&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

-------