BEPA
           United States
           Environmental Protection
           Agency
          Office of Air Quality
          Planning and Standards
          Research Triangle Park NC 27711
EMB Report 80-ELC-7
March 1981
           Air
         Arc Furnace -
Revision
Argon Oxygen
Deca rburiza tion

Emission Test Report
Altech Specialty Steel
Corporation
Albany, New York

-------
              EMISSION TEST REPORT
       AL TECH SPECIALTY STEEL CORPORATION
              WATERVLIET, NEW YORK
                  ESED NO. 79/9
                EMB NO. 80-ELC-7
                      by

          PEDCo Environmental, Inc.
              11499 Chester Road
           Cincinnati, Ohio  45246
           Contract No. 68-02-3546
            Work Assignment No. 2
                  PN 3530-2
                Task Manager

              Dennis Holzschuh
     Emission Measurement Branch, MD-13
 Emission Standards and Engineering Division
OFFICE OF AIR QUALITY PLANNING AND STANDARDS
    U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA  27711
                  July 1981

-------
                            CONTENTS

                                                            Page

Figures                                                       iv
Tables                                                        vi
Quality Assurance Element Finder                             vii
Acknowledgment                                              viii

1.   Introduction                                           1-1

2.   Process Operation                                      2-1

3.   Summary of Results                                     3-1

          Particulate matter                                3-3
          Particle size                                     3-21
          Visible and fugitive emissions                    3-33
          Fabric filter dust samples                        3-40
          Fluoride, chromium, lead, and nickel              3-45

4.   Sampling Sites and Test Methods                        4-1

          Site l--Uncontrolled north EAF and north AOD      4-1
          Site 2—Uncontrolled South EAF and south AOD      4-7
          Site 3—Fabric filter outlet                      4-7
          Velocity and gas temperature                      4-12
          Molecular weight                                  4-13
          Particulate matter                                4-13
          Particle size distribution                        4-15
          Visible and fugitive emissions                    4-17
          Fabric filter dust samples                        4-18

5.   Quality Assurance                                      5-1

6.   Standard Sampling and Analytical Procedures            6-1

          Determination of particulate emissions            6-1
          Determination of particle size distribution       6-8

References                                                  R-l

Appendix A     Computer printouts and example calculations  A-l

Appendix B     Field data                                   B-l
                               11

-------
Appendix C



Appendix D



Appendix E



Appendix F



Appendix G
       CONTENTS (continued)








Sample recovery and analytical data



MRI process summary



Calibration procedures and results



Quality assurance summary



Project participants and activity log
Page




C-l




D-l




E-l




F-l




G-l
                              ill

-------
                             FIGURES

Number                                                      Page

 3-1      Average Particle Size Results for Uncontrolled
            Emissions, Site No. 1                           3-25

 3-2      Particle Size Results for Uncontrolled Emissions
            During Various Furnace Operations, Site No. 1   3-26

 3-3      Average Particle Size Results for Uncontrolled
            Emissions, Site No. 3                           3-28

 3-4      Average Particle Size Distribution of Fabric
            Filter Dust Samples                             3-43

 4-1      Control System Schematic, Top View                4-2

 4-2      Control System Schematic and Location of Sam-
            pling Sites, Elevation View                     4-3

 4-3      Control System Configuration and Location of
            Sampling Sites                                  4-4

 4-4      Inlet Sampling Locations                          4-6

 4-5      Sampling Site No. 3, the Fabric Filter Outlet     4-9

 4-6      Sampling Location at Site No. 3, the Fabric
            Filter Outlet                                   4-10

 4-7      Location of Sampling Points at Site No. 3, the
            Fabric Filter Outlet                            4-11

 5-1      Audit Report Sample Meter Box                     5-7

 5-2      Audit Report Sample Meter Box                     5-8

 5-3      Audit Report Sample Meter Box                     5-9
                                IV

-------
                       FIGURES (continued)

Number                                                      Page
1			IL-.-TL                                                         _ - -

  5-4     EPA Method 5 Dry Gas Meter Performance Test
            Data Sheet                                      5-10

  5-5     EPA Method 5 Dry Gas Meter Performance Test
            Data Sheet                                      5-11

  5-6     EPA Method 5 Dry Gas Meter Performance Test
            Data Sheet                                      5-12

  6-1     Particulate Sampling Train Schematic              6-4

  6-2     Particle Size Distribution Sampling Train at
            Site 1                                          6-11

  6-3     Particle Size Distribution Sampling Train at
            Site 3                                          6-12

-------
                             TABLES

Number                                                      Page

 2-1      EAF and AOD Production Summary                    2-4

 3-1      Samples Collected at Al Tech Specialty Steel      3-2

 3-2      Summary of Gas Stream Characteristics             3-6

 3-3      Filterable Particulate Emission Summary           3-7

 3-4      Filterable Particulate Collection Efficiency      3-10

 3-5      Particulate Emission Factors                      3-12

 3-6      Particulate Emission Factors Based on Production  3-13

 3-7      Summary of Particle Size Distribution and
            Fractional Efficiency                           3-29

 3-8      Summary of Visible and Fugitive Emissions         3-35

 3-9      Comparison of Melt Shop Visible Emissions to
            Process Operation                               3-37

 3-10     Summary of Trace Element Analyses on Fabric
            Filter Dust Samples                             3-41

 3-11     Summary of Supplemental Analyses for Fluoride,
            Chromium, Lead, and Nickel                      3-47

 5-1      Field Equipment Calibration                       5-3

 5-2      Dry Gas Meter Audit Results                       5-6

 5-3      Filter Blank Analysis                             5-13

 5-4      Reagent Blank Analysis                            5-15

 5-5      Trace Element Audit Results                       5-16
                               VI

-------
                QUALITY  ASSURANCE  ELEMENT  FINDER
(1)   Title page

(2)   Table of  contents

(3)   Project description

(4)   Project organization and responsi-
     bilities

(5)   QA objective for measurement data
     in terms  of precision,  accuracy,  com-
     pleteness,  representativeness,  and
     comparability

(6)   Sampling  procedures

(7)   Sample custody

(8)   Calibration procedures  and frequency

(9)   Analytical  procedures

(10)  Data reduction,  validation,  and
     reporting

(11)  Internal  quality control checks and
     frequency

(12)  Performance and  system audits and
     frequency

(13)  Preventive  maintenance  procedures and
     schedules

(14)  Specific  routine procedures used
  .  to assess data precision, accuracy and
     completeness of  specific measurement
     parameters  involved

(15)  Corrective  action

(16)  Quality assurance  reports to management
                                                   Location
                                               Section    Page
             11

   1        1-1


Appendix F  F-2
Appendix F  F-2

Section 6   6-1

Appendix C  C-l

Appendix E  E-.1

Section 6   6-1


Appendix F  F-3


Appendix F  F-4


Appendix F  F-3


Appendix F  F-5
Appendix  F   F-4

Appendix  F   F-5

Appendix  F   F-6
                               vn

-------
                         ACKNOWLEDGMENT






     Messrs. William Terry and Lalit Banker of Midwest Research



Institute, the New Source Performance Standards contractor,



monitored the process operation during the test series, assisted



in the coordination of tests to process conditions, and provided



most of the information contained in Section 2 of this report.



Mr. Art Stienstra, Sr., of Al Tech Specialty Steel Corporation



was available to coordinate plant activities, and Mr. Dennis



Holzschuh, Task Manager for the U.S. Environmental Protection



Agency, was on site:to monitor the test series.

-------
                            SECTION 1

                          INTRODUCTION


     During the week of April 6, 1981, PEDCo Environmental per-

sonnel conducted an emission sampling program at the steel melt

shop operated by Al Tech Specialty Steel Corporation (Al Tech) in

Watervliet, New York.  The purpose of this test program was to

provide data for assessing the need for revising present New

Source Performance Standards (NSPS) for electric arc furnaces

(EAF) to include argon-oxygen decarburization (AOD) furnaces.

     This plant was selected for source testing for the following

reasons:

     1)   It exhibits best available control -technology.

     2)   The emissions capture and control equipment is repre-
          sentative of the industry.

     3)   Both AOD and EAF furnaces are controlled by the same
          device.

     4)   Emission data could be obtained by nonstandard sampling
          techniques on the typical positive-pressure fabric
          filter.

     Particulate matter concentrations and mass emission rates

were measured at two inlet sites and one outlet site.  Tests at

the two inlet sites were conducted according to U.S. Environ-

mental Protection Agency (EPA)  Reference Method 5.*  The sampling
*
 40 CFR 60, Appendix A, July 1, 1980.
                               1-1

-------
train used for outlet tests was as described in Method 5, but


procedures were adapted to the large exhaust area and correspond-


ing low gas velocity.


     Inlet and outlet tests for particulate matter were performed


simultaneously to enable determination of control efficiency as


well as values for controlled and uncontrolled emissions.  Flue


gas flow rates, temperature, and composition were measured in


conjunction with these tests.  In addition, particle size distri-


bution samples were collected simultaneously at one inlet site


and at the fabric filter outlet.  Method 9* procedures were used


to evaluate visible emissions (VE) from the melt shop and fabric


filter outlet throughout the test.  Visual determinations of


fugitive emissions (FE)  from the fabric filter dust handling


system were recorded according to the proposed Method 22.**


Samples of dust collected by the fabric filter were obtained for


analysis of particle size distribution and elemental composition.


Tests took place simultaneously at all sites, including visible


and fugitive emission sites.  A representative from the NSPS


contractor assisted in coordinating the tests with process opera-


tions.   Subsequent analyses were performed on two outlet particle


size samples and two fabric filter dust samples to determine the


concentration of fluoride,  chromium, lead, and nickel.


     This report documents the activities and results of the test


program.  Section 2 describes the processes that were tested and


the operating conditions during the sampling period.  Section 3
  40 CFR 60, Appendix A, July 1, 1980.
**
  Federal Register, Vol. 45, No. 224, November 19, 1980.


                                1-2

-------
presents the results and discusses them, whereas Section 4



describes the sampling sites and general test procedures used.



Section 5 briefly outlines quality assurance measures and audit



results.  Section 6 details the particulate matter and particle



size distribution sampling and analytical procedures.  The



appendices contain computer output and example calculations



(Appendix A), field data (Appendix B), sample recovery and



analytical data (Appendix C),  MRI process summary (Appendix D),



calibration procedures and results (Appendix E), a quality



assurance summary (Appendix F), and a list of project partici-



pants (Appendix G).
                               1-3

-------
                            SECTION 2



                        PROCESS OPERATION






     Al Tech operates a total of two EAF's and two AOD's in the



production of many different grades of steel, including stainless



steel and high-temperature alloys.  Each EAF and AOD has a rated



capacity of 27.3 Mg (30 tons) and produces an average heat of 29



Mg (32 tons).  The facility was operating 24 hours per day, 5



days per week, during the test series, but this schedule normally



fluctuates between 3 and 5 days per week, depending on product



demand.



     The north AOD vessel and EAF No. 89 are located in the north



end of the melt shop, and EAF No. 90 and the south AOD vessel are



located in the south end of the shop.  Typically, EAF No. 89



feeds the north AOD, and EAF No. 90 feeds the south AOD.  If



necessary, however, the north EAF can feed the south AOD vessel,



and vice versa.  Normal melt shop operation consists of charging



and backcharging the EAF with cold scrap and fluxes; meltdown;



tapping the EAF; charging molten metal, fluxes, and alloys into



the AOD vessel; refining in the AOD vessel; and tapping the



refined metal into a ladle, which is then transferred to the



teeming aisle.
                               2-1

-------
     The EAF charge materials consists of 80 percent scrap and 20



percent additives.  The scrap used at this plant is relatively



clean, contains a high percentage of stainless steel, and is not



pretreated.  The initial additives are generally lime, charge



chrome, ferronickel, and iron ore.  Other materials may be added



during the melting phase.  A minimum amount of oxygen, if any, is



blown into the molten metal while it is in the EAF, because



refining is accomplished primarily in the ADD.  Hot metal is



tapped into a ladle, weighed, and transferred to the AOD vessel.



The average EAF heat time  (tap to tap)  is 3.7 hours.



     Hot metal from the EAF is charged into the AOD vessel,



where major alloy additions are made.  The additions include high-



carbon magnesium, high-carbon chrome, nickel, silicon, molyb-



denum, and aluminum.  The argon-oxygen-nitrogen gas mixture is



blown into the molten bath through tuyeres located in the bottom



of the vessel until the carbon in the metal has been oxidized to



specification.  Reduction mixes are added to remove sulfur and



other impurities.  The refined molten metal is tapped from the



AOD into a ladle and transferred to the teeming area where the



hot metal is poured into ingot molds.  After it is tapped, the



AOD remains idle until another hot metal charge from the EAF is



ready.  The AOD heat (charge through tap) takes about 1.5 to 2



hours, but can last longer if more refining is needed.



     In the emission capture system, two hoods are provided above



each electric furnace and each AOD vessel.  The electric furnace



hoods are situated so that one hood captures most of the





                               2-2

-------
emissions from charging and melting, whereas the other captures



most of the tapping emissions.  The ADD hoods are situated



so that each one captures half the refining emissions.  A divert-



er stack directs the AOD refining emissions toward the hoods to



avoid excessive drift of the emissions as a result of cross



drafts.  Each hood has a set of automatic dampers, which can be



closed to direct more suction to another hood.  The dampers on



the hoods on the charging side of each EAF are always open.  The



dampers on the tapping side open automatically during tapping and



remain closed the rest of the time.  The dampers on the hood of



the AOD vessels are always open because the automatic mechanism



no longer functions.



     Combined exhaust gases from all four furnaces are ducted to



a Wheelabrator-Frye, positive-pressure fabric filter to remove



particulate matter.  The gas flows through the fabric filter at a



rate of approximately 280 m /s (600,000 acfm) and exits via a



monovent type of exhaust.  Mechanical shakers clean the bags in



each compartment at periodic intervals.  Al Tech has assigned two



people to the regular maintenance of the fabric filter.  A



periodic visual inspection is made, and broken bags are changed



when needed.



     Technical data on the process and control system are in-



cluded in Table 1 of Appendix D.



     The operations of the four furnaces and the fabric filter



were monitored by Lalit Banker and William Terry of Midwest



Research Institute  (MRI).  Table 2-1 presents a summary of




                              2-3

-------
               TABLE 2-1.   EAF  AND  AOD  PRODUCTION  SUMMARY3
PEDCo
run No. ,
date
1

4/7/81







2

4/8/81







3

4/9/81






Furnace
North AOD

89 EAF

90 EAF


South AOD


North AOD

89 EAF

90 EAF


South AOD


North AOD

89 EAF

90 EAF

South AOD


Partial or completed
heats tested
Heat No.
98033
08440
98047
98068
08440
08444
08445
98047
08444

98045
98064
98064
98039
08464
08446
08447
08464
08446

98090
98091
98091
98044
08451
08465
08452
08451

Heat
tlme.b
minutes
85
100
190
225
440
240
235
no
155

90
65
210
299
185
225
300
150
230

225
225
340
280
350
225
e
160

Metal
produced
Mg tons
28.0 30.9
27.9 30.8
27.7 30.5
28.1 31.0
25.4 28.0
30.2 33.2
27.2 30.0
28.9 31.8
33.9 37.4
Total
27.9 30.8
30.8 34.0
29.9 33.0
29.9 33.0
28.6 31.5
29.5 32.5
28.8 31.8
32.7 36.1
32.2 35.5
Total
29.1 32.1
30.4 33.5
28.1 31.0
24.9 27.5
30.4 33.5
32.5 35.8
.
35.0 38.6
Total
Tested production
rated
Mg/h
8.7

8.5

6.5


5.2

28.9
8.5

7.1

7.7


6.9

30.2
4.9

7.8

9.8

5.7

28.2
tons/h
9.6

9.4

7.2


5.7

31.9
9.4

7.8

8.5


7.6

33.3
5.4

8.6

10.8

6.3

31.1
 aCompiled from process data  in Appendix D and field data  (test times) in
 Appendix 8.
b Charge-to-tap time for AOD's; tap-to-tap time  for EAF's.

c EAF production  is the weight of metal transferred to an AOD; AOD production
  1s the final  tap weight.
dTested production rates were determined by dividing the total weight of
 metal  produced during a test by the sampling time.  The weight of metal
 produced by a  given furnace during a test was calculated by first
 dividing the minutes of normal operation actually sampled (from charging
 through tapping, not Including delays or patching) by the total  minutes
 of normal operation 1n the heat, and then multiplying by the weight of
 metal  produced during the entire heat.
eThe only normal  operation of this heat sampled  was the tap.
                                       2-4

-------
production data for the test series.  The EAF heat times varied



from 3.1 to 7.3 hours (tap to tap),  and the weight of metal



produced per heat ranged from 24.9 to 32.5 Mg (27.5 to 35.8



tons).  The AOD heat times varied between 1.1 and 3.8 hours



(charge through tap), and the weight of metal produced per heat



varied between 27.9 and 35.0 Mg (30.8 and 38.6 tons).  The



variations in heat times were related in part to delays caused by



one or more cranes that did not operate properly.



     Production rates were calculated to determine if the tests



were conducted during representative operating conditions.  The



tested production rate for each furnace was determined by divid-



ing the total weight of metal produced during a test by the



sampling time.  The metal produced by a given furnace during a



test was calculated by first dividing the minutes of normal



operation actually sampled (from charging through tapping, not.



including delays or patching) by the total minutes of normal



operation in the heat, and then multiplying by the weight of



metal produced during the entire heat.  For comparison, the



normal production rate for all four furnaces operating at once



was calculated at 31.4 Mg/h  (34.6 tons/h), based on an average



heat time of 3.7 hours and an average metal production of 29 Mg



(32 tons) per furnace.  The same heat time was used for the EAF's



and AOD's, because one AOD heat normally occurs for each EAF



heat.  The tested production rate and the normal production rate



both include the effects of delays and intervals between heats



and should therefore be comparable.   The average equivalent



                               2-5

-------
production rate for the test series was 29.1 Mg/h (32.1 tons/h),



which indicated that the tests were conducted while the process



was operating at 93 percent of normal production rate.



     Tests were delayed at various times during the series to



avoid sampling when emissions were not representative of normal



operation.  On April 7, the test was stopped for 10 minutes



because EAF No. 89 was idle for 33 minutes waiting for the pit



crane to clean the tap runner.  At 3:00 p.m. on April 8, the test



was stopped for approximately 15 minutes because two of the



furnaces were not operating.  On April 8, the test was stopped



before the end of the heat in EAF No. 89 because the furnace bay



crane lost power.  On April 9, the start of the test was delayed



for about 30 minutes because the furnace bay crane temporarily



lost power again.



     The fabric filter was operating normally during the test



period.  The fabric filter control panels were monitored hourly



for the duration of the tests.  Test personnel periodically



closed off a baghouse compartment for a short period to move



their sampling equipment.  Because one compartment is always in



the cleaning mode, this meant two compartments were not in



operation during that short period of time.  The pressure drop in



each compartment, the inlet gas temperature, the fan amperage for



each of the three fans, and the number of compartments cleaning



or closed off were observed to assure normal fabric filter opera-



tion.  The pressure drop varied between 1.0 and 1.5 kilopascals



(4 and 6 in. H2O) throughout the tests.  All three fans were







                               2-6

-------
operating at full power during the tests.  The flue gas inlet



temperature fluctuated from 43° to 54°C  (100° to 130°F) during



the test period.



     The test conditions were representative of normal plant



operation, and the data should be useful.



     During the tests, emissions periodically filled the melt



shop.  Some of the emissions were deflected when cranes passed



above the furnaces and vessels and when the cranes were in



position for a charge or tap.  The cross drafts that developed



from open doors in the scrap bay and tapping pit also sometimes



deflected emissions from the hood.  The inoperable automatic



dampers on the AOD hoods were always open and prevented their



optimal use.  Although these dampers could be operated manually,



they were left open throughout the tests in line with normal



operating procedure.  Although the emissions were not always



completely captured, the shop always cleared in 5 to 10 minutes.



Plant workers indicated that emission capture was better when the



capture and control system was new and the automatic AOD dampers



were operating properly.



     Emissions generated by the EAF's were greatest during melt-



down, but emissions were also significant during tapping.  Emis-



sions from the AOD were greatest during the initial stages of the



heat when the oxygen concentration in the blowing gases was the



highest.  Emissions from the AOD during blowing appeared to be



equal to or greater than EAF meltdown emissions.  For long periods



of time, however, the AOD vessel is in a nonblowing position






                               2-7

-------
while operators wait for sample results to determine what alloys



are needed and how much more blowing is necessary to meet final



specifications.
                                2-8

-------
                            SECTION 3



                       SUMMARY OF RESULTS





     This section details results obtained from the emission test



program.  Particulate matter tests were run simultaneously at two



inlet sites and the fabric filter outlet, and particle size



distribution tests were run at one inlet site and at the outlet.



Visible emissions from the melt shop and fabric filter outlet



were evaluated concurrently with particulate test runs, as were



fugitive emissions from the fabric filter.  Fabric filter dust



samples were collected and analyzed for trace elements and



particle size distribution.  Table 3-1 summarizes the type and



number of samples that were collected.



     In brief, uncontrolled particulate matter concentrations



averaged 245 milligrams per dry normal cubic meter (mg/dNm ) at



20°C and 101 kilopascals  (kPa), or 0.1076 grains per dry standard



cubic foot (gr/dscf) at 68°F and 29.92 in.Hg.  At the outlet,



particulate concentration averaged 3.46 mg/d$m  (0.0015 gr/dscf),



to yield a 98.6 percent control efficiency.  Both levels of



concentration were in the range of expected values, which were


                                           1 2
based on previously reported data on EAF's. '



     Individual 6-minute set averages of visible emissions from



the melt shop ranged from 0 to 15 percent opacity during charging
                                3-1

-------
                              TABLE 3-1.  SAMPLES COLLECTED AT AL TECH SPECIALTY STEEL
Sampling site
No. 1 - North
No. 2 - South
No. 3 -
Fabric fil-
ter outlet
Shop exit
Fabric filter
dust handling
system
Sample type
Particulate
Particle
size
Particulate
Particulate
Particle
size
VE
VE
FE
Dust
Sampling
method
EPA 5
High-capacity
impactor
Impactor
EPA 5
Modified EPA 5
Impactor
EPA 9
EPA 9
EPA 22
Grab
Number3
of
samples
3
3
3
3
3
3
2C
2C
2c
3
....
Time for
each sample
~ 6 h
~ 3 h
-20 min
- 6 h
~ 5-1/2 h
~ 5 h
- 7 h
~ 7 h
~ 5 h
1 per
day
Additional analysis
Type
Condensibles
b
Condensibles
Organic and
inorqanic
Condensibles

Trace metals,
particle
size^
No.
3
3
3

3
Method
Gravimetric
Gravimetric
Back-half E/C
extract

SSMS,d Coulter
 I
ro
     Does not include preliminary, blank, or duplicate runs.
     Two samples were analyzed later for fluoride by EPA Method 13B and for chromium,  lead,  and nickel  by atomic
     absorption.
    GThe third run could not be performed because of unfavorable weather conditions.
     Spark source mass spectroscopy.

-------
and tapping operations.  Average opacities of individual sets



ranged from 0 to 22 percent during refining and other process



operations.  Visible emissions from the fabric filter outlet were



zero percent opacity, even after compartment cleaning cycles.



     These and other results are presented and discussed in



detail in the following sections.  Results are grouped by emis-



sion type.  The sections in each group describe the sampling



scheme used at each site, summarize data and results, and discuss



the results.





3.1  PARTICULATE MATTER



     Two inlet sites and the fabric filter outlet were tested



simultaneously.  Site 1 represented emissions from the north EAF



(designated by the plant as No. 89) and the north AOD; Site 2



represented emissions from the south pair of furnaces (including



EAF No. 90); and Site 3 represented fabric filter outlet emis-



sions.



3.1.1  Sampling Scheme



     Tests at all sites commenced simultaneously and ran for



approximately the same time until respective traverses were



completed, about 5 to 6 hours.  This procedure enabled calcula-



tion of control efficiencies and emission factors.  Each test run



at Site 1 was to have included two integral EAF heats (from



initial charge through the final tap)  of approximately 3 to 3.5



hours each and two complete AOD heats of approximately 1.5 to 2



hours each.  Because a malfunctioning crane caused shifts in the
                               3-3

-------
normal sequence of operation, the second and third tests were

stopped prior to completion of the second EAF heat, and the third

test included only portions of the AOD heats.  Nevertheless, the
               A
actual tests were generally representative of integral EAF heats.

So that tests at Site 2 could be conducted simultaneously with

other tests, the sample times were not coordinated to include an

integral number of heats, as the operating schedule of the south

furnaces was staggered from that of the north furnaces.  Outlet

tests at Site 3 also could not be coordinated to represent an

integral number of heats (because of the different furnace

schedules), but they were conducted in conjunction with tests at

other sites.  Fabric filter cleaning cycles were sampled in the

normal fashion as they occurred.

     The NSPS contractor representative, who was on site to

monitor process operations, assisted in the coordination of tests

with process conditions.  Based on his observations, tests were

interrupted if one of the EAF's experienced an operational delay

of longer than 20 minutes.   Delays from 1 to 1.5 hours could be

tolerated for the AOD's without interrupting tests, because that

amount of AOD downtime would normally occur within the time frame

of an EAF heat.

     Particulate matter sampling and analytical procedures at

both inlet sites followed those described in EPA Methods 1, 2, 3,

and 5 of the Federal Register.*  A Method 5 sampling train and
 40 CFR 60, Appendix A, July 1, 1980.
                               3-4

-------
analytical procedures were used on the outlet tests, but the



sampling procedures were modified by sampling at a constant rate,



at a site that did not meet minimum Method 1 criteria, and at



fewer points than specified by Method 1.  The sampling rate was



based on the estimated average velocity of the gas stream at the



sampling location, which was calculated by dividing the total



inlet flow rate measured by Method 2 by the total exhaust area



represented by the sampling cross section.  Three tests were run



at each site.  Integrated gas samples were collected once at each



site (according to Method 3) to verify that the gas streams were



essentially air.  Additional molecular weight determinations were



not made.



3.1.2  Gas Conditions and Particulate Emissions



     Summaries of the measured stack gas and particulate emission



data are presented in Tables 3-2 and 3-3.  Volumetric flow rates



are expressed in actual cubic meters per second (m /s) and actual



cubic feet per minute (acfm) at stack conditions.  Flow rates



corrected to zero percent moisture and standard conditions [20°C



and 101 kPa  (68°F and 29.92 in.Hg)]  are expressed as dry normal



cubic meters per second (dNm /s) and dry standard cubic feet per



minute (dscfm).  Average stack gas velocities are expressed in



actual meters per second (m/s)  and actual feet per second (ft/s)



at stack conditions.  Particulate concentrations in Table 3-3



are reported in milligrams per dry normal cubic meter and



grains per dry standard cubic foot.   Emission rates are expressed



in kilograms per hour and pounds per hour.  The product of the
                              3-5

-------
                TABLE  3-2.  SUMMARY OF GAS STREAM CHARACTERISTICS*
Run No.
AIP-1
AIP-2
AIP-3
Date .
(1981)
4/7
4/8
4/9
Average
A2P-1
A2P-2
A2P-3
4/7
4/8
4/9
Average
A3P-1
A3P-2
A3P-3
4/7
4/8
4/9
Average
Flow rate
dNm3/s
130.0
121.8
127.2
126.3
129.9
129.3
130.4
129.9
259.9
251.1
257.6
256.2
dscfm
275,500
258,200
269,600
267,800
275,200
273,900
276,300
275,100
550,700e
532,100
545,900
542,900
Temperature
°C
43
47
42
44
46
48
45
46
42
48
46
45
°F
110
117
107
111
116
118
114
116
108
119
114
114
Moisture,
%
0.81
0.89
1.25
0.98
0.65
0.69
1.29
0.88
0.41
0.68
1.17
0.75
Velocity0
m/s
19.1
18.2
19.0
18.8
19.3
19.3
19.7
19.4
2.4
2.4
2.5
2.4
ft/s
62.8
59.8
62.2
61.6
63.3
63.4
64.7
63.8
3. Of
7.9
8.2
8.0
Flow rate
m3/s
139.6
133.0
138.4
137.0
140.7
141.1
143.9
141.9
275.7
273.0
282.0
276.9
acfm
295,800
281 ,900
293,300
290,300
298,100
298,900
304,800
306,600
584,100f
578,600
597,600
586,800
 Average C0£ <0.2%, 02 = 20.3%.

 Dry normal cubic meters per second at 20°C and 101  kPa, and dry standard
 cubic feet per minute at 68°F and 29.92 in.Hg.

cVelocity at stack conditions.

 Flow rate at stack conditions.
g
 The outlet standard flow rates were determined by summing flows at both
 inlet sites.

 Outlet gas velocities and actual  flow rates are based on measured inlet flow
 rates converted to outlet conditions.
                                     3-6

-------
          TABLE 3-3.  SUMMARY OF FILTERABLE PARTICULATE EMISSIONS
Site
Inlet No. 1
north
furnaces
Run
No.
'A1P-1
A1P-2
A1P-3
Average
Inlet No. 2
south
furnaces
A2P-1
A2P-2
A2P-3
Average
Combined
inletsb
1
2
3
Average
Outlet No. 3C
A3P-1
A3P-2
A3P-3
Average
Concentration9
mg/dNm3
217
182
523
307
198
240
111
183
208
212
315
245
3.27
2.88
4.24
3.46
gr/dscf
0.0948
0.0794
0.2288
0.1343
0.0867
0.1047
0.0485
0.0800
0.0908
0.0924
0.1397
0.1076
0.00143
0.00126
0.00185
0.00151
Mass emission rate
kg/h
102
80
240
140
93
111
52
86
194
191
292
226
3.06
2.61
3.94
3.20
Ib/h
224
176
529
309
205
246
115
188
428
421
643
498
6.75
5.75
8.68
7.06
Isokinetic
rate, %
108
102
102

104
104
105



!07d
106
102

 Milligrams per dry normal cubic meter at 20°C and 101  kPa, and grains per
 dry standard cubic foot at 68°F and 29.92 in.Hq.
L                                              J
 Combined inlet mass emission rates represent the sum of results for Sites 1
 and 2; concentrations are weighted averages based on standard flow rates.

C0utlet mass emission rates are based on measured outlet concentrations and
 total  inlet standard flow rates.

 Isokinetic sampling rates for outlet runs were calculated by using average
 gas velocities that are based on inlet flow rates converted to outlet
 conditions.
                                     3-7

-------
concentration and the volumetric flow rate is the mass emission



rate.  The particulate data represent filterable material col-



lected in the sample probe and on the filter, both of which were



heated to approximately 121°C  (250°F).  The isokinetic rate is



the ratio of the velocity of the sample gas stream entering the



nozzle to the local stack gas velocity, expressed as a percent-



age.



     The volumetric flow rates at inlet Sites 1 and 2 averaged



126 dNm3/s (268,000 dscfm) and 130 dHm3/s (275,000 dscfm).



The outlet flow rate averaged 256 dNm /s  (543,000 dscfm).  The



flow for each outlet run represents the sum of the standard flow



rates measured at the two inlet sites.



     The actual flow rate at inlet Site 1 averaged 137 m /s



(290,000 acfm)  at 44°C (111°F) and less than 1 percent moisture,



and was equivalent to a gas velocity of 19 m/s (62 ft/s).  The



actual flow rate at inlet Site 2 averaged 142 m /s (301,000 acfm)



at 46°C (116°F)  and less than 1 percent moisture, which equalled



a gas velocity of 19 m/s  (64 ft/s).  The outlet flow rate aver-



aged 277 m3/s (587,000 acfm)  at 45°C (114°F) and less than 1



percent moisture, which represented an average gas velocity of



2.4 m/s (8.0 ft/s)  at the sampling cross section.  Outlet veloci-



ties and flow rates are based on inlet standard flows and mea-



sured outlet stack conditions of temperature, pressure, and



moisture.



     For calculation purposes the stack gases were essentially



air.  For one run at each site, the carbon dioxide (C02)



                               3-8

-------
concentration averaged less than 0.2 percent, and the oxygen con-



centration was 20.3 percent by volume.



     The combined inlet concentration of filterable particulate



matter averaged 245 mg/dNm  (0.1076 gr/dscf).  This is the



average of concentrations measured at both inlet sites, weighted



according to measured standard flow rates.  At inlet Site 1 the



average particulate concentration was 307 mg/dNm  (0.1343



gr/dscf), and at Site 2 the average particulate concentration was



183 mg/dNm  (0.0800 gr/dscf).   These concentrations correspond to



mass emission rates of 140 kg/h (309 Ib/h) at Site 1, 86 kg/h



(188 Ib/h) at Site 2, and 226  kg/h (498 Ib/h) total uncontrolled



emissions.



     The outlet particulate concentration averaged 3.46 mg/dNm



(0.00151 gr/dscf), with a corresponding mass emission rate of



3.20 kg/h (7.06 Ib/h).  The concentration results for each run



are actually representative of the exhaust stream from four of



the eight fabric filter compartments, but emission rates are



based on the total system flow.



     All isokinetic sampling rates were between 102 and 108



percent.   The outlet values are based on average gas velocities



at the sampling cross section, which were calculated from mea-



sured inlet flow rates and outlet stack conditions.



3.1.3  Control Efficiencies and Emission Factors



     Control efficiencies were calculated by dividing the dif-



ference between the outlet and weighted inlet particulate con-



centrations by the weighted inlet value.  Table 3-4 presents a





                               3-9

-------
                               TABLE 3-4.   FILTERABLE PARTICULATE COLLECTION EFFICIENCY
Run
1
2
3
Average
Inlet concentration3'
mg/dNirr
208
212
315
245
gr/dscf
0.0908
0.0924
0.1397
0.1076
Outlet concentration
mg/ dNnP
3.27
2.88
4.24
3.46
gr/dscf
0.00143
0.00126
0.00185
0.00151
% efficiency
98.4
98.6
98.7
98.6
U)

M
O
       Weighted  average from  Sites  1  and  2.

       Milligrams  per  dry normal cubic meter at  20°C and  101  kPa,  and  grains  per dry standard cubic foot at
       68°F  and  29.92  in.Hg.

      cPercent efficiency = Cin1et  -  Coutlet
                                 'inlet
                                            x  100.

-------
summary of filterable particulate concentrations and indicates



the fabric filter collection efficiency for each run.  Control



efficiencies were 98.4, 98.6, and 98.7 percent on the three test



days.



     Table 3-5 presents particulate emission factors for uncon-



trolled and controlled emissions, which were calculated by



dividing the appropriate hourly mass emission rate by the corre-



sponding furnace metal capacity.  Results are reported in kilo-



grams per hour per megagram of furnace capacity (kg/h per Mg)



and in pounds per hour per ton (Ib/h per ton).  Based on a total



capacity of 109 Mg (120 tons) for the four furnaces, the emission



factor for uncontrolled emissions averaged 2.1 kg/h per Mg, or



4.2 Ib/h per ton.  It was possible for two runs at Site 1 to be



conducted over an integral number of heats.  Each run consisted



of two EAF heats and two AOD heats.  Emission factors for these



tests should be more representative of uncontrolled emissions on



a per-heat basis.  Results for the two runs averaged 1.67 kg/h



per Mg (3.33 Ib/h per ton) based on a metal capacity of 54.5 Mg



(60 tons ).  The average controlled emission factor for all four



furnaces was 0.03 kg/h per Mg (0.06 Ib/h per ton).



     Emission factors shown in Table 3-6 are based on actual



production data.  Results were calculated by dividing the fil-



terable mass emission rate by the corresponding equivalent



tested production rate.  Emission factors are reported in kilo-



grams per megagram (pounds per ton) of metal produced.  The



average uncontrolled emission factor was 7.8 kg/Mg  (15.6 Ib/ton),





                              3-11

-------
                   TABLE 3-5.  PARTICULATE  EMISSION  FACTORS'
Run No.
AlP-lb
1
AlP-2b
2
3
Average
Uncontrolled
kg/h per Mg
1.88
1.78
1.47
1.75
2.08
2.07
(1.67)
Ib/h per ton
3.73
3.57
2.93
3.51
5.36
4.15
(3.33)
Run No.
1
2
3
Average
Controlled
kg/h per Mg
0.028
0.024
0.036
0.029
Ib/h per ton
0.056
0.048
0.072
0.059
 Factors  are  based  on  emissions  per  unit  of  furnace metal  capacity  in  kilograms
 per  hour per megagram (pounds per hour per  ton).
 Tests were conducted  for  an  integral  number of  heats  and  represent emissions
 from the north furnaces at a metal  capacity of  54.4 Mg  (60  tons).   All  other
 runs represent total  emissions  from the  four furnaces at  a  metal capacity  of
 109  Mg (120  tons).

'Values in parentheses (  ) are for the integral  heat tests on  the north
 furnaces.
                                    3-12

-------
          TABLE 3-6.  PARTICULATE EMISSION FACTORS  BASED  ON  PRODUCTION1
Run No.
1
2
3
Average
Metal production
rateb
Mg/h
28.9
30.2
28.2
29.1
tons/h
31.9
33.3
31.1
32.1
Emission factor0
Uncontrolled
kg/Mg
6.7
6.3
10.4
7.8
Ib/ton
13.4
12.6
20.7
15.6
Controlled
kg/Mg
0.11
0.09
0.14
0.11
Ib/ton
0.21
0.17
0.28
0.22
 Calculated by dividing the filterable mass emission rate by the  corresponding
 average metal production rate.

bFrom Table 2-1.

cKilograms per megagram (pounds per ton) of metal  produced.
                                     3-13

-------
based on a production rate of 29.1 Mg/h (32.1 ton/h).  At the


same production rate, controlled emissions averaged 0.11 kg/Mg


(0.22 Ib/ton).

3.1.4  Discussion


     In general, the particulate tests were conducted according

to schedule.  No problems were encountered with the sampling


equipment, and the few problems associated with the process

operation were considered minor.  The report does not include the

results of preliminary tests that were conducted at each site to

compare particulate loadings with estimated sampling times and to

eliminate any problems associated with test coordination or

physical sampling maneuvers.  This section discusses validity of

results, deviations in test methods and calculations caused by


the fabric filter site configuration, and effects of process


operations.

     The primary purpose of the long sampling time was twofold:

(1) to collect approximately 25 to 50 mg in the front-half of the

outlet sampling train so as to minimize handling and weighing

errors; and  (2) to satisfy NSPS minimum requirements for sample

time and volume.  The actual catch weights were between 17 and 25

mg, which were considered satisfactory.  The actual minimum

sampling time and volume were 5.3 hours and 5.9 dNm  (208 dscf).

These met the minimum criteria of 4 hours and 4.5 dNm  (160 dscf)

set forth in Subpart AA of the Federal Register.*
*
 40 CFR 60, Subpart AA, July 1, 1980.
                               3-14

-------
     The back-halves of each inlet and outlet run were analyzed



for condensible matter.  These results are included in the com-



puter printouts in Appendix A, but are not summarized here



because they are considered to be biased.  The results do not



agree well with expected values based on previously reported



tests at similar installations.  Probable cause of the biases is



believed to be the long sample line used between the heated



filter and first impinger, even though it was Teflon-lined.



     The inlet flow rates measured at Sites 1 and 2 were within



6 percent of each other, which indicated that the three induced-



draft (I.D.)  fans were operating effectively to promote the



equal capture of emissions from each half of the melt shop.  The



sum of the inlet flows compared very well with the system design



flow rate of 280 m /s  (600,000 acfm), as all runs were within 4



percent of this .value.



     The outlet volumetric flow rates, dry and at standard con-



ditions, were assumed to be equal to the sums of the inlet



standard flow rates.  This assumption was necessary because the



site configuration made it impossible to obtain accurate velocity



data at the outlet.  Flows at stack conditions were calculated



from the standard flow rates by use of measured values of tem-



perature, pressure, and moisture at the outlet.



     The procedure just described is typical for tests at posi-



tive-pressure fabric filters without stacks, and assumes no air



leakage.  Air inleakage at the I.D. fans was probably no more



than 10 percent, as evidenced by the agreement between measured





                               3-15

-------
and design flows.  In addition, air leakage through the open



grating at the bottom level of the fabric filter was minimized by



covering the openings during the test period.  At Al Tech these



open gratings represented sources of dilution air.  Comparison of



inlet and outlet moisture contents seemed to indicate some



dilution, but the limits of accuracy for the moisture determina-



tions at the 1 to 2 percent level are probably greater than the



reported differences.  The close agreement between inlet and



outlet gas temperatures indicated that any entry of dilution air



was negligible.  A significant inflow of ambient air would have



caused a temperature decrease, which was not evident.  For these



reasons, the amount of dilution air that entered the system was



considered to be minimal, and results reported for the outlet



flow rates should be representative of actual conditions.



     The fabric filter site configuration required that several



modifications be made to EPA reference methods for their use in



the outlet tests.  Although they could not be analyzed precisely,



the effects of these deviations on outlet particulate concentra-



tion results were considered to be relatively minimal.  The two



deviations from Method 1 were the use of a sampling location less



than two equivalent duct diameters downstream from the nearest



disturbance and sampling at fewer than the minimum number of



points.  The three deviations from Method 5 were the lack of



velocity monitoring at individual sampling points, use of a



constant sampling rate at all points during a given test, and



traversing only half of the large exhaust area for each test.





                              3-16

-------
These deviations, which generally would apply when testing



positive-pressure fabric filters without stacks, are discussed



in the following paragraphs.



     The outlet site configuration did not provide any sampling



location capable of meeting minimum Method 1 criteria.  The



throat of the monovent was selected as the sampling location,



primarily because it was the smallest cross-sectional area avail-



able.  This location would be less prone to concentration biases



caused by faulty bags, and would provide the highest gas veloci-



ties.



     The reason for sampling four rather than eight points per



compartment was to lessen the possibility of biasing results.



Because moving from one traverse point to another during sampling



required test personnel to enter a compartment while the fabric



filter was operating, extraneous dust could have been stirred up



by bumping the probe against nearby beams or by personnel activ-



ity, and results could have been biased if any such dust had



entered the nozzle during sampling.  By sampling only four com-



partments and using extreme care during point changes, we were



able to avoid these potential problems in all of the tests.



     It should be noted that a better sampling approach for this



type of fabric filter configuration would be to sample from out-



side on the baghouse roof and to use ports located in the mono-



vent throat.  This would reduce the possibility of sampling



extraneous dust and shorten the time required to change traverse



locations.  More points and compartments per run could then be





                                3-17

-------
sampled.  Because safe access to the roof was not readily avail-



able, ports were not installed at this plant.



     A constant sampling rate was used for the outlet tests



because individual point velocities could not be accurately



measured to make isokinetic sampling rate adjustments.  The



preliminary traverse of three compartments verified the inac-



curacy of velocity measurement attempts.  Most of the very low



velocity heads were readable on an expanded scale manometer;



however, turbulence at the low flows caused fluctuations and



frequent negative readings.  Thus, an average gas velocity at the



outlet sampling location was calculated from previous data



obtained at the two inlet sites and used to set a constant sam-



pling rate.  The two inlet flow rates were calculated and



totaled; then the sum was converted from standard conditions of



temperature and pressure to outlet conditions.  This flow was



then divided by the total cross-sectional area of all eight



compartments represented by the sampling plane.  The resultant



average gas velocity was assumed to represent each traverse



point.  For each outlet test, inlet data obtained from the pre-



vious day's activities were used to estimate the average veloc-



ity.  The constant isokinetic sampling rate to be used that day



was then calculated from this estimated velocity.  For data



recording and computer calculation purposes, an equivalent



velocity head was calculated and entered on the field data sheets



for each traverse point as if it had actually been measured.
                                3-18

-------
Other parameters entered on the data sheets were measured accord-



ing to normal procedures.



     All outlet computer calculations in Appendix A for flow



rate, emission rate, velocity, and isokinetics are based on these



estimated velocities; however, results reported in tables and



text have been adjusted to reflect measured rather than estimated



values.  This was accomplished by dividing the measured total



inlet flow rate by the flow rate used initially to estimate the



outlet velocity.  Results were adjusted by applying this ratio,



either directly or inversely as appropriate, to computer outputs



that were based on estimated values.  These calculations are



shown in Appendix A.



     The average isokinetics indicated for the outlet tests were



all within the acceptable range (100 + 10 percent).  These iso-



kinetic calculations included the assumption that the average



velocity for the four compartments tested per run was approxi-



mately equal to the average velocity for all eight compartments.



We believe this is a reasonable assumption  (+_ 10 percent),



although the middle compartments may have had slightly higher



velocities than the end ones.  Another factor affecting iso-



kinetic and velocity results was the fabric filter cleaning cycle



time.  Reported results do not account for the time when only



seven compartments were operating instead of eight, but the



greatest effect would occur if one compartment were always off



line.  This would reduce reported outlet isokinetics by a factor



of 0.875 and increase gas velocities by the inverse of this




                               3-19

-------
factor.  Isokinetics would be between 89 and 94 percent; there-



fore, emission results should not be affected significantly by



this consideration.



     Although average outlet isokinetics were acceptable, values



at individual traverse points could be much different, depending



on the local gas velocities.  The range of actual isokinetic



variation was difficult to gauge without valid point velocity



data, but an estimate of plus or minus 50 percent seems realis-



tic.  The overall isokinetic rates, however, were within speci-



fied limits, and any point-specific biases should tend to be



averaged.  Even so, it is expected that the precision of the



method used (constant sampling rate) is less than that of Method



5, which requires isokinetic sampling at each point.  Results



should be viewed in this light, even though they were fairly



consistent and appear to be representative.



     Evaluation of the process data furnished by MRI and actual



sampling times indicated that operation conditions were repre-



sentative during the tests.  The equivalent tested production



rate averaged 93 percent of the maximum production rate of the



four furnaces.  Although the metal production of individual



furnaces varied considerably, the overall production rates for



each test were within 10 percent of each other.  A periodically



malfunctioning crane caused a shift in the normal sequence of



furnace operations and several extended furnace delays.  Tests



were interrupted when the emissions were considered to be sig-



nificantly affected.  The results of Tests 1 and 2 did not seem





                               3-20

-------
to be affected by the delays or variation in furnace operations,



but Test 3 results may have been.  The concentration at Site 1



increased and the concentration at Site 2 decreased, but these



changes could not be related to specific process activities.



Indeed, the equivalent production rates for Test 3 showed the



reverse:  a decrease for Site 1 furnaces and an increase for Site



2 furnaces.  The overall increase in uncontrolled emissions was



verified by the corresponding increase in outlet emissions, but



the overall production rate for Test 3 was slightly lower than



the previous tests.  The increase in Test 3 emissions caused a 12



percent increase in the average outlet concentration, which was



considered relatively insignificant.  Overall results were



therefore taken as being representative of normal operating



conditions.





3.2  PARTICLE SIZE



     Tests for particle size distribution were conducted at Site



1 (the north furnaces)  to represent uncontrolled emissions, and



at Site 3 to represent controlled emissions.  These tests were



performed in conjunction with particulate matter tests.



3.2.1  Sampling Scheme



     Inlet particle size tests were conducted over an entire heat



cycle to represent average emissions, and during shorter inter-



vals, to represent different process modes.  Tests during inte-



gral heats were initiated at the beginning of a charge for



either the North AOD or EAF No. 89, whichever occurred first.
                               3-21

-------
Tests continued for one heat and concluded at the end of tapping



operations at EAF No. 89, which yielded a sampling time of



approximately 3 to 3.5 hours.  This time should have included an



entire ADD heat of between 1.5 and 2 hours.  Because a malfunc-



tioning crane caused as shift in the sequence of normal opera-



tions, the first test included almost two ADD heats, and the



third test included portions of two different AOD heats.  The



shorter particle size runs were performed at various times during



the second half of each particulate test.  The sampling times for



these shorter tests were approximately 20 minutes, adjusted as



necessary to obtain proper loadings.



     Outlet particle size distribution samples were collected



simultaneously with each particulate test, which yielded a 5-



hour sampling time for each run.  Each test was conducted in the



same four fabric filter compartments as the coinciding particu-



late tests, but at different times to minimize interferences.



Because of the overlapping schedules between north and south



furnaces, no attempt was made to represent integral heats.



Fabric filter cleaning cycles were sampled as they occurred.



     The integral heat runs at the inlet were coordinated with



process operations ^.with the assistance of the NSPS contractor



representative.   Based on his observations, tests were inter-



rupted as necessary to avoid sampling during unrepresentative



conditions.



     Andersen Mark III Cascade Impactors were used to collect the



shorter inlet samples and all of the outlet samples, and an




                               3-22

-------
Andersen Heavy Grain Loading Impactor was used to obtain integral



heat samples.  All inlet samples were collected at an average



velocity point in the north furnace duct.  Velocity data were



obtained periodically during each run by the use of Method 2



equipment.  Outlet samples were run in duplicate, with each



impactor positioned at one of the four sampling points per



compartment.  The sample time for each run was divided equally



among the same four compartments in which the particulate tests



were being conducted.  Because accurate velocity data could not



be obtained at the outlet site, estimated velocities calculated



from previous inlet data were used to set approximate isokinetic



sampling rates.  The results of three runs for each type of



sample are included in the report.



3.2.2  Particle Size Distributions and Fractional Efficiencies



     Cumulative distribution curves represent the total weight



percent of particulate matter smaller than the indicated aerody-



namic particle diameter in micrometers.  Each distribution curve



was plotted manually and represents the best-fit curve through



individual and average test data points.  Each data point was



plotted manually and indicates both the 50 percent effective cut-



size of each impactor stage and the cumulative weight percent of



material collected in subsequent stages.



     The three cut-points for each Andersen Heavy Grain Loading



Impactor test at Site 1 were determined graphically from informa-



tion supplied by the manufacturer.  Cut-points for the eight Mark



III Impactor stages were calculated by computer programs
                               3-23

-------
contained in "A Computer-Based Cascade Impactor Data Reduction



System,"  (CIDRS) developed for EPA by Southern Research Institute



(SRI).   All particle size results are based on a particle



density of one gram per cubic centimeter.  Data reduction and



intermediate result calculations for both types of impactors were



performed by the CIDRS programs with moisture contents obtained



from simultaneous particulate tests.  All calculations and



results are included in Appendix A.



     Figure 3-1 shows the average cumulative distribution curve



for uncontrolled emissions over an entire heat cycle.  Actual



results were limited to three cut-points between 1.6 and 13 ym,



but the curve was extrapolated down to a diameter of 0.5 ym for



comparison purposes.  The average distribution indicated that 50



percent by weight of uncontrolled particulate emissions consisted



of particles with aerodynamic diameters of 0.55 ym or less.



Eighty percent by weight had diameters less than or equal to



10 ym.



     Figure 3-2 shows the results of three runs conducted at Site



1 during various times of process operation.  The three cumula-



tive distribution curves distinctly indicate a short-term varia-



tion of emissions.  Run 1 represented EAF charging emissions, Run



2 represented emissions during the EAF melting phase, and Run 3



represented emissions from both an AOD and an EAF.  The percent



by weight of emissions that had diameters equal to or smaller



than 10 ym varied among the three runs from 54 to 29 to 74 per-



cent, respectively.  Only the run representing AOD and EAF



emissions was similar to average integral heat results.



                               3-24

-------
to
 I
to
01
              WJ
              M.t
            <->  t
               1.1
                                        irt
                                                   rtt
                                                      i	iQ.EppHaai; JUCEJU.^
                                                           itt

                                                                                                      ttr
                                                                                                               —i
                                               1.0                            10.0
                                                AERODYNAMIC PARTICLE SIZE, micrometers        A1PS-1  O
                                                                                         A1PS-2  A
                                ,                                                         A1PS-3  0

                                                                                         	 EXTRAPOLATED

                          Figure  3-1.  Average  particle  size results for uncontrolled
                                              emissions, Site No.  1.
100

-------
Ul
I
ro
                                                                              r EAF MELT  AND ADD
                                             AERODYNAMIC PARTICLE SIZE, micrometers
A1PM-1A O
A1PM-2B A
A1PM-3B D
                        Figure 3-2.   Particle size  results for uncontrolled  emissions
                                during  various furnace operations,  Site No.  1.

-------
     Figure 3-3 shows the average distribution curve for the



outlet samples.  Results indicated that 50 percent of the mass



emissions consisted of particles having aerodynamic diameters of



5 ym- or less.  Sixty-six percent by weight had diameters smaller



than or equal to 10 ym.



     Table 3-7 presents the fractional collection efficiencies



for various size ranges.  Weight percents in each size range were



determined from the average inlet and outlet cumulative distri-



bution curves plotted in Figures 3-1 and 3-3.  These percentages



were multiplied by the average inlet and outlet particulate



concentrations shown in Table 3-3 to calculate controlled and



uncontrolled mass loadings in the respective size ranges.



Fractional efficiencies were calculated for each size range by



dividing the difference between inlet and outlet concentrations



by the inlet values.  Efficiency ranged from a high of 99.7 per-



cent for particles smaller than 0.5 ym in diameter to a minimum



of 96.2 percent for particles between 5 and 10 ym in diameter;



the overall efficiency was 98.6 percent.



3.2.3  Discussion of Results



     Results of all the tests are not reported.  Some runs were



not included because of poor isokinetics, undesirable stage



loadings, or problems related to impactor assembly.



     When evaluating these results, one should remember that



particle sizes are in terms of aerodynamic diameters based on a



particle density of 1 g/cm .  Cumulative distribution curves



based on physical diameters and actual density would be shifted






                               3-27

-------
•*.•
•*.•
tt.l
M
M
Ul
ISl
M Oft
«/> "
o
UJ
5"
*— 1
o
ae
*""" M
1 »
>fi
53 »
IE M
to
£ «•
W _ M
1 1°
fO t-
8
CUMULATIVE PERCEN
— M «i • 8
M
t.t
•.1








=
q
— J
3













J
L 	 	 I.
t:.:::., il
L 	 I
^Tt7 I'll JT3
iTF-i.: ^f-f?4- ?*f




—*— t :•• *- *JJ-
-*...».-«.- ..*. « , — ^,
— * ' • T * t • — *
-J! I-*.. — ~ — —
S^-
&d^^
-r^f-r^-
_*_...x.. :;



c
^







.ill..
iiin
1+- t^+ri
l—. — L 	 L* — j-.
= ?i^ ^
+ £ 5 ffl



n 'i , min*
; "'t

Jggr^b; .= = g.
. i 1 4l J>Mt •




: rrj — -r: ~4- -f
» -*_^ ^_ -^ — — [7
r — Er = = =
;SE=^-=3
5t553"-^
•* 7**; ,X — T- * ^
. , • r
ilSSIl!!;
* ;' 	 I1; '•• "•
Trttrssr^g
-. *^M f4. .*. -*. — .
^ „, ^ jit — — i
1.K llil ..4. — — I
' • ! ' '

null!
r r ~
. t •

_J...J..J
iL. _
IT* T




S rrr-
lj k« —



*• *»4- —
ifW
JT

ibi








...


*



U

!

,ii!










1





...
I.1

ill







_..!.

!
-r-r: -nft






... — u



-F
d




•H




M




r-




- -f
^4 • • .
: J±:
th^?±
=ifiS

| 	 U -f4-J.
I I
---4-4-
i
f±t-trr
" T "^l




Ti
TJ
:m
t .lit U^r
r^Lr
•5nl
r ii
ii i|i

I L.
:$$
rl^
t-;i 2|
4- 4- • + ' '|i
::t^
rn
- *-*j* -*—
4> • . . *1»




I rTj-n;
:^rr


, i
;|l!

1
:T?
:fi
• Ii


•U-
T'
^
•Y1
-*i»
H-
^
~r*i








1 i
II'

]i
T
1_
\K
'!'


*t^
M— •

^




N








-









. 1

,
:
^r;


~.
— 14
— U
c;i
if..
-^
iiz n




I
^E =
m





1
(
1 ''!

li |(
.. *±i.
^.T
. \


-rt — ••
n •"•*:
JJ
9=^:





	

r*»f
t !1
ll'

















***t
:
'!?- * i







-

x





-


,i


^



•-
^
^










._ 	 I
___ 	 I
T-- 4TI
	 Tt-

. i

fel:i±I
;--{•! -*t*f
rK^T^
ifH:il
— r-t-^ ' i i
III 3.T7T


: 1 1 1
-till -4,
: , t • U
--j-pTtir



..|. n UIJT

ijllti^lZ
ntrcrln:^:
— rtr— -n-~
r^1 f " if T"
; : i ii ' • ,;,



H^H^
4..I -£. ' '
-rtf 7— -r1 —
^2n;?tE=
^nr-jr-^.
— ^-* J^*- — —
( . .. 1.

*-4i-*tr ^T-
1 ' j t i ' [ .
mtc^i
rttr^T
-U^. m zti zn
*0.<-t.f«J — -.—I.,..
ill I -! it
in i ^^

,11'






•r~
=1



"





L.,.,iIl£ElLlIi
' illl

1
~W
ST'Tf

" |r
-LUii
-#
-.-v

>- r^i

Sis
,f 1
11 ill1
	 p^


ill
1 II

u IF
_.£ in
1 ii


r-4. ,*.






I'! :"•


[•i M
I






...












.J




-










Lili^L
-
-*t
_ * *
- t
_•
-
— s
— •
1.0 10.0 100
AERODYNAMIC PARTICLE SIZE, micrometers A3PS-1AO
A3PS-ZAA
A3PS-3AD
Figure 3-3.  Average particle size results for controlled emissions, Site No. 3.

-------
                    TABLE 3-7.   SUMMARY OF PARTICLE SIZE DISTRIBUTION AND FRACTIONAL  EFFICIENCY
 I
to
vo

Cumulative weight percent8 less Inlet
than larger stated size Outlet
Weight percent In stated size Inlet
range Outlet
Partlculate concentration , Inlet
In stated size range, mg/dNm Outlet
gr/dscf Inlet
Outlet
Fractional collection effi-
ciency0 In stated size range
Aerodynamic particle size range, micrometers
D<0.5
49
10
49
10
120
0.346
0.0527
0.0002
99.7
0.510 urn
-
20
34
49.0
1.18
0.0215
0.0005
97.6
Total
100
100
100
100
245
3.46
0.1076
0.0015
98.6
              Weight percents are taken from plots of average distributions.

              Partlculate concentration = total emissions x weight % in stated size range; concentrations given in terms
              of milligrams per dry standard cubic meter at 20°C  and 101 kPa  and  grains per dry standard cubic foot at
              68°F and 29.92 in. Hg.  Total  concentrations were taken from Method 5 test results.

              Collection efficiency = Inlet  concentration - outlet concentration   x 100
                                               inlet concentration

-------
to the left toward smaller sizes if the actual density were



greater than 1 g/cm , and vice versa.  A quick approximation of



the physical diameter can be obtained by dividing the reported



aerodynamic diameter by the square root of the actual particle



density.  For example, the specific gravity of the fabric filter



dust samples was about 3.3 g/cm .  Using this particle density



would increase the amount of controlled emissions smaller than



1 ym from roughly 20 to 30 cumulative weight percent.



     As expected, results of the Mark III blank run at the outlet



indicated that stack gases did not react with the glass fiber



filter media to create false weight changes.



     Results representing average uncontrolled emissions over an



entire heat cycle generally agreed with expected distributions



based on previous EAF data.  Particulate concentrations indicated



by the first two distribution tests compared favorably with the



results of particulate tests conducted over similar time frames.



Results of the third distribution test did not agree with the



third particulate test, but this lack of agreement was believed



to be representative of the short-term variation in emissions.



The isokinetic sampling rates for the three runs (104, 100,



and 112 percent, respectively)  were considered acceptable.  The



Heavy Grain Loading Impactor sampling rates were all within the



limits suggested by the manufacturer, and results are believed to



be within generally expected limits of accuracy.



     The particle size tests performed during various times of



process operation show that uncontrolled emissions are not




                               3-30

-------
consistent over the short term.  Although these results gen-


erally can be related to process operations,  one should remember


that the actual sampling times do not precisely coincide with


only one specific process mode.  For example, Run 1 was conducted


during an EAF charge, but the time required to obtain an ade-


quate sample was longer than the charging operation.  As a


result, Run 1 overlapped into the melting phase of EAF operation.


Also, the mode of ADD operation (and relative emission genera-


tion) during Run 3 was not clear.   Evaluation of the appropriate


heat sheet indicated that the AOD may not have been generating


any emissions during the actual sampling period.


     Comparison of particulate concentrations among the Mark III


runs showed disagreement with the relative observation that


emissions during EAF meltdown were greater than emissions during


an EAF charge.  For this reason, and because of the dilution


effects when only one furnace was operating,  particulate concen-


trations were not reported.  The average isokinetic sampling


rates (107, 110, and 106 percent)  were all considered acceptable,


and impactor sampling rates were within suggested operating


limits.  In addition, the distributions are not considered to be


biased toward larger sizes by the use of glass fiber collection

                        4
media.  This possibility  was discounted, as the filter material


did not seem to increase the collection efficiencies of upper


impactor stages because most of the captured material was either


in the cyclone precutter or on the lower stages.  Results,
                              3-31

-------
therefore, are taken to be representative of the variation in


uncontrolled emissions.


     Outlet particle size distributions showed a higher number of


large particles than expected.  It was expected that 80 to 90


cumulative weight percent of emissions would be of particles with


aerodynamic diameters of approximately 2.5 pm or less.  One of


the several plausible explanations for the apparent discrepancy


is the possible bias caused by increased efficiency of the upper

                                                  4
impactor stages when glass fiber filters are used.   Examination


of analytical results indicated that this may have occurred,


but it could not be verified.  The adjustment to outlet results


for this type of bias would reduce the aerodynamic diameter from


5 to 3.5 ym at a cumulative weight of 50 percent; however, this


adjustment would not completely account for the difference


between actual and expected results.


     A second explanation could be related to inaccuracies of the


method at very low sample weights; however, results are con-


sidered to be acceptable, based on several observations.  The


particulate concentrations indicated by the particle size runs


were between 65 and 85 percent of results from simultaneous par-


ticulate tests, which is good agreement for the two different


methods.  The increase in emissions indicated by the third par-


ticulate test is verified by a corresponding increase in particle


size emission results.  The close agreement among the three


cumulative distribution curves is further evidence that particle


size results are representative.  If the stage sample weights had



                              3-32

-------
been too low for accurate determinations,  the three curves



probably would have shown more variability and flatter slopes.



All of these observations suggest that results were within



expected limits of accuracy for particle size distributions



tests.



     The most likely explanation for the higher than expected



number of large particles is related to electrostatic charge.



Small charged particles could have agglomerated to form particles



of larger diameter.  Evidence of a high electrostatic charge



inside the fabric filter was discovered while the open gratings



at the bottom level of the filter were being covered.  For this



reason, results of particle size tests are considered to be



representative of actual conditions during the tests and reflect



possible particle agglomeration due to electrostatic charges.



     The fractional efficiencies reported in Table 3-7 also could



be affected by the possible particle agglomeration just described.



Particle agglomeration would explain the relatively low control



efficiencies indicated for the larger particle sizes compared



with the abnormally higher efficiencies for smaller sizes.  Large



particles formed after filtration would indicate false filtering



efficiencies for the different size classifications; however, the



overall efficiency would be unaffected.





3.3  VISIBLE AND FUGITIVE EMISSIONS



     Visible emissions from the melt shop and fabric filter



monovent were evaluated simultaneously with particulate matter
                               3-33

-------
tests.  In accordance with Method 9* procedures, emissions were


observed in 6-minute sets, and individual opacity readings were


recorded at 15-second intervals.  Fugitive emissions from the


fabric filter were evaluated periodically throughout the test


series according to procedures outlined in the proposed Method


22.**  Fugitive emissions were recorded as the cumulative minutes


of any emissions visually detectable during 20-minute observation


periods.


3.3.1  Results


     Table 3-8 summarizes the visible emissions detected during


charging, tapping, and other furnace operations.  The melt shop


was divided into a north and a south segment for evaluation


purposes.  Visible emissions from the north segment were attrib-


uted to the north furnaces, and vice versa.  When one of the


north furnaces was charging, only the emissions from the north


segment of the melt shop were considered.  Tabulated emission


data for each furnace type and process mode include the total


number of data sets for both melt shop segments; for example, the


five data sets during Run 1 for EAF charging represented three


sets from the north furnaces and two from the south.  During Run


1, individual 6-minute set averages ranged from 0 to 4 percent


opacity for charging and tapping operations.  Average opacities


of individual sets ranged from 0 to 13 percent for refining and


other operations.  During Run 2, individual set averages during
  40 CFR 60, Appendix A, July 1, 1980.
**
  Federal Register, Vol. 45, No. 224, November 18, 1980.



                               3-34

-------
           TABLE 3-8.   SUMMARY OF VISIBLE AND FUGITIVE  EMISSIONS0

                               Melt shopb
Date
(1981)
4/7







4/8





Total




Run
No.
1







2










Furnace
type
EAF

ADD

EAF/AOD

EAF/AOD

EAF

ADD

EAF/AOD

EAF/AOD




Process
mode
Charge
Tap
Charge
Tap
Tap/Charge
Charge/Tap
Refining
and other
Charge
Tap
Charge
Tap
Refining
and other
Charging
and
tapping
Refining
and other
Number of
sets
5
5
4
2
1
1
60

8
5
4
3
74

38


134

Range of
readings,
% opacity
0
0
0-5
0
0-10
0
0-25

0-20
0-25
0-20
0-10
0-25

0-25


0-25

Range of
set averages,
% opacity
0
0
0-3
0
4
0
0-13

0-12
0-15
0-8
0-5
0-22

0-15


0-22

                         Fabric filter outlet
Number of
of sets
73
Range of readings,
% opacity
0
Range of set
averages, % opacity
0
                 Fugitive emissions from fabric filter
Accumulated observation
period, minutes
595
Minutes
0
Accumulated emission time,
% of observation period
0
Data were collected during 7 hours of process operation on April 7, 7
hours on April 8, and 20 minutes on April  9.  Unfavorable weather conditions
prevented additional readings on April  9.

Each set of readings represents emissions  from either the north or south
segment of the melt shop.  Data for Run No. 3 could not be obtained because
of unfavorable weather conditions on April 9.
                                   3-35

-------
charging and tapping ranged from 0 to 15 percent opacity.



Average opacities of individual sets ranged from 0 to 22 percent



during refining and other operations.  No data were obtained



during Run 3 because of adverse weather conditions.



     Table 3-9 lists the time and average opacity for each 6-



minute set of visible emissions data obtained at the melt shop.



The heat sheets in Appendix D were used to determine the times of



various process operations for comparison to emissions.  None of



the emissions seemed to be caused by abnormal operations.



     No visible emissions from the fabric filter monovent exhaust



were detected at any time during the test series, even after



compartment cleaning cycles.  A total of 73 six-minute sets of



data were collected.



     The fabric filter structure was observed for a total of 595



minutes.  No fugitive emissions were detected at any time.



3.3.2  Discussion



     The higher periods of melt shop emissions could be related



to daily activities that normally occur during the first part of



the day shift.  No significant emissions were detected after 1130



hours on April 7 or 1230 hours on April 8, 1981.  Examination of



the available process information did not yield support for this



hypothesis, however.



     The low opacity data for the fabric filter outlet were



supported by the low particulate concentration results.
                               3-36

-------
  TABLE 3-9.   COMPARISON OF MELT  SHOP  VISIBLE EMISSIONS TO PROCESS OPERATION
Process mode3
EAF


N-T

N-C







NC, STi
S-CTL
S-T2B
K
s-cb



N-T

N-AE
N-C



N-RT-
N-RT
N-RT

L.
ADD


N-C
N-C








N-Tb

N-C






S-Cb
N-T






S-Tb


NC,STl&.SBB
S-T2b


j-i,


S-C



Run number, date,
set
Run No. 1
10:15 -
10:27 -
10:33 -
10:43 -
10:49 -
10:55 -
11:01 -
11:07 -
11:13 -
11:25 -
11:37 -
11:49 -
12:01 -
12:13 -
12:25 -
12:37 -
12:49 -
1:01 -
1:13 -
1:25 -
1:37 -
1:49 -
2:01 -
2:13 -
2:25 -
2:37 -
2:49 -
3:01 -
3:13 -
3:25 -
3:37 -
3:49 -
4:01 -
4:13 -
4:25 -
4:37 -
4:49 -
5:01 -
5:13 -
time
, 4/7/81
10:20 a.m.
10:32
10:38
10:48
10:54
11:00
11:06
11:12
11:18
11:30
11:42
11:54
12:06 p.m.
12:18
12:30
12:42
12:54
1:06
1:18
1:30
1:42
1:54
2:06
2:18
2:30
2:42
2:54
3:06
3:18
3:30
3:42
3:54
4:06
4:18
4:30
4:42
4:54
5:06
5:18
Average opacity,
%
North

0
4
3
0
0
13
9
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
South

0
4
3
12
11
13
0
2
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
(continued)
3-37

-------
 TABLE  3-9  (continued)
Process modea
EAF


N-C
N-C


N-AE.S-TI
N-AE.S-BB
S-BB
S-BB
S-T2





N-RWP
N-RWP,S-C
N-RWP, S-C
N-C









N-T

N-C


S-Ti
S-BB
S-BB
S-T2


s-cb
S-AE
ADD

N-Cb










S-C

N-T















N-C
S-T




,i-,

S-C



Run number, date,
set
Run No. 2
9:45 -
9:51 -
9:57 -
10:03 -
10:09 -
10:15 -
10:21 -
10:27 -
10:33 -
10:39 -
10:45 -
10:51 -
10:57 -
11:09 -
11:15 -
11:20 -
11:27 -
11:33 -
11:39 -
11:45 -
11:51 -
11:57 -
12:03 -
12:09 -
12:19 -
12:25 -
12:37 -
12:49 -
1:01 -
1:13 -
1:25 -
1:37 -
1:49 -
2:01 -
2:13 -
2:25 -
2:37 -
2:49 -
3:01 -
3:15 -
3:27 -
3:39 -
time
, 4/8/81
9:50 a.m.
9:56
10:02
10:08
10:14
10:20
10:26
10:32
10:38
10:44
10:50
10:56
11:02
11:14
11:20
11:26
11:32
11:38
11:44
11:50
11:56
12:02 P.m.
12:08
12:14
12:24
12:30
12:42
12:54
1:06
1:18
1:30
1:42
1:54
2:06
2:18
2:30
2:42
2:54
3:06
3:20
3:32
3:44
Average opacity,
%
North

8
10
5
12
22
15
5
8
10
6
2
1
2
5
4
6
10
12
8
0
13
20
12
8
12
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
South

0
10
5
12
22
15
5
8
10
6
2
0
0
0
0
0
10
12
8
8
13
20
12
8
12
2
0
0
0
0
0
0
0
0
0
0
0
0
0
2
0
0
(continued)
                                    3-30

-------
TABLE 3-9  (continued)
Process mode3

EAF

N-Cb




ADD

S-PS
S-PS




Run number, date,
set time
Run No. 2, 4/8/81
3:51 - 3:56 p.m.
4:03 - 4:08
4:15 - 4:20
4:27 - 4:32
4:39 - 4:44
Average opacity,
%
North

0
0
0
0
0
South

0
0
0
0
0
 N = North furnace,  S = South furnace,  C  =  Charge, T = Tap, CTL = Clean tap
 ladle, AE = Add or  adjust electrode, RT  =  Repair tap spout, BB = Burn bottom,
 RWP = Repair water  pipe,  PS = Patch seam.
^Process mode actually began 3 to  6 minutes prior to indicated set time.
                                      3-39

-------
3.4  FABRIC FILTER DUST SAMPLES



     Samples of dust collected by the fabric filter were obtained



daily from the dust-handling system just below the central junc-



tion of the screw conveyors.  Samples were collected in a manner



that did not interfere with other ongoing tests.  The laboratory



split each sample into two fractions for separate analyses of



trace elements by spark source mass spectroscopy (SSMS) and for



particle size distribution by Coulter Counter.



3.4.1  Trace Elements



     Table 3-10 summarizes the results of SSMS analyses on the



three dust samples.  Concentrations are given in micrograms of



element per gram of sample.  Less than (<) and greater than (>)



marks are used to denote concentrations outside the quantifica-



tion limits for the particular element and sample analysis.  The



minimum detection limits for the majority of elements ranged from



0.1 to 0.4 yg/g; major constituents are listed as >1000 yg/g.



Results for several elements are not reported, and indium was



added to each sample as an internal standard.  Elements are



listed alphabetically for convenience.  The analytical results



included in Appendix C are listed in order of decreasing atomic



number.



3.4.2  Particle Size Distribution



     Figure 3-4 shows the best-fit cumulative distribution curve



for the three dust samples.  This curve represents the weight



percent of particulate matter smaller than the indicated physical



particle diameter (in micrometers).  Each data point was





                               3-40

-------
              TABLE  3-10.  SUMMARY OF TRACE ELEMENT ANALYSES ON
                         FABRIC  FILTER DUST SAMPLES
Element
Aluminum
Antimony
Arsenic
Barium
Beryllium
Bismuth
Boron
Bromine
Cadmiun
Calcium
Carbon
Cerium
Cesium
Chlorine
Chromium
Cobalt
Copper
Dysprosium
Erbium
Europium
Fluorine
Gadolinium
Gallium
Germanium
Gold
Hafnium
Hoi mi urn
Hydrogen
Indium
Iodine
Iridium
Iron
Lanthanum
Lead
Lithium
Lutetium
Magnesium
Manganese
Mercury
Molybdenum
Neodymium
Concentration, yg/g (ppm weight)
Sample 1
>1000
160
420
>1000
0.3
350
42
170
150
>1000
NRa
21
7
>1000
>1000
340
>1000
<0.4
<0.4
<0.4
>1000
<0.4
660
100
O.4
3
<0.4
NR
STDb
2
<0.4
>IOOO
21
>IOOO
160
<0.4
>1000
>1000
NR
>1000
5
Sample 2
>1000
190
410
>1000
0.1
420
23
94
180
>1000
NR
19
14
>1000
>1000
190
>1000
<0.1
<0.1
0.4
>1000
<0.1
810
55
<0.1
1
<0.1
NR
STD
3
•*0.1
>1000
25
>IOOO
9
<0.1
>1000
>1000
NR
>1000
3
Sample 3
>1000
44
120
>1000
0.2
44
53
43
132
>1000
NR
6
3
>1000
>IOOO
430
>1000
<0.2
<0.2
<0.2
>1000
<0.2
330
41
*0.2
1
«0.2
NR
STD
3
•P0.2
>1000
6
>1000
79
1000
>1000
NR
>1000
1
(continued)
                                     3-41

-------
TABLE 3-10 (continued)
El ement
Nickel
Niobium
Nitrogen
Osmium
Oyxgen
Palladium
Phosphorus
Platinum
Potassium
Praseodymium
Rhenium
Rhodium
Rubidium
Ruthenium
Samarium
Scandium
Selenium
Silicon
Silver
Sodium
Strontium
Sulfur
Tantalum
Tellurium
Terbium
Thallium
Thorium
Thulium
Tin
Titanium
Tungsten
Uranium
Vanadium
Ytterbium
Yttrium
Zinc
Zirconium
Concentration, yg/g (ppm weight)
Sample 1
>1000
190
NR
<0.4
NR
<0.4
>1000
<0.4
>1000
2
<0.4
<0.4
170
<0.4
4
1
120
>1000
<4
>1000
>1000
>1000
3
3
<0.4
12
8
<0.4
410
>1000
100
6
850
<0.4
15
>1000
130
Sample 2
>1000
56
NR
<0.1
NR
<0.1
840
<0.1
>1000
1
<0.1
<0.1
530
<0.1
2
0.1
66
>1000
66
>1000
>1000
>1000
<0.9
5
<0.1
<0.1
5
<0.1
220
710
54
3
r >1000
<0.1
8
>1000
39
Sample 3
>1000
52
NR
<0.2
NR
<0.2
970
<0.2
>1000
1
<0.2
<0.2
200
<0.2
2
0.5
28
>.1000
220
>1000
800
>1000
1
2
*0.2
2
2
<0.2
82
660
50
2
>1000
<0.2
16
>1000
36
 Not reported.
""internal  standard.
                                     3-42

-------
                                                                                                        t-*i
U)
             «.i
                                                    PARTICLE SIZE,  micrometers
                              Figure 3-4.  Average particle size  distribution  of
                                              fabric filter dust samples.

-------
plotted manually from differential distribution data reported by



the laboratory.  The average curve indicated that 50 percent by



weight of collected dust consisted of particles with physical



diameters of 2.2 ym or less.  Ninety-three percent by weight had



diameters of less than 10 ym.



3.4.3  Discussion



     The concentrations of several trace elements seem to vary



considerably, and this could be related to different specifica-



tions of the metal in the furnaces.  It should be noted, however,



that SSMS is more of a qualitative than quantitative analytical



technique.   [The results of an audit sample (Section 5)  bear this



out; they indicate that reported element concentrations are only



accurate within a factor of +3.]



     When evaluating the particle size results, one should note



that the cumulative distribution curves are based on physical



diameters rather than aerodynamic diameters as reported for



emission tests.  If desired, an approximation of the aerodynamic



diameters can be made by multiplying the reported physical



diameters by the square root of the actual particle density.



Using the specific gravity analysis results of 3.3 g/cm  would



decrease the amount of dust smaller than 3 ym from approximately



70 to 35 cumulative weight percent.



     The reported cumulative weight distributions also assume



that all particles have the same density.   This assumption was



necessary to convert particle volume data measured by the Coulter



Counter to a weight basis.




                              3-44

-------
     The Coulter Counter results were compared with a theoretical

size distribution based on emission test results at the inlet and

outlet sites.  This theoretical cumulative weight curve was esti-

mated from the average size distributions, fractional efficien-

cies, and mass loadings listed in Table 3-7.   The Coulter Counter

and theoretical curves both indicated that 75 cumulative weight

percent of the collected dust consisted of particles with aero-

dynamic diameters of approximately 6 jam or less.  The curves

differed considerably at smaller sizes, but this may have re-

sulted from the agglomeration of particles in the outlet gas

stream.

     The particle size distribution samples were originally sub-

jected to Banco analysis, but agglomeration of the particles

during analysis prevented an accurate determination.


3.5  SUPPLEMENTAL ANALYSES FOR FLUORIDE, CHROMIUM, LEAD, AND
     NICKEL

     Several outlet samples and fabric filter dust samples were

analyzed for particulate fluoride content by procedures described

in EPA Method 13B*, and for chromium, lead, and nickel content by

Atomic Absorption Spectrophotometry.  These analyses were per-

formed subsequent to the completion of originally scheduled lab-

oratory work to better quantify emission levels indicated as

greater than 1000 yg/g by the SSMS analyses on the fabric filter

dust samples.
 40 CFR 60, Appendix A, July 1, 1980.
                              3-45

-------
     Separate fluoride analyses were performed on two acetone



rinses and one set of filters from the outlet particle size



samples, two of the fabric filter dust samples, and appropriate



blanks.  Metal analyses were performed on two sets of outlet



particle size samples (acetone rinse and filters combined), two



dust samples, and appropriate blanks.  The fourth outlet sample



was obtained from a duplicate test run.



     Laboratory results of fluoride and metal analyses on the



outlet samples were reported as milligrams of pollutant.  The



species concentration (in micrograms per gram) was calculated by



dividing the mass of pollutant by the mass of particulate matter



reported in earlier gravimetric analyses.  The laboratory results



for dust samples were reported in concentrations of milligrams



per gram, which were easily converted to micrograms per gram.



The two concentration results of each species and sample type



were averaged.  Average concentrations were multiplied by average



filterable particulate emission results to determine pollutant



gas stream concentrations, mass emission rates, and emission



factors.  For this purpose, results of dust sample analyses were



assumed to be representative of uncontrolled emissions.  Results



are summarized in Table 3-11.



     The outlet fluoride results are based on acetone rinse



analyses only because the filter analysis had a high blank value



of fluoride.  This was caused by the filter material, which was



glass fiber instead of paper as specified by Method 13B.  The



total filter blank value was 2.3 times larger than the net




                              3-46

-------
                                TABLE 3-11.  SUMMARY OF SUPPLEMENTAL ANALYSES FOR
                                     FLUORIDE, CHROMIUM, LEAD, AND NICKEL
                                              Uncontrolled emissions0
Pollutant
species
Fluoride
Chromium
Lead
Nickel
Concentration _
ug/g of solid
47,200
39,200
8,400
16,300
mg/dNnr3
12
9.6
2.1
4.0
gr/dscf
0.0051
0.0042
0.0009
0.0018
Emission rate
kg/h
11
8.9
1.9
3.7
Ib/h
24
20
4.2
8.1
Emission factors
kg/h/Mg
0.098
0.081
0.017
0.034
Ib/h/ton
0.20
0.16
0.035
0.068
kg/Mg
0.37
0.31
0.066
0.13
Ib/ton
0.74
0.61
0.13
0.25
                                                Controlled emissionsb
Fluoride0
Chromium
Lead
Nickel
31,600
17,400
5,800
7,600
0.11
0.060
0.020
0.026
0.00005
0.00003
0.000009
0.00001
0.10
0.056
0.019
0.024
0.22
0.12
0.041
0.054
0.0009
0.0005
0.0002
0.0002
0.0019
0.0010
0.0003
0.0004
0.0035
0.0019
0.0006
0.0008
0.0070
0.0038
0.0013
0.0017
aBased on average uncontrolled particulate emissions and average of analyses  on  two dust samples,  assuming
 that the concentration in the uncontrolled gas stream is the  same as  in  the  collected  dust.
 Based on average controlled particulate emissions and an average of analyses on two outlet samples.
C8ased on analyses of acetone rinses only; the glass fiber filter analysis  had a high blank weiaht
 of fluoride and was not used.

-------
fluoride on the filter, which increased the possibility of error



in the results.  Because the amount of fluoride in the acetone



rinse was much smaller than that indicated by the filter analy-



sis, a small error in the filter results would have a significant



impact on total fluoride.  For these reasons, and because the



filter result indicated a much higher concentration of fluoride



than did the acetone rinse results, the filter analysis was



disregarded.  If the filter result is correct, the outlet



fluoride concentration would be 56,700 yg/g instead of the 31,600



yg/g indicated in the table.  This would not compare favorably



with the 47,200 yg/g of fluoride measured in the dust samples



because it would contradict the trend of the other results,



which indicate lower pollutant concentrations in the outlet



samples than in the dust samples.
                              3-48

-------
                            SECTION 4



                SAMPLING SITES AND TEST METHODS






     This section describes the sampling sites and outlines



the various test methods that were used to characterize par-



ticulate matter emissions, particle size distributions, visible



and fugitive emissions, and fabric filter dust samples.  The



schematics of the air pollution control system presented in



Figures 4-1 and 4-2 identify the relative locations of each



sampling site.  Figure 4-3 presents several photographs of the



control system configuration and sampling sites.






4.1  SITE 1—UNCONTROLLED NORTH EAF AND NORTH AOD



     Uncontrolled emissions from the north furnaces were sampled



for particulate matter and particle size distribution.  Site 1



was located in the 3.0-m  (10-ft) diameter duct between the point



where the north EAF and AOD ducts meet and the junction of north



and south ducts.  Two sampling ports, 90 degrees apart, were



located 2.9 diameters downstream and 1.3 diameters upstream of



45-degree bends, as shown in Figure 4-4.  Forty-four traverse



points were used to sample the cross-sectional area of the duct



for particulate matter, with 22 points on each traverse diameter.




Each particulate run covered two consecutive EAF heats.  Tests




were started at the beginning of charging operations and





                               4-1

-------
                                                           NORTH ADD
                   SITE
                  NO.  1
                      EXISTING CATWALK
                   SITE
                   NO. 2
           FABRIC FILTER

           (8 COMPARTMENTS)
                                                            SOUTH ADD
Figure 4-1.   Control system  schematic,  top view.
                        4-2 .

-------
                                                                                     CURED nor
 I
U)
            Figure  4-2.   Control  system schematic and location of  sampling sites, elevation view.

-------
NORTH AOD (LEFT) AND
NORTH EAF HOOD DUCTS
                                         SOUTH EAF (LEFT)  AND SOUTH
                                               AOD HOOD DUCTS
            SAMPLING LOCATIONS AT THE NORTH EAF/AOD DUCT
            (LEFT, SITE NO.  1) AND THE SOUTH EAF/AOD DUCT
            (SITE NO.  2)

   Fiqure  4-3.  Control  system configuration and  location of  sampling  sites.
                                 4-4

-------
01

         FABRIC  FILTER AND MONOVENT.   SAMPLING
         LOCATIONS ARE INSIDE  EACH  COMPARTMENT
         NEAR THE BASE OF THE  MONOVENT.
FANS AND COMBINED INLET DUCT
             Figure 4-3.  Control  system configuration and  location of  sampling  sites  (continued)

-------
TRAVERSE
POINT NO.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
DISTANCES
cm
15.0
22.4
30.0
38.1
46.7
55.9
66.3
77.7
91.4
107.6
131.1
196.6
220.5
236.2
249.9
261.4
271.8
280.9
289.6
297.9
305.6
312.7
(in.)
( 5.9)
( 8.8)
( 11.8)
( 15.0)
( 18.4)
( 22.0)
( 26.1)
( 30.6)
( 36.0)
( 42.3)
( 51.6)
( 77.4)
( 86.8)
( 93.0)
( 98.4)
(102.9)
(107.0)
(110.6)
(114.0)
(117.3)
(120.3)
(123.1)
                                          CROSS SECTION
                                                      PARTICULATE
                                                  -TRAVERSE DIAMETERS
                                                                          10.2 cm (4 1n.)
                                                                               PORT
                                                    304.8 cm (120 in.)
                                                        STACK I.D.
                         NORTH
                          PORT0
                     PARTICLE,-.
                     SIZE PORT0
                         SOUTH
                          PORT°
             4 m  (13 ft)
      GAS  FLOW
     FROM  NORTH
    —	
     AOD AND EAF
SITE  NO.  1
45 deg.  BEND
NEAR JUNCTION
TOP VIEW  OF
INLET  SITES
             8.8 m (29 ft)
   45 deg. BEND AT
    CREST OF ROOF
             4m (13 ft)
               8.8 m  (29 ft)
            -*-!
(
\
(
}

NORTH
PORT0
EXTRA o
PORT w
SOUTH -
PORT °
U-»-J

GAS FLOW
^ FROM SOUTH
AOD AND EAF

f
)
SITE NO. 2
(
)
                         61 cm (2 ft)
                           OFF AXIS
                  Figure 4-4.   Inlet  sampling  locations.
                                       4-6

-------
continued through tapping.  Initial sampling of each traverse




point lasted 4 minutes.  At the completion of a full traverse,



the nozzle was positioned at a point of average velocity and



sampling continued until the end of the first heat.  A new



traverse was then initiated for the second heat, and each point



was sampled for 4 minutes.  By the end of the second EAF heat, a



minimum of two complete traverses had been conducted.  Actual



sampling time, which depended on heat times, ranged from 352 to



399 minutes.



     Particle size distribution samples were collected at a point



of average velocity near the centroid of the duct.  A separate;



port was used to minimize interferences with the particulate



matter tests.  Sampling times ranged from 17.5 minutes for the



Andersen Mark III samples to 207 minutes for the Andersen Heavy



Grain Loading Impactor samples, which covered an integral heat.






4.2  SITE 2—UNCONTROLLED SOUTH EAF AND SOUTH ADD



     Site 2 was similar to Site 1, but only particulate matter



tests were conducted at this site.  Forty-four traverse points



were sampled for 8 minutes each, to yield a total sampling time



of 352 minutes per run.  These tests were coordinated with tests



at Site 1, not with process conditions of the south furnaces.






4.3  SITE 3--FABRIC FILTER OUTLET



     Controlled emissions from the fabric filter serving all four



furnaces were sampled for particulate matter and particle size



distribution.  The cleaned gases from each compartment exit






                               4-.7

-------
through the common monovent located atop the fabric filter in a


configuration typical of positive-pressure fabric filters.


Method 1* criteria could not be met at this site; therefore, the


throat of the monovent was chosen as an optimum sampling location


because it represented the smallest cross-sectional area.  This


small area would not only provide the highest gas velocities,


but also offer less chance for bias due to faulty bags.


     Figure 4-5, a top view of the fabric filter arrangement,


shows the general location of sampling points.  Figure 4-6, an


end view of the fabric filter, shows the location of the sampling


plane used in each compartment with respect to the site configu-


ration.  Figure 4-7 gives specific dimensions for the sampling


plane cross-sectional area and the location of sampling points


in a typical compartment.  It should be noted that the dimensions


of the sampling plane were slightly different from those of the


actual monovent throat because the tests were conducted just


below the throat.  Other dimensions are given for the compartment


widths and panel offsets created by structural I beams.  As the


figure indicates, dimensions for the sampling cross section and


the exhaust opening in the two end compartments were different


from those in the six middle compartments.  Based on these


measurements, the total cross-sectional area of the sampling

                                                            2
plane in all eight compartments was calculated to be 113.2 m


(1218.8 ft2).
*
 40 CFR 60, Appendix A, July 1, 1980.
                               4-8

-------
 I
vo






_ _J


8










7

-35 4 m (116 ft)

TNI FT Dl CMIIM

1 1
1 1
| MONOVENTv
1 i V«
1 \
1 	 1 	 ' 	 * — 1
	 j 	 1 	 !
OUTLINE OF EXHAUST ARE
AT TOP OF COMPARTMENTS
1 i
I— 4- — I J
1 ^
1 1
| '
i i
6 j 5 | 4
1 !
COVERED WALKWAY






\/^
L ^^ _
1. -«. -


3







SAMPLE
POINTS ~
'

2







V 	
4 1
-*
3 2


CMPT.
NO. 1
"J










k


"^-*
^~-
A

SEE SECTION K-K
IN FIGURE 4.3-2


A
^_J
                                                   TOP VIEW OF FABRIC FILTER
1  ACCESS DOOR PER
   COMPARTMENT
                                                 RUN 1 - COMPARTMENT NOS. 1 THROUGH 4
                                                 RUN 2 - COMPARTMENT NOS..3 THROUGH 6
                                                 RUN 3 - COMPARTMENT NOS. 5 THROUGH 8
                                Figure 4-5.   Sampling Site No.  3, the fabric  filter outlet.

-------
            SEE SECTION B-B
            IN FIGURE 4.3-3
WALKWAY
 WITH
ACCESS
 DOOR
      METHOD 5
  PROBE AND HEATED
     FILTER BOX
   SUSPENDED  FROM
CENTER OF RIDGE BEAM
SAMPLE
 LINE
       ^/IMPINGtRS AND METERING EQUIPMENT
                                                    RAISED
                                                    CENTER
                                                    WALKWAY
                                                                           MONOVENT
                                                                               GAS FLOW
                                       GRATING  LEVEL ABOVE BAGS
                                      	13.6 m (44.7  ft)	
        H = 3.4 m (11 ft)
        W = WIDTH OF EXHAUST OPENING  AT SAMPLING PLANE  =  3.51 m (11.5 ft)
        ^ = 87.8^cm (2.88ft)
                           SECTION A-A, FABRIC  FILTER END VIEW FACING SOUTH
         Figure 4-6.   Sampling  location  at  Site  No.  3, the fabric  filter  outlet.

-------
                               EDGE OF EXHAUST OPENING AT SAMPLING PLANE
                                                                                          MONOVENT PANEL
                                                                                            SEPARATES
                                                                                             EXHAUSTS
                                                                                             BETWEEN
                                                                                           COMPARTMENTS
  3.51 m
(11.5 ft)
                                                       4 SAMPLING POINTS
                                                        PER COMPARTMENT
                                 END PANELS SEPARATING COMPARTMENTS
                                	  4.42 m (14.5 ft)  	
              •DIMENSIONS  IN BRACKET [ ] ARE FUR THE TWO END COMPARTMENTS. IF DIFFERENT.

               NOTE:   BAG  ROWS ARE IN-OPEN AREAS BETWEEN CATWALKS.

                               TOP VIEW OF FABRIC FILTER EXHAUST AREA
                               WITH MONOVENT NOT SHOWN.  ONLY ONE OF
                               EIGHT COMPARTMENTS IS SHOWN.


            Figure  4-7.  Location  of  sampling points  at Site  No.  3, the
                                        fabric  filter outlet.

-------
     Each particulate test consisted of sampling four points in

each of four compartments (for a total of 16 points); at 20

minutes per point, this yielded 320 minutes of sampling time.

The tests began at the same time as the inlet tests and concluded

at the end of the 16-point traverse.  Figure 4-5 shows the

specific compartments tested during each run.  At the end of

three runs, each compartment had been tested at least once.

     Each particle size distribution sample was collected at one

sampling point in each of four compartments, which yielded a

total sampling time of 300 minutes.  These samples were collected

simultaneously with particulate matter tests and in the same

compartments; however, both probes were not in the same compart-

ment at the same time.

     During each test, the entry of dilution air through the open

grating at the bottom level of the fabric filter was minimized by

covering the gratings with kraft paper and boards.  The gratings

in all four compartments to be tested on a given day were covered

during the preceding day to allow fabric filter conditions to
                                                       (
equilibrate.


4.4  VELOCITY AND GAS TEMPERATURE

     A type S pitot tube and an inclined draft gauge manometer

were used to measure the gas velocity pressures at the two in-

lets.  Velocity pressures were measured at each sampling point

across the duct to determine an average value.  Measurements

were taken in accordance with procedures outlined in Method 2 of
                              4-12

-------
the Federal Register.*  Velocities at the outlet site were



calculated from inlet flow rates and the size of the outlet area.



The temperature at each sampling point was measured by using a



thermocouple and potentiometer.






4.5  MOLECULAR WEIGHT



     Flue gas composition was determined in accordance with



procedures described in Method 3.*  An integrated bag sample was



collected at each site during the preliminary runs on Monday, and



an Orsat Gas Analyzer was used to analyze the bag contents for



oxygen and carbon dioxide.  Since these results verified that the



gas streams were essentially air, additional samples were not



collected.






4.6  PARTICULATE MATTER



     Method 5* was used to measure particulate concentrations at



the two inlet sites.  All tests were conducted isokinetically by



traversing the cross-sectional area of the stack and regulating



the sample flow rate relative to the gas velocity in the duct as



measured by the pitot tube and thermocouple attached to the



sample probe.  Each sampling train consisted of a heated, 316



stainless steel-lined probe, a heated 87-mm (3-in.) diameter



glass fiber filter  (Gelman Type AE), a Teflon sample line, and a



series of Greenburg-Smith impingers followed by an umbilical line



and metering equipment.  At the end of each test, the nozzle,




probe, and filter holder portions of the sample train were
 40 CFR 60, Appendix A, July 1, 1980.
                               4-13

-------
acetone-rinsed.  The acetone rinse and filter media were dried at



room temperature, desiccated to a constant weight, and weighed on



an analytical balance.  Total filterable particulate matter was



determined by adding the net weights of the two sample fractions.



Any condensate in the sample line was drained into the impinger



section of the sampling train.  After the amount of water col-



lected in the impingers was measured, the contents were recovered



and gravimetrically analyzed for condensible matter by evaporating



the solutions in an oven at 105°C.



     Method 5 equipment and modified sampling procedures were



used for particulate tests at the fabric filter outlet.  The  :



sampling train was similar to those used at the inlet sites



except for the lack of a pitot tube and the use of a glass-lined



probe.  Tests were conducted at a constant sampling rate based on



the estimated average velocity of the entire sampling area.  This



average velocity was calculated by first converting the total



flow rate measured at the two inlet sites to outlet conditions of



temperature, pressure, and moisture, and then dividing by the



total outlet sampling area.  The resultant average velocity was



assumed to represent each sampling point and was used to calcu-



late an average isokinetic sampling rate.  The heated probe and



filter assembly was suspended from the center of the ridge beam



in a fabric filter compartment, as shown in Figure 4-6.  The



nozzle was positioned at each of the four sampling points in a



compartment by rotating the probe and filter assembly.  When the



sampling points were changed, care was taken to avoid stirring up
                              4-14

-------
any dust or bumping the probe against any structural members.

Compartment cleaning cycles were sampled as they occurred.  Com-

partment gas flows were interrupted while test equipment was
moved from one compartment to another, but conditions were

allowed to equilibrate for several minutes before sampling

resumed.  Outlet samples were recovered and analyzed in a manner
similar to inlet particulate samples, except that the impinger

solutions were analyzed for organic and inorganic matter by
ether-chloroform extraction.
     Sampling times and volumes for the outlet particulate tests

exceeded the respective minimum requirements of 4 hours and 4.5
dNm  (160 dscf)  specified in Subpart AA of the Federal Register.*

4.7  PARTICLE SIZE DISTRIBUTION
     Particle size samples at the inlet site were collected with
an Andersen Mark III Cascade Impactor and an Andersen Heavy Grain
Loading Impactor (HGLI).  The Mark III is an in-stack, multistage

cascade impactor that yields a total of eight particle cut-sis:es
ranging, nominally, from 0.5 to 15 ym.  Substrates for this
impactor were 64-mm diameter glass fiber filters.  The Mark III
was used to collect samples over time intervals of approximately
20 minutes.  The HGLI is an in-stack multistage impactor designed

specifically to allow longer sampling times at high grain load-

ings.  The three nominal cut-points are 2, 5, and 10 ym.  The

only filter in the HGLI is a glass fiber thimble used as the

backup stage.  This impactor was used to collect samples over an
 40 CFR 60, Subpart AA, July 1, 1980.
                               4-15

-------
entire EAF heat, which was approximately 3.5 hours.  A cyclone



precutter was attached to the front of each type of impactor to



remove larger particles and to avoid the use of buttonhook



nozzles.  Because the sampling rate could not be adjusted to



obtain the 15-ym cut-point of the cyclone precutter, the weight



of particulate collected by the cyclone was added to the weight



in the first stage of the respective impactor.



     All inlet samples were collected at a point of average



velocity near the centroid of the duct.  The isokinetic sampling



rate was based on initial measurements of velocity pressure and



temperature.  Constant cut-point characteristics were maintained



during sampling, but velocity pressures and temperatures were



measured periodically at the sampling point to evaluate the



actual variation in isokinetics.  Nozzles were selected to keep



sampling rates in the recommended range of 8.5 to 21 liters per



minute (0.3 to 0.75 acfm).   Each filter was recovered, desic-



cated, and weighed on an analytical balance.  Acetone rinses of



appropriate stages were evaporated, desiccated, and weighed.



     Particle size samples at the outlet were collected by using



a Mark III impactor fitted with a straight nozzle.  The impactor



and probe were suspended in a fabric filter compartment from the



center of the ridge beam,  in a manner similar to that used on



particulate matter testing equipment.  Each sample was collected



for an equal amount of time at one point in four different com-



partments.  The initial isokinetic sampling rate was based on the



calculated average velocity of the entire sampling area.
                                4-16

-------
Constant cut-point characteristics were maintained throughout
each test, and gas temperatures were measured with a thermocouple
attached to the impactor probe.  Each filter was recovered,
desiccated, and weighed on an analytical balance.  The inlet
chamber and nozzle were brushed and rinsed with acetone and the
rinse was evaporated, desiccated, and weighed.

4.8  VISIBLE AND FUGITIVE EMISSIONS
     Certified observers recorded visible emissions from the melt
shop and fabric filter monovent according to procedures described
in EPA Method 9.*  Data were taken in 6-minute sets (simultane-
ously with particulate tests),  and individual readings were re-
corded in percent opacity at 15-second intervals.  Intermittent
rest periods were taken to prevent eye fatigue; however, as long
as emissions were visually detectable, readings were continued
until a break was absolutely necessary.  The emission points were
casually monitored during break periods, and readings were
resumed if emissions greater than zero opacity were noticed.
     Fugitive emissions from the fabric filter dust-handling
system were observed according to the proposed Method 22.**
Emissions were recorded as the cumulative amount of time that
any fugitive emissions were visually detectable during a 20-
minute observation period.  Several observation periods were
recorded during the test series.
     Observers were positioned on the side of a hill, approxi-
mately 75 meters (250 feet) southwest of the baghouse.
  40 CFR 60, Appendix A, July 1, 1980.
  Federal Register, Vol. 45, No. 224, November 14, 1980.
                              4-17

-------
Adverse weather conditions prevented visual emission observations



during the third test.





4.9  FABRIC FILTER DUST SAMPLES



     Samples from the dust-handling system were obtained just



below the central junction of the screw conveyors that connect



individual hoppers.  Because the hoppers were emptied only once



a day, a single grab sample was taken on each test day.  For each



sample, approximately 1 liter of dust was collected in a glass



jar that had been rinsed with dilute nitric acid in the labora-



tory.  Upon return to the laboratory, each sample was split into



two fractions;  one for trace element analysis, and one for



particle size distribution analysis.



     Spark Source Mass Spectroscopy was the analytical technique



used for qualitative examination of the presence of approximately



70 elements.  A known concentration on indium was added to each



sample prior to ionization.  All elements were ionized with



approximately equal sensitivity.  A photographic plate was used



to record the mass spectra.  The plate was examined and the



response of each element was related to that of indium.  Relative



sensitivity factors based on previous analyses of standards were



used to compensate for the variation in response of the photo-



plate for different elements.



     The Coulter Counter technique was used to determine particle



size distributions after problems associated with particle



agglomeration prevented the initial attempts by Bacho analysis.



For the Coulter analysis, particles in each sample were suspended




                              4-18

-------
in a sodium chloride electrolytic solution.  Electrical current



passed from one immersed electrode through a small aperture to



another electrode.  As a particle passed through the aperture, it



displaced a volume of electrolyte and changed the electrical



current by an amount proportional to the size of the particle.



The volume and number of particles were used to establish a



differential distribution by volume.  Assuming all particles were



of equal density, the volume distribution also represented a



weight distribution.
                               4-19

-------
                            SECTION 5
                        QUALITY ASSURANCE

     Quality assurance is one of the main facets of stack sam-
pling because the end product of testing is to produce repre-
sentative emission results.  Quality assurance guidelines provide
detailed procedures and actions necessary for defining and pro-
ducing acceptable data.  Four documents were used in this test
program to provide the required guidance to help ensure the
collection of acceptable data and determine when data quality is
unacceptable.  These documents are the source-specific test plan
prepared by PEDCo and reviewed by the Emissions Measurement
Branch; the EPA Quality Assurance Handbook Volume III, EPA-600/4-
77-027b; the draft PEDCo Environmental Emission Test Quality
Assurance Plan; and the PEDCo Environmental Laboratory Quality
Assurance Plan.  The last two quality assurance plans are PEDCo's
general guideline manuals, which define the standard operating
procedures followed by the company's emission testing and the
laboratory groups.
     Appendix F provides more detail on the Quality Assurance
procedures, including QA objective; data reduction; quality
control checks; performance and system audits; preventive main-
tenance; precision, accuracy, and completeness; corrective
action; and quality assurance reports to management..
                               5-1

-------
     Relative to this specific test program, the following are

the steps that were taken to ensure that quality data were ob-

tained by the testing and analytical procedures.

     0    Calibration of field sampling equipment.  (Calibration
          guidelines are described in more detail in Appendix E.)

     0    Train configuration and calculation checks.

     0    Onsite quality assurance checks, such as sample train,
          pitot tube, and Orsat line leak checks.

     0    Use of designated analytical equipment and sampling
          reagents.

     Table 5-1 lists the sampling equipment used to conduct

particulate and particle sizing tests, along with calibration

guidelines and limits.  In addition to the pre- and post-test

calibration, a field audit was performed on the dry gas meters by

the use of critical orifices calibrated and supplied by the EPA.

The audit results in Table 5-2 show that all dry gas meters used

for this test series were within limitations stipulated in EPA

Method 5.  Dry gas meter performance test procedures and field

audit sheets are shown in Figures 5-1 through 5-6.

     Between runs, onsite preliminary calculation checks were

performed to verify isokinetic sampling rates and to compare

moisture contents, flow rates, and other parameters with expected

values.  These checks indicated that the tests were being con-

ducted properly.

     As a check of the reliability of the method used to analyze

the particulate matter and particle size filters, sets of blank

filters that had been preweighed in the laboratory were resub-

mitted for replicate analysis.  Table 5-3 summarizes the results

                              5-2

-------
                                               TABLE  5-1.   FIELD  EQUIPMENT  CALIBRATION
tn
u>
Equipment
Meter box
Meter box
Meter box
Meter box
Meter box
Meter box
P1tot tube
PI tot tube
P1tot tube
1.0.
No.
FB-3
FB-5
FB-2
FB-7
FB-6
FB-8
251
252
253
Calibrated
against
Wet test meter





Standard pi tot tube


Allowable
deviation
AY prea + 0.020
AH@ + 0.15
AY pt>stbrjL 0.05





A Cp + 0.01


Actual
deviation
-0.003
-0.09
+0.012
-0.008
-0.05
+0.019
-0.007
-0.08
-0.02
-0.009
-0.06
+0.009
+0.001
-0.06
+0.002
-0.010
+0.06
+0.018
0.003
0.003
0.001
Within
allowable
limits
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
Comments
Y pre = 1.008
7 post = 1.020
T pre = 1.056
7 post - 1.075
7 pre = 1.007
7 post = 0.987
7 pre = 1.004
7 post = 1.013
7 pre = 0.973
7 post = 0.975
7 pre = 0.982
7 post = 0.999
Cp = 0.81
Cp = 0.81
Cp = 0.80
(continued)
                    "Allowable deviation AY  pretest = +0.02 7 pretest.
                    Allowable deviation AY  post-test = +0.05 7 pretest.

-------
                  TABLE  5-1 (continued)
Ul
 I
Equipment
Thermocouple
Thermocouple
Thermocouple
Thermocouple
Thermocouple
Digital
Indicator
Or sat
analyzer
Trip balance
Barometer
1.0.
No.
149
138
129
128
254
219
222
208
207
142
198
227
Calibrated
against
ASTM reference
thermometer




Millivolt signals
Standard gas
Type S weights
NBS-traceable
barometer
Allowable
deviation
+ 1.5*




0.5*
+0.5*
+ 0.5 g
0.20 1n. Hg
post-test
Actual
deviation
-0.68
-0.54
+0.69
-0.35
-0.61
Avg. 0.1*
Avg. 0.16*
Avg. -0.10*
2.5°F
-0.2*
0.0 g
0.00 1n.
Hg
Within
allowable
limits
/
/
/
/
/
/
/
/
/
/
/
/
Comments





Actual deviation 1s
an average of eight
temperature points;
No. 207 tested and
calibrated by manu-
facturer
CO 1s highest
deviation


                (continued)

-------
TABLE 5-1 (continued)
Equipment
Dry gas
thermometer




Probe nozzle






I.D.
No.
FB-2
FB-3
FB-5
FB-6
KB-7
FB-8
3-111
2-101
A2PS-P
A1PS-2
A1PM-2B
A1PM-1B
A1PM-2A
A3PS-P
A3PS-1B
Calibrated
against
Reference thermom-
eter type ASTM 2F
or 3F




Call per






Allowable
deviation
+5°F




On + 0.004 in.






Actual
deviation
1 1.5°F
0 1.7°F
I 1.2°F
0 1.9°F
I 2.7°F
0 1.6°F
1 0.6°F
0 1.6°F
I 3.5°F
0 1.4°F
I 1.0°F
0 2.3°F
0.001
0.002
0.001
0.001
0.003
0.003
0.002
0.001
o.obi
Within
allowable
limits
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
Comments
1 = inlet thermom-
eter
0 = outlet thermom-
eter






Nozzles for particle-
size tests were
labeled according to
run numbers





-------
TABLE 5-2.   DRY GAS METER AUDIT RESULTS
Meter box No.
FB-2
FB-3
FB-5
FB-6
FB-7
FB-8
Calibrated against
Critical orifice No. 2
Critical orifice No. 2
Critical orifice No. 2
Critical orifice No. 1
Critical orifice No. 1
Critical orifice No. 1
Deviation, %
+ 3.3
+ 2.0
+ 0.4
- 0.3
+ 2.7
- 0.2
                 5-6

-------
                                                            M.'DIT REPORT GAVPt.R MPTEr!
                    Dace
                    Baromcteric  pressure ( Phal.» In Hg )  3O
                                           bar'
                    Crifice number
                    Trlfice K  factor
                                        6",
                                                                              Client
                                                                                       0 5
                                                          Meter box number   f~&

                                                          Pretest Y	/.OO~f
                                                          Auditor .£>
Ul
 I
-J
Crlticc
manometer
reading
AH
In H20
Z.4
ury gas
meter
reading
'$.
O(e?g>.ooo
O83..000
rr>' eas
meter
volume
V
r?
/if .000
Temperatures
Ambient
T /T
ai af
°F
5-y
^
Avcracc
a
°F
^V
Inlet
Tmi
°F
^
^
Outlet
nio
°F
sy
55-
Average
T
m
°F
St. -2-Z
Samp linf;
time
0
min
/&.V
V
"std
ft3
/4.i*t>
V
m
act
ft3
/v-*/
Percent
error
Jt-Z.Z'/c
      = (  17.647 )( V  )( Y )(  P.   -I All/13.6 )/(
                                                                         A60 )
                     "act = (  1203  )( 0 )( K )( Pbar )/( Ta + 46° >'
                                                             z.V
                    Vmstd=(  17.647 )(  /V,*~>)
                     "act"(  12°3
error = ( V     - V
          mstd    "act
                                              100
                                                        act
                                                                         100
                                                  Figure 5-1.   Audit  report  sample meter  box.

-------
                                                            A'.!DIT REPORT 3AKPI.F. METEI!
                    Date
                              V- 7 -
                     Barometerlc pressure (  P.   ,  in Hg )


                     Orifice number	2
                    rrifice K factor
                                                         Client    U^>	


                                                         Meter box number  /*jfl ^3


                                                         Pretest Y
                                                         Aud i to r £> ^
(Jl

 I

00
Oriticc
manomcier
reading
AH
in H20
**«•
ury gas
meter
reading
ft3
JV7^
&t.~c
I'ry gas
meter
volume
V
t?3
•/-f.ooo
Temperatures
Ambient
T /T
ai af
°F
£4
<^
Average
T
a
°F
**
Inlet
°Fl
fl
W
Outlet
T
nio
°F
^6
^
Average
m
°F
**
Sampling
time
0
min
Z
V
-std
ft3
/^
V
mact
ft3
,735,
Percent
error
^
-------
                                                       AUDIT REPCIFIT
                                                                          ^:^:T!^r; r/\x
               Date
                Barometerlc pressure (  Pfe f.  In Hg ) ^O • 3


                Orifice number   -2.
               Crlflce K factor
                                                                         Client
                                                          Meter box number



                                                          Pretest Y	^

                                                                   s~\  /

                                                          Auditor
                                                                                    - 7-
                                                                                     ft^-awes.
Ul

 I

VO
entice
manoraeter
reading
AH
In H20
z>i*
ury gas
meter
reading
'£•
TS'S-Ot
-770.000
rry gas
meter
volume
V
m
ft3

//2 . t?067
Temperatures
Ambient
T /T
al af
°F
&>
bo
Average
a
°F
t?o
Inlet
Tn,l
°F
^6
V
Outlet
T
mo
°F
fa£~
k


-------
                                       CM MTNOO S BUT CAS M1EH MM
                                                              :( TtSI MU SMIt
(Jl
I
tat. Y/7/J7
iaraaatric
NKar In N
Pratatt V
Pollutant Coda
Prattora. fc^,. 3O • Vfc -MQ 77J> Survav Nuabar
•. Pfe A^Oae>
"735.1 Oi





Cat voluaa
dry gat
•atar
W..

Art) lent
•c
60 F /^.<

Teaperitu
0
Inlet
•"'
9y
8«V




ra
•y git aattr
Outlet'
lao
•F
'7 »"
Ty


'Average
la
•F
'---
VI « Initial raadlng af dry gat aatar.
Vf • final rtadtng af dry gat aatar.
to • ¥f - »l « iraluaa af gat patting through dry gat aatar
»irs
f T£ • tUfl)
Tha «a1«a af 0.4M7 It abtainad fn» ^ '
•feara: •.«•? It tka camrartton farior fro
Tttd • *»••
0.02B117
• ft1 te •'



Saapling
tiaa
0. Bin.
,7^a-


Vacwai
tatting
In. Hg
20



Came tad
gat voluoa
«Mttd)
• *^oVO
~~) C 3 e<


^ OCA-t^o-va^
SlgMtura
                         Figure  5-4.   EPA Method 5 dry gas  meter performance test  data sheet.

-------
            tat.*
t
dry gat
wtar
va>

/ 9-. 5



^ 	 ... _
r • • ' 	 ~
AaDitnt
la
'JU7 ^.
55




Teiyeratu
	 o
Inlat
•"

-------
                      •'•
IM NUHOO 4 Ot» C*S WTEI Hit
                               IfSI Ml* SHUT
cn
 I
(-•
N)
»u V/7/S-,
^ft-f. r ,_. 	
••iMatrlc PrefMir*. tV JO. Vjfc, •«
mutt r
,. Pfr 9
. 7*35-
•
T**t
Ho.
1
f
1
Orlflc*
MMMttr
reading
AH.
In. hjO
5.3


Cat voluM
dry gat
Mttr
V,M
ft1
feblO.^bo
£J3- Voc?




•ollutant Codt
Ig 773 Survt
Parti
v Number
cipantt 10

OH f let Ho. /&>(•*, J-

Cat voluM
dry gat
Mttr
-1
wo lent
la
•c
^^.^
Taaptratu
r.:.-. o
Inltt
•r
~?y
7 •




r«
ry gat Mttr
Outlet"
1*0
rf£
.'r*


•"
,»

Sailing
tiM
0. «tn.
27 /J*0"
'•^*o


VacuM
tatting
in. Hg
/9



C*rr*ct*d,
Mt Ml«M
.Z-^l .

x * is r

             VI • initial reading of dry gat Mttr.
             Vf • final reading of dry gat Mttr.
             Va • Vf - VI * MluM of gat patting through dry gat Mttr

                                       AH
             v._f.g).|t • (*«499/)CMi)(V)(>^kK_ * 13.6)
             ^^	 	__?y

             «• -I- .f •Wu .^Md fro. ^ • ° 0»J»

                   0.0717 It the CMMrtlM (actor fro* ft-1 ta •'
                   T«U'
                   •ttd
                                 Figure 5-6.   EPA  Method 5  dry gas  meter  performance test data  sheet.

-------
                     TABLE 5-3.   FILTER BLANK ANALYSIS


Type of filter
Participate: 87-mm
Gelman A/Ea
Andersen Mark,
III Impactor0







Andersen Mark
III Impactor .
Blank test run0






Andersen Heavy.
Grain Loading
Impactor, (HGLI)
Thimble0


Filter No.

2165
Y94
Z51
Y66
Y75
Y88
Z39
Y76
Z53
B179
X31
X32
X33
X34
X35
X36
X37
W60
B222
39
61


Tare
weight,
mg

362.3
127.8
136.0
126.8
134.9
127.2
136.0
127.4
136.6
180.5
148.9
137.6
145.6
138.6
151.1
138,9
150.2
135.2
188.0
2470.6
2176.6


Blank
weight,
mg

362.6
127.8
136.3
126.8
135.0
127.4
136.3
127.4
136.8
180.6
149.1
137.8
146.0
139.0
151.3
139.2
150.4
135.4
188.7
2480.4
2178.6


Net
weight,
mg

+ 0.3
0.0
+ 0.3
0.0
+ 0.1
+ 0.2
+ 0.3
0.0
+ 0.2
+ 0.1
+ 0.2
+ 0.2
+ 0.4
+ 0.4
+ 0.2
+ 0.3
+ 0.2
+ 0.2
+ 0.7
+ 9.8
+ 2.0




Comments



















c




 Expected deviation,  +0.5 mg.

 Expected deviation,  +0.3 mq.
•s  '                  """'
'The high net weight  is  probably a  result of very small  particles leaking
 around the prefilter.
 Expected deviation,  +0.5 mg.
                                    5-13

-------
of the blank filter analyses.  These results show good data



reproducibility from an analytical standpoint.



     A blank run was performed at the fabric filter outlet to



determine whether stack gases reacted with the filter media to



produce erroneous results.  This was accomplished by placing a



backup filter in front of a normally prepared impactor and then



sampling in the usual manner.  Table 5-3 lists results of the



blank run, which shows that stack gases did not significantly



affect filter media.



     In addition, blanks were taken to check the quality of



reagents used to recover and analyze particulate and particle



size samples.  Table 5-4 summarizes the results of these blank



analyses, which show that most reagents met designated specifica-



tions for quality.



     A trace element audit sample was analyzed, along with the



fabric filter dust samples, to check the accuracy of the SSMS



analytical procedures.  The audit sample was taken from Standard



Reference Material No. 1633, "Trace Elements in Coal Fly Ash,"



which was obtained from the National Bureau of Standards.  The



results (shown in Table 5-5) indicate that the analyses were



within a factor of three of true values, which is the expected



limit of SSMS accuracy.



     Sampling equipment, reagents, and analytical procedures for



this test series followed and met all necessary guidelines set



forth for accurate test results in Volume III of the Quality
                              5-14

-------
                     TABLE 5-4.   REAGENT BLANK ANALYSIS
Type of blank
Participate blanks:
Acetone
Acetone
Water
Particle size blanks:
Acetone
Acetone

Acetone
Analytical blanks:
Ether/chloroform
Water
Container
No.

3982A
3989A
3983A

2526A
2531A

2549A

BT459 (org)
BT459 (aq)
Volume of
blank, ml

237
227
378

382
190

518

150
250
Weight after
evaporation and
desiccation,3
mg/g

+ 0.01
+ 0.007
+ 0.004

+ 0.004
+ 0.02

+ 0.005

+ 0.005
+ 0.002
Comments






0.01 used in
calculations




Tolerance:   +0.01  mg/g.
                                     5-15

-------
                   TABLE 5-5.   TRACE ELEMENT AUDIT  RESULTS
Element
Arsenic
Cadmium
Chromium
Copper
Lead
Manganese
Nickel
Selenium
Uranium
Vanadium
Zinc
Concentration, yg/g
NBS-certifieda
61+6
1.45 + 0.06
131 + 2
128 + 5
70 + 4
493 + 7
98 + 3
9.4 + 0.5
11.6 + 0.2
214 + 8
210 + 20
Measured
120
1
150
71
110
170
54
16
18
99
220
Percent b
difference
+100
- 30
+ 10
- 40
+ 60
- 70
- 40
+ 70
+ 60
- 50
0
SRM No. 1633, "Trace Elements in Coal  Fly Ash".
Percent difference =
                     ^"
                         actual
deviation is +200%, -70% (+ factor of 3).
x 100, to the nearest 10%.   Expected
                                  5-16

-------
Assurance Handbook.5  Therefore, test results reported in this



document should be within the expected accuracies of the method



used.
                               5-17

-------
                            SECTION 6

            STANDARD SAMPLING AND ANALYTICAL PROCEDURES


     This section describes the test methods, sampling equipment,

and analytical techniques that were used in this test program for

determination of particulate matter and particle size distribu-

tion.


6.1  DETERMINATION OF PARTICULATE EMISSIONS

     In this test program, the sampling and analytical procedures

used to determine particulate emissions at Sites 1 and 2 were

those described in Method 5 of the Federal Register.*

6.1.1  Sampling Apparatus

     The particulate sampling train used in these tests met

design specifications established by the EPA.  The sampling

apparatus, which was assembled by PEDCo personnel, consisted of

the following:

     Nozzle - Stainless steel (316) with sharp, tapered leading
     edge and accurately measured round opening.

     Probe - Stainless steel (316) with a heating system capable
     of maintaining a minimum gas temperature of 121°C (250°F) at
     the exit end during sampling.

     Pitot Tube - A type S pitot tube that met all geometric stan-
     dards was attached to a probe to monitor stack gas velocity
     pressure.
 40 CFR 60, Appendix A, July 1, 1980,
                               6-1

-------
     Temperature Gauge - A Chromel/Alumel type-K thermocouple  (or
     equivalent)was attached to the pitot tube, in an inter-
     ference-free arrangement, to monitor stack gas temperature
     within 1.5°C (5°F) using a digital readout.

     Filter Holder - The filter holder was made of Pyrex glass,
     with heating system capable of maintaining a filter tem-
     perature of approximately 121°C (250°F).

     Filter - An 87-mm (3-in.) diameter glass fiber filter
     (Gelman A/E) was used.

     Draft Gauge - The draft was measured with an inclined manom-
     eter (made by Dwyer) with a readability of 0.25 mm  (0.01
     in.) H20 in the 0 to 25 mm (0 to 1 in.) H-O range.

     Impingers - Four Greenburg-Smith design impingers were con-
     nected in series with glass ball joints.  The first, third,
     and fourth impingers were modified by removing the tip and
     extending the tube to within 1.3 cm (0.5 in.) of the bottom
     of the flask.

     Metering System - The metering system consisted of a vacuum
     gauge,  a leak-free pump, thermometers capable of measuring
     temperature to within 1.5°C (5°F), a calibrated dry gas
     meter,  and related equipment, to maintain an isokinetic
     sampling rate and to determine sample volume.  The dry gas
     meter was made by Rockwell, and the fiber vane pump was made
     by Cast.

     Barometer - An aneroid type barometer was used to measure
     atmospheric pressures to 0.3 kPa (+0.1 in.Hg).

6.1.2  Sampling Procedure

     After the sampling site and minimum number of traverse

points were selected, the stack pressure, temperature, moisture,

and range of velocity head were measured according to procedures

described in the Federal Register.*

     Approximately 200 grams of silica gel were weighed and

placed in a sealed impinger prior to each test.  Glass fiber

filters were desiccated for at least 24 hours to a constant
 40 CFR 60, Appendix A, Methods 1, 2, 3, or 4, July 1, 1980


                               6-2

-------
weight and weighed to the nearest 0.1 mg on an analytical balance.

One hundred milliliters of distilled water was placed in each of

the first two impingers; the third impinger was initially empty;

and the impinger containing the silica gel was placed next in

series.  The train was set up as shown in Figure 6-1.  The sam-

pling train was leak-checked at the sampling site prior to each

test run by plugging the inlet to the nozzle and pulling a 50-

kPa (15-in.Hg) vacuum, and at the conclusion of the test by

plugging the inlet to the nozzle and pulling a vacuum equal to

the highest vacuum reached during the test run.

     The pitot tube and lines were leak-checked at the test site

prior to each test run and at the conclusion of each test run.

The check was made by blowing into the impact opening of the

pitot tube until 7.6 cm (3 in.) or more of water were recorded on

the manometer and then capping the impact opening and holding it

for 15 seconds to assure it was leak-free.  The same procedure

was used to leak-check the static pressure side of the pitot

tube,  except suction was used to obtain the 7.6-cm (3-in.) H20

manometer reading.  Crushed ice was placed around the impingers

to keep the temperature of the gases leaving the last impinger at

20°C (68°F) or less.

     During sampling, stack gas and sampling train data were

recorded at each sampling point and whenever significant changes

in stack flow conditions occurred.  Isokinetic sampling rates

were set throughout the sampling period with the aid of a nomo-

graph or calculator.  All sampling data were recorded on the

Emission Testing Field Data Sheet.
                               6-3

-------
       1.9-2.5 cm
      (0.75-1 irO
 1.8 cm (0.75-1 in.)
                       "H
— v
cd — 	 ^^.




"*
                    PROBE
                    PITOT TUBE
                                          THERMOMETER
        NOZZLE^
                  STACK WALL


                     LI   PROBE
 "S" TYPE
   PITOT
   TUBE

THERMOCOUPLE
                HEATED  >
                FILTER^
                  r ----- -n
                       |
THERMOMETER
                                                           100 ml.  OF  WATER
TEMPERATURE
 INDICATOR      THERMOMETERS
                 o
 CALIBRATED
  ORIFICE
                                                          -^.CONTROL
                                                           |   VALVES
                                                          hJU
                     MANOMETER
      VACUUM\LINE


    GAUGE
                Figure 6-1.   Schematic of participate samplinq train.

-------
6.1.3  Sample Recovery Procedure

     The sampling train was carefully moved from the test site to

the cleanup area.  The volume of water from the first three

impingers was measured, and the silica gel from the fourth

impinger was weighed to the nearest 0.1 gram.  Sample fractions

were recovered as follows:

     Container No. 1 - The filter was removed from its holder,
     placed in a petri dish, and sealed.

     Container No. 2 - Loose particulate and acetone washings
     from all sample-exposed surfaces prior to the filter were
     placed in a polyethylene jar, sealed, and labeled.  Par-
     ticulate was removed from the probe with the aid of a brush
     and acetone rinsing.  The liquid level was marked after the
     container was sealed.

     Container No. 3 - A minimum of 200 ml of acetone was taken
     for the blank analysis.  The blank was obtained and treated
     in a similar manner as the acetone washing.

     Container No. 4 - After being measured, distilled water in
     the impinger section of the sampling train was placed in a
     polyethylene container.  The impingers and connecting glass-
     ware were rinsed with distilled H20, and this rinse was
     added to the container for shipment to the laboratory.

     Container No. 5 - A minimum of 200 ml of distilled water was
     taken for the blank analysis.  The blank was obtained and
     treated in a similar manner as the water rinse.

     Container No. 6 - An unused glass fiber filter was taken for
     blank analysis.

     All pertinent data were recorded on the Sample Recovery and

Integrity Data Sheet.

6.1.4  Analytical Procedures

     The analytical procedures used were those described in the

Federal Register.*

     Container No. 1 - The filter and any loose particulate
     matter were desiccated in the petri dish for 24 hours to a
     constant weight and then weighed to the nearest 0.1 mg.
 40 CFR 60, Appendix A, July 1, 1980.

                               6-5

-------
     Container No. 2 - The volume of acetone washings was mea-
     sured and transferred to a tared beaker.  The sample was
     evaporated to dryness at ambient temperature and pressure,
     desiccated for 24 hours to a constant weight, and weighed
     to the nearest 0.1 mg.

     Container No. 3 - The volume of acetone blank was measured
     and transferred to a tared beaker.  The blank was evaporated
     to dryness at ambient temperature and pressure, desiccated
     for 24 hours to a constant weight, and weighed to the near-
     est 0.1 mg.

     Container No. 4 - The volume of the impinger contents and
     distilled water rinse was measured and transferred to a
     tared beaker.  The sample was evaporated to dryness at 105°C,
     desiccated to a constant weight, and weighed to the nearest
     0.1 mg.

     Container No. 5 - The volume of distilled water blank was
     measured and transferred to a tared beaker.  The blank was
     evaporated to dryness at 105°C, desiccated to a constant
     weight, and weighed to the nearest 0.1 mg.

     The term "constant weight" means a difference of no more

than 0.5 mg or 1 percent of total weight less tare weight, which-

ever is greater between two consecutive readings, with no less

than 6 hours of desiccation between weighings.  All analytical

data were recorded on the Analytical Particulate Data Sheet.

Acetone and water blank data were recorded on respective blank

data sheets.

6.1.5  Modifications to Standard Procedures for Site 3 Tests

     Several modifications were made to the standard sampling and

analytical procedures to conduct particulate matter tests at Site

3.  Some sampling equipment and analytical procedures were also

different, but were not method deviations.  These modifications

and differences are as follows:
                              6-6

-------
Sampling Apparatus--

     Probe - The probe was of borosilicate glass with a heating
     system capable of maintaining a minimum gas temperature of
     121°C (250°F)  at the exit end during sampling.

     Pitot Tube - A pitot tube was not used.


Sampling Procedure--

     The sampling site and number of sampling points did not meet

minimum Method 1* criteria, but were selected according to

practical considerations.  The stack pressure, temperature, and

moisture at this site were measured according to standard pro-

cedures, but accurate velocity pressures could not be determined

at each point.

     The sampling train was prepared, assembled, and leak-checked

according to standard procedures, except that a pitot tube was

not used.

     During sampling, stack gas temperature and sampling train

data were recorded at each sampling point.  An average isokinetic

sampling rate was set initially, based on the estimated average

velocity for the entire sampling cross-sectional area.  This

estimated velocity was calculated from data that had been ob-

tained at Sites 1 and 2 and adjusted for the differences in

cross-sectional area, temperature, pressure, and moisture.  The

resultant value was assumed to represent the average velocity at

each Site 3 sampling point.  The average isokinetic sampling rate

was held constant throughout the sampling period.  All sampling
 40 CFR 60, Appendix A, July 1, 1980,
                                6-7

-------
data, including the estimated average velocity pressure, were

recorded on the Emission Testing Field Data Sheet.

     Initial calculations of emission results were based on esti-

mated velocities and later adjusted to reflect actual velocities

measured at Sites 1 and 2 during simultaneous tests.


Sample Recovery Procedure—

     The sample recovery procedures were the same as those for

Sites 1 and 2.


Analytical Procedures—

     Container No. 4 - The volume of distilled water and water
     rinse was measured and transferred to a separatory funnel.
     The sample was extracted three times with diethyl ether,
     and each time the water was drained back into the original
     sample container and the ether into a clean, tared beaker.
     The sample was then extracted three times with chloroform,
     and each time the chloroform was drained into the beaker
     with the ether.  After the final extraction, the water
     portion was drained into a separate tared beaker, evaporated
     to dryness at 105°C, desiccated, and weighed to a constant
     weight to obtain the condensible inorganic content.  The
     ether/chloroform portion was evaporated to dryness at
     ambient temperature, desiccated, and weighed to a constant
     weight to obtain the condensible organic content.

     Container No. 5 - The distilled water blank was treated in
     an identical manner as Container No. 4.  The aqueous frac-
     tion was used as a water blank, and the organic fraction was
     used as an ether/chloroform blank.

     All other procedures for the determination of particulate

emissions were as used in tests at Sites 1 and 2.


6.2  DETERMINATION OF PARTICLE SIZE DISTRIBUTION

     Three different configurations of in-stack cascade impactors

were used to collect samples for particle size distribution
                                6-8

-------
measurements.  The following sampling and analytical procedures

were used.

6.2.1  Sampling Apparatus

     The source sampling train used in these tests met design

specifications established by the EPA.  Assembled by PEDCo per-

sonnel, it consisted of:

     Nozzle - Stainless steel (316) with sharp tapered leading
     edge and accurately measured round opening.

     Temperature Gauge - A Chromel/Alumel type-K thermocouple  (or
     equivalent) was attached to the probe to monitor stack gcis
     (impactor)  temperature to within 1.5°C (5°F) using a digital
     readout.

     Metering System - The metering system consisted of a vacuum
     gauge, a leak-free pump, thermometers capable of measuring
     temperature to within 1.5°C (5°F), a dry gas meter with 2
     percent accuracy, and related equipment to maintain an
     isokinetic sampling rate and to determine sample volume.
     The dry gas meter was made by Rockwell, and the fiber vane
     pump was made by Cast.

     Condenser - The condenser consisted of a moisture-removal
     device capable of maintaining a temperature less than 20°C
     (68°F) and an attached thermometer to monitor temperature.

     Impactor - An Andersen Mark III with eight stages and a
     backup filter was used at Sites 1 and 3.   An Andersen Heavy
     Grain Loading Impactor with three stages and a backup filter
     was used at Site 1.  A cyclone precutter was attached to the
     front of each impactor used at Site 1.

     Barometer - An aneroid type barometer was used to measure
     atmospheric pressures to 0.3 kPa (+0.1 in.Hg).

6.2.2  Sampling Procedure

     The stack pressure, temperature, moisture, and velocity

pressure of the selected sampling site were measured with Method

5 equipment according to procedures described in the Federal
                               6-9

-------
Register.*  One or more points representing average velocity were


selected as sampling points.


     Each type of impactor was assembled appropriately.  Assembly


of the Andersen Mark III  (Mark III) involved alternating the


stage plates, collection media, flat crossbars, and Inconel


spacer rings so as to provide eight cut-sizes.  The collection


substrates were Reeve Angel 934 AH glass fiber filters that had


been heated in a 204°C  (400°F) oven for 1 to 2 hours, desiccated


for 24 hours to a constant weight, and weighed to the nearest 0.1


mg on an analytical balance.


     Assembly of the Andersen Heavy Grain Loading Impactor  (HGLI)


involved inserting a glass fiber thimble in the backup stage and


threading together the various parts of the third-stage cyclone


and first-and-second stage jet-impaction chambers.  The glass


fiber thimble had been desiccated for 24 hours to a constant


weight and weighed to the nearest 0.1 mg on an analytical balance.


     If used, the cyclone precutter was threaded together and


attached to the front of the impactor.


     The sampling train was assembled as shown in Figure 6-2 or


6-3.  It was leak-checked at the sampling site prior to each test


run by plugging the inlet to the impactor (or cyclone precutter,


if used)  and pulling a 50-kPa (15-in.Hg) vacuum.  Once the


desired vacuum was reached, the leakage rate was checked at the


dry gas meter for 1 minute.  If the leak rate was less than 0.6


liter/min (0.02 cfm), the train was considered ready for
*
 40 CFR 60, Appendix A, Methods 2, 3, or 4, July 1,. 1980.

                               6-10

-------
                                     METER BOX
                                  TEMPERATURE
                                   INDICATOR
                                                      CYCLONE
                                                     PRECUTTER
                                                  THERMOCOUPLE
         PROBE TUBE
                                                           NOZZLE
Figure 5-2.   Particle  size distribution sampling train at Site 1.
                             6-11

-------
                PROBE TUBE
                                             METER BOX
                                               TEMPERATURE
                                                INDICATOR
Figure 6-3.   Particle  size distribution sampling train at Site 3.
                             6-12

-------
sampling.  Any excessive leaks were corrected before the train



was used.  The impactor was then placed at the selected sampling



point and allowed to preheat for several minutes before sampling



began.  While the impactor was preheating, the nozzle was capped



or pointed away from the gas flow.  A leak-check was not per-



formed after the test run so as to avoid the possibility of



dislodging the particles on individual stages.



     During sampling, stack gas and sampling train data were



recorded at regular intervals based on the length of the run.



Velocity pressure data at Site 1 were obtained periodically using



separate Method 5 equipment.  Average velocities at Site 3 were



estimated from previous data measured at Sites 1 and 2.  The



isokinetic sampling rate was set initially, and constant cut-



point characteristics were maintained throughout the sampling



period.  Preliminary impactor runs were made at each site to



determine the Mark III sampling times required to allow uniform



loading on the backup filter and to prevent loadings of greater



than 10 mg on any one stage.  All sampling data were recorded on



the Impactor Testing Field Data Sheet.



6.2.3  Sample Recovery Procedure



     After the test was completed, the impactor was removed from



the probe and carefully moved to a designated cleanup area while



still in an upright position.  The impactors were recovered as



follows:
                               6-13

-------
     Mark III;

     Container No. 1 - Particulate in the nozzle and inlet cham-
     ber was removed by brushing and rinsing with acetone into a
     polyethylene container, which was sealed and labeled.

     Containers No. 2 through 10 - Each filter was removed from
     its stage and carefully placed in a petri dish.  Loose
     particulate from the bottom side of the previous stage
     plate,  the Inconel spacer, flat crossbar, and the top side
     of the plate directly under the filter were brushed into the
     same petri dish as the respective filter.  Each petri dish
     was sealed and labeled.

     Container No. 11 - If the cyclone precutter was used, par-
     ticulate from all sample exposed surfaces except the in-
     terior of the cyclone exit tube was brushed and acetone-
     rinsed into a polyethylene container, which was sealed and
     labeled.  Particulate from the interior of the cyclone exit
     tube was added to Container No. 1.

     Heavy Grain Loading Impactor With Cyclone Precutter;

     Containers No. 1 through 5 - Particulate from all sample-
     exposed surfaces after the cut-point of the preceding stage
     and prior to the cut-point of a given stage was brushed and
     rinsed with acetone into a polyethylene container.  After
     the container was sealed and labeled, the liquid level was
     marked.

     Container No. 6 - The glass fiber thimble was carefully
     removed from the holder and placed in a glass jar.  The jar
     was then sealed and labeled.

     All pertinent data were recorded on Sample Recovery and

Integrity Data Sheets.

6.2.4  Analytical Procedures

     Filters - Each glass fiber filter or thimble and any loose
     particulate matter were desiccated in respective sample
     containers for 24 hours to a constant weight and weighed to
     the nearest 0.1 mg on an analytical balance.

     Acetone Rinses - The volume of each acetone washing was
     measured and transferred to a tared beaker.  The sample was
     evaporated to dryness at ambient temperature and pressure,
     desiccated for 24 hours to a constant weight, and weighed to
     the nearest 0.1 mg.
                               6-14

-------
     The term "constant weight" means a difference of no more



than 0.5 mg or 1 percent of total weight less tare weight, which-



ever is greater between two consecutive weighings, with no less



than 6 hours of desiccation between weighings.  All analytical



data were recorded on Andersen Impactor or HGLI Particulate



Analytical Data Sheets.



6.2.5  Blanks



     Several unused glass fiber thimbles and a complete set of



unused Mark III filters were returned to the laboratory in



their respective containers.  Approximately 200 ml of the acetone



used for sample recovery was taken as a blank.  In addition, a



blank test run was conducted with the Mark III impactor to deter-



mine if stack gases had reacted with the filter media to cause



false weight changes.  In the blank run a backup filter was



placed in front of a normally assembled impactor to filter out



all particulate matter so that only the stack gases would contact



the filter media.



     All blanks were recovered and analyzed in the same manner as



the actual samples.  Data were recorded on the respective blank



data sheets.
                                6-15

-------
                           REFERENCES
1.   U.S.  Environmental Protection Agency.   A Review of Standards
     of Performance for Electric Arc Furnaces in the Steel In-
     dustry.  EPA 450/3-70-033,  October 1979.

2.   U.S.  Environmental Protection Agency.   Background Informa-
     tion for Standards of Performance:  Electric Arc Furnaces in
     the Steel Industry.  EPA 450/274017b,  October 1974.

3.   Southern Research Institute.  A Computer-Based Cascade
     Impactor Data Reduction System.  Prepared for U.S. Environ-
     mental Agency under Contract No. 68022131.   March 1978.

4.   University of Florida.  Use and Limitations of In-Stack
     Impactors.  Prepared by the Department of Environmental
     Sciences for U.S. Environmental Protection Agency under
     Grant No. R803692-02.  February 1980.

5.   U.S.  Environmental Protection Agency.   Quality Assurance
     Handbook for Air Pollution Measurement Systems, Volume III.
     EPA-600/4-77-027b, August 1977.
                                R-l

-------