary of Visible and Fugitive Emissions 3-33
3-11 Summary of Trace Element Analyses on Fabric
Filter Dust Samples 3-35
3-12 Summary of Supplemental Analyses for Fluoride,
Chromium, Lead, and Nickel 3-41
5-1 Field Equipment Calibration 5-3
5-2 Dry Gas Meter Audit Results 5-6
5-3 Filter Blank Analysis 5-12
5-4 Reagent Blank Analysis 5-14
VI
-------
TABLES (continued)
Number Page
5-5 Trace Element Audit Results 5-15
5-6 Trace Element Audit Results 5-16
VII
-------
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
viii
Location
Section Page
11
1 1-1
Appendix F F-2
Appendix F F-3
Appendix D D-l
Appendix C C-l
Appendix E E-l
Appendix D D-l
Appendix F F-4
Appendix F F-5
Appendix F F-4
Appendix F F-6
Appendix F F-5
Appendix F F-6
Appendix F F-7
-------
ACKNOWLE DGMENT
Mr. William Terry of Midwest Research Institute, the New
Source Performance Standards contractor, monitored the process
operation during the test series, assisted in the coordination of
tests with process conditions, and provided the information
contained in Section 2 and Appendix D of this report. Messrs.
Larry Geiser and George Michael of Carpenter Technology Corpora-
tion helped to coordinate plant activities.
xiv
-------
SECTION 1
INTRODUCTION
During the week of April 28, 1981, PEDCo Environmental per-
sonnel conducted an emission sampling program at the steel melt
shop operated by Carpenter Technology Corporation (Cartech) in
Reading, Pennsylvania. The purpose of this test program was to
provide data for assessing the need for revising current New
Source Performance Standards (NSPS) for electric arc furnaces
(EAF) to include argon-oxygen decarburization (AOD) furnaces.
The No. 2 AOD at Cartech was selected for source testing for
the following reasons:
1) It utilizes best available control technology.
2) The emissions capture and control equipment is repre-
sentative of the industry.
3) Emissions from a single AOD are controlled separately
from those of other furnaces.
4) Emission data could be obtained by standard sampling
techniques at the desired locations.
Particulate matter concentrations and mass emission rates
were measured at one inlet and one outlet site according to U.S.
Environmental Protection Agency (EPA) Reference Method 5.*
Inlet and outlet tests for particulate matter were performed
simultaneously so that control efficiency as well as values for
40 CFR 60, Appendix A, July 1980.
1-1
-------
controlled and uncontrolled emissions could be determined. Flue
gas flow rates, temperature, and composition were measured in
conjunction with these tests. In addition, particle size dis-
tribution samples were collected simultaneously at the inlet and
outlet sites. Method 9* procedures were used to evaluate visible
emissions (VE) from the melt shop and fabric filter outlet
throughout the test series. Fugitive emissions (FE) from the
fabric filter dust handling system were determined visually
according to the proposed Method 22.** Samples of dust collected
by the fabric filter were obtained for analysis of particle size
distribution and trace element 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 operations. Two
filterable particulate samples taken at the outlet and two fabric
filter dust samples were subsequently analyzed for concentrations
of fluoride, chromium, lead, and nickel.
This report documents the activities and results of the test
program. Section 2 describes the process that was tested and the
operating conditions during the sampling period. Section 3
presents and discusses the results. Section 4 describes the
sampling sites and general test procedures. Section 5 briefly
outlines quality assurance measures and audit results. Section
6 gives details of the sampling and analytical procedures for
*
40 CFR 60, Appendix A, July 1980.
**
Federal Register, Vol. 45, No. 224, November 19, 1980.
1-2
-------
determining particulate matter concentrations and particle size
distribution. The appendices contain computer output and example
calculations (Appendix A), field data (Appendix B), sample re-
covery and analytical data (Appendix C), the Midwest Research
Institute (MRI) process summary (Appendix D), calibration pro-
cedures and results (Appendix E), a quality assurance summary
(Appendix F), and a list of project participants (Appendix G).
1-3
-------
SECTION 2
PROCESS OPERATION
Cartech's Reading plant has five EAF's and two AOD's. The
plant is capable of producing about 400 different grades of
steel for use in numerous industries (e.g., electronics, auto-
motive, appliance, aerospace, and industrial equipment). Each
of the five EAF's has a rated capacity of 13.6 megagrams (Mg) (15
tons) and produces an average heat of 15.4 Mg (17 tons). The No.
1 AOD has a rated capacity of 15.4 Mg (17 tons). The No. 2 AOD
has a rated capacity of 18.1 Mg (20 tons), but typically refines
a 15.4-Mg (17-ton) charge of molten metal. During the test
series the facility was operating at normal capacity (three
shifts per day, 5 days per week).
The No. 2 AOD vessel normally operates continuously, with
only a 5- to 10-minute delay between a tap and a subsequent
charge of molten metal. The molten metal charge comes from one
of two EAF's (designated as "C" and "E"). The time lapse between
tap and charge is short because no refractory gunning is per-
formed on the AOD. Longer delays occur periodically, when
maintenance is performed on the AOD vessel or when vessel
charging is delayed because a crane is not available. Delays
2-1
-------
in tapping could occur if the continuous caster were still
casting the metal from the No. 1 AOD or if the continuous caster
were broken down.
Each heat in the No. 2 AOD vessel consists of three general
stages. The first stage begins with the charging of molten metal
from either of two EAF's (C or E), which is followed by the
addition of the fluxing agent (lime). The vessel blow begins
almost immediately after the charge, with an oxygen-to-argon
ratio of 3:1 [930 normal cubic meters per hour (Nm /h) of oxygen
to 310 Nm /h of argon, or 33,000 standard cubic feet per hour
(scfh) to 11,000 scfh]. Some of the heats also use nitrogen in a
ratio of 3:1:1 (oxygen:argon:nitrogen) for the first phase of the
heat. After a gas blow of 15 to 30 minutes, the vessel is turned
down for a temperature check. If the temperature is close to
1923 K (3182°F), alloys are added. The type and weight of the
alloy additions depend on final product specifications.
During the next stage of the heat (approximately 1 hour),
the orientation of the vessel alternates between an upright
position (for blowing or stirring the molten metal with oxygen,
argon, or nitrogen gas) and a turned-down position (for tempera-
ture measurement, sample acquisition, and alloy and flux addi-
tion) . The oxygen-to-argon ratio for blowing during this stage
is 1:3. The AOD vessel operators make the necessary alloy and
flux additions, which are determined by mathematical calculations
based on the gross weight of the heat. Near the end of the heat,
2-2
-------
the slag is poured off into a slag pot to remove the lime and
impurities that are chemically bound in the slag.
Final chemical additions are made at the end of the heat,
after the final sample results are available. After the alloys
are melted into the bath by stirring with argon or nitrogen gas,
the molten metal is tapped into a ladle for transfer to the
continuous caster area. The AOD shop and the continuous caster
building are separated by a sheet metal wall suspended from the
ceiling to a level of about 6.1 meters (m) [20 feet (ft)] above
the floor. The molten metal ladle from the AOD must pass under
this wall to the ladle-stirring area before it is delivered to
the single-strand continuous caster.
Figure 2-1 is a schematic of the process and control system.
The process emissions generated during the heat are captured by
a canopy hood built into the roof trusses approximately 13 m
(42.7 ft) above the mouth of the vessel. The fumes are directed
to the canopy hood by a movable diverter hood located 1.5 m (5
ft) above the mouth of the AOD vessel. The diverter hood swings
out of the way during charging and tapping operations. The shop
roof above the No. 2 AOD is closed, and any fugitive emissions
not captured by the canopy hood remain inside the building and
are drawn into two openings in a scavenger duct located in the
peak of the roof between the AOD canopy and the continuous caster
area. The scavenger duct openings are not hooded.
The canopy hood and scavenger ducts are combined to form a
main inlet duct, which is then split into two ducts. Each duct
2-3
-------
NO.l AOD
AND EAF'S
N
\
,,
1 1 /
J 1 /
"'
/' NO. 2 AOD -^H
/ CANOPY
\
\
\ CATWALK
-^j" ivi rvi
7* -* i IAJ |/\)
1 /^ \ y
/ SITE NO. 3 SCAVENGER DUCT OPENINGS^
MELT SHOP |
— SITE NO.l
i
CONTINUOUS
CASTING
SHOP
I.D. FANS
REVERSEI1
AIR FANL
SITE NO.?
10
!
i
T) !
4) I
-J I
DP
NC.2 AOD
BAGHOUSE
\ TWO PORTS IN EACH STACK,
f 45 deg. FROM x TO CATWALK
SITE NO.4
Figure 2-1. Process and control system schematic at Cartech.
2-4
-------
has a 522-kilowatt (700-horsepower) fan that routes the emissions
to the inlet plenum of the fabric filter. The emissions from the
continuous-caster cutting torch are normally routed to the fabric
filter by a small duct and booster fan. The EPA requested that
this duct be closed off during the source test, however, to
prevent emissions from the cutting torch and the AOD from mix-
ing.
The AOD gases are treated in a positive-pressure Carborundum
fabric filter to remove the particulate matter. The cleaned
exhaust gases exit via five short stacks on the fabric filter.
Table 2 in Appendix D presents technical data on the No. 2 AOD
and the associated control device.
William Terry of MRI monitored the operation of the vessel
and the fabric filter. The AOD vessel operated normally during
the source tests, with only a few short delays. Table 2-1
presents a summary of production data for the test series. Tap-
to-tap times ranged from 1.4 to 2.3 hours, and the weight of
metal produced per heat ranged from 14.9 to 18.8 Mg (16.4 to 20.7
tons). The average production rate during the tests was 9.2 Mg
per hour (10.1 tons per hour), not including process delay times.
The average time between heats was less than 5 minutes, and
testing continued during these periods unless maintenance was to
be performed or the AOD operators indicated there would be a
delay. The testing was stopped on April 28, 1981, when emissions
from a fire at the No. 1 AOD drifted over to the No. 2 AOD.
2-5
-------
TABLE 2-1. NO. 2 ADD PRODUCTION SUMMARY*
PEDCo
Run
No.
1
2
3
Date
(1981)
4/28
4/29
4/30
Complete
heats
sampled,
Heat No.
1
2
3
4
5
Average
1
2
3
4
Average
1
2
3
4
5
Average
Tap-to-
tap
time,
minutes
86d
108
1000
104e
105
101
121
94f
137T
95
112
115
115
97
102n
879
103
Metal .
produced
Mg
16.4
17.7
15.7
15.3
15.1
16.1
15.5
15,9
16.1
17.2
16.1
15.4
18.8
15.1
16.1
14.9
16.1
tons
18.1
19.5
17.3
16.9
16.6
17.7
17.1
17.5
17.7
19.0
17.8
17.0
20.7
16.6
17.8
16.4
17.7
Process weight0
rate
Mg/h
11.4
9.8
9.4
8.9
8.6
9.5
7.7
10.2
7.1
10.9
8.6
8.1
9.8
9.3
9.5
10.3
9.3
tons/h
12.6
10.8
10.4
9.8
9.5
10.5
8.5
11.2
7.8
12.0
9.5
8.9
10.8
10.3
10.5
11.3
10.3
Compiled from data in Appendix D (Table 4 and Attachment 1).
Final billet weights, reported in megagrams and tons.
cProcess weight rates were calculated by dividing the weight of metal pro-
duced by the tap-to-tap time. Averages are based on average metal produced
and heat time to yield a weighted average.
These are actually charge-to-charge times because one tap time was unavail-
able.
eSeven minutes was subtracted from the actual time to reflect a delay result-
ing from the unavailability of a crane.
Thirteen minutes was subtracted from the actual time to reflect a delay
caused by slag pot removal. The long heat time resulted from problems in
attaining metal specifications.
^Five minutes was subtracted from the actual time to reflect a delay caused
by pulling a collar.
2-6
-------
The fabric filter was observed approximately once an hour
and was found to be operating normally. Indicator dials in the
fabric filter control room showed which compartment was closed
for cleaning, the amperage on each of the two fans, the inlet gas
temperature, and the amperage of the reverse-air fan. The
amperage of the reverse-side fan was typically 100 to 105 when
the fan was in the cleaning operation and 60 when it was not in
the cleaning mode. The other indicator dial readings are in-
cluded in Attachment 1 of Appendix D.
The test conditions were representative of normal opera-
tions and should provide useful data on controlled and uncon-
trolled emissions from an individually controlled ADD vessel.
Emissions were most significant during the blowing operation, and
the flow to the fabric filter appeared to be adequate to capture
almost all of the emissions in the canopy hood. The emissions
that were not captured by the canopy drifted northward toward the
continuous caster area, where they were captured by the scavenger
ducts. Charging and tapping emissions were minimal, and most
were captured by the canopy hood.
2-7
-------
SECTION 3
SUMMARY OF RESULTS
Particulate matter and particle size distribution tests were
conducted simultaneously at the inlet and outlet of the fabric
filter. Visible emissions from the melt shop and fabric filter
outlet were evaluated concurrently with particulate tests.
Fugitive emissions from the fabric filter dust-handling system
were evaluated periodically. 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 167 mg/dNm (0.073 gr/dscf). At the fabric filter out-
let, particulate concentrations averaged 0.66 mg/dNm (0.00029
gr/dscf) which indicates a 99.6 percent control efficiency. Both
concentration levels were in the range of expected values based
1 2
on reported data on EAF's. ' Outlet emissions, however, were
3
significantly lower than results of previous tests at this site.
The opacity of melt shop visible emissions averaged zero
percent over the test series; in fact, no emissions were visually
detected at any time. These results attested to the efficient
capture of emissions during charging, tapping, and other process
3-1
-------
TABLE 3-1. SAMPLES COLLECTED AT CARTECH
Sampl ing site
No. 1 -
Inlet
Ho. 2 -
Outlet
Ho. 3 -
Scavenger duct
Ho. 4 -
Torch cutter
Melt shop
Fabric filter
dust handling
system
Sample type
Particulate
Particle size
Particulate
Particle size
VE
Velocity
Velocity
VE
FE
Dust
Sampling
method
EPA 5
High capacity
impactor
Impactor
EPA 5
Impactor
EPA 9
EPA 2 -
Average point
EPA 2
EPA 9
EPA 22
Grab
Number
of
samples3
3
3
3
3
3
3
3
1
3
,'>
Time for
each sample
-7-1/2 h
-b h
5-15 minutes
~8 h
7-1/2 - 9 h
-1-5 h
-7-1/2 h
10 minutes
-1-5 h
20 minutes
1 per day
Additional analysis
Type
Organic and
inorganic con-
densibles
Trace metals,
particle size
No.
3
3
Method
Back half E/C
extract
SSMS.C Coulter
aDoes not include preliminary, blank, or duplicate runs.
bTwo samples were analyzed later for fluoride (by EPA Method 13B) and for chromium, lead, and nickel
(by Atomic Absorption Spectrophotometry).
GSpark source mass spectroscopy.
-------
operations. The opacity of visible emissions from the fabric
filter outlet averaged zero percent, which is indicative of ef-
ficient control of particulate matter in terms of opacity.
These and other results are presented and discussed in
detail according to emission type.
3.1 PARTICULATE MATTER
The fabric filter inlet duct (Site 1) and the fabric filter
outlet (Site 2) were tested simultaneously for particulate
matter. Site 1 represented uncontrolled process and fugitive
emissions from the No. 2 AOD; Site 2 represented controlled
emissions from the same source.
The fugitive portion of uncontrolled emissions was repre-
sented by Site 3, the scavenger duct. The velocity in this duct
was monitored for the duration of the particulate tests to detect
any changes in flow rate and corresponding emission capture effi-
ciency. Uncontrolled emissions from the torch cutter operation
(represented by Site 4) were not sampled. After initial gas flow
measurements had been made at Site 4, the torch cutter duct
was blocked (at EPA's request) to prevent those emissions
from entering the control system. This accomplished two things:
1) only emissions related to the AOD were tested, and 2) the
gas flow measurements at Site 1 represented the total net gas
flow to the fabric filter (not counting gas recirculated by the
reverse-air cleaning system).
3-3
-------
Particulate tests at Sites 1 and 2 were conducted over
approximately five ADD heats. Because the large number of heats
covered by the sampling period reduced the significance of
testing for integral heats (i.e., charging through tapping),
tests were commenced at any convenient time during a heat. Test-
ing started at both sites simultaneously and continued until the
respective traverses were completed, (about 7.5 to 8 hours).
This procedure made possible the calculation of control effi-
ciencies and average emission factors. Fabric filter cleaning
cycles were sampled as they occurred.
The NSPS contractor representative, who was on site to moni-
tor process operations, helped to coordinate the tests with
process conditions. Based on his observations, tests were
interrupted whenever the AOD experienced an operational delay or
conditions were unrepresentative.
Particulate matter was sampled and analyzed according to
procedures described in EPA Methods 1, 2, 3, and 5 of the
Federal Register.* Each outlet test consisted of traversing all
five stacks. Site 1 did not meet minimum Method 1 criteria, but
previous velocity profiles obtained by Cartech personnel indi-
cated the site was acceptable. At Sites 3 and 4 velocity was
measured according to procedures described in EPA Method 2.
Three particulate tests were conducted at Sites 1 and 2. Three
velocity determinations were made at Site 3, and one was made at
40 CFR 60, Appendix A, July 1980.
3-4
-------
Site 4. Integrated gas samples were collected once at Sites 1
and 2 (according to Method 3) to verify that the gas streams were
essentially air. Additional molecular weight determinations were
not made.
3.1.1. Gas Conditions and Particulate Emissions
Summaries of the measured stack gas and particulate emission
data are presented in Tables 3-2 through 3-5. 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 ex-
pressed in actual meters per second (m/s) and actual feet per
second (ft/s) at stack conditions. Particulate concentrations
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
concentration and the volumetric flow rate is the mass emission
rate. The filterable particulate data represent material col-
lected in the sample probe and on the filter, both of which were
heated to approximately 121°C (250°F). The condensible organic
and inorganic fractions represent material that passed through
the filter and was collected by the impinger section of the
sampling train at approximately 20°C (68°F). The isokinetic rate
3-5
-------
TABLE 3-2. SUMMARY OF GAS STREAM CHARACTERISTICS
AT SITES 1 AND 2a
Run No.
C1P-1
C1P-2
C1P-3
Date
(1981)
4/28
4/29
4/30
Average
C2P-1
C2P-2
C2P-3
4/28
4/29
4/30
Average
Flow rate
dNm3/s
130.6
138.7
133.3
134.2
59.0
63.0
68.5
63.5
dscfm
276,800
293,800
282,400
284,300
125,100
133,400
142,200
134,600
Temperature
°C
57
51
50
53
55
55
49
53
°F
135
124
122
127
130
131
119
127
Moisture.
%
1.1
1.7
0.9
1.2
0.3
1.4
0.5
0.7
Velocity0
m/s
21.5
22.7
21.5
21.9
6.0
6.5
6.9
6.5
ft/s
70.5
74.3
70.5
71.8
19.8
21.4
22.6
21.3
Flow rate
m3/s
150.0
158.1
149.9
152.7
66.5
72.0
76.1
71.5
acfm
317,800
335,000
317,600
323,500
141,000
152,800
161,300
151,700
Average C02 £0.6%, 02 = 19.4%. Sites 1 and 2 are the inlet and outlet,
respectively.
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.
Note: Outlet flowrates are values measured in the five stack exhausts. The
difference between reported inlet and outlet flowrates is the net gas
flow that escapes by means other than the stacks; that is, through the
partially open grating at the bottom level of the bags and through
other small openings on the clean side of the exhaust. These losses
are a result of the back pressure on the exhaust system caused by the
small stack outlet area.
3-6
-------
TABLE 3-3. SUMMARY OF GAS STREAM CHARACTERISTICS
AT SITES 3 AND 4a
Pun No.
C3-1
C3-2
C3-3
Date
(1.981)
4/28
4/29
4/30
Average
C4V-1
4/27
Flow rate
dNm3/s
34.0
34.0
34.5
34.2
6.5
' dscfm
72,030
72,060
73,030
72,370
13,830
Temperature
°C
53.4
52.8
47.8
51.3
33.4
°F
128
127
118
124
92
Moisture,
%c
1.5
1.5
1.5
1.5
1.5
Velocityd
m/s
15.9
15.9
16.1
16.0
32.7
ft/s
52.2
52.2
52.9
52.4
107.3
Flow rate
m3/s
34.5
34.5
35.0
34.7
6.6
acfm
73,130
73,160
74,140
73,480
14,040
Average C02 and 02 estimated at 0.6 percent and 19.4 percent, respectively.
Site 3 is the scavenger duct, and Site 4 is the torch cutter.
3Dry 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.
'Estimated.
Velocity at stack conditions.
"Flow rate at stack conditions.
Note:
Scavenger duct flowrates are included in inlet values reported in
Table 3-2. The flow through the torch cutter duct was blocked
during the test series, but the reported data were obtained
prior to testing.
3-7
-------
TABLE 3-4. SUMMARY OF FILTERABLE PARTICULATE EMISSIONS DATA
AT THE INLET, SITE NO. 1
Run
No.
C1P-1
C1P-2
C1P-3
Average
Date
(1981)
4/28
4/29
4/30
Concentration3
mg/dNm3
149.9
141.1
210.9
167.3
gr/dscf
0.0655
0.0617
0.0921
0.0731
•Mass
emission rate
kg/h
70.5
70.5
101.2
80.7
Ib/h
155.4
155.3
223.0
177.9
Isokinetic
rate,
%
108
107
101
OJ
I
00
^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.
-------
TABLE 3-5. SUMMARY OF PARTICULATE EMISSIONS DATA AT THE OUTLET, SITE NO. 2
Run
No.
C2P-1
C2P-2
C2P-3
Average
Date
(1981)
4/28
4/29
4/30
Concentration8
Filterable
mg/dNm
0.827
0.493
0.650
0.657
gr/dscf
0.000365
0.000217
0.000284
0.000289
Condenslble
Ore
mg/dNm3
0
0.069
0.289
0.119
anic
gr/dscf
0
0. 000030
0.000126
0. 000052
Inoraanlc
mg/dNm3
0.372
0.259
0.449
0.360
gr/dscf
0.000162
0.000113
0.000196
0.000157
Mass emission rate
Filterable
kg/h
0.390
0.246
0.312
0.316
Ib/h
0.859
0.542
0.687
0.696
Condenslble
Orgai
kg/h
0
0.035
0.138
0.058
1C
Ib/h
0
0.077
0.305
0.127
Inorc
kg/h
0.175
0.130
0.216
0.174
anic
Ib/h
0.385
0.286
0.475
0.382
I so-
kinetic
rate, i.
98.0
98.7
97.9
U)
vr>
"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 1n.Hg.
''Outlet mass emission rates are based on measured outlet concentrations and total Inlet standard flow
rates.
-------
is the ratio of the velocity of the sample gas stream entering
the nozzle to the local stack gas velocity, expressed as a
percentage.
The volumetric flow rate at the combined inlet (Site 1)
averaged 134 dNm /s (284,000 dscfm). The outlet flow rate
averaged 64 dNm /s (135,000 dscfm), which was significantly lower
than it should have been. This discrepancy is believed to be a
result of the excessive back pressure on the exhaust system,
which was caused by the small cross sectional area of the outlet
stacks. Because of this back pressure, a large volume of cleaned
exhaust gas exited the fabric filter through various openings in
the structure as well as through the stacks. These flow measure-
ments are discussed in more detail later in this section.
The actual flow rate at inlet Site 1 averaged 153 m /s
(324,000 acfm) at 53°C (127°F) and approximately 1 percent mois-
ture, and was equivalent to a gas velocity of 22 m/s (72 ft/s).
The actual flow rate measured at outlet Site 2 averaged 72 m /s
(152,000 acfm) at 53°C (127°F) and less than 1 percent moisture,
which represented a gas velocity of 6.5 m/s (21 ft/s).
For calculation purposes, the stack gases were essentially
air. During one run at each site, the carbon dioxide (CO-) con-
centration averaged less than 0.6 percent, and the oxygen con-
centration was 19.4 percent by volume.
Table 3-3 presents the results of velocity and flow measure-
ments at the scavenger duct (Site 3) and the torch cutter duct
(Site 4). For calculation purposes, gas composition data were
3-10
-------
estimated from results of the particulate tests at Sites 1 and 2.
The flow rate measured in the scavenger duct averaged 34 dNm /s
(72,000 dscfm), which is included in the total flow reported for
Site 1. At a stack temperature of 51°C (124°F), the actual flow
rate averaged 35 m /s (73,000 acfm) and represented a gas veloc-
ity of 16 m/s (52 ft/s). Measurements of the normal flow rate in
the torch cutter duct made before the duct was blocked off showed
that the actual flow rate was 6.6 m /s (14,000 acfm) at 33°C
(92°F), which represented a gas velocity of 33 m/s (107 ft/s).
This flow was equivalent to 6.5 dNm /s (13,800 dscfm).
Tables 3-4 and 3-5 present particulate emission results. At
Site 1 the average filterable particulate concentration was 167
mg/dNm (0.0731 gr/dscf), with a corresponding uncontrolled mass
emission rate of 80.7 kg/h (178 Ib/h). Condensible fractions
were not determined at this site so as to avoid biases that could
be caused by the long sample line used between the filter and
first impinger.
At the outlet, the filterable particulate concentration
averaged 0.66 mg/dNm (0.00029 gr/dscf). The organic and inor-
ganic condensible concentrations averaged 0.12 mg/dNm (0.00005
gr/dscf) and 0.36 mg/dNm (0.00016 gr/dscf), respectively. The
reported mass emission rates are based on the total flow rates
measured at inlet Site 1 rather than on the flow rates measured
at the outlet, which were biased low. Thus, they provide a
realistic estimate of the total particulate matter exiting the
fabric filter. The filterable, organic condensible, and
3-11
-------
inorganic condensible emission rates averaged 0.32, 0.06, and
0.17 kg/h (0.70, 0.13, and 0.38 Ib/h), respectively.
Isokinetic sampling rates ranged between 101 and 108 per-
cent at the inlet and were either 98 or 99 percent at the out-
let.
3.1.2 Control Efficiencies and Emission Factors
Control efficiencies were calculated by dividing the dif-
ference between the outlet and inlet particulate concentrations
by the inlet value. Table 3-6 presents a summary of filterable
particulate concentrations and indicates the fabric filter
collection efficiency for each run. Control efficiencies were
99.4, 99.6, and 99.7 percent on the three test days.
Table 3-7 presents filterable particulate emission factors
for uncontrolled and controlled emissions in terms of emission
rate per unit of furnace metal capacity. Factors were calculated
by dividing the appropriate hourly mass emission rate by the
furnace capacity of 18.1 Mg (20 tons). Results are reported in
kilograms per hour per megagram of furnace capacity (kg/h per Mg)
and in pounds per hour per ton (Ib/h per ton). The emission
factor for uncontrolled emissions averaged 4.5 kg/h per Mg, or
8.9 Ib/h per ton. The average controlled emission factor was
0.018 kg/h per Mg (0.035 Ib/h per ton).
The emission factors shown in Table 3-8 are based on actual
production data. Results were calculated by dividing the fil-
terable mass emission rate by the corresponding average produc-
tion rate. Emission factors are reported in kilograms per
3-12
-------
TABLE 3-6. FILTERABLE PARTICULATE COLLECTION EFFICIENCY
Run
1
2
3
Average
Inlet concentration
mg/dNm33
150
141
211
167
gr/dscfb
0.0655
0.0617
0.0921
0.0731
Outlet concentration
mg/dNm^
0.827
0.493
0.650
0.657
gr/dscf
0.000365
0.000217
0". 000284
0.000289
% efficiency0
99.4
99.6
99.7
99.6
u>
Milligrams per dry normal cubic meter at 20°C and 191 kPa.
Grains per dry standard cubic foot at 68°F and 29.92 in.Hg.
°Percent efficiency = Cin1et - Coutlet
Cin1et
x 100.
-------
TABLE 3-7. PARTICULATE EMISSION FACTORS BASED ON FURNACE CAPACITY*
Run No.
1
2
3
Average
Uncontrolled
kg/h per Mg
3.90
3.90
5.59
4.46
Ib/h per ton
7.77
7.76
11.2
8.91
Controlled
kg/h per Mg
0.022
0.014
0.017
0.018
Ib/h per ton
0.043
0.027
0.034
0.035
Factors are based on emissions per unit of furnace metal capacity in kilo-
grams per hour per megagram (pounds per hour per ton). The furnace capacity
is 18.1 Mg (20 tons).
3-14
-------
TABLE 3-8. PARTICULATE EMISSION FACTORS BASED ON PRODUCTION0
Run No.
1
2
3
Average
Metal production
rateb
Mg/h
9.5
8.6
9.3
9.1
tons/h
10.5
9.5
10.3
10.1
Emission factor0
Uncontrolled
kg/Mg
7.42
8.20
10.9
8.84
Ib/ton
14.8
16.3
21.6
17.6
Controlled
kg/Mg
0.041
0.029
0.034
0.035
Ib/ton
0.082
0.057
0.067
0.069
Calculated by dividing the filterable mass emission rate by the corresponding
average metal production rate.
'From Table 2-1.
'Kilograms per megagram (pounds per ton) of metal produced.
3-15
-------
megagram (pounds per ton) of metal produced. The average uncon-
trolled emission factor was 8.8 kg/Mg (18 Ib/ton), based on a
production rate of 9.1 Mg/h (10.1 ton/h). At the same production
rate, controlled emissions averaged 0.035 kg/Mg (0.069 Ib/ton).
3.1.3 Discussion
In general, the particulate tests were conducted according
to schedule. Only minor problems were encountered with the sam-
pling equipment and the process operation. The report does not
include the results of preliminary tests that were conducted at
Sites 1 and 2 to compare particulate loadings with estimated
sampling times and to eliminate any problems associated with test
coordination or physical sampling maneuvers. This section dis-
cusses validity of results, the discrepancy between inlet and
outlet flow rates, 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 filterable catch weights, which
ranged between 5.7 and 8.9 mg, were considerably lower than
desired. The actual minimum sampling time and volume were 8
hours and 10.8 dNm (380 dscf), which met the minimum criteria of
4 hours and 4.5 dNm (160 dscf) set forth for EAF's in Subpart AA
of the Federal Register.*
40 CFR 60, Subpart AA, July 1980.
3-16
-------
The analyses of filterable particulate matter should provide
results within the expected limits of accuracy for Method 5.
Several factors support this conclusion: 1) all repeat weighings
of filters and acetone rinses were within 0.2 mg, which is closer
than the 0.5 mg allowed by the method; 2) the acetone used was
within method specifications (impurities were 0.006 mg/g versus a
tolerance of 0.01 mg/g); 3) the analysis of the particulate
filter blank agreed within 0.4 mg of the original tare weight
(Method 5 gives no criterion for this); and 4) a glass-lined
probe was used to minimize possible sample biases. The maximum
possible margin for error associated with net acetone blank
adjustments, filter blanks, and repeat weighings was estimated to
be about 30 percent. Actual error was probably much lower on
the average; therefore, results were considered to be acceptable.
Condensible concentrations measured at the outlet compared
favorably with expected values based on previously reported tests
at EAF's and AOD's. Test results indicated that a significant:
portion (30 to 50 percent) of total emissions was being collected
in the impinger section of the sampling train.
Sampling equipment problems were minor. Although the probe
heat failed during Run 1 at the inlet site, it was heating
properly during about 30 percent of the test, and a comparison of
Run 1 results with those of the other two runs indicated that
particulate concentrations had not been significantly affected.
Sampling at the outlet during Run 3 was delayed for a short
period while the filter holder heating system was being repaired.
This delay did not affect the results.
3-17
-------
Inlet flow rates measured at Site 1 agreed very well with
the system design flow rate of 142 m /s (300,000 acfm); all
runs were within 12 percent of this value. This satisfactory
agreement supported the decision to sample at Site 1 despite its
failure to meet minimum Method 1 criteria.
All flow rates measured at the scavenger duct were within
2.5 percent of the design flow of 35 m /s (75,000 acfm). Com-
parison of the combined inlet and scavenger duct flow rates
indicated that approximately 25 percent of the control system's
capacity was used to capture fugitive emissions and 75 percent
to capture primary process emissions.
Although the system for the capture of torch cutter emis-
sions represented only 5 percent of the total volumetric capacity
of the entire No. 2 AOD control system, the EPA requested that
it be disconnected for the duration of the test series so that
only AOD emissions could be determined. Based on visual obser-
vations, particulate loading at the torch cutter operation was
much lower in magnitude than the AOD process emissions. There-
fore, the impact of this source on total fabric filter emissions
under normal operating conditions is probably insignificant.
Measured outlet flow rates were approximately half of those
expected. Actual velocity measurements are considered to be
representative of conditions at the time of the tests and accu-
rate within the limits of EPA Method 2. This conclusion is
supported by the following: 1) the site met minimum Method 1
3-18
-------
criteria, 2) all procedures of Method 2 were followed properly,
3) checks for cyclonic flow were negative during both the test
series and the pretest survey, 4) all five stacks were traversed
during each test, 5) Cartech's previous measurements at this site
indicated similar low flow rates, and 6) inlet flow rates were
validated.
The difference between measured inlet and outlet flow rates
is believed to have resulted from an excessive back pressure on
the exhaust system. Because the cross-sectional area provided by
the five stacks is so small, some exhaust gases must exit through
other available openings. Among the possible exit points were
the interior walkway gratings at the bottom level of the bags.
Even though Cartech personnel had covered these openings with
Masonite panels prior to the test series, the low stack flow
rates indicated that some gases still managed to escape. It was
also evident that, contrary to conditions at similar installa-
tions, these gratings were not sources of air inleakage at this
facility. Other possible gas exit points included the seams
between the compartment walls, roof, and access doors. All
possible exit points were considered to be on the clean side of
the fabric filter because there was no visually detected leakage
of particulate-laden gases. Because the measured stack flow
rates were biased low and the measured outlet particulate con-
centrations were considered to be representative, the total inlet
flow rate was used to calculate a more realistic estimate of the
3-19
-------
outlet mass emission rate. The gas flow of the reverse-air
cleaning system was not added to the inlet flow because the
reverse-air system recirculated cleaned exhaust gas and did not
represent an increase in net flow out of the fabric filter.
Also, based on the agreement between the measured and design
inlet flows, air inleakage at the induced draft (I.D.) fans was
probably negligible. Although comparison of inlet and outlet
moisture contents seemed to indicate some dilution, accuracy
limitations of moisture determinations at the 1 to 2 percent
level are probably greater than the reported differences. The
close agreement between average inlet and outlet gas temperatures
indicated that any entry of dilution air was less than 10 per-
cent. There was an apparent discrepancy between the inlet and
outlet gas temperatures measured during Run 2, but it was not
significant because the difference was within the 1.5 percent
criterion specified in Method 2. Because the flow rate measured
at Site 1 was considered representative of the net system flow,
the outlet mass emission rates reported in the text and tables
reflect this flow rate. Calculations used to adjust the computer
output are shown in Appendix A. These calculations did not
affect any other parameters.
The concentration of particulate matter measured at the
stack outlets was considered to be representative of the gases
escaping through openings other than the stacks. Although the
measured outlet particle size distributions indicated that the
number of large particles present was greater than expected,
3-20
-------
these distributions were thought to be biased by a combination of
factors (which are discussed later in this section): 1) increased
collection efficiency of upper impactor stages, 2) weighing
errors, and 3) particle agglomeration. A similar type of particle
size distribution (i.e., greater than expected number of large
particles) also was shown by a separate test series at an EAF/AOD
installation that did not show a loss of gas flow between inlet
and outlet test sites. This comparison tends to support the con-
clusion that an apparent high number of large particles does not
necessarily indicate a biased particulate concentration. In
addition, observations of the fabric filter structure and stack
outlets did not reveal any visually detectable differences in
opacity of the gases exiting the stacks and the gases escaping
via other means. Therefore, the particulate concentrations mea-
sured at the stack outlets are considered to be representative.
Evaluation of the process data furnished by MRI seemed to
indicate that the No. 2 ADD system was operating normally during
the test series. The several short process-related delays in
testing did not seem to affect emission results. The increase in
uncontrolled emissions indicated by Run 3 was substantiated by
the results of the particle size test during the same period;
however, the increase could not be related to specific process
activities. Overall, process operations were relatively con-
sistent. The average production rates for each test were within
10 percent of each other and compared favorably with normal
3-21
-------
values. Emission results were therefore considered to be repre-
sentative of normal operations.
3.2 PARTICLE SIZE
Tests for particle size distribution were conducted at Site
1 to represent uncontrolled emissions, and at Site 2 to represent
controlled emissions. These tests were performed in conjunction
with particulate matter tests.
Inlet particle size tests were conducted over three entire
heat cycles to represent average emissions, and during shorter
intervals to provide supplemental data. Tests during integral
heats were initiated at the beginning of a charge, continued for
three heats, and concluded at the end of tapping operations.
This yielded a sampling time of approximately 5 hours, which was
necessary to collect an adequate sample in the impactor that was
used. The shorter particle size runs were performed at various
times during the particulate test. The sampling times for these
shorter tests were approximately 5 to 15 minutes, adjusted as
necessary to obtain proper loadings.
Particle size distribution samples were collected at the
outlet simultaneously with each particulate test, which yielded
a sampling time of between 7.5 and 9 hours for each run. All
five stacks were sampled during each run in a manner that mini-
mized interferences with the coinciding particulate tests. Fabric
filter cleaning cycles were sampled as they occurred.
3-22
-------
The NSPS contractor's representative assisted in the coor-
dination of the integral heat runs at the inlet with process
operations. Tests were interrupted when he considered it neces-
sary to avoid sampling during unrepresentative conditions.
Andersen Mark III Cascade Irapactors were used to collect the
shorter inlet samples and all of the outlet samples, and an
Andersen Heavy Grain Loading Impactor was used to obtain integral
heat samples. All inlet samples were collected at an average
velocity point in the combined inlet duct (Site 1). Outlet
samples were run in duplicate, with the impactors positioned at
average velocity points in each of the five stacks. The sample
time for each run was divided equally among the five stacks.
Velocity data were obtained periodically during each run by the
use of Method 2 equipment. The report presents the results of
three runs for each type of sample.
3.2.1 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
represents the best-fit average curve through test data points.
Each data point was plotted manually and indicates both the 50
percent effective cut-size of each impactor stage and the cumula-
tive 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
3-23
-------
Ill Impactor stages were calculated by computer programs con-
tained in "A Computer-Based Cascade Impactor Data Reduction
System" (CIDRS), developed for EPA by Southern Research Institute
4
(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. Results of the integral heat tests
and shorter interval tests agreed very well with each other and
were plotted on the same graph. The average distribution indi-
cated that 50 percent by weight of uncontrolled particulate
emissions consisted of particles with aerodynamic diameters of
1.2 ym or less. Approximately 81 percent by weight had diameters
of 10 ym or less.
Figure 3-2 shows the average distribution curve for the
outlet samples. Results indicated that approximately 50 percent
of the mass emissions consisted of particles having aerodynamic
diameters of 10 ym or less. Only 20 percent by weight had diam-
eters of 3.0 ym or less.
Table 3-9 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 3-2. Controlled and
3-24
-------
en
M.0
tt.»
Kl
•-«
\S\
I/)
ts>
C1M2 C1H1
C1M3 V C1H2
C1M6 O C1H3
1.0
AERODYNAMIC
10.0
PARTICLE SIZE, micrometers
Figure 3-1. Average particle size results for uncontrolled emissions, Site 1.
-------
10
N>
1.0 10.0
AERODYNAMIC PARTICLE SIZE, micrometers
Figure 3-2. Average particle size results for controlled emissions, Site 2.
-------
TABLE 3-9. SUMMARY OF PARTICLE SIZE DISTRIBUTION AND FRACTIONAL EFFICIENCY
ro
Cumulative weight percent less
than larger stated size3
Weight percent in stated size
range
Particulate concentration in
stated size range, b mg/dNm3
(gr/dscf)
Fractional collection effi-
ciency in stated size range0
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Aerodynamic particle size range, micrometers
D<0.5
15
9
15
9
25.0
0.059
0.0110
<0. 000026
99.8
0.599.9
1.010 urn
-
18
47
30.1
0.309
0.0132
0.000135
99.0
Total
100
100
100
100
167
0.657
0.0731
0.000289
99.6
aWeight percents are taken from plots of average distributions.
Filterable particulate concentration = total emissions x weight percent 1n stated size range; concentrations
given in terms of milligrams per dry normal cubic meter at 20°C and 101 kPa and grains per dry standard cuoic
foot at 68"F and 29.92 in.Hg. Total concentrations were taken from Method 5 test results.
Cr n »4 tt- • Inlet concentration - outlet concentration lnn
Collection efficiency = inlet concentration ~ x 100'
-------
uncontrolled mass loadings in the respective size ranges were
calculated by multiplying these weight percentages by the average
inlet and outlet particulate concentrations shown in Table 3-6.
Fractional efficiencies were calculated for each size range by
dividing the difference between inlet and outlet concentrations
by the inlet values. The overall efficiency was 99.6 percent;
the range was from a high of 99.9+ percent for particles between
0.5 and 1.0 ym in diameter to a minimum of 97.6 percent for
particles between 5 and 10 ym in diameter.
3.2.2 Discussion of Results
All test run results 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 keep in mind 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
to the left toward smaller sizes if the actual density were
greater, 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 3.8 g/cm . Using this particle density would increase
the amount of controlled emissions of particles smaller than 5 ym
from roughly 30 to 50 cumulative weight percent.
3-28
-------
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 compared very well with results of the shorter
tests conducted during the 3:1 oxygen-to-argon blow phase of
operation. Because of this agreement, both sets of data were
plotted on the same graph. The resultant distribution generally
agreed with expected results based on previous EAF/AOD data.
Except for the first set of runs, the particulate concentrations
indicated by the integral heat and shorter particle size tests
compared favorably with the results of particulate tests con-
ducted over similar time frames. The differences between the
first set of runs may be related to the variation in emissions
over the short term.
The isokinetic sampling rates for five of the six reported
inlet tests were between 91 and 101 percent. The isokinetic
sampling rate for one of the shorter Mark III runs was 67 per-
cent, but the distribution results agreed with the other runs.
The sampling flow rates were all within limits suggested by the
manufacturer, and results are believed to be within generally ex-
pected limits of accuracy.
Particle size distributions at the outlet showed a higher
number of large particles than expected; 80 to 90 cumulative
weight percent of the particles were expected to have aerodynamic
diameters of about 2.5 ym or less. One of several plausible
3-29
-------
explanations for the apparent discrepancy is the possibility of
bias as a result of increased efficiency in the upper impactor
stages when glass fiber filters are used. Although examination
of analytical results indicated that this may have occurred, it
could not be verified. Adjusting outlet results for this type of
bias would not completely account for the difference between
actual and expected results, however.
A partial explanation of the results could be related to the
very low sample weights. Despite the long sampling times (7.5 to
9 hours) and large sample volumes [5.6 to 9.8 dNm (198 to 347
dscf)], the total catch weights were low (between 2.5 and 4.2
mg). Approximately 30 to 55 percent of the catch weights were
collected in the acetone rinse of the nozzle and inlet chamber.
Because each total catch represents 10 separate analyses (9
filters and 1 rinse), weighing errors could have contributed to
the discrepancy between actual and expected results, but would
not completely account for the difference. The particle size
distributions indicated by the three runs were in relative agree-
ment, and the particulate concentrations indicated by two of the
runs were within 65 to 90 percent of the results from simultane-
ous particulate tests (which is good agreement for the two
different methods).
Another explanation could be related to particle agglomera-
tion caused by electrostatic charge. Although the existence of
such a charge was not verified at this site, it has occurred
before at an EAF/AOD installation. This agglomeration could
3-30
-------
account for the discrepancy between expected and actual dis-
tribution results.
Sampling procedures probably had no effect on results. All
the isokinetic sampling rates averaged between 99 and 102 per-
cent, and all sampling flow rates were within the limits sug-
gested by the manufacturer.
The results probably were affected by a combination of the
following: 1) increased collection efficiency of the upper
stages, 2) weighing errors, and 3) particle agglomeration.
Therefore, the reported distributions for outlet emissions are
believed to be biased, and describing them by mean particle size
and geometric standard deviation would be misleading.
3.3 VISIBLE AND FUGITIVE EMISSIONS
Evaluation of visible emissions from the melt shop roof and
five fabric filter stacks took place simultaneously with par-
ticulate concentration tests. Emissions were observed in 6-
minute sets, and individual opacity readings were recorded at 15-
second intervals according to Method 9* procedures. Fugitive
emissions from the fabric filter dust-handling system were
evaluated periodically 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.
40 CFR 60, Appendix A, July 1, 1980.
**
Federal Register, Vol. 45, No. 224, November 18, 1980.
3-31
-------
3.3.1 Results
Table 3-10 summarizes the results of all visible and fugi-
tive emission observations. No emissions were visually detect-
able at any time during normal operations. Most of the data
were collected during Run 1 because adverse weather conditions
prevented additional observations during Runs 2 and 3. A total
of 69 six-minute observations were made at the melt shop during
13 hours of process operation covering all modes of furnace
operation. A total of 68 six-minutes sets of opacity data were
collected at the fabric filter outlet. The fabric filter dust-
handling system was observed for a total of 60 minutes.
3.3.2 Discussion
Capture of melt shop emissions during charging, tapping, and
other process operations was efficient. It is important to note
that only the area of the melt shop surrounding the No. 2
ADD was observed; the continuous casting shop and the No. 1
AOD were not. Several of the 6-minute set times did not coincide
exactly with actual charging and tapping times. Because emission
points were casually monitored during break periods, and readings
were to be resumed if emissions greater than zero percent opacity
were noticed, the average opacity for these charging and tapping
periods was considered to be zero percent. During the 13 hours
of operation, a total of seven charges and seven taps were
observed. The emissions caused by a fire at the No. 1 AOD during
Run 1 were recorded but not included in the summary.
3-32
-------
TABLE 3-10. SUMMARY OF VISIBLE AND FUGITIVE EMISSIONS0
Melt shop
Date
(1981)
4/28
4/29
4/30
Run
No.
1
2
3
Point of
emissions
Roof
Roof
Roof
Number of
sets
48
8
13
Range of
readings,
% opacity
0
0
0
Range of
set averages,
% opacity
0
0
0
Fabric filter outlet
Number of
of sets
68
Range of readings,
% opacity
0
Range of set
averages, % opacity
0
Fugitive emissions from fabric filter dust handling system
Accumulated observation
period, minutes
60
Accumulated emission time
minutes
0
% of observation period
0
Data were collected during 9.5 hours of process operation on April 28, 1.5
hours on April 29, and 2 hours of April 30. Unfavorable weather conditions
prevented additional readings on April 29 and 30.
D0n April 28 an abnormal situation at the No. 1 ADD, which occurred between
4:38 and 4:43 p.m., caused visible emissions at the No. 2 ADD. The average
opacity for that period was 6 percent.
3-33
-------
The lack of visible emissions from the fabric filter outlet
indicated efficient control of particulate matter in terms of
opacity. No emissions were detected at any time, even after
compartment cleaning cycles, and these results were supported by
the low particulate concentration results.
Fugitive emission data for the fabric filter dust-handling
system may be misleading in that the system probably was not in
operation during most of the observation period.
3.4 FABRIC FILTER DUST SAMPLES
Samples of dust collected by the fabric filter were obtained
daily from the waste container into which the hopper screw
conveyors emptied. 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 par-
ticle size distribution by Coulter Counter.
3.4.1 Trace Elements
Table 3-11 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 limit for most of the elements was 0.1 yg/g,
although it was as high as 0.8 yg/g in some cases. Major con-
stituents are listed as >1000 yg/g. Results for several elements
3-34
-------
TABLE 3-11. SUMMARY OF TRACE ELEMENT ANALYSES ON
FABRIC FILTER DUST SAMPLES
Element
Aluminum
Antimony
Arsenic
Barium
Beryllium
Bismuth
Boron
Bromine
Cadmium
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
Sample 1
> 90
7
19
430
<0.1
70
2
8
27
>1000
NRa
2
1
510
>1 000
>1000
>1000
< 0.1
< 0.1
< 0.1
>1000
< 0.1
75
9
< 0.1
< 0.1
< 0.1
NR .
STDb
0.4
< 0.1
>1000
0.9
>1000
3
< 0.1
>1000
> 920
NR
>1000
< 0.4
Goncentration, yg/g (ppm weight)
Sample 2
> 130
10
27
140
< 0.1
45
5
11
10
>1000
NR
6
2
740
>1000
>1000
>1000
< 0.1
< 0.1
0.3
>1000
< 0.1
110
13
< 0.1
< 0.8
< 0.1
NR
STD
0.6
< 0.1
>1000
5
>1000
8
< 0.1
>1000
>1000
NR
>1000
0.6
Sample 3
> 96
8
20
260
< 0.1
33
2
8
7
>1000
NR
2
1
540
>1000
>1000
>1000
< 0.1
< 0.1
0.2
>1000
< 0.1
80
10
< 0.1
< 0.1
< 0.1
NR
STD
0.5
< 0.1
>1000
2
>1000
6
< 0.1
>1000
>1000
NR
>1000
' 0.5
(continued)
3-35
-------
TABLE 3-11 (continued)
Element
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
Sample 1
>1000
21
NR
< 0.1
NR
< 0.1
390
< 0.1
>1000
0.2
< 0.1
< 0.1
16
< 0.1
< 0.1
< 0.1
280
>1000
84
> 230
65
> 560
< 0.3
0.7
< 0.1
2
< 0.8
< 0.1
16
>1000
89
<0.6
270
<0.1
0.7
>1000
3
Concentration, vg/g (ppm weight)
Sample 2
>1000
17
NR
< 0.1
NR
< 0.1
560
< 0.1
>1000
0.9
< 0.1
< 0.1
51
< 0.1
1
< 0.1
640
>1000
140
> 330
53
> 810
< 0.1
2
< 0.1
6
2
< 0.1
26
680
73
0.9
99
< 0.1
1
>1000
4
Sample 3
>1000
12
NR
< 0.1
NR
< 0.1
170
< 0.1
>1000
0.4
< 0.1
< 0.1
21
< 0.1
0.5
< 0.1
290
>1000
40
< 250
31
> 600
< 0.1
0.7
< 0.1
5
< 0.8
< 0.1
19
150
85
0.7
180
< 0.1
0.7
>1000
6
Not reported.
Internal standard.
3-36
-------
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 3-3 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 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 1.6 ym or less. Ninety-eight percent by weight had
diameters of less than 10 ym.
3.4.3 Discussion
Concentrations of several trace elements seem to vary
considerably. Such variation could be related to the different
specifications of the metal in the furnaces. It should be
remembered, however, that the SSMS analytical technique is more
qualitative than quantitative. The results of the audit samples
shown in Tables 5-5 and 5-6 bear this out in that 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
3-37
-------
CO
I
OJ
00
RESULTS OF DUST SAMPLE ANALYSES
0-1 6-2 0-3
CALCULATED DISTRIBUTION BASED ON
COLLECTION EFFICIENCIES AND MASS
ADTNG<:
1 1 NX I m t H N 1>U
1.0
10.0
PARTICLE SIZE, micrometers
Figure 3-3. Average particle size distribution of fabric filter dust samples.
-------
emission tests. 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 3.8 g/cm would decrease the amount
of dust smaller than 5 ym from approximately 95 to 80 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.
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
estimated from the average size distributions, fractional effi-
ciencies, and mass loadings listed in Table 3-9. The Coulter
Counter and theoretical curves both indicated that 65 cumulative
weight percent of the collected dust consisted of particles with
aerodynamic diameters of approximately 2.0 ym or less. The
curves differed considerably at other sizes, but this may be
related to the different test methods.
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
40 CFR 60, Appendix A, July 1980.
3-39
-------
by Atomic Absorption Spectrophotometry. These analyses were per-
formed subsequent to the completion of originally scheduled lab-
oratory work for better quantification of emission levels that
SSMA analyses on the fabric filter dust samples had indicated as
being greater -than 1000 yg/g.
Separate fluoride analyses were performed on two acetone
rinses and one filter from the outlet filterable particulate
samples, two of the fabric filter dust samples, and appropriate
blanks. Metal analyses were performed on two outlet filterable
particulate samples (acetone rinse and filter combined), two dust
samples, and appropriate blanks. The 'fourth outlet sample was
obtained from the shorter preliminary 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 summarizes in Table 3-12.
3-40
-------
TABLE 3-12. SUMMARY OF SUPPLEMENTAL ANALYSES FOR
FLUORIDE, CHROMIUM, LEAD, AND NICKEL
Uncontrolled emissions
Pollutant
species
Fluoride
Chromium
Lead
Nickel
Concentration
yg/g of solid
15,200
49,200
2,200
21,500
mg/dNm^
2.5
8.2
0.37
3.6
gr/dscf
0.0011
0.0036
0.0002
0.0016
Emission rate
kg/h
1.2
4.0
0.18
1.7
Ib/h
2.7
8.8
0.39
3.8
Emission factors
kg/h/Mg
0.068
0.22
0.0098
0.096
Ib/h/ton
0.14
0.44
0.020
0.19
kg/Mg
.0.13
0.43
0.019
0.19
Ib/ton
0.27
0.87
0.039
0.38
Controlled emissions
Fluoride0
Chromium
Leadd
Nickel
10,300
7,600
<2,300
6,800
0.0068
0.0050
<0.0015
0.0045
0.000003
0.000002
<7 x 10"7
0.000002
0.0033
0.0024
<0.0007
0.0021
0.0072
0.0053
<0.0016
0,0047
0.0002
0.0001
<0. 00004
0.0001
0.0004
0.0003
<0. 00008
0.0002
0.0004
0.0003
<0. 00008
0.0002
0.0007
0.0005
<0.0002
0.0005
Based on average uncontrolled particulate emissions and the 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 the average of analyses on two outlet samples.
cBased on analyses of acetone rinses only; the glass fiber filter analysis had a high blank weight
of fluoride and was not used.
Lead concentrations were below the analytical detection limits for these samples; numerical values
for emissions are based on the minimum detectable mass of lead.
-------
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 two times larger than the net
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 analysis,
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. Should the filter result be correct, the outlet
fluoride concentration would be 40,700 yg/g instead of the 10,300
yg/g indicated in the table. This would not compare favorably
with the 15,200 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.
The mass of lead in each outlet sample was less than the
minimum detection limit. The concentration of lead (in micro-
grams per gram) was therefore calculated by dividing the minimum
detectable mass of lead by the particulate sample weight. A
less-than mark (<) is used in the table to indicate that emis-
sions are based on the equivalent minimum detection limit.
3-42
-------
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. An over-
all schematic of the No. 2 AOD process and control system was
shown earlier in Figure 2-1.
4.1 SITE 1—INLET
Uncontrolled process and fugitive emissions from the No. 2
AOD were sampled at Site 1 for particulate matter and particle
size distribution. As shown in Figure 4-1, this site was in the
264-cm (104-in.) square main inlet duct downstream of the junc-
tion of the canopy and scavenger ducts, and upstream of the point
where the duct is split by the two I.D. fans. Seven new sampling
ports in the side of the duct were located 1.1 equivalent diam-
eters downstream of the scavenger duct inlet and 0.4 diameter
upstream of a 90-degree bend, as shown in Figure 4-2. Although
the site did not meet minimum Method 1 criteria, analysis of
velocity traverse data indicated an acceptable flow distribution
without cyclonic characteristics. Figure 4-3 shows the location
of the 49 sampling points used to traverse the duct cross-
4-1
-------
K)
NO.l AOD
AND EAF'S
K
j
i /
i /
i /
i i /
i/
K. —
GAS
FLOW
264
(104
in.fi
V
v
N
/ NO. 2 AOD
' CANOPY
\ SITE 3 (VELOCITY C
\
•-*£-& f («4J F*
f ' T (58 InJ sq.
X^ «92 en 722 0"
/ " (273 InJ *™ (284 InJ
j
297 cm (117 InJ
' SITE 1 (SAMPLING LOCATION)
i , 107 cm (42 1n)
GAS FLOW ». »
^ I I \ I /
L: *-f —
*—
-------
FIVE EXISTING
PORTS ON TOP
(NOT USED)
NEW
SEVEN
PORTS
DAMPERS
264 cm (104
MELT SHOP ELEVATION - FACING SOUTH
Figure 4-2. Sampling location for uncontrolled AOD emissions, Site 1.
-------
15 cm <
(6 in.)
39 cm
(15.2 in.)
i
40 cm
(15.7 in.)
\
40 cm
(15.7 in.)
l
39 cm
(15.2 in.)
i
38 cm
(15 in.)
i
38 cm
(15 in.)
i
15 cm .
(6 in.) i
;
fA^^H
L
— nr~
UL_
_rr~
rl—
-H
^^**^^toJ
To o
L- 18.8 cm
(7.4 in.)
O 0
1 2
0 0
0 0
o o
0 0
O 0
* 11.1 cm (4.
264
(104
. ^
cm
in.)
J
o o \ o
37.5 cm -^
(14.8 in.)
O 0 0
345
o o o
POINTS EQUALLY SPACED
o o o
0 0 O
0 0 O
o o o
4 in.)
O O T
18.8 cm -J
(7.4 in.)
0 O
6 7
0 0
0 0
O 0
O 0
o o
i
26
(104
1
\ cm
in.)
PORTS B, C, AND D ARE 15.2 cm (6 in.) IN DIAMETER,
OTHERS ARE 10.2 cm (4 in.)
INLET DUCT
Figure 4-3. Location of sampling points at Site 1.
-------
sectional area for the particulate matter tests. During one
complete traverse each point in the equal matrix was sampled for
9 minutes, which yielded a total sampling time of 441 minutes per
run. Tests were initiated sometime during a heat and continued
until the traverse was completed. Sampling was interrupted
during process delays or unrepresentative operating conditions,
but not for the short intervals ("5 minutes) between heats.
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 5 minutes for the
Andersen Mark III samples to 310 minutes for the Andersen Heavy
Grain Loading Impactor samples, which covered three integral
heats. These integral heat runs were inititated at the beginning
of a charge and continued through the end of the third subsequent
tap.
4.2 SITE 2—FABRIC FILTER OUTLET
Controlled emissions from the No. 2 ADD fabric filter were
sampled at Site 2 for particulate matter and particle size
distribution. This site consisted of five 168-cm (66-in.)
diameter stub stacks aligned down the center of the baghouse
roof. As shown in Figures 4-4 and 4-5, each stack exhausted
cleaned gases from 2 of the 10 compartments. Two sampling ports
in each stack were located 2.0 duct diameters downstream of the
gas entry point and 0.4 diameter upstream of the stack exit. At
4-5
-------
CONTINUOUS
RAIN CAP
TWO TEST PORTS AT
EACH OF FIVE STACKS
SHEET METAL SEPARATOR
TOP LEVEL OF BAGS
BAGHOUSE ELEVATION - FACING EAST
Figure 4-4. Fabric filter, Site 2.
4-6
-------
REV
irncr ATDJ
rtKat-fUKr
FAN L
5
4
^
3:
i
i
3
2
1
i
i
i
51
A
B
M:
_i
B
^-_ —_____.
0! .
NC.2 AOD
FABRIC FILTER
^ \ TWO PORTS IN EACH STACK,
C 45 deg. FROM x TO CATWALK
PORTS ARE 10-cm (4-inJ DIAMETER HOLES
ONE OF FIVE STACKS
POINT
NUMBER
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
LOCATION
2.5 cm (1 in.)
5.4 cm (2.1 1nJ
9.2 cm (3.6 In.)
13.3 cm (5.2 In.)
17.8 cm (7.0 in.)
22.2 cm (8.7 in.)
27.0 cm (10.6 la)
32.7 cm (12.9 in.)
38.4 cm (15.1 in.
45.7 cm 18.0 in)
54.3 cm 21.4 in.)
67.3 cm 26.5 in.)
101.0 cm (38.7 in.)
113.3 cm (44.6 in.)
121.9 cm (48 in.)
129.2 cm (50.9 in.)
135.2 cm (53.2 in.)
140.6 cm (55.4 in.)
145.4 cm (57.2 in.)
149.9 cm (59 in.)
154.3 cm (60.7 in.)
158.4 cm (62.4 in.)
162.2 cm (63.9 in)
165.1 cm (65 in.)
Figure 4-5. Location of sampling points at the
fabric filter outlet, Site 2.
4-7
-------
these distances Method 1 criteria required 48 particulate sam-
pling points. Figure 4-5 shows the location of sampling points
and the orientation of traverse diameters. Each particulate test
consisted of traversing all five stacks. Each point was sampled
for 2 minutes, which yielded a total sampling time of 480 min-
utes. The tests began at the same time as the inlet tests and
continued until all stacks had been traversed. Cleaning cycles
were sampled as they occurred, and tests were interrupted during
process delays and unrepresentative operating conditions.
Each particle size distribution sample was collected for
an equal amount of time at one sampling point in each of the
five stacks. During Runs 1 and 2, each stack was sampled for 90
minutes, which yielded a total run time of 450 minutes. Because
the sample catch weights for these runs were low, each stack was
sampled for 108 minutes during Run 3, which yielded a total sam-
pling time of 540 minutes. All particle size sampling occurred
simultaneously with particulate matter tests, but all probes were
not in the same stack at the same time.
Sometime before the test series, Cartech personnel had
covered all of the gratings at the bottom level of the fabric
filter with Masonite panels to minimize the loss of exhaust gas
flow through these gratings.
4.3 SITE 3—SCAVENGER DUCT
The velocity in the scavenger duct was monitored at Site 3
during each test. As shown in Figure 4-1, the site was located
4.9 equivalent duct diameters downstream of the last fugitive
4-8
-------
emission capture point and 4.7 duct diameters upstream of a 60-
degree bend. Figure 4-6 shows the 21 traverse points used to
obtain initial velocity data. An average velocity point was
selected and monitored for the duration of each test at Site 1.
4.4 SITE 4—CONTINUOUS CASTING TORCH CUTTER
Emissions from this source are normally controlled by the
fabric filter. The EPA requested that flow through this duct be
prevented during the test series so that tests would be repre-
sentative of AOD emissions only. The normal flow through the
torch cutter duct was measured prior to the tests to determine
its impact on the control system. (Figure 4-1 shows the approx-
imate location of Site 4.) As shown in Figure 4-7, two sampling
ports, 90 degrees apart, were located 1.5 duct diameters down-
stream and 0.5 duct diameter upstream of 45-degree bends. Twenty
points were used to traverse the duct cross section.
4.5 VELOCITY AND GAS TEMPERATURE
A type S pitot tube and an inclined draft gauge manometer
were used to measure the gas velocity pressures at each site.
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 the Federal
Register.* The temperature at each sampling point was measured
with a thermocouple and digital readout.
40 CFR 60, Appendix A, July 1980.
4-9
-------
33 cm
(13 in.)
40.6 cm
(16 in.)
i
<
40.6 cm
(16 in.)
33 cm
(13 in.)
i
^
— C
A
B
— C
c
— c
H
^ 147 cm ^
(58 in.)
. . , 14.6 cm 7.3 cm .
J* h~H(5.75 in.) (2.9 in.H
Toooooooooo
*~7.3 cm
(2.9 in.)
oooooooooo
POINTS EQUALLY SPACED
oooooooooo
*• 1 1 . 1 cm
(4.4 in.)
i
*
147
cm
(58 in.)
t
SCAVENGER DUCT
Figure 4-6. Location of velocity traverse
points at Site 3.
4-10
-------
TO BAGHOUSE
TORCH
CUTTER
-»* GAS FLOW
FAN
PORTS ARE 5 cm (2 in.)
DIAMETER HOLES
50.8 cm
(20 in.)
DIAMETER
POINT
NUMBER
1
2
3
4
5
6
7
8
9
10
LOCATION
2.5 cm (1 in.)
4.1 cm (1.6 in.)
7.6 cm (3.0 in.)
11.4 cm (4.5 in.)
17.5 cm (6.8 in.)
33.4 cm (13.2 in.)
39.3 cm (15.5 in.)
43.4 cm (17.1 in.)
46.6 cm (18.4 in.)
48.3 cm (19.0 in.)
Figure 4-7. Location of traverse points at Site 4.
4-11
-------
4.6 MOLECULAR WEIGHT
Flue gas composition was determined in accordance with pro-
cedures described in Method 3.* An integrated bag sample was
collected at Sites 1 and 2 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.
4.7 PARTICULATE MATTER
Method 5* was used to measure particulate concentrations at
Sites 1 and 2. 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. The outlet sampling train consisted of a heated,
glass-lined probe, a heated 87-mm (3-in.) diameter glass fiber
filter (Gelman Type AE), and a series of Greenburg-Smith imping-
ers followed by an umbilical line and metering equipment. The
inlet sampling train was similar except that the probe was lined
with 316 stainless steel, and a Teflon sample line was used
between the filter and first impinger. At the end of each test,
the nozzle, probe, and filter holder portions of the sample train
were acetone-rinsed. The acetone rinse and filter media were
dried at room temperature, desiccated to a constant weight, and
*
40 CFR 60, Appendix A, July 1980.
4-12
-------
weighed on an analytical balance. Total filterable particulate
matter was determined by adding the net weights of the two sample
fractions. The amount of water collected in the impinger section
of the sampling train was measured (any condensate in the sample
line used at the inlet was first drained into the impingers).
On the outlet train, the impinger contents were recovered and
analyzed for organic and inorganic condensible matter by ether-
chloroform extraction.
Sampling times and volumes for particulate tests at the out-
let exceeded the respective minimum requirements of 4 hours and
3
4.5 dNm (160 dscf) specified for EAF's in Subpart AA of the
Federal Register.*
4.8 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-
sizes 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
5 to 15 minutes. The HGLI is an in-stack multistage impactor
designed specifically to allow longer sampling times at high
grain loadings. The three nominal cut-points are 2, 5, and 10
ym. The only filter in the HGLI is a glass fiber thimble used as
40 CFR 60, Subpart AA, July 1980.
4-13
-------
the backup stage. This impactor was used to collect samples over
an interval that included three entire AOD heats (approximately 5
hours) .
A cyclone precutter was attached to the front of each type
of impactor to remove larger particles and to avoid the need to
use 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 isokinetic rate. 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 a
Mark III impactor fitted with a straight nozzle. Each sample was
collected for an equal amount of time at an average velocity
point in each of the five stacks. The isokinetic sampling rate
was based on initial measurements of velocity pressure and gas
temperature. Constant cut-point characteristics were maintained
4-14
-------
throughout each test, but gas temperatures and velocity pressures
were measured periodically at the sampling points to evaluate the
actual variation in isokinetic sampling rate. 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.
4.9 VISIBLE AND FUGITIVE EMISSIONS
Certified observers recorded visible emissions from the melt
shop roof and fabric filter stacks in accordance with procedures
described in EPA Method 9.* Data were taken in 6-minute sets
(simultaneously with particulate tests), and individual readings
were recorded in percent opacity at 15-second intervals. Inter-
mittent 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 immediately if any opacity was noted.
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.
40 CFR 60, Appendix A, July 1980.
**
Federal Register, Vol. 45, No. 224, November 18, 1980.
4-15
-------
Observers were positioned near the parking lots, approxi-
mately 60 meters (200 feet) southeast of the fabric filter.
Adverse weather conditions reduced the amount of time available
for visual emission observations during the second and third
tests.
4.10 FABRIC FILTER DUST SAMPLES
Samples from the dust-handling system were obtained from the
waste container into which all the hoppers were emptied. 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 laboratory. Upon return to the
laboratory, each sample was split into two fractions: one for
analysis of trace elements and one for analysis of particle size
distribution.
The Spark Source Mass Spectroscopy technique was used for
qualitative examination of approximately 70 elements. A known
concentration on indium was added to each sample before it was
ionizied. All elements were ionized with approximately equal
sensitivity. A photographic plate used to record the mass
spectra 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 photoplate for different elements.
The Coulter Counter technique was used to determine particle
size distributions. Particles in each sample were suspended in a
4-16
-------
sodium chloride electrolytic solution, and electrical current
passed from one immersed electrode to another electrode through
a small aperture. 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.
4-17
-------
SECTION 5
QUALITY ASSURANCE
Quality assurance (QA) is one of the main facets of stack
sampling 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. The four guideline documents used in
this test program were a 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; a draft
of the 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 guide-
line manuals, which define the standard operating procedures
followed by the company's emission testing and laboratory groups.
Appendix F provides more detail on the Quality Assurance
procedures, including the 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
-------
With regard to this specific test program, the following
were steps taken to ensure that quality data were obtained 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 leak checks on
the sample train, pitot tube, and Orsat line.
0 Use of designated analytical equipment and sampling
reagents.
Table 5-1 lists the sampling equipment used to conduct
particulate loading and particle size tests, along with calibra-
tion 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 stipu-
lated in EPA Method 5. Dry gas meter performance test procedures
and field audit sheets are shown in Figures 5-1 through 5-5.
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 were used to ensure that the tests were
conducted 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
i
u>
Equipment
Meter box
Meter box
Meter box
Meter box
Meter box
Pi tot tube
Pi tot tube
Pi tot tube
Pi tot tube
Pi tot tube
Pi tot tube
I.D.
No.
FB-2
FB-3
FB-4
FB-6
FB-8
179
180
185
187
189
192
Calibrated
against
Wet test meter
Standard pitot tube
Allowable
deviation
AY prea + 0.020
AH
-------
TABLE 5-1 (continued)
01
i
Equipment
Thermocouple
Thermocouple
Thermocouple
Thermocouple
Digital
Indicator
Or sat
analyzer
Trip balance
Barometer
I.D.
No.
129
164
256
259
124
125
219
232
198
225
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 In. Hg
post-test
Actual
deviation
+0.70
-1.4
-0.2
-0.3
Avg. 0.13%
Avg. 0.22%
Avg. -0.10%
0.1%
0.0 g
0.02 in.
Hg
Within
allowable
limits
'
'
S
/
/
'
'
'
Comments
OK In range
of use
OK In range
of use
0? and C02 are the
higher deviation
(continued)
-------
TABLE 5-1 (continued)
Ul
Equipment
Dry gas
thermometer
Probe nozzle
I.D.
No.
FB-2
FB-3
FB-4
FO-6
FB-8
CIM-3
6XXX
C2M-4
C2M-2
C2M-P
2-103
CIH-3
Calibrated
against
Reference thermom-
eter type ASTM 2F
or 3F
Call per
Allowable
deviation
+5°F
On +_ 0.004 in.
Actual
deviation
I 1.2°F
0 1.4°F
I 1.1 °F
0 2.0°F
I 1.1°F
0 1.0°F
I 0.8°F
0 1.4°F
I 1.08F
0 2.1°F
0.002
0.000
0.000
0.001
0.001
0.002
0.002
Within
allowable
limits
'
J
J
J
<
'
'
*
'
Comments
I = inlet thermom-
eter
0 = outlet thermom-
eter
Nozzles for particle-
size tests were
labeled according to
run numbers
Allowable deviation AY pretest = +0.02 Y pretest.
Allowable deviation AY post-test = +0.05 T pretest.
-------
TABLE 5-2. DRY GAS METER AUDIT RESULTS
Meter box No.
FB-2
FB-3
FB-4
FB-6
FB-8
Calibrated against
Critical orifice No. 6
Critical orifice No. 7
Critical orifice No. 9
Critical orifice No. 10
Critical orifice No. 8
Deviation, %a
- 2.5
- 0.71
- 1.78
+ 0.99
- 1.72
Expected deviation < +5%.
5-6
-------
AUDIT REPORT SAMPLE METER BOX
Date
Client
iarometerlc pressure ( Pfc r> in Hg )_
Orifice amber ft* £?
Orifice K factor *
Meter box number_
Pretest T
Auditor i
1.007
Ui
Ortiice
reeding
AH
in HO
/-7k ,
Dry ges
•eter
reeding
Wl
ft3
wry ges
•eter
Temperatures
Average
Inlet
•OT
Outlet
©ft
Average
- < IJ.647 )( V )( Y )(
Vt • ( 120J
15
AH/13.6 )/(
15
T
460 ) ^
•~* - ( "»**7 x
-act
is
error - < » - V )( 100 )/( V ) - (/>.
100 )/<
••td "act
act
Sampling
time
0
•In
ft
IC5.7Y8
ft
act
3
Percent
error
Figure 5-1. Dry gas meter audit.
-------
AUDIT REPORT SAMPLE METER BOX
01
00
Date
Cllent
BaroMterlc pressure (P. , In Hg )_
Orifice number "ft 3-
Orifice K factor ;g"/O.SC,
-V
Meter box number
Pretest T
Auditor
3
on i ice
reading
AH
in HjO
2.2
Dry gas
•etcr
reading
Vi/Yf
ft3
ury ga«
•eter
voluoa
V
AeDient
T. A.
\'
Teoperatures
Average
Inlet
Outlet
Average
SHpltng
tl*e
0
•In
ft
It**
ft
act
3
Percent
error
Vd
t
- < 11.6*7 )( V )( Y )(
M 0 )C K X
- ( 17.647 )( /J2-00
».( 1203 )( /6.V
-
-------
AUDIT REPORT SAMPLE METER BOX
I
vo
Dace
Client
IO % I£_
fu . . - (17.647 )( V )( T)"<'p'r' + AH/0.6 )/( T + 460 )
atd • oar •
\mct - < 1203 )( 0 )( K )( Pfc., )/( T. + 460 )*
?»,ta • ( 17-**7
'•«•
•rror - ( f_ - V )( 100 )/( V ) - (-r. ,<>*)( 100 )/( -,
std act act
)*
BaroMterlc pressure ( f\ri
Orifice number ^ Q
Orifice K
on i ice
reading
AH
la H.O
. i.-w .
In He ) 85- B2.5
factor H-T1& ^lO"*
tfry g'aT"
•eter
reading
ft3
538. i H
5^-W
Dry gas
•eter
VOlUBB
V
f?
io.«>rt,
Teat
X"
"-
f!>'
—
Percent
error
•u*
Figure 5-3. Dry gas meter audit.
-------
AUDIT REPORT SAMPLE METER BOX
Date.
la
7 I
iteric pressure ( Pb , In Hg ) JL9.
Orifice number t*/O
Orifice K factor.
Meter box nuaber
Pretest T
Auditor _,_ 2
cn
M
o
Orl(ice
Bun meter
reading
AH
in H.O
ury gea
met.tr
reading
ft3
»ry gas
•eter
voluam
V
MBDient
T. /T_
ereture*
Average
T
Inlet
T .
Outlet
Average
T
~atd
'•«.
».
'•-.
error
-(17.647 )( V^ )( T )(
- ( 1203 )( 0 )( JC )( Pb<
• ( 17.647 )( /J./O )(
* ( 1203 )(
100
td
act
+ AH/13.6 )/( T + 460 )
)/( T + 460 )
*
X 100
act
Sailing
0
in
it
ft
act
3
Percent
error
Figure 5-4. Dry gas meter audit.
-------
AUDIT REPORT SAMPLE METER BOX
Date,
•era
/
Client
terlc pressure (
Orifice nuaber
Orifice K factor
, in HI ) -JQ
Meter box nuaber
Pretest T
Auditor
/-/S -
Grille*
•unoneter
reeding
AH
In HO
wry g««
meter
reeding
ury ges
meter
volUM
V
r?
AaDlen
Teoperatures
Average
Inlet
Outlet
Average
tlM
0
•In
it
ft
«et
3
Percent
error
75"
75 '
76
•gtd - ( 17.6*7 )( V^ )( T )( Pfcar + AM/13.6 )/( TB + 4*0 )
v is i.-7^>io* z?.(, 75" L
•Mt - ( 1203 )( * )( K )< Pb-r )/( Ta + 460 )'
21.736.
-( 17.647
•act * ( 1203
error - ( V
-V )( 100 )/( V ) - (.1? X 100 )/( II 05 )
std act act
Figure 5-5. Dry gas meter audit.
-------
TABLE 5-3. FILTER BLANK ANAYLSIS
Type of filter
Particulate: 87-mm
Gelman A/Ea
Andersen Mark
III Impactorb
Andersen Mark'
III Impactor
Blank test runb
Andersen Heavy
Grain Loading
Impactor. (HGLI)
Thimble d
Filter No.
0002029
W-39
W-42
X39
W38
W35
W34
W31
W32
B217
W07
W08
W05
W06
W03
W04
W01
W02
B407
4-BU658
5-BU659
6-BU660
Tare
weight,
mg
363.0
148.7
136.7
149.2
136.6
147.2
137.8
147.6
135.8
188.1
149.0
139.2
150.2
139.4
149.0
138.6
149.0
138.0
198.3
1878.1
2120.2
2329.2
blank
weight,
mg
363.4
149.2
136.2
149.6
136.8
147.3
138.0
148.8
136.0
188.0
148.8
139.1
149.6
138.8
149.2
138.8
149.0
137.8
198.8
1880.6
2122.6
2330.0
Net
weight,
mg
0.4
0.5
-0.3
0.4
0.2
0.1
0.2
1.2
0.2
-0.1
-0.2
-0.1
-0.6
-0.6
0.2
0.2
0.0
-0.2
0.5
2.5
2.4
2.8
Comments
c
Expected deviation, +0.5 mg.
Expected deviation, +0.3 mg.
cBoth initial tare weighings agreed within +0.2 mg, as did both blank
weighings.
Expected deviation, +5.0 mg.
5-12
-------
of these blank filter analyses. Except for one particle size
filter, 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. These results show that stack gases did not sig-
nificantly affect filter media.
Blanks also 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. These
results show that all reagents met designated specifications 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 (SRM) No. 1633, "Trace Elements in Coal Fly
Ash," which was obtained from the National Bureau of Standards
(NBS). The results (shown in Table 5-5) indicate that, except
for manganese, the analyses were within a factor of three of true
values, which is the expected limit of SSMS accuracy. The
laboratory performed its own internal audit by analyzing a sample
taken from SRM No. 1632, "Trace Elements in Coal," which was also
obtained from the NBS. The results (shown in Table 5-6) indicate
5-13
-------
TABLE 5-4. REAGENT BLANK ANALYSIS
Type of blank
Particulate blanks:
Acetone
Water
Particle size blanks:
Acetone
Analytical blanks:
Ether/chloroform
Container
No.
1228A
1229A
3823A
BU630
Volume of
blank, ml
537
500
221
150
Weight after
evaporation and
.desiccation,
mg/ga
+0.0061
+0.0052
+0.0074
0.004
Comments
Tolerance: +0.01 mg/g.
5-14
-------
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
31
0.5
170
49
59
74
27
4
13
110
96
Percent .
difference
- 50
- 70
+ 30
- 60
- 20
- 80
- 70
- 60
+ 10
- 50
- 50
SRM No. 1633, "Trace Elements in Coal Fly Ash."
ercent difference =
""^
actual
deviation is +200%, -70% (+ factor of 3).
x 100, to the nearest 10%. Expected
5-15
-------
TABLE 5-6. TRACE ELEMENT AUDIT RESULTS
Element
Arsenic
Cadmium
Chromium
Copper
Lead
Manganese
Nickel
Selenium
Thallium
Uranium
Vanadium
Zinc
Concentration, yg/g
NBS certified^
5.9 + 0.6
0.19 + 0.03
20.2 + 0.5
18 + 2
30 + 9
40 + 3
15 + 1
2.9 + 0.3
0.59 + 0.03
1.4 + 0.01
35 +_ 3
37 + 4
Measured
5
0.7
10
12
4
46
9
2
<0.1
1
17
15
Percent .
difference
- 20
+270
- 50
- 30
- 90
+ 10
- 40
- 30
>- 80
- 30
- 50
- 60
*SRM No. 1632, "Trace Elements in Coal."
Percent difference =
measured - actual
actual
deviation is +200%, -70% (+ factor of 3).
x 100, to the nearest 10%. Expected
5-16
-------
that, except for cadmium, lead, and thallium, the analyses were
within the range of expected 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
Assurance Handbook. 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 for determination of
particulate matter and particle size distribution.
6.1 DETERMINATION OF PARTICUALTE EMISSIONS
The sampling and analytical procedures used to determine
particulate emissions 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. (A glass-lined probe was used
at Site 2.)
Pitot Tube - A type S pitot tube that met all geometric
standards was attached to a probe to monitor stack gas
velocity pressure.
40 CFR 60, Appendix A, July 1980.
€-1
-------
Temperature Gauge - A Chrome1/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) by the use of a digital readout.
Filter Holder - The filter holder was made of Pyrex glass
and had a heating system capable of maintaining a filter
temperature 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 Dwyer
manometer 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 consisting 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 was used to maintain an iso-
kinetic 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 400 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 1980,
6-2
-------
weight and weighed to the nearest 0.1 mg on an analytical bal-
ance. 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 or
Figure 6-2. Before each test run the sampling train was leak-
checked at the sampling site 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
before and after each test run. The check was made by blowing
into the impact opening of the pitot tube until the manometer
indicated 7.6 cm (3 in.) or more of water 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.) H-O 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 1n.)
1.8 cm (0.75-1 in.)
« THERMOCOUPLE
* PROBE
*•« PITOT TUBE
• NOZZLE,
"S" TYPE
PITOT
TUBE
THERMOCOUPLE
STACK WALL
UPROBE
THERMOMETER
HEATED
FILTER
i
THERMOMETER
TEMPERATURE
INDICATOR
CALIBRATED
ORIFICE
-7^
^-,,
MANOMETER
\
100 ml. OF WATER
THERMOMETERS
o
VACUUMVLINE
o
DRY GAS
METER
Figure 6-1. Participate sampling train used at Site 1.
-------
i
en
1.9-2.5 cm
(0.75-1 in.)
1.8 cm (0.75-1 in.)
-THERMOCOUPLE
• PROBE
=t-* PHOT TUBE
HEATED AREAv /FILTER HOLDER
NOZZLEtf^r
STACK WALL
U PROBE
'S" TYPE
PITOT
TUBE
THERMOCOUPLE
THERHOMETER
\ / IMPINGERS 4
\ / T/*rmiYmnAYIi ^^
TEMPERATURE
INDICATOR THERMOMETERS
\ / ICE MATER BATH
^-
100 ml OF WATER
BYPASS
VALVE
MAIN VALVE
VACUUM GAUGE
_£_
VACUUM LINE
VACUUM PUMP
Figure 6-2. Particulate sampling train used at Site 2.
-------
6.1.3 Sample Recovery Procedure
The sampling train was moved carefully 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.
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 1980.
6-6
-------
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 - For tests at Site 1, the content of this
container was stored for future reference. For tests at
Site 2, 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 con-
tainer 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/chloro-
form portion was evaporated to dryness at ambient tempera-
ture, desiccated, and weighed to a constant weight to obtain
the condensible organic content.
Container No. 5 - For tests at Site 1, the content of this
container was stored for future reference. For tests at
Site 2 the distilled water blank was treated in an identical
manner as Container No. 4. The aqueous fraction was used as
a water blank, and the organic fraction was used as an
ether/chloroform blank.
Container No. 6 - The blank filter was treated in an iden-
tical manner as the filter in Container No. 1.
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.
6-7
-------
6.2 DETERMINATION OF PARTICLE SIZE DISTRIBUTION
Three different configurations of in-stack cascade impactors
were used to collect samples for particle size distribution mea-
surements. 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 the following:
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
gas (impactor) temperature to within 1.5°C (5°F) by the use
of a digital readout.
Metering system - The metering system consisting 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 was used 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 of 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 2. 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).
6-8
-------
6.2.2 Sampling Procedure
The stack pressure, temperature, moisture, and velocity
pressure of the selected sampling site were measured with Method
5 equipment in accordance with procedures described in the
Federal 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 bal-
ance.
It used, the cyclone precutter was threaded together and
attached to the front of the impactor.
The sampling train was assembled as shown in Figure 6-3 or
Figure 6-4. It was leak-checked at the sampling site prior to
each test run by plugging the inlet to the impactor (or cyclone
*
40 CFR 60, Appendix A, Methods 2, 3, or 4, July 1980.
6-9
-------
PROBE TUBE
METER BOX
TEMPERATURE
INDICATOR
CYCLONE
PRECUTTER
THERMOCOUPLE
-------
METER BOX
TEMPERATURE
INDICATOR
w
Figure 6-4. Particle size distribution sampling train used at Site 2.
6-11
-------
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 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
capped or pointed away from the gas flow. A leak-check was not
performed 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 were obtained periodically using separate
Method 5 equipment. 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
When the test was over, the impactor was removed from the
probe and carefully moved to the designated cleanup are while
6-12
-------
still in an upright position. The impactors were recovered as
follows:
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
plates, the Inconel spacer, the 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 con-
tainers 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,
6-13
-------
desiccated for 24 hours 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 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 determine 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 par-
ticulate 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.
6-14
-------
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 450/2-74-017b, October 1974.
3. Carpenter Technology Corporation. Reports of Emissions
Testing Performed August 22, 1978, on Carbonnndum Baghouse -
No. 2 AOD. September 1978.
4. Southern Research Institute. A Computer-Based Cascade
Impactor Data Reduction System. Prepared for U.S. Environ-
mental Protection Agency under Contract No. 68022131,
March 1978.
5. 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.
6. U.S. Environmental Protection Agency. Quality Assurance
Handbook for Air Pollution Measurement Systems. Vol. III.
EPA-600/4-77-027b, August 1977.
R-l
------- |