EPA-600/2-77-213
October 1977
Environmental Protection Technology Series
SAMPLING AND ANALYSIS
OF COKE-OVEN DOOR EMISSIONS
Industrial Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental Protec-
tion Agency, have been grouped into nine series. These nine broad categories were
established to facilitate further development and application of environmental tech-
nology. Elimination of traditional grouping was consciously planned to foster technology
transfer and a maximum interface in related fields. The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL PROTECTION TECHNOLOGY
series. This series describes research performed to develop and demonstrate instrumen-
tation, equipment, and methodology to repair or prevent environmental degradation from
point and non-point sources of pollution. This work provides the new or improved tech-
nology required for the control and treatment of pollution sources to meet environmental
quality standards.
REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved for
publication. Approval does not signify that the contents necessarily reflect the views and
policies of the Government, nor does mention of trade names or commercial products
constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Information
Service, Springfield, Virginia 22161.
-------
/AN ! 9 1978
JNiTED 3':'AT£::; t:WVi R O \< >..•'! h NT-•.!.. PROTECTION AGENCY
DATE: January 17, 1978
SUBJECT: Final Report "Sampling and Analysis of Coke-Oven Door
Emissions" EPA-600/2-77-213
From: Robert C. McCrillis /) clty~t.> f
Metallurgical Processes Branch
TO: Distribution
Enclosed for your use is one copy of the subject report. This work was
completed for EPA by Battelle-Columbus Laboratories under EPA contract
68-02-1409, Task Numbers 16 and 34.
The work described consisted of the development of a test method for
collecting emissions from coke-oven doors and then collecting and
analyzing emission samples from a production coke-oven. Analysis
included organic analysis by IR spectroscopy, gas chromatigraphy-mass
spec (GC-MS) , thin film liquid chroma tography (TLC) and high resolution
mass spec (HRMS) and bioassay analysis of bacterial mutagenesis (Ames
test) and mammalian cell cytotoxicity (RAM test) .
The test method did not preclude the deposition of sample on the interior
surfaces of the equipment ahead of the filters and adsorber. This
deposited material was not included in any of the subsequent analyses.
Therefore, the analyzed samples represent an unknown and perhaps atypical
fraction of the total emission.
..If you have any questions or comments please call me at 629-2733.
-------
Errata EPA-600/2-77-213
pg. 7 fourth paragraph, second line should read, ".... positive
mutagenic response to 0.5 ul of the sample."
pg. 36 (a) second full paragraph, second sentence, Table 1
should be Table 2.
(b) third full paragraph, first sentence should read,
".... is the fact that the average particulate
emission rate for Test 2 was 3 times greater than
the emission ...."
-------
EPA-600/2-77-213
October 1977
SAMPLING AND ANALYSIS
OF COKE-OVEN DOOR EMISSIONS
by
R.E. Barrett. W.L. Margard,
J.B. Purdy, and P.E. Strup
Battelle-Columbus Laboratories
505 King Avenue
Columbus, Ohio 43201
Contract No. 68-02-1409
Task No. 34
Program Element No. 1AB604C
EPA Project Officer: Robert C. McCrillis
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, N.C. 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D.C. 20460
-------
FINAL REPORT
on
SAMPLING AND ANALYSIS OF COKE OVEN-DOOR EMISSIONS
to
U.S. ENVIRONMENTAL PROTECTION AGENCY
(Contract 68-02-1409, Task 34)
by
P. E. Strap, J. B. Purdy,
W. L. Margard, and R. E. Barrett
November 4, 1977
BATTELLE
Columbus Laboratories
505 King Avenue
Columbus, Ohio 43201
-------
CONTENTS (Continued)
References 79
Appendices
A. Thermal Analysis of Coke Oven Doors 80
B. Heat Transfer Analysis for Hood 110
C. Field Data and Calibration Sheets 122
D. Reconstructed Gas Chromatograms, Electron Impact
lonization Mass Spectra, and Methane lonization
Mass Spectra of Emissions from Coke Oven Doors. . . . 123
-------
FIGURES
Page
1. Sketch of Test Equipment Arrangement on Top of Oven
No. 41, "A" Battery 10
2. Sampling Arrangement for Particulate and Gas Samples. ... 13
3. Particulate Emission Concentration vs Time Into
Coking Cycle 37
4. Particulate Emission vs Time Into Coking Cycle 38
5. Absorbance Infrared Spectra 45
6. Absorbance Infrared Spectra 46
TABLES
1. Identification of Test No. 2 Samples 18
2. Test No. 2 Flow for Air Input, Hi-Vol Sampler, and
Duct Orifice 31
3. Summary Data, No. 1 Test, Coke Oven Door Leakage 32
4. No. 2 Test, Coke Oven Door Leakage 34
5. Level 1 Analyses, Gravimetric Results 40
6. Level 1 IR Analysis of LC Filter Fractions 41
7. Level 1 IR Analysis of LC Adsorber Fractions 42
8. GC-MS Analyses of Selected Coke Oven Door Sample
Fractions 47
9. Semiquantitation of Selected Coke Oven Door Sample
Fractions 49
VI
-------
ABSTRACT
Emissions generated by leakage from a coke oven door were collected
and sampled during a 16-hour coking cycle. Extensive analyses, including
organic and bioassay analyses, were conducted on selected fractions of the
emission samples.
A sealed-hood was fabricated to fit over the door of a coke oven so
that gases leaking past the door during the coking cycle would be contained and
representative samples could be obtained. Additional criteria for the hood
included not severely altering the normal door leakage and not interfering
with coke oven operation. Initial tests of one hood design suggested modi-
fications which were incorporated into the final design. The final hood was
used for conducting two sampling runs at an operating coke oven.
Analysis of the coke oven samples included:
1. Particulate emissions determination
2. Trace metal analyses
3. Gas analyses
4. Organic analyses by IR spectroscopy, GC-MS, TLC, and
HRMS on entire samples or LC fractions of the samples
5. Bioassay analyses of bacterial mutagensis and
mammalian cell cytotoxicity.
Results of the particulate mass emission determination show that a
considerable variation can exist in emissions of coke ovens from cycle to
cycle. Also, results of the bioassay analyses confirmed that the samples were
mutagenic, as the chemical analyses would lead one to expect.
This report was submitted in fulfillment of Contract No. 68-02-1409
by Battelle-Columbus under the sponsorship of the U.S. Environmental Protection
Agency. This report covers the entire effort and was completed on September 30,
1977.
111
-------
CONTENTS
Abstract iii
Figures vi
Tables vi
1. Introduction 1
2. General Conclusions 4
Conduct of Test 4
Interpretation of Results 6
3. Sampling Procedures 9
Sampling Equipment 9
Conduct of Sampling Runs 14
Condition of Sampling Equipment After Two Tests .... 16
Comments on Sampling Procedure. 17
4. Analytical Procedures 18
Mass Determination 19
Preparation of Adsorbent Column and
Filter Samples 19
Level 1 Type Analysis 20
Gas Chromatography-Mass Spectroscopy
Analysis of Selected Samples 21
Analysis by Thin Layer Chromatography 23
Analysis by High Resolution Mass Spectrometry 24
Trace Metals Analysis 24
Analysis of Flask Samples of Gases 25
Bacterial Mutagenesis Bioassay 26
Mammalian Cell Cytotoxicity 28
5. Results 30
Door Leakage and Flow Data 30
Particulate Mass Emissions 31
Results of Level 1 Analyses 39
Gas Chromatgraphy-Mass Spectroscopy Results 44
Thin Layer Chromatography Results 52
High Resolution Mass Spectrometry Results 52
Trace Metal Results 52
Gaseous Emissions 52
Results of Bacterial Mutagenesis Analysis 63
Results of Mammalian Cell Cytotoxicity Studies 68
Discussion of Results 77
Recommendations 73
IV
-------
TABLES (Continued)
10. TLC Quantitation Results for BaP 53
11. Low Voltage Mass Spectrometric Results Obtained
on Fiber Samples 54
12. Low Voltage Mass Spectrometric Results Obtained
Adsorber Samples 55
13. High Resolution Mass Spectrometric Results 56
14. SSMS Analysis of Filter Samples 57
15. Mass Spectrographic Analyses of Coal & Coke'^) 59
16. Oven No. 41 "A" Gas Phase, Test No. 1 . . . 60
17. Oven No. 41 "A" Gas Phase, Test No. 2 61
18. Gaseous Emission Estimates 64
19. EPA Coke-Oven Samples(5) 65
20. EPA Coke-Oven Samples 69
Vl'l
-------
SECTION 1
INTRODUCTION
In connection with a program to develop and evaluate concepts
for minimizing emissions from coke-oven doors (EPA Contract No. 68-02-
1439), Battelle's Columbus Laboratories (BCL) initiated development of a
system for measuring these emissions. The basic equipment consists of a
metal hood that extends over the buckstays adjacent to an oven door, thus
capturing emissions from around the door and permitting these emissions
to be channeled to an exit duct where they can be conducted to measuring
devices.
The initial hood design was tested on a Koppers coke oven at
Empire-Detroit Steel in Portsmouth, Ohio (see Appendix A, Figure 1).
Although the system was judged to operate satisfactorily, it was believed
that use of the sampling hood caused the oven door temperature to rise
above normal, due to the hood eliminating natural convection currents
that normally flow over the door and provide cooling.
To determine the amount of temperature rise above normal caused
by the hood, an oven door at Portsmouth was instrumented with eight
thermocouples and door temperatures were then measured throughout a com-
plete coking cycle of approximately 15 hours (see Appendix A, Figures
2-8). A subsequent test with the sampling hood in place indicated a
temperature rise above normal of about 28 C (50 F) during a 2-hour test.
A heat transfer analysis made of the system showed that the temperature
rise could be minimized by painting both sides of the hood black and
installing vertical fins on the outside of the hood to enhance heat
transfer from the hood to the ambient air.
Following this development work on the sampling hood, EPA
authorized emission measurement tests to quantify particulate and gaseous
emissions due to coke-oven-door leakage and to establish operating pro-
-------
cedures and system reliability under real-life conditions. When these
measurements were authorized, coke production at Portsmouth had been cur-
tailed to the extent that samples could not be obtained from a represen-
tative coking cycle.
After delays while waiting for resumption of normal coke
production at Portsmouth, it was decided to seek another test site.
Republic Steel Corporation gave permission for emission measurement runs
to be conducted at their Poland Avenue coke plant in Youngstown, Ohio.
Two sampling runs were conducted at coke oven No. 41, Battery A, at this
Republic plant during the week of March 29 through April 2, 1976.
During the test period, the following samples were collected:
1. Particulate samples; the samples were collected
on a filter by using a Hi-Vol sampler.
2. Adsorbent column samples of volatile species; these
were collected on a Tenax adsorbent column.
3. Flask samples for gas analysis
Extensive analyses were conducted on the samples collected from
this test, as follows:
Sample
Filter Samples
Extract of filter and adsorbent
column samples
Gas Samples
Analyses
Mass, by drying and weighing
Trace metals by spark-source
mass spectroscopy
Level 1, by infrared spectro-
scopy
Compounds by gas chromatography-
mass spectroscopy
BaP by thin-layer chromatography
(EPA)
Compounds by high resolution-
mass spectroscopy (ERDA)
Bacterial mutagenesis (Stanford
Research Institute)
Mamalian cell cytotoxicity
(Northrup Services)
Compounds by mass spec and C!C
-------
Analytical efforts were conducted by Battelle-Columbus, unless otherwise
indicated.
This report describes the preparation leading up to the field
test, conduct of the field test, and presents results of the analyses.
Where possible, conclusions regarding the procedures used for sampling
and analysis, and regarding the results of this test, are presented.
-------
SECTION 2
GENERAL CONCLUSIONS
CONDUCT OF TEST
Sampling
Generally, the use of a finned hood over the door of a coke oven to
contain door leakage gases and provide for sampling of these gases worked
successfully in that samples were collected without damage to the coke
oven or severe distruption to plant activities.
Improvements that should be considered for future efforts on sampling
emissions from coke oven doors include:
1. Use of a larger blower so that more air could be fed into the
hood. This would more accurately represent the normal air
flow with a breeze, etc., and would provide better
cooling of the oven door.
2. Redesign of the hood to permit insulating the outside
surfaces to reduce condensation of volatile organics
on the inside of the hood while providing adequate
door cooling by increased air flow.
3. Insulation of ducts carrying gases withdrawn from the hood
to reduce condensation .
4. Obtaining a continuous printout or, at least, frequent read-
ings of air and gas flow rates so that leakage, etc., can be
calculated. Because flow rates may change as filters become
loaded, it is desired to obtain readings at intervals no
greater than about 10 to 15 minutes early in the test. For
the later stages of the test, hourly readings should surfice.
5. For a portion of a run, or for another trial of this hood
system, it might be useful to add a tracer gas to the supply
air so that leakage might be estimated more accurately.
-------
6. To determine emission rates (mass of total particulate or certain
compounds or elements per unit time) , all material coming from
the coke oven should be known. That would require collecting
condensed material off of the inside of the hood and the gas
ducts and attributing this material to the gas flow in some way.
Three problems this introduces are (a) how to clean the hood
and ducts safely when the residue being washed off may contain
carcinogenic species, (b) how to clean the very large hood
surfaces, and (c) how to equate the condensed mass to gas flow
- obviously, more of the condensed mass would be collected early
in the test.
The rather large differences in emission rates for the two runs on the
same coke oven cell suggest that single tests will not be useful in pre-
cisely defining emissions from these sources.
Analysis
Due to the presence of large quantities of organics in the coke oven
samples, and the increasing concern over health hazards related to organic
materials, the possibilities for modifying or extending the analysis pro-
cedure are limited only by the imagination and the budget. Obviously, it
would be desirable to conduct semiquantitation of all LC fractions of each
sample. However, for this single test such an approach could have in-
creased the number of GC-MS analyses to 88, rather than the 12 that were
conducted. Also, for the semiquantitation analyses conducted there are
significant unidentified fractions (e.g., 81 percent for Sample Al, Frac-
tion 5) - of what does the missing material consist? Further, it might be
desirable to conduct the HRMS, the trace metal analyses, and the bioassay
analyses on LC fraction of samples, rather than on the entire samples.
Again, material removed (in some future test) from the inside of the hood
and gas ducts should be analyzed.
To facilitate interpretation of future test results, the quantity of
sample used for all analyses should be precisely defined and blanks should
be provided and run for all analyses.
-------
INTERPRETATION OF RESULTS
Mass emissions from coke oven door leakage vary greatly over the
cycle, ranging from over 1000 mg/min during the first hour to less than Img/
min during the latter portion of the cycle. The fraction of the sample
collected in the filter was about 90 percent for the first hour of Test 2,
but dropped to about 45 percent for the period from about 8 hours to 13
hours into the test.
A major observation is the variation in mass emissions from one run
to another. For example, the average mass emission rate for the first
hour of the test was 410 mg/min, whereas, the mass emission rate during
the same period for Test 2 was 1003 mg/min. These data suggest that the
care taken in sealing the door can have a major effect on emissions, and
that estimates of emission rates (or factors) based on a small number of
runs could be inaccurate.
Chemical Analysis Results
The coke oven samples, as expected, were very complex mixtures of
organic compounds. Even after separating the samples into 8 fractions
by liquid chromatography (LC), the fractions were still complex mixtures.
Because of the complex nature of the samples and the fact that several of
the specifies identifications were of a qualitative nature, it is impossible
to make exact comparisons of results achieved by the various chemical
analyses. However, the LC fractions, the GC-MS analyses, and the HRMS
analyses produced results that generally were consistent.
Significant emissions were found of several organic compounds that
have been identified as carcinogenic, as follows:
benzo pyrenes
dibenzo anthracene
benzo phenanthrene
benzo anthracenes
benzo fluoranthenes
indeno pyrene.
The presence of these compounds is adequate to account for the positive bio-
assay results.
-------
Bioassay Results
Complex mixtures such as the coke-oven samples submitted for bio-
logical analysis contain large numbers of materials which may be mutagenic,
cytotoxic, or both. The chemical analysis of the samples indicated that
mutagenicity was to be expected. The samples contained numerous poly-
cyclic hydrocarbons which have been confirmed to be both mutagens and
carcinogens. Benzo fluoranthene and benzo-a-pyrene are but examples.
Mixtures are always difficult to assay. Sometimes a sample may give
a positive mutagenic response at one concentration but the mutation rate
may fall off dramatically at the next higher concentration due to the toxic
effect of the same material, or a second material effecting the mutagenesis
of the first or the viability of the mutants. A large number of synergistic
or antagonistic interactions may take place which can exaggerate or diminish
net bioassy responses.
The fact that Ames assay found mutagens in all of the samples serves
to confirm the chemical analyses. The samples were found to be moderately
mutagenic at the very dilute concentrations analyzed. Thus the data
really suggests that the particulates entrapped on the filters were heavily
laden with mutagens when the total sample is taken into account.
In general, both the filter and the adsorbent column sample extracts
showed a moderate but positive mutagenic response 0.5 pi of the sample.
The 0.5 yl sample represents only 0.0007 percent of the extract of the
filter which was submitted for bioassay analyses. The extract sample was
obtained from one-eighth of the total filter area. When more of the sample
extract was analyzed the mutagenic response was increased until the concentra-
tion was sufficient to elicit a toxic response. It thus becomes quickly
apparent that the total sample contained thousands of times larger amounts
of mutagenic substances.
It has been shown by Commoner-'- that the mutagenicity of a complex
mixture of environmental pollutants can be the sum of the mutagenicity of
the individual fractions. Such a possibility may account for the quantity
of mutagenic substances in the filter samples. In this investigation it
-------
was ascertained that the quantity of benzo-a-pyrene (BaP) in the filter
extract representing the first hour of the coking cycle was approximately
700 ug. The mutagenic detection level for BaP is approximately 1 pg per
plate with the Salmonella tester strains. Therefore the BaP is present in
the filter extract samples in 100 fold less concentration than the mutagens
detected. Thus it is suggested that those mutagens which were detected
were either present in much greater quantities than BaP, were mutagenic
in significantly smaller quantities, or an additive effect of the total
mutagenic compounds was operative, as Commoner suggests.
The results "of the rabbit alveolar macrophage assay must be considered
on a preliminary basis only. The particulate was unevenly distributed on
the filter medium, due to uneven deposition aggravated by the loss of large
quantities during transit. Thus, reliable representative quantitative
samples could not be obtained. Clean unused filter media was not available
as a control.
The relatively clean border of the filters was used and found to be
nontoxic to the macrophages. The samples cut from the center of the filter
media caused a significant reduction in cellular viability.
-------
SECTION 3
SAMPLING PROCEDURES
SAMPLING EQUIPMENT
A new sampling hood with vertical corrugations acting as cool-
ing fins, and black paint applied to both sides, was constructed for the
tests at Youngstown, Ohio. The new features incorporated were a result
of analytical and field work reported in Appendices A and B. After being
checked out at BCL for satisfactory operation, the hood was taken to
Youngstown and used for the program to measure emissions from a coke-oven
door.
*
Figure 1 shows the sampling hood in place clamped to the
buckstays on either side of the oven from which emissions were to be
sampled. As shown in Figure 1 the hood consisted of 5 full sections and
a single half section, all held together by piano-type hinges. On the
back of each section, placed vertically along the edge, were magnetic
tape strips which aid in maneuvering the hood into place and holding it
to the buckstays. However, the magnetic strips were not capable of hold-
ing the hood firmly in place and preventing gas leakage. Special clamps
attached to lengths of angle iron were used to firmly clamp each side of
each hood section to the buckstay. All edges of the hood and the hinges
were sealed with furnace tape and then further sealed with furnace cement
to prevent leakage of gases out, or of air into, the cavity behind the
hood. Also shown in Figure 1 is the layout of the sampling equipment on
top of the coke-oven battery during the test.
* Regulations at Republic in Youngstown prohibited the taking of any
photographs at the test site. Therefore, Figure 1 is an artist's
sketch showing the general layout of the sampling hood over a coke-
oven door.
-------
Vorioc lor Hi-Vol blower
speed control
Flow to Hi-Vol
Hi-Vol somplers
Orifice pressure drop
Monomclcr lo mcosure pressure
beneoth hood
• Adsorber flow melcr
Blower
Pressure lap
Furnace cement
seal
Compressed
air supply
Hinges covered
wit'i furnace tape
Hood sections,
corrugated and '
pointed black to
dissipate heat
Edges of hood
sealed with
furnace tape
and cement
Orifice
Hood exhaust duct
Metered o.r supply to
bottom of overt door
Neighboring coke oven doors
FIGURE 1. ARTIST'S SKETCH OF TEST EQUIPMENT ARRANGEMENT ON TOP OF
OVEN NO. 41, "A" BATTERY - liEPUtlUC STEEL CORPORATION,
YOUNGSTOWN, OHIO
10
-------
A supply of compressed air was delivered through a 1-inch pipe
to a horizontal air distributor below the bottom of the oven door. The
distributor consisted of a 1-inch diameter length of pipe with drilled
holes that allowed air to blow upward over the door, mix with the door
leakage gases, and pass on to the suction blower. The compressed air
fed to the door originated from an air compressor. Ordinarily, a plant
supply line could be used as the air supply, but at Youngstown it was
necessary to rent a compressor for air supply because the plant compressed
air supply system had excessive water in the lines. The compressed air
supply to the hood was carefully metered by a rotometer.
The principle used to measure the leakage of emissions from
around the coke-oven door was based on measurement of air fed under the
hood and measurement of total gas flow away from the hood. This required
that the blower draw gases away from the door area enclosed by the hood
at the same rate as gases were generated by the air supply plus the door
leakage, thus maintaining a zero gage pressure behind the hood. Positive
pressure could cause leakage of gases from behind the hood and negative
pressure could induce air leakage into the hood enclosure. As shown in
Figure 1, a pressure tap was installed in the top plate seal, near the
top of the buckstays. This pressure tap was connected to an inclined
manometer sensitive to 0.001-inch of water, and efforts were made to keep
the pressure reading on the manometer at zero to assure no leakage into
or out of the door area. During the sampling period, the highest pres-
sure differential recorded on the inclined manometer was 0.56 mm (0.022-
inch) H^O, and most of the time it was kept at zero. There was every
indication that significant leakage did not occur during testing; all
sealing materials were intact at the end of the test.
As a result of having the hood sealed in place and maintaining
a zero pressure differential across the hood wall, all emissions leaking
past the coke-oven door were induced to exit the hood through the duct
identified in Figure 1 as "Hood exhaust duct". As the gases passed
through this duct, part of the flow was pulled off by the Hi-Vol sampler;
this part was measured and recorded at the Hi Vol. The remaining gases
flowed on toward the blower and the flowrate was measured at an orifice
mounted in the line. Thus, the gas flow measured at the Hi Vol plus the
11
-------
flow measured at the orifice gave the total flow being drawn from within
the hood enclosure.
Between the orifice and the hlower there was an adjustable
valve which could be used to restrict the flow of gas going to the
blower. Also between the adjustable valve and the blower, a tee was
installed in the duct; one leg of the tee was an open leg which contained
an adjustable butterfly valve. By opening the butterfly valve, unmetered
air was drawn to the blower, thus reducing the quantity of gases pulled
from the hood to the blower.
The capacity of the blower limited the amount of metered air
that could be fed to the air distributor beneath the hood (even with the
butterfly'valve in the closed position). When the volume of metered air
plus oven gas leakage from the door equaled the capacity of the blower
plus the gas drawn to the Hi-Vol sampler, no more air could be supplied
because the additional air would increase the pressure beneath the hood
and cause leakage from beneath the hood.
It would be desirable that the blower have greater capacity so
that much more air could be fed beneath the hood. Whereas the blower used
3
for this effort had a capacity of about 300 m /hr (180 scfm), a capacity
3
of at least 500 m /hr (300 scfm) would be desirable to provide additional
cooling for the door. Also, the additional air would further dilute
gases going to the Hi-Vol sampler which should cause less plugging of the
filter on the Hi-Vol.
It was anticipated that filters on the Hi-Vol sampler might
have to be changed frequently at the beginning of the coking cycle;
therefore, two Hi-Vol samplers were provided. Quick connect and discon-
nect flexible duct connections were provided so that gas sampling could
be quickly switched from one Hi-Vol to the other. In addition, to pro-
vide better chart records, each Hi-Vol was modified by installing a
special motor giving one complete chart revolution per hour instead of
the normal one revolution per 24 hours.
Figure 2 is a schematic diagram that shows the arrangement of
the Hi-Vol sampler and related sampling equipment. Flow from the hood
passed through the 0.20 x 0.25 m (8 x 10-inch) Hi-Vol filter. Down-
stream of the filter there was a tap which connected to an adsorber
12
-------
Tar film on inside
surfaces after tests
After tests Tygon tubing
dark and discolored
but no heavy deposits
8" x 10" filter
Adsorber
Gas sample
pump
Blower and motor section
Positive pressure area
Purge line from
positive pressure
side of blower
Flow from
hood sample
Transducer tap for
measuring flow
FIGURE 2. SAMPLING ARRANGEMENT FOR PARTICULATE AND GAS SAMPLES
13
-------
column, a dry gas meter or flow meter, and a gas sampling pump in that
order. The adsorber column contained 35/60 mesh Tenax-GC, which has the
capability of removing the heavy molecular weight hydrocarbons from a
gas sample passing over it. After passing through the adsorber column,
the gas passed through a dry gas meter which was used to measure the
quantity of gas sampled by the adsorber column and a pump which main-
tained the sample flow.
A tee was installed in the sample line immediately downstream
of the adsorber. The exit leg of the tee had a stopcock and a tapered
fitting which enabled an evacuated flask sample bottle to be attached
for taking a sample of the gas that has just passed through the adsorber.
As shown in Figure 2, a purge line (consisting of rubber tubing terminat-
ing with a stopcock) was attached to the positive side of the blower-
motor section of the Hi-Vol sampler. Before and during attachment of
the flask sampler, gas from the hood was blown over the connections to
help ensure that air was not collected in the sampling flask.
The hood-exhaust duct, from the top of the hood to the blower,
was not insulated as it was thought that the relatively high flow through
this duct would not result in excessive sample loss via condensation on
duct walls. The line from the tie in the hood-exhaust duct to the Hi-
Vol sampler was insulated. In retrospect, it would probably have been
better to heat and insulate the entire duct between the hood and the
Hi-Vol sampler.
CONDUCT OF SAMPLING RUNS
Two sampling runs were conducted at the Republic Steel Corpor-
ation Coke Oven No. 41, Battery A, during the week of March 29 through
April 2, 1976. The first test was conducted on March 31, beginning at
3:10 a.m. and concluding at 6:45 p.m. The second test was begun on
April 1 at 1:03 p.m. and completed on April 2 at 2:15 a.m. The second
test was about 3 hours shorter than the first due to process demands; the
coke was pushed after a shorter coking time.
-------
No effort was made to select either a well-sealed or poorly-
sealed door on which to conduct the test. Rather, the intent was to
sample a "typical" door. A coke oven near the end of the battery was
selected for the test to minimize the disruption of the test due to
plant equipment moving past the sampling area.
The operation of an individual coke oven at this plant is
approximately as follows:
• push previous charge
• oven remains open for 5 to 15 minute interval
• put door in place
• charge oven with fresh coal (up to 5 minutes)
• seal charging doors
• coking period of 12 to 16 hours
• doors removed for 15 to 30 minute interval prior to push
• push coke (about 2 to 3 minutes)
The hood was put in place and sampling begun as soon as possible after
the door was closed. Thus, sampling was begun during the charging opera-
tion, but did not include the entire charging period. Sampling was
continued until about 15 to 30 minutes before the end of the coking
cycle. Sampling was stopped to allow sufficient time to remove the hood
and sampling equipment before coke pushing was to begin.
Prior to sampling runs, it was empirically decided that the
coking cycle of approximately 16 hours should be divided into time seg-
ments for purposes of taking samples for analysis. The length of these
six sampling segments were assigned as follows:
Segment No. Length of Segment, hours
1 1
2 1
3 2
4 4
5 4
6 4
15
-------
Length of the segments was based on the knowledge that emissions from
coke oven doors decreased as time into the cycle increases. Therefore,
to obtain suitable sample quantities for analytical purposes, longer
sampling times are required as the coking cycle progresses. A flask
sample was planned midway through each of the six segments designated;
thus, flask samples were taken at about 0.5, 1.5, 3, 6, and 10 hours
elapsed time and at some convenient point during the last sampling segment,
Concerning the sampling and deposits on the duct, some deposits
were observed on all of the duct with the heaviest deposits being found
on the duct between the tee and the blower. It was noted that the temp-
erature drop from the hood exit to the Hi-Vol sampler did not exceed
93 C (200 F). The Hi-Vol filter temparature was 82 to 107 C (180 to
225 F) during the sampling period.
CONDITION OF SAMPLING EQUIPMENT AFTER TWO TESTS
At the conclusion of the two Youngstown tests, the five full
sections of hood subject to direct radiant heat from the oven door were
covered with a thin coat of black tarry material giving the appearance of
a coat of black spray paint. The top hood section, about half the height
of other sections and subject to less radiant heat, had a thicker coat of
tar appearing more like a brush coat of black paint.
The metal lid that covered the top of the buckstays had a thick
paint-like coat of tar with a run buildup around the gas exit hole.
The 4-inch flexible duct leading from the buckstay lid to the
metal ductwork contained a progressively heavier coating from beginning
to end, with the exit end having a coating like shiny, heavy black lacquer.
The 2-inch flexible duct leading to the Hi-Vol sampler con-
tained progressively heavier tar deposits with run deposits up to 1/16-
inch thick at the exit end.
The sampler metal cap upstream of the filter had an inside
coating appearing as a heavy black spray paint.
Tygon tubing that carried gas samples from downstream of the
filter to the adsorber column had some discoloration but not the deposit
evidenced in other parts of the system.
16
-------
The metal surface of the sampler downstream from the filter
had some deposit similar to very light spray paint.
The 4-inch metal duct leading to the exhaust fan had heavy
deposits like brushed paint. When the fan was laid on its side after
tests, with an approximately 6-foot-long metal duct projecting verti-
cally upward, tar deposits 1/16-inch thick were formed on the fan parts
from warm tar dripping from the duct walls.
COMMENTS ON SAMPLING PROCEDURE
In general, the sampling hood and associated sampling equipment
performed satisfactorily during the two Youngstown tests. The one equip-
ment weakness at Youngstown was in the exhaust blower that pulled the
gases from behind the sampling hood. A greater blower capacity would
permit a larger flow of air to be passed over the oven door for cooling
purposes, and more dilution of leakage gases would reduce rapid filter
loading. Also, heating and insulating the duct between the hood and the
tee would reduce loss of condensibles.
The intent of the sampling system that was used was to provide
a constant flow at the Hi-Vol sampler and at the POM sampling point.
Fluctuations in gas leakage would be compensated for by adjusting the
butterfly valve controlling bleed air to blower system. An alternative
would have been to sample particulate (and POM) from the duct to the
blower and to have omitted the Hi-Vol sampler. However, to do this, the
sampling rate would have to be adjusted to provide isokinetic sampling
with changes in air flow rate. The latter system would be slightly more
cumbersome to operate, but, theoretically, would provide a more logical
sampling procedure in that sample collection rate would match leakage rate.
17
-------
SECTION 4
ANALYTICAL PROCEDURES
This section of the report describes the procedures used to
analyze the samples obtained during the sampling effort conducted on
the Republic Steel Corporation coke oven. Initial analysis conducted
on the coke-oven samples was a gravimetric determination of the mass
collected on the filters. Subsequently, the extract from the filters
and the adsorbent samples was subjected to a number of detailed
analyses related to determining the environmental hazard of coke-oven
emissions.
Because Test 2 contained 4.1 times the particulate concen-
tration as Test 1 and a more accurate recording of test parameters was
made, Test 2 was chosen as the test run to be analyzed. Table 1
identifies the filter and adsorber column samples for Test No. 2. The
sample numbers shown in Table 1 are used in reporting most of the
analytical results.
TABLE 1. IDENTIFICATION OF TEST NO. 2 SAMPLES
Sampling Time
1303-1411
1414-1506
1506-1708
1708-2105
2105-0215
Compressor
air supply
(Blank)
Filter Nos.
119-131
132-136
137-147
148-155
156-161
BCL Code
A IF
A2F
A3F
A4F
A5F
Adsorber Nos.
2.1, 2.2
2.3, 2.4
2.5, 2.6
2.7, 2.8
2.9, 2.10,
2.11
TP.4
BCL Code
Al
A2
A3
A4
A5
A6
18
-------
Section 5 presents the results obtained in conducting the analyses
described in this section.
MASS DETERMINATION
The mass of deposit collected on the filters was determined by
desiccating the filters for 24 hours and then weighing them. Tare weights
of the desiccated filters were subtracted from the final values to determine
the mass of material collected.
To the extent possible, the samples were stored in a cooled dark
location prior to analysis. Desiccation and weighing were conducted at
room temperature, and the weighing was done in the presence of light.
However, the precautions taken should have protected most compounds from
photo reaction.
PREPARATION OF ADSORBENT COLUMN
AND FILTER SAMPLES
From the group of filters representing a portion of a specific run,
each filter was divided into quarters using a long-bladed lab spatula. Filters
to be used for certain analyses were not to be extracted; these filter
portions were set aside. The remaining filter sections were Soxhlet extracted
sequentially with methylene chloride and methanol until the solvent around the
Soxhlet thimble remained clear (approximately 3 days). Extracts were combined
and reduced in volume. The extracted samples were then divided and, together
with the filter sections that had not been extracted, were distributed to
Battelle (BCL), U.S. Bureau of Mines (USBM), and EPA as follows:
• BCL (Organic) 1/4 of each filter group, extracted
as above.
« BCL (Inorganic) 1/4 of each filter group, no extraction.
• USBM (Organic) 1/4 of each filter group, extracted
as above.
19
-------
*
• EPA (Bioassay) 1/8 of each filter group, no extraction;
1/8 of each filter group, extracted as above
The Tenax-GC adsorbent columns were Soxhlet extracted with pentane
for a period of 24 hours. The appropriate extracts were combined, reduced
in volume and one-fourth of each extract was distributed as follows:
BCL for organic analysis by Level 1 and GC-MS
EPA bioassay analysis by Ames (mutagenicity) and Ram
(cytotoxicity) tests
USBM for Organic Analysis by HRMS
BCL saved for possible future analysis
LEVEL 1 TYPE ORGANIC ANALYSIS
Level 1 organic analysis was conducted on each set of filters
and adsorbent column samples using the analysis procedure described in the
EPA publication listed as Reference 2. The objective of this Level 1
analysis is to semiquantitate the predominant classes of organic compounds
present. This is achieved by subjecting the extracted sample to liquid
chromatography (LC) using solvent gradient elution. Each sample is separated
into eight fractions containing different organic classes which may present.
Because of the large amounts of material collected on the 0.20 x 0.25 m
(8 x 10 inch) filter, only 1/600 (0.25 ml) of the sample was used for liquid
chromatography. All the material extracted from the adsorbent columns was
subjected to this analysis.
Gravimetric Analysis
Following liquid chromatography, the solvent was evaporated from
each fraction in air until a constant weight was achieved. After recording
the weights, the fractions were redissolved in methylene chloride prior to
analysis by infrared spectroscopy. Results are reported in Table 5.
*Following removal by EPA of small segments for the cytotoxicity tests the
filter group was extracted with cyclohexane for BaP analysis.
20
-------
Infrared Spectroscopy
Infrared spectra were obtained on the eight liquid chromatographic
fractions of each of the 11 samples. The samples were prepared for the in-
frared analysis by depositing a film (from Me^Cl™ solutions) of each frac-
tion on an infrared transmitting (KBr) crystal. The spectra were obtained
using a Fourier Transform infrared (FT-IR) system. FT-IR systems use an
interferometer (instead of a monochromator) to generate the spectral data
in terms of an interferogram. A dedicated computer then makes the Fourier
Transform of the interferogram (light intensity versus time) to the more
usable infrared spectrum (light intensity versus wavelength). The sensi-
tivity of the interferometer coupled with the data handling capability
of the dedicated computer are two of the advantages of FT-IR systems
over conventional dispersive infrared spectrophotometers. The use of
FT-IR systems is not a normal part of Level 1 type analysis, but the
ability to subtract spectra (see subsequent discussion) provides quali-
tative identifications that are not possible without this capability.
The liquid chromatographic separation roughly separates the
samples into classes or types of compounds, but each class is still a
complex mixture of many compounds. Since resolution is not sufficient
to isolate individual compounds, it is extremely difficult to identify
individual compounds by IR, or even distinct compound types. However, by
using FT-IR (with the capability to subtract the spectra of consecutive
fractions), such specific identifications are possible.
GAS CHROMATOGRAPHY-MASS SPECTROSCOPY ANALYSIS OF SELECTED SAMPLES
Following infrared analysis of the 88 sample fractions, repre-
sentative filter and adsorber samples from each LC class fraction were
chosen to undergo further analysis using gas chromatography-mass spectro-
scopy (GC-MS) to obtain a semiquantitation of the amount of various species
present. The GC-MS semiquantitation procedure used was similar to procedures
described in Reference 2. The samples were chosen on the basis of their
representation of the classes of compounds found to be common in each class
fraction as determined by IR. Whenever IR indicated any difference in
21
-------
these class fractions, these differing samples were also subjected to GO-MS.
Fractions selected for GC-MS are indicated with an asterisk in Tables 6 and 7.
Gas chromatography separation was carried out using a 1.8 m
(6-ft), 3 percent OV-101 column programmed from 70 to 300 C at 4 C min
Both, chemical ionization (CI) and electron impact (El) mass spectra were ob-
tained using a Finnigan 3200 quadrupole mass spectrometer.
CI mass spectra were obtained using methane as the reagent and
GC carrier gas. The mass spectra thus obtained are usually characterized
by a prominent protonated molecular ion (M+l) together with adduct ions
at (M+29) and (M+41) due to the addition of C H and C H in the mass spec-
trometer. The advantage of CI mass spectrometry is that it is usually pos-
sible to obtain reliable molecular weight data on unknown compounds. Since
CI is a relatively low-energy process compared to El, it generally pro-
duces less fragmentation which is characterized by the loss of neutral mole-
cules and relatively stable ions.
El, while less sensitive than CI, is the most common method for
obtaining mass spectra. Since El is a higher energy process, it produces
more fragmentation giving rise to a more unique mass spectrum for any
compound and, thus, is more useful for spectra matching. Reference mass
spectra are usually available using the El mode.
Mass spectral interpretation was further aided by a spectral
matching routine whereby El mass spectra are compared directly with more
than 30,000 known spectra in the Battelle mass spectral data bank. The
10 best spectral matches are printed out based on a dissimilarity index
of the fraction of unmatched intensities. Compound matches with a
dissimilarity of 0.3 or less and with concurrence of the reference
spectrum from an 8-Peak Spectral Index were considered to be identified
with a high degree of certainty.
Following El and CI GC-MS analysis, the individual components in
each fraction were assigned structures on the basis of CI molecular weight
determinations and fragmentation patterns, and on the basis of El reference
22
-------
spectra and spectral matching. Where several structural isomers were
possible, no attempt was made to identify each isomer since this was be-
lieved to be beyond the scope of this work. A tabulation of the compounds
identified in each sample examined is given in Table 8 in Section 5.
Semiquantitation for the identified GC-MS components were achieved
by ratios of the sum of all measurable peak heights in a given sample to
the height of the known GC component. The ratio was then multiplied by the
total amount of material in the fraction. Whenever an identified component
did not exhibit an easily measurable GC peak, the reconstructed gas
chromatogram was expanded. The expanded GC component area was then measured,
divided by the amount of expansion and proportioned by the sum of the total
peak heights as obtained before expansion. This ratio was then multiplied
by the total amount of material in the fraction as above.
This semiquantitation procedure assumes that all the components
in a given sample pass through the GC column and that they are recorded as
a measureable peak in the gas chromatogram. Although this requirement is
not always met, the procedure will give an indication as to the relative
amounts of each species present in a given sample.
The results of the semiquantitation analysis are presented in
Table 9 of Section 5.
ANALYSIS BY THIN LAYER CHROMATOGRAPHY
Quantitation for benzo-a-pyrene only was performed by Dr. Joseph
Bumgarner (EPA) on the filter extracts using thin layer chromatography (TLC).
These filter extracts were prepared by Dr. Bumgarner by extracting filter
portions designated for cytotoxicity testing. Very small segments were
first removed for the cytotoxicity work; the remainder was then extracted
with cyclohexane. The TLC plates were scanned using a Perkin-Elmer MPF-3
fluorescence spectrophotometer for benzo-a-pyrene using an excitation wave-
length of 388nm and read at an emission wavelength of 430nm.
*See Mannalian Cell Cytotoxicity write up (p.28) for description of
condition of samples.
23
-------
Recovery studies based on spiked blanks showed an average
recovery of 98.9 ± 5 percent. The limit of detection is O.lng haseoon the
standard of a peak being twice the background.
ANALYSIS BY HIGH RESOLUTION MASS SPECTROMETRY
High resolution mass spectrometric (HRMS) analyses were conducted
on 1/4 of the methylene chloride filter extracts and 1/4 of the adsorbent
sampler extracts. This work was performed by Dr. A. G, Sharkey of the
U. S. Energy Research and Development Administration.
Low voltage mass spectrometry was initially performed on all
samples to identify the molecular ions prior to analyses by HRMS.
TRACE METALS ANALYSIS
The elemental composition of emissions from coke oven doors was
determined by spark-source mass spectroscopy (SSMS) of samples .from Test
No. 2. To make this analysis, the 1/4 portions of unextracted 8" x 10"
filters and of a blank filter were subjected to SSMS.
Each of the fiber glass filter samples (A1F through A5F) and
blank (A6F) were extracted in hot HCl'HNOo. The extract which contained
the collected sample as well as some dissolved fiber glass was taken to
dryness and mixed with graphite in the weight ratio of 3 sample to 1 gra-
phite. The graphite is added to provide electrical conduction and to aid
in forming a sound sample briquette for analysis by SSMS.
Standardization of SSMS was carried out by comparing the SSMS
data with data for high concentration elements from analysis by optical
emission spectroscopy.
Additionally, SSMS was conducted on 100 mg of coal and coke as
obtained from the Youngstown facility. These analysis were conducted simi-
larly to that of the glass filter samples above.
24
-------
ANALYSIS OF FLASK SAMPLES OF GASES
All samples were analyzed on an "as received" basis by the mass
spectrometer to provide information on mass to charge ratio of 1 through
100. Each sample was then analyzed by gas chromatography using both a ther-
mal conductivity detector and a flame ionization detector. Sample No. 1,
Test No. 2 was concentrated by using a liquid nitrogen trap to remove the
noncondensables while pumping to approximately 2 torr pressure. This con-
centration step removes oxygen, nitrogen, argon, methane, and carbon mon-
oxide and leaves the condensable compounds in the trap. The trap is then
warmed to evaporate the condensables and condensables are analyzed by mass
spectrometry. This concentration step increases sensitivity and accuracy of
analysis for C,. to C7 hydrocarbon compounds. As the hydrocarbon detected
using the concentration step were not significantly greater than without this
step, it was not included in the analysis procedure for remaining samples.
The specific gases identified by each technique is as follows:
Mass spectroscopy
-nitrogen
-oxygen
-argon
-carbon dioxide
-hydrogen
-benzene
-toluene
Gas chromatography-thermal conductivity detector
-carbon monoxide
-methane
Gas chromatography in flame ionization detector
-acetylene and ethene
-ethane
-propene
-propane
-1-butene
-iso-butane
-n-butane
25
-------
These techniques are sensitive to about 50 ppm, 10 ppm, and 0.1 ppm,
respectively.
BACTERIAL MUTAGENESIS BIOASSAY
Stanford Research Institute examined seven coke-oven samples
(A1F, A3F, ASF, Al, A3, A5, and A6) for their potential bacterial
mutagenesis. The assay was conducted according to the recommendations
of Ames using four of his testor strains with and without the rat
liver microsome induced system. Each sample was examined twice on
separate days by a single plate incorporation dose response assay. The
following was transcribed directly from their report to EPA (Ref.5) with
only minor format changes to conform with the rest of this report.
The Salmonella typhimurium strains used were all histidine
auxotrophs by virtue of mutations in the histidine operon. When these
histidine-dependent cells are grown on a minimal media petri plate
deficient in histidine, only those cells that revert to histidine
independence (his+) are able to grow (mutants). The spontaneous
mutation frequency of each strain is relatively constant, but when a
mutagen is added to the agar the mutation frequency is increased 3 to
100 times. In addition to having mutations in the histidine operon,
all the indicator strains have mutations in the lipoplysaccharide coat
(rfa~) and deletions that cover a gene involved in the repair of uv
damage (uvrB~). The rfa~ mutation makes the strains more permeable to
large molecules, thereby increasing their sensitivity to these molecules.
The uvrB~ mutation decreases repair of some types of chemically damaged
DNA and thereby enhances sensitivity to some mutagenic chemicals.
Strain TA1535 is reverted to histidine prototrophy (his"1") by many
mutagens that cause base-pair substitutions. TA100 is derived from
o
TA1535 by the introduction of the R factor plasmid pKMlOl. The intro-
duction of this plasmid, which confers ampicillin resistance to the
strain, greatly enhances the sensitivity of the strain to some base-pair
substitution mutagens. Mutagens such as benzyl chloride and 2-(2-furyl)-
26
-------
3-(5-nitro-2-furyl) acrylamide (known as AF2) can be detected in plate
assays by TA100 but not by TA1535. The presence of this plasmid also
makes strain TA100 sensitive to some frameshift mutagens—e.g. ICR-191,
benzo-a-pyrene, alfatoxin B^, and 7,12-dimethylbenz-a-antracene. Strains
TA1537 and TA1538 are reverted by many frameshift mutagens. TA1537 is
more sensitive than TA1538 to mutation by some acridines and benzanthracenes,
but the difference is quantitative rather than qualitative. TA98 is
derived from TA1538 by the addition of the plasmid pKMlOl, which makes
this strain more sensitive to some mutagens.
For each experiment, an inoculum was taken from a frozen
repository (-80°C), grown overnight at 37°C in a nutrient broth consisting
of 1 percent tryptone and 0.5 percent yeast extract. After stationary
overnight growth, the cultures are shaken for 3 to 4 hours to ensure
optimal growth. Each culture was checked for sensitivity to crystal
violet. The presence of the rfa~ mutation makes the indicator strains
sensitive to this dye, whereas the parent strain, rfa+, is not sensitive
to the dye. However, the mutation is reversible, leading to the accumula-
tion of rfa+ cells in the culture. Therefore, the cells must be tested
routinely to ensure their sensitivity to crystal violet. Each culture
also was tested by specific mutagens known to revert each test strain
(positive controls).
Some carcinogenic mutagens (e.g., dimethylnitrosamine) are inactive
unless the bacteria strains are converted to their active form by being metabolized
Ames has described the metabolic activation system used. '^ Adult male rats
(250 to 300 g) are given a single 500 mg/kg intraperitoneal injection of a
polychlorinated biphenyl (Aroclor 1254). Four days after the injection,
the animals' food was removed. On the fifth day, the rats were killed and
the liver homogenate prepared.
The assay itself was conducted in the following manner. To sterile
13 X 100mm tubes containing 2 ml of 0.6 percent agar containing .05 mM histidine
and .05 mM biotin were added in order
0.05 ml of indicator organisms
0.5 ml of metabolic activation mixture (when used)
27
-------
Up to 50 yl of a solution of the test chemical.
This mixture was stirred gently and then poured onto glucose minimal agar
having the following composition:
15 g of agar, 20 g of glucose, 0.2 g of MgS04'7 H20, 2 g of
citric acid monohydrate, 10 g of fc^HPO^, and 3.5 g of NaHNffy
PO^ 'H20 per liter.
After the top agar had solidified, the plates were incubated at 37C for 2 days.
The number of his revertants is counted and recorded. Some of the
revertants are routinely tested to confirm that they are his , require biotin,
and are sensitive to crystal violet (rfa~).
All of the samples were analyzed as they were received.
Dilutions of the samples were made in dimethylsulfoxide (DKSO). The
amount of sample added per plate is expressed as the pi of sample in
50 yl of DMSO. The results of these investigations are reported in
Tables 19 and 20 (Presented in next Section).
MAMMALIAN CELL CYTOTOXICITY
Preliminary cytotoxicity testing of filter samples A1F, A2F, A3F,
A4F, and ASF was preformed by Northrup Services utilizing cultured rabbit
alveolar macrophages . The filter segments as received had suffered some
degradation as a result of handling and shipping. Some of the particulate
material had flaked off the filters and was lying loose in the containers.
As is often the case in filters collected in the field, the particulate was
not evenly distributed all over the surface of the filter. This situation
was aggravated by the loss of some of the particulate. An attempt was made
to remove the particulate from the filter however, it was impossible to do
this without also removing some of the filter fibers that were imbedded in
the particulate, thus it was decided that the test would be conducted by
removing squares of material from the filters and utilizing these squares in
the test. Therefore, small, medium, and large squares were removed from each
28
-------
filter in an effort to develop a dose response. However, due to the
heterogeneous distribution of particulate on the filters, the dose response
data generated are highly questionable and are not reported.
No control filters were provided which lead to the decision that
the relatively clean edge of each filter would be removed and used as a
control.
29
-------
SECTION 5
RESULTS
Test No. 1 was in some ways a preliminary test, in that problems in re-
cording data, changing filters and adsorbers, and in recording flows could
not be anticipated as they could for Test No. 2. Test No. 2 was better
organized and the various flows were more accurately recorded; therefore, of
the two tests completed at Youngstown, Test No. 2 (on April 1-2) was chosen
for the more complete analysis, including particulate mass and gas analysis
and, in most instances, the data obtained in Test No. 2 are used in presenting
results.
Although the two sampling runs were made on the same oven, the concentra-
tion of emissions was greatly different for the separate runs made only about
a day apart. Test No. 2 showed an overall particulate emission rate about
three times that measured in Test No. 1, and total light hydrocarbon concen-
trations about two times as great as measured for Test 1. The fact that such
wide variations in emissions occur on a single oven served to illustrate the
wide range that can exist between emissions data obtained on different coke
ovens.
DOOR LEAKAGE AND FLOW DATA
Table 2 gives a summary of flow data taken at irregular periods through-
out Test No. 2. Test No. 2 started at 1303 April 1, and Table 2 shows that
o
leakage was relatively high around the door with a value of 99 m /hr (58 cfm),
at 1320, or 17 minutes later. Theoretical calculations indicate that if an
0.8 mm (1/32-inch) crack is present around half of the perimeter of a door
4.1 m (13.5 ft) high and 0.7 m (2.25 ft) wide, leakage would be 217 m3/hr
30
-------
TABLE 2. TEST NO. 2 FLOW FOR AIR INPUT, HI-VOL
SAMPLER, AND DUCT ORIFICE
Date
and
Clock
Time
4/1-1320
4/1-1600
4/1-1737
4/1-1910
4/1-2237
4/2-0203
Hood Air
Input
(Rotometer
Flow}scfm'a
14.5
37.3
41.0
41.0
41.0
41.0
Hi-Vol Sampler Flow
Orifice
Gas Flow,
' scfm
47.6
49.0
50.5
45.8
46.8
45.8
Uncorrected,
scfm
29.2
15.6
16.0
26.3
30.7
30.0
/ v \
Corrected,^0'
scfm
24.5
13.1
13.4
22.2
25.8
25.2
Coke Oven
Door Leakage'0'
scfm(c)
58
25
23
27
32
30
3
m /hr
99
42
39
46
54
51
(a) At 29.92" Hg and 70 F
(b) Net flow is gross Hi-Vol corrected for calibration factor.
(c) Door leakage = Hi-Vol flow + Orifice flow - Hood air input.
3 *
(128 ft /min ), assuming a gas pressure of 8 mm H.O in the oven, and a gas
temperature at the crack of 316 C (600 F). Table 2 shows that leakage had
3 3
declined to 42 m /hr (25 ft /min) about three hours after the start of the
3 3
test, leakage remained in the range of 39 to 54 m /hr (23 to 32 ft min)
for the remainder of the test.
PARTICULATE MASS EMISSIONS
Although adsorber flows and overall data from Test No. 1 were not as
accurate as Test No. 2, filter weights and flows from the Hi-Vol samplers
were available. Accordingly, Table 3 summarizes the filters and their
weights, and the flow of gases through the filter from Test No. 1.
Table 4 summarizes filter weight and flow data from the Hi-Vol samplers
for Test No. 2. Door leakage in Test No. 2 was much greater than the door
leakage in Test No. 1. Table 3 shows that Test No. 1 required 18 filters
* Gas pressure = 8 mm = 0.011 psi; temperature at leakage point = 600 F
Area of opening around door = (27 + 4.5) x 1/32 x 1/12 x 1/2 = 0.04 ft
Gas density p = 0.075 x (460 + 70)/(460 + 600) = 0.0375
Then velocity V = 96.3 x (0.011/0.0375) = 52.2 ft/sec
and volume Q + AV = 0.041 x 52.2 x 60 = 128 ft3/min.
31
-------
TABLE 3. SUMMARY DATA, NO. 1 TEST. COKE OVEN DOOR LEAKAGE
REPUBLIC STEEL, YOUNCSTOWN, OHIO A/1-2/76
Approximate
Hours
Clock Time Into Filter
On
3:10
3:21
3:30
4:12
4:25
5:05
6:05
6:35
7:10
8:10
9:45
10:45
11:10
12:25
1:10
2:10
3:10
5:10
Off
3:21
3:30
4:12
4:25
5:05
6:05
6:35
7:10
8:10
9:45
10:45
11:10
12:25
1:10
2:10
3:10
5:10
6:45
Test
1
2
3
4
5
8
10
11
12
14
16
Number
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
Average
Uncorrected
Filter
Flow As
Recorded
on Hl-Vol .
Chart. cfmU;
32
34
38
36
38
39
38
39
38
39
39
39
37
40
40
41
43
41
Filter
Flow
Corrected With Hl-Vol
Filter Flow, Calibration Dilution
... ' Correction, Correction
scfm(b) 8cfmXcJ Factor^)
24
26
29
27
29
29
29
29
29
29
29
29
28
30
30
31
32
31
24
22
29
23
29
24
29
24
29
24
29
24
28
26
29
28
30
28
.58
.57
.61
.58
.67
.79
.80
.59
.62
.59
.62
.59
.61
.51
.53
.52
.53
.52
Net
Leakage
Flow
Through
Hi-Vol,
scfra
14
13
18
13
19
19
23
14
18
14
18
14
17
13
15
15
16
15
Time of
Flow
Through
Filter,
minutes
11
9
42
13
40
60
30
35
60
95
60
25
75
45
60
60
120
95
Total
Time,
minutes
11
20
62
75
115
175
205
240
300
395
455
480
555
600
660
720
840
935
Leakage Flow
Through Filter,
ft3 o,3
154
117
756
169
760
1140
690
490
1080
1330
1080
350
1275
585
900
900
1920
1425
4.361
3.313
21.410
4.786
21.523
32.285
19.541
13.877
30.586
37.666
30.586
9.912
36.108
16.567
25.488
25.488
54.374
40.356
TOTALS
15.6
18
935
-------
TABLE 3. SUMMARY DATA, NO. 1 TEST, COKE OVEN DOOR LEAKAGE
REPUBLIC STEEL, YOUNCSTOWN, OHIO 4/1-2/76 (Continued)
Particulate Partlculate
Approximate Weight Concentration
Hours on In Leakage Gas,
Clock Time Into Filter Filter.
On Off Test Number grams mg/ra
3:10 3:21 100 5.9494 1364
3:21 3:30 101 9.3877 2834
3:30 4:12 1 102 10.0692 470
4:12 4:25 103 3.8342 801
4:25 5:05 2 104 5.7531 267
5:05 6:05 3 105 5.0010 155'
6:05 6:35 106 1.9813 101
6:35 7:10 4 107 5.6064 404
3:10 8:10 5 108 3.1818 104
8:10 9:45 109 2.7937 74
9:45 10:45 110 1.3152 43
10:45 11:10 8 111 0.8054 81
11:10 12:25 112 1.2681 35
12:25 1:10 10 113 0.6783 41
1:10 2:10 11 114 0.0528 2
2:10 3:10 12 115 0.0482 2
3:10 5:10 14 116 0.0488 1
5:10 6:45 16 117 0.0616 2
TOTALS 15.6
(a) Average flow from Hl-Vol flow chart as recorded during period
(b) Average gas temperature reading (208 F) and average barometer
correction factor of 0.77; corrected to 29.92" Hg and 70F.
(c) Odd numbered filters used with No. 1 Hi-Vol and even numbered
calibration correction in the range operated of -16 percent;
which it operated.
(d) Dilution factor • Hi-Vol, cfm + Orifice, cfm - cfm air input
Hi-Vol, cfm + Orifice, cfm
(e) Period refers to discrete test periods of 1 hr, 1 hr, 2 hr, 4
Particulate
Weight
for(e)
Period. '
grams
25.4063
9.5873
12.5887
8.0961
2.0474
0.1104
57.8362
Leakage
Flow
Through
Filter for
Period (e)
n,3
29.084
26.309
65.703
108.75
103.651
94.730
428.227
Particulate
Concentration
in Leakage Gas
for Period/6'
mg/m
874
364
192
74
20
1
135
Particulate
Emission
Rate,
mg/min
409.8
180.9
100.7
33.7
8.5
0.5
61.9
of flow through filter.
reading (29.01" Hg) at filter during test used to establish
filters used with No. 2 Hi-Vol; No. 1 Hl-Vol required a
No. 2 did not require calibration correction in the range at
under hood
hr, 4 hr, and 4
hr.
-------
TABLE 4. MO. } TEST. COKE OVEN DOOR LEAKAGE
REPUBLIC STEEL. YOUNCSTOWN, OHIO
4/1-2/76
Cloc
ON
1303
1400
1416
1555
1654
1741
1755
IB25
1912
1922
2000
2105
2150
2225
2325
2351
0100
< Time
OFF
1304
1405
1506
1607
1708
1755
IE2S
1912
1922
2000
2105
2150
2225
2325
2351
0100
0215
Approximate
Hour a
Into
Teat
1
2
3
4
5
6
7
8
9
10
11
12
13
Filter
Number
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
Uncorr^ctcd
Filter
Flow Aa
Recorded
on Hl-Vol
Chart, clsO*'
38
28
38
28
37
20
35
30
32
30
36
23
30
32
32
40
30
32
30
35
30
28
32
38
30
34
36
32
25
30
30
34
33
36
33
37
35
30
35
38
30
38
40
Corrected
Filter
.cfift>
29
22
29
22
28
15
27
23
25
23
28
22
23
25
25
31
23
25
23
27
23
22
25
29
23
26
28
25
19
23
23
26
25
28
29
28
27
23
27
29
23
29
31
Flew
Through
With Hl-Vol
Calibration
Correction,
aofm(e)
29
IB
29
18
28
13
27
19
25
19
28
18
23
21
25
26
23
21
23
23
23
18
25
24
23
22
28
21
19
19
23
22
25
24
29
24
27
19
27
24
23
24
31
Dilution
Correction
Factor(d)
.81
.78
.81
.78
.81
.76
.81
.78
.80
.78
.81
.78
.79
.79
.80
.80
.79
.79
.79
.79
.79
.78
.DO
.49
.48
.47
.52
.48
.45
.45
.48
.43
.46
.45
.45
.41
.44
.37
.44
.41
.41
.41
.47
N«t Leak-
age Flow
Through
Hl-Vol,
ncfai
23
14
23
14
23
10
22
IS
20
15
23
14
18
17
20
21
18
17
18
18
18
14
20
12
11
10
15
«• 10
9
9
11
9
12
11
13
10
12
7
12
10
9
10
13
Time of
Flov
Through
Filter ,
plnutaa
1
7
8
16
10
10
11
7
14
6
11
12
12
11
12
12
14
17
16
14
30
47
10
38
65
45
35
60
26
69
7J
Total
Tl n«
I IDA .
mlnutca
1
2
6
9
25
34
37
40
43
48
52
57
63
70
78
94
104
114
125
132
146
152
163
175
187
198
210
222
236
253
269
283
313
360
370
408
473
518
553
61.3
639
708
783
Leakagi
Through
ft3
23
1.4
92
42
368
90
66
45
60
75
92
70
108
119
160
336
160
170
198
126
252
84
220
144
132
110
180
120
126
153
176
125
360
517
130
3.10
780
3 -.5
420
600
2 34
690
112}
Flov
Filter,
m
0.65
0.40
2.61
1.19
10.42
2.55
1.87
1.27
1.70
2.12
2.61
1.98
3.06
3.37
4.53
9.52
5.10
4.81
5.61
3.57
7.14
2.38
6.23
4.08
3.74
3.12
5.10
3.40
3.57
'..33
4.93
3.57
10.20
14.04
3.63
10.76
22.09
8.92
11.89
16.99
6.62
H.54 •
31.86
•rorALs
13.2
43
783
-------
TABLE 4. HO. 2 TEST, COKE OVEN DOOR LEAKAGE
REPUBLIC STEEL, YO'JNCSTOWH, OHIO
4/1-2/76 (Continued)
1
Clock Tlmo
Approxlait*.
ON OFF Hour*
Into
Te«c
1303
1400
1456
1555
1654
1741
1755
13?5
1912
1022
200J
2105
2150
2225
2325
2351
0100
TOfALS
13
1405
1506
1607
1708
1755
IB25
1912
1922
2000
2105
2150
277'i
2325
2351
0100
0215
1
2
3
4
s
6
7
8
9
10
11
12
13
13.2
Filter
Number
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140 '
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
15/
158
159
160
161
43
Partlculet*.
WetGht •
on
Filter,
grama
0.0162
2.3429
1.5054
7.6181
9.9949
10.0939
4.4263
5.1016
4.8536
5.3571
4.2589
4.9861
2.4750
5.0502
7.46/.0
7.9094
5.4523
5.5386
5.2461
3.5385
3.5813
3.1257
3.3512
3.3360
2.9323
2.6086
2.1009
2.2416
2.2274
2.6734
1.9769
2.1100
3.1065
2.2147
1.07H1
2.8207.
3.0858
2.0096
1.5319
1.3301
0.5264
0.2584
0.0761
Pertlculat*
Concentration,
in Leakage
Caa,
mg/n3
948
5057
577
6402
959
3?58
2307
4017
28S5
2527
1632
2518
009
1499
1648
831
10IS9
1151
935
911
502
1313
538
818
784
836
412
659
624
617
397
591
305
151
233
262
140
225
129
73
80
13
2
Parclculac*
Weight
for Period .<•> P«lod (•) .g/.tB
29.37
30.39
29.01
18.93
12.88
24.84
14.44
22.09
8.92
28.88
6.62
19.54
31.86
277.77
2082
1115
763
640
523
214
270
140
225
99
80
13
2
3S5
1002.5
555.6
363.6
198.3
143.8
69.1
81.2
47.5
'.4.7
30.1
20.2
3.7
1.0
195.1
Footnote* •«»• u TABLE 3, p«g« 33.
-------
over a period of 935 minutes, collected 57.8 grams of particulate from a
calculated gas flow through the filters due to door leakage of 428 cubic
meters, and indicates an overall particulate concentration in the door
leakage gas of 135 mg/m3. By comparison, Table 4 shows that Test No. 2
required 43 filters over 783 minutes, collected 154 grams of particulate
from total door leakage gas flow through the filters of 278 m3, and
indicates an overall particulate concentration in the door leakage gas
o
of 555 mg/m .
Figure 3 is a graphic presentation of particulate concentrations for
the door leakage gas plotted against time in hours into the test. Figure 4
shows the emission rate in mg/min, for Test Nos. 1 and 2. For these figures,
each curve point is established by data for one sampling segment and is
plotted against the time mid-point of that time segment.
It is important to recognize that gas flow values for the Hi-Vol sampler
presented in both Tables 3 and 4 are values obtained from the recorder charts
on the samplers. Conversely, Hi-Vol flow values shown in Table 1 are in-
stantaneous values recorded during the test.
One important result from Tables 3 and 4 is the fact that the overall
particulate concentration for Test 2 was 4.1 times as great as the emission
rate measured in Test No. 1. There are a number of possible reasons for this
wide difference in particulate emission rate, none of which can be supported
by very specific data. The most probable explanation is that the oven door
did not fit on as tightly for Test No. 2 as had been the case for Test No. 1.
It appeared to members of the research team that the oven door and the door
frame did not receive as thorough a cleaning prior to the coking cycle of
Test 2 as in the case of Test No. 1. However, there was no comparative mea-
sure available for cleaning of deposits from the oven door.
Other factors that might affect door leakage are oven temperature and
pressure. No detailed data on these parameters were available for both tests,
however, operating personnel reported that S-mm-^O back pressure was main-
tained on all batteries. In response to a request for temperature and pres-
sure information, the following coke-side flue temperatures were measured at
approximately 1500 hours April 1.
36
-------
2400
0
6 8 10
Time From Beginning of Test, hours
12
14
16
FIGURE 3. PARTICIPATE EMISSION CONCENTRATION FOR DOOR-LEAKAGE GAS
VS TIME INTO COKING CYCLE
-------
1000 -
OJ
OO
4 6 8 10 12
Time From Beginning of Test, hours
FIGURE 4. PARTICULATE EMISSION VS TIME INTO COKING CYCLE
14
16
-------
Temperature,
Flue No. _C F
1 1338 2440
2 1371 2500
3 1382 2520
4 1382 2520
5 1382 2520
6 1371 2500
7 1382 2520
8 1393 2540
9 1382 2520
10 1360 2480
11 1360 2480
12 1349 2460
13 1327. 2420
In general, operating personnel reported pusher side flue temperatures to be
1316 C (2400 F) and coke side to be 1371 C (2500 F). They also mentioned
that, in the past, normal back pressures had been as high as 14-mm H20.
RESULTS OF LEVEL 1 ANALYSES
Gravimetric-Analysis
Table 5 presents results for the gravimetric analysis conducted on the
filter and adsorber samples as part of the Level 1 analysis. Results are pre-
sented in mass units per hour. Each filter weight was corrected for the
total amount of material that would have been obtained if the entire sample
extract had been analyzed.
Infrared Spectroscopy
The interpretation of the infrared spectra (and subtracted spectra) of
the 88 samples (8 fractions of 11 samples) are summarized in Tables 6 and 7.
For several of the fractions, the spectra indicated only the presence of in-
organic Si-0 components. From past samples we have demonstrated that these
inorganic components arise from background components and not from the col-
lected sample. Thus, the samples where only inorganic material has been ob-
served have been marked insufficient sample. From the method of preparing
the sample for infrared analysis and from the method of running the spectra,
this means that less than 100 nanograms of organic material was present.
An example of the FT-IR technique and the subtractive procedure is the
identification of carbozole in Fraction 3 of several of the samples; this is
39
-------
TABLE 5. LEVEL 1 ANALYSES, GRAVIMETRIC RESULTS
Fraction
Filter
A1F
A2F
A3F
A4F
ASF
Adsorber
Al
A2
A3
A4
A5
A6(*)
1
8.28
2.79
1.52
.88
.31
1966.40
3275.03
1000.43
1202.53
581.42
20.98
2
22.
20.
11.
4.
•
2704.
1429.
1097.
879.
1067.
3.
3
71
36
70
99
15
20
73
24
45
07
26
7.
4.
3.
1.
•
159.
213.
81.
10.
44.
1.
93
95
44
33
33
29
51
54
43
56
01
4
1
6.62
3.75
2.60
1.20
.26
Hi
335.43
613.83
189.34
137.74
106.57
.96
5
;/hr
2.88
3.74
1.40
.88
.23
;/hr
497.72
571.13
127.58
271.89
231.11
1.10
6
.87
.96
1.68
.58
.08
103.42
143.35
106.22
27.09
17.17
.91
7
10.
9.
3.
1.
•
321.
293.
96.
75.
64.
6.
09
25
94
39
30
35
83
69
91
89
37
8
.50
.61
.74
.10
.01
5.08
7.37
2.60
3.66
2.89
.18
(a)
Compressor Air Blank.
40
-------
TABLE 6. LEVEL 1 IR ANALYSIS OF LC FILTER FRACTIONS
Fraction
Sample
A1F
A2F
A3F
A4F
A5F
I
. Aliphatic Hv
Fused
Aliphatic Hv
Fused
Aliphatic HV
Fusee
*
Aliphatic ti\
Fusee;
Aliphatic Hv
Fus
2
drocarbons
Ring Aromati
(a)
drocarbons
Ring Aromati
drocarbons
Ring Aromati
drocarbons
jling Aromati
drocarbons *
ed Ring Aroma
3
:s
Possible
Carbazole
cs
Carbazole
C=N(d)
cs
Carbazole
*
CSN
(a)
tics v '
Ketone
4
*
Phenol (c>
c=N(d)
s=Nw;
Phenol
Phenol
Phenol
5
Phenol
(b)
*
Pheno
(b)
A
(b).
A
_E
6
or Amine
(b)
romatic Carbo:
comatic Carbo
Aromatic Ca
Acid
*
cf-pT PhfV^glaf
7
Aromatic
Carboxylic
Acid(b^
Aromatic
Carboxylic
Acid
*
cylic Acid
cvlic /\rid
rboxylic
^
8
Insuff icier
Sample
Insuff icier
Sample
Insuff icier
Sample
Insuff iciei
Sample
Insuff iciei
Sample
t
t
t
.t
it
(a) Possibly contains pyrene and/or benzpyrenes.
(b) Possibly S compounds.
(c) Possibly carbazole types.
(d) Possibly C^N or (less likely) C=C.
* Sample subjected to GC-MS analysis.
-------
TABLE 7. LEVEL 1 IR ANALYSIS OF LC ADSORBER FRACTIONS
Fraction
Sample
Al
A2
I
A3
A4
A5
A6
1
Aliphatic
Hydrocarbo
Aliphatic
Hydrocarbo
Napht-ha
A
Aliphatic
Hydrocarbo
Nflnhrha"
Aliphatic
Hydrocarbo
Mo^t-,l-V,^1
11J.pllLU.ii-l.
Aliphatic
Hydrocarbc
Naphtha
Aliphatic
Hydrocarbi
2
is
Fused Riru
Aromatics
(a)
is
e.ne
Fn«(=rl Rinp
Aromatics (a
is
'HP
"Ril eprj 1^ -f no
AromaticsX3
ns
CJQp
Fused Ring
Aromatics <-a
*
ns
ene
Fused Ring
Aromatics (a
ns A
Silicons
3
Nitrile ' Nnr(
nj . Ket^n0 A
nols
Ketone tr
Phthalat
ester
-coni . Ket<
cient
le
7
le B
le B
np R
A
np R
e *
ne B
Non-conj .
Ketone
A
Phthalat<
8
Insuf fi
cient
Sample
T
cient
Sample
Insuffi-
cient
Sample
Insuffi-
cient
Sample
Insuffi-
cient
Sample
Insuffi-
cient
Sample
-------
Footnotes for Table 7.
(a) Pyrenes and benzpyrenes are likely in this fraction.
(b) Probably primarily carbazole.
(c) Nitrile is most likley but the possibility of C-C cannot be excluded
in dramatic p subs nitrile is most likely.
(d) Can be conjugated ketone, quinone or mixture of both.
* Sample subjected to GC-MS analysis.
43
-------
illustrated in Figure 5. The top spectrum is an absorbance spectrum of Frac-
tion 4 of Sample A1F; the bottom spectrum is an absorbance spectrum of Frac-
tion 3 of Sample A1F. The two spectra in the middle of Figure 4 are the
resultant of subtracting the spectrum of Fraction 3 from two different ratios
of Fraction 4. These two subtracted spectra show:
(1) Carbazole is the main component of Fraction 3
(2) Similar components in both Fraction 3 and Fraction 4
(3) Correct intensities of Fraction 4 by removing the
contribution (carbazole) of Fraction 3, thus aiding
in the identification of Fraction 4.
Another example of the subtraction technique is given in Figure 6. At
the top of this figure is an absorbance spectrum of Fraction 6 of Sample A2.
At the bottom is a spectrum of Fraction 3 of this same sample. Note that
these spectra show many similarities making it difficult (because the separa-
tion might not be clean) to identify the unique components of each fraction.
In the middle of Figure 6 is the resultant of subtracting the spectrum of
Fraction 5 from that of Fraction 6. The absorption pointing up and down near
1700 cm clearly determines that:
(1) Fractions 5 and 6 have different (but very similar)
carbonly absorptions
(2) The positions (frequencies) of each carbonyl vibration
coupled with the other observed bands in each fraction
identifies both as non-conjugated ketones, and each
fraction as having a different ketone structure. These
ketones were designated as Ketone A and Ketone B,
GAS CHROMATGRAPHY-MASS SPECTROSCOPY RESULTS
The major species determined by the GC-MS analysis of selected samples
are listed in Table 8. Examples of El and CI reconstructed gas chromatograms,
with the more abundant species labeled, are shown in Appendix C. Individual
mass spectra for the CI GC-MS analysis are also given in Appendix C.
Results obtained by the semiquantitation procedure for identified GC-MS
components are listed in Table 9.
44
-------
A. Fraction 4 of AlF
B. 15 percent Frac-
tion 4 - 100
percent Fraction 3
C. 75 percent Frac-
tion 4 - 100
percent Fraction 3
D. Fraction 3 of AlF
1800
1600 1000 600 cm
FIGURE 5. ARSORBANCE INFRA RKD SPECTRA
_ 1
m A.
45
-------
1
!
i '
.
i
;
. . .. .
i -
. • — ---
|
V'\.N
1
1
I
!
;
j
1
\
\ i
i
,
1
, i
i
it i
*'
1
\
i ...
(j ;
I '
If*
_
,V
•
i
J
I
'''"
..-.
|
\ '• / *
'
_ .
i
l;x ,
'
'
1
]
i
1
;.._.
i
.1...-
•
i
i,
._
-
A. Fraction 6 of A2
B. 100 percent Fraction
6 - 100 percent
Fraction 5
C. Fraction 5 of A2
1800 1400 1000
FIGURE 6. ABSORIiANCK INFRARED SPECTRA
(>6o cm"1
-------
TABLE 8. GC-MS ANALYSES OF SELECTED COKE OVEN DOOR SAMPLE FRACTIONS
Fraction Sample Compounds Identified^
Adsorbent Samplers
1 A2 Naphthalene, methylnaphthalenes, dihydro-
naphthalene, C2~naphthalene
2 A5 Naphthalene, methylnaphthalenes, anthracene,
acenaphthalene, methyl acenaphthalene/
fluorene, biphenyl acenaphthene/
methyl biphenyl, tetrahydronaphthalene,
indene
3 Al Indole, naphthylisocyanide, carbazole
methyl carbazole
3 A5 Carbazole, methyl, carbazole
4 Al C2~phenol
5 Al C2~phenol, C3~phenol
6 A2 No positive identification
7 A3 Quinoline, phthalate, acridine
7 A3 No positive identification
8 All No GC components
Filters
1 A3F Anthracene, methylanthracene, fluorene,
methylfluorene, fluoranthene, methyl-
fluoranthene, pyrene, methylpyrene,
phthalate
2 ASF Anthracene, methylanthracene, fluoranthene,
methylfluoranthene, pyrene, methylpyrene,
benzphenanthrene, chrysene, benzopyrene,
benzofluoranthene, methyl chrysene/
methyl benznathracene, methyl benzopyrene/
methyl benzfluoranthene, coronene
3 A3F Carbazole, methylcarbazole, benzanthrone,
cyanofluorene, dimethyl carbazole
-------
TABLE 8. GC-MS ANALYSES OF SELECTED COKE OVEN DOOR SAMPLE FRACTIONS
(Continued)
(a\
Fraction Sample Compounds Identifiedv '
A1F (2)C3-phenol, C2-benzaldehyde, naphthol,
methylnaphthol, hydroxyfluorene, methoxy-
fluorene, benzanthrone
5
6
7
8
2
3
7
A1F
A5F
A4F
A1F
A6
A6
A6
No identification
Filters
Quinoline, phthalate
No GC components
Compressor Air Supply (Blank)
No positive identification
Naphthalene, anthracene
Phthalate
Cat
v ' See text.
-------
TABLE 9. SEMIQUANTITATION OF SELECTED COKE OVEN DOOR
SAMPLE FRACTIONS
Percent of
Fraction Sample Compound Total Emission
1
2
3
4
A3F Fluorene
Methyl fluorenes
Anthracene
Methyl anthracenes
Pyrene
Methyl pyrenes
Fluoranthene
Phthalate
ASF Anthracene
Fluoranthene
Pyrene
Methyl pyrenes/methyl
fluoranthenes
Dimethyl pyrenes /dimethyl
fluoranthenes
Benzo phenanthrenes
Chrysene
Methyl chrysenes/benzoanthracenes
Dibenzo anthracenes
Benzo fluoranthenes
Benzo pyrenes
Methylbenzopyrenes/methylbenzo-
f luoranthenes
Indeno pyrene
Benzo perylene
Coronene
A3F Carbazole
Methyl carbazoles
Dimethyl carbazoles
Cyanofluoranthene
Benzanthrone
Phthalate
A1F C2 benzaldehydes
03 phenol
Methyl allyl phenol
Methyl indenone
Naphthol
Methyl naphthols
Hydroxy fluorene
Methoxy fluorene
Benzanthrone
2
2
21
12
10
10
12
5
5
15
10
4
1
4
21
4
2
15
11
1
4
4
1
29
13
6
5
3
3
6
7
3
5
7
14
7
1
6
Emission
Rate,
mg/hr
30
30
320
180
150
150
180
75
7.5
22
15
6.0
1.5
6.0
31
6.0
3.0
22
16
1.5
6.0
6.0
1.5
95
43
20
16
10
10
400
460
200
330
460
920
460
65
400
-------
TABLE 9. SEMIQUANTITATION OF SELECTED COKE OVEN DOOR
SAMPLE FRACTIONS (Continued)
Fraction
7
1
2
3
3
4
5
Sample Compound
A4F Quinoline
Phthalate
A2 Naphthalene
Methyl naphthalenes
Dimethyl naphthalenes
Dihydroxy naphthalene
A5 Indene
Naphthalene
Methyl naphthalene
Tetrahydro naphthalene
Acenaphthalene
Biphenyl/acenaphthrene
Methyl biphenyl/methyl
acenaphthrene/f luorene
Anthracene
Methyl anthracene
Dimethyl anthracenes
Trimethyl anthracenes
Fluoranthene
Pyrene
Methyl pyrenes/methyl fluor-
anthenes
Benzophenanthrene
Chrysene
Al Indole
Methyl indole
Carbazole
Methyl carbazole
Naphthyl isocyanide
A5 Carbazole
Methyl carbazole
Phthalate
Al C2 phenol
Alkyl phenol
Al C2 phenol
Co phenol
Percent of
Total Emission
19
16
35
26
8
5
10
33
19
3
11
6
6
3
1
0.05
0.05
1
1
0.1
0.05
0.07
19
6
25
3
25
57
16
16
53
20
11
8
Emission
Rate,
mg/hr
265
220
1100
850
260
160
100
350
200
30
120
60
60
30
10
0.5
0.5
10
10
1.0
0.5
0.7
30
9.0
40
5.0
40
25
7.0
7.0
180
68
55
40
50
-------
TABLE 9. SEMIQUANTITATION OF SELECTED COKE OVEN DOOR
SAMPLE FRACTIONS (Continued)
Fraction Sample Compound
7 A5 Quinoline
Methyl quinolines
Dimethyl quinolines
Acridine
Methyl acridines
Benz acridines
3 Phenyl propanilide
Methyl 3 phenyl propanilide
Phthalate
Percent of
Total Emission
15
3
5
7
4
1
10
6
53
Emission
Rate,
mg/hr
10
2.0
3.0
4.5
2.5
0.6
6.5
4.0
34
51
-------
THIN LAYER CHROMATOGRAPHY RESULTS
Results of the quantitation of BaP as determined by TLC are listed in
Table 10. This quantitation was performed only on the filters extracted by
EPA with cyclohexane.
HIGH RESOLUTION MASS SPECTROMETRY RESULTS
The results of the HRMS analyses conducted by Dr. Sharkey are listed in
Tables 11, 12, and 13. The nine formulas, listed in Table 13, include highly
carcinogenic PNA's. Few of these possible structures were found in the ex-
tracts of the absorber column, while there were strong indications for all
except 3-methyl-cholanthrene (mass 268) in the filter extracts.
TRACE METAL RESULTS
Results of trace metals emissions, as determined by the SSMS analysis
described earlier, are presented in Table 14. Each element is reported in
yg above the blank for the entire sample collected.
Differences among samples in detection limits of a given element arise
from differences in items such as degree of plate fogging and/or plate sen-
sitivity and sometimes differences in sample composition. In these analyses,
differences are related to plate characteristics. Also, detection limits for
many of the elements are mandated by the elemental blank levels of the glass
fiber. This is especially true for the alkali and alkaline earth elements as
well as for some of the transition elements such as Zn and Zr.
SSMS results for analyses of the coal and coke samples are presented in
Table 15.
GASEOUS EMISSIONS
The light hydrocarbons and inorganic components detected in the flask
samples for Test No. 1 and 2 are presented in Table 16 and 17, respectively.
These results are calculated on a dry basis; the amount of water detected is
shown at the bottom of each table. Water was much higher for Test No. 2 re-
sults which might be due to the fact it rained during this test.
For compounds present at concentrations lower than the sensitivity of the
analysis technique, values are reported as less than 0.1 ppm or less than 50
ppm of the component was present. In the first instance a more sensitive
52
-------
TABLE 10. TLC QUANTITATION RESULTS FOR BaP
Sample pg mg/hr
A1F 710 5.0
A2F 440 4.0
A3F 700 2.7
A4F 540 1.1
A5F 230 .35
53
-------
TABLE 11. LOW VOLTAGE MASS SPECTROMETRIC RESULTS OBTAINED
ON FIBER SAMPLES
Major Structural Type,
Including Alkyl Derivatives
({i-Naphthalenes
6-Ring Peri
3-Rings
1 , 12-Benzoperylenes
Benzonaphthothiophenes
5-Ring Cata
Acenaphtyhlenes + Fluorenes
5-Ring Peri
Dibenzothiophenes
<|>- 3-Rings or Binaphthyls
4-Ring Cata
4-Ring Peri
Coronenes
Dinaphthothiophenes
AIF
8.8
.3
16.5
1.3
2.0^
1.6
8.6
8.3
3.5^
3.5
10.7
17.9
.2
-
Total lonization, percent
A2F A3F A4F A5F
8.5
1.7
10.0
4.0
2.2^
3.9
5.4
15.1
2.2^
4.9
12.1
17.8
.5
-
9.6
.4
12.4
2.0
2.3
2.2
4.5
11.6
2.5
4.2
13.4
23.7
-
.6
9.2
.9
7.4
3.2
2.3
3.2
2.9
13.4
1.9
5.1
20.8
19.4
.2
-
8.2
1.9
4.2
4.8
2.2
3.1
4.5
16.0
1.6
4.5
18.1
21.4
.5
-
(a) Percent total ionization assumes equal sensitivity for all
components. Valid for comparison of similar samples.
(b) Confirmed by HRMS.
(c) Not confirmed by HRMS.
Note: 3-Ring compounds are anthracene and phenathrene
4-Ring peri compounds are pyrene and fluoranthene
4-Ring cata compounds are chrysene, 3.4-benzphenanthrene,
tetraphene, and tetracene
5-Ring peri compounds are perylene, benzo(a)pyrene,
benzo(e)pyrene
5-Ring cata compounds includes 11 5-ring compounds
6-Ring peri compounds includes 11 6-ring compounds
54
-------
TABLE 12. LOW VOLTAGE MASS SPECTROMETRIC RESULTS OBTAINED ON
ADSORBER SAMPLES
Benzenes
Indanola/Benzothiophenes
(((-Naphthalenes
Phenols
3-Rings
Acenaphthylenes + Fluorenes
5-Ring Cata
Acenaphthenes + Biphenyls
Dibenzofurans
Naphthalenes
Dibenzothiophenes
Indenes
4-Ring Cata
Indans/Tetralins
Benzofurans
4-Ring Peri
5-Ring Peri
Al
3.4
2.4
.9
9.9
6.3
8.5
.1
4.3
3.2
42.1
1.0
7.0
.6
3.2
2.7
1.7
-
A2
4.6
1.6
.3
19.5
4.1
5.3
-
3.9
-
42.2
.5
6.9
-
1.8
2.0
1.0
-
Total
M
11.7
1.7
.4
17.6
2.6
3.6
-
2.6
-
39.5
.2
11.6
-
2.8
3.7
.9
-
(a)
lonization, percent
A4_ A5_ A6
4.2
2.8
.5
18.1
3.0
5.4
-
4.0
-
44.7
.2
8.3
.2
2.2
2.7
1.6
-
4.5
3.1
.6
16.4
4.0
6.1
-
4.4
-
47.0
.3
7.4
.2
2.1
1.1
1.7
.1
(a) Percent total ionization assumes equal sensitivity for all
components. Valid for comparison of similar samples.
Note: See note on Table 11.
55
-------
TABLE 13. HIGH RESOLUTION MASS SPECTROMETRIC RESULTS
Sample Identification
m/e Formula
228 C10H, .
18 12
252 C_-H10
20 12
254 ConHn/
20 14
256 C20H16
267 C.-H.-N
20 13
268 C21H16
278 C.-H,,
22 14
279 C01H10N
21 13
302 Cn,E..
?4 14
Al
A
•
A
•
A
G
A
O
A
G
L
0
A
r\
^^
A
O
A
o
A2
A
O
A
L.;
A
O
A
O
A
U
A
G
A
0
A
0
A
o
A3
/.
0
A
0
A
O
A
O
A
•u
A
O
A
O
A
O
A
o
A4
A
•
A
O
A
O
A
O
A
O
A
0
A
O
A
0
A
o
A5
A
9
A
G
A
0
A
O
A
G
A
O
A
O
A
O
A
o
A6
A
(._)
A
vJ
L
O
A
O
A
G
A
G
A
O
A
O
A
o
Sample Identification
A1F
A
*
A
a
A
0
A
©
A
O
a
A
O
A
&
A
C
D
A
O
A2F
A
G
A
O
A
O
A
O
A
O
B
A
O
A
C
A
G
n
A
O
A3F
A
e
A
G
A
6
A
©
A
O
B
A
G
A
©
A
O
a
A
G
A4F
A
O
A
O
A
G
A
G
A
O
•
A
O
A
O
A
0
o
A
G
A5F
A
O
A
O
A
O
A
O
A
O
•
A
O
A
O
A
O
a
A
O
O - Detected by HRMS (photoplate)
O - Not detected by HRMS (photoplate)
B - Detected by HRMS peak matching
D - Not detected by HRMS peak matching
A - Detected by HRMS (computer run) data system
A - Not detected by HRMS (computer run) data system
56
-------
TABLE 14. SSMS ANALYSIS OF FILTER SAMPLES
Element
LI
Be
F
Na
Kg
Al
P
S
K
Ca
'Jr. •
7i
y
Or
M.T
Te
(Jo
N.i.
Cii
i'.n
C;,
Ce
As
r,e
?,c
Kb
Sr
Y
Zr
Mb
Mo
Ru
Kh
ra
AS
Cd
A1F
1.5
<10
<0.15
<0.058
<0.058
<0.15
<0.15
A4F
<0. 00010
<0.52
<0.10
<7900
<260
<370
<2.6
<520
<1000
<1600
<1.0
<2.6
<0.52
<0.52
<0.52
26
<0.52
<1.6
0.52
I--2
'0.52
<] .0
<1.0
<0.52
<0.52
<] .6
±1.6
<0.052
<1.0
<0.052
<0.52
<0.26
<0.1
<0.1
<0.26
<0.26
A5F
<0.
<350
<0.
<25,000
<870
<1,200
<8.
<1,700
<3,500
<5.200
<3.
<3.
<1.
1.
-------
TABLE 14. SSHS ANALYSIS OF FILTER SAMPLES (Continued)
Ln
CO
Element
In
Sn
Sb
Te
I
Cs
Ba
La
Ce
Pr
Nd
Sm
Eu
Gd
Tb
Dy
Ho
Er
Tm
Yb
Lu
Hf
Ta
W
Re
Os
Ir
PC
Au
Hg
Tl
Pb
Bi
Th
U
B
A1F
<5
500
<10
<1
0.5
<1
<500
50
100
<1
<3
<2
<2
<2
<2
<2
<2
<2
<2
<2
<2
<1
<2
1
<1
<1
<1
<1
<1
<1
<1
1,000
15
<0.5
<0.5
<2,000
A2F
<5
<20
<10
<1
<0.5
<1
<500
<1
<1
<1
<3
<2
<2
<2
<2
<2
<2
<2
<2
<2
<2
<1
<2
<1
<1
<1
<1
<1
<1
<1
<1
<100
5
<0.5
<0.5
<2,000
A3F
US
<5
<20
<10
<1
<0.5
<1
<500
<1
<1
<1
<3
<2
<2
<2
<2
<2
<2
<2
<2
<2
<2
<1
<2
<1
<1
<1
<1
<1
<1
<1
<1
<100
<0.5
<0.5
<0.5
<2,000
A4F
<5
<20
<10
<1
<0.5
<1
<500
<1
<1
<1
<3
<2
<2
<2
<2
<2
<2
<2
<2
<2
<2
<1
<2
<1
<1
<1
<1
<1
<1
<1
<1
<100
<0.5
<0.5
<0.5
<2,000
ASF
<5
<20
<10
<1
<0.5
<1
<500
<1
<1
<1
<3
<2
<2
<2
<2
<2
<2
<2
<2
<2
<2
<1
<2
<1
<1
<1
<1
<1
<1
<1
<1
<100
<0.5
<0.5
<0.5
<2,000
A1F
<0.079
7.9
<0.16
<0.016
<0.0079
<0.016
<7.9
0.79
1.6
<0.016
<0.047
<0.031
<0.031
<0.031
<0.031
<0.031
<0.031
<0.031
<0.031
<0.031
<0.031
<0.016
<0.031
0.016
<0.016
<0.016
<0.016
<0.016
<0.016
<0.016
<0.016
16
0.24
<0.0079
<0.0079
<31
A2F
<0.16
<0.64
<0.32
<0.032
<0.016
<0.032
<16
<0.032
<0.032
<0.032
<0.095
<0.064
<0.064
<0.064
<0.064
<0.064
<0.064
<0.064
<0.064
<0.064
<0.064
<0.032
<0.064
<0.032
<0.032
<0.032
<0.032
<0.032
<0.032
<0.032
<0.032
<32
0.16
<0.016
<0.016
<64
A3F
Pg/g
<0.15
<0.58
<0.29
<0.029
<0.015
<0.029
<15
<0.029
<0.029
<0.029
<0.087
<0.058
<0.058
<0.058
<0.058
<0.058
<0.058
<0.058
<0.058
<0.058
<0.058
<0.029
<0.058
<0.029
<0.029
<0.029
<0.029
<0.029
<0.029
<0.029
<0.029
<29
<0.015
<0.015
<0.015
<58
A4F
<0.26
<1.0
<0.52
<0.052
<0.026
<0.052
<26
<0.052
<0.052
<0.052
<0.16
<0.10
<0.10
<0.10
<0.10
<0.10
<0.10
<0.10
<0.10
<0.10
<0.10
<0.052
<0.10
<0.052
<0.052
<0.052
<0.052
<0.052
<0.052
<0.052
<0.052
<52
<0.026
<0.026
<0.026
<100
ASF
<0.87
<3.5
<1. 7
<0.17
<0.087
<0.17
<87
<0.17
<0.17
<0.17
<0.52
<0.35
<0.35
<0.35
<0.35
<0.35
<0.35
<0.35
<0.35
<0.35
<0.35
<0.17
<0.35
<0.17
<0.17
<0.17
<0.17
<0.17
<0.17
<0.17
<0.17
<17
<0.087
<0.087
<0.087
<35
-------
TABLE 15. MASS SPECTROGRAPHIC ANALYSES OF COAL & COKE
(b)
Element
Li
Be
B
F
Na
Mg
Al
Si
P
S
Cl
K
Ca
Sc
Ti
V
Cr
Mn
Fe
Co
Ni
Cu
Zn
Ga
Ge
As
Se(a)
Br
Rb
Sr
Y
Zr
Nb
Mo
Ru
Rh
Pd(a)
Ag
Cd
In(a)
Sn
Coal
30
5
100
5
3000
5000
-v5%
VLO%
300
3000
100
500
2000
30
5000
100
100
300
^5%
20
20
30
30
30
2
10
<5
10
5
2000
100
200
2
10
<2
<1
<1
0.5
10
<0.5
1
Sample
Coke
Designation
Element
10 ppm Sb
3
100
5
3000
2000
^3%
VLO%
300
5000
200
2000
^2%
30
3000
50
20
300
M.%
20
100
30
50
20
20
20
<20
20
10
3000
100
200
10
20
<5
<1
<10
5
10
<2
100
Te
I
Cs
Ba
La
Ce
Pr
Nd
Sm
Eu
Gd
Tb
Dy
Ho
Er
Tm
Yb
Lu
Hf
Ta
W
Re
Os
Ir
Pt
Au
Hg
Tl
Pb
Bi
Th
U
Coal
<0.5
<1
<1
1
100
10
20
1
10
3
<1
1
<1
3
<0.5
<1
<1
<1
<2
<3
<2
<1
<1
<2
<3
<5
<2
<5
<2
1
<1
<3
<1
Coke
2 ppm
3
5
20
1000
300
200
100
200
10
3
10
1
10
1
5
<1
5
0.5
<1
<2
<2
<1
<5
<3
<5
<2
<5
<2
10
<1
5
3
(a) Memory from previous sample.
(b) Total weight of sample analyzed 100 mg.
59
-------
TABLE 16. OVEN NO. 41 "A" GAS PHASE, TEST NO. 1
Sample No. & (Hrs )
Component
Nitrogen
Oxygen
Argon
Carbon dioxide
Carbon monoxide
Hvdrogen
Methane
Acetylene & ethene
Ethane
Propene
Propane
1-Butene
Iso-Butane
n- Butane
Benzene
Toluene
Total low MW hydrocarbons
Water
a. Number of hours into the
* Less than 0.1 ppm.
1 (0.5)
% ppm
77.40
20.60
.95
.09
.10
.44
3200
200
200
50
20
8
2
6
100
62
3848
<•!
run is shown
Results Reported as Volume Percent & Volume ppm, Dry Bases
2 (1.5) 3 (3) 4 (6) 5 (10) 6 (13)
% ppm % ppm % ppm % ppm % ppm
77.8 79.4 78.0 78.0 78.0
20.9 19.4 20.8 20.9 20.9
.95 .99 .96 .96 .95
.08 .06 .05 .05 .05
.03 .004 .02 .01 .005
.16 <.l .08 .04 .03
900 60 400 100 31
50 3 200 2 .6
30 2 9 2 .4
9 .3 .4 .2 *
3 .1 .2 .1 *
1 ft * * *
*****
.5 * * * *
** ** ** ** *ft
ft* ft* *ft ft* ft*
993.5 65.4 609.6 104.3 32
.2 .2 .6 .3 .2
in parenthesis.
-------
TABLE 17. QVEN NO. 41 "A" GAS PHASE, TEST NO. 2
Results Reported as Volume Percent & Volume
Sample No. & (Hrsa)
Component
Nitrogen
Oxygen
Argon
Carbon dioxide
Carbon monoxide
Hydrogen
Methane
Acetylene & ethane
Ethane
Propane
Propane
1-Butene
Iso-Butane
n-Butane
Pentene
Pentane
Hexane
Benzene
Toluene
Carbonyl sulfide
Carbon disulfide
Total Low MW hydrocarbons
Water
a. Number of hours into the
* T.PRQ t-Vian 0.1 nr>m .
1 (0.5)
% ppm
76.3
20.5
.95
.24
.16
.81
.69
.16
.08
.02
84
67
12
32
7
16
3
3
A
10
1
9724
3.8
run is shown
2 (1.5)
% ppm
82.9
15.9
1.03
.13
.003
<.l
<2
A
A
A
A
A
A
A
A A
AA
AA
AA
AA
AA
AA
<2
0.5
in parenthesis.
3 (3)
% ppm
81.7
16.9
1.00
.16
.03
.07
30
60
50
60
30
0.3
A
A
AA
AA
AA
AA
AA
AA
AA
230
4.6
4 (6)
% ppm
80.4
18.3
.98
.17
.02
.11
.04
9
20
2
0.8
0.4
A
A
AA
AA
AA
AA
AA
AA
AA
432.2
2.2
ppm, Dry Bases
5 (10)
% ppm
80.5
18.2
.99
.17
.02
.07
.04
40
7
0.3
0.3
A
0.4
0.7
AA
AA
AA
AA
AA
AA
AA
448.7
0.5
6 (13)
% ppm
79.3
19.5
1.0
.07
<0.3
<.l
.01
0.1
A
A
A
A
A
A
AA
AA
AA
AA
AA
AA
AA
.1
0.4
Less than 50 ppm.
-------
analytical technique was used; the technique used for those components re-
ported as less than 50 ppm is not capable of detecing a specific quantity less
than 50 ppm. Sample No. 1 of Test No. 2 was concentrated which made it pos-
sible to detect much lower concentrations of C5 to Cy hydrocarbon and other
inorganic gases, but few additional compounds were detected.
The gas analysis results support the contention that much greater leakage
occurred around the oven door in Test No. 2 compared to Test No. 1. The data
for analysis of gaseous grab samples reported in Tables 16 and 17, showed
generally higher hydrocarbon contents for Test No. 2 samples compared to Test
No. 1 samples. The inorganic products of combustion (carbon dioxide, carbon
monoxide, and hydrogen) showed approximately twice the amount for Test No. 2.
During Test No. 2, the oxygen content of the gas dropped as low as 15.9 volume
percent (Sample 2) and gradually increased to 19.5 percent by the end of the
run. This low oxygen content (20.9 percent in normal air) was evident in
spite of air addition of about 68 m^/hr (40 cfm). This is in contrast with
Test No. 1 which showed a lower oxygen content of 19.4 percent for Sample 3
with Samples 2, 4, 5, and 6 having an oxygen content close to normal air. In
all cases of low oxygen (except Sample 1) for Test No. 2, the inert components
of the air (nitrogen and argon) increased above normal which indicates pro-
ducts formed from the oxygen consumption were high molecular weight compounds,
or the filter coated with tars and/or the adsorber were removing more of the
volatile components before they reached the flask sample.
During the running of Test No. 1, the total light hydrocarbon content de-
creased with time (as expected since the volatile components should be driven
off earlier in the run). Sample 3 was the only exception in this trend.
During Test No. 2, Sample 2 showed no hydrocarbons despite the high oxygen
consumption and Sample 6 showed less than Sample 4 and 5. It would appear
that some of the hydrocarbons were removed from these samples before reaching
the flasks, probably by adsorption on a caked filter or in the adsorption
bulb.
A partial list of compounds that could be detected by the gas analysis
techniques used but not found in the coke oven samples are listed below. The
lower limit of detection for these compounds for the concentrated sample
(Sample 1, Test 2) is 0.1 ppm, and 50 ppra for the other samples.
62
-------
Sulfur dioxide Hydrogen cyanide
Hydrogen sulfide Nitrogen oxides
Hydrogen chloride GI to C3 alcohols
Methyl chloride Acetone
Quantifying Gaseous Emissions
The quantifying of gaseous emissions cannot be done precisely for the
coke oven samples. The reason that such quantitation cannot be done is ap-
parent disagreement between gas flow data and the results of the gas analyses.
For example, field test data for Test No. 1 show that the air flow into
the hood (as measured at the rotameter) ranged from 15 to 35 scfm, and was
mostly between 27 and 35 scfm. Flows out of the hood were 41 to 47 scfm at
the orifice and 27 to 30 scfm at the Hi-Vol filter. By addition, the total
flow out was about 68 to 77 scfm. Subtracting the air flow into the hood, 27
to 35 scfm, would leave estimated leakage values of 33 to 50 scfm. Thus, the
leakage is calculated to be 100 to 150 percent of the inlet air flow.
However, examining the gas analyses, the high values for oxygen and
nitrogen (averaging 20.6 and 78.1, respectively) suggests that the flask
samples consisted of mostly air. This suggests that the leakage rate
through the oven door may have been much lower than the rate of air flow
into the hood via the compressor and any leakage at the hood seals.
The best estimate that can be made of gaseous emissions can be made by
assuming that the total gas flow out (via the orifice and the Hi-Vol sampler)
was measured fairly accurately. Then, by using the flow rate and the gas
analysis data, estimates of gaseous emissions can be made. This has been done
*
and is reported in Table 18 .
RESULTS OF BACTERIAL MUTAGENESIS ANALYSIS
Table 19 shows the results of the average of two experiments with the
tester strains on samples A1F, A3F, and ASF. In these assays, the number of
* Because the gas flow rate data was not taken simultaneously with the gas
samples, average gas flow rates for the entire test were used to calculate
gaseous emissions.
63
-------
TABLE 18. GASEOUS EMISSION ESTIMATES
Test No.
Total Estimated Gas
Flow Out, scfm
Total Estimated Gas
Flow Out, Nm^/min
Sample Number
Emission, mg/min
Carbon monoxide
Methane
Acetylene
Ethane
Propene
Propane
1-Butene
Iso-Butane
n-Butane
Benzene
Toluene
Pentene
Pentane
Hexane
Carbonyl sulfide
Carbon disulfide
1
73.0
2.06
1 2 3 A 5 6
2,392 717 96 478 239 120
4,382 1,233 82 548 137 42
462 115 7 462 51
513 77 . 5 23 5 1
180 32 1 1 1
75 11 <1 1 <1
38 5
10
30 2
787
488
2
72.2
2.04
1
3,790
9,361
3,659
2,033
712
313
318
59
157
20
41
98
22
51
6
23456
71 711 474 474 <1
<3 41 543 543 136
137 21 91 <1
127 51 18
213 7 1
112 3 1
1 2
2
3
-------
TABLE 19. EPA COKE OVEN SAMPLES
(5)
Compound
Metabolic
Activation
Negative control
Positive controls
g-Prop!olactone
2-Anthramine
Coke-oven sample A1F
Microli ters
of Sample
(in 50 ul of
DM SO)
Average of Two Experiments
0 ,05
0 .01
0 .Ul
0 .01
0 .05
0 . 10
0 .50
1.0
2. 5
5.0
0 .01
0 .05
0 . 10
0 .50
1.0
2.5
5.0
Histidine-Posi tive Revertants per
TA1535
22
19
201
9
16
19
14
23
31
32
10
28
22
24
16
48
54
TA1537
4
4
4
130
1
2
3
S
8
6
20
4
6
15
45
63
74
135
TA153S
10
13
15
1563
9
8
8
9
17
30
35
9
14
17
86
116
144
226
TA98
18
17
32
1371
14
17
17
42
81
96
117
21
26
46
153
243
373
415
Pla te
TA100
86
77
234
08
65
59
61
7-1
68
88
92
88
84
120
158
1GG
228
-------
TABLE 19. EPA COKE-OVEN SAMPLES (Continued)
Compound
Metabolic
Activation
Negative control
Positive controls
(3-Propiolactone
2-Anthramine
Coke-oven sample A3F
Microliters
of Sample
(in 50 p,l of
Average of Two Experiments
+
+
0 .05
0 .01
0 .01
0 .01 ,
0 .05
0 .10
0 .50
1.0
2.5
5.0
0 .01
0 .05
0 .10
0 .50
1.0
2.5
5.0
Histidine-Positive Revertants per Plate
TA1535
22
19
201
13
17
20
15
20
22
17
17
17
20
..19
19
34
40
TA1537
4
4
4
130
2
3
2
7
6
5
10
1
6
8
32
53
56
94
TA1538
10
13
15
1563
6
9
8
13
10
9
17
10
13
9
45
71
135
152
TA98
18
17
32
1371
22
17
24
28
37
27
69
25
27
37
91
159
183
293
TA100
86
77
234
64
68
76
83
81
79
74
95
71
86
117
144
146
156
-------
TABLE 19.
EPA COKE-OVEN SAMPLES(5) (Continued)
Compound
Metabolic
Activation
Negative control
Positive controls
g-Propiolactone
2-Anthramine
Coke-oven sample A5F
Microli ters
of Sample
(in 50 (j.1 of
DMSO)
Average of Two Experiments
0.05
0.01
0.01
o.oi'
0.05
0.10
0. 50
1.0
2.5
5.0
10.0
0.01
0.05
0. 10
0.50
1.0
2.5
5.0
10.0
Histidine-Positi ve Revertnnts per Plate
TA1535
25
24
213
22
30
34
27
25
19
17
22
13
13
21
17
16
•41
15
35
TA1537
6
3
4
199
6
7
5
6
9
7
15
toxic
7
5
10
13
30
33
54
toxic
TA1538
11
10
15
2029
6
5
12
9
17
10
23
29
9
7
13
21
34
49
85
99
TA98
23
29
33
1924
14
25
18
19
42
57
115
119
18
25
29
44
81
128
22G
172
TA100
98
104
263
91
68
71
82
. 74
60
93
71
103
101
7S
89
103
116
172
127
-------
revertants per plate was increased ten-fold over the negative control in
strains TA1537, TA1538, and TA98, and two-fold in strain TA100. Metabolic
activation (rat-liver homogenate) increased the number of mutants above that
observed without activation. The results suggest that coke oven samples A1F,
A3F, and A5F are moderately mutagenic and that a frame-shift mutation occurred
since TA1535, which is reverted by baser-pair substitution mutagens, was not
affected. Mutagenic activity was enhanced when the metabolic activitation
system was added to the test. The relative order of mutagenic activity was
A1F >A3F >A5F.
Table 20 presents the results of the experiments with ^. typhimurium on
samples Al, A3, A5, and A6. In the first experiment, 0.1 ml of each sample
was removed and dissolved in 0.9 ml of DMSO (50 yl = 5 yl of sample). Dilu-
tions were then made so that each dose was in 50 yl of DMSO. In the second
experiment, 0.4 ml of sample was dissolved in 1.6 ml DMSO (50 yl = 10 yl of
sample) for samples A5 and A6.
Samples Al, A3, and A5 were weakly mutagenic on strains TA1537, TA1538,
TA98, and TA100 when a metabolic activation system was added. In the first
experiment, the number of histidine revertants per plate increased 2-4 fold
over the control plates. In the second experiment, only strain TA100, with
metabolic activation, had a 2-fold increase in revertants. These samples
also appear to cause frame shift mutations. Sample A6 was not mutagenic in
either of the tests.
RESULTS OF MAMMALIAN CELL CYTOTOXICITY STUDIES
Due to the nature of the filter samples, it was not possible to quanti-
tate the dose or concentration of effluent material the cells were exposed to.
Control filters were not available for testing. The edge of the filter did
not appear to have been exposed to the coke oven emissions and was found not
to be toxic to the alveolar macrophages. The center of the filters A1F, A2F,
A3F, A4F, and ASF all caused a significant reduction in cellular viability.
In each case, between 5-30 mg of filter was utilized in the cytotoxicity
test. It was not possible to quantitate the cytotoxic response or compare
relative toxicity of these samples since the emissions were not evenly dis-
tributed on the filters and the amount of emissions present in the test was
no t known.
68
-------
TABLE 20. EPA COKE-OVEN SAMPLES
Compound
Metabolic
Activation
Negative control
Positive controls
g-Propiolactone
2-Anthramine
Coke-oven sample Al
Microli ters
of Sample
(in 50 ul of
DMSO)
0 .05
0 .01
0 .01
0 .01
0 .05
0 . 10
0 .50
1.0
5.0
0 .01
0 .05
0 . 10
0 . 50
1.0
5.0
Hlstidine-Positive Revertants per Plate
TA1535
22
19
329
28
19
31
16
23
24
11
7
14
15
14
12
TA1537
7
3
201
8
8
3
6
14
5
5
13
8
7
9
47
TA1538
15
6
16
1625
9
10
17
11
15
8
12
12
9
14
20
68
TA9S
29
34
29
1705
23
16
16
12
15
9
23
26
39
33
28
124
TA100
107
128
360
82
92
80
88
58
115
77
81
105
82
95
162
-------
TABLE 20. EPA COKE-OVEN SAMPLES (Continued)
Compound
Metabolic
Activation
Experiment 2
Negative control
Positive controls
AF-2
2-Anthramine
• Coke-oven sample Al
Microliters
of Sample
(in 50 ul of
DM SO)
Histidine-Positive Revertants per Plate
TA1535 TA1537 TA1538
TA98 TA100
0 .005
0 .01
0 .01
0.1
0.5
1.0
2.5
5.0
10.0
0.1
0.5
1.0
2.5
5.0
10.0
14
13
408
20
387
19
22
11
8
8
7
12
16
10
10
12
7
9
8
694
21
543
6
3
1
5
5
6
15
12
6
10
20
18
9
14
1915
17
876
16
12
7
4
11
6
20
18
32
26
18
20
17
22
1007
28
697
13
9
10
9
12
1
13
27
19
12
17
toxic
110
118
624
139
880
115
102
105
91
115
100
150
154
155
159
253
245
-------
TABLE 20. EPA COKE-OVEN SAMPLES (Continued)
Compound
Metabolic
Activation
Negative control
Positive controls
P-Propiolactone
2-Anthramine
Coke-oven sample A3
-f
Microliters
of Sample
(in 50 p,l of
DMSO)
Histidine-Posi ti.ve Revertants per Plate
0 .05
0 .01
0 .01
0 .01
0 .05
0 . 10
0 .50
1.0
5.0
0 .01
0 .05
0 .10
0 .50
1.0
5.0
TA1535
22
19
329
29
29
31
24
25
16
5
10
16
10
7
15
TA1537
7
3
201
1
2
6
6
5
5
7
5
5
10
7
10
TA1538
15
6
16
1625
8
9
15
12
0
9
10
12
13
14
16
28
TA98
29
34
29
1705
19
If)
1<>
20
12
18
2iJ
26
27
34
33
62
TA100
107
128
360
83
84
79
89
72
67
8-1
82
85
90
74
116
-------
TABLE 20. EPA COKE-OVEN SAMPLES (Continued)
Compound
Metabolic
Activation
Experiment 2
Negative control
Positive control
AF-2
2-Anthramine
Coke-oven sample A3
Microliters
of Sample
(in 50 p,!
. of DMSO)
0.005
0.01
0.01
0.1
0.5
1.0
2.5
5.0
10.0
0.1
0.5
1.0
2.5
5.0
10.0
Histidine-Positive Revertants per Plate
TA1535
14
13
408
20
387
27
18
20
12
11
7
15
13
8
7
10
15
TA1537
9
8
694
21
543
11
9
5
4
9
6
11
10
5
13
12
77
TA1538
9
14
1915
17
S7G
6
7
4
5
3
10
19
14
26
21
26
35
TA9S
1"
->2
1007
28
697
9
20
23
18
17
12
29
21
27
38
53
40
TA100
110
118
624
139
880
131
110
98
126
85
91
122
128
177
164
198
211
-------
TABLE 20. EPA COKE-OVEN SAMPLES (Continued)
Positive controls
p-Prop!olactone
2-Anthramine
Coke-oven sample A5
+
+
0.05
0.01
0.01
0.01
0.05
0.10
0.50
1.0
5.0
.01
.05
.10
.50
1.0
5.0
Histidine-Posi tive Revertcints per Plate
TA1535
22
19
329
14
17
23
27
19
14
14
5
10
7
18
7
TA1537
7
3
201
2
5
5
6
G
toxic
4
5
6
11
5
18
TA1538
15
6
17
1625
18
11
9
7
10
4
12
18
15
IS
22
40
TA98
20
34
29
1705
23
19
16
27
13
1-1
17
24
27
25
31
73
TA100
107
128
3fiO
77
78
88
82
7f>
68
86
80
88
62
9-1
123
-------
TABLE 20. EPA COKE-OVEN SAMPLES (Continued)
Compound
Metabolic
Activation
Microliters
of Sample
(in 50 ^1
of DMSO)
—i
.p-
•Experiment 2
•Negative control
Positive controls
AF-2
2-Anthramine
Coke-oven sample A5
0.005
0.01
0.01'
0.1
0.5
1.0
2.5
5.0
10.0
0.1
0.5
1.0
2.5
5.0
10.0
Histldine-Positive Revertants per Plate
TA1535
14
13
408
20
387
19
24
15
14
toxic
7
17
23
14
9
toxic
TA1537
9
8
694
21
543
16
G
11
5
toxic
13
11
11
7
4
toxic
TA1538
9
14
1915
17
876
17
14
11
13
toxic
14
18
25
19
18
toxic
TA9S
17
22
1007
28
697
30
11
10
toxic
15
14
29
toxic
TA100
110
118
62-1
139
880
161
142
139
96
toxic
131
191
210
193
107
toxi c
-------
TABLE 20. EPA COKE-OVEN SAMPLES (Continued)
Compound
Metabolic
Activation
Negative control
Positive controls
p-Propiolactone
2-Anthramine
Coke-oven sample A6
Microli ters
of Sample
(in 50 ,_,!
of DMSO)
4-
0 .05
0 .01
0 .01
0 .01
0 -05
0 .10
0 .50
1.0
5.0
0 .01
0 .05
0 .10
0 . 50
1.0
5.0
His tidine-Posi tive Revertcmts per Plate
TA1535
19
20
280
22
15
22
14
9
16
15
6
15
20
9
14
TA1537
5
8
375
5
5
6
3
4
2
8
4
5
7
4
12
TA1538
12
20
14
1545
3
8
10
10
10
9
7
9
6
16
3
13
TA98
22
31
32
16G1
22
28
30
20
18
6
27
16
28
30
24
22
TA100
137
104
315
IDS
77
102
73
SO
62
98
105
81
126
90
100
-------
TABLE 20. EPA COKE-OVEN SAMPLES (Continued)
Compound
Metabolic
Activation
Experiment 2
Negative control
Positive control
g-Propiolactone
2-Anthramine
Coke-oven sample A6
Mlcrpliters
o'f Sample
(in 50 p. 1
of DMSO) .
0 .05
0 .01
0 .01
t
0 .10
0 .50
1.0
2.5
5.0
10.0
0
0
1.
10
50
.0
2.5
5.0
10.0
Histidine-Ppsitive Revertants per Plnte
TA1535
27
28
97
20
17
15
14
17
18
29
27
31
. 15
28
26
TA1537
5
3
4
196
2
3
4
3
2
2
4
2
4
2
4
8
TA153S
7
15
14
2432
7
7
4
7
5
13
9
8
5
7
16
6
TA98
17
24
36
2143
26
16
18
12
16
9
31
17
22
15
27
20
TA100
8S
80
166
74
60
64
60
58
40
59
53
60
63
67
64
-------
DISCUSSION OF RESULTS
As would be expected during devolatilization of organic materials, such
as coal, the rate at which materials are emitted is high initially and then
decreases with time. Figure A and Table 5 shows that this phenomena was
observed at the coke plant - particulate emissions were greater than 1000
mg/min during the first hour and decreased to less than 1 mg/min near the
end of the coking cycle. Thus, even though filters and adsorbent columns
were changed more frequently during the early part of the test, the filters
and adsorbent columns used during the early part of the coking cycle were
much more heavily loaded than those used later in the cycle.
The large variation is mass emission between the two tests raises
questions concerning using results of a small number of runs to determine
emission factors or emission estimates for these sources.
Highly complex sample mixtures such as coke oven effluents are ex-
tremely difficult to analyze qualitatively, even more so quantitatively.
The Level-1 analytical strategy followed here has enabled a semi-quanti-
tative determination of the predominate classes of organic components
present in the samples obtained.
Infrared analysis of the Level-1 fractions failed to detect any
significant differences in the composition of the emissions during the
coking cycle. However, the LC fractions obtained in the Level-1 analysis
remain highly complex and, therefore, not easily amenable to detailed in-
frared analysis.
GC-MS analyses were conducted on representative fractions, as deter-
mined by infrared analysis; these analyses also proved difficult due to
the highly complex nature of the fractions yielding insufficient GC re-
solution. In many cases, the mass spectral data obtained for these unre-
solved GC components proved too complex for adequate interpretation.
Nevertheless, the predominate classes of organic components that were
identified by infrared analysis were subsequently identified by GC-MS.
HRMS further confirmed the presence of many components identified
by GC-MS. In addition, HRMS detected dibenzothiophenes and a C H N
component which were not observed using GC-MS.
77
-------
Bioassay results for the total sample are of limited value, since
infrared and GC-MS confirmed the presence of a large number of known
hazardous species such as polynuclear aromatic hydrocarbons present in
these samples. Since many of such compounds are mutagenic and/or cyto-
toxic, we would clearly expect the positive bioassay results obtained in
this study.
RECOMMENDATIONS
Further studies which involve the chemical analysis and bioassay of
coke emissions should include provisions for obtaining higher resolution
fractionation of the samples. This would include HPLC size and class
separation which would yield fractions which were more amenable to success-
ful infrared and mass spectral analyses.
Bioassay tests should be conducted after the LC or, preferably, HPLC
fractionation. They would then give an indication as to which fractions
merit further analysis, giving a more complete overview of such coke oven
emissions.
78
-------
REFERENCES
1. B. Commoner. Chemical Carcinogens in the Environment. Chapter 4 in
Identification and Analysis of Organic Pollutants in Water. Ed. by
L. H. Keith, Ann Arbor Science Publ., Ann Arbor, MI, 1976.
2. Jones, P. W., Graffeo, A. P., Detrick, R., Clarke, P. A., and
Jakobsen, R. J. Technical Manual for Analysis of Organic Materials
in Process Streams. EPA-600/2-76-072. U.S. Environmental Protection
Agency, Office of Research and Development, Washington, D.C.,
March, 1976.
3. Kier, L. D., Yamasaki, E., and Ames, B. N. Detection of Mutagenic
Activity in Cigarette Smoke Condensate. Proc. Nat. Acad. Sci. USA.
71:4159-4163, 1974.
4. Ames, B. N., McCann, J. and Yamasaki. Methods for Detecting Carcinogens
and Mutagens with the Salmonella/Mammalian-Microsome Mutagenicity Test.
Mut. Res., 31:347-364, 1975.
5. Simmon, V. F., Poole, D. In Vitro Microbiological Mutagenicity, Studies
of Seven Coke-Oven Samples. Stanford Research Institute Report No.
LSU-3493. Submitted to EPA, Research Triangle Park, NC, January 6, 1977,
6. Waters, M. D., et al. Environmental Research, 9:32-47, 1975.
79
-------
APPENDIX A
THERMAL ANALYSIS OF COKE OVEN DOORS
80
-------
SUMMARY REPORT
on
THERMAL ANALYSIS OF COKE OVEN DOORS
(Contract No. 68-02-1409, Task 16)
to
U.S. ENVIRONMENTAL PROTECTION AGENCY
June 12, 1975
by
G. R. Whitacre, S. E. Miller, and J. B. Purdy
BATTELLE
Columbus Laboratories
505 King Avenue
Columbus, Ohio 43201
Franklin County
81
-------
TAPiLli OF CONTENTS
Pa.qe
INTRODUCTION 1
SUMMARY AND CONCLUSIONS 4
DISCUSSION OF THE PROGRAM 8
Test Setup •. . . . . 8
Results of Test Runs 14
Theoretical Analysis 20
Calculation Procedure 20
Results 23
LIST OF TABLES
Table 1. Summary of Results of Heat Transfer Analysis 7
Table 2. Ambient Temperature in Portsmouth Area During Period
of Coke Oven Door Temperature Measurements (April 24-25,
1975) as Recorded by Huntington, West Virginia,
Weather Service 13
Table 3. Average Door Temperatures During Empty Oven Period and
Over a Complete 15-Hour Coking Cycle 15
LIST OF FIGURES
Figure 1. Sampling Hood Installed on Coke Oven No. 6 with Recorder
Showing on Top of Battery Above Hood 2
Figure 2. Thermocouple Locations, No. 6 Coke Oven Door 5
Figure 3. Closeup of Thermocouples No. 4 and 8 Showing Leads to
Recorder, and Section of Purge Air Pipe . 9
Figure 4. Closeup of Thermocouple No. 2 Showing Detail of Quick
Disconnect Plug in Thermocouple Lead 9
Figure 5. No. 6 Coke Oven Showing Eight Thermocouple Locations. . . 10
Figure 6. Top Plate of Sampling Hood Installed on No. 6 Coke Oven
Showing Thermocouple Leads and Purge Air Pipe in Place. . 10
Figure 7. Twelve Point Recorder Installed on Top of Battery Above
No. 6 Oven 12
Figure 8. Closeup of Temperature Recorder 12
Figure 9. Temperature Record of Four Thermocouples, Right Side of
No. 6 Coke Oven Door, Approximate 15-Hour Cycle,
April 24-25, 1975 16
82
-------
LIST OF FIGURES (Cont'd)
Figure 10. Temperature Record of Thermocouples No. 5, 7, and 8,
with Sampling Hood in Place Followinj.; No. 6 Oven
Push, April 25, 1975 19
Figure 11. Calculated Maximum Door Temperature Versus Purge
Gas Flow Rate 24
Figure 12. Calculated Maximum Hood Temperature Versus Purge
Gas Flow Rate 25
Figure 13. Total Heat Flux Dissipated and Division of Heat Flux
Between Purge and Ambient Air 26
83
-------
SUMMARY REPORT
on
THERMAL ANALYSIS OF COKE-OVEN DOORS
to
U.S. ENVIRONMENTAL PROTECTION AGENCY
from
BATTELLE
Columbus Laboratories
by
G. R. Whitacre, S. E. Miller, and J. B. Purdy
June 6, 1975
INTRODUCTION
As part of a study of concepts for minimizing emissions from
coke-oven doors (Contract No. 68-02-1439), a system was developed at
Battelle for measuring the emissions escaping from oven doors. Figure 1
shows the hood used to capture emissions from around the oven door.
In its original form, the hood had a natural aluminum shiny surface on
both inside and outside, and the outside surface was covered with a 1-inch
thickness of glass fiber backed on its outside surface with aluminum
foil. The system was tested on a Koppers oven at Empire-Detroit Steel
in Portsmouth, Ohio. The system performed satisfactorily. However, with
the hood installed on the oven buckstays, and natural convection over
the oven door cut off, the surface temperature of the door rose about
65 C (150 F) above a similar spot temperature on a nearby door. The
comparison door was charged 10 to 15 minutes later than the test door.
The temperature rise caused by installation of the hood was
undesirable for two reasons: (1) uncontrolled temperature rise might
cause damage to an oven door, and (2) excessive temperature above normal
could result in abnormal samples of emissions collected for analysis
through expansion of leakage cracks or through thermal change of leakage
products.
84
-------
FIGURE 1. SAMPLING HOOD INSTALLED ON
COKE OVEN
85
-------
It was proposed thai: certain control steps be taken to control
temperature rise, and that normal door temperature levels be established
by a test run over a complete normal coking cycle. Proposed control
steps were:
(1) Remove glass fiber insulation from the hood
(2) Increase the amount of purge air released beneath
the hood
(3) Blacken both sides of the hood to increase heat
radiation
(4) Remove hood for a cooling-off period when oven
door temperatures rise to a predetermined point.
In order to determine the effectiveness of the proposed modifi-
cations for control of door temperature, EPA authorized Task 16. This task
provided for measurement of door temperatures throughout a normal col:ing
cycle. It also authorized a heat-transfer analysis to provide the following
information:
(1) Quantification of heat flux, under normal operating
conditions, identifying the proportion of heat loss due
to convection and radiation
(2) On a theoretical basis, determine what the increase
in door temperature would be with the testing hood
installed in the original test condition. Compare
these calculations to conditions actually observed.
(3) Theoretically determine the temperature effect that
darkening the inside of the shield and removing the
insulation will have without increasing the purge-gas
flow rate.
(4) Show the temperature effect, including the above modi-
fications, of increasing the purge-gas flow rate
throughout the range of 30 cfm to 90 cfm. A graph of
gas flow rate versus total heat dissipation is expected.
(5) On a theoretical basis, show the temperature effect of
intermittent removal of the hood indicating approximate
rate of door temperature decrease after removal.
86
-------
A meeting to 1:0.view 1 h J r; work war, held on June 3 with Mcsrs.
Richard Rovang and Norman Plakt: of I1!PA. Two suggestions were made by
Mr. Pinks that might Further improve heat transfer and reduce door
temperatures when the sampling hood is in place. Thus, a sampling run
might be made over an entire coking cycle with door temperatures remaining
close to those of a normal coking cycle. The suggestions were:
(1) Install small fins (about 1/4 inch high) in a vertical
position on the inside and possibly on the outside of
the hood sections to aid in heat transfer
(2) Provide air circulation in back of the hood by use of
several fans or blowers.
It was agreed that a brief study of these approaches would be
made prior to any extensive sampling program, to determine the degree of
improvement they may provide and to help in design and placement of
the fins. Also, it was agreed that this brief investigation should be
made a preliminary part of any sampling program.
Door temperatures were measured by eight thermocouples through-
out one 15-hour coking cycle. Figure 2 is a schematic illustration of the
door size and the relative locations of the thermocouples.
The No. 7 thermocouple, 267 cm (105 inches) from the bottom
of the door, registered the highest peak and average temperatures. The
peak temperature recorded was about 213 C (416 F) and the average
temperature value for No. 7 thermocouple was 206 C (403 F).
Following one complete coking cycle with recorded temperatures,
the sampling hood was installed for a period of 2 hours and 20 minutes.
As shown by Figure 1, the insulation had been removed from the hood and
its surfaces were, painted black. Purge air was fed under the hood at
a rate of 1.7 m /rain (60 ft /m.i.n) . Temperature on No. 7 thermocouple rosi
from 204 C (400 F) to 241 C (465 F), an increase rate of 0.26 C/min (0.46
87
-------
?f> -in.
••']
7/4
•
R a
33 cm
"13 in'J
JL-, OH
»f/ irui
_>. . .
vD-D
#1
O -H
oo m
ro r-i
E c
O -i-l
CM
OO
»
E C
O -H
/
E C
O -H
CM OO
CM
-------
F/min) . The hood wan then ruuoved .-md in 1 hour the temperature had de-
creased to 271 C (430 10, a rate of decrease oC 0.67 C/min (0.58 F/min).
From Lcmpera l.vires recorded during the test run on the coke
oven door, a theoretical heat analysis was made. It consisted of:
heat loss due to convection and radiation, door temperatures with various
hood emissivities and purge air flows, and the door temperatures with
insulation on the hood.
Table 1 summarizes the temperatures and heat flux calculated
3 3
under conditions of 0.85, 1.70, and 2.5 m /min (30, 60, and 90 ft /min) air
flow in combination with emissivities (e) of 0.9 and 0.2 on both sides of
the hood. Calculations were also made for an air flow of 1.7 rrr/min
3
(60 ft /min] with emissivities of 0.9 and 0.2 applied to the inside and
outside surfaces of the hood respectively, and then 0.2 and 0.9 applied
to the inside and outside surfaces, respectively.
3
The door temperature calculated with 1.7 m /min purge air flow
and with both sides of the hood painted black (e = 0.9) was 248 C (478 F);
this temperature as measured during the test run after 2 hours and 20 minutes
was 241 C (465 F). At 241 C the hood was removed with the temperature
still slightly on the rise. Therefore, the calculated door temperature
of 248 C (.478 F) appears to be reasonable.
Calculations show that the total heat dissipated by convection
3 3
levels off at about 2.5 m /min (90 ft /min) of purge air. Calculated
door temperature with insulation on the outside surface was 399 C (750 F).
Over the purge air range of 0.85-2.5 m /min (30-90 ft /min), calculated
door temperatures vary from 343 C (650 F) to about 306 C (582 F) with
shiny aluminum surface on both sides of the hood; by painting only the
outside surface black, door temperature is lowered only about 1.7 C (3 F).
Painting the inside surface black and leaving the outside shiny drops
the door temperature 54 C (97 F), and painting both inside and outside
surfaces black lowcrr; door temperature 66 C (119 F) .
Hood temperature is less affected than door temperature by
cmiss ivity change duo to p.'iinting both hood surfaces black. At a purge -
3 3
air flow of 0.85 m /min (30 ft /min), hood temperature is 141 C (286 F)
89
-------
TABLE i. SUMMARY OF RESULTS OF HEAT TRANSFER ANALYSIS
Air Flew.
m3/-in
and
ft" /-in sin eout
O.S5 0.9 0.9
30
1.7 0.9 0.9
60
2.5 0.9 0.9
50
O.S5 0.2 0.2
30
1.7 0.2 0.2
60
2.5 0.2 0.2
90
1.7 0.9 0.2
60
1.7 0.2 0.9
60
With Insulation on Hood:
0.85 0.2 0.2
30
^door ,
Degrees ,
C and F
261.0
502.0
248.0
478.0
239.6
463.3
343.1
649.5
314.0
597.0
306.0
582.0
260.0
500.0
201.0
594.0
393.0
740.0
^air ,
Degrees ,
C and F
101.8
215.2
69.1
156.5
54.2
129.5
128.0
263.0
85.4
185.8
64.0
148.0
77.0
170.5
79.0
175.0
172.5
342.5
^hood,
Degrees ,
C and F
127.0
260.5
134.1
237.3
106.2
223.1
141.4
286.5
242.0
222.0
148.1
298.5
89.7
193.5
284.2
543.5
^insulation Heat
Outside
Surface q to
Degrees , purge
C and F air
1078.7
341.9
1389.1
440.3
1556.0
493.2
1394.2
441.9
1770.9
561.3
1902.8
603.1
1574.3
499.0
1632.1
517.3
75.3 1917.3
167.5 607.7
pi,... /w/m and
FlaX I.BU;/ft2hr
q to
ambient
air
1687.3
534.8
1422.0
450.7
1276.5
40 4 . 6
1111.2
352.2
826.6
262.0
722.8
229.1
1191.6
377.7
969.8
307.4
425.3
134.8
Total q
2766.3
876.8
2808.6
£90,2
2S34.4
S9S.4
2505.4
794.1
2598.1
823.5
2624.6
831.9
2769.8
877o9
2603.5
825.2
2345.4
743.4
c = emissivity factor;
Cubic feet x 0.0283 = cubic
T = temperature, C and F; q = heat flux, w/m2 and Btu/ft2hr.
meters; Heat flux, Btu/ft2hr x 3.155 = w/m2 .
-------
\v.r:ui.s ] !'.7 C ('.'OH I-') v.-licn belli riidc.s oC Ihc hood .-ire pai.ntcd. An purge
air flow is increased, hood temperature under the two cnu'ssivity condi-
3 3
ti.onr; becomes more nearly equal and at 2.5 ni /mi n (90 ft /inin) it is equal
tinder the two purge-air flow conditions.
The tests and theoretical analyses show that removal of hood
insulation and increased emissivity from paintiny both sides of the
hood black are effective1, in lowering door temperature. Although normal
door temperature is not maintained by these steps alone, theoretical
3
analysis shows that at 1.7 m /rnin purge-air flow door temperature should
be about 248 C 0478 F). The actual test run of over 2 hours made on
April 25 verifies this calculated temperature; a maximum temperature of
240 C (464 F) was recorded with indications that the temperature was
starting to stabilize. A run of 3 hours' duration under these test condi-
tions could undoubtedly be completed, and the test just completed shows
that hood removal results in a rapid lowering of door temperature. The
increase in door temperature of 36 to 42 C (65 to 75 F) that would
probably result from the type of test run just completed, would not, In
the opinion of Battelle chemistry analytical authorities, have a harmful
effect on samples obtained. Metallurgical experts advise that this
much temperature rise in this temperature range will not be harmful to
the oven door.
It is possible that further improved heat dissipation such as
from fins on the hood may enable a run to be made over a complete cycle
without removing the hood.
DISCUSSION OF THE PROGRAM
Test Setup
Eight thermocouples, numbered and located as shown in Figure 2,
were installed on the door of No. 6 coke oven at the Koppcrs battery of
Empire-Detroit, Portsmouth, Ohio. Figures 3, 4, and 5 show details of
91
-------
FIGURE 3. CLOSEUP OF THERMOCOUPLES NO. 4 AND 8 SHOWING LEADS TO
RECORDER, AND SECTION OF PURGE AIR PIPE (UPPER RIGHT CORNER)
FIGURE 4. CLOSEUP OF THERMOCOUPLE NO. 2 SHOWING DETAIL OF
QUICK "DISCONNECT PLUG IN THERMOCOUPLE LEAD
-------
FIGURE 5. NO. 6 COKE OVEN SHOWING
EIGHT THERMOCOUPLE LOCATIONS
FIGURE 6. TOP PLATE OF
SAMPLING HOOD
INSTALLED ON NO. 6
COKE OVEN SHOWING
THERMOCOUPLE LEADS
AND PURGE AIR PIPE
IN PLACE
93
-------
thermocouple Jnr; tallal'ions. The couples were 20 gage Chromel-alumel
wires joined by a welded beau. At the point on the door where temperature
was to be read, a small hole about 0.3 cm (1/8 inch) deep was drilled.
A thermocouple was Inserted in the hole and the bead was peencd with a
center punch so as to fill the hole and make the couple secure. To
further secure the couple a spot several inches from the couple bead
was cleaned by abrasion and a stainless steel foil strap was welded over
the insulated couple wires. In Figures 3, 4, and 5, the cleaned spots
can be seen as bright strips. Best detail appears in Figure 4 which
also shows in good detail the disconnect plug that enabled thermcouple
lead wires to be quickly pulled up out of the way when the door had to
be removed to push the oven.
Figure 6 shows thermocouple leads passing up through the gas
exit in the sampling-hood top plate. Figure 7 is a general view of the
12 point Honeywell recorder in position on top of the battery. Figure 8
is a closeup of the recorder that shows the general trace of temperatures
being recorded. Since only eight temperature points were being measured
in the door, the re;;;.:rJning four (Nos. 9, 10, 11, and 12) were shorted
in the back of the recorder; thus, they recorded air temperature in the
back of the recorder.
A record of ambient temperature in the area was obtained from
the Huntington, West Virginia,weather service, located 30 miles east
and 23 miles south of the Portsmouth, Ohio,coke-oven battery. Table 2
summarizes these temperatures from 1:00 p.m., April 24, through 1:00 p.m.
on April 25.
The thermocouple installation was completed on No. 6 oven door
the morning of April 24. The push on No. 6 oven was completed at 2:07 p.m.
and the door was back in place at 2:U8 p.m. Thermocouple leads were
reconnected and recording was started at 2:15 p.m., April 24. The oven
was charged at 2:28 p.m. and tin.1; charge was pushed at 5:35 a.m. on the
morning of April 25.
Although door temperature measurements with the hood in place
were not a previously plannr-.d part of this pros'.r.'iw, time and ccjuipiiicnt
94
-------
FIGURE 7. TWELVE POINT RECORDER INSTALLED OK
TOP OF BATTERY ABOVE NO. 6 OVEK
FIGURE 8. CLOSEUP OF TEMPERATURE RECORDER
-------
TABLE 2. ANCIENT TEMPERATURE IN PORTSMOUTH AREA DURING
PERIOD OF COKE-OVEN DOOR TEMPERATURE MEASUREMENTS
(APRIL 24-25, 1975) AS RECORDED BY IIUNTINGTON,
WEST VIRGINIA, WEATHER SERVICE(a)
Time
Thursday, April 24 1:00 p.m.
2:00 p.m.
3:00 p.m.
4:00 p.m.
5:00 p.m.
6:00 p.m.
7:00 p.m.
8:00 p.m.
9;00 p.m.
10:00 p.m.
11:00 p.m.
12:00 a.m.
Friday, April 25 1:00 a.m.
2:00 a.m.
3:00 a.m.
4:00 a.m.
5:00 a.m.
6:00 a.m.
7:00 a.m.
8:00 a.m.
9:00 a.m.
10:00 a.m.
11:00 a.m.
12:00 p.m.
1:00 p.m.
Temperature,
C
17.2
17.2
17.2
17.2
17.8
17.2
17.2
16.7
16.7
16.7
15.6
14.4
13.3
13.3
12.8
12.8
13.3
13.3
13.9
14.4
15.0
16.1
16.1
15.6
17.2
Temperature,
F
63
63
63
63
64
63
63
62
62
62
60
58
56
56
55
55
56
56
57
58
59
61
61
60
63
(a) Located 30 miles east and 23 miles south of coke oven battery.
96
-------
were available so several hours of temperature measurements were made
under sampling-hood conditions.
No high-volume sampler and filters were used in these test
runs. Since only door temperatures were of interest, purge air and
door emissions merely exhausted through the hood top plate to the atmo-
sphere.
Results of Test Runs
Table 3 presents a summary of average door-temperature values,
without the sampling hood, during a 15-minute period before charging and
after the door had been put on. It also gives the average temperature
value for each of the eight thermocouples during the complete coking cycle
for the April 24-April 25, 15-hour period.
Figure 9, plotted over the span of three pages, shows the curves
for temperature plots of couples No. 5, 6, 7, and 8. Only temperatures
from the four couples on the right side of the door were plotted because
the right side of the door was consistently the hottest side. Figure 9
shows that the hottest spot was at couple No. 7. It started out at about
199 C (390 F), gradually increased to a maximum of about 213 C (416 F)
after 5 hours, leveled off to 209 C (409 F) for about 1.5 hours and then
very gradually decreased to around 199 C (390 F) at the end of the coking
cycle. Couple No. 6 leveled off and started to decline in temperature
more quickly than No. 7. No. 5 and No. 8 declined near the start of the
cycle and leveled off after about 3.5 hours.
Figure 10 is a plot of temperatures from thermocouples 5, 7,
and 8 with the sampling hood installed and immediately after the hood is
removed. It shows temperatures with the hood in position for 2 hours
and 20 minutes, and the temperature decrease over a period of an hour
after the hood was removed. The plot of No. 7 couple in Figure 10
indicates that temperature falls more rapidly when the hood is removed
than it rises when the hood is installed. Theoretical analysis discussed
later shows a good correlation between the 241 C (465 F) maximum temperature
recorded with the hood in place and the calculated value of 248 C (478 F)
at a purge air rate of 1.7 m /min (60 ft /min).
97
-------
TABLE 3. AVERAGE DOOR TEMPERATURES WITHOUT HOOD
DURING EMPTY OVEN PERIOD AND OVER A
COMPLETE 15-HOUR COKING CYCLE
Period
Covered
Average Temperature, C and F
Thermocouple Number
12345678
4/24/75, 2:15
2:30 p.m.:
Empty Oven
181
357
181
357
189
373
182
360
182
360
191
375
201
394
171
340
4/24, 2:28 p.m.-
4/25, 5:35 a.m.:
Complete 15-Hour
Coking Cycle
170
339
172
342
195
383
174
345
171
340
186
368
206
403
162
324
98
-------
o
0)
tc
o
o
o
Q.
E
c;
H
-• 204
2 -400-
I
177;
-350'-
E
Q.
OO
rg
Thermocouple number designation: -J-5
•iH 0
ifti +
I
2:00 p.m.
^
...p.c , A,^ . • -3 y>n-o-oa^ .
. : • •"'"" V , ^-^
•*-..
149
.^,nn
-t" ' ^-4 ' •
April 24, 1975 ; . ' I » 1 I I
i, . , ; , '...i — ; : — : — ! — i — i : _j — :,. , ,•.
3:00 p.m.
4:00 p.m.
5:00 p.m.
6:00 p.m.
,r , 1 L.
7:00 p.m.
j
8:00 p.m.
FIGURE 9. TEMPERATURE RECORD OF FOUR THERMOCOUPLES,
WITHOUT HOOD, RIGHT SIDE OF NO. 6 COKE-OVEN
DOOR, APPROXIMATE 15-HOUR CYCLE, APRIL 24-25, 1975
-------
O
o
T)
C
I
23 2[
450-
'•
\
: • •
!
i
i
i
i
.. .. i „ . . ;
: ' i
Thermocouple number de:
; : ; ;
: i : : •
05
V
a
tc
c
rt
o
»j
~
4_)
rj
t-i
CJ
a.
e
a
C-!
SJ
o
0
a
i
i
i
i
204!. t>-c"'
400r
i
t
F
j
I
|
; •>J-
177!
350-
i *
: /\
8:00 p.m.
->^v--
----Cl -•-
• '
*-.£-" x0'^
•" ••^•'•--rt V
_A
c
r;--J-! 1 I- : I I r-*-^-<-*-T-r-* *
April 25, 1975
i
9:00 p.m.
10:00 p.m.
11:00 p.m.
12:00 a.m.
1:00 a.m.
2:00 a.m.
FIGURE 9. (Continued)
-------
ll.
C
o
v>
O
Jj
CJ
O
o
C,
5-1
O
O
Thermocouple number designation:
in
n
•o
O
1 .
in
m
C
•H
CO
60
CO
T3
O
to
Ll
CO
"u
C
(U
°
v6 a
^7 "o
?* 8 +
2:00 a.m.
3:00 a.m.
25, 1975
' 4:00 a.m.
\,
,v;
! 1
5:00 a.m.
••'.!/ ' :
\ '
t
6:00 a.m.
FIGURE 9. (Continued)
-------
-3
C
ct
CJ
t-
C)
360
500-
2
"o
232
450
1/1
o
ij
O
*r-<
Q.
C
O
O
Q
204 ,-' o- o-- d
400-
/ " ' •
177
350.__j. _;.__-_-S — ^
^ j r
• -^r
149-
rtr\n: — - .
6:00 a.m.
,.0
Thermocouple number designation: ->:5
; "°
i i—i
.— i
w
AJ
: 1 , W
i • ! , C
• ' . *H
"O
O
1 : O
. | . . . : _ . J=
CO
C
•r-l
r— 1
CL
E
^tj
/>- OT
/ X
/ . *
/ V
r-.
0 :..;'. i
1-1 1
u-j : :
'• •
i-i 1 '
•H | ' ;
E ' '
o
o ' . *^r*
\Q - ^S**~
•'
•£ ^
•H ^C
5 ~s"
/ '
/
i ^;-
j~y~~~
•': — 0''
T3
>
O
E
-------
Th co ret.ic.-11 An •'ij_)^ji_s^
A parametric analysis was conducted to determine the effectiveness
of potential modifications to reduce the temperature of the coke-oven door
with the hood in place. However, first it was necessary to determine the
thermal condition of the door under normal operating conditions. Using the
temperature data obtained in the field test, the proportion of heat loss
due to natural convection and radiation and the thermal resistance of the
door were determined. Quasi-steady-state, one-dimensional, thermal analyses
were then performed with the hood in place to determine the predicted
increase in door temperature with the original testing hood. Finally, the
effect on temperature and heat flux was determined parametrically over a
range of purge-gas flows and emissivities with the insulation removed. A
theoretical analysis of the transient behavior of the door following hood
removal was not conducted due to potential complexity. However, the transient
behavior was determined by an additional test conducted following the normal
operation cycle and is described elsewhere in this report.
Calculation Procedure
In order to keep the analysis essentially one-dimensional, it was
necessary to chose a location on the door at which the calculations would be
performed. (An alternate procedure would be to use averages over the entire
door, but this was rejected as combining and masking too many different
effects such as purge gas temperature rise and end effects.) The location used
was thermocouple location No. 7 which is 2.67 meters (8-3/4 feet) from the
bottom of the door. This was chosen because it was farthest from end effects
and away from latches which act as cooling fins and further complicate the
analysis. Thus, it was the maximum temperature recorded. In addition, it
had the best defined flow because the upward flow at that location had been
unobstructed for approximately 1.2 meters (4 feet) below.
In order to be quasi-stcady-state,the average of the temperature
at location No. 7 over the entire cycle, 206 C (403 F), was used as the
103
-------
basis for the normal operation calculations. From this temperature and
an average ambient temperature, 15.6 C (60 F) , the normal operation heat
fluxes at this location were calculated using accepted heat-transfer correla-
f ~\
tion from McAdams^ ' . The natural convection heat flux from a vertical
2 -2-1
plate was 1379 w/m (437 Btu ft hr~ ) and was in the turbulent regime so
that the characteristic length cancelled out of the correlation. The
radiation using a door emissivity of 0.6 and a view factor of 1.0 was
1562 w/m2 (495 Btu ft"2 hr"1) . This gave a total heat flux of 2941 w/m
(932 Btu ft"2 hr"1).
To calculate the effect of the hood installation and other
changes, it was assumed that the average inside wall surface temperature
would not be affected by the changes since it is essentially fixed by the
process. To determine this steady-state inside-surface temperature, the one-
dimensional thermal resistance of the door was calculated from available
thermal conductivity data. The door consisted of 34.3 cm (13-1/2 in.) of
first quality silica firebrick with a thermal conductivity of 190 w/m
(1.10 Btu ft" hr F ) plus 5.1 cm (2 in.) of H.B. #28 castable refrac-
tory with a conductivity of 0.38 w/m K (0.22 Btu ft" hr" F" ) for a total
2-1 2_ -1
thermal resistance of 0.3146 m Kw (1.785 hr ft F Btu ). This gives a
calculated inside-surface temperature of 1131 C (2067 F) which was held
fixed for all subsequent calculations. This wall surface temperature is
(c)
within the range of 1010 to 1149 C (1850 to 2100 F) given by Preston
for typical coking operation.
For all calculation with the hood in place, steady-state heat
balances were performed on the hood and at the door surface. The heat
transferred through the door was convected to the purge air and radiated
to the hood. The heat received by the hood was radiated and convected to
the outside ambient air as well as some additional convection to the purge
air. Then insulation was added on the hood, inside and outside hood
(a) McAdams, W. H., Heat Transmission, 3rd Edition, McGraw-Hill, New York,
1954.
(b) w = watts; m = meter; K = degrees Kelvin.
(c) Preston, E., "Carbonization of Coal and Gas Making", Mechanical Engineers'
Handbook. Edited by L. S. Marks, 5th Edition, McGraw-Hill, 1951, p 817.
104
-------
temperatures were calculated and the temperature difference determined by
the heat flux to the ambient and the thermal resistance of the insulation.
The purge air temperature rise was also calculated using total heat convected
to the purge-air from the inlet to location No. 7. The temperature differ-
ences at location 7 were used in this calculation. Although the door wall,
hood, and air temperatures all decrease in the direction of the inlet at
the bottom, the temperature differences should remain nearly constant, so
use of location 7 temperature differences is a reasonable approximation.
Iterations were performed on the various temperatures until all the heat
balances and other conditions were satisfied.
The convection to the purge air flow was a combination of natural
and forced convection. The Reynolds number for the purge air flow in the
unobstructed passage was 1800 to 6100 which is in the transition regime
from laminar to turbulent flow. Therefore, the transition correlation of
(a)
Kroll given in McAdams was used. However, the natural convection heat
transfer coefficient for vertical surfaces was always much larger than the
forced convection coefficient. For combined natural and forced convection
in laminar flow, Martinelli and Boelter derived and confirmed a relation-
ship of the form
h , . , = 3/ h3 . + h]
combined natural — forced
This relationship was used to combine the convection in the present calculations
although the flow was in the transition regime. Because the natural convection
h was larger than the forced h, the effect on h directly of increasing purge
air flow was quite small. However, the increased air flow decreased the air
temperature rise which did increase the temperature difference and the natural
convection h so that there was an effect of purge-gas flow rate.
(a) McAdams, loc. cit., p. 240.
(b) Mart.inclli, R. C. and Boelter, L.M.K., University of California Publication
in Engineering, Vol. 5, No. 2, Berkeley, 1942, pp. 23-58.
105
-------
Results
Table 1 shows a summary of results of the various cases calculated.
The emissivity of 0.2 is representative of the aluminum hood or aluminum foil
used with the original hood.. The emissivity of 0.9 is representative of
black paint on the aluminum hood. All the results apply to thermocouple loca-
tion No. 7 and, therefore, are approximately the maximum to be expected. The
total heat flux as well as the division between the heat transferred to the
purge air and the ambient are also given. All the various terms in the
heat balances were calculated and are available in the detailed calculation
sheets.
Figure 11 shows the calculated maximum door temperature versus
purge-gas flow rate for the various cases. It does not appear possible to
decrease the door temperature to the normal operating temperature of 260 C
(403 F) by increasing the purge-gas flow rate or the emissivity of the
surfaces. Painting the outside only has little effect on decreasing the
door temperature. On the other hand, painting the inside only has a
significant effect. The predicted value of 248 C (478 F) for both sides
painted and a purge-gas flow rate of 1.7 m /min (60 cfm) is close to the
value of 241 C (465 F) measured in the test of the No. 7 thermocouple
which had not yet quite reached its quasi-steady-state value after about
2 hours with the hood on.
Figure 12 shows the calculated maximum hood temperature versus
purge-gas flow rate for the same cases. With equal emissivity the maximum
hood temperature is not very dependent on purge-gas flow rate or the
emissivity itself. However, there are significant changes with unequal
emissivities.
Figure 13 shows the total heat flux dissipated and the division
between heat transfer to the purge air and to the ambient for the cases with equal
emissivities of 0.9 only. The total heat flux has decreased very little from the
normal operation value of 2953 w/m^ (936 Btu ft'^hr"^-) and is nearly indepen-
dent of purge gas flow rate.
106
-------
-o
c
rt
o
CO
v
-------
288
o
CO
-------
CM
0)
sc
900;.—•
2524
800 I
2209
700
>,
i—I
0)
u 1893'
o. 600
01
0)
i-i
x;
CM
JJ
U-l
3
4J
03
1578
500
1262
400
947 .
300:
631
200
316
100
0
.849
30
Total Heat Diasipcitc-d
e Inside = 0.9
e Outside = 0.9
1.132 1.415 1.698 1.981 2.264
40 50 60 70 80
Purge Gas Flow Rate, m3/min and ft3/min
2.547
90
FIGURE 13. TOTAL HEAT FLUX DISSIPATED AND DIVISION OF
HEAT FLUX BETWEEN PURGE AND AMBIENT AIR
109
-------
APPENDIX B
HEAT TRANSFER ANALYSIS FOR HOOD
110
-------
SUMMARY REPORT
on
EFFECT OF FINS ON COKE-OVEN
HOOD HEAT TRANSFER
to
ENVIRONMENTAL PROTECTION AGENCY
October 9, 1975
by
G. R. Whitacre
BATTELLE
Columbus Laboratories
505 King Avenue
Columbus, Ohio 43201
111
-------
SUMMARY REPORT
on
EFFECT OF FINS ON COKE-OVEN
HOOD HEAT TRANSFER
to
ENVIRONMENTAL PROTECTION AGENCY
from
BATTELLE
Columbus Laboratories
by
G. R. Whitacre
October 9, 1975
INTRODUCTION
To permit sampling of emissions from coke-oven doors, a hood must
be used to contain the emissions. The addition of a hood in the vicinity of
the coke-oven door serves to reduce heat loss from the door and, thus, in-
creases the door temperature. It has been proposed that fins might be added
to the hood to increase heat loss from the hood and thereby, reduce the
temperature of the hood and of the door. A thermal analysis was conducted to
determine the effectiveness of hood fins and whether they should be placed
on the inside or outside of the hood. In addition, optimum fin spacing and
heights were determined.
CALCULATION PROCEDURE
The analytical thermal analysis techniques described in a
previous report"' were used to predict steady state maximum door and
••'"Whitacre, G.R., Miller, S.E., and Purdy, J.B., "Thermal Analysis of
Coke Oven Doors", Summary Report to EPA, June 12, 1975.
112
-------
hood temperatures. Only the modification and additions necessary to in-
clude the effect of the fins and optimize their spacing and height are
discussed here. There are two limiting analysis procedures available to
handle natural convection fin heat transfer. One limit applies to small
fins which are spaced far enough apart that the boundary layers which form
around one fin do not interact with the boundary layers from the adjoining
fins. In addition, it is assumed (for vertical fins) that the heat transfer
coefficient on the fin surfaces including the corners is the same as the
coefficient on the plane base surface. This heat transfer coefficient is
determined from a Grashof number correlation based on vertical hood height.
The only effect of the fins is to increase the surface area for heat transfer.
However, the fin surface area is multiplied by the fin effectiveness due
to heat conduction in the fin before this product is added to the
remaining base area to obtain the effective total area for heat transfer.
For aluminum fins with the same gage as the hood (16) the fin effectiveness
*
was over 99 percent for all the cases analyzed. Therefore, fin thickness
could be changed with practically no effect on fin effectiveness.
For this limiting case of small fins with no interaction, the
natural convection is in the turbulent regime on both sides of the hood.
Although the purge flow on the inside of the hood enclosure is provided at
3 3
a rate of 1.7 m /min (60 ft /min), natural convection is still the dominant
mode of heat transfer and the procedure used in combining the forced and
natural convection is described in the previous report. Essentially the
heat transfer coefficient is determined by the turbulent natural convection
but the purge air temperature rise affects the heat transfer by reducing
the available temperature difference. The analysis procedure used to
determine the fin effectiveness for this limiting case is identical to
**
that described in the previous report. The surface area on both the in-
side and outside of the hood was increased in steps representing the
*
Fin effectiveness is defined as the ratio of the amount of heat dissipated
by the fin. surface to that which would be dissipated if the fin surface
were held at the temperature of the fin base.
**
Ibid.
113
-------
addition of fins, and new wall, hood, and air temperatures were determined
by performing iterative calculations until the energy (heat balance)
equations were satisfied.
The other limiting case for vertical natural convection fins is
to consider the flow between the fins as flow in an enclosure with con-
stant wall temperature. (For aluminum fins the high fin effectiveness
is equivalent to a constant wall temperature.) This case has been well
analyzed both experimentally and analytically by Elenbaas for both parallel
* **
plates and for enclosures of various shapes and aspect ratios. In
addition Elenbaas derives an expression to optimize the fin spacing for
maximum heat transfer as a function of aspect ratio of the enclosure.
This optimum results from a decrease in the heat transfer coefficient as
the fin spacing decreases in competition with an increase in the surface
area with decreased spacing. Treating the space between the fins as an
enclosure, the optimum fin spacing is given by
, .
Grm Pr = 0.72 (YRe) /J
where
r = hydraulic radius with optimum spacing
1 = vertical distance
3
Gr = Grashof number based on r
ro m
Pr = Prandtl nun.i. :r of fluid.
The flow resistance parameter Y Re is a function of the aspect ratio and
varies from 24 for an infinite parallel plate to 14.225 for a rectangular
enclosure with an aspect ratio of 1.0. The complete variation with aspect
ratio is given in the second Elenbaas paper.
In calculating the effect of fins in the enclosure limit,
various aspect ratios were selected and the opt 'mum spacing calculated by
* Elenbaas, W., "Heat Dissipation of Parallel Plates by Free Convection",
Physica, Vol. 9, No. 1, January, 1942, pp 1-28.
** Elenbaas, W., "The Dissipation of Heat by Free Convection. The Inner
Surface of Vertical Tubes of Different Shapes of Cross-Section", Physica
Vol. 9, No. 9, September, 1942, pp 865-873.
114
-------
the Elenbaas formula. In order to simulate a full enclosure for this
limiting case, the aspect ratio was assumed to be equal to the ratio of
twice the fin height to the distance between the fins. Physically, this
assumption is equivalent to the actual fins with no leakage flow in or out
past the plane of the fin tips. Also, the hydraulic radius as used by
Elenbaas is twice the cross-sectional area divided by the wetted perimeter.
This hydraulic radius is twice the value normally used in classical treat-
ments of flow in enclosures.
After the optimum fin spacing was calculated for the various
aspect ratios the heat-transfer coefficient was determined from the Elenbaas
correlation at the optimum point. The correlation is
hr 1 ,
Nu = -r21- = 0-385 (Y Re) '
m k
where
Nu = is the Nusselt number at the optimum spacing
h = is the heat transfer coefficient and
k = is the thermal conductivity of the fluid.
The heat transfer rate obtained in this manner is based on the temperature
difference between the fin and the inlet air temperature and the effect
of the air temperature rise in the flow direction is included in the
derivation and verification of the correlation. Because of the small
spacing between fins and the interaction between boundary layers the flow
remains in the laminar flow regime so the correlation and all the results
calculated for the enclosure limit are for laminar flow.
Using the heat-transfer coefficients and optimum spacings cal-
culated for the various aspect ratios, the heat balance iterations previously
formulated were repeated with the various fin arrangements on both sides
of the hood. For the fins on the outside, the procedure was essentially
the same as described, but several changes were necessary for the fins placed
on the inside. Because it was assumed that there was no flow through the
plane of the fin tips the inside flow was divided into two parts. In order
115
-------
to calculate the flow between the fins the theoretical solution of Bodoia
and Osterle was used since it included a prediction of the amount of flow
and also showed good agreement with the experimental data of Elenbaas.
The heat transfer by convection from the door surface was determined using
only the difference between the total purge air flow and flow between the
fin since only this flow was available to cool the door directly.
The fins also affect the radiation heat transfer by increasing
the effective emissivity of the surface through the cavity effect. This
increase in emissivity was included using the data from Sparrow and Gregg.
RESULTS
Table 1 shows a summary of results for the various fin configura-
tions considered. All of these results were obtained for an assumed purge
3 3
air flow of 1.7 m /min (60 ft /min) and a hood emissivity of 0.9 (painted
black) on both sides. The first line in the table repeats the results
obtained previously (June 12, 1975 Summary Report) for the case without
fins. Where an aspect ratio is given, the calculation procedure was based
on the enclosure technique and the fin spacing is the optimized spacing
(second limiting case discussed). Where no aspect ratio appears, no
interaction was assumed and the area ratio was used to increase the appropriate
heat transfer surface (first limiting case discussed). The total heat flux
as well as the division between the heat transferred to the purge air and
the ambient environment are also given. All the various terms in the heat
balances were calculated and balanced by iteration and are available in
detailed calculation sheets.
Figure 1 shows the calculated maximum door temperatures for
various fin configurations. Somewhat surprisingly, the fins on the inside
Bodoia, J. R., and Osterle, J. F., "The Development of Free Convection
Between Heated Vertical Plates", Journal of Heat Transfer, Trans. ASME,
Series C, Vol. 84, No. 1, February, 1962, pp 40-44.
.,„,, Sparrow, E. M., and Gregs, J. L. , "Radiant Emission from a Parallel-
Walled Groove", Journal of Heat Transfer, Trans ASME, Series C, Vol. 84,
No. 1, February, 1962, pp 40-44.
116
-------
TABLE 1. SUMMARY OF RESULTS OF HEAT TRANSFER ANALYSIS
Total Area Fins
B.ise Area Location
1 . 0 None
1.5 Inside
2.0 Inside
3.0 Inside
2.40 Inside
2.86 Inside
4.68 Inside
1.5 Outside
2.0 Outside
3.0 Outside
1.95 Outside
2.86 Outside
4.68 Outside
Spacing
cm
in
-
2.54
1.0
2.54
1.0
2.54
1.0
2.36
0.93
2.18
0.86
1.93
0.76
2.54
1.0
2.54
1.0
2.54
1.0
?..72
1.07
2.16
0.85
1.91
0.75
Height
cm
in
-
0.64
0.25
1.27
0.5
2.54
1.0
1.65
0.65
2.03
0.80
3.56
1.40
0.64
0.25
1.27
0.5
2.54
1.0
1.30
0.51
2.01
0.79
3.51
1.38
Enclosure Door
Aspect C
Ratio F
247.8
478.0
246.9
476.4
246.1
475.0
245.3
473.6
1.5 244.0
471.2
2.0 244.3
471.8
4.0 248.6
479.5
244.3
471.7
241.7
467.0
238.1
460.5
1.0 251.7
485.0
2.0 248.3
479.0
4.0 244.2
471.5
Door
C
F
114.1
237.3
111.3
232.3
109.6
229.2
106.9
224.4
105.0
221.0
102.6
216.6
96.7
206.0
102.9
217.3
94.1
201.4
81.7
179.0
124.2
255.5
208.5
240.5
102.8
217.0
f w/m
lBtu/ft Hr
q to
PUTRC Air
1389.1
440.3
1429.5
433.1
1464.6
464.2
1520.1
481.8
1564.2
495.8
1607.8
509.6
1701.2
539.2
1331.4
422.0
1290.1
408.9
1234.9
391.4
1430.8
453.5
1395.8
442.4
1330.8
121.8
q to
Ambient
1122.0
450.7
1381.6
437.9
1347.5
427.1
1295.8
410.7
1257.3
398.5
1211.5
384.0
1103.0
349.6
1486.6
471.2
1537.4
487.3
1607.5
509.5
1365.2
432.7
1407.8
446.2
1487.6
471.5
Total
q
2811.1
891.0
2811.1
891.0
2812.1
891.3
2815.9
892.5
2821.5
894.3
2819.3
893.6
2804.2
888.8
2818.0
892.3
2827.5
896.2
2842.4
900.9
2796.0
886.2
2803.6
888.6
2818.4
893.3
-------
00
460
2.0 3.0 4.0
AT/AB, Ratio of Total Area to Base Area
FIGURE I. CALCULATED MAXIMUM DOOR TEMPERATURE FOR VARIOUS FIN CONFIGURATIONS
-------
had very little effect on the maximum door temperature. In fact, practically
all of the slight decrease shown for fins on the inside with turbulent
natural convection, i.e., the no interaction limit, was a result of the in-
crease in effective emissivity caused by the cavity effect. As the ratio
of the areas increases, the actual door temperature should switch from the
no interaction limit curve to the laminar flow enclosure curve. The poor
performance of fins placed on the inside of the hood can be explained by
a combination of factors. First, the heat transfer from the door to the
hood is radiation dominated and convection is only secondary. Second, the
inside fins do increase the heat transfer to the purge air and decrease
the hood temperature as shown in Table 1. However, because of the interactions
between the hood, wall, and purge air heat balances there is a detrimental
effect of the lower hood temperature which balances the heat transfer gain
and can even result in an increase in door temperature. This gives a
significantly lower heat transfer rate to the ambient both by radiation
and natural convection from the outside of the lower temperature
hood. In addition, the increased flow and heat transferred in the en-
closure between fins reduces the available purge air flow and resulting
natural convection directly from the door to the purge air. Thus, it
appears that fins placed on the inside of the hood are not worth the cost
and effort and may even be detrimental.
The results for fins placed on the outside of the hood are quite
different, however. For the no interaction assumption with turbulent
natural convection, there is a beneficial effect of decreasing door
temperature with increasing fin area. There are no major counterbalancing
effects although the heat flux to the purge air does decrease due to the
lower wall and hood temperatures. This effect of outside fins with tur-
bulent natural convection does not come close to reducing the door tempera-
ture to its normal operating temperature of 206 C (403 F) however.
When the limiting analysis of a laminar flow enclosure is applied
to the outside fin case there is a significant shift in results with an
increase in wall temperature (compared with the no fin case, A /A = 1)
for area ratios less than about 3. This is caused by a much lower heat-
transfer coefficient inside the enclosure with laminar natural convection
119
-------
than exists on a vertical surface with turbulent natural convection. This
lower coefficient is not offset by the increased surface area until an
area ratio of 3 is reached. This decrease in heat-transfer coefficient
would be a real effect as the. interaction of the flow between fins in-
creases. The actual wall temperature curve probably leaves the turbulent
curve somewhere between an area ratio of 2 to 2-1/2 and should blend into
the laminar flow enclosure curve somewhere between an area ratio of 3 to
4 as shown by the shaded area in Figure 1. This transition is only an
estimate so it appears that short fins with an area ratio no greater than
2.0 to minimize flow interaction and maintain turbulent flow are most
desirable. The optimum spacing for an area ratio of 2 is approximately
2.54 cm (1.0 in) resulting in a fin height of 1.27 cm (0.5 in).
*
A recent paper by Pnueli indicates that an improvement in heat
transfer can be achieved by a moderate inclination from vertical for natural
convection fins. The initial decrease in length for a number of fins
terminating in the vertical edge increases the heat-transfer coefficient
enough to more than offset the cosine law decrease in Grashof number. How-
-1/4
ever, this effect is based on a heat-transfer coefficient varying with 1
which is true for laminar natural convection. For turbulent natural con-
vection we are trying to insure in the present case the coefficient is
independent of L and thus there is no gain (only the cosine law loss)
by inclining the fins.
CONCLUSIONS
The conclusions from the heat transfer analysis on the effect
of fins on coke oven door heat transfer are as follows.
Vertical fins placed on the inside of the sampling hood are
not effective in reducing door temperature.
Short vertical fins placed on the outside of the sampling hood
can reduce the door temperature by a maximum of approximately 6 C out of
•
Pnueli, David, "Optimization of Inclined Convcctive Fins", Journal of Hont
Transfer, Trans ASME, Series C, Vol. 96, No. 4, November, 1974, pp 545-547,
120
-------
a total temperature rise of 42 C from the normal operating condition as
indicated in the June 12, 1975, Summary Report.
Deep vertical fins on the outside can actually cause an increase
in the door temperature because of the trapping of flow between fins and
the resulting transition to laminar flow.
If fins are used on the outside of the hood they should be
approximately 1.27 cm (0.5 in) high on 2.54 cm (1.0 in) spacing and can be
the same gage as the hood.
For this case, inclination of the fins from vertical provides
no increase in heat transfer.
121
-------
APPENDIX C
FIELD DATA AND CALIBRATION SHEETS
(Copies of field data and calibration sheets may be
obtained from R.C. McCrillis MD-62, US Environmental
Protection Agency, Research Triangle Park, N.C. 27711)
122
-------
APPENDIX D
RECONSTRUCTED GAS CHROMATOGRAMS. ELECTRON IMPACT IONIZATION
MASS SPECTRA. AND METHANE IONIZATION MASS SPECTRA OF
EMISSIONS FROM COKE OVEN DOORS
123
-------
1 «"•"'
d Q
3 :
Q T
fi
S
»-*'
B
ff
10 20 3(1 10 SO GO 70 80 30 ICCl
I i I __(_ I ^t __?._ I I I
-- Tetra hydronaphthalene
Methyl naphthalenes
-- Biphenyl/acenaphthrene
-- Biphenyl/acenaphthrene
Fluorene/methyl acenaphthrene/methyl biphenyl
-- Anthracene
Methyl anthracenes
Fluoranthene
Pyrene
FIGURE D-l. RECONSTRUCTED GAS CHROMATOGRAM; GC-MS ANALYSIS OF
COKE OVEN DOOR EMISSIONS, SAMPLE A5-2
124
-------
SPECTRUM 3 - 2
OVEN R5-2
£P
»c^^
UJO
-------
SPECTRUM 8 - 7
8
«r«
8.
OVEN flS-2
10 S3 60 70 80 90 100 110 120 130 140 150 160 170 1«J
M/E
126
-------
SPECTRUM 19-22
8-
8-
|Pu
18.
!8-
8.
COKE OVEN RS-2
in.|m.(l).MIM,|i>ll|lnl|inl(l)).|MMl.|l|)ln.|
10 SO 60 70 80 90
M/E
.|M,....,...| I
103 110 120 133 110 193 163 170 183 19C
127
-------
SPECTRUM
- 30
bRu
I*.
8.
8.
COKEWEH RS-2
tfr
*
40 S3 60 70 80 90
M/E
100 110 120 130 1
-------
SPECTRUM 35-34
OVEN RS-2
a.
8.
^p
fr f^» —
UJO
EG? -
bR.
**_
8_
10 SO 60 70 80 90 100 110 120 130 HO 150 160 170 180
h/E
129
-------
SPECTHUM 16 - 18
8
Rj
8
jP
IS
OVEN PE-2
bR
I*
Is
•
8
o
Ttf
•'""I
"'I " '• "I' '•
93 63 70
M/E
80 90 103 110 120 130 110 1S0 160 170 180 19C
130
-------
SPECTRUM SB - SB
COKE OVEN flS-2
^C CD
BC ^^
UJO
ft*® -
bR_
J—Ih
•'""I
S3 60 70 80 90
M/E
100 110 120 130 110 ISO 160 170 180 13?
131
-------
SPECTRUM 64 - 62
8
COKE OVEN flS-2
£P
HL^^ —•
UJO
QW _
bR.
fcj
»8.
8.
o
*3 SB 60 70 80 90 100 110 120 130 1*3 ISO 160 170 180 19C
M/E
132
-------
SPEOHLM 83-81
8
a.
CONE OVEN
•W S3 63 7CT 83 93 133 113 123 133 140 1SB 168 173 183 13C
M/E
133
-------
SPECTRUM 71 -
COKE OVEN RS-2
VJFJ
CL ^^ ^--
ujo
8 -
bR.
fc.
8.
10 SO 60 70 80 30 100 110 120 130 110 ISO 160 170 180 13G
M/E
' 134
-------
SPECTRUM 123 - 127
8
OVEN FS-2
tfr
**T
10 50 60 70 80 90 100 110 120 130 110 153 160 170 163 13C
M/E
135
-------
Mi;'
il
7J '•
"r "
f. ;:-
'
c:
rMethyl pyrenes
Methyl fluoranthenes
i Dimethyl pyrenes
' Dimethyl fluoranthenes
Benz(c)phenanthrene
=:= -- Chrysene
•) Methyl chrysenes
J Methyl benzanthracenes
=-.-=— -- Benzofluoranthenes
~^r^=^ -- Benzo pyrenes
Methyl benzpyrenes
Methyl benzfluoranthenes
-- Indenopyrene
-- Benroperylene
" Dibenzoanthracene
-- Coronene
FIGURE D-2. RECONSTRUCTED GAS CHROMATOGRAM;
GC-MS ANALYSIS OF COKE OVEN DOOR
EMISSIONS, SAMPLE A5F-2
136
-------
SPECTRUM 211 - 205
g
8.
CDKE OVEN El FS-F2
4r"4
f
\\
JH,
TTTJ.TTTJTTTMrTr.|,»'|"" iTM™Mp'1|nT1|rTTTrT*Y"N""|"'M"
-------
SPECTRH 223 - 216
8
8.
CCKE OVEX El FE-F2
feS.
8J
o
70 80 90 100 110 120 130 ItO 150 160 170 180 190 200 210
M/E
138
-------
SPECTHJM H9 - 115
CCKE 0\e< El PS-F2
IP-
I*.
!«..
•
s^
o
1
SO 60 70 80 90 100 110 120 130 110 ISO 160 170 1:
M/E
139
-------
9PECTRJM 239 - 237
8
CCKE ON/EN El flS-F2
:P
18-
fe8
i«
s
o
iWTTWTtTTmTTtTI
,,„..,„„,
170 180 190 200 210 220 230
M/E
140
-------
SPECTFLM 212 - 215
8
8.
COKE CMEN El PG-F2
O
CD _
18.
170 180 190 200 210 220 Z30
M/E
141
-------
SPECTFLM 219 - 2SO
COKE OVEN El F&-FZ
fc8_
«,
8.
8-
170 180 193 203 210 220 230 2*3 2SS
M/E
142
-------
SPECTPLM 270-267
3
CCKEOVEN El R5-f2
190 200 210 220 230 210 293 26O
M/E
143
-------
SPECTRUM 281 - 277
8
COKE OVEN El FS-F2
I
180 198 238 218 223 233 213
M/E
144
-------
283-287
OVEN El
oj _
w _j
o
13.
.^^^..-T^TT^^rV^J. ,,-, ,^TTT,.TTT^
200 210 220 230 2-W 250 260
r
-------
SPECTRUM 382 - 297
g
COKE OVEN El RS-F2
iP-
bs.
tt!»
8_j
210 223 233 2*3 233 260 270
M/E
146
-------
SPECTRUM 329 - 32S
8
CCKEOVEN El R5-F2
JlUp^pJll
u
210 220 230 2-K3 233 260 270
M/E
147
-------
SPECTPLh 338 -
3
CDKE CNEN El flS-FZ
A
210 220 230 2-H3 2SO 260 270 280
M/E
148
-------
SPECTRA 391 -
g
8.
COME OVEN El RS-F2
2SO 263 273 280 290 303
M/E
149
-------
-383
8
8-
CC3KEOVEN El RS-F2
i8_
•
R.
250 273 288 233 303 310
M/E
150
-------
a
#}
aj
8]
»-«~
8-
-
2
»-••
2
s:
»-•"
a
-f
81
10 20 30 10 LO- GO TO CO 90 ICO
J J I I I I I I
C3 Phenol
n
f C2 Benzaldehydes
-- Methyl allyl phenol
Methyl naphthols
-- Hydroxy fluorene
-- Benz enthrone
FIGURE D-3. RECONSTRUCTED GAS CHROMATOCRAM; GC-MS ANALYSIS
OF COKE OVEN DOOR EMISSIONS, SAMPLE A1F-4
151
-------
SPECIRH 18 - IS
g
El OKE OVEN
8-
8.
i^
fC
.
0
V J ••
o
0 _
I
Illl 1 I ll I Jl
Mll|llll|)ltl|l
20 30
M/E
1,
9
ill j !
9 60
,J
70
|j
N«i|ili1
80 £
Lll
ii|iui|ii i
0 IOC
||
I ii|ini
1 HC
|
\
|
1
..,....,
120 13
|IIJMII|IIII llll|l
0 110 ISO
152
-------
SPECTRUM JI - 8
g
*"* -1
8.
8.
•J Ws ^
jk
8.
0
o .
1
ci UJTVC.
i 1
20 30
H/E
uvcn r
III,
10 3
111 K.
I ll
9 6C
i
1
J
II.
70
tliimnlt
80 9
ill ,1
0 IOC
I
J
1 lilt
110 1
1 ....... ll
20 130
!ll
110
153
-------
SPECTRUM 21-20
160
8
EI COKE CWEN R1F-C
18.
8.
,,M,,,m
20 ao 10 so
M/E
£0 70 60 90 103 110 120 130 110 ISO
154
-------
SPECTRUM 27 - 25
El COKE CNQ4
160
20
30 10
M/E
90 100 110 120 130 110 ISO
155
-------
SPECTRUM 32-33
EI CCWE OVEN fllF-IE
j
^
20 30 10 SO 60 70 60 90 100 110 120 130 110 ISO
M/E
156
-------
SPECTRUM 31-33
El COKE OVEN R1F-1E
|.
1|.
|.
l|
|.
.|.
|ll..|IM.|l
ISO
20
30 tO
M/E
SO 60 70 80 90 100 110 120 130 110
157
-------
SPECTRUM -12-39
8
8.
EI OKE OVEN RIP-IE
dtW
mt
iWtf
J|mnJ
Antmittl
itntt
iiH|miitl«ijiiii/TTrnriTT tit(iniMi i|rM'irTii[iiiiimi|iii iriri|iiii|iiii|'iii|>
-------
SPECTRUM
8
8.
EI COC OVEN RIF-IE
'-
8-
mml
it
miW
l|i.ii|....|i.. j • ••|i»i| . • ....|.... ii..|.ni|. |.i..|i,,i| .1 |tn,|i. i , Tjrrr,,! i
-------
8
SPECTHLM S-t - SI
EI COKE OVEN R1F--1E
|U|L
v,,,,|,.,,r
110 120 133 1K3 ISO
160
-------
SPECTFLM 56 - GO
El COKE OVEN
IB.
18.
o
M J
i. I
JfL
110 120 130 110 ISO 160
M/C
161
-------
SPECTRLM 60-80
8-
o
03 -.
:£
IS.
El CCKE OVEN
te°
10
O
iS?J
110 120 130 110 ISO 160 170
M/C-
162
-------
SPECTFLM 78-83
8
El COKE OVEN fllF-^E
8.
18.
v-rrn 11 T | tr»irTri|r TI n» i^t *«•!
MO ISO 160 170 180 130 2OO 210
K/P
.163
-------
SPECTFLM 87 - 127
El COKE OWEN
ISO 160 170 ISO 190
M/C-
-------
its -
El COKE 0\€N
8-1
I I!
ISO 160 170 180 190 200 210 220 230 210
M/P
165
-------
SPECTRM 166 - 163
El CCME OVEN fllF-»E
bo
VJ ^
So
2ZO
i
4
2Sfl 2©3 270 280
166
-------
0 10 20 S-'J «;j o
0
Q
oJj
SJ
ej
^**i
B-j
»^"^
^*
Q.
».*•*
0-
**»"•
a.
***
pA
Ej
»*"
0
v*'
a
3"
*-*"
a
»-*'
0
8
ro
*-»
&-.
Hi
Ei"
C3 '/O W £3 jet)
Naphthyl isocyanides
-- Carbazole
Methyl carbazoles
FIGURE D-4. RECONSTRUCTED GAS CHROMATOGRAM; GC-MS
. ANALYSIS OF COKE OVEN DOOR EMISSION,
SAMPLE A1-3
167
-------
SPECTRUM 16-13
8
COKE CMEN El FU-3E
I*-
R.
JJ
20 ao 10 SB eo TO so so 100 no 120
H/E
168
-------
98-102
B
OHECWEN El m-3E
V • ^ ^
8.
8.
!*-
18.
i
£
o
0 ..
tTmrnmlni
'"'""I
1 1 '
„.„! iMiiiiiiuiiiiiiinmnmniini mnmiiniimiiiniiM mim
....... . ..J....r...r...|....r...,....|...., ...,..„,.„.,.„.,....,.. .,....,
20 30 10 SO
H/E
70 80 90 100 110 120 130 110 ISO 160
169
-------
its -103
8
CCKEOWENEl W-3E
8.
_
.*-
is.
8.
o
e _
HHiiHmHin
H/E
170
-------
SPECTRUM 199 -201
8
«p-»
8.
COKE OVEN El A1-3E
18.
£-
1
*r
20 30 10 SB 60 70 80 90 100 110 120 130 110 150 160 170
H/E
171
-------
TECHNICAL REPORT DATA
(Please read Inductions on ilic reverse before completing)
I. REPORT NO.
EPA-600/2-77-213
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Sampling and Analysis of Coke-oven Door Emissions
5. REPORT DATE
October 1977
6. PERFORMING ORGANIZATION CODE
7. AUTHORIS)
8. PERFORMING ORGANIZATION REPORT NO
R.E.Barrett, W.L.Margard, J.B.Purdy, and
P E
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Batte lie-Columbus Laboratories
505 King Avenue
Columbus, Ohio 43201
10. PROGRAM ELEMENT NO.
1AB604C
11. CONTRACT/GRANT NO.
68-02-1409, Task 34
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Task Final: 8/75-8/77
14. SPONSORING AGENCY CODE
EPA/600/13
,5. SUPPLEMENTARY NOTES iERL_RTp pr0ject officer for this
Mail Drop 62, 919/541-2733.
report is Robert C. McCrillis,
. ABSTRACT
Tne rep0rt gives results of extensive tests of selected fractions of samples
of emissions generated by leakage from a coke oven door during a 16-hour coking
cycle. The tests included: particulate emissions determination; trace metal analyses;
gas analyses; organic analyses by IR spectroscopy , GC-MS, TLC, and HRMS on
entire samples or on LC fractions of the samples; and bioassay analyses of bacterial
mutagenesis and mammalian cell cytotoxicity. The particulate mass emission deter-
mination showed that coke oven emissions can vary considerably from cycle to cycle.
The bioassay analyses confirmed that the samples were mutagenic, as implied by
the chemical analyses. A sealed hood was fabricated to fit over the coke oven door,
so that gases leaking past the door during the coking cycle would be contained and
representative samples could be obtained. Additional criteria for the hood included
not severely altering the normal door leakage and not interfering with coke oven
operation. Initial tests of one hood design suggested modifications which were incor-
porated into the final design. The final hood was used for conducting two sampling
runs at an operating coke oven.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS C. COSATI i K'ld/CrO'.i|)
Air Pollution
Coking
Ovens
Leakage
Sampling
Analyzing
Organic Chemistry
Bioassay
Air Pollution Control
Stationary Sources
Coke Oven Doors
Particulate
13B
13H
13A
07C
06A
iy. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS i/iiij Kcporn
Unclassified
21. NO. ul' I'AdES
17S
20. SECURITY CLASS (This
Unclassified
22.
EPA Form 2220-1 (9-73)
172
------- |