EPA-650/2-74-062
July 1974
Environmental Protection Technology Series
-------
EPA-650/2-74-062
COKE OVEN CHARGING
EMISSION CONTROL
TEST PROGRAM-
-VOLUME I
by
R. W. Bee, G. Erskine, R. B. Shaller,
R. W. Spewak, A. Wallo, III, andW. L. Wheaton
The Mitre Corporation
Westgate Research Park
McLean, Virginia 22101
ROAPNo. 21AFF-004
Program Element No. 1AB013
Interagency Agreement F192628-71-C-002
Contract 68-02-0650
EPA Project Officer: R. V. Hendriks
Control Systems Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
July 1974
-------
This report has been reviewed by the Environmental Protection Agency
and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Agency,
nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
11
-------
CONTENTS
Abstract
List of Figures
Acknowledgements
Sections
Volume I
I Conclusions
Identification of Coke Oven Charging Emissions
Comparison of Emissions from Wilputte and AISI/EPA
Charging Systems
Technology of Coke Oven Emissions Measurement
Feasibility of a Coke Oven Compliance Monitoring System
II Introduction
III Approach
General
Continuous Measurement System
Sensors and Probes
Manual Sampling
Optical Program
Field Test Program
IV Data Handling Procedures
Input Data
Mathematical Treatment of Data
V Results
Comparison of Continuous Measurement and Manual Gas
Sampling Methods
Results of Particle Size Analysis
Statistical Evaluation of Size Distribution Data
Comparison of Particulate Emissions from Wilputte and
AISI/EPA Car
Mass Emissions
Supplementary Analyses
Optical System Program Results
Page
111
vii
xviii
1
1
3
5
8
11
13
13
13
20
29
33
38
48
48
58
81
81
96
109
109
111
117
126
Select Project Documentation List
163
-------
CONTENTS (CONTINUED)
Page
Volume II*
VI Appendices A-l
A - Volume Flow Curves and Associated Data A-l
B - Manual Sampling B-l
C - Computer Programming Documentation C-l
D - Leaking Coke Oven Doors D-l
E - Particulate Size Distribution E-l
* Volume II of this document is available through EPA's Control
Systems Laboratory.
vi
-------
FIGURES
Number Page
1 Wilputte Larry Car Emissions Guide 15
2 Wilputte Larry Car Emissions Guide 16
3 AISI/EPA Larry Car Emissions Guide 17
4 AISI/EPA Larry Car Emissions Guide 18
5 Wilputte Emissions Guide Sensor Duct 21
6 Wilputte Stack and AISI/EPA Emissions Guide Sensor Duct 22
7 Cross-Over Bridge Shed Area 23
8 Gas Handling System 26
9 Simplified System Component Configuration 35
10 Film Format 37
11 Test Program Schedule 39
12 Daily Test Schedule Outline 41
13 Continuous Sensor Analog Voltage Print-Out 49
14 Gas Analyzer Strip Chart Data 52
15 Gas Temperature and Pressure Strip Chart Data 53
16 Final Data Print-Out in Engineering Units 55
17 Typical Volume Flow Graph Versus Charging Procedures 60
i
18 Coke Oven Particulate Distribution 103
19 Log Normal Plot of Ranges 1 Thru 7 104
20 Log Normal Plot of Ranges 8 Thru 13 105
21 Brink Data 108
22 Typical System Geometry !41
23 Secondary Emissions Detector Arrangement 143
vii
-------
FIGURES (CONTINUED)
Number Page
24 Emission Conditions and Calculated RQ Values 146
25 Sample Optical Data Analyses Print-Out 149
26 Test 7 Data 155
27 Test 9 Data 157
28 Test 17 Data 158
29 Test 20 Data 160
30 Test 25 Data 161
A-l Test OB, #2 Wilputte Guide (Volume Flow in Cu. Ft/Min.) A-7
A-2 Test 1, #2 Wilputte Guide (Volume Flow in Cu. Ft/Min.) A-8
A-3 Test 1A, #2 Wilputte Stack (Volume Flow in Cu. Ft/Min.) A-9
A-4 Test IB, #2 Wilputte Guide and Stack A-10
(Volume Flow in Cu. Ft/Min)
A-5 Test 8A, #3 Wilputte Stacks (Volume Flow in Cu. Ft/Min.) A-ll
A-6 Test 2, Actual and Standard Volume Flow A-13
A-7 Test 3 A-15
A-8 Test 4 A-16
A-9 Test 5, Actual and Standard Volume Flow A-17
A-10 Test 6, Actual and Standard Volume Flow A-19
A-ll Test 7, Actual and Standard Volume Flow A-21
A-12 Test 8, Actual and Standard Volume Flow A-23
A-13 Test 9, Actual and Standard Volume Flow A-25
A-14 Test 10, Actual and Standard Volume Flow A-27
A-15 Test 11, Actual and Standard Volume Flow A-29
A-16 Test 12,.Actual and Standard Volume Flow A-31
A-33
viii
-------
FIGURES (CONTINUED)
Number Page
A-17 Test 13, Actual and Standard Volume Flow A-35
A-37
A-18 Test 14, Actual and Standard Volume Flow A-39
A-19 Test 15, Actual and Standard Volume Flow A-41
A-20 Test 16, Actual and Standard Volume Flow A-43
A-45
A-21 Test 17, Actual and Standard Volume Flow A-47
A-22 Test 18, Actual and Standard Volume Flow A-49
A-23 Test 19, Actual and Standard Volume Flow A-51
A-53
A-24 Test 20, Actual and Standard Volume Flow A-55
A-25 Test 21, Actual and Standard Volume Flow A-57
A-58
A-26 Test 22, Actual and Standard Volume Flow A-61
A-27 Test 23, Actual and Standard Volume Flow A-63
A-28 Test 24, Actual and Standard Volume Flow A-65
A-29 Test 25, Actual and Standard Volume Flow A-67
A-30 Test 26, Actual and Standard Volume Flow A-69
A-31 Test 27, Actual and Standard Volume Flow A-71
A-73
A-32 Test 28, Actual and Standard Volume Flow A-75
B-l The Particulate Sampling Train B"2
B-2 The Bag Sampler B~5
B-3 Sampling Trains B~6
B-4 The Total Hydrocarbon Analyzer Set-Up B~7
B-5 Mass-Particulate Sample Process B-15
B-6 Number of Particles Sized Per Total Area Viewed Versus B-34
Particle Diameter for Sample Number 1051
ix
-------
FIGURES (CONTINUED)
Number Page
B-7 Number of Particles Sized Per Total Area Viewed Versus B-35
Particle Diameter for Sample Number 1072
B-8 Number of Particles Sized Per Total Area Viewed Versus B~36
Particle Diameter for Sample Number 1076
B-9 Number of Particles Sized Per Total Area Viewed Versus B~37
Particle Diameter for Sample Number 1109
B-10 Number of Particles Sized Per Total Area Viewed Versus B"38
Particle Diameter for Sample Number 1222
B-ll Size Distribution by Particle Diameter for Samples B-42
B-12 Volume Percent Versus Particle Diameter for Samples B-43
B-13 Background Curves for Andersen Cascade Impactor - B-66
Limestone at 0.25 cfm
B-14 Background Curves for Andersen Cascade Impactor - B-67
Limestone at 0.66 cfm
B-15 Background Curves for Andersen Cascade Impactor - B-67
Iron Oxide at 0.247 cfm
B-16 Background Curves for Andersen Cascade Impactor - B-67
Iron Oxide at 0.65 cfm
B-17 Calibration Curve for Large Cyclone Using Limestone B-68
B-18 Calibration Curve for Small Cyclone B~68
C-l Coke Oven Charge Data Recording Overview c~2
C-2 Coke Oven Data Processing Overview c~3
C-3 Input Data Record Format - Program "BILL" c"6
C-4 Output Data Record Format - Program "BILL" C~7
C—8
C-5 Output Test Header Record Format - Program "BILL" C~10
C-6 Functions of Computer Program "BILL" C~12
C-7 Linear Threshold Values C~14
C-8 Data Transfer Details C~15
-------
FIGURES (CONTINUED)
Number Page
C-9 Functions of Computer Program "ANDY" C-16
C-10 Output Data Record Format - Program "ANDY" C-20
C-21
C-ll Data Transfer Details C-192
D-l Leaking Coke Oven Doors D-2
D-2 Instrument Layout, J & L P-4 Battery - Pittsburgh Works 1973 D-3
D-3 Cross-Sectional View of Leaking Coke Oven Door Collecting D-4
and Measurement Equipment
D-4 Collecting hood D-5
D-5 Installing Collecting Hood D-5
D-6 Measurement Duct D-7
D-7 Measurement Duct in Place on Oven D-8
D-8 Particulate Sampling of Leaking Door Measurement Duct D-9
E-l Larry Car Charging Coke Oven E-3
E-2 Carrousel Interior Exposed E-5
E-3 Carrousel in Operational Configuration E-6
E-4 Housing with Two Particulate Samplers E-8
E-5 Cumulative Size Distribution - Particle Number Percent E-14
E-6 Cumulative Size Distribution - Weight Percent E-15
E-7 Efficiency Curves E~17
E-8 Test 27, 150 X E~27
E-9 Test 27, 150 X E-27
E-10 Test 28, 150 X E-28
E-ll Test 28, 150 X E-28
E-12 Test 28, 150 X E-29
E-13 Brink Data E-35
E-14 Coulter Analysis E~37
xi
-------
TABLES
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Gas Chromatograph Parameters for MITRE Gas Analyses
Summary of Coke Oven Tests
Data Acquisition System - Channel Assignment
Variables Used in Volume Flow
Variables for Gas Analysis
Constitutents Measured by the Continuous and Manual
Sampling Systems
so2
H2
CO
co2
THC
°2
NO Plus NH,
x 3
NO
X
Emission Volumes for Wilputte Charging
Emission Volumes for AISI/EPA Charging Including Test 21
Emission Volumes for AISI/EPA Charging Excluding Test 21
Percent Reduction of Gaseous Emissions
Constituent Concentrations (Measured)
Andersen Analysis
Size Distribution Due to Coke Oven Charging
Size Range
Page
32
40
51
62
66
82
83
84
85
86
87
88
89
90
94
94
95
95
97
101
102
110
XIX
-------
TABLES (CONTINUED)
Number Page
23 Anisokinetic Mass Emissions (Wilputte) 112
24 Anisokinetic Mass Emissions (AISI/EPA) 113
25 Calculated Isokinetic Mass (Mass in Grains for Wilputte Car) 115
26 Calculated Isokinetic Mass (Mass in Grams for AISI/EPA CAR) 116
27 Tar Concentrations 118
28 Benzpyrene Analysis 120
29 Elemental Analyses and Supporting Coal Analyses 121
30 Coal Analysis (Ultimate and Proximate) 122
31 Conversion of Transmission Values to Ringelmann Values 147
A-l Temperature Record for the Coke Oven Tests A-2
A-2 Preliminary Test Temperatures A-3
A-3 Volume Flow A-4
A-5
A-6
B-l Summary of Samples Taken in the Field B-ll
B-12
B-13
B-14
B-2 Mass Results - Field Data B-17
B-3 Mass Analytical Results B-18
B-4 Mass Concentration Results B-19
B-5 Andersen Analysis Summary B-21
B-6 Andersen Analysis Summary B-22
B-7 Andersen Analysis Summary B-23
B-8 Andersen Analysis Summary B-24
B-9 Andersen Analysis Summary B-25
B-10 Andersen Analysis Summary B-26
xiii
-------
TABLES (CONTINUED)
Number
B-ll Andersen Analysis Summary
B-12 Andersen Analysis Summary
B-13 Andersen Analysis Summary
B-14 Andersen Analysis Summary
B-15 Andersen Analysis Summary
B-16 Cross Reference Information - AISI/EPA Larry Car
B-17 Optical Size Distribution Data on Sample Number 1051
B-18 Optical Size Distribution Data on Sample Number 1072
B-19 Optical Size Distribution Data on Sample Number 1076
B-20 Optical Size Distribution Data on Sample Number 1009
B-21 Optical Size Distribution Data on Sample Number 1222
B-22 Cumulative Percent Optical Size Distribution Data
B-23 Particle Diameter Versus Cumulative Percent (By Volume)
B-24 Results of Tar Analysis
B-25 Benzpyrene Analysis
B-26 Wavelengths and Flame Gases Used in the Elemental Analysis
B-27 Sample Combinations Used in Elemental Analysis
B-28 Filter Number 1
B-29 Filter Number 2
B-30 Aluminum
B-31 Antimony
B-32 Barium
B-33 Beryllium
B-34 Cadmium
B-35 Calcium
xiv
Page
B-27
B-28
B-29
B-30
B-31
B-32
B-39
B-39
B-40
B-40
B-41
B-41
B-45
B-46
B-48
B-49
B-51
B-52
B-52
B-53
B-53
B-53
B-54
B-54
B-55
-------
TABLES (CONTINUED)
Number Page
B-36 Cobalt B-55
B-37 Chromium B-55
B-38 Copper B-56
B-39 Gallium B-56
B-40 Germanium B-56
B-41 Lead B-57
B-42 Iron B-57
B-43 Magnesium B-57
B-44 Manganese B-58
B-45 Mercury B-58
B-46 Molybdenum B-59
B-47 Potassium B-59
B-48 Nickel B-59
B-49 Selenium B-60
B-50 Sodium B-60
B-51 Strontium B-50
B-52 Thallium B-61
B-53 Titanium B-61
„-'
B-54 Tin i B-61
B-55 Vanadium B-62
B-56 Zinc B-62
B-57 Coal Analysis (Ultimate and Proximate) B-64
B-58 Sieve Analysis B-65
B-59 H2S Analytical Results B-70
B-60 NO (NOJ Analytical Results B-71
ji £»
XV
-------
Number
B-61
B-62
B-63
B-64
B-65
B-66
B-67
B-68
C-l
TABLES (CONTINUED)
NH, Analytical Results
CN Analytical Results
Phenol Analytical Results
S02 Analytical Results
Pyridine Analytical Results
Gas Chromatograph Parameters for Gas Analysis
Calculations for Gas Analysis
02, N2, C02, THC, CB4, CO and HZ Results
Wilputte Volume Flow/Location, NO.
C-2 AISI/EPA Volume Flow/Location, NO^
C-3 Wilputte Volume Flow/Location, C02
C-4 AISI/EPA Volume Flow/Location, C02
C-5 Wilputte Volume Flow/Location, NO
C-6 AISI/EPA Volume Flow/Location, NO
C-7 Wilputte Volume Flow/Location, CO
C-8 AISI/EPA Volume Flow/Location, CO
C-9 Wilputte Volume Flow/Location, THC
C-10 AISI/EPA Volume Flow/Location, THC
C-ll Miscellaneous Constants Card
C-12 Test Duct Card
C-13 Test Time Definition Card
C-14 Duct Area Card
C-15 Free Fall Velocity Card
xvi
Page
B-72
B-73
•B-74
B-75
B-76
B-78
B-79
B-80
C-151
C-15 2
C-153
C-154
0155
C-156
C-157
C-158
C-15 9
C-160
C-161
C-162
C-163
C-164
C-168
C-169
C-170
C-171
C-172
-------
TABLES (CONTINUED)
Number Page
C-16 Anisokinetic Weight Percent Card C-173
C-17 Sample Velocity Weight Card C-174
C-18 Input Tape Format C-175
C-19 Data Field Format C-177
C-20 Particulate Correction Factors, Q C-185
C-186
C-21 Mass Loading Computations (Sample) C-187
C-22 Data Record Structure C-189
D-l Test #1 - Door Leak Experiment D-ll
D-2 Test #2 - Door Leak Experiment D-12
D-3 Test #3 - Door Leak Experiment D-13
D-4 Test #4 - Door Leak Experiment D-14
E-l Deposition Efficiency for Sampler at Coke Oven E-12
E-2 Test 22 E-13
E-3 Test 24A E-19
E-4 Test 26A E-20
E-5 Test 27 E-21
E-6 Test 28 E-22
E-7 Numerical Number Percent Versus Size Range E-24
E-8 Cumulative Percent for Particles > 18 u - Uncorrected E-25
E-9 Cumulative Percent for Particles > 18 y - Corrected E-26
E-10 Test #1 - Boom E-31
E-ll Test #2 - Boom E-32
E-12 Test #3 - Boom E-33
E-13 Test #4 - Emission Guide E-34
xvii
-------
ACKNOWLEDGEMENTS
Numerous people contributed to this project, from early planning through
execution and data reporting. A guiding influence in steering the
activities within the bounds of Jones & Laughlin plant regulations, and
the EPA requirements, was R. V. Hendriks, EPA Project Officer. Person-
nel at J & L who cooperated are too numerous to list, although T. R.
Greer and E. Renninger are to be noted for their special contributions.
Gases and particulates were manually sampled during the tests by Midwest
Research Institute, under EPA contract. Additional analytical support
was provided by R. Statnick of EPA.
This project spanned a period of nearly three years, and accordingly,
a number of MITRE personnel were assigned to the project as it developed
through various phases. K. Yeager and G. Erskine shared the Group
Leader responsibilities, while J. Hoffman and R. B. Shaller each in turn
acted as Task Leader. R. W. Bee conducted the Optics portion of the
program, while A. Wallo III was responsible for sensors and gas/
particulate samples with assistance from R. W. Spewak. W. L. Wheaton
was responsible for the continuous monitoring instrumentation with
assistance from J. Findley, J. Miller, R. Reale, and W. R. Robinson.
R. C. Kuehnel assumed the data processing responsibility, and support
was provided in this area by J. Morris. Additional assistance and ad-
vice was provided by A. F. Epstein, E. Jamgochian, B. A. Stokes and
J. B. Truett.
xviii
-------
SECTION I
CONCLUSIONS
Measurements were successfully made defining the. emission characteristics
of the coke oven charging operations for the P4 battery of The Jones and
Laughlin Pittsburgh works. Both the old Wilputte and the production
prototype AISI/EPA larry car charging operations at this battery were
characterized in terms of gaseous emissions and particulates released to
the atmosphere. Both continuous monitoring and manual sampling techniques
were used in a specially designed test program to obtain the required
information. Additionally, optical measurements were made to determine
the technical feasibility of a compliance monitoring system based on
optical measurements.
IDENTIFICATION OF COKE OVEN CHARGING EMISSIONS
Particulate mass emissions from the Wilputte Larry Car were defined from
10 charging operations. These mass emissions are reported in terms of
particulates released from combinations of six major emission points.
From the data on these emission points, a composite total mass emission
of 815 grams per charge was derived for the Wilputte car. Mass emissions
were also defined for the production prototype AISI/EPA larry car from
three major emission points from four separate charging operations. A
composite total mass emission figure of 120 grams per charge was derived
from the data from the four charging operations.
The major gaseous pollutants emitted during charging that were measured
are: total hydrocarbons, carbon dioxide, carbon monoxide, nitrogen oxides,
sulfur dioxide, hydrogen sulfide, methane, ammonia, phenol, cyanide.
Based on analysis of particulate samples collected from the charging
emissions, there is a possibility that carcinogenic materials may be
generally present in emissions released during coke oven charging. This
conclusion is based on the detectable presence of benzpyrene in the tar
portion of a large number of particulate samples. Although not all
chemical forms of benzpyrene have been identified as carcinogens,
-------
benz(oc)pyrene has been so labeled. In at least one large composite sample
of collected particulate material, analytical techniques have established
the presence of benz(oc)pyrene and/or benz(e)pyrene and other known or
potential carcinogens. Based on the representative nature of the sample
involved, it is reasonable to assume that the same constituents will be
generally found in the tar fractions of other coke oven emissions. To
further characterize the quantities of materials involved, tar concentra-
tions in particulate samples were found to range from approximately 30%
to 90%, with an average of 57%. Additionally, benzpyrene analysis of the
tar fractions showed concentrations ranging from 260 ppm to 18,000 ppm
(1 ppm is equivalent to 1 |i gram benzpyrene/gram of tar) .
An elemental analysis was undertaken to identify the trace constituents
of the coal charged and the resulting emissions. The intended emphasis
was the identification and quantification of hazardous elements, partic-
ularly heavy metals. The emissions analysis was performed on particulate
sizing equipment catches, and was limited by the sample size, as well as
available analytic techniques. These limitations resulted in a large
number of instances where constituent concentrations, if they did exist,
were below detectable limits. The emission constituent concentrations
which were successfully measured and reported are consistent with the
constituents identified in the elemental analysis of the charging coal.
Some of the more important elements identified in detectable concen-
trations in the emitted particulate material were Cu, Fe, Pb, and Zn.
Due to the variances in particulate sample sizes, it is not possible
to make a generalized statement concerning constituent concentrations
in the emissions. It should be pointed out, however, that no constituent
concentrations in excess of what can be reasonably explained by coal
constituents was identified in the particulate material. In order to
obtain more definitive information on particulate constituents, much
larger samples distributed as a function of particle size must be obtained
and analyzed. The particle size information is important in assessing
the results, since this property will determine whether the particles
fall or settle out quickly (i.e., particles >200|i), or behave similar to
a gas in the atmosphere (i.e., <3 u).
-------
A particle size distribution was derived based upon a composite of
samples taken from emissions released by both the Wilputte and the
AISI/EPA cars. The size distribution found was characterized as a bi-
populate lognormal distribution; with a distinct grouping of the finer
particles containing 47% of the sample weight and a distinct grouping of
the larger particles containing 53% of the sample weight. The finer
particle size grouping was found to have a mass mean diameter of 8.5 y
and a standard deviation of 2.5 y; whereas the larger particle size
grouping was found to have a mass mean diameter of 235 y and a standard
deviation of 3.9 y. It was also concluded that the tar portion of the
particulate sampling was derived primarily from the small diameter por-
tion of the particle size distribution.
As an additional baseline measurement, gaseous concentrations were deter-
mined for the longer term emissions from leaking seals on the pusher side
doors of the oven. Data was collected from four separate measurements
of leaking door emissions. The primary gaseous constituent found in
these emissions was total hydrocarbons, for which an average emission
value of .35 ACFM was found. This measured emission value for hydro-
carbons translated to an average value of 1.2 pounds of hydrocarbon
released over a 19 hour coking cycle per ton of coal charged to the oven.
This figure of 1.2 pounds of hydrocarbon per ton of coal coked is obviously
for an oven with doors leaking at a constant maximum rate. Since the
majority of the doors do not leak at this rate, this emission rate
should not be construed as a normal level of emissions from all oven doors.
COMPARISON OF EMISSIONS FROM WILPUTTE AND AISI/EPA CHARGING SYSTEMS
It was determined that the average total particulate mass emission from
the Wilputte car was 815 grains per charge. The AISI/EPA car averaged
120 grams per charge, a reduction of approximately 85% from the average
particulate emission level measured from the Wilputte Larry car.
The major gases emitted during charging with the Wilputte Larry car con-
sisted of; total hydrocarbons (33.77 scf/charge), C0£ (29.4 scf/charge),
CO (17.49 scf/charge) and NOX (.11 scf/charge). Other gaseous constituents
-------
that were found in the emis.sions included; S02 (maximum concentration
of 232 ppm), H2S (maximum concentration of 42 ppm), methane (maximum
concentration of 4 ppm), ammonia (maximum concentration of 130 ppm) ,
phenol (maximum concentration of 31 ppm) and cyanide (maximum concen-
tration of 16 ppm).
The gaseous emissions from the AISI/EPA larry car were measured for 10
separate charging operations and are compared to similar Wilputte
measurements below. The major gases emitted during charging with the
AISI/EPA larry car consisted of; total hydrocarbons (20.43 scf/charge),
an improvement of 40% over the Wilputte emissions, CO (6.28 scf/charge),
an improvement of 64%, (X>2 (3.82 scf/charge), an improvement of 87%,
and NOX (.021 scf/charge), an improvement of 82% over Wilputte emissions.
Other gases found in these emissions included; S02 (maximum concentration
of 25 ppm), and methane (maximum concentration of 1.8 ppm).
The percentage reductions of the various particulate and gaseous emission
constituents are not uniform primarily because of the variations in
emission reactions which occur during charging operations. Examples
might be the presence or absence of flame, variations in dilution prior
to sampling, and changes in volume flow causing similar changes in
reaction rates. It is felt, however, that a comparison based on total
mass emitted is reasonable and justified.
The elemental analysis of various particulate samples showed some degree
of uniformity in constituent concentrations. This would indicate that
the volume of trace elements emitted during a particular charge would be
directly related to total particulates emitted. Based on this premise,
it is reasonable to assume that reductions in particulate emissions from
the Wilputte to the AISI/EPA car connote like reductions in total volume
of trace elements (particularly heavy metals) emitted.
I
The primary carcinogenic materials, various types of benzpyrene, display
reasonably uniform concentrations in the tar fraction of particulate
samples. Therefore, it is reasonable to assume that a'n argument similar
to the above would be valid for the appraisal of reductions in total
-------
volume of carcinogens emitted (i.e., reductions in carcinogens could
be considered equivalent to reductions in total tars).
TECHNOLOGY OF COKE OVEN EMISSIONS MEASUREMENT
Although a number of technical problems were encountered during coke
oven testing, an extensive effort was made to minimize these problems
and obtain valid data.
Sixteen separate gaseous constituents were measured by either manual
sampling methods or the continuous measurement instrumentation system
or both. Of these sixteen gases, eight gases were analyzed by both
manual methods and the continuous measurement system, and good agreement
was found between the methods for four of the major gases (CO, C02>
total hydrocarbon, and 02)- Comparisons between the measurement methods
for two of the gases (S02 and l^S) showed significant discrepancies
which were identified as being due to chemical interferents giving a
"false high" reading from the continuous measurement systems. These
interferents (polycyclic organic materials were suspected) preclude
accurate measurement of S0£ and H2S with the best current method for
continuous monitoring of these constituents - ultraviolet absorption.
Additional testing must be performed to identify the nature of the
chemical species which interfere with this instrumentation technique
before the technique can be applied to the task of monitoring coke oven
emissions.
Manual sampling was used for certain components (particulates, CIfy, NH3,
HCN, pyridine, and phenols) because the manual sampling method was the
best and/or only technique which could be applied. In other cases (CO,
C02j S02, H2S, total hydrocarbon, H2 and ©2) , manual sampling was used
as a supplementary method for comparison with the results of the con-
tinuous measurement instrumentation. For this latter group, good agree-
ment was found between results from manual sampling and the continuous
data, except for S02 and H^S as noted above. The manual sampling data
for these two constituents was judged to be the most valid obtained
during the tests and is used in the comparison of emissions. Two
-------
conclusions were developed concerning the suitability of manual sampling
techniques applied to the measurement of coke oven emissions. First,
the manpower required to perform the sampling was substantial when com-
pared with that required for operation of the continuous measurement
instrumentation system. Second, the time required to obtain analytical
results on the samples was considerable, and therefore, the results
could not be utilized for test-by-test modification of procedures (as
continuous measurement results were used).
Several problems were encountered in the implementation of an emission
measurement system. The first problem relates to the concept of iso-
kinetic sampling and the difficulty of applying this concept to a highly
variable emission source such as a larry car. The approach followed
was to sample at a constant controlled rate close to the predicted
average stack velocity and to correct the collected mass based on the
actual measured velocities to a value that would have been obtained if
isokinetic sampling had been followed. To do this, it was necessary
to gather continuous data on the velocity and/or volume flow at the
sampling point. It was concluded that this was the best approach when
considered against the alternatives of continuously and manually
adjusted sampling rates (to match stack velocities), and systems which
automatically adjust sampling rates. Based on a review of the results
obtained this appears to be a valid and successful approach.
In the measurement of the emissions released in coke oven charging, all
emission points could not be measured during one charging sequence. The
approach followed was to sample at several of the emission points on
separate charges, and to combine these results into a composite mass
representative of the total mass emissions of the charging operation.
This approach to the measurement of mass emission was the best method
within the time and economic constraints of the test program. After
review of the data, it was concluded that there was sufficient variability
in emissions from point-to-point and charge-to-charge, so as to preclude
single point testing as a means of determining the total mass released
during a single charge with any degree of accuracy.
6
-------
The initial test objectives called for the performance of three tests
per day. However, a number of factors encountered after the start of
scheduled testing made the accomplishment of this goal impossible.
These factors included unscheduled equipment maintenance and repair,
problems in production "non-interference" scheduling of ovens within
instrumentation accessibility limits, post-test instrumentation and
data acquisition system turn around delays, and unexpected severe
worker fatigue.
The high temperatures and open flame prevalent in the coke oven environ-
ment often caused extensive damage to aarious system components necessi-
tating unscheduled repair or replacement. The unpredictable nature of
these occurrences also caused problems in replacement component avail-
ability.
The instrument lines to the battery top were capable of reaching a block
of 20 consecutive ovens. However, during an eight hour shift, only one
pass would normally be made through the block, exclusively charging
either the odd or the even numbered ovens. This reduced by a factor of
1/2 the number of available ovens in the accessible block and spaced
the charges of interest approximately 45 minutes apart. It was often
found necessary to delay the scheduled charging of an oven within the
accessible block to allow the completion of test preparation. Such
scheduled disruptions could only be tolerated on a limited basis with-
out affecting normal production.
Assuming no obvious damage to oven top instrumentation, the necessity
remained to check the continuous monitoring instrumentation and data
acquisition system components for proper operation. In addition, con-
siderable time was required when manual sampling was scheduled to turn
around manual sampling equipment between consecutive tests.
Lastly, workers experienced an unusually high level of fatigue after a
test period, primarily attributed to the rapid work pace necessary to
limit production interference (.tests were generally completed in less
than 20 minutes), high ambient temperatures (100°+), and poor air quality.
-------
The data which forms the basis for conclusions contained in this report
are judged to be consistent and repeatable in the context of test
requirements and the operational environment. This assessment is
founded on a careful examination of basic data element test averages
for volume flow, temperature, and measured constituent concentrations
including particle size distribution. For example, the average volume
flow values for a Wilputte stack emissions point show an average flow
rate of approximately 200 cubic feet per minute over 15 individual tests,
while the maximum and minimum average values for this same group of
tests are approximately 290 and 130 scfm, respectively. The temperatures
in the gas streams during these tests showed similar consistency with
an approximate average of 600° Rankine. The corresponding maximum and
minimum temperature averages were approximately 550° and 2000° Rankine
(in the presence of flame). Although almost constant fluctuations were
observed in volume flow, the data obtained appears consistent with all
observations of the charging process.
Good agreement between the continuous monitoring system and data obtained
through manual sampling is evident with the exception of H2S/S02 data.
The discrepancies noted here were several orders of magnitude and were
traced to chemical interferents. In addition, the concentrations
measured were consistent with similar data reported by other investigators.
A number of various techniques and equipments were used to obtain partic-
ulate size information. When the various sources of data are compared
using particle statistics criteria, good agreement is shown in support
of the composite particle size distribution reported here. It is felt,
however, that valuable information might be obtained through the addi-
tional refinement of particulate sampling techniques and devices, and
thorough analysis of the resulting data.
FEASIBILITY OF A COKE OVEN COMPLIANCE MONITORING SYSTEM
During the conduct of this test program, consideration was given to the
applicability of this measurement approach to compliance monitoring.
One factor which appears to transcend most other considerations is the
8
-------
unique design efforts required to implement this measurement approach.
Much of the time spent in preparing for the test operation was devoted
to tasks of fitting the measurement system to the particular battery
and larry car being tested (i.e., installation of sampling lines, signal
cable, facility modifications, field fitting of emission guides, etc.).
From this it was concluded that the measurement system and the concept
of using this system could not easily be applied to the task of emission
measurement of a larger population of coke ovens. Rather, the concept
is tailored for specific application to the particular battery and larry
car being tested.
The technical feasibility of a compliance monitoring system based upon
optical measurements was established, with qualifications. The optical
measurement system consisted of a fluorescent light bar source and a
35 mm sequence camera. Micro-densitometer readings of the film images
of the light bar were then used to give a measure of the emissions of
the charging operation based on the light transmission characteristics
of the emission volume.
Light transmission or its inverse, opacity, can be mathematically related
to particulate material contained in the emission plume based on the
portion of source light an individual particle removes from the light
reaching the detector or camera. Determination of total mass emitted
using this approach depends on several assumptions. It is necessary to
assume that the majority of the particles are larger than 3 y in diameter
so as to avoid problems caused by Mie scattering. It must be assumed
that the particle size distribution'is constant over time and that the
distribution can be described accurately by its geometric mass mean
radius value. In order to make the final total mass calculation, the
emission plume remains constant over short periods (5 seconds in this
calculation) such that total mass emitted can be related to vertical
plume rise.
Technical and economic constraints precluded a charge-by-charge com-
parison of the particulates measured by direct means, with estimates
of mass emission as determined by the optical measurement system.
-------
Instead, a comparison was made for six separate charging operations
(four Wilputte and two AISI/EPA), between the measured flow of gaseous
emissions (scf/second at one second intervals) and the particulate
mass rate across the length of the light bar as calculated from micro-
densitometer readings using the appropriate algorithms Cmeasured as
grams particulate/meter vertical distance at five second intervals).
In such a comparison, evidence of a correlation between optical
measurements and the mass loading in the emission plume would appear
as similarities in the trends of each measurement (i.e., comparison of
peaks, and increasing and decreasing trends of the curves). In general,
these trends and relationships were found, and are reviewed in detail
in the body of the report.
Although reasonable correlation was found, problems encountered in
accurately characterizing the particulate characteristics and the
variability of the size distribution caused serious doubt concerning
the ability of the system to accurately and consistently measure total
mass emitted for compliance purposes. An alternative approach was
developed and discussed in the body of this report. This approach
involves the use of the basic system elements to implement a refinement
of Ringelmann technique accounting for duration and volume of an
emissions plume.
10
-------
SECTION II
INTRODUCTION
The atmospheric environmental problems associated with coke manufactur-
ing are a serious concern to the industry and to anyone in the near
vicinity of a coke plant. The American Iron and Steel Institute (AISI)
has made studies over a number of years addressing the environmental
health problems associated with coke oven emissions. The results of
these early studies led most steel companies to initiate programs re-
lated to control of coke oven emissions. One particular aspect of the
emissions control program was improved charging techniques - a subject
receiving increasing attention from AISI and the industry. As a result
of this common interest, AISI and Jones & Laughlin Steel Corporation
signed an agreement in March 1969 whereby J & L would manage the AISI
coke oven charging program. In June 1970, AISI signed an agreement with
the Air Pollution Control Office (later designated Office of Research
and Development) of the Environmental Protection Agency under which half
the cost of the program would be born by EPA.
The AISI coke oven charging program was to culminate in a prototype of
a new larry car that would result in smokeless charges and have an en-
vironmentally controlled cab for the safety and comfort of the operator.
In order to determine the improvement in atmospheric emissions due to
the new prototype larry, EPA engaged the services of The MITRE Corpora-
tion in April 1971 to conduct a test and evaluation program. The over-
all objectives of the test program as agreed to by EPA and The MITRE
t
Corporation were as follows:
• quantify the atmospheric pollutants resulting from the
charging operation in the coking process,
• provide a comparative evaluation of the pollution
abatement system (AISI/EPA larry car versus existing
Wilputte Larry Car),
• determine the feasibility of a compliance monitoring
system concept based upon optical measurements.
11
-------
By July 1971, construction of the new car was started on the south end
of the P4 battery at the J & L Pittsburgh Works. In December 1971, the
new car first operated under power.
During this period, effort under the MITRE contract consisted of the
design and fabrication of emission guides to fit around the drop sleeves
of each larry car so as to channel emissions that would be normally
vented to the atmosphere; the design and installation of a van mounted
continuous monitoring instrumentation system for monitoring these
emissions; the preparation of manual sampling and analytical specifica-
tions; and the design and fabrication of equipment for the optical
measurement program.
In September 1972, MITRE performed two days of preliminary testing at
J & L to determine the feasibility of using certain sensors and to
determine gross flow parameters. Additional preliminary tests were
performed in May 1973, and full tests on the oven began in June 1973
and concluded at the end of August. Data reduction and report produc-
tion efforts were then initiated in September of 1973 and were completed
with the submission of this final project report in March 1974.
As an adjunct to the test activities carried on at the J & L facility,
observations of two new charging cars with features similar to the J & L
car were scheduled at two other coking operations in March and April
of 1974. The purpose of these observations was to collect data on the
operation of these cars including production data, reliability data, and
a qualitative assessment of emissions from the charge through visual
observations. Observations were scheduled at the Weirton, West Virginia
and Granite City, Illinois plants of The National Steel Corporation.
However, due to operational problems and program schedules, observations
at Granite City were indefinitely postponed.
Observations carried out at the Weirton Steel Division of The National
Steel Corporation provided much additional information useful in any
assessment of improvements and new technology currently in use at
operating coke plants. The data obtained during these observations has
been compiled and is published under separate cover.
12
-------
SECTION III
APPROACH
GENERAL
In order to compare the improvement in atmospheric emissions resulting
from the use of the new AISI larry car, the MITRE approach was to
channel all emissions from around the drop sleeves and Wilputte stacks
through ductwork of defined cross-sections for flow and particulate
measurement and gas quantification. These data were then integrated to
obtain a typical grain loading and gas mass flow during a charge by each
type larry car and, in turn, a measurable change due to the new design.
This section describes the Emissions Guides used for channeling emissions
from around the drop sleeves, the sensors and probes used on the oven,
and the instrument van where continuous gas analyses were performed and
where data were recorded. Manual gas and particulate samples were ob-
tained, and this sub-program is also described in this section.
A study was conducted to establish the feasibility of determining or
inferring the mass loading of particulate matter during a charge by
means of an optical (camera) system. This experiment is also described
in this section.
Finally, the test program, schedule, and operating procedures used in
the field measurement program are described.
CONTINUOUS MEASUREMENT SYSTEM
This system was designed to allow a continuous (one reading per second)
measurement of flow and selected gases. The various components required
to satisfy this requirement are described in this section.
Emissions Guides
These are structures that were designed and fabricated to fit around the
drop sleeves of the larry cars and channel the emissions that are usually
vented to the atmosphere. The guides for each larry car are different
because the configuration of the drop sleeves is different. In both
cases, heavy emissions escape following normal charging when the drop
13
-------
sleeve is raised up from the charging port on the oven floor. Conse-
quently, the guides were designed so that moving parts would not leak
emissions and still resist the heat of the gases and any local flames.
Wilputte Larry Car Emissions Guides - The duct which contained the emis-
sions from the Wilputte car and provided a controlled measurement loca-
tion was configured in the form of an annulus extending from the oven
top upward around the drop sleeve assembly. The inner wall of the
annulus was sealed to the bottom of the drop sleeve by means of a flexi-
ble fire retardant skirting. The outer wall of the annulus was located
7" outside the inner wall and extended from the top of the oven around
the drop sleeve assembly. The outer sleeve was sealed to the top of the
oven by the weight of the guide resting on its lower edge. As can be
seen in Figure 1 and 2, the annulus is elliptical in shape and smaller
at the bottom than at the top. This configuration provided an expanding
volume in the vertical direction which allowed the hot emissions to rise
and expand with minimal interference. Portions of the guide were blocked
off by deflectors, allowing measurements to be made in one well-defined
measurement duct. In order to provide a smooth emissions flow within
the measurement area, vortex breakers were placed inside the duct at an
appropriate distance from the sampling area. The breakers, in the form
of wide mesh screens, appeared to improve the flow characteristics in
the measurement ducts. The design of the guide was such that emission
suppression or inducement was minimized.
In addition to emissions from the vicinity of the drop sleeves, emissions
escaping from the three stacks on the Wilputte car were monitored.
Guides for these emissions were merely cylindrical "stovepipe" extensions
on the stacks.
AISI/EPA Larry Car Emissions Guide - This system consisted of two hoods
connected by a flexible flameproof section for each of the three drop
sleeve assemblies (Figures 3 and 4). The inner hood was designed to
surround the drop sleeve and contain the emissions when the sleeve is
in the lowered position. The hood was sealed at the top to the center
14
-------
FIGURE 1
WILPUTTE LARRY CAR
EMISSIONS GUIDE
15
-------
5"x20" EXHAUST PORT
(12.5 cm x 50cm)
FLEXIBLE
SKIRTING
FIGURE 2
WILPUTTE LARRY CAR EMISSIONS GUIDE
16
-------
FIGURES
AISI/EPA LARRY CAR
EMISSIONS GUIDE
17
-------
12" DIA. EXHAUST PORT (30 cm
INNER
HOOD
SKIRTING
FIGURE 4
AISI/EPA LARRY CAR EMISSIONS GUIDE
18
-------
flange on the drop sleeve assembly and extended downward and completely
around the drop sleeve assembly to a point slightly above the oven
surface.
A flexible woven fabric was attached to the lower edge of the inner hood
and extended downward to contact the oven top along a line completely
surrounding the lowered drop sleeve. The flexible wall was held in con-
tact with the oven top by the weight of the angle iron fabric frame that
rested on the oven floor. The hood was designed to minimize the inter-
nal empty volume while still allowing the movement of contained emissions
to a single measurement duct located at the top and forward side (away
from the lid lifting mechanism) of the inner hood. The measurement duct
extended some distance to obtain a smooth flow of emissions, thus pro-
viding a region in which accurate measurements could be made. The small
hood volume optimized the entrapment of emissions experienced during the
portion of the charging cycle when the drop sleeve is lowered. Vortex
breakers were placed at the beginning of the measurement duct.
At the completion of the charging cycle, the drop sleeves were raised
and the lid mechanisms replaced the charging port covers. During this
period, the outer hood, sitting on the oven surface and connected to the
inner hood by the lower end of the fabric wall, confined the emissions
and guided them upward to the measurement duct on the inner hood. In
order to allow normal operation of the lid lifter mechanism, the rear
portion of the outer hood was left open from the lower edge of the inner
sleeve to the oven surface. The lid lifter mechanism then passed through
this opening and replaced the charging port cover in the normal manner.
A partial seal of this opening was achieved by a flexible fabric skirt
attached to the back of the inner hood above the lid lifting mechanism
and directly behind the lid lifter magnet assembly. The inner portion
of the hood was balanced so as to allow the drop sleeve to be raised or
lowered while the hood was in place.
19
-------
SENSORS AND PROBES
Flow measurements and sampling require sensors and equipment located at
the emissions guides and stacks. Other hardware was required on the
oven to obtain the needed data. The mechanical, electrical, and elec-
tronic equipment located on the oven is described below.
Metal plates were fabricated and permanently attached to the sides of
the ducts of the emissions guides, and the sensor heads were mounted on
these plates in such a fashion that they extended into the flow of
emissions from the ducts. This design allowed for all delicate sensors
and circuitry to be handled separately from the emissions guides for
ease of cleaning and maintaining, as well as for protection against
rough handling. Equipment that was mounted on the plates consisted of:
pitot tubes for gas velocity determination, thermocouples for tempera-
ture determination, probes for collecting gaseous emissions, filter
bases for collecting particulates, and heaters for maintaining the
temperature of the gas and particulate probes. Photographs of the sen-
sor ducts are shown in Figures 5 and 6.
Sensors and sampling probes were connected to additional equipment
located off the oven floor in or near the small shed on the pipe bridge
opposite Oven 3-5 (see Figure 7). These connections were made through
wires and heated tubing that either transmitted electrical signals,
differential pressure, or gas samples. These wires and tubing were made
up into bundles, termed cables, and run from the emissions guides and
stacks via overhead booms and towers to the pipe bridge area.
Two booms were used in the testing area and were attached to the pipe
bridge shed: one closest to the coal bin supported cables for testing
the emissions guides of the Wilputte car and the one farthest from the
coal bin supported cables for testing either the Wilputte stacks or the
AISI/EPA emissions guides.
The small shed at the end of the pipe bridge cross-over main opposite
Oven 3-5 housed equipment required to convert temperature and pressure
probe outputs into electronic signals, and also served as the location
20
-------
SENSORS AND PROBES
Flow measurements and sampling require sensors and equipment located at
the emissions guides and stacks. Other hardware was required on the
oven to obtain the needed data. The mechanical, electrical, and elec-
tronic equipment located on the oven is described below.
Metal plates were fabricated and permanently attached to the sides of
the ducts of the emissions guides, and the sensor heads were mounted on
these plates in such a fashion that they extended into the flow of
emissions from the ducts. This design allowed for all delicate sensors
and circuitry to be handled separately from the emissions guides for
ease of cleaning and maintaining, as well as for protection against
rough handling. Equipment that was mounted on the plates consisted of:
pitot tubes for gas velocity determination, thermocouples for tempera-
ture determination, probes for collecting gaseous emissions, filter
bases for collecting particulates, and heaters for maintaining the
temperature of the gas and particulate probes. Photographs of the sen-
sor ducts are shown in Figures 5 and 6.
Sensors and sampling probes were connected to additional equipment
located off the oven floor in or near the small shed on the pipe bridge
opposite Oven 3-5 (see Figure 7). These connections were made through
wires and heated tubing that either transmitted electrical signals,
differential pressure, or gas samples. These wires and tubing were made
up into bundles, termed cables, and run from the emissions guides and
stacks via overhead booms and towers to the pipe bridge area.
Two booms were used in the testing area and were attached to the pipe
bridge shed: one closest to the coal bin supported cables for testing
the emissions guides of the Wilputte car and the one farthest from the
coal bin supported cables for testing either the Wilputte stacks or the
AISI/EPA emissions guides.
The small shed at the end of the pipe bridge cross-over main opposite
Oven 3-5 housed equipment required to convert temperature and pressure
probe outputs into electronic signals, and also served as the location
20
-------
LIFTING
BAIL
THERMOCOUPL
JPORT AND CLAMP FOR
PARTICULATE PROBE
GAS PORT
FLAME
ARRESTOR
PITOTTUBE
PRESSURE RELIEF VALVE
FIGURES
WILPUTTE EMISSION GUIDE
SENSOR DUCT
21
-------
LIFTING
BAIL
THERMOCOUPLE
FLAME ARRESTOR
PORT FOR PARTICULATE PROBE
PITOT
TUBE
PRESSURE RELIEF VALVE
lll'.i .illii
FIGURE 6
WILPUTTE STACK AND AISI/EPA EMISSION GUIDE SENSOR DUCT
-------
METEOROLOGICAL
TOWER
FIGURE?
CROSS-OVER BRIDGE SHED AREA
23
-------
for combining gas samples from all measuring points into a "common duct"
for manual sampling and further transmission to the instrument van. The
shed also served as a central location for communications via citizen
band radio between the test director, instrument van, and the optics
balcony, and was also the power distribution point for the heated probes
and gas lines. The porch-like platform in front of the shed was the
location from which the booms were manipulated.
Pressure transmitters were housed in a 24" w x 30" h x 18" metal box,
located adjacent to the shed. Control switches for activating the gas
and particulate control valves were also in this box. Temperature
transmitters were secured to a 19" wide instrument rack inside the shed.
The "common duct" manifold that combined gas flows was housed in a
12" x 12" x 36" heated box, located in the shed.
A meteorology station located on the top of the shed had associated
electronic boxes in the instrumentation van.
Cables, wires, and heated tubing were placed in a cable trough located
across the top of the pipe bridge where they dropped down to ground
level to the instrumentation van at the base of the bridge tower.
Instrument Van
Gas collected at the sensor ducts and the output of the temperature and
pressure sensors was processed by electro-chemical-mechanical instru-
mentation and the results recorded for subsequent analysis. The pro-
cessing center was a 36' van filled with seven racks of electronic
equipment to perform these functions and was located at the base of the
pipe bridge tower as shown in Figure 7. Three specific functions were
performed in the van: gas analysis, determining physical properties in
the vicinity of the measurement points, and data recording. Instrumen-
tation used is briefly described here, and a more detailed description
can be found in the Test Plan, MITRE WP-10434 and in MTR 6566, "A Con-
tinuous Monitoring System for Coke Oven Emissions Due to Charging."
24
-------
Gas Analysis System
The continuous gas measurement system consists of a number of specific
gas analyzers of two basic types: raw gas analyzers which accept a gas
sample as it comes from the emissions guides, and conditioned gas
analyzers which require preconditioning in the form of gas drying and
cooling.
The gas sampling system draws a representative sample of emitted gas
from the emissions guide and the Wilputte stack (when testing the
Wilputte Larry Car). The gases from the emissions guide probes and the
Wilputte stack probes are combined at the common duct in the pipe bridge
shed. All components in contact with the sample gas are held at ele-
vated temperature to prevent condensation.
The common duct consists of a plenum containing a mixed representative
sample of emission from the selected sources. The gas sample for the
wet chemical analysis (manual sampling) equipment is extracted from this
plenum. The remaining gas flowing through the common duct is fed through
heated sample lines to the analyzers located in the instrumentation van.
Upon entering the instrumentation van, the gas flows through a heated
manifold, from which a sample is extracted for the raw gas analyzers.
The remaining gas flows through a refrigerated condenser where it is
cooled and condensables are removed. The dried, cooled gas is then
pumped under slightly positive pressure to the conditioned gas analyzers.
A pressure regulator and bypass was provided to minimize the gas transit
time. A schematic diagram of this system is shown in Figure 8.
Raw Gas Analyzers
A number of .the gas analyzers operate on the gas sample under stack
conditions* The gaseous constituents monitored in this manner include:
SO , H S, THC, NO, NO and. H-0 vapor. Each instrument incorporated an
£ Z. A £
independently heated sample handling system for controlling the flow
through each analyzer.
The H S analyzer is a dual gas instrument, manufactured by Peerless In-
strument Company and operated on the principal of ultraviolet absorption.
25
-------
HEATED FILTER
ASS'Y
SPAN GAS
FIGURE 8
GAS HANDLING SYSTEM
-------
The band selected for H2S has an SO- interference factor, but since
the analyzer was also measuring S02 on a continuous basis, a correction
circuit was employed which subtracted out the SO- interference, provid-
ing a true output of H?S. The analyzer incorporates three ranges pro-
viding 3,000, 300, and 100 ppm by volume full-scale.
The THC (Total Hydrocarbons) analyzer operates on the flame ionization
principle whereby the gas sample is burned in an ionization chamber
using a mixture of zero air and pure H_ as fuel. The combustion pro-
cess produces free ions, which are detected by an ion detector. The
detector produces a current whose magnitude is a function of the number
of ions reaching the collector; therefore, the detector current is a
function of the total hydrocarbon content of the gas sample. The
analyzer utilizes an input gas dilution, system to eliminate interference
caused by the HL present in the sample gas.
The analyzer for Nitrous Oxide is a chemiluminescent device. In the
reaction
NO + 03 •+• N02 + hv , (1)
light (X - .6-3u) is emitted when electronically excited NO- molecules
revert to their ground state, and the intensity is proportional to the
NO concentration.
The analyzer for Nitric Oxide is similar to that for NO, but includes a
hot catalyst for the reduction of NO- to NO. The resulting measurement
is NO (NO + NO-). The quantity of NO.is subtracted electronically,
Xf £ '•
from the NO value, producing an output in terms of NO concentration.
X 4b
The Water Vapor analyzer is a MSA Lira 200 HR Non-Dispersive Infra Red
(NDIR) instrument. It has a diual range configuration with 10% and 50%
moisture full—scale ranges.
Conditioned Gas Analyzers
These gas analyzers are supplied with a cooled, dried sample gas stream.
A refrigerated condenser with a stainless steel condensation chamber is
employed to pump the dried gas through the analyzers. A dry gas sample
27
-------
handling system was employed for each analyzer, providing flow monitor-
ing, throttling control, and calibration gas insertion. Those gases
monitored are: H2> Q^ CO, and CO..
The analyzer for Hydrogen is a Beckman 7C Thermal Conductivity Analyzer.
The analyzer for Oxygen is a Beckman Model 742, utilizing a polaro-
graphic oxygen sensor.
The analyzer for Carbon Monoxide is a MSA dllA 202, non-dispersive in-
frared instrument. The instrument is a three-range device, operating
at 3%, .3%, and .03% by volume full scale.
The analyzer for Carbon Dioxide is a MSA LIRA 202 non-dispersive infra-
red instrument. The instrument is a four range device, operating at 10%,
3%, 1%, and .3% by volume full scale.
Physical Properties Measurement System
The physical properties of the gas effluent emitted at each source were
monitored. These properties include temperature, static pressure, and
differential pressure, and are used to compute the volume flow of the
emission and in processing the data to convert the gas measurements to
STP.
Temperature measurements were monitored by using Type K (Chromel-Alumel)
thermocouples in conjunction with Leeds and Northrup Model 1992 tempera-
ture transmitters in a temperature range of 0-2,200°F. The temperature
transmitter has a response of 1 millivolt/2°F.
The differential pressure measurements were made using a series of "S"
Type pitot tubes, connected to CGS Datametrics Model 536 Barocel pres-
sure transducers. The range of differential pressure is 0-10.0" W. C.
with resolution of .0005" W. C.
Meterological parameters were monitored on a continuous basis. These
parameters include ambient temperature, humidity, barometric pressure,
wind speed, and wind direction.
28
-------
Data Recording
The data produced by the gas analyzers and the physical property mea-
surement instruments were recorded, in IBM-360 compatible format, on
magnetic tape. Selected parameters were recorded in analog form on a
series of strip chart recorders. In addition, each parameter could be
monitored, independently, by a digital volt meter.
The data acquisition system was manufactured by Data Graphics Corpora-
tion, to MITRE specifications. The system provides 70 analog input
channel capability, time of day clock, and manual data input. The
analog-to-digital converter provides an accuracy and resolution of .005%
of full scale and is capable of operating at a data rate of 200 con-
versions/second.
Selected data parameters were graphically recorded on strip charts. The
recorders, manufactured by the MFE Corporation, are multi-channel
devices using rectilinear heat writing and providing 29 channels of
recording capability.
MANUAL SAMPLING
The sampling of gas and particulate by manual methods was done to com-
plement the continuous monitoring and optical portions of the program.
Samples were obtained at the same measurement points on the emissions
guides and Wilputte stacks as the emissions were channeled during nor-
mal charging and simultaneously with the continuous measurement tests.
A brief summary of this part of the test program follows below, with a
more detailed description found in MITRE WP-10179, "Manual Sampling and
Analytical Requirements for the Coke Oven Charging Emissions Test
Program," 16 February 1973, and MTR-6288, "Manual Sampling System for
the Coke Oven Charging Emissions Test Program," December 1972.
Gas Sampling System
Gases drawn through the probes of the sensor duct and into the overhead
boom cables to the pipe bridge were mixed in the "Common Duct" manifold.
The probes, cables, and common duct were maintained at elevated temper-
atures to prevent condensation of hydrocarbons and other tars. The
29
-------
heated common duct manifold was used as the source for the manual gas
samples, and a heated tube extended from this common duct to a heated
box containing four Teflon bags. During the test of a normal charge,
the four bags were sequentially filled with the gases from the common
duct. The heated box containing the four bagged samples was removed to
the portable lab where each sample was preprocessed and partially
analyzed on site to minimize the possibility of degradation of any con-
stituents. The preprocessing was as follows:
• Samples containing NZ, H2, 02, CO, CO^ and CH, were trans-
ferred to glass containers for later analysis.
S02 was absorbed using an impinger and held for later
analysis.
• NO was analyzed on site.
ji
• THC was analyzed on site.
• NH_ was absorbed by an impinger for later analysis.
• HCN was absorbed by an impinger train for later analysis.
• H2S was absorbed by an impinger train for later analysis.
• Specific hydrocarbons (pyridine and phenols) were placed in
trains for later analysis.
Later analysis of the samples was performed in the Kansas City Labora-
tories of Midwest Research Institute by prescribed techniques.
Analytical Methods
In this section, the analytical methods are described for the determina-
tion of gaseous concentrations in the manual sampling program.
CO, THC. Methane - A Varian Aerograph Gas Chromatograph, Model 1420-10,
equipped with a 5 ml. gas sample loop and a high sensitive thermal con-
ductivity detector, was used throughout for this work. These gas deter-
minations were made in the field lab due to the unstable nature of the
gases. The sample loop system was evacuated to as low a pressure possible
with a Welch Duo-Seal vacuum pump. For calculation purposes, this
pressure drop was considered to be atmospheric pressure. The sample was
30
-------
passed into the sample loop and the sample pressure measured. Using a
molecular sieve 5A column, the gases were separated under the instrument
conditions listed in Table 1.
0 , N£, CO and H^ - These gas concentrations were determined in the
identical method as just described, except that the analyses were per-
formed at the MRI laboratory in Kansas City at the completion of the
field work..
H2S - ASTM D 2725-70, "Methylene Blue Method;" exceptions to method -
Neutral CdSO, solution (140 g/liter) used in place of zinc acetate ab-
sorbing solution.
NO . - Federal Register Method 7 (Volume 37, Number 247), "Determination
jft.
of Nitrogen Oxide Emissions from Stationary Sources."
NH3 - Analytical Chemistry 39. 971 (1967),. "Catalyzed Berthelot Reaction."
CN - ASTM D 2036-72, Method C. Colorimetric (chloramine-T/Pyridine-
Barbituric Acid).
Phenol - ASTM D 1783-70. Aminoantipyrine-Ferricyanide Method.
S0_ - Federal Register Method 6 (Volume 36, Number 247), "Determination
of Sulfur Dioxide Emissions from Stationary Sources."
Pyridlne - Gas Chromatographic Analysis. Instrument - MicroTek Gas
Chromatograph, Model 220, equipped with a flame ionization detector.
Column - 5% Theed on Chromasorb G, 80/100 Mesh» 9' x 1/4" Cu. Column
Temperature - Isothermal at 85°C. Carrier Gas - N_. Carrier Gas Flow
Rate - 20 ml/min.
Particulate Sampling System
Particulate sampling was done at each emissions guide and Wilputte stack.
The sampling nozzles were located in the measurement ducts parallel to
the expected flow directions. Each nozzle was connected directly to a
heated cyclone and filter or an Andersen Sampler. This sampling system
further consisted of a modified EPA particulate sampling train employing
quasi-isokinetic sampling.
31
-------
TABLE 1
GAS CHROMATOGRAPH PARAMETERS FOR MITRE GAS ANALYSES
Gas
Column
Column Temperature, C
Detector Temperature, C
Carrier Gas
Carrier Gas Flow Rate,
ml/min
Bridge Current, ma
°2
(a)
60
195
He
100
200
N2
(a)
60
195
He
100
200
co2
(a)
250
195
He
100
200
H2
(b)
25
140
N2
60
150
(a) Molecular Sieve 5A, 5 ft x 1/8 in., s.s.
(b) Molecular Sieve 5A, 15 ft x 1/8 in. s.s.
32
-------
The samples were analyzed for mass loading, size distribution, and
elemental analysis. The weight of each sample was correlated with the
volume of the gas sampled from the stream to determine the anisokinetic
particulate concentration to which theoretical corrections were applied
to produce a calculated isokinetic concentration. Samples, obtained
from the Andersen impactor and captured on the MITRE designed carrousel
sampler (see MITRE WP-10480), as well as those collected by the Brink
impactor, were used for size analysis. Methods used were aerodynamic
classification and optical microscopy. The particulate samples obtained
were analyzed for the following elements:
Al Cu Mo Se
As Fe Na Sn
Ba Ga Ni Sr
Be Ge P Ti
Ca Hg Pb Tl
Cd K S V
Co Mg Si Zn
Cr Mn Sb U
Coal Sampling System
Coal Samples were obtained while testing both larry cars. Analyses per-
formed on these samples were size distribution, elemental, proximate,
and ultimate.
OPTICAL PROGRAM
One objective of the MITRE test program was to investigate the feasi-
bility of developing simple monitoring systems and techniques and to
define an inexpensive compliance monitoring system for evaluating emis-
sion sources with similar characteristics.
The proposed system of monitoring was to cause minimal interference with
the normal operation of this coke plant and was to be adaptable to various
plant configurations and conditions. Optical techniques have been
successfully applied in numerous situations involving a requirement to
monitor emissions to determine pollution and air quality. Because of
this previous use, because it did not interfere with coke production and
33
-------
would be adaptable, and because it could be operated in parallel with
the continuous monitoring system, an optical monitoring system was de-
signed to satisfy this requirement.
The three essential elements in any optical measuring system are:
• a source of radiant energy (the sun, light bulbs, lasers,
etc.) or a target which reflects radiant energy,
• energy collection and focusing devices (lenses, mirrors),
• energy detectors (the eye, photoelectric cells, photo-
graphic film, etc.) plus amplification, display, and re-
cording devices.
The basic type of light source which best satisfied the system require-
ments is the flourescent tube. In order to achieve the desired vertical
width for the light source, a parabolic reflector was placed behind two
standard 1500 MA 96" parallel flourescent tubes which achieved desired
intensity as well. The tubes and reflector were housed in a "light bar"
which had a diffuser lens mounted in front of the unit to provide a
more uniform light output. The bar was mounted 25 feet above the oven
floor on the south end of the battery (see Figure 9). The main section
extended across the entire width of the oven, and "wing sections" ex-
tended outward and downward.
The collecting, focusing, and detecting device selected is a modified
35 mm sequence camera capable of providing a frame rate of one frame per
second. The camera was mounted on the first balcony of the coal in on
the oven and was equipped with a 200 mm lens such that the light bin
completely filled the field of view. The camera was also equipped with
a time-of-day clock which recorded time on each frame. Additionally,
a near-field light (18" flourescent lamp) was positioned in front of the
camera balcony for orienting the camera after each setup.
During operation of the optical system, sequential photographs were taken
of the light bar as the emissions plume from the larry car crossed the
path between the camera and the bar. Each frame of the film contains
an image of the large linear light source (light bar). In recording the
34
-------
LIGHT BAR
CONTROL SWITCH
CAMERA AND
*• LIGHT BAR
REMOTE CONTROL
CIRCUITS
FIGURE 9
SIMPLIFIED SYSTEM COMPONENT CONFIGURATION
35
-------
image of the light bar, the film detected any diminishment of light bar
brightness caused by particulate or aerosol material in the light path.
This diminishment was recorded as a decrease in density of the light bar
image on the film. By properly choosing the exposure values for the
film and accurately processing the exposed film, the amount of light
diminishment was determined using a micro-densitometer to analyze
image density changes of the individual frames.
Each frame contains three light bar segments as shown in Figure 10 .
Each segment has two distinct areas. The bottom or black (this drawing
represents a negative film frame) area represents the surface of the
light source. The top or white area is a flat surface of the light bar
that was painted flat black to reduce any reflected light coming from
this area. All three segments are identical in construction. The
light source image appears ^350y high on the film frame. The flat black
area appears ^400y high and adjoins the lighted area directly above or
to the outside of it. Micro-densitometer analysis of both light and
dark areas of each of the three segments of the light source image can
be accomplished in the following manner.
The micro-densitometer was adjusted so as to have a 50 x 195y aperture.
A single scan was made across each area (lighted and black individually)
of each of the three light bar segments. During the scan, a specified
number of data points were recorded. A total of 539 data points evenly
spaced were taken across the lighted and the black areas of the longer
or horizontal bar section. The data obtained on each scan was placed on
magnetic tape in the form of a binary density scale reading for each
data point. Each scan sequence was further annotated on the magnetic
tape with header information including sequence number, test number,
type of frame and frame number information.
This brief summary of the optical program has been condensed from MITRE
WP-10149, "Design of an Optical Emissions Measurement System for Coke
Oven Monitoring," by R. W. Bee, December 14, 1972, and MTR 6596, "Optical
Emissions Measurement Program Development."
36
-------
bJ
\
140555
140556
DDDannanaDaDDnnacinaaDnnDnnD
FIGURE 10
FILM FORMAT
-------
FIELD TEST PROGRAM
Scheduling the tests on the battery in a reasonable sequence required
coordination of many activities. First, the installation and site
modifications to the new AISI/EPA larry car had to be completed; second,
the procurement of critical equipment by EPA was required; third, the
development and verification of the instrumentation by MITRE was
necessary; and, finally, the EPA/AISI commitment to be completed at an
early date was to be honored. A further requirement, to avoid inter-
fering with J & L coke production, was also incorporated into the
schedule. The system to measure the number of required parameters at
the number of required measurement points, verify the measurement, con-
vert the data to electrical signal and record these data for later pro-
cessing required a large amount of equipment and an extensive amount of
time for installation and checkout. Preliminary testing and data veri-
fication were also required and field tests, preliminary analysis, and
dismantling had to be considered. These activities covered a five-to-
six month period, during which five to six weeks of field system tests
were conducted. The overall program schedule is shown graphically in
Figure 11, and the final summary of the tests actually performed is
provided in Table 2 .
Only those ovens in the Immediate vicinity of the pipe bridge and
within reach of the overhead cable booms were monitored during the tests.
Since an average time of approximately 15 minutes per oven was required
between chargings, 11 ovens in the vicinity of the pipe bridge were
normally charged during a four-hour period in any one daylight shift.
MITRE avoided interfering with the charging schedule, and advantage was
taken of the standard charging sequence and the normal shutdown at
shift changes and noon to establish a tentative test routing. Based on
this information, and experience on the battery, a Test Plan Outline for
one day's activities was followed as shown in Figure 12.
Test Procedure
Various sampling techniques were employed during the time period when the
larry car charged coal into the oven. These techniques and systems,
38
-------
INSTALLATION
PARTICULATE DETERMINATION QUANTIFICATION.
CONTINUOUS MONITORING "AND OPTICAL TESTS -
MANUAL SAMPLING PRELIMINARY TEST
DATA ANALYSIS AT MITRE AND MRI
DATA ANALYSIS AT MITRE-
CONTINUOUS MONITORING AND OPTICAL TESTS
AIS I/EPA TESTS • —
DISMANTLE
t>vt
«BHi^i?i^.?M *o ao oo p. o. o. a.
33 g 33933 2 3 3 <"> «> « «
l"t"---
MANUAL SAMPLING SET-UP |"~l
WILPUTTE TESTS'
58 n
CONTINUOUS MONITORING AND OPTICAL TESTS 1 |
FIGURE 11
TEST PROGRAM SCHEDULE
DATA ANALYSIS AT MITRE
CONTINUOUS MONITORING AND OPTICAL TESTS - 1|
-------
TABLE 2
SUMMARY OF COKE OVEN TESTS
Date
May 7
9
10
30
31
June 18
19
20
21
22
25
26
27
28
29
July 2
3
4
5
11
12
12
Aug. 6
7
8
8
9
9
13
13
14
15
15
16
16
16
17
17
22
22
22
23
23
23
24
MITRE
Number
OA
OB
1
1A
IB
2
3
4
5
6
7
8
8A
9
10
11
12
13
14
15
16
17
DL-1
DL-2
18
19
DL-3
20
21
22
DL-4
DL-4A
23
24
24A
25
26
26A
27
28
28A
28B
MRI
Number
Wl
W2
W3
W4
W5
W6
W7
US
W9
W10
Wll
Kl
£2
K3
DL-1
DX-2
W12
K4
Oven
3-12
3-13
3-7
3-5
3-7
3-6
3-3
3-8
3-3
3-6
3-10
3-11
3-8
3-8
3-6
3-10
3-11
3-8
3-9
3-15
3-6
3-6
3-4
3-4
2-27
2-27
3-3
3-5
2-26
3-4
3-3
3-11
2-23
2-23
3-12
3-3
3-11
3-1
3-9
3-13
2-23
3-4
3-10
3-10
Time
7:30
4:45
10:30
4:00
5:30
1:00
11:15
5:00
12:00
2:00
4:00
12:00
5:20
12:00
4:00
4:00
3:00
10:00
4:20
1:30
12:00
12:00
2:30
11:00
4:00
11:30
1:00
11:30
12:00
4:00
2:00
3:30
4:00
2:00
5:00
12:00
3:00
5:00
12:30
3:00
5:00
9:00
Points Monitored
K Ullputte Guide
*2 Wllputte Guide
12 Wllputte Guide
#2 Ullputte Stack
f2 Ullputte Guide & Stack
3 Wllputte Guides
12 Wllputte Guide & Stack
#1 Ullputte Guide & Stack
12 Wllputte Guide & Stack
13 Wilputte Guide & Stack
3 Wilputte Stacks
3 Wilputte Stacks
3 Wllputte Stacks
3 Ullputte Stacks
3 Wllputte Guides
3 Wllputte Guides
3 Wilputte Guides
#2 Wilputte Stack & Guide
11 Wilputte Stack & Guide
fl & #2 AISI/EPA Guide
3 AISI/EPA Guides
3 AISI/EPA Guides
Leaking Door
Leaking Door
3 AISI/EPA Guides
3 AISI/EPA Guides
Leaking Door
3 AISI/EPA Guides
3 AISI/EPA Guides
*2 Wllputte Stack
Leaking Door
Leaking Door
12 Wilputte Stack
3 AISI/EPA Guides
#2 Wllputte Stack
3 AISI/EPA Guides
*2 Nilputte Stack
t2 Wllputte Stack
12 AISI/EPA Guide
12 Wilputte Stack
Coke Side Door
Ascension Pipe
Remarks
Plow data only - no mag. tape
Flow data only - no mag. tape
Preliminary test for MRI - bad
flame out
Plow data only - no mag. tape
Flow data only - no mag. tape
No test - boom cables shorted
MRI activated manually - bent
boom
Larry car breakdown - no test
Diff. press, pitot plugged
Dlff. press, pitot plugged
J & L cancelled test
No mag. tape - no gas sampled
Gas leaks
Guide gas probe plugged
12 probe & thermocouple data N.6.
Early part of test not recorded
03 sensor burned out
Bad seal on 13 drop sleeve
Low flow - fans burned out
MITRE Carrousel Partlculate
Sampler
Bad fire on 93
Poor gas data
No gas data obtained
Partlculate sample during push
Partlculate filter
40
-------
: TIME TO TEST STARTS (IEHE -fr-l-2 Hit.
-10 MIS. -IS MIX. -S Kttt.
m /
COORDINATOR \
III IKt
TLhT
DIKtCTOK
: LARRY
CAR
CONTINUOUS
:«M1TORU<;
SiL
OPTICAL
MtASURL-H) V) S
HEUSSIOSS 1
GUIDES
|_|i»5t«lB«;:irATiosl__
~ v« 1
H - H
-J
1— I MftTiaUAlC 1—
tfiLpum 2 J
UA1LY
PREPARATIO:;
PHASE
AUM WAV
KOPPEKS 2 i. ,
AWAY AWAY
UUPUTTt
-1
I CAR otrr OF
i SERVICE
KOPPERS INSTALL
EMISSION GUIDES
POSITION OVER
LOAD COAL — CHARGING PORT-
DROP SLEEVE DO'/N
post no:; CAR
C.EARG[;iC PORT
INSTALL
PACKAGES
PACKAGES
T...AL CHECK OUT
^JmLLtCTI'T C3UIP 1"S-»LLEtl
^| '
JI'ISTALL PART,
— C
41
-g
Icoi
*""*" 1 sv<
~~\ CA."
ARCE
1
i
LA!'"OFTE "R'PLAC
-
i
i
1
i
|
CORD
TA
1
AECT
D>LES
~i
i
LECT
PLCS
T
___
ERA
-
-
Roovr
SENSORS &
TUBES
RTWVF
SCN'tIRS
'
RJM3VE SA,'
ronp. FRO
n. GUIDE
ECALIG
Tn
10 SfORAOC
SI^W«
LVTION |
— 1
Hi-LLAii.
STOPE
A;;D SOPORS
1
A-» >*«,, |
FilK SIUPIENT
OR ANALYil.-.
A'lPLtS Piuii>Ml>
FILM LABELED FOR
SulP 1LVT TO
'RUCti'iOR
AXD
REPAIR
STANDBY
FOR
::E\T
TES1
FIGURE 12
DAILY TEST SCHEDULE OUTLINE
-------
the continuous gas sampling system, the manual gas sampling system, the
particulate sampling system, and the optical system are described in
this section.
Continuous Sampling
The MITRE Test Director obtained from J & L a list of ovens to be
charged during the test day. Based on this data, a schedule was pre-
pared which showed tentative times that the "MITRE dedicated" ovens
were to be charged. Based on this schedule, the step-by-step actions
for the tests were scheduled.
Two hours prior to the test, all heaters and gas samplers were turned
on. One hour prior to the test, all other equipment was turned on.
Gas sample analyzers were calibrated per manufacturer's instructions and
placed in a "standby" mode. Temperature and pressure sensors were
tested for response, and all flow lines were checked for pressure leaks.
Emissions guides for the Wilputte car were temporarily stored beside
the railing along the coke side of the battery near the oven to be
tested. The sensor ducts were made ready for the test and hung on the
boom. An aluminimum ladder was stored in a convenient location.
During installation of test equipment on the battery, the Wilputte larry
car was modified by installing the following equipment:
• Stack sensor modification rings
• Drop sleeve modification rings.
The AISI/EPA. car required no modification.
Initial plans to test all stacks and emissions guides of the Wilputte
larry car at one time were reduced in scope due to the time required to
install equipment prior to charging, the time allowed by J & L for
testing the car, and the excess exposure of personnel to heat and smoke
during a prolonged period. This plan reduced to testing either the
three emissions guides, the three stacks, or one emissions guide and
42
-------
its corresponding stack. Testing the AISI/EPA car was done by monitor-
ing all three emissions guides at one time. The step-by-step procedure
for each of these three test situation is given below.
In testing with the Wilputte Emissions Guides, the larry car proceeded
to the coal bin in the usual fashion, was loaded with coal, and was
then disengaged, moved to the charging port and the drop sleeve was
lowered to the top of the oven. At this point, the charging cycle was
halted, and two men installed the emissions guide on the pusher side
drop sleeve and then proceeded to install the center emissions guide and
the coke side emissions guide. A third man then detached the sensor
ducts from the boom and attached them to the emissions guides. After
coordinating the quenching sequence of the coke car, to assure that ex-
cess steam was not emitted during the charging, the Test Director com-
municated via citizen band radio with the camera operator, the manual
sampling crew chief, the boom operator, and the instrumentation van
operator; and when all parties were in a state of readiness, he or the
lead man on the oven floor commenced a ten point countdown to instruct
the larry man to start the charge. This countdown was broadcast to all
parties associated with the test so that the exact starting time for
the charge was known. The charging cycle was resumed at this point, and
the oven was charged in the usual fashion. The larry man reported to
the lead MITRE man on the oven floor when this charge was complete. This
completion time was relayed via radio to all test personnel. After all
coal had been deposited in the oven, the drop sleeve was raised for ap-
proximately one minute, after which time personnel removed all instru-
ment housings from the emissions guides, and the guide handlers removed
the guides so that the larry could move far enough for the lid man to
replace the oven lids. The larry car was then released to continue its
normal charging duties. Emissions guides, sensor ducts, booms, and
cables were removed from the oven floor near the test area, so that
they would not interfere with the normal operation of the J & L personnel.
Instrument probes, wires, and tubes were visually inspected, and
emissions guides were examined. Any clogging, burning, or warping was
repaired by the boom man and guide handlers prior to the next test.
43
-------
Wilputte Stack Tests
For these tests, the larry car was again positioned over the charging
ports, and the drop sleeve was lowered to the oven top. The charging
cycle was halted, and a ladder was raised against the pusher side hopper.
The sensor duct was then removed from the boom, carried up the ladder,
and installed in the stack cap. The ladder was then moved to the center
hopper and the sensor duct installed in the stack cap. In a similar
fashion, the sensor duct was installed on the coke side stack cap. After
all the sensors were in place, the countdown was initiated and charging
proceeded as for the emissions guide tests. At the conclusion of the
charge, the larry car was backed off the oven enough to allow replacing
the lids. After the lids were replaced, the ladder was used to climb
up on the larry to remove the sensor housings. After the larry had been
released, all equipment was inspected and repaired in preparation for
the next test.
The emissions guides for the AISI/EPA larry car consists of two
parts: the rigid inner hood and flexible outer skirting. The inner
hood was installed while the car was stored at the south end of the
battery and the drop sleeve lowered to the floor. The outer hood was
placed around the charging ports of the oven to be tested while the
larry was off the floor. The larry car proceeded to the coal bin for
coal after the inner hood had been attached and the drop sleeve raised.
The car then proceeded to the oven to be tested, was positioned over the
ports, and was then placed in a standby mode while testing equipment
was attached. The outer flexible skirting was then attached to the inner
rigid hood. The sensor ducts were then inserted into the openings in
the inner hood and firmly attached so that raising and lowering the drop
sleeves would not cause them to disengage. The countdown and charging
then proceeded as for the previously described tests. At the comple-
tion of the charge, the larry car operator signaled the Test Director,
who broadcasted this information to all parties involved. The emissions
guides were constructed so as to remain in place while the lids were
automatically replaced. After replacement of the lids, the outer skirt
44
-------
was then removed, the larry car was returned to storage at the south
end of the battery, and the inner hoods were removed.
After the last test of the day, the emissions guides were stored at the
south end of the battery, well out of the way of normal activities, and
the booms were securely tied back.
Optical Measurements
At the beginning of each test day, initial setup of optical equipment
was necessary. This involved the transport of the camera and camera
control unit from its storage place in the office van to the balcony of
the coal bin. During inclement weather, installation of a protective
hood or shelter on the balcony was accomplished prior to movement of the
equipment. Also, prior to transport of the camera, film was placed in
the camera and camera mechanical operation was checked. The first step
in initial assembly of the system was to attach the camera body to the
camera mount on the balcony rail. Next, the lens was mounted on the
camera and preliminary alignment of the camera was accomplished. Next,
the camera control unit was set up and connecting cables were installed.
The A.C. power supply for the camera and control was connected and power
applied to the system. The camera was then checked on manual pulse
operation and the presence of automatic exposure pulses was confirmed.
Next, the correlation data display was synchronized with the data acqui-
sition system by using the manual set button of the display and the
intercommunications circuit to the van. After synchronization, the dis-
play was allowed to run continuously during the test day. Next, fine
alignment and focusing of the camera was accomplished. This step re-
quired the illumination of the light bar which normally remained on
night and day during a test week. The lens opening was set and the
camera frame rate switch and shutter speed adjustment checked for proper
settings.
The camera was then ready for the first test. The procedure described
above required 30 minutes and was started at least 45 minutes prior to
the expected test time to allow proper coordination and temperature
stabilization of all equipments involved. At this point, a written
45
-------
record of equipment configurations and settings was made for the test
record. When all jequipment had been checked and configuration data re-
corded, a cloth cover was placed over the camera and lens to protect it
from particulates and moisture in the air.
While the larry car was moving into position for the charge, the camera
cover was removed, and a final visual check of the equipment was made.
Ten to twelve exposures would be taken during this time to obtain a
clear background base reference for the test.
Camera operation was initiated at the conclusion of the countdown by
the Test Director. Exposure continued for five minutes or until only
four feet of film remained on the supply roll. At this point, the
camera was stopped to provide a section of unexposed film on which ex-
posure calibration could be performed.
After the test, the connecting wires to the camera were disconnected,
and the lens was removed and stored in its carrying case. The camera
was next removed from its mount and placed in its carrying case and
transported to the van dark room where the film was removed and placed
in a storage container along with one copy of the test configuration
data sheet. The container was then placed under refrigeration to await
shipment to the processor.
Next, the camera was inspected and cleaned, if necessary, in preparation
for the next test. When another test was contemplated for the day, a
fresh roll of film was threaded into the camera and the proper mechanical
operation checked. If it was the last test of the day, the camera was
placed in its storage container and left in the van until the next test
day. After the camera had been checked, cleaned, and stored in the van
after the last test, the camera control unit and associated components
were removed from the balcony and stored in the van.
Manual Gas Sampling
Prior to the test-charging operation, manual sampling personnel set up
the gaseous sampling system on the pipe bridge. The heated lines from
the sampling trains and common duct to the heated manifold were connected
46
-------
and preheated to operating temperature. All components of the system
(pumps, lines, connectors, solenoid valves, and heaters) were tested to
see that they were in operating condition.
Once pre-test activity was completed, the manifold line was connected to
the common duct. The sequential sampling system was then manually
turned on at the conclusion of the start countdown for charging.
Once turned on, the sequential sampling system was- manually controlled
by predetermined time intervals. The sequential sampling trains then
collected grab samples at selected intervals. When the charge was com-
plete, the manual sampling personnel stopped the sequential cycle as
the last grab sample bag was filled.
Particulate Sampling
Prior to the test, sampling nozzles and filter assemblies were mounted
on the sensor ducts. The impinger trains, pumps, and regulators were
mounted on the instrument carts and prepared for operation, and were set
up along the coke side of the battery near the oven to be charged. The
lines from the filter to the impingers were connected. All quick dis-
connect connectors were checked to see if they were in operating order.
The lines were checked for leaks and/or breaks, as well as preheating
to proper temperature.
As each emissions guide was brought into place and the sensor duct con-
nected, the manual sampling team brought the assigned instrument cart
up to the guide and connected the input heated lines of the impingers to
the filter assembly. This procedure was followed for each emissions
guide and was performed consecutively as the guides were set up.
After the test was completed, the lines between the impingers and the
valves were disconnected. At the same time the line between the filter
and impinger was disconnected from the sensor ducts. The instrument
carts were then moved to the coke side of the oven. Once the emissions
guides or stack sensors were removed, the filter, lines, and nozzles
were cleaned.
47
-------
SECTION IV
DATA HANDLING PROCEDURES
Data obtained during the field portion of the program were consolidated
at MITRE headquarters, where it was subjected to critical quality con-
trol, corrected for travel time delays, and analyzed by computer methods.
The results of the analyses were then presented in terms of pollutants
emitted from the two larry cars being tested. This section describes
the steps taken to develop these larry car emissions characteristics.
INPUT DATA
Test data were obtained from the Continuous Monitoring, Manual Sampling
and Optics Programs. With minor exceptions, all data required extensive
analysis at a location other than the test site before meaningful infor-
mation could be derived. The original data, in its various forms, is
discussed here.
Continuous Monitoring Data
The continuous monitoring data were obtained every second from the sen-
sors during the tests and were recorded on magnetic tape and strip
charts. The data consisted of gas analysis, flow sensor output, meteor-
ological parameters and certain control and status information. To
utilize the magnetic taped data, a program was prepared for reading the
tape and printing the data in a form that could be read and evaluated.
In the output of this program, the units defining the data are in
voltage form. The print-out from this program was inspected to prove
that the basic data, as produced by the sensors, was recorded on the
tape accurately. The second-by-second variations in output voltages can
be observed on the print-out, and the absence of data becomes readily
apparent.
An example of one of these print-out sheets is shown in Figure 13, which
shows data from Test 13, for the center Wilputte stack and guide. The
first line on the print-out is a data-time group [year, month, day, hour
(EST) and minute]. The second line consists of headings for the data
48
-------
VE/u: 1*
SEC 13J7
LH V-IITS
oj-ii: 11
01-14840
02-163BJ
03-00500
04- 'I'll )
05-94 7i»
06-13*72
97-1713-1
06-U5140
09-10170
10-01470
11-03420
12-11 2 1'J
13-07090
14-03554
15-90518
16-06730
17-01340
18-01240
19-00570
20-00760
21-01180
22-01278
23-16152
24-01080
25+16381
26+16381
27+00001
28+16381
29-00180
30+00003
31+00483
32+09241
33-00100
34-004*0
35+16381
36+00009
37-00554
38+00001
39-00410
40+00021
41+JOOOl
42-30860
43*16381
44*00009
45-00638
46«163ol
47*16331
48-15570
49*16361
53-00700
OOOOUOOO
OOOOO'iOl
OOJO'J.-O.)
00001 101
00003000
"1NTH
src 14)7
CH V III if>
•10-OSOb'J
01-J4d60
02-1631)
03-J0446
0*-O1092
Ob-0*21 J
06-027')d
07-07073
OH-02880
09+OOOCI1
10-11632
11-07378
12-11250
13-07090
14-03530
15-00530
16-0»730
17-01350
18-01331
19-00570
20-00760
21-91160
22-01280
23-16140
24-0107*
25+16381
26+163dl
27+00001
26+16361
29-00180
30+00003
31+00491
32+09291
33-00210
34-00452
35+16381
36+00011
37-00550
38+00001
39-00410
40+00021
41+OOOU1
42-00650
43+16381
44+>J0009
45-00714
46*16381
47+16381
48-15570
49*16331
53-00690
oooooaoo
.» ')00100
J1 1)01 03
ij 1 ) ) )01 •)
OOUOUOU'J
JULY i
si.-: iji.7
f.H VCLT1
11- J* ).'•)
U-04.IS?
•J2-163HO
03-OV*00
J4-01192
J5-03712
I6-U2090
17-07151
08-0^554
')9+»0>'Jl
10- jlal')
!!-••) !)!?•>
12-01220
13-07100
14-01530
15-00530
16-06730
17-013*0
19-01290
19-J05BO
20-J0150
21-01180
22-01278
2J-16152
2*-01050
25+16381
26+16381
27+00001
28+16381
29-10180
30+OOC03
31+00*91
32+09721
33-002*0
34_0044o
35+16381
36+00001
37-00550
38+00001
39-00410
40+00021
41+00001
42-00840
43+16381
44+ >00 13
45+OC001
46+16381
*7tlt38<
*8- l»5b )
*9+16381
53-00?00
00000000
100UCi.CO
lioio * z •
000 I-J" 1 >
oooococo
W 1.! IM
Sc( 16J7
Crl V')LTS
i-i-147'ia
Jl--l«*0<)0
12- lo3iH
03-003*0
lt-01260
15-0?210
J6-014oO
07-0734?
08-02260
09+00005
11-01630
11-08172
12-01246
13-07100
14-03580
15-00556
16-06720
!7-nl3*0
13-01210
19-00570
20-00756
21-01180
22-01280
23-16140
24-01050
25+16381
26+16381
27+00001
23+16361
29-00180
30+00001
31+00501
32+09661
33-00234
34-00442
35+10381
36+00009
37-00552
36+00001
39-00412
4O+ 00021
41+00001
42-00332
43+16381
44+00013
45+00001
46+16381
47+16381
*)-15572
*9*16381
M-00700
JO'JOOOOO
J'JJOIOOO
n yio j Ju
to JOOOO'l
JOOOOOOO
II ••<• Mil
b: (. 17J7
TH VH.TS
irj-^/t5,jo
.Jl-04730
J2-16300
03-03280
04-01*40
35-029'jO
06-00922
07-07100
08-02052
09-00210
1 J-01920
11-08450
12-01250
U-07100
14-03570
15-00528
16-06730
17-01368
18-01240
19-00570
20-0075*
21-01180
22-01278
23-16152
24-01050
25+16381
26+16381
27+00001
23+16361
^>9_f),5l92
30+00001
31*00501
32*09701
33-00074
34-004*8
35*16381
36+00007
37-005*0
38+00001
39-00410
40+00021
41+00001
42-00820
43+16381
44+00011
45-00680
46+16381
*7+163Bl
48-15580
49*16381
53-00700
00000,000
) JL>)H030
1 K-OOOX'
"> ) '"I'loil
03000000
' I'j
Srf, 1H')7
cn VPUTS
)r>-o*394
01-0*4 JO
12-15922
03-00240
0*-11372
05-0^540
06-00500
07-07166
08-01780
09+00001
10-01678
H-08490
12-01238
13-07100
14-03558
15-00520
16-06720
17-01358
18-01320
19-00580
20-00764
21-0118.1
22-01278
23-16152
24-01088
25+16381
26+16381
27+00001
28+16381
29-00180
30+00033
31+00513
32+09841
33+00001
34-30444
35+16361
36+000 'J3
37-00552
38+00001
39-00410
40+00021
41+00001
42-OOH32
43+16381
44+00013
45-00638
46+16361
*7+l«81
*6-1557»
*9+1638l
53-00700
OOOOOOOO
10O 3 ^310
01 OOJ 1C*
J"10i") J J
00011131
JrC 1VU7
CH VI LTI
l^i-.lf, 2SO
-11-0*6*0
12-1 ?432
J3 -00220
•14-014*2
15-02274
v<)6-'-02i(;
07-07354
08-01 5SO
09+00001
10-0127B
11-089S4
12-01250
13-07100
14-03568
15-005*0
16-06752
17-OU60
18-01314
19-00570
20-00768
21-01160
22-01286
23-16140
24-01050
25+16361
26+16381
27+00001
28+16381
29-00180
30+00001
31+00513
32+09981
33-00050
34-00440
35*16381
36+00003
37-00554
38+00001
39-00410
40+00021
41+00001
42-00810
43+16381
44+00013
45-00570
46+16381
47+16361
48-15570
49+16381
53-00700
OOOOOOOO
10040000
OjiiO J01n
lllilQ jnilp
luoooooo
Sir joj/
CH VHTS
00-04U42
01-1*434
02-15032
03-01112
04-014»0
05-02030
06-00043
07-07330
06-01390
09*00001
10-01063
11-09160
12-01234
13-07100
14-03550
15-00520
16-06726
17-01364
18-01210
19-00608
20-00760
21-01160
22-01278
23-16150
2*-Ol08*
25 + 16381
26+16381
27+00001
28+16381
29-00180
30+00001
31+00513
32+10293
33-00000
34-00444
35+16381
36*00005
37-00552
38+00001
39-00410
40+00021
41+00001
42-00809
43+16381
44+00013
45-00720
46+16381
47+163B1
48-15590
49+16381
53-00690
00000300
00000010
OO10>>'>1 .)
uooojnio
OOOOOO'H
;.K. V107
i"rt vni TS
00-03840
01-34292
12-14650
13-00158
•34-J1476
05-01790
06-00000
07-07440
08-01278
09+00001
10-00962
11-08474
12-01250
13-07112
14-03518
15-00526
16-06730
17-01362
18-01240
19-00592
20-00760
21-01180
22-01280
23-16152
24-01086
25+16361
26+16381
2 7+00001
26+16381
29-00190
30+00001
31+00515
32+10241
33+00001
34-00442
35+16361
36+00007
37-00550
38+00001
39-00410
40+00021
41+00001
42-00792
43+16381
44+00013
45-00770
46+16381
47+16381
48-15570
49+16381
53-00700
OOOOOOOO
OOOOOOOO
J'.llilOOOO
•) J'JUOOOil
10000000
SPC 2207
CH VOl TS
00-03600
01-(>4?04
02-14180
03-00116
04-01460
05-01570
06-OOU01
07-07*70
08-01100
09+00021
10-00998
11-06284
12-01238
13-07100
14-03518
15-00514
16-06722
17-01360
18-01314
19-00570
20-00756
21-01190
22-O1280
23-16152
24-01080
25 + 16381
26+16381
27+00001
28+16381
29-00180
30+00001
31+00521
32+10405
33+00001
34-00444
35+16381
36+00003
37-00554
38+00001
39-00412
40+00021
41+00001
42-00790
43+16381
44+00013
45-00780
46+16381
47+163J1
48-15568
49+16361
53-00700
OOOOOOOO
OOOOOOOO
00000010
OOOUOOOO
OOOOOOOO
SfC 2307
CH vr
-------
columns further down the page. The time is in seconds to the second
decimal place, therefore, the first two digits are the only significant
ones for this .analysis. The third line is also a heading line for the
columns, indicating channel number and voltage levels. The remaining
lines are data channel numbers and their respective voltage levels,
ranging from 0-1.638/volts, with the channel assignments as given in
Table 3.
The channel assignments given in Table 3 were made before experience
was gained on the oven and several significant field modifications were
performed. Of most signficance is the fact that a maximum of only three
measurement points were monitored at one time as opposed to the nine as
originally planned. Many temperature and pressure channels were, there-
fore, not used. A threshold system to start and stop gas flow was
found to be impractical and, therefore, those channels were also not
used. These channels continued to record on the data acquisition system,
but the values shown on the print-outs are noise recorded on the unused
channels, and were ignored in later work with the data.
Another source of continuous monitoring data was the strip charts. All
gas analyzer data and all flow related data (pitot pressure and tempera-
ture) were recorded on strip charts as well as on magnetic tape. Exam-
ples of this data are given in Figures 14 and 15, which show outputs
from four gas analyzers and four channels of output data for AP and T
for Test 13. The print-outs described above were compared on a channel-
by-channel basis with the strip charts and corrections made to reconcile
any differences. In those cases where the data acquisition channel was
determined to be faulty, the data were then taken from the strip charts
and punched on IBM cards for insertion in the analyses program.
As a next step, the output of gas analyzers (located in the instrument
van) was correlated with the thermocouple and pitot tube output (located
on the oven floor). The time difference between these two readings is
due to the travel time of the gas to pass through the tubing to the
analyzers and the process time in the analyzers. (During the Field Test
Program, sample gases were inserted in the lines on the oven and the time
50
-------
TABLE 3
DATA ACQUISITION SYSTEM - CHANNEL ASSIGNMENT
Channel No.
00
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Constituent
4
NO
THC
H2
°2
CO
co2
Wind Speed
Wind Direction
Relative Humidity
Ambient Air Temp.
Barometric Pressure
Volume Flow
Threshold Sum
#1
#2
" #3
#4
#5
rt #6
#7
#8
#9
Channel No. Const
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
53
St.P
AP
T
St.P
AP2
T.2
St.P
APS
T3
St.P
AP
T
AP
T
AP
AP
T6
T
AP
AP9
T
T
Manu.
Coi
Threi
.Gas '
in Van
51
-------
1 ' ' ! !
•:•- f
J_L
• ' ' :.! r rru' i
MFE CORPORV'ON
H.
i i
./'
FIGURE 14
GAS ANALYZER STRIP CHART DATA
52
-------
•
SALEM, NEW HV.U'SHIRE. U.S.*.
;
. . i ;
. ,
;
! j .
'
1 !
i
1 ,
.
1
FIGURE 15
GAS TEMPERATURE AND PRESSURE STRIP CHART DATA
53
-------
at which the analyzer responded was noted. This was done for all gases
analyzed, and the delay varied from 16 to 30 seconds.)
After the specific "gas channels" had been time-shifted, the data were
then ready to be converted from voltages to engineering units. This was
done by applying the manufacturers calibration data (which was period-
ically checked at the site) and observed effects, adding the proper
units and reformatting the print-out page into groups of rows and print-
ing increments of 15 seconds on a page. The actual corrections and
calibration factors applied are not included in this report, but have
been published separately as MITRE Working Paper WP-10445, "Conversion
and Correction Factors for Coke Oven Emissions Data."
The resulting print-out for part of Test 13 is shown in Figure 16. Here
the columns are better separated and identified for the 15 seconds of
data. The first two columns along the left hand side are channel
identifying number and identifying constituent. The first ten rows are
either percent or parts per million of the indicated gas. The next
twelve rows (four used in Figure 16) are allocated for pitot tube pres-
sure and associated measurement duct temperatures in inches of water and
degrees Fahrenheit; actual channel assignment for each measurement point
is also given in.MITRE WP-10445, "Conversion and Correction Factors for
Coke Oven Emissions Data." The next four lines (Ch. 10 through 14) are
meterological data: wind speed in miles per hour, wind direction in de-
grees from true north, ambient temperature in Fahrenheit and barometric
pressure in inches of mercury. Thresholds 1 and 2 (Ch. 17 and 18) were
not used. Channels 19 through 24 indicate points on the larry car that
were being monitored; Emissions Guide (EG) on the Pusher Side (P), Middle
(M) and Coke Side (C), and Stacks (ST) on the Pusher, Middle and Coke Sides
(P, M, C). Channel 48 indicates when manual gas samples were being drawn.
The magnetic tape that produced this final print-out was then used as the
starting point for the final data analysis. A complete listing of the data
was sent to the Environmental Protection Agency, Office of Research and
Development, National Environmental Research Center, Control Systems
Laboratory.
54
-------
- (ft CCKE CVEH TESTS
F/CE
MITRE TEST NlMEtP 12
MRI TEST MJHftf
SEC
00 ( H20 2.6
nt ecu cn7 imc.n
02 PPM H2S 365 l.C
03 PPM NO 0.0
n.4 ppu NPV 7*.4
OS X THC 0. 29.00
n TH i
18 TH 2
20 EG^M ***nw****i
21 EG-C
22 ST-P
23 ST-H ***QN****>
24 ST-C
5.1 3,5
CC1 096
ti.e ai.e
2f.CC 29.CC
>»OK***»*»ON****
3.1
81.8
29. CO
**ON«4**'
2.3
81.8
29.CC
>*CN****«
ttCKttt+l
1.9
81.8
29.00
*ON****
1*7
fU3*7
81.8
29.00
*»ON****
1.8 2.7
81.8 81.8
29.00 29.00
**ON******ON***
2.6
fl**»
£1.8
29.00
**»CN****<
2.9 3.2
059 037
81.8 £1.6
29.00 29.00
>*CN******GN***«
2.S
077
El.£
29.00
'**ON»*»»
2.3
•9C1
81.8
29.00
037 '
81.8
29.0D
**ON»*****ON***
..«-&««A^«.r&4**
FIGURE 16
FINAL DATA PRINTOUT IN ENGINEERING UNITS
-------
Manual Sampling Data
The manual samples of gas and particulate were obtained by Midwest Re-
>
search Institute (MRI) under separate contract to the Environmental
Protection Agency. The samples were transported to MRI's laboratory in
Kansas City for final analysis, and the results were transmitted to
MITRE for incorporation in the overall data study of the coke oven emis-
sions. Farticulate samples were also collected as part of the manual
sampling effort, and the analyses of these samples are also incorporated
in the overall study.
Gas Samples
Results of the gas analysis were in the form of tables. Analytical re-
sults for pyridine, SCL, cyanide, phenol, NH,, NO , and H~S are listed
£» -J X £t
on computer print-out charts. The results of the analysis for 0., N.»
CO, C02, CH,, H- and total hydrocarbons are listed on a hand-printed
table. These print-outs and tables are provided in Appendix C of this
report.
Particulate Samples
Samples obtained at the Emissions Guides and the Stacks by MRI were
analyzed at their laboratory. The data were reported in the form of
computer print-outs, copies of laboratory notebook sheets and hand
written mass loading tables. This data is included in Appendix B.
Particulate samples obtained by the MITRE carrousel sampler were analyzed
at the MITRE Washington operations facility. This sampler, analytical
procedures and a complete presentation of data has been published as
MITRE Working Paper WP-10480, "Direct Impaction Particulate Collecting
Carrousel." The form of these results is a series of tables and size
distribution curves, which are provided in Appendix E.
Coal Samples
Proximate analyses were performed on the four coal samples obtained.
This work was done by Industrial Testing Laboratory of Kansas City under
subcontract to MRI. Their findings are reported as "Certificate of Coal
56
-------
Analysis", tables listing the constituents measured, and are enclosed
in Appendix B.
The ultimate analysis, elemental analysis and size determination were
presented as tables and are also enclosed in this appendix.
Optical Data
The record of light transmission from the light bar passing through the
emission plume to the detector (camera) was recorded on the photographic
film in the form of varying light bar image densities. The exposed film
was sent to a photographic processing laboratory where a sensitometric
exposure was placed on an unexposed portion of the film, and the film
was developed. The sensitometric exposure provides a basis for associat-
ing an observed light bar image optical density with a value of light
intensity producing the image exposure. Using the values thus obtained,
it is possible to determine the value of light obscuration. This infor-
mation was then used to estimate the mass loading values of the observed
emission plume.
The data on optical densities of the light bar images was provided in two
forms by the film processing facility. The microdensitometer electron-
ically digitized the measured optical density value for a specific point
(area of 50 by 200 microns) and placed this value in a binary number
format on a magnetic tape. The microdensitometer also simultaneously
produced a continuous strip chart record of measured optical density.
As stated previously, the data on the tape was contained in the form of
eight place binary numbers ranging from 0 to 255. One horizontal scan
across the light bar image produced 539 individual binary density
records corresponding to 539 points on the scanned image. A second pass
across the black surface area produced 539 more density records which
correspond directly with the light bar records for purposes of comparison
and mass loading calculations. A third scan containing 539 records was
made for some frames as a part of an experiment to investigate the Mie
scattering effect. This scan was across a short section of light source
having color filters placed on the diffusion lens to approximate
57
-------
monochromatic light. This unit was located on top of and at one end of
the primary light bar structure.
The data tape, generated on a manually controlled data logger, was read
as a binary number by a standard nine-track computer tape drive, but
was not directly compatible with computational or display portions of
the data processing program. Because of the format of the tape, records
to be processed were combined with header information obtained from the
hand-written tape log supplied with each raw data tape. The header in-
formation for each individual scan of 539 points consisted of a sequence
number, test number, clock time for frame, and type of frame. This in-
formation was combined on a separate program data input tape for further
processing.
A second data input consisted of two place tables in punch-card form.
These tables provided necessary correlation for conversion from density
values to relative exposure values (light intensity values, assuming a
constant exposure time) for a particulate processing run. The table was
generated by the processing laboratory using 21 density values obtained
from the sensitometric exposure wedge. A computer program fitted a
curve to these 21 points and produced the 300 evenly spaced (in light
intensity) values of optical density in the range of the sensitometric
exposure. A second table of 300 corresponding exposure values was cal-
culated by the program from the density information. The program used to
generate the exposure values is described in detail in MITRE Working Pa-
per 1045, "Correction and Conversion Factors from Coke Oven Emissions Data."
MATHEMATICAL TREATMENT OF DATA
The processes of converting the data that was described above into
meaningful terms such as pounds of pollutant emitted, mass loading, etc.,
are generally straightforward. An effort was made to perform these
functions with automatic data processing hardware, although this project
was a one-time effort and programming requirements are out of proportion
to the amount of data processed. This was intentional so that the
analytical routines developed for this project could be made available
to anyone with a future similar requirement.
58
-------
Concentration and Volume Flow Variability
The pattern of emissions produced by a coke oven charging operation
differs radically from a source which could be monitored by standard
manual sampling techniques. As a result, real time information is re-
quired for data reduction procedures. For the brief period of time
during which the charging operation lasts, differential pressure (AP),
temperature (T), and constituent concentration vary erratically. Values
can change from 0 to .1" HJD AP, 100° to 1000°F temperature, or zero to
full scale concentration in a matter of seconds. Figures 14 and 15 are
strip charts showing some typical fluctuations in AP and concentrations.
Figure 17 shows flow with respect to time and charging procedures for a
typical test.
These figures demonstrate the importance of a continuous monitoring sys-
tem for measuring the type of emission fluctuation that results from a
charging operation. It is obvious from Figure 17 that a manual measure-
ment would be too insensitive for the emission guide flows and too slow
for the rapid variations in the stack emissions to produce an accurate
volume of emissions. An example of the effects of using real time con-
centration values as opposed to an average concentration value for cal-
culating emission volume can be shown in Test 21 (see Appendix C for
data). A volume of THC from Test 21 was 33 scf as calculated using real
time values while the use of average concentrations and flow produced
19.2 scf Cv40% error).
Volume Flow Parameters
A program to determine volume flow and velocity of coke oven emissions
was the first of the data reduction programs implemented. Both the
gaseous and particulate systems depend on velocity and volume flow for
final calculations of mass and/or volume.
The volume flow program uses the equation:
AP.(t) [T±(t) + 460] (29.92)
. (t) - (174) x/i V — P.(t) - Gk (2)
59
-------
o»
o
LIDS REPLACED
FIGURE 17
TYPICAL VOLUME FLOW GRAPH VERSUS CHARGING PROCEDURES
-------
to determine the volume flow, F.(t), at location i for a given test.
Constants and variables required in this equation are listed in Table 4.
Equation 2 was used to determine the in-stack real time volume flow
(CFM) at each emission point and these values were then stored for use
with the particulate sampling program.
A second real time volume flow adjusted to standard conditions (SCFM) ,
F.'(t), (70°F @ 1 atm) was required for use with the continuous gas and
manual gas calculations where t
FCt) P(t)
•P f
[T±(t) + 460] °R 29.92" Hg
Then the total volume (CF) emitted at location i, ET. , under stack con-
ditions is given by:
ETi = £ At[F..(t)] (4)
t=tl
and under standard conditions (SCF) :
E ' = £ At[F »(t)l (5)
The real time volume flow in standard and stack conditions F.'(t) and
F. (t) was stored on magnetic tape, but the output format was in graphic
form only. The total volume values E. and E ', however, were both stored
and output numerically. The average temperature (F°):
T. = £ T.(t)/x (6)
1 t=t X
was also computed and printed during this operation, as well as the high
and low temperatures.
61
-------
TABLE 4
VARIABLES USED IN VOLUME FLOW
X = Pitot tube factor, a value from .95 to .7
?
. _ .69 ft for Wilputte Emissions Guide
i ~ .85 ft2 for Stacks and AISI/EPA Guides
P.(t) = Static pressure - Channel 14 (inches Hg)
G, = Specific gravity of the gas relative to air
T = Total time of test, in most cases, will be
described by Channel 48 (in seconds)
t., = Start of test, Channel 48 will read on (1 volt)
AP (t) = Differential pressure at time t and location i
(inches H00)
T (t) = Temperature at time t location i (F°)
At = Time interval (seconds)
62
-------
Continuous Gas Monitoring System Parameters
The gaseous concentrations are given for standard conditions and the
concentrations, C.(t), of the gas constituents are assumed to include
% H_0 in total volume. Gases measured after drying (H0, 0 , CO, CO :
£• £.£<£»
channels 06 to 09, respectively) were corrected to wet gas concentra-
tions by:
C.(t) = C., (t)
y jdryv
100 -
100
(7)
where, C., (t) = volume % in dry gas as measured by respective instru-
ments .
To determine the volume flow (SCFM) of each gas constituent, j, that is
measured in the stack under standard conditions (70°F, 1 atm) the
equation:
E.j
-------
where
C.. = Concentration of j at ambient
_ Number of Locations: F.(t) < F
Number of Locations
FT is a constant which was determined after the first run of the volume
flow data was analyzed.
Equation 9 (correction for ambient dilution) was only used for channels:
S02 - 01 THC - 05
H2S - 02 CO - 08
NO - 03 C02 - 09
N0x - 04
Then the total volume (SCF) of constituent j measured directly was:
E. = E At[E.(t)] (11)
and the total volume corrected for continuous sampling is
E ' = £ At[E '(t)] . (12)
• *
The real time volume per unit time [E.(t) and E.'(t)] for each constituent
j was displayed in graphic form. The total volumes in correct and un-
corrected form (E.1 and E.) was listed numerically. The average concen-
trations C (t) and C.f(t) were computed by
J J
C.(t)/T (13)
64
-------
and
_ T
-------
TABLE 5
VARIABLES FOR GAS ANALYSIS
F."*(t) - from volume flow program (SCFM)
- % H20 - 00
- ppm S02 - 01
- ppm H2S - 02
C4(t) - ppm NO - 03
C..(t) = C,.(t) - ppm NO,, - 04
C > '
c6(t) -
c7(t) -
/
c8(t) -
c9(t) -
c10(t) -
Number
rr X
% THC
%TI
"o
2.
%o2 -
% CO
% co2 -
of F±'(t) <
05
06
07
08
09
FT
Number of Locations Monitored
F = A constant to be determined after
volume flow data has been reduced
66
-------
Pg = Standard pressure, mmHg;
?a = Atmospheric pressure, mmHg;
S = Volume % of gas in standard.
(2) Ratio of standard gas at atmospheric pressure per instru-
ment response unit =
R - V /IRU , where
sa s'
R = Ratio, volume of standard gas per instrument
reponse unit;
IRU = Instrument response units (either peak area or
height) for standard gas.
(3) Volume of unknown gas at atmospheric pressure =
V = (R) (IRU), where
lio. tl
V = Volume of unknown gas at atmospheric pressure;
IRUu= Instrument response units for unknown gas.
(4) Total sample volume at atmospheric pressure =
Vts = (5) (Pv/Pa), where
V = Total sample volume;
ts
P = Pressure of sample, mmHg.
(5) Concentration of unknown gas, volume percent =
Volume percent of unknown = (V /V ) 100.
U3. to
Size Analyses - The primary methods used to determine a particle size
distribution consisted of a combination of Andersen size distribution
r
and optical sizing performed in samples taken from the MITRE Carrousel.
The Andersen samples were taken in the measurement ducts. At the comple-
tion of a test, the Andersen sampler was hand carried to the field labor-
atory and allowed to cool down to a stable temperature. The sampler was
then disassembled. The plates and rings were then weighed on site. The
sampler was cleaned and readied for the next test and the washings were
stored and labeled for later reference.
67
-------
The data from the field weighings were later analyzed at the MRI Kansas
City facilities. Standard computer programs were used by MRI to produce
size distribution information in ranges from about 20 y to .5 y and in-
formation on gross weight percent of the sample above and below these
boundaries. The data are listed in Appendix C. They were then combined
and reduced with MITRE data as described in another section. The orig-
inal Andersen distribution was calculated assuming a 1 gm/cc density;
however, the final distribution given was calculated assuming a 1.2 gm/cc
density.
Samples collected by the MITRE Carrousel were analyzed by optical tech-
niques. Representative slides were selected from each test and analyzed
for particle size distribution and number density. The microscope
reticle was calibrated with 39ym and 28pm pollen. Particles were
grouped in one of seven size ranges using the calibrated reticle with
the "Ferets diameter" method. The ranges are:
1 - 4.6 to 9.2 urn 5 - 73.6 to 147.2 ym
2 - 9.2 to 18.4 ym 6 - 147.2 to 230 ym
3 - 18.4 to 36.8 ym 7 - 230 ym .
The particles were sized optically with a B & L microscope at 300, and
150 powers. The magnified samples were displayed and counted on a
closed circuit television system coupled optically to the microscope.
Through the use of a polaroid camera attachment, pictures were period-
ically taken.
A particle count of about 500 (termed a "batch")* was required at each of
the powers. Areas scanned (fields-of-view) were selected at random to
produce a representative sample, until the batch was complete. Each
* Lodge, J. P. Production of Controlled Test AtmospheresIn: Air Pro-
duction, Volume II, Stern, A. C. (ed.) New York and London, Academic
Press, 1968. p. 465-481
68
-------
slide to be analyzed was chosen such that a field-of-view had less than
200 particles** in the countable ranges for any given power. The areas
2 2
of field-of-view at 150 power and 300 power are 965 ym and 241 ym ,
respectively.
A batch counted in the 300 power range includes particles between 4.6 um
and 147.2 ym (excluding ranges 6 and 7). The 150 power batches include
particles greater than 9.2 ym (excludes range 1). To combine the results
for 300 power and 150 power batches sized for the same test, the number
of particles in the omitted ranges were calculated from the ratios of the
other batches. For example, the number of particles in range 1 of
batch 2 would be calculated as the ratio:
"ll'S Ni2
"12 - -r^2— (15)
N E the number of particles in (unmeasured) range 1 of batch 2,
N^ E the number of particles in range 1 of batch 1,
N«2 S the number of particles in range i of batch 2,
N S the number of particles in range i of batch 1.
The number of fields-of-view completing a batch were counted and recorded
and the area calculated. The density of the slide was then calculated
as a ratio between the number of particles counted to area scanned. The
location of each field-of-view was recorded to the nearest tenth of a mm
using the movable stage position scale. All slides were labeled and
stored for later access.
** Silverman, L., Billings, C. E., First, M. W. Particle Size Analysis
in Industrial Hygiene. New York and London, Academic Press» 1971.
p. 106-107
69
-------
The theoretical collection efficiency of the microscope slides was in-
vestigated to determine system biases relative to particle size and
emission velocity.
The collection efficiency corrections were applied to the respective
size ranges. Though the optical sizing of the coke oven particles in-
cluded ranges from 4.6 ym and larger, the data analyzed will include
the size ranges where the particles were larger than 18.4 ym. Ranges 1
and 2 were excluded from final computations because:
a) it was found that the collection efficiency drops off
rapidly below 20 ym;
b) the Andersen sampler adequately sized particles up to 20 ym; and
c) it was difficult to confidently identify the smaller
particles due to foreign materials and grease anomalies
on the slides.
Each slide was also analyzed for particles larger than 1000 y with a ten
power magnifier. The number of particles larger than 1000 y was counted
on each slide and the density for particles >1000 y was calculated and
recorded.
All data obtained from optical sizing are presented in Appendix E which
includes:
a) Particulate counts for each test,
b) Particle density for each slide for particles larger than 18 y,
c) Particle density of particles larger than 1000 y,
d) Calculated percentage of particles versus size range,
e) Cumulative number and weight percents versus size range.
Secondary methods of size analyses are described in Appendices B and E.
The methods included:
Brink
Sieve and Sedegraph
Coulter counter
Optical of Andersen and cyclone cateles.
70
-------
Mass Loading - Due to oven constraints and the unique features of emis-
sions produced during a charging operation, particulate sampling was
done anisokinetically, where the sampling velocity was held constant,
independent of stack velocity. To bracket the error created by aniso-
kinetic sampling and determine theoretical isokinetic values, modified
procedures explained in detail in MTR-6288, "Manual Sampling System,
Coke Oven Emissions Test Program," were used and are summarized here.
The theoretical isokinetic mass, EM., can be calculated by the algorithm:
A n
i
J- „ T-"i » ,^, ^ £j_gj
1=1
where
M. = mass of sample collected at location i,
%W. = the weight percent of total mass at a given range,
A. = area of stack at duct i,
a. = area of sampling nozzle at duct i,
Q.- = anisokinetic correction factor for particle size range 1
at location i from a given test.
The correction factor is found as a function of particle size, sampling
velocity and stack velocity and is calculated from:
t=l
and
£or V"
71
-------
The inertial parameter a...(t) can be determined from:
- exp[-4.5A,.(t)]
where
and
X±1(t) = C.508)Fi(t)Y1/(Aig) (20)
. . _ , k. . .69 ft. for Wilputte Emissions Guide
A± = Area of location x ^ ffc2 fm ^^ ^ AISI/EPA Guide
a. = Area of sampling nozzle
%W = Weight percent of mass in size range i at location i.
The values and ranges were calculated beforehand
from Andersen and Optical sizings and assumed con-
stants for all tests. Volume flow data are needed to
determine %W. - .
T = Total time of test measured by Channel 48
V . = Sampling velocity to be computed for each sampling
location for each test
F.(t) = Volume flow at stack conditions computed and stored
during volume flow calculations, ft^ /minute
2
g = Acceleration due to gravity* 980 cm/second
Yi = Average free fall velocity of particles in size
range 1 (in cm/ second)
Ideally, when employing this sampling method in the stack velocity or
volume flow, F..(A) should never go to zero. Correction factors for
F.(t) = 0 are inadequate. However, the condition did exist on a number
of tests and was handled by setting Q., = 0 for small particle ranges
where F (t) = 0 and Q... = 1 for the larger particle ranges.. The
rationale behind this concept is simply small particles (those which act
as gases) remain in the stack after the flow has gone to zero and are
72
-------
sampled. Hence, they mistakenly add some amount of weight to the sample.
The larger particles fell out rapidly and similarly are not sampled. The
amount of mass produced by the large particles is correctly zero for
F1(t) = 0 implying Qi:L(t) = 1.
The particle mass distribution is described by %W where 1=1, 2, ... 13
(number of size ranges). It is calculated as a function of anisokinetic
weight percents found for each test duct i and is found by:
%W Q
I!
%W.
and the average weight percent is ranged over all tests, n, for range 1
is:
n
ZW = £ %Wi;L/n . (22)
1=1
These values for weight percent were calculated and used for aniso-
kinetic corrections, as well as to determine a representative mathe-
matical frequency distribution model.
Optical Monitoring System Parameters
Data analysis for the optical monitoring system was divided into three
phases. The first phase consisted of the necessary operations to con-
vert data in the form of a photographic image to information about the
source light transmission characteristics of discrete -areas of the emis-
sion plume. The second phase included the calculation of mass loading
estimations from the transmission values for discrete areas and the com-
bination of these"estimations to provide mass loading data across the
entire light source image at some instant in time. The third phase in-
volved the estimation of total mass emitted over time by multiplying
the instantaneous mass loading estimations by the velocity at which the
emission plume is passing upward across the light source. The calcula-
tions and quantities involved in each phase will be discussed below.
73
-------
Quantitative data was obtained from the photographic record through the
use of a microdensitometer having a'digitized output. The machine mea-
sured the optical density of discrete areas on the source film image and
produced a record on magnetic tape in the form of an eight-place binary
number. The 256 integral values possible in an eight-place binary num-
ber are uniformly distributed between two bounding optical density
values as part of the initial adjustment of the microdensitometer. The
bounding values for the optical density range were selected so that the
lower value fell below the observed optical density for the photographic
image of the black portion of the light source in the absence of emis-
sions. The upper bound of optical density was chosen so that it was above
the maximum observed photographic image density of the lighted portion of
the light source in the absence of emissions. Further, the photographic
film was processed so that a close to linear relationship existed be-
tween light source image optical density values and the log exposure
values which produced them.
The relationship between the exposure values was established for each
film roll through a 21-step density versus exposure tablet placed on
each roll prior to processing. Each step associated a relative exposure
value with a resulting optical density. From the 21 tablet steps se-
quentially spaced at .15 log exposure units apart, a vendor-supplied
computer program fitted a cubic curve of density versus exposure values.
The program then continued to assign individual relative exposure values
to each of the binary density values between 0 and 255 as represented on
the magnetic tape.
The output values of this program were provided in the form of a punch
card deck and the magnetic tape. The exposure values were positioned in
the table matrix so that their sequential position number (address) cor-
responded to the binary number value associated with that particular
density. Thus, using the binary value as an address in the matrix, the
relative exposure value was found as the numerical contents of that
address.
74
-------
Transmittance "T" is defined as f/fo, where f and fo are the values for
light intensity incident upon the system detector in the presence and
absence of a plume, respectively. The relative exposure values are a
product of light intensity and time:
EV = Meter Candle Seconds.
In this system, distance in meters, which determine apparent intensity
and exposure time in seconds, was held constant. Thus, a change in EV
can be considered to be a change in luminance (candles) caused by a
change in plume transmission characteristics.
In order to facilitate the calculation of "T" from exposure values, a
lower exposure value associated.with zero source light was established.
The density readings of the black area in the absence of emissions was
taken to represent this value. This condition existed on the designated
"base frame" of each test. There were, however, slight differences in
individual density values in the scan record of this area. These could
be caused by machine inconsistencies, anomalies in the photographic
films, or small quantities of emission somewhere along the line of sight
causing a plume-air light return. In order to obtain a representative
zero value, all measured densities across the unobscured black portion
of the bar were averaged as follows:
P
D° = ~,D li (23)
where D -. was the binary density number for a point; and P was the
total number of points scanned across the black area on the base frame.
This value was established for each roll of film or test run. This den-
sity value is assumed to be the image density value of the light source
if transmission were to go to zero and there were no other scattered
light, which appeared to come from the light source position on the frame.
75
-------
This density value was then located in the density versus exposure table
and a corresponding exposure value was determined. This exposure value
was then substituted in the equation for determination of the reciprocal
of transmittance:
i f - f v
*
T f - f
a b
where f , is always the exposure value corresponding to 5 , f is the
\ju O Ocl
exposure value corresponding to the unencumbered point density of the
light source and t& and ffc are equal to the exposure values correspond-
ing to the point image density of adjacent light source and black areas
respectively, in the presence of the emission plume.
The exposure values f 0, f , and f were obtained by using the binary
o^ a D
optical density value of each, record on the magnetic tape as an address
in the exposure value matrix table. The contents of that address is the
corresponding relative exposure value for direct substitution in the
formula above.
As the reciprocal of transmittance was found for each point in the scan,
the natural log of that value was determined and added to an accumula-
tion. When transmission calculations on all the 539 points in a frame
scan were completed, the contents of the log 1/T accumulator was multi-
plied by the constants representing particulate characteristics.
The equation for the calculation of the mass emission rate for each film
data frame is:
M = KAL
where $ represents the mass density of the particles, K is defined as the
specific particulate volume divided by the light extinction coefficient
ratio and AL is the lateral dimension of a small area of the light
source (termed a point) which is represented by each of the measured
76
p
^ In
i=k
[foA
-------
density values (539) across the light source. The values of 1/T were
calculated for each of the measured density points across the light
source and summed as ln(l/T) as described above.
The values of iji directly determines the calculated mass volume from par-
ticulate volume determined by the opacity measurement. If the particles
collected were made up of a single material or compound, the value of
would be simply the mass density of that single material. The particle
population collected on the coke oven was not, however, composed of a
single material. An analysis of particulate samples collected in particle
sizing equipment shows that several different materials were present in
the population. There is, however, a predominance of two materials.
They are carbon (as it occurs in coal) and heavy hydrocarbons or coal
tars. A further analysis shows that the larger particles are composed
mostly of carbon with some tar causing considerable particle agglomera-
tion. The smaller particle size ranges, on the other hand, are composed
primarily of tar globules which are probably condensed from the existing
gas as it reaches the ambient air. The particulate size distribution
from the particle sizing equipment further explains the apportionment
of the material between the two ranges. The combined weight of the
stages shows that 53.4% of the total average sample mass is contained
between 9.4 p and M.OOO y, while 46.6% is contained in the ranges 6.4 p
and below. Using this information, a representative particle mass den-
sity value was estimated. The value used in mass rate calculations for
3
the optical system is 1.2 gm/cm . It is quite simple, however, to de-
termine mass rates for other assumed values of density, since this factor
is purely multiplicative in the mass rate equation.
The determination of an appropriate value for the constant K is compli-
cated by the complex particle material makeup and the distribution of
particle size. The value of K is mathematically evaluated using the
equation:
77
-------
K = ± -
K 3
exp -
21n2 o
dr
(26)
rQE(d,m) exp -
21n2 a
g
dr
in which r is the radius of a particle, r is the simple geometric mass
8n
mean radius and a the standard deviation of a lognormal particle size
g
distribution, and 0_ is the extinction coefficient assigned or CalCU-
Ci
lated for size distribution. It was shown by Conner and Hodkinson* and
Pilat and Ensor** that (X, can reasonably be set equal to 2 for large
Ci
particles (larger than 2-4 y) without serious error in calculations.
This approach cannot, however, be used for small particles since their
Q value is much more strongly influenced by their material makeup,
their complex index of refraction, and the wavelength of the light source
which they attenuate.
Measurements and analysis of the particle size distribution have shown
that the size frequency curve is bi-populate in nature. It has also been
demonstrated through graphic and chi-square tests that each half of the
combined curve generally fits a lognormal distribution.
The upper portion of the combined curve representing the larger particles
has a geometric mass mean radius of ^100 y. This portion of the parti-
cle population can be expected to have a value of Q_ approximately equal
to two. The lower portion of the combined curve, however, has a geo-
metric mass mean radius of .85 y. With this mean radius, it is improper
to assign a value of z for Q_ of these particles.
* Conner, W. D. , Hodkinson, J. R. Optical Properties and Visual Effects
of Smoke Stack Plumes. U. S. Department of Health, Education, and
Welfare, Cincinnati, Ohio. Publication #PB 174-705. 1967.
** Ensor, D. S., Pilat, M. J. Calculation of Smoke PlumeOpacity from
Particulate Air Pollutant Properties. Journal of Air Pollutant Con-
trol Association. Volume 21, Number 8, August 1971.
78
-------
In order to develop a K value for the total particle population, K was
determined for each of the two particulate mean values and the observed
variance about that mean. The upper portion around 100 p was assumed to
be made up primarily of carbon with an index of refraction of 1.96-0.66 i.
The lower portion of the population was assumed to be primarily coal tar
with an estimated value for the complex index of refraction of
-4
1.54 - 1.9x10 i. The values of K found for each half were combined
using the particle number distribution as a weighing factor.
Particle number distribution showed 4% of the total particles to be in
the upper distribution, while 46% was contained in the lower portion.
The K value for the upper portion was found to be 28, while the lower K
value was estimated to be .22. The combining of these two values using
the associated population percent as a weighing factor yields a combined
value of 1.33 for K. This value was used in the mass rate calculations.
Since this factor is purely multiplicative, mass rate for other K values
can be easily estimated.
The last constant, AL, is the lateral width of a single sample point.
The microdensitometer was set up to take 539 equal width samples across
the 48.' light source. The equation for AL is therefore:
AL = 48' x 3048 M/ft . ^ ^ (2y)
This is the value for AL used in the mass rate calculations across the
horizontal bar. The same AL can be used for rate calculations, involving
the vertical "wing" light sources.
In order to calculate a total mass emitted over any period of time dur-
ing a charge, the instantaneous value of mass emission rate is multiplied
by the plume vertical velocity or plume rise velocity. In the original
system concept, the value or values for plume velocity was to be deter-
mined using motion pictures of the actual plumes. This approach was
modified slightly to utilize the system sequence camera instead of a
motion picture camera.
79
-------
In order to measure the vertical progress of the plume, an illustrated
scale was constructed and placed on one of the equipment towers approxi-
mately 30 feet above the top of the oven. The bar consisted of ten
small light bulbs spaced one foot apart. As viewed from the coal bin
balcony, the scale started about one foot below the light bar and ex-
tended upward about nine feet ab'ove it. This scale was used to calibrate
the visible vertical area included in each frame which extended about
30 feet above the image of the bar.
In order to prevent special distortion, emission plumes were chosen
which were rising near vertically (not being blown by wind) in the
vicinity of the illuminated scale (about the same distance from the
camera). An additional consideration in the selection of emission
periods was the visibility of the illuminated scale during at least some
portion of the period recorded on film. It was not necessary that the
bar be visible the entire period since only two or three frames per
period were required to calibrate the viewing screen.
The technique used in photographing the plume involved observing the
emissions until a vertically rising plume developed with a number of
distinguishable features such as puffs. The sequence camera was then
operated at either one or five frames per second for some period of time -
usually from 30 seconds to one minute. The record of the vertical motion
was thus made at known time intervals.
The resulting film was processed and viewed using a standard 35mm pro-
jector and a square grid as a viewing screen. The projector was adjusted
with respect to the screen until the distance between two adjacent bulbs
on the scale image correspond to some convenient number of viewing screen
grid dimensions. This allowed a direct estimation of the vertical rise
distance of some observed feature on the plume.
A total of 60 observations were made on five different charges. The arith-
metic mean value was calculated for those observations. This value was
6.63 feet/second free rise velocity for the plume in the vicinity of the
light source. A one a value for the population was also calculated. These
values were used for mass emission calculations in the optical program.
80
-------
SECTION V
RESULTS
This portion of the document summarizes the final results of the data
analyses. A comparison is provided of gaseous emissions released by the
AISI/EPA car and Wilputte car along with a comparison of gaseous emissions
that were measured by both the continuous and manual sampling systems.
The results of the particulate analyses are provided in two parts. First,
particle size information is given for a number of sizing methods used
during the program. An overall frequency size distribution is also deter-
mined. Following this, a comparison is given between the total particu-
late mass emitted during a charging operation of the AISI/EPA car to that
emitted by the Wilputte car.
The optical monitoring results are also presented. The results are
evaluated and compared to the in-stack measurement system results. The
feasibility of this system for general use in source monitoring is also
discussed.
COMPARISON OF CONTINUOUS MEASUREMENT AND MANUAL GAS SAMPLING METHODS
The continuous monitoring system was used to obtain gas measurements on
27 charging operations, 25 of which produced useful data. The manual
sampling system was in operation on 16 of the 27 tests, and data from 10
of the tests were analyzed.
Table 6 lists the constituents measured by the continuous and manual
sampling systems. The data from manual and continuous gas analyses is
presented in Appendices B and C, respectively. Of the 16 gases, eight
gases were analyzed by both the continuous and manual systems. In the
presentation of results, either the continuous or manual values were used,
based on the comparison of the systems as described below. The results
for both manual and continuous analyses are shown in Tables 7 through 14.
In these tables, the columns labeled "Interval Average" refer to average
values calculated over the period during which manual gas sampling was
taking place. Under the "MITRE Values" heading, the "Interval Average"
81
-------
TABLE 6
CONSTITUENTS MEASURED BY THE CONTINUOUS AND
MANUAL SAMPLING SYSTEMS
Constituent Continuous
CO X
CO- X
2
SO- X
2
H_S X
2.
THC X
CH.
4
NO X
NO X
X
NO
2
NHL
3
H00 X
2
HCN
H0 X
2
0. X
2
pyridine
phenols
Manual
X
X
X
X
X
X
X
X
X
X
X
X
X
X
82
-------
TABLE 7, S02
MITRE*
/MRI#
5/W4
6/W5
7/W6
10/W10
11/W11
19 /Kl
21 /K3
23/W12
24 /K.4
* Detectabl
** Approxima
MITRE VALUES IN PPM
TEST
AVERAGE
427.9
76.1
496.9
221.8
388
120.1
1453.4
3.8
188.9
e Limit - Mini
ted Values - I
INTERVAL
AVERAGE
<380 *
< 69.35
450 **
125 **
425 **
19.22
450 **
2.35
250 **
mum detectab
ata not comp
HIGH
1088.1
293.6
857.5
454.2
654.1
381.6
3485.7
14.0
833.3
Le quant
Letely r
LOW
0
0
256.2
0
0
4.7
8.4
0
18.8
Lty cala
:duced d\
MRI VALUES IN pptf
INTERVAL
AVERAGE
7.033
159.99
28.69
• 9,55
22.33
23.63
14.85
28.34
ilated for sam{
e to obvious ;
BAG
A
<4.5 *
87.51
< 4.5 *
< 9.7 *
< 7.3 *
25.4
21.03
<23.2 *
< 6.3 *
le size
nstrument
BAG
B
9.2
232.47
< 19.16 *
< 6.6 *
24.75
9.72
< 6.5 *
^Deration;*!
BAG
C
<7.4 *
69.1
< 4.5 *
21.32
48.25
<50.37 <
nfnh] pfflfi
BAG
D
< 22 *
< 19.8 *
17.84
< 15.53*
-------
TABLE 8,
oo
1
MITRE*
/MRI#
5/W4
6/W5
7/W6
8/W7
10/W10
1/WH
19 /Kl
21/K3
23/W12
24/K4
•
MITRE VALUES IN %
TEST
AVERAGE
.1
0
0
0
0
0
0
2.2
.2
.1
INTERVAL
AVERAGE
.0497
0
0
.01
0
0
0
1.57
.175
.081
HIGH
1.0
0
0
.2
0
0
0
9.0
2.3
.6
LOW
0
0
0
.1
0
0
0
0
0
0
MRI VALUES IN %
INTERVAL
AVERAGE
.0933
.315
1.66
.54
.38
.977
.0175
2.038
.875
1.9
BAG
A
0
.17
1.8
.48
.21
2.13
.01
.41
.19
.48
BAG
B
.28
.46
.01
.13
.51
.3
.03
.36
1.56
BAG
C
0
.43
1.23
.42
.06
.02
3.54
1.56
3.32
BAG
D
4.39
.32
1.42
.01
3.84
-------
TABLE 9, CO
MITRE*
/MRI#
5/W4
6/W5
7/W6
8/W7
19 /Kl
21/K3
23/W12
24 /K4
MITRE VALUES IN %
TEST
AVERAGE
.68
1.08
1.18
1.22
.41
1.12
.69
.41
INTERVAL
AVERAGE
.6
1.0
1.2
1.2
.1
1.3
.7
.6
HIGH
2.02
2.15
2.88
3.34
2.36
2.53
2.95
1.63
LOW
.02
.58
.13
.20
.04
.08
.10
.01
MRI VALUES IN %
INTERVAL
AVERAGE
.02
.99
1.15
.63
.02
.93
.27
.26
BAG
A
.01
.77
1.53
.34
.02
.5
.2
.1
BAG
B
.04
1.22
.12
.17
.02
.15
.33
.43
BAG
C
.01
1.4
1.34
.03
1.96
BAG
D
1.53
.68
.01
1.11
00
tn
-------
TABLE 10, CO
oo
o\
MITRE*
/MRI#
5/W4
6/W5
7/W6
8/W7
10/W10
11/W11
19 /Kl
21/K3
23/W12
24/K4
9/W9
MITRE VALUES IN %
TEST
AVERAGE
.0
3.2
2.1
1.2
.7
.8
.2
.4
.0
.3
.0
INTERVAL
AVERAGE
.1
3.1
2.1
1.3
.3
.9
.0
.5
.0
.6
.0
HIGH
.4
8.1
6.3
4.5
2.0
1.9
.9
1.1
.0
1.8
.0
LOW
.0
.6
.0
.0
.0
.0
.0
.0
.0
.0
.0
MRl VALUES IN %
INTERVAL
AVERAGE
.21
2.7
1.9
.9
.7
.6
.1
.7
.4
.6
.2
BAG
A
.06
2.33
.96
.74
.29
.86
.07
.22
.42
.15
BAG
B
.28
3.06
2.14
.27
.56
.07
.14
.16
.4
.95
BAG
C
.29
2.77
1.4
1.2
.15
.12
1.24
BAG
D
1.84
1.3
1.39
.11
1.14
-------
TABLE 11, THC
MITRE*
/MRI#
6/W5
.7/W6
19 /Kl
21/K3
23/W12
24/K4
MITRE VALUES IN %
TEST
AVERAGE
.73
1.57
1.12
3.62
3.00
1.78
INTERVAL
AVERAGE
.69
1.23
.14
3.45
2.50
2*55
HIGH
1.84
6.34
7.00
7.00
13.96
6.98
LOW
.00
.00
.00
.00
.00
.00
MRI VALUES IN %
INTERVAL
AVERAGE
.93
1.44
.09
5.77
.87
1.51
BAG
A
.68
.8
.11
7.41
.26
.44
BAG
B
1.18
.15
.09
.52
1.48
2.59
BAG
C
2.08
.10
8.1
BAG
D
2.72
.06
7.05
00
••J
-------
TABLE 12, <)„
MITRE*
/MRI#
5/W4
6/W5
7/W6
8/W7
10/W10
11/W11
19/K1
21/K3
23/W12
24/K4
MITRE VALUES IN %
TEST
AVERAGE
18.1
16.7
15.1
14.7
14.4
16.5
18.9
17.5
18.0
19.6
INTERVAL
AVERAGE
18.19
16.96
15.22
15.14
15.4
16.25
19.46
17.72
17.98
16.37
HIGH
18.7
21.6
19.0
18.7
16.9
18.9
19.6
19.6
19.3
20.3
LOW
15.0
5.8
9.7
7.5
11.0
13.9
15.8
15.3
13.9
18.3
MRI VALUES IN %
INTERVAL
AVERAGE
16.29
13.52
14.72
15.87
15.69
15.64
16.13
15.075
16.195
15.39
BAG
A
16.56
15.75
15.23
15.58
16.1
15.33
16.03
14.85
16.05
15.76
BAG
B
15.71
11.29
16.59
16.33
16.27
16.01
16.42
16.41
16.34
BAG
C
16.6
12.99
15.96
14.7
16.22
15.96
14.20
15.02
BAG
D
14.06
15.6
15.0
16.11
14.84
00
00
-------
TABLE 13, NO PLUS NH
x 3
MITRE*
/MR!#
5/W4
6/W5
7/W6
8/W7
10/W10
11/W11
19 /Kl
21/K3
23/W12
24/K4
* Detectable
MITRE VALUES IN PPM
TEST
AVERAGE
22.2
42.2
124.9
68.8
57.9
82.7
3.4
113.7
21.4
.6
i Limit - Minii
INTERVAL
AVERAGE
22
40.9
95.7
65.2
50.6
91.2
.5
42.5
25.1
1.6
urn detectab]
HIGH
47.9
103.8
483.8
144.3
120.5
365.4
14.0
319.0
70.3
11.3
e quanti
LOW
0.0
1.4
6.0
0.0
0.0
0.0
0.0
0.8
0.0
0.0
ty calcu
MRI VALUES IN PPM
INTERVAL
AVERAGE
< 4.89 *
< 37.07 *
<140. *
<208.0 *
< 4.4 *
< 37.3 *
< 8.16 *
< 22.4 *
< 8.3 *
< 31.1 *
.a ted for samp.
BAG
A
.e size
BAG
B
BAG'
C
BAG
D
00
-------
TABLE 14, NO
MITRE #
/MRI#
5/W4
6/W5
7/W6
8/W7
10/W10
11/W11
19/K1
21/K3
23/W12
24 /K4
* Detectable
MITRE VALUES IN PPM
TEST
AVERAGE
22.2
42.2
124.9
68.8
57.9
82.7
3.4
113.7
21.4
.6
Limit - Minimi
INTERVAL
AVERAGE
22
40.9
95.7
65.2
50.6
91.2
.5
42.5
25.1
1.6
m detectable
HIGH
47.9
103.8
483.8
144.3
120.5
365.4
14
319.0
70.3
11.3
quantit
LOW
0.0
1.4
6.0
0.0
0.0
0.0
0.0
0.8
0.0
0.0
y calcul
MRI VALUES IN PPM
INTERVAL
AVERAGE
< 3.5 *
34.5
<124.7 *
<151.7 *
< 2.18 *
< 29.6 *
< 1.96 *
< 4.7 *
< 2.38 *
< 1.94 *
ated for sampl
BAG
A
< 2.16 *
<12.48
< 2.19 *
< 2.17 *
< 2.17 *
< 2.16 *
< 1.96 *
< 1.96 *
2.79
< 1.95 *
! size
BAG
B
6.2C
56.59
6.3
< 2.18 *
< 2.18 *
< 2.15 *
< 1.97 *
11.26
< 1.97 *
< 1.93 *
BAG
C
<2.17 *
<2.19 *
301.28
<2.18 *
3.09
<1.94 *
2.8
BAG
D
488.4
301.28
110.95
<1.96 *
2.78
vo
o
-------
values are an average of five second averages (see Appendix C, page C-123)
during the manual sampling period. "Test Average" values are averages of
five second averages over the total test period. Under the "MRI Values"
heading, the "Interval Average" is a simple average of the measured con-
centrations in the sample bags. Thus, only the "interval" values are
comparable, whereas the "test" value represents an average over a longer
period of time.
Also included in these tables are values of detectable limits for individ-
ual samples. This value is a function of sample size and is included to
indicate that concentrations of the particular constituent below this
level may have been present, but would not have been detected by the
analytical techniques utilized. Thus, this value does not indicate the
absolute absence or verified presence of the constituent.
In general, good agreement was observed between manual and continuous
data for CO, C02, THC, and 0^. For these four gases, it was felt both
systems were producing results that agreed within the constraints of the
system. Generally, large disagreement for these gases on specific tests
could be traced to some specific problem occurring during the test. Good
comparisons were also found between the results of each test and the
general qualitative descriptions of the emissions recorded on the voice
tapes. The emission volumes determined for each of these gases were cal-
culated using the continuous results. This was done because more con-
tinuous measurements were performed than manual, and hence, offered a
large statistical sample.
The continuous SO- and H«S data when compared with the manual data were
^ ^
found to be orders of magnitude higher. As a result, a number of labora-
tory tests were performed. These tests established the fact that certain
aromatic hydrocarbons, probably present in coke oven emissions (especially
phenol), interfere with the UV bands used to measure E^S and S02; (215 m
and 280 my, respectively). Therefore, even with the constraint of a
smaller statistical sample, it was necessary to use the manual results to
determine the typical SO- and H_S emission volumes per charge.
91
-------
Similarly, interference problems, as well as calibration problems, for
the H« analyzer resulted in considerable disagreement between manual and
continuous data, and it was concluded that manual data should be used for
any computation requiring hydrogen.
Comparison of NO data shows values for certain tests in agreement, while
j£
others differ by as much as an order of magnitude. The data from both
systems were compared to qualitative descriptions of each test, and it
was determined that there were inconsistencies in the manual sampling re-
sults. As a result, it was concluded that the continuous measurements
would be used for NO .
3C
In summary, of the eight gases measured by both systems, the continuous
results were used in the data reduction procedures for CO, CO-, THC, 0_,
and NO ; and the manual data were used for H_S, S09 and H_.
X £ £, £m
COMPARISON OF GASEOUS EMISSIONS FOR WILPUTTE AND AISI/EPA CAR
During the field measurement program, it was not possible to monitor all
emission points simultaneously. Therefore, the estimate of total gaseous
emissions released by the car was prepared using aggregate data on mea-
sured gaseous concentrations from the series of individual tests. The
average volume flow for each emission point was calculated for each test
and summed to determine the average total volume flow of gases emitted
during a charge. These values were multiplied by the average concentra-
tion of the individual gases and by the test times to calculate the
volume of gaseous constituents emitted per charge. All volume calcula-
tions are based on average test times of 2.61 minutes and 3.48 minutes
for the AISI/EPA and Wilputte car, respectively. A charge or test is
defined as the period of time beginning when the larry car starts to un-
load its coal and ending when the lids are replaced by the automatic lid
lifters (for the AISI/EPA car) or 15 seconds after all the coal is dumped
(for the Wilputte car). Since the Wilputte car had to move away from' the
ports before the lids were replaced, and since monitoring could be per-
formed only when the car was in place, the Wilputte test time was extended
to allow for the emissions that would normally occur before the lidman
replaced the lids. Fifteen seconds was considered the very minimum time
92
-------
in which a man could replace the lids and only under the very best con-
ditions. Table 15 lists the average volumes for gases monitored by the
continuous measurement system for the Wilputte car and Table 16 lists
them for the AISI/EPA car. The. measured and corrected values represent
flows and volumes calculated using measured and corrected (Equation 10)
concentrations, respectively. The measured value can be interpreted in
most cases as a lower limit value, while the corrected would indicate
the upper limit of the-emission levels. It can be assumed that the
volume of emission escaping through the measurement points would lie in
a range somewhere between the measured and corrected values. Hence,
the closer the measured value is to the corrected value, the more confi-
dence can be placed in the results. When the measured value is very
close to the corrected value, this is an indication that the error
caused by dilution of the sample with ambient air was minimized.
During testing of the AISI/EPA car, a combination of conditions caused
number 3 drop sleeve during Test 21 fail to seal. The emission guide
surrounding the drop sleeve presented detection of the faulty seal
until the charge had begun. Hence, no alternative measures could be
taken to correct this condition. Therefore, this test was not considered
a fair representation of normal AISI/EPA emissions, and Table 17 lists
AISI/EPA volumes excluding Test 21 values from the averages. The re-
sults are presented with and without Test 21 included, so as to give a
representation of average emissions under poor conditions and under a
more typical set of conditions which might require a reasonable amount
of care to obtain.
Table 18 is based on corrected values and compares the Wilputte car to
the AISI/EPA car for NO , NO, CO., CO and THC. As noted on Table 18,
X • £•
the volume of NO , NO, C0_, CO and THC was shown to decrease signifi-
x 2
cantly with the new car. Emission reduction ranged from 14 to 99%, with
the greatest reduction being in NO emissions. The smallest reduction of
these five constituents was still very significant, THC average emission
volume for the new car averaged between 14% and 40% less than the other
car.
93
-------
TABLE 15
EMISSION VOLUMES FOR WILPUTTE CHARGING
Constituent
NO
X
NO
co2
CO
THC
Average Volume Emitted/Charge (SCF)
Measured
.1107
.0783
29.00
16.94
33.28
Corrected
.1144
.0802
29.36
17.49
33.77
TABLE 16
EMISSION VOLUMES FOR AISI/EPA CHARGING INCLUDING TEST 21
Constituent
NO
X
NO
co_
2
CO
THC
Average Volume Emitted/Charge (SCF)
Measured
.0456
.0059
4.05
8.18
26.69
Corrected
.0492
.0061
4 ,71
9.04
29.12
94
-------
TABLE 17
EMISSION VOLUMES FOR AISI/EPA CHARGING EXCLUDING TEST 21
Constituent
NO
X
NO
co2
CO
THC
Average Volume Emitted/Charge (SCF)
Measured
.0169
.0008
3.08
5.30
17.09
Corrected
.0211
.0011
3.83
6.28
20.43
TABLE 18
PERCENT REDUCTION OF GASEOUS EMISSIONS
; •*,
Constituent
NO
X
NO
co2
CO
THC
(Wilputte Volume-AISI/EPA Volume) /Wilputte Volume x 100
% Reduction
56 . 8
92.4
83.9
48.3
13.7
% Reduction w/o Test 21
81.6
98.6
86%9
64.1
39.5
95
-------
Table 19 lists the average, maximum and minimum concentration for all
the gaseous pollutants measured. The first five constituents were
measured by the continuous monitoring system. The remaining seven con-
stituents were measured by the manual sampling system. It should be
noted that the averages of gases measured by the manual sampling system
was based on 10 tests for the Wilputte car and three tests for the
AISI/EPA, while the continuous system was operational on 18 Wilputte
tests and 10 AISI/EPA tests.
Data for SO-, H.S, CH,, NH-, phenol, CN and pyridine are presented in
Appendix C. For a substantial number of tests, measurement of these
seven constituents produced concentrations below the detectable limit.
It was felt, due to the small number of tests on which these consti-
tuents were measured and the large number of times they were below de-
tectable limits, that no calculation of volume would be truly repre-
sentative of their emissions. Instead, the comparison of cars was made
on a measured concentration basis. Table 19 shows that the maximum
concentration measured, excluding Test 21, was larger for the Wilputte
car in every case. When Test 21 is involved in the comparison, the
concentrations were reduced for the AISI/EPA car in every case except
for the constituents of H~S and CH,.
In general, gaseous emissions from the AISI/EPA car were significantly
reduced between 14% and 99% for NO , NO, C0_, CO and THC, while for the
X ^
other seven pollutants measured, a qualitative evaluation indicates
reductions of better than 15% in all cases.
RESULTS OF PARTICLE SIZE ANALYSES
Along with the obvious health reasons for interest in the size distri-
bution of coke oven particulate matter, its characterization is re-
quired to predict the physical behavior of particles relative to aero-
dynamics and optical characteristics.
96
-------
TABLE 19
CONSTITUENT CONCENTRATIONS (MEASURED)
CONSTITUENT
NO (ppm)
NO (ppm)
ro (7\
V->U _ 1 /o 1
2
CO (%)
THC (%)
S02 (ppm)
H2S (ppm)
CH4 (%)
NH3 (ppm)
Phenol (ppm)
CN (ppm)
Pyridine (ppm)
WILPUTTE
Average Maximum Minimum
38.9 484 0.0
15.6 336 0.0
.5 8.1 0.0
.69 3.76 0.0
1.6 13.98 0.0
* 232.5 BDL
* 42.5 BDL
.62 4.3 .01
* 130.6 BDL
* 31.1 BDL
* 16.5 BDL
* BDL BDL
AISI/EPA
Without Test #21
Average Maximum Minimum
9.6 70.2 0.0
.13 20.3 0.0
.11 1.5 0.0
.21 2.2 0.0
.59 11.4 0.0
* 25.4 BDL
* BDL BDL
.53 1.77 .01
* BDL BDL
* BDL BDL
* BDL BDL
* BDL BDL
With Test #21
Maximum Minimum
319 0.0
52.8 0.0
1.5 0.0
2.53 0.0
11.4 0.0
48.3 BDL
72.27 BDL
7.79 .01
12.37 BDL
17 . 72 BDL
5.48 BDL
BDL BDL
\0
A large number of tests were below detectable limits (BDL) and no average
could be computed.
-------
The particle size distribution determined is also required for use with
anisokinetic mass loading and volume flow data to produce a set of
calculated isokinetic mass values, and with optical data (opacity mea-
surements) to calculate theoretical mass emissions.
In the calculation of theoretical isokinetic mass data from anisokinetic
mass values, the size distribution of the sample is a critical factor.
Anisokinetic sampling will result in correct particulate concentrations
if the particles are aerodynamically small. For most conditions,
particles <5 y may be considered as small particles. The mass of large
particles (those which move independently to gas stream) will produce
the correct concentration values if sample volume calculations are made
using stack velocity instead of sampling velocity. For the average
particle density of coke oven emissions, particles <200 y can be con-
sidered large particles which move independent of the gas stream. How-
ever, for particles between 5 and 200 y, anisokinetic sampling can pro-
duce results far different from isokinetic sampling. Hence, to approxi-
mate isokinetic results, corrections are required and are partly depen-
dent on the particle size distribution. The relations and calculations
are discussed in the Analytical Methods section of this report.
A particle number distribution is also required to relate absorption and
scattering effects measured by the optical measurement system to the
mass of particulate matter in the plume.
Five methods of particle analysis were used during this test program:
Method Range Measured Number of Tests*
Anderson ^20y * .5y 10
Brink -v 3y ^ .3y 2
Sieve & Sedegraph 325 mesh to .6y 2
Coulter ^lOOy to ly 1
Optical 500y to 4y 9
98
-------
Though data from all the methods were compared and considered, the
Andersen data and optical analysis data of samples taken by the MITRE
Carrousel were the primary methods used for development of the particle
distribution. The following paragraphs explain the rationale for this
decision, as well as discuss the methods and sample handling techniques.
Both the Andersen and the Brink samplers use aerodynamic sizing methods.
The sample is simultaneously collected and sized. Weighing of the col-
lection stages is performed within a few hours of the tests, and no
redispersion of the sample is needed. The Andersen sampler was used on
a total of ten tests (all were performed in the stacks or guides). Due
to underloading or overloading or anomalies in stack velocities, five
of the tests produced suspected data and were not used in the final
compilation of the data. The Brink sampler was used a number of times,
producing data from six tests. Three of the six tests concerned sam-
ples taken from emission plumes 20 feet above the car; two were of leak-
ing door tests and only one test concerned sampling from the emission
guides. The emission guides and stack areas are of most importance,
since the mass loading samples were taken from these areas.
Sieve, sedegraph, Coulter and some optical analyses were performed on
samples redispersed after various forms of collection. It is intuitively
felt that dissolving the particulate matter could seriously alter the
particle size distribution by reducing aggregation and fleecing, and
hence produce biased results.
This number indicates the total number of both "good" and suspect
samples taken at emission guides and stacks. Test of leaking doors
of free plumes, etc., are not listed here.
99
-------
A number of the optical sizings were performed on samples collected on
field constructed equipment (MITRE Carrousel) specially designed to col-
lect the particles and retain them in a form that could be directly used
for optical size analyses. The resulting optical size distribution was
used to develop a weight distribution and combined with the Andersen
results to produce a typical particle size distribution.
Data from the Andersen and MITRE Carrousel optical analyses are listed
respectively in Appendices B and £. The Andersen sampler was used on
ten tests. Only five of the ten were acceptable for determining a dis-
tribution function. Table 20 lists all ten tests, location of samplers
and reasons why certain tests were suspect and, hence, not included in
the average. The values of the remaining "good" tests were combined
with the optical results and corrections applied to the anisokinetic
distribution to determine an average calculated isokinetic distribution
shown in Table 21.. The cumulative distribution of Table 21 is plotted
in Figure 18. Inspection of the figure implies that the distribution
could fit a bi-populate lognormal distribution* To check this, the distri-
bution was broken into two separate distributions - one between Range 1
and Range 7 (containing ^46.6% of sample by wt.) and the other Range 8
to Range 12 (containing ^53.4% of sample by wt.). These were then nor-
malized (Table 21 ) and plotted on log probit paper (Figures 19 and 20).
As expected, a straight line can be drawn through the data points-im-
plying that both curves can be approximated by log-normal distributions.
The lower range curve has a mass mean diameter of about 8.5 ym and a
sigma of about 2.5, whereas the larger range curve has a mass mean
diameter around 235 ym and a sigma of about 3.9. The overall mass fre-
quency distribution can be described by:
100
-------
TABLE 20
ANDERSEN ANALYSIS
Test Number
and Location Status and Reason
7-1 Test was acceptable.
8-2 Test was acceptable.
Test was suspect due to stage over-
8A - 3 loading (the 2/3 stage had more
than 120 mg on it).
9-1 Test was suspect, load on stage
was very light (<.2 mg) and emis-
sion velocity was zero during much
of the test making calculation of
theoretical correction factors
difficult.
10-6 Test was acceptable, loading was
light.
11-5 Test was acceptable, loading was
lighter than ideal.
23-2 Test was acceptable, one stage
suffered minor overloading.
20 - K2 Test was acceptable.
21 - K3 Test was suspect, stages were
over-loaded (as much as 590 mg).
24 - Kl Test was suspect, stages under-
loaded (some as light as .02 mg).
101
-------
TABLE 21
SIZE DISTRIBUTION DUE TO COKE OVEN CHARGING
Assigned
Range Numbers
Range Mean
Particle
Diameter p*
Weight % in
Size Range
Cumulative
Weight % <
Upper Range
Bound
1
Filter
14.7
14.7
2
.584
4.3
19.0
3
.883
6.2
25.2
4
1.387
11.8
37.
5
2.631
6.5
43.5
6
4.313
1.9
45.4
7
6.412
1.2
46.6
8
9.433
1.7
48.3
9
14.791
4.5
52.8
10
45
3.3
56.1
11
112
11.0
67.1
12
273
18.7
85.8
13
700
14.1
99.9
NORMALIZED SEPARATE POPULATIONS OF BIPOPULATE DISTRIBUTION
Lower Mode
Weight % <
Upper Range
Bound
Upper Mode
Weight % <
Upper Range
Bound
31.5
40.7
54.0
80.6
93.3
97.4
100.0
3.2
11.6
17.8
38.4
73.5
99.9
46.6%
of
Sample
53.4%
of
Sample
* Assumed 1.2 gin/ml density. Some overlap of ranges occurs in sampling/sizing equipment. As a result, the mean
value must be derived from theoretical range bounds.
-------
1000
800
600
400
200
100
80
60
40
20
Q
H 10
S 8
1
0.8
0.6
0.4 -
0.2
0.1
0.01
•A
•A
• A
KEY
A UPPER LIMIT CUMULATIVE
• LOWER LIMIT CUMULATIVE
• A
• A
0.2
10 20
40 60
80
95
99
99.9 99.99
% OF SAMPLE > SIZE
FIGURE 18
COKE OVEN PARTICULATE DISTRIBUTION
103
-------
1,000
800
600
400
200
100
60
40
20.
10
8
6
KEY
A UPPER LIMIT CUMULATIVE
• LOWER LIMIT CUMULATIVE
0.01
0.2
10 20 40 60 80
% OF SAMPLE > SIZE
95
99
99.9 99.99
FIGURE 19
LOG NORMAL PLOT OF RANGES 1 THRU 7
104
-------
100
80
60
40
20
10
8
2
1
0.8
0.6
0.4
0.2
0.1
0.01
KEY
A UPPER LIMIT CUMULATIVE
• LOWER LIMIT CUMULATIVE
0.2
10 20 40 60 80
% OF SAMPLE > SIZE
95
99
99.9 99.99
FIGURE 20
LOG NORMAL PLOT OF SIZE RANGES 8 THRU 13
105
-------
f (D) = W.,fLCD) + W2fu(D)
D 2TT Inc^ ^ I 2 Inc.
ln2(D/y2)
ln2(D/y)
(28)
where it is assumed f (D) is independent of f (D)
Li U
f(D) = mass frequency distribution of bi-populate lognormal e
f1(D)= mass frequency distribution lower lognormal curve
f (D)= mass frequency distribution of upper lognormal curves
D = diameter of range
W- = weight % of lognormal portion of bi-populate centered
about .85 y (46.6%)
W_ = weight % of lognormal portion of bi-populate centered
about .235 y (53.4%)
y.. = mean of lower curve (.85 y)
y_ » mean of upper curve (235 y)
a = sigma of lower mode, 2.5
a_ = sigma of upper mode, 3.9
This distribution function would seem to give the best possible general
description of the coke oven charging particulate emissions based on the
collected data. However, it should be noted that this distribution is
based on an average density of 1.2 gm/ml. Though the measured distribu-
tion fits the bi-populate lognormal form reasonably well, it is based on
only five tests which cannot be considered the most ideal statistical
example. Therefore, the bi-populate lognormal frequency distribution is not
106
-------
presented as the absolute particle distribution, but rather as a simpli-
fied mathematical model to be used as a representation of coke oven
charging particulate distribution.
Appendix E gives the particle distributions from Brink samplers for
emission guide, boom, and leaking door tests. Due to the limited range
of the Brink sampler, the data concerns only the lower lognormal portion
of the bi-populate distribution and, hence, can only be .compared against
this portion of the curve. Furthermore, only Brink Test #4 can be used
as a comparison because it was the only test performed at the emission
points of the charging operation. Figure 21 is a plot of the cumulative
weighted percent of Test 4. Assuming a lognormal distribution, a straight
line drawn through the points produce a mean of about 1.2 y and a sigma
of 2.2. This is in good agreement with the calculated lower distribu-
tion of a mean '^.85 y and sigma ^2.5. In addition, the Brink data are
all based on an assumed density of .8 gm/ml and the Andersen data are
based on 1.2 gm/ml. Equating the densities of both methods would tend
to bring them into closer agreement.
The Coulter counter produced a volume mean of ^27 y. The sample was
taken by scraping particulate matter off a 1/4" wire screen which had
been exposed to the emission plume of the Wilputte stack.
Optical analyses were performed on portions of the samples removed from
the cyclone and Andersen sampler. These samples were suspended in ben-
zene when removed from the sampler and were later suspended in isopropyl
alcohol just prior to sizing. The number mean diameter was found to be
between 5.5 and 3 ym or around 30 ym mass mean diameter.
The sieve and sedegraph analyses were performed on particulate matter
trapped in the case of the MITRE Carrousel Sampler (not on the collection
slides of the sampler) . The size distribution of these samples was found
to be bi-populate. Seventy percent of the sample mass was greater than 43 ym
(325 mesh by sieve) and 30 percent less than 3 y with a mean of about 1.3 y
for the portion of the sample less than 3 u. The sedegraph analysis was
performed using water as the medium. This is in fair agreement with the
107
-------
10
8
g
0.8
0.6
O.S
0.4
0.2
0.1
0.2
10 20 40 60 80 90
WEIGHT X OF SAMPLE > SIZE
95
99
99.8 99.9 99.99
FIGURE 21 _
BRINK DATA
-------
Andersen, Brink, and MITRE Carrousel analysis.
STATISTICAL EVALUATION OF SIZE DISTRIBUTION DATA
The arithmetic averages for the acceptable tests including anisokinetic
and calculated isokinetic values are shown in Table 22. Chi-square (x2)
tests were performed on both anisokinetic and theoretical isokinetic
averages against the respective values for individual tests. As shown
in the table, the theoretical isokinetic values produced equal or lower
2 2
X values, hence, higher confidence levels. All x tests on the cor-
rected data pass the goodness of fit test while only two of the aniso-
2
kinetic tests pass the test. To pass the x test implies only that
there is no reason to doubt the distribution fits by the hypothetical
distribution.
The bi-populate lognormal distribution was determined from the values of the
arithmetic average of all the corrected isokinetic values and passes
2
the x goodness of fit test.* However, the data do suggest the distri-
bution varies between guide and stack. The distributions from the
emission guide tend to shift the mean of the lower mode toward a lower
mean and higher standard deviation, while the stacks tend to the other
extreme. Though these differences were noted, there was statistically
insufficient data to produce separate distribution for the guides and
stacks. Therefore, the bi-populate distribution previously described was
used to approximate particulate emissions in all cases.
COMPARISON OF PARTICULATE EMISSIONS FROM WILPUTTE AND AISI/EPA CAR
A number of analyses were performed on the particulate mass samples. The
primary analyses (for mass emitted) was performed on all samples and cor-
rections (Equation 16) for errors due to anisokinetic sampling were ap-
plied. Secondary investigations include analyses for:
* Herdan, G. H., Smith, M. L., Hardwlck, W. H., Conner, P. Small Par-
ticle Statistics. London, Butterworths, 1960. p. 122-125
109
-------
TABLE 22
SIZE RANGE
Assigned Range Numbeis
Range Mean Part. Diam.
Average Anisokinetic
Distribution*
Average Calculated
Isokinetic Distribu-
tion**
iipopulate Lognormal
Approximation at
Mean Upper- 23 5y,
ffn=3.9, Mean Lower=
.85, c?L«2,5
1
Filt.
14.65
14.7
12.58
2
.58
4.57
4.3
8.39
3
.88
7.12
6.2
8.85
4
1.39
14.2
12.4
8.85
5
2.63
6.73
5.9
5.22
6
4.31
2.1
1.9
1.93
7
6.41
1.29
1.2
.84
8
9.43
1.89
1.7
.73
9
14.79
5.13
4.5
2.74
10
45
3.53
3.3
6.62
11
112
11.8
11
12.28
12
275
17.0
18.7
13.99
13
700
11.34
14.1
16.92
Test
07
08
10
11
23
Sipopulate Lognormal
Chi-Square
Average Anisokinetic
17.12
29.52
23.69
21.16
11.25
Test For***
Average Isokinetic
11.04
20.80
16.37
12.96
11.31
* Actual particulate sampling was conducted
at a constant sampling velocity indepen-
dent of variations in stack velocity.
** Theoretical corrections were applied
based on instantaneous emission velocity
measurements to yield a calculated iso-
kinetic distribution
2
*** For 12 degrees of freedom if x is larger
than 21.03 the probability becomes sig-
nificant that the test does not fit the
distribution.
-------
% tar content
benzpyrene content
density
elemental
The major objectives of the secondary analysis were to develop baseline
data on coke oven emissions with emphasis on the detection of potentially
hazardous substances such as carcinogens and heavy metals, and an assess-
ment of the state of the art in techniques for making such measurements
and determinations. This analysis is necessary in any assessment of the
importance and potential effects of coke oven emissions. These analyses
were performed on only a portion of the total number of samples.
MASS EMISSIONS
The average calculated isokinetic mass emitted by the Wilputte car was
814.7 gm per charge, while the like value for the AISI/EPA car was 120
gm per charge. This is a reduction of about 85.2% from Wilputte to
AISI/EPA emissions. These calculated isokinetic mass values were obtained
by applying a theoretical correction process to the anisokinetic data.
All particulate samples were collected under anisokinetic conditions.
Tables 23 and 24 present this particulate emissions data for the Wilputte
and AISI/EPA car, respectively. The values in these tables were calcu-
lated by the following equation which assumes isokinetic conditions:
Ai
EM± - — M± (29)
where EM.» = total mass emitted at location I during the charge,
A£ = area of measurement duct,
Uj - area of sampling nozzle,
M.J = mass of sample collected at location I.
The tables show the average mass emission assuming isokinetic conditions
to be 808.9 gm/charge for the Wilputte car and 245.5 gm/charge for the
AISI/EPA car. These results are incorrect, however, since isokinetic con-
ditions did not exist.
Ill
-------
ro
TABLE 23
ANISOKINETIC MASS EMISSIONS
WILPUTTE
TEST
NO.
2
5
6
7
8
Q
10
11
23
8A
Average
STACK
1
201.5
144.7
42.0
65.0
113.3
STACK
2
250.6
182.0
123.6
294.2
257.3
82.4
198.4
STACK
3
134.5
132.4
269.3
355.3
975.9
374.7
GUIDE
1
41.7
37.6
39.65
GUIDE
2
36.8
28.7
67.9
34.0
41.9
GUIDE
3
38.3
26.2
50.6
48.7
40.9
TOTAL
808.9
-------
TABLE 24
ANISOKINETIC MASS EMISSIONS
AISI/EPA
TEST
NO.
19
20
21
24
Average
STACK
1
N/A*
-
STACK -
2
N/A*
STACK
3
L_ N/A*
GUIDE
1
30.8
9.52
3.3
14.54
GUIDE
2
108.1
4Q.7
62.6
12.2
58.15
GUIDE
3
•na.7
2652.5**
206.9
172.8
TOTAL
245.5
u>
* The AISI/EPA car does not have coking stacks, and as a result these
columns are not applicable
** Test 21 not included in average
-------
A theoretical approximation of isokinetic mass values may, however, be
obtained by applying Equation 16 to the anisokinetic data. Tables 25
and 26 present the calculated isokinetic values of mass emitted for the
AISI/EPA and Wilputte cars, respectively, and from the basis for the
85.2% reduction figure.
A failure of the number 3 drop sleeve to seal during Test 21 may have
been caused by a cocked emission guide. However, detection and hence,
resealing of the drop sleeve was prevented by the emission guide. There-
fore, Location 3 on Test 21 for the AISI/EPA car was omitted from all
the average mass emission calculation. The magnitude of the mass, mea-
sured at 21-3, is another indication that it is not representative of a
normal charge. It is almost two orders of magnitude higher than the
next largest mass value for the AISI/EPA car.
Out of the 10 tests performed on the AISI/EPA car, no charge other than
Test 21 demonstrated emissions of this magnitude.
The accuracy of the calculated isokinetic values in Tables 25 and 26
is difficult to assess. However, a comparison of the anisokinetic mass
value to the calculated isokinetic values can be used to give an evalua-
tion of the relative errors induced by anisokinetic sampling. It can be
assumed that if the calculated value is close to the measured value, the
error is minimized and the mass is representative of the emissions. As
the difference between the calculated mass and measured mass increases,
the values become less reliable. But they are still quite useful, for
the difference between the values is a measure of the direction and mag-
nitude of the error due to anisokinetic sampling. For example, in
Test 10, Guide 1, the anisokinetic value was 41.7 gm; the isokinetic
value was 42.9 gm; the masses are very close. One might, therefore,
assume sampling was near isokinetic and if the sampling velocity
(193.4/min) and the average stack velocity (204.3 ft/min) are considered,
it is obvious conditions are near isokinetic.
114
-------
TABLE 25
CALCULATED ISOKIHETIC MASS
MASS IN GRAMS FOR WILPUTTE CAR
TEST
NO.
2
5
6
7
8
9
10
11
23
8A
Average
-
STACK
1
117.8
132.7
20.2
35.9
76.65
STACK
2
410.6
125.8
77. 4
234.3
141.9
52.5
173.62
STACK
3
221.9
94.1
182.0
213.6
1028.2
434.95
GUIDE
1
42.9
48.4
45.65
GUIDE
2
19.0
28.4
-
71.2
42.0
40.15
GUIDE
3
30.1
29.3
56.9
58.0
43.58
TOTAL
81,4.7
-------
TABLE 26
CALCULATED ISOKINETIC MASS
MASS IN GRAMS AISI/EPA LARRY CAR
TEST
NO.
19
20
21
24
Average
STACK
1
N/A*
STACK
2
N/A*
,
STACK
3
N/A*
,
GUIDE
1
14.38
3
8.3
1.3
8.0
GUIDE
2
50.92
23.20
45.1
5.68
31.5
GUIDE
3
74.26
1736.78**
86.78
80.5
TOTAL
120.0
* The AISI/EPA car does not have coking stacks, and as a result these
columns are not applicable
** Test 21 not included in average
-------
Conversely, there is a large difference between sampling velocity
(470.5 ft/min) and stack velocity (209.9 ft/min) for Test 7, Stack 1. As
expected, there is a large difference between the mass values (~41.6%).
If this rationale is extended by comparing the anisokinetic and isokinetic
values for average mass/charge, the Wilputte anisokinetic and isokinetic
averages differ by less than 1%, while the AISI/EPA values differ by
about 50%. This would imply the Wilputte data offers a somewhat higher
confidence level than the AISI/EPA.
However, the 50% difference found with the AISI/EPA data is still rela-
tively good, considering the sampling problems encountered with coke
oven charging emission monitoring. The data indicates the constant
sampling velocity method employed at the coke oven is more precise than
originally expected.
SUPPLEMENTARY ANALYSES
Supplemental analysis was performed on particulate material collected
from coke oven emission streams as well as prepared coal prior to oven
charging. The intent was to establish the quality and quantity of
trace constituents in the material and access the technique for these
determinations. The body of data presented below represents the results
of this analysis. The difficulties or limitations applicable to each
area are discussed.
Tar roughly defined as the particle fraction soluable in benzene was
found to be a major constituent of the collected material. The average
tar concentration for a particulate sample was 57.1%. No separation
was made between the Wilputte and AISI/EPA emissions due to the small
number of samples analyzed for tar.
Tar analysis was performed on 20 samples which were primarily Andersen
catches. In most cases, the analysis was done on the entire sample
collected excluding impinger catches. However, a few samples were
separated into parts (i.e., in Test 21-3). The Andersen plates were
done separately from probe tip and the Andersen front. Table 27 lists
the test number and % Tar as measured in the entire sample.
117
-------
TABLE 27
TAR CONCENTRATIONS
Test Number
7
8
8A
9
10
11
19
21
Average
Location
1
2
3
1
6
4
1
3
Tar Percentage
29.2
56.1
25.9
75.0
86.2
42.1
64.8
54.5
57.1
118
-------
Polycyclic organic matter present in coke oven emissions will be found
chiefly in the particulate fraction which has been defined as tar.
Several of the more widely accepted carcinogens are species of benz-
pyrene. In order to establish and characterize the presence of this
general group, Benzpyrene Analysis was performed on the tar portion of
a group of samples. These tests indicate that benzpyrene is present in
amounts ranging from 18,000 ppm to less than 260 ppm. (Ippm equivalent
to 1 |JL gram benzpyrene/gram of tar).
Sample sizes and lack of available analytical techniques precluded a
more detailed examination of these samples. The data obtained is presented
in Table 28 and gives the benzpyrene content for samples analyzed prior
to and including the filter. A few analyses were performed on the impinger
catches and are presented in Appendix B, but all measurements of impinger
tests were below the detectable limits. The maximum concentration was
18,000 ppm found on the Andersen plates of Test 21-3. The Andersen front
and probe tip also produced measurable values. The results indicate the
general presence in the emissions of a species containing widely accepted
carcinogenic substances in relatively heavy concentrations. Further con-
clusions are impossible because of the limited data.
An elemental analysis was undertaken to identify the trace constituents
of the coal charged and the resulting emissions. The intended emphasis
was the identification and quantification of hazardous elements, partic-
ularly heavy metals. The emissions analysis was performed on particulate
sizing equipment catches, and was thus limited by the sample size, as
well as available analytic techniques. These limitations resulted in a
large number of instances where constituent concentrations, if they did
exist, were below detectable limits. The emission constituent concentra-
tions which were successfully measured and are reported in Table 29 are
consistent with the constituents identified in the elemental analysis
of the charging coal. The sample numbers can be cross-referenced with
Table 30 to determine the portion of the sample that was analyzed.
119
-------
TABLE 28
BENZPYRENE ANALYSIS
Test
Number
7
8
8A
8A
9
10
10
11
11
19
20
21
21
21
23
Location
1
2
3
3
1
6
4
4
5
1
2
3
3
3
2
Sample
Number
1045
1060
1066
1067
1095
1102
1105
1111
1114
1208
1213
1222
1223
1229
1241
Description
Andersen Front Probe Tip
Andersen Front Probe Tip
Andersen Plates
Andersen Front
Andersen Front Probe Tip
Andersen Front Probe Tip
Probe Tip Cyclone
Probe Tip Cyclone
Andersen Front Probe Tip
Probe Tip Filter
Andersen Plates
Andersen Plates
Andersen Front
Andersen Back
Andersen Plates
Total Benzpyrene
(Pg>
30
70
100
B.D.L.
B.D.L.
30
B.D.L.
B.D.L.
B.D.L.
B.D.L.
B.D.L.
22,000
520
B.D.L.
220
Benzpyrene Concentration1
(ppm)
510
1,000
560
<260
<630
560
<1,300
<1,000
<590
<980
<1,400
18,000
3,000
<610
1,700
* yg of Benzpyrene/gram of Tar
-------
TABLE 29
ELEMENTAL ANALYSES AND SUPPORTING COAL ANALYSES
Al SI/EPA DATA
Te.t Supl«
HuBbet HiMB«r
19-1
20-2
21-1
1208
1213
1222
1221
1224
1229
1230
Al B»
Z pp.
5.11
•.172
• 380
•290
<»70
112
•.952
Se C. a
PP" PP» PP"
•.96 «11
'.83 no
•.16 cl.9
.42 260
•.40 25
<2.4 400
•4.2 28
CB Cr
PP. pp.
azr.
750
15
160
170
860
Cu F«
pp. Z
90
750
<6.9
80
.312
180
.572
=« Ge Hg 1C Kg
PP" PP» HP. Z »B.
83
.27
3.6
80
61
28
Hn KB
DBB BBB
140
410
•2.8
26
97
170
260
Ht HI
BB. BBB
770
.152
11
110
270
.182
1.102
Pb Sb
230
.392
15
.172
130
.592
• 810
Se
•680
• 520
'210
• 160
•260
•3300
• 2900
Sn
•710
•560
•86
•61
•280
•.11Z
•.31Z
Sr Tl Te V
• 480
O70
36
28
190
540
2100
Zn
3.31
19
.112
.222
.192
WIlPUTTt DATA
7-1
B-2
8«-l
-3
9-1
10-6
-4
11-5
-4
23-2
1045
1060
1067
10M
1095
11(12
1105
1114
1111
1261
2.27 4.12
0.39 4.72
1.06 <1.7Z
.36 <0.3Z
3.74 15Z
1.73 III
<.32Z
280 152
4.52
.612
•8.8 1.14
-------
to
to
TABLE 30
COAL ANALYSIS (ULTIMATE AND PROXIMATE)
Test
% Carbon
% Hydrogen
% Nitrogen
% Sulfur
% Ash
% Oxygen
% Moisture
% Volatile
Matter
% Fixed
Carbon
(BTU/lb)
Heat of
Combustion
8A
72.94
5.34
1.16
1.12
6.90
12.54
6.82
30.34
55.94
13.142
As Received
10 23
73.23
5.29
1.13
1.13
6.56
12.66
7.32
30.50
55.62
13.121
73.89
5.13
1.27
1.13
7.36
11.22
5.46
31.90
55.28
13.140
24
72.91
5.11
1.35
1.27
7.40
11.96
6.25
30.59
55.76
13.162
8A
78.28
4.92
1.25
1.20
7.40
6.95
—
32.56
60.04
14.103
Dry
10
79.01
4.83
1.22
1.22
7.08
6.64
—
32.91
60.01
14.157
23
78.16
4.78
1.34
1.19
7.78
6.75
—
33.74
58.48
13.905
24
77.77
4.71
1.44
1.35
7.89
6.84
—
32.63
59.48
14.039
-------
Some of the more important elements identified in detectable concen-
trations in the emitted particulate material were Cu, Fe, Pb, and Zn.
Due to the variances in particulate sample sizes, it is not possible
to make a generalized statement concerning constituent concentrations
in the emissions. It should be pointed out, however, that no constituent
concentrations in excess of what can be reasonably explained by coal
constituents also reported in Table 29, was identified in the particulate
material. In order to obtain more definitive information on particulate
constituents, much larger samples distributed as a function of particle
size must be obtained and analyzed. The particle size information is
important in assessing the results, since this property will determine
whether the particles fall or settle out quickly (i.e., particles >200(J-).
or behave similar to a gas in the atmosphere (i.e., <3ji).
Further coal analysis was performed as support for the particulate
analysis. Table 30 reports the ultimate and proximate analyses of the
coal as received and dry. The size analysis is listed in Appendix B.
One hundred percent of all four samples passed the 1" round sieve size
and always less than 6% of the sample passed the number 200 square
sieve size. The average size was between square #8 and square #30 sieve
size.
The density of the particulate matter was investigated at the MITRE lab
where float tests indicated the specific gravity of the solid portion
of the sample ranged between 1.6 and .9 relative to water. Generally,
better than 50% of the solid particles were greater than 1.3. The tar
portion of the sample was assumed to have a specific gravity similar to
that of coal tar between 1 and .85- To simplify calculations using
density throughout this document, 1.2 g/ml was used as the approximate
density. This is probably on the high side of the average. However,
there was not enough data concerning the density of the tar portion of
the sample to justify any other value.
As a part of the supporting analyses, a sample of the collected particu-
late was provided by EPA to the Columbus Laboratories of Battelle. The
sample consisted of a number of black granules, and a loaded 5 cm
123
-------
diameter filter. Several of the granules were placed in a probe tube
and inserted into the MS9 mass spectrometer. Eight high-resolution mass
spectra were obtained starting at 60°C and terminating at 350 C. The
spectra were dominated by hydrocarbon fragments and a strong phthalate
ester peak at mass 149.024. This latter peak was thought to be attri-
butable to a contaminant from the plastic bottle top of the container.
No.evidence of carcinogens was found in this analysis. However, ions
with mass 228.094 (CnoH.0) and 252.094 (Cor.H10) were detected. These
lo I/ /U 12
peaks correspond to the molecular ions for benzphenanthrenes and benz-
(a)pyrene, or their structural isomers. It is, of course, impossible
to determine from high resolution spectra whether the carcinogens are
actually present or whether the peaks are due to their noncarcinogenie
isomers. Detailed chromatographic studies would be required to confirm
the presence or absence of the carcinogenic PNA's. Unfortunately, ex-
traction of the few remaining black granules with methylene chloride
yielded insufficient soluble material to permit gas chromatographic-mass
spectrometric analysis.
The sample filter was Soxhlet extracted with methylene chloride for 30
hours, and the extract concentrated to approximately 0.25 ml. A portion
of this extract was injected into a MS9 probe sample tube, and the sam-
ple was thermally volatilized. Six high resolution mass spectra were
obtained, starting at a probe temperature of 60 C and terminating at
300 C. Despite repetition of these high resolution runs, the exceedingly
complex mixture and high number of compounds present precluded detection
of the most important reference calibration peaks. Consequently, no
useful high resolution data was obtained from this sample.
The filter extract was subsequently subjected to gas chromatographic-
mass spectrometric analysis, using both OV-17 and Dexil 300 columns, in
an attempt to detect OSHA carcinogens and carcinogenic polynuclear
aromatic hydrocarbons, respectively. The total mass chromatogram, ob-
tained using OV-17 (150 to 280 , programmed at 6 min ), exhibits a
large number of quite well resolved peaks. In only one instance was
there a possibility of the presence of a carcinogen, benzidine. Three
124
-------
lower polynuclear hydrocarbon, anthracene, pyrene, and fluoranthene,
were readily identified from their mass spectra and chromatographic data.
In order to examine the possibility of the loaded filter extract containing
known polynuclear aromatic hydrocarbon carcinogens, gas chromatographic-
chemical ionization mass spectrometry was carried out using a 10-foot
2-1/2 percent Dexil 300 at 260°C, programmed at 1°C min~ to 300°C. The
combination of mass spectra and chromatographic data clearly indicated
the presence of benz(c) phenanthrene (potent carcinogen), benz(a)
anthracene (carcinogen), a benzfluoranthene isomer (possible carcinogen),
a benzfluoranthrene isomer (possible carcinogen), benz(a)pyrene (potent
carcinogen) and/or benz(e) pyrene, and cholanthrene (carcinogen).
125
-------
OPTICAL SYSTEM PROGRAM RESULTS
The primary objective of the optical measurement program is to determine
the feasibility of a compliance monitoring system for charging emissions
based on the technique of plume optical transmittance measurements. This
assessment of feasibility is logically divided into three areas; tech-
nical feasibility, implementation feasibility, and operational feasibility.
The technical feasibility of the optical system concerns the demonstrated
ability to measure plume transmittance with conventional light sources
and photographic recording techniques. The basic measurement concept
has been established under laboratory conditions by several other in-
vestigators. MITRE's optical measurement program extended this work to
establishment of feasibility under field conditions for polidisperse
emission plumes with high dynamic characteristics. The second area,
implementation feasibility, concerns the availability and practicality
of equipment and materials needed to implement and operate the system in
the environment of an operational coke oven. The feasibility of this
approach for other extended emission sources should be investigated
independently, however, the coke oven environment is sufficiently severe
and demanding so as to provide a basis for a valid implementation feasi-
bility test. The last area, operational feasibility, concerns the cost
of fabricating, installing and operating a number of such systems on
typical coke ovens, where operating costs cover the upkeep and maintenance
of the equipment, as well as the cost of film densitometry and data
analysis. Other factors to be included in a feasibility assessment are
the accuracy and dependability of the system, how information from this
system might be used in a compliance system, and what equipment config-
urations might be applicable for such a system.
Technical Feasibility
In order to provide a theoretical background for the proposed measurement,
the basic theory of particle and light interaction had to be shown to be
compatible with the expected polydisperse characteristics of the coke
oven emissions. Such a compatibility can be shown mathematically and
was developed in the preliminary optical system report, MTR-6546. In
126
-------
addition to the basic transmission measurement theory, a theoretical
technique for the detection of non-source light interference (plume air-
light) was developed and explained. The theory showed how the system
could be self-compensating for long term variations in light source
conditions and ambient lighting conditions.
The objective of most optically oriented systems has been to estimate
mass loading value by dealing with a known volume of intervening space
between light source and detector. The intent of this system is to es-
timate total mass emitted. With such an intent, the geometry of the
system is relaxed and variability in size of the emission plume is
diminished in importance. The theory of this system has been to estimate
the mass in a cross-sectional area in the path of the light source. In
order to produce an estimation of total mass emitted, a measurement of
the vertical rise speed of the plume must be made. The mass rate indi-
cated on a single photographic frame multiplied by the rise velocity was
shown theoretically to be a reasonable estimation of total mass emitted
over time. The characteristics of plume dynamics for the charging cycle
were studied and reasonableness of plume consistency over a period of a
few seconds established. In addition, it was found that a characteristic
velocity for plume rise rather than a continuously measured value did
not seriously affect the validity of the measurement system.
A study of photographic material characteristics was conducted to
establish the availability of a suitable film for the proposed measure-
ments. The study showed that the material having the widest exposure
range and longest linear portion of the exposure curve was one of the
most common photographic films. This fact simplified the selection of
film for use in this system. The choice of film was then considered in
the determination of the required intensity for the system light source.
This investigation showed that there were several common light sources
available which could produce the required intensity.
A limitation basic to the proposed system configuration was the probable
^dependability of mass calculations in situations where the transmission
fell below 10%. This is because in such a situation, multiple scattering
127
-------
has an increased likelihood of occurrence. A lower limiting value for
transmittance was selected after consideration of the distance between
the light source and detection, the acceptance angle of the detector,
and the expected nature of the emissions. Based on this limitation,
the system will calculate a maximum mass value for the point correspond-
ing to the lower limit and indicate that the calculated value is > the
actual mass rate.
It is now felt at the conclusion of our test period that all theoretical
aspects of the system have proved to be feasible within the constraints
expressed at the outset of the project. Some questions have arisen
stemming from the highly variable nature both in size and composition of
the particulate as to the ability of developing average characteristics.
However, such a limitation would have equal impact on other mass measure-
ment approaches and is not considered a detriment unique to this optical
system concept.
Implementation Feasibili ty
During the selection, fabrication, and assembly of system components,
one criteria which was stressed was off-the-shelf availability. The
major components of this system meet this criteria. The selection of
components also considered the adaptability of components to other loca-
tions or environments. This objective was accomplished to a lesser
degree.
The test system utilized a standard sequence camera which was modified
to provide a secondary exposure of a digital clock. This system proved
to be a virtually trouble free recording device. Although the camera
was not run in an unattended mode, its performance was such that the
feasibility of unattended operation was established to a great extent.
All tests performed on the camera to establish its reliability yielded
positive results. Specifically, the test for accuracy and repeatability
of shutter timing shows a repeatability of better than ±4% of the marked
values at 1/1000 and 1/50 second speeds. On continuous runs of 12 and
24 hours at one frame per second exposure rate, no malfunctions occurred
in the camera system. Modification which might be desirable to increase
128
-------
the film capacity of the camera should not affect the basic reliability
and operability of the unit.
For extended periods of operation (one month) camera equipment would
require a minimum of a "dust proof" housing with a suitable transparent
window which can be cleaned regularly without optical damage. Such an
enclosure could be constructed using a NEMA water tight box with a hard
plate glass window installed. A cleaning and maintenance routine
similar to those used for pen-type pressure and temperature recording
instruments in use in the oven environment should be sufficient for
reliable operation.
The camera was operated using an AC power supply during the field test
period. The basic camera, however, was designed to operate from a
12 volt DC power source, such as a battery. The camera was tested in
the laboratory using a 12 volt battery and found to operate in essen-
tially the same manner with no detectable degradation in performance.
Such a capability provides operational flexibility needed in situations
where temporary operation in a remote or difficult to reach location is
required.
Additional details of tests performed on the camera system can be found
in Appendix C of this report.
The camera control used in the test configuration was somewhat more com-
plex than would be required for an operational system. This complexity
was the result of control flexibility needed in performing the tests.
The test, however, demonstrated the feasibility of the control concept.
Specifically, a control circuit based on a clock locked to the AC line
provides a satisfactory time base for the system. This clock consists
of integrated circuit counting modules which drive both the digital dis-
play recorded on each film frame and exposure control. A fixed exposure
rate (one frame per one or five seconds) would be wired into the system.
During the test period, occasional electrical noise would cause the clock
counter to advance. The environment of the coke oven contains much high
power electrical equipment which can generate electrical spikes on the
129
-------
AC power line. For this reason, additional attention should be given to
the power supply of the control unit to eliminate this interference.
This problem is not, however, considered serious and should be. relatively
easy to solve.
The camera control unit is designed to operate from either 110 volt AC
commercial power or 12 volt DC power. In the DC operational mode, the
camera exposure rate is continuously adjustable using an RC circuit.
The digital display then becomes a counter operating on the frequency of
the exposure pulse. Should a realtime clock display be desired on DC
operation, a time standard oscillator operating at radio frequency must
be added. Such an oscillator would provide a standard with accuracy of
better than ±5 seconds in 24 hours. All other functions of the control
remain the same.
The system light source is the one component which must be considered
"custom made" for the specific observation site. The unit fabricated
for this test utilized all off-the-shelf electrical lighting components
with the exception of the reflector surface. The intent of the special
reflector was to obtain maximum apparent source width with uniform light
output across this width. The parabolic reflector proved to be very
efficient in the achievement of this objective. Densitometry traces
across the width of test frames showed excellent uniformity. It was
originally expected that a 10 inch width might be required to provide
for densitometry alignment error. Experience has shown, however, that
the densitometry trace can accurately follow an image of a source having
a width of less than 200 p. For this reason, it is felt that the special
reflector could be replaced by 4 or 5 parallel fluorescent tubes, each
having a diameter of 1-1/2 inches.
The stroboscopic effect of fluorescent tubes was discussed in the docu-
ment containing initial design considerations. It was decided that the
simplest method to eliminate this problem was to use a shutter speed that
would capture one complete cycle of light source supply current. Test
exposures using this method showed minimal variation in image density
less than ±5% exposure change for the overall system. It is felt, however,
130
-------
that this might be improved by using a DC supply voltage for the fluor-
escent tubes. Discussions with tube manufacturers indicate that such
operation is possible and can be implemented with few changes. These
consist primarily of isolating a filament supply source and provision
for periodic polarity reversal to prevent breakdown of the tube through
the plating effects present in DC operation. It should also be noted
that the light output of the unit would be increased for a given supply
voltage and any shutter speed could then be used without stroboscopic
related problems.
The system light source is physically the largest system component. In
addition to this fact, its configuration is affected more by the physical
details of the observation site than any other unit. The light source
for the test system was designed and fabricated to provide maximum
flexibility both at the test location and for possible reuse at some
other site. Even with this in mind, major structural components would
probably require considerable modification for reuse. It is difficult
to imagine a suitable "portable" unit which would be designed for quick
assembly and disassembly. It is expected, therefore, that each site to
be observed would require a unit designed for that location and per-
manently installed on suitable support structures.
All other details of the light source configuration are considered
feasible and desirable and would be recommended for inclusion in speci-
fications for new units.
The environment of a coke oven is quite severe on all types of equipment.
The pervasive heat, corrosive gases, and dust/grit, act upon all equip-
ment in the area. It was necessary to select or develop system com-
ponents which could function continuously in this environment. During
the course of the tests, no significant problems were encountered in
maintaining the various system components selected for system implemen-
tation. The system light source was left in place and is still operating
as a lighting facility for the PA battery.
131
-------
The densitometry analysis of film from this system was the area where
the most difficulty in performance was encountered. It is well within
the capability of standard densitometers to measure density with the
required accuracy for small image areas. There are also many machines
which produce a computer compatible magnetic tape output. The primary
difficulty with such machines is that they are not normally used to
selectively scan a small portion of the image area, but rather to scan
the complete area and allow computer processing to locate and recon-
struct the area of interest. Although such an approach might be appli-
cable to our system data, it was decided that the necessary computer
programming for image recognition was excessive in terms of the desired
test results. A rather specialized machine with the ability to provide
usual image alignment, selectability of small scanning areas and computer
compatible magnetic tape output was selected for film analysis. The
resolution, accuracy, and stability of the machine were somewhat greater
than would be required for our analysis. As a result of its unique
features and versatility, the machine availability was less than desirable
and the operational cost higher than expected. Film analysis using this
machine and associated operational practices is not considered to be
feasible for any large scale operation of the system.
Film processing and calibration is an important phase of system operations.
Although the actual processing of the film is not complex, very few
service facilities were found which would guarantee the control necessary
in the processing operation. This is explained by the fact that such
control is not normally necessary in processing for standard image
photography and the amount of processing which we projected for the con-
duct of the test was not sufficient to persuade a supplier to change his
operations to accommodate our needs. It was determined, however, that
the facility providing densitometry analysis could provide the required
processing on a custom basis. They were also equipped to provide the
i
sensitometric exposure calibration. It was decided that total film
analysis service at one location was more desirable than a possible re-
duction in film processing costs.
132
-------
Although, not feasible for large scale implementation under the cost
structure applied during this test period, the basic processing tech-
niques are considered feasible. It is reasonable to assume that given
some amount of assured business, a service organization would be willing
to establish a facility where the required processing could be done on
a routine basis.
Operational Feasibility
The operational feasibility of this system for coke oven monitoring is
difficult to assess for two reasons. First, the standards for emission
compliance are not accurately defined or generally expressed in direct
quantitative terms. Second, there are few, if any, alternative systems
against which to compare the technique on a cost and operability basis.
The camera equipment required for such a system is available as an
off-the-shelf item. The cost of this type of equipment ranges from
$1500 to $8000 per unit. The primary difference in the various units is
the level of versatility and ruggedness. The less expensive units pro-
vide acceptable performance and are suggested for system implementation.
Since it is feasible to use one type of camera for a wide variety of
site configurations, the modification required to place the secondary
exposure of the digital clock display on the frame might be offered as
a standard option on that type. The cost and availability of the equip-
ment discussed above is considered reasonable and feasible for such a
system.
The camera control system was a custom designed unit but could be greatly
simplified when a routine is established for system operation. Reliability
of the system is such that no change in the basic components is necessary.
An improvement in line noise interference is needed but this should not
affect the complexity, reliability or cost of the unit. It is estimated
that a control suitable for system operation could be constructed in
quantities of 10 or more for under $300. Because of the limited quantity,
they would be produced by a custom or small run fabrication facility.
The basic concept and design is considered practical and feasible for
133
-------
system application with a per—unit cost of up to $500. If the cost
should exceed this figure, it might be advisable to utilize a commer-
cially available digital clock and display and an electromechanical ex-
posure control.
The physical arrangement of the light source for the system is highly
dependent on the characteristics of the monitoring site. The support
structure for the source must be designed to allow normal operation of
the coking plant. This involves minimization of interference to the
oven top so as to provide the access necessary in the maintenance and
operation of the battery. The structure must span the battery and ex-
tend over each end by several feet. The side sections must extend down-
ward as close as possible to the oven surface. The support structure can
be expected to represent at least half of the total light source cost.
The test activity showed that fluorescent lamps of the "very high output"
type provide suitable light sources for this system. Placing four such
bulbs side by side should provide suitable source width and eliminate
the need for the rear reflecting surface. If these lights are operated
from an A.C. source, the required case, lens, electrical components and
fabrication are estimated to cost $250 per 8 foot section. To operate
these lamps on B.C. current would add approximately $50 per 8 foot sec-
tion to the cost. The costs would be the same for the side or "wing"
sections of the source.
In general, coke ovens runs 45 to 55 feet wide. The light source should
be placed above the larry car height which runs between 15 and 20 feet.
With these assumptions, the typical installation would consist of 10 to
12 eight foot light sections. The total cost including a supporting
structure, is estimated at between $3500 arid $6000.
The light source should require a minimum of maintenance which can be
performed by regular plant electrical maintenance personnel. Assuming a
' i
continuing requirement for emission surveillance and compliance monitor-
ing and the costs of continuous equipment or trained personnel for
visual observation, the cost of the light source and supporting structure
is considered economically feasible. This assessment also takes into
134
-------
account the direct operating cost of electricity supplied to the unit.
Processing of the film from this system must now be accomplished on a.
"custom" basis. The processing cost currently runs 29 cents per foot
of film processed. At approximately 750 frames of data per 100 foot of
film, the processing cost would be approximately 3.9 cents per data
frame. At an exposure rate of one frame per second for approximately
2.5 minutes needed to charge a coke oven, 150 frames of data would be
taken. The processing cost per average charge observed would then be
approximately $5.85.
If some volume of business could be guaranteed to a service organization,
the cost could probably be reduced from 20% to 40%. However, the current
cost of 3.9 cents per frame is not considered prohibitive for a system
which would operate only during charges or during periods of high emission
detected by some secondary system.
Densitometry analysis of the film output from the measurement system pre-
sents the greatest problem in current system operational feasibility.
Because of the unique nature of the scanning requirement, the use of a
highly flexible and sophisticated microdensitometer is required. As a
result, the cost of analysis is high and the response time for this
service is relatively slow because of the large demand for the instrument
in question to perform other work. The standard rate set by the par-
ticulate supplier used in the test program is $30 per hour for the in-
strument and $16 per hour for the instrument operator. The machine must
be set up for the particular scanning job which requires 15 to 30
minutes. This time is charged as operating time at the regular rate.
At these rates, the data frames scanned during the course of the tests
cost between $10 per frame including all test frames, calibration,
special uniformity checks and set up, or $15 per frame counting only
test data frames scanned as production runs.
The use of a microdensitometer can only be considered feasible if the
machine can be dedicated to this particular type of film analysis. This
would allow the machine to be modified to facilitate quick and accurate
135
-------
mounting of the system film size. It would also eliminate the "set up"
operation since it is reasonable to assume that film from different ob-
servation sites would be very similar in terms of light source image
size and density range. The utilization of such a machine would also
lower the required skills of the operator. Since the machine could be
made automatic once a scan is initiated, the operator would only be re-
quired to initially position the film in the machine and monitor the
operation for obvious malfunctions. Output from the machine would be
placed in computer compatible form on magnetic tape.
A machine capable of performing the required analysis is estimated to
cost between $5,000 and $8,000. The machine could probably scan one
complete frame of data in three to five minutes. If film was taken or
analyzed at one frame every five seconds, the typical coke oven charging
operation extending over three minutes would involve 36 frames. The
optimal analysis time of the data frames would then be about two hours.
To completely cover all oven charges on one battery during a typical
eight hour shift (40 charges), the scanning time would be about 80 hours
of machine time. Such coverage and analysis is not considered operation-
ally feasible. Both time and money are considered to be excessive for
such coverage. Should complete surveillance be considered necessary, a
comparison would have to be made between this system and other systems
with similar capabilities with particular emphasis placed on the data re-
trieval aspects of system operation.
If one hypothesizes the use of analysis by exception, the system analysis
feasibility is greatly enhanced. As an example, if the camera is only
activated during periods of excessive visable emissions, the amount of
analysis required could be greatly reduced. The analysis load could be
still further reduced if some specific number of film frames were to be
established as a basis for non-compliance [i.e., visible emissions in
I
three consecutive frames (15 second period) exceeding some set level].
Then analysis could be reduced to 20 to 30 minutes machine time for any
charge during which violation is suspected.
136
-------
The job of initial emission detection could be accomplished by several
long path transmissometers crisscrossing the oven surface above the
level of the larry car top. When one or more of these units detected
some lower level of opacity, they would cause activation of the camera
unit which in turn would take some predetermined number of data frames.
This record could be correlated with over activities through the opera-
tion time logs.
Under the above assumptions, the use of a microdensitometer is considered
feasible for use in data acquisition in the optical monitoring system.
Data processing of the densitometer output is a fairly straightforward
task. The algorithm involves simple arithmetic and a relatively small
amount of core memory is required to execute the program. The input to
the program can be totally on tape and the output can be a single line
giving the mass rate for a particular frame or a more detailed report
displaying the quantities involved for each scanned point across the
bar. During the course of the test data analysis, the latter option was
chosen to allow a more detailed examination of the data and the results
of the computations. Although helpful in gaining an understanding and
confidence in system operation, the additional printed information is
not necessary to the operational output of the system.
Computer compatible data (in a form which can be directly input to the
computer) is feasible and indeed necessary to the efficient operation of
any such system. The amount of raw data obtained from this system would
be unmanageable in printed or strip chart form for any operational ex-
tension of the concept.
System Accuracy
Because of the multiple steps involved in system operation, opportunity
for error is increased substantially over any single step measurement.
In addition, some assumptions must be made to allow the application of
this technique to a practical situation. The most important assumption
is that the size of the emission particles can be accurately represented
by some size distribution and that that distribution remains relatively
137
-------
constant over the test period and to a lesser extent constant from one
test to another.
Based on the information originally available on particulate sizing for
coke oven emissions, a much larger particle size was anticipated. Under
varying size distributions for larger particles, the effective light
removal capabilities of a particle remain essentially constant at a
value equal to about two times its cross-sectional area. For small
particles, however, the effective light removal capability of a given
particle varies between one and four as a function of particle material,
wavelength of source light, and size of particle.
Coke oven emissions were found to be very dynamic in nature. The
quantity of emissions varied widely over the total charging period, be-
tween source points in the charging system, and between the old and new
charging cars. In addition, the visual characteristics of the emission
varied widely from a puffy black smoke generally associated with visible
flames around the car, to thick brown or mustard colored plumes which
appeared to be the raw gases driven off the coal with little or no com-
bustion taking place. Any investigation as to how each type of emission
behaved in terms of optical system considerations was beyond the scope
of this investigation. .A further complication in the measurement process
was the bi-populate nature of the particle size distribution. This fact
negated the simple analysis of a single distribution and the establish-
ment of a single mean diameter value to use in volume calculations.
In order to facilitate measurement calculations, the two distributions
were considered separately. Through analysis of the particulate material,
it was determined that the larger particles were made up primarily of
carbon or coal particles, while the smaller particles were composed
primarily of tar. Using this information, curves of K values versus
particle size were selected for these materials, and the geometric mass
i
mean radius of each separate distribution established. The representative
value for K was determined from the curves using the mean radius. The
two values were combined to produce a single value for computation by
weighing each value as a function of the particle sample mass distribution*
138
-------
The above discussion is included as an example of how difficult the
accurate determination of particle size distributions can be, and what
assumptions must be made if the distribution is found not to fit some
standard form such as a single normal or lognormal curve. In addition,
if the particle size tends toward values below 10 y in diameter, the mass
values calculated for assuming one mean diameter can vary by a factor of
four with very small changes in the particle distribution and thus
in the actual effective mean diameter.
Actual experience with some particular source or type of source can
greatly improve the accuracy of the system through increased knowledge of
the typical optical properties of the emission. The system could only
approach 100% accuracy if the emissions were all of the same material
with similar optical properties (surface conditions) and all particles
were of the same size. With minimal knowledge of the particulate
characteristics, only order-of-magnitude accuracy can be expected for
total mass measurements. However, relative measurements performed on
some small source which exhibits reasonably constant particle character-
istics (either observed or measured) can be expected to have an accuracy
factor on the order of 2-5. It is this range of accuracy which appears
reasonable for measurements made with the experimental system.
To make the operation of this system feasible, two other assumptions must
be made which are multiplicative in terms of mass measurement errors.
The basic assumptions are that a representative value can be determined
for particle density and emission rise velocity, and that these values
once determined, remain relatively constant. In both situations, it is
difficult to conceive of a case where the selected value for either
quantity would introduce an error greater than 25% and indeed, the error
actually experienced would probably be <10%.
Analysis of particle samples show that the primary constituents are tars
and carbon or coal. The densities have ranged between <\,.9 On/cm and
1.4 Gm/cm3. The apportionment between the two materials has shown reason-
able consistency. Based on these observations, a representative density
of 1.2 Gm/cm3 was established. In the presence of very low emissions, we
139
-------
would expect the material make-up to tend towards a predominance of tar
and the selected value would tend to be high. If emissions were par-
ticularly heavy indicating considerable combustion and turbulence in
the oven, the emissions would probably tend toward carbon and the
selected value would tend to be low.
The selected value for emission rise velocity was based on a large num-
ber of photographic observations of plume behavior. Few inconsistencies
in behavior were rated. This is probably because the position of the
plume, when observed by the system, is removed from the larry car stacks
and hopper structure, and its behavior is primarily determined by ambient
air conditions which remain relatively constant.
The exception to this is when a relatively strong crosswind is blowing.
The plume in this situation can have a horizontal velocity ^ the vertical
velocity. In such a situation, the plume will probably pass across one
of the light source vertical wing using units rather than the horizontal
light source. In such situations, it might be necessary to make some
wind measurements to obtain a reliable value for emission velocity pass
the vertical bar units. In any case, the representative vertical velocity,
when applied to the horizontal bar under minimal wind conditions, should
cause minimal error in system measurements.
A second area of system error apart from the necessary quantitative assump-
tions, is the inability of the system to see or detect all emissions
occurring on the oven battery during charging. This deficiency is a
result of the geometry of a feasible system installation on a typical
coke oven. In Figure 22 the field of view of the camera with respect to
the light bar and the oven surface is shown. As the field of view nar-
rows toward the camera, additional portions of the oven surface and space
above that surface are excluded from detection by the system. The drawing
shows what are considered ideal locations in terms of image size and de-
tection locations. The camera could be moved further away from the
source to provide an enlarged field of view, but this would have the
effect of increasing the differences in plume behavior caused by the vari-
ations in distance from emission source for different positions on the
140
-------
LIGHT SOURCE
LARRY CAR.
COKE OVEN SURFACE
TOP VIEW
LARRY CAR
COKE OVEN SURFACE
SIDE VIEW
CAMERA
CAMERA
FIGURE 22
TYPICAL SYSTEM GEOMETRY
-------
oven. It would also tend to decrease the photographic Image size which
in turn makes densitometry more difficult. The light source might be
increased in size, but this quickly approaches the practical limitations
of available support for the structure. Additional cameras could be
added, but the costs of implementation and operation would be increased
proportionately.
It also appears possible, based on test experience, that light emissions
occurring around the base of the car could become so diffused before
rising to a detectable point above the car that the system would have
considerable difficulty distinguishing them from heavy background or
ambient.
In general, the conclusion must be drawn that this monitoring concept in
configurations similar to the experimental system cannot be expected to
provide 100% coverage or positive detection of any light emissions.
Change to the system configuration can be made, but this would result in
substantial increases in both cost of experiment and operation.
System Configuration. Operation, and Data Analysis
The configuration of system components in the implementation of this
measurement concept does' not provide a large latitude of possibilities.
The detector (camera) must be positioned on one side of the emission
plume and a light source must be positioned on the opposite side. The
use of a reflector in place of the light source considerably complicates
the system implementation, as well as data interpretation, and is not
felt to be a practical alternative. As we stated previously, some second-
ary system for initial emission detection should be considered. This
would allow the optical system to operate on an exception basis rather
than continuously or on every charging sequence. Such a system might in-
clude relatively simple transmissometers operating over long paths over
i
the oven top. These paths might use a single reflector at one epd in
order to minimize the number of units required. A simple path configur-
ation is shown in Figure 23.
142
-------
-OVEN TOP
ooo
ooo o
-LIGHT BEAM PATHS
FIGURE 23
SECONDARY EMISSION DETECTOR ARRANGEMENT
-------
The system operating in this mode could function unattended for periods
of days or weeks, assuming an 80% or greater compliance record and large
capacity film magazines. The primary attention need by the system would
be the occasional cleaning of the light source and camera/associated
optics, the changing of film magazines, and checks of data clocks and
secondary detection system operation.
In comparison to a Ringelmann assessment of emissions, the experimental
optical system can provide additional data with higher quality. As ap-
plied to coke oven observations, the Ringelmann technique cannot account
for the quantity or size of the emission plume. This is important in
determining the total emission because of the wide variation in point
source characteristics on and around the larry car. The coke stacks
for instance, release large quantities of dense opaque emissions while
small leaks around the drop sleeve may release equally dense emissions,
but of a much lower volume. The situation where a large number of small
leaks release a large quantity of diffused emissions as contrasted to a
small dense emission from a stack is also not adequately differentiated.
As understood for current compliance monitoring, the Ringelmann observer
would time the period over which he observed a certain opacity, not noting
the quantity (plume size) of emission at that opacity. This could
obviously give a distorted picture of the actual emission volume in
comparison to observations made on other charges. The optical system
should be able to provide information that more truly reflects the
actual quantity of emission and allows a better basis for comparison of
this quantity between individual charging activities. The system has
also been shown to adjust for conditions beyond the control of the
Ringelmann observer, such as light conditions and color of the emission.
As previously stated, the optical system, although not providing a highly
accurate value of absolute mass emitted, should provide a reliable basis
for relative mass measurements within the system. If, however, the
quantitative aspects of mass measurement are not sufficiently accurate
or reliable for use in a compliance system, the basic transmission (the
reciprocal of opacity) measurement capability of the system provides an
144
-------
improvement over Ringelmann in terms of accounting for emission volume.
By dividing up the light source into many point sources, the system makes
the equivalent of many instantaneous Ringelmann assessments over a wide
area with the advantage that variables, such as emission color or
ambient lighting, have been eliminated or compensated for.
An example can be given to demonstrate this capability. Assume a light
source whose photographic image can be divided into 500 equal sections
for purposes of density measurement. If the bar is 50 feet long, then
each of the 500 spots would represent a .1 foot wide area of the light
source. Suppose that transmission values were determined for each of
the 500 points along the bar, and that corresponding Ringelmann numbers
were assigned to each of the points. These numbers could then be summed
to provide a composite value (referred to here as a Ringelmann "Quotient"
- RQ) which would take into account the size or extent of the emission
plume or condition. A perfectly clear bar would produce a sum of zero,
corresponding to a Ringelmann of "zero." Assuming a Ringelmann 1 condi-
tion across the entire bar, the sum would be 500. If a Ringelmann 2 con-
dition exited across the bar, the sum would be 1000. The other Ringelmann
sums would be proportional for constant Ringelmann numbers. The system
upper cut-off limit at 90% opacity would yield a Ringelmann sum of 2000,
which would also be obtained for an 80% opacity condition.
Since conditions can be expected to vary from point to point on the bar,
the Ringelmann sum would be expected to vary as a function of emission
quantity between 0 and 2000. Since a value for one photographic frame re-
presents an instantaneous rate, the sum values for several frames might
be totaled and divided by the number of frames to provide a rate average
over time. As an example of how a Ringelmann Quotient would work for
actual data, two optical system data frames of contrasting emission con-
ditions along with their assigned Ringelmann Quotient (RQ) and calculated
mass emission rates are presented in Figure 24. Additionally, the values
used to compute the RQ from transmission data are presented in Table
31. For the test system, 539 spot transmission measurements were made
yielding a possible maximum RQ value of 2156. This value would
145
-------
MASS RATE = 14 .4151 g. An
FIGURE 24
EMISSION CONDITIONS AND CALCULATED RQ VALUES
146
-------
TABLE 31
CONVERSION OF TRANSMISSION VALUES
TO RINGELMANN VALUES
% Transmission
100
90
80
70
60
50
40
30
20
10
% Opacity
0
10
20
30
40
50
60
70
80
90
Corresponding Ringelmann
1/T Value Assigned
1
1.11 :
1.25
1.43
1.66
2.00 :
2.50
3.33 :
5.00
10.00J
0
1
2
3
4
147
-------
correspond to a Ringelmann condition 4 measured at all 539 points along
the light source. The test system cannot, however, reach this absolute
value since about 30 points fall on gaps in the light source and are
assigned 0 mass values. It is easy, however, to see the difference be-
tween the two RQ values in comparison to the emission conditions
pictured. It can be seen that such a technique might offer an improved
capability in assessing the amount of emissions actually released to the
air during a charging operation.
Data Processing Output
The basic output of the data analysis for the optical monitoring system
is a total mass rate value per unit vertical height of the emission
plume. This value is in turn made up of 539 individual mass rate values
calculated for corresponding spots along the length of the light source.
In an operational situation, the only number of interest is the mass rate
associated with a particular frame. For purposes of system performance
assessment, however, additional information was included on the data
analysis print-out for the selected tests. Figure 25 presents one page
of a typical analysis print out.
The header information contains frame identification consisting of test
number and frame time in hours, minutes, and seconds. The base and scan
numbers used in "housekeeping" refer to the base frame number on the
data tape and the scan number performed against that particular base.
Each of the seven line groups on a page present data from fifteen con-
secutive points in the left to right scan of the light source. A total
of six pages are required to present the 539 points from one complete
scan of the light bar. A seven space column associated with a single
point contains the base exposure value (no emission) for that point on
the light source labeled "BASE L". The base exposure value for the
corresponding spot on the black portion of the bar is obtained by
averaging black area readings over some specific number of points at
the end of the light source and designating this average as a constant
to be used in all calculations involving that base frame. This value
is presented as part of the header information and is labeled "ZERO PAL",
148
-------
« LI5HT
UATA II
BASE L
0-TA L
I/I
LNI1/I 1
tL MASS
DATA I.
rtA SE L
DATA L
1/T
LNI1/TI
FL 1AiS
FLAGS
iMTA (.1
b*SF I
I'-iTA L
1/T
I1MA I-
*JASt L
OUT* L
LM1/TI
F.U 1«SS
FLAGS
UATA 0
U/ISL L
II4TA L
1/7
1,1(1/11
61 MAS.
FLAGS
IM7. l>
DATA L
1/T
L ,11/11
5L MASS
Fl/GS
H«T4 0
BASF I
IK.Tt L
I/I
FL MASS
FLAGS
TEST 7
UAH GAP
1.906
9.903
4.712
3.146
1.14612
J.050
1.JS1
9.484
4.477
1*J¥>»
1.35C
'6
l.d n
3.64154
J.')2fl
l.ojo
6.952
0.01/
Jl
7.397
6.95i
1.201
O. JOd
16
ZFRU PAL
• PAL >
1.962
9.903
4.477
3.509
1.25542
0.054
l.dil
8.955
4.125
1.15058
(.'. 050
1.797
1.348
2.398
l.OOC
O.OOOO-)
1.-J62
5^739
1.851
6.335
5.1)33
1.823
1. 60C39
U.026
6. -14?
O.old
7.172
0.010
HL »
2.020
10.048
4.4 I.I
J. 7o'»
1.32(80
0.01,1
l.V 16
•'•Vj'j
t. 325
1.18110
0. 05 1
S. 02'J
1. 797
2.529
1.000
I.S51
d.335
5. 6a j
1.823
0.6)039
0. 126
7.172
0.015
7.2o4
O.'llo
1.077
DL > BL f OL
2.023
10.655
4.633
3.666
1.29909
0.056
1.9.10
9.084
4.!o:»
3.024
1.10645
0.04H
1.962
3.529
3.324
1.801
1.50H31
2.02.1
8.570
0.029
1.797
8.955
6.123
1.821
0.59935
0.026
7.2B4
O.U14
7.204
1.213
O.UOd
1.962
10.655
4.633
3.5K6
1.27710
0.055
1.962
8.456
4.477
2.934
1.0/634
0.047
1.906
5.929
4.249
2.071
0.72795
1.051
5.379
OlU32
1.851
9.215
6.323
1.820
0.59838
0.026
6.952
0.016
fl.335
;. 172
0.013
H4 tH
< DO IHUISEI
1.962
10.815
4.633
3.646
1.29363
0.056
1.942
8.827
4.555
2.990
1.09513
0.047
1.962
7.284
4.712
2.257
1.851
9.084
5. ',46
1.797
9.084
6.424
1.730
0.54834
0.024
6.952
0.014
a. 2 16
7.284
1.906
10.499
4.400
3.77K
1.32915
0.058
1.962
d. 702
4.633
2.855
1.04907
0.045
1.906
7.982
4.955
2.265
0.01769
1.962
8.702
O.020
1.851
9.348
6.631
1.730
0.54837
0.024
7.512
6.843
0.010
1.636
6.952
INlSS: 15:23:29 mi** 2 " s-J
? 01 < tlU UNVALfnl t 1/r > PCUT a 8L <. AVG fl 1 rMVUI II
1.962
10.346
3.57'>
1.27401
1.0S5
1.962
8.702
4.417
3.032
1. 10914
0.048
1.906
8.456
5.121
2.295
0. S3CM3
1.962
0.335
H. '129
1.851
9.404
6.631
1.759
0.56460
U.024
6.631
1.256
0.010
1.636
6.737
1.906
10.346
4.477
1.2S247
0.056
.'.r;. 1
4.477
3.030
1.10849
0.048
1.962
8.335
5.121
2.297
1.962
8.578
1.031
1.851
8. 6? 7
6.737
1.586
0.46152
11.020
6.222
1.112
I. »1.
1.616
7.512
7.112
1.905
10.346
4.^77
1.2H247
'1.056
2. 144
7.902
4.633
2.774
1.02026
O.O44
1.962
8. 335
5.121
?.297
0.83181
1.851
3.327
0.031)
1.6H9
9.215
6.952
1.547
0.43604
0.019
6.123
1.011
1.74-,
7. .-47
7.172
1.416
10.119
4.4/7
1.26612
0.055
?.'!'' 1
7. 745
4.477
2.714
0.99041
0.041
I.<;o2
II.K27
5.292
2.3?8
1.797
8.H27
5.516
1.7^
9.215
6. 7' 7
1 .6 j )
0.40038
0.121
6.424
7.172
(..•J52
1.906
9.903
4.477
1.23355
0.153
•.U 20
/.061
4.325
2.597
0.95430
0.041
••..120
a. 578
5.379
2.233
1.797
0. 02 n
l.l-.l
(.. ^2 t
1 .f '»3
". 526"*5
0.1123
1.0119
». 7*7
6.1,31
1.906
9.903
4.4MO
1.26386
1.055
2.120
5.55*
3.803
2.404
n. 87721
0.03ft
1.962
«. 456
5.379
2.160
1.J16
1.797
n 1 7O2
6.631
1.577
1.45568
0.020
(,.527
1.611
0.6.11
1.^
9.34H
4. "i "' "i
1.J5J11
•» in t
1.4 .1
*.1°-.
-i.lS'
T. 760 ''o
1. 'I''!
\m'l »'
9.0^
7*0?'
1.1-"»l
p.'"-,'.
0. i1*1"'
UAU
rU")0''
6.717
1* ci*ti
^.^34S7
rt»ni'«
,
l.f- •-
',."4^
l.-4.,»o
t - .
'. -M
• .'-.i"
'.III
»*l » 1
if
'>. ' t
' ' • • ' '
I. "1
' , •'• 1
"V •' •
U- '-'
•!j >"*"f
f • , • i '.
I* *•*• '
>• *r> * ';
0. •{')
}. ••>
' * fl 1 w
i. !. 1
., i^
* I IGHT
OATA 1;
UASL L
OATA t.
1/T
LNIl/ri
EL f ASS
FL4GS
rtsr 7
1, AK GAP
1.639
7.397
lI^Ol
1.1J319
t.ooe
16
2CRJ PAL *
1.742
7. d64
7.172
). 22333
0.310
1. 7-t3
j.099
7.2J4
1.267
0.23691
1.010
1. 177
1.687
7.864
1.284
1.213
'1.19327
0.0-08
1.743
8.335
7.172
l.i-7
J. 2'»"3 1
O.i<13
"' :"{'\\^
1.743 1.61-6 !.<>><.
8.216
7.2H4
l.il-(
0.25152
O.Mil
7.9d2
6.952
1.29V
0.261o6 '
0.011
7.i i.
-------
meaning a value for zero plume air light.
Exposure value information for the point obtained in the scan of that
particular data frame is labeled "DATA D" for the value of the black
area and "DATA L" for the value of the lighted area. The three exposure
values plus the value for "ZERO PAL" are used to calculate a reciprocal
value for transmittance which is printed and labeled "1/T". The value
labeled (LN(1/T)" is the natural log of the 1/T value and is presented
as an interim quantity to be saved for possible re-run of the data. The
quantity labeled "EL MASS" is the calculated mass rate in grams per
meter vertical plume size for that particular point. This value is ob-
tained using constants whose values are .assigned and listed at the start
of each data reduction run. In order to obtain a total mass rate for a
complete frame, "EL MASS" values for each of the 539 points are summed.
This value is presented on the last page associated with the particular
scan as "TOTAL FRAME MASS RATE." The last line of each group summarizes
special situations which may occur on a particular frame. An abbreviated
explanation of the "FLAGS" is presented as part of the header informa-
tion for each page. The "*" signifies a point at which the value for
"BASE L" falls below some pre-set value. This situation occurs for
points falling on the gaps between bulbs of the light source and is
generally 5-8 points long. In general, no mass rate is calculated for
these points and the value of "EL MASS" is set to zero.
The "4-" denotes a situation in which the frame plume-air light value,
"DATA D", is high when compared to a corresponding base light value
(BASE L). This situation can logically occur under two different con-
ditions. The first occurs when a gap value for "BASE L" is compared to
a slightly high value of "DATA D", caused by light smoke over the black
area. The second occurs when very heavy smoke of a bright or highly
reflective color occurs and the smoke image density appears brighter
than the light source in the base frame. Few situations of the latter
type were detected and in general, indicated that the light source re-
mained dominant for most conditions of plume air light. A "$" signifies
that a value for "DATA L" was found to be larger than a corresponding
150
-------
value for "BASE L." This occurs in the presence of heavy emissions where
the light source is still visible, giving rise to an addition of plume
air light to the apparent light source brightness. It may also occur to
a lesser degree in a no emission situation due to system inaccuracies.
To differentiate, the difference between the two values is compared to a
threshold value. If it is less than that value, a "
-------
frame if its start point does not coincide with that of the base frame.
The alignment is based on the position of the gaps in the base and data
frames and in general, require little or no shifts. The total print-
out provides a convenient method for scanning individual data frames
for areas of interest such as observed areas of high or low transmission
or suspected interferences. In short, it provides a valuable tool in
the understanding and assessment of optical system performance.
Test Data
During the course of the overall test program, 21 tests were photographed
by the optical system which were also observed using the continuous
monitoring system. In addition to the 21 tests in common, a number of
other charges were photographed using the optical system to provide test
data for that system. The original intent at the outset of testing was
to process and analyze all optical system output for comparison with the
continuous monitoring results. This became impractical in the light of
higher than expected processing and analysis costs and unexpected but
necessary alterations in the original test schedule. As a result, the
selection of a sub-set of optical film was made and analysis accomplished
prior to the processing and analysis of the continuous monitoring system
data.
The original test concept involved monitoring all emission source points
simultaneously (six points on the Wilputte car and three points on the
AISI/EPA car). The combination of volume flow from all points on a test
would have provided a reasonably good second-by-second picture of visible
emission volume against which the optical system data could be compared.
Problems encountered during the course of testing precluded the measure-
ment of all points on the Wilputte car simultaneously and made necessary
the use of aspirating fans on the three AISI/EPA car test points with the
result that the absolute volume flow was a function of the fans rather
than the volume of emission.
An alternative selection of tests where all three stacks on the Wilputte
car were monitored for volume flow or all three guides of the AISI/EPA
car were monitored for gas concentrations was made. The rationale was
152
-------
that during the charging period for the Wilputte car, a high percentage
of the total emissions escaped through the stack. Also, that on the
AISI/EPA car, the concentrations of selected emission constituents such
as total hydrocarbons could be used as an indicator of visible emissions
volume at that instant in time. A further selection of specific frames
was made by visual observation of the test film to include periods of
highly contrasting emissions (high versus low volume visible) and
transients. The results of this comparison are presented in the follow-
ing material.
Presented in Figures 26 through 30 are graphs of actual volume flow, in
cubic feet per second, combined for all three stacks on the Wilputte
larry car or measured concentrations in % total hydrocarbons for the
three combined guides on the AISI/EPA car. The second trace on each graph
presents calculated mass emission rates as measured by the optical moni-
toring system. These values expressed in grams per meter vertical plume
size, are plotted on the same time base to show event correlation. After
inspection of the film, and comparison of photographed events with the
data acquisition system record, it was obvious that some loss in syn-
chronization between the optical system clock and data acquisition sys-
tem had occurred. This was probably caused by noise pulses causing the
optical system clock to jump ahead in time a few seconds. The events
displayed were graphed using the time displayed by the clock rather than
arbitrarily correcting this time. In general, however, it appears from
several observations that in the absence of wind, events in the Wilputte
stacks are seen by the optical system with about a five second delay
and events in the AISI/EPA guides are seen by the optical system with
a five to ten second delay.
Also, since the continuous monitoring system volume flow measurements
are taken in a confined stack on a one second basis, abrupt changes in
flow can be detected and represented in a very precise fashion. The
optical system, however, detects the emission after it has had an oppor-
tunity to diffuse and, in general, become more spread out over both
time and space. The analysis interval for the optics system was selected
153
-------
at five seconds as opposed to the one second frame rate so abrupt changes
cannot be detected with the same resolution as the volume flow measure-
ment system. Association of event records displaced by an abnormally
long period in time according to the two clock values, will be noted in
the text discussing that test. Also, trend arrows showing the smoke
conditions before, after, and between optical data points are included
on the graphs. These were placed on the basis of visual observations
of the subject data frames and are not intended to be quantitative, but
instead to aid in the correlation .of the two sets of results.
Shown in Figure 26 is data obtained during Test 7- The trace of volume
•
flow represents a typical example of conditions during a Wilputte car
test. The optical system data is divided into two parts, each of which
cover an observed transient condition during the charge. The large
volume flow peak occurring just after the start of observations was de-
tected as a less abrupt peak in mass rate by the optical system with some
time delay. Visual inspection of the optical system film showed a smoke
condition corresponding to approximately the next to last calculated
point of the first trace persisting until the first part of the second
optical system trace. In the second trace, the actual shape of the volume
flow curve was more closely matched by the optical system results. This
is probably because the changes in trend of volume flow were less abrupt
and of a more consistent nature. It is also interesting to note the
optical system saturation point ("optical-system maximum value") with re-
spect to the measurements displayed here. Roughly speaking, a 10% trans-
mission of 90% opacity condition covering the complete length of the bar
would correspond to a mass rate of 53 grams/meter, allowing for the gaps
between bulbs in the light source and not including the "wing" light
source units. One can easily see that the smoke densities typically ex-
perienced during a Wilputte car charge could approach or exceed the
measurement capability of the optical system. In such cases, the rate
would have to be expressed as 53 grams/meter or greater. It would 'ap-
pear, however,-that this condition would lie well outside any standard
for visible emissions which might be established. The volume flow data
154
-------
OPTICAL SYSTEM
MAXIMUM VALUE
OPTICAL SYSTEM
DATA POINTS
OPTICAL TREND
INDICATORS
15 22 40 15 23 00
15 25 00
TIME
FIGURE 26
TEST 7 DATA
155
-------
available from the continuous monitoring system ends with, an increasing
flow rate. This trend is also observed in the optical system data as
an almost totally obscured light source after the last measured point.
This condition corresponds to the maximum measureable value for the
optical system.
Presented in Figure 27 is data obtained during Test 9. Again, the
optical data is divided into two sections covering obvious transient
flow conditions. During the first part of this test, a strong cross
wind caused a large portion of the emission to pass undetected below the
light source. The data frames displayed in Figure 24 are from Test 9
and show the effects of the wind conditions. The arrow shown on the
second data point indicates that the measured value was much lower than
the actual emission because of the cross wind condition. During the
second portion of the optical data trace, the wind subsided allowing a
truer measure of the actual emission condition. Lower peak values were
recorded because of the wind condition and the resulting emission dis-
persion. The flow characteristics are similar to Test 7 showing an
initial abrupt high flow rate, decreasing toward the middle of the test
and then building up again toward the end.
Figure 28 shows data collected during Test 17. This test was performed
on the AISI/EPA car and shows total hydrocarbon concentrations rather
than volume flow data plotted against optical system data points. The
optical data is divided into three parts on this test. The first and
the last parts have been identified as quench steam interference using
the photographic film record and voice tape commentary. The center op-
tical system trace shows the detection of the heavy emission flow near
the center point of the test. The peak values are much lower than those
noted for the Wilputte tests because of the lower quantity of emission
and the fact that the smoke source is at the oven surface under the car.
As a result, the emission has been diffused to a greater extent by the
time it rises to the light source detection area. The trend of the data
recorded by the optical system closely follows the density of emission as
indicated by the concentration of total hydrocarbons. The values
156
-------
w
co
w
Nl
H
CO
w
CJ
H
W
OPTICAL SYSTEM
MAXIMUM
VALUE
MEASURED
VOLUME FLOW
OPTICAL SYSTEM
DATA POINTS
OPTICAL TREND I
INDICATORS
10-
11 44 54
11 45 00
11 46 00
11 46 20
TIME
FIGURE 27
TEST 9 DATA
157
-------
53
50
10-
w
N
40
30
§2
20
Ol
00
10
11 13 00
20
-OPTICAL SYSTEM
MAXIMUM VALUE
KEY
TOTAL
HYDROCARBON
CONCENTRATION
__.<>._ OPTICAL SYSTEM
DATA POINTS
.OPTICAL TREND
^INDICATORS
11 16 00
FIGURE 28
TEST 17 DATA
-------
detected are well below the saturation point for the optical system.
Figure 29 presents data taken during Test 20. During the first part
of this test, a large quantity of emissions escaped through an open
"chuck door" and ascension pipe cap. This smoke rose past the light
source and in preliminary selection of optical data to be analyzed was
mistaken as emissions from the drop sleeve area. The optical data frames
selected showed a rapid rise in density during the first part of the
trace followed by a rapid decrease in density during the second part of
the optical data trace. This condition correlates with a slight increase
in recorded THC concentration, but since the emission occurred through
unmonitored points, was not recorded directly by the continuous monitor-
ing system. The primary emission around the drop sleeve occurred later
in the test but was of fairly low volume. The sequence of events and
the source of emissions was determined by examination of the optical sys-
tem film data. The data is presented to show the ability of the optical
system to track rapidly changing emission conditions but also points out
a deficiency in the system in that it has no way of differentiating be-
tween emission sources. Emissions originating from an ascension pipe or
oven door will appear as part of the total emission detected if they pass
between the light source and the camera. It might be necessary to
develop some rather arbitrary rules to cover this situation since it may
not always be possible to identify the exact source of the detected
emission.
Figure 30 presents data obtained during Test 25 performed on the AISI/
EPA car. During the first part of this test, an unusually heavy emission
release occurred around the drop sleeve area. This event was easily
identified on the optical system film record, but the optical system
digital clock time associated with the event was approximately one
minute later than the data acquisition time. The only reasonable explan-
ation was a missetting of the optical system .data clock in the minute
position. A rather remote possibility exists that a line noise pulse
caused the optical system clock to jump ahead one minute as opposed to
one second. A check of the two observed emission periods on the optical
159
-------
s
t-5
PM
Pi
£
s. 8
o ^>
o
co
JO
50
40
30
20
10
0
- 10
o
13
"9
- 8
7
- 6
5
- 4
3
- 2
1
.
A
\
\- OPTICAL SYSTEM
? MAXIMUM VALUE
1
;b i
1 (
I 4 \
1 \
1 I
1 1
i 1
I
I
1
'i KEY
i TOT
1 HYD
\ CON
\ — o — OPTI
I DAT/
, ». OPT:
A « INI)]
I 1
I
/ t
- 1 '^
N
- ^ //\
&L
ROCARBON
CENTRATION
CAL SYSTEM
L POINTS
[CAL TREND
[CATORS
11 08 00
20
40
11 09 00
20
40
11 10 00
20
40
11 11 00
TIME
FIGURE 29
TEST 20 DATA
-------
53
50
30
11 15
FIGURE 36
TEST 25 DATA
-------
data films show good agreement in spacing when compared to the continuous
monitoring system record. The steam condition also detected by the op-
tical system was identified as occurring about one minute before the
test providing a further basis for a time adjustment. Assuming that ,the
second portion of the optical trace should be associated with the first
peak in THC concentration and that the third portion of the trace should
be associated with the high THC concentration peak, good agreement is
observed in the data comparison. The third portion of the optical sys-
tem trace does not exhibit as high a value as might be expected from the
THC trace, but examination of the data film showed only a highly diffused
emission condition existing in the area of the light source at that time.
It is possible, however, that a mild wind condition at that time might
have blown the bulk of the diffused emission out of the camera view. It
is difficult to identify sharp emission peaks in a diffused smoke condi-
tion, but the general concentration conditions are indicated by a number
of frames selected over an extended period of time.
Several conclusions can be drawn based on the optical data completely
analyzed and the large amount of optical data film manually examined.
First, the system is capable of detecting trends in total emissions using
opacity measurements. Second, the mass concentrations appear reasonable
in light of the data available for comparison. Third, the system has
obvious limitations in measuring very heavy emission concentrations or
very high concentrations which are highly dispersed or which may be blown
laterally beneath the light source and thus pass undetected. In con-
trast, the minimal interference caused to plant operations and the sys-
tem potential for unattended operation offer positive reasons for con-
sideration of this concept for application to compliance monitoring.
162
-------
SELECT PROJECT DOCUMENTATION LIST
MTR 6055
MTR 6182
MTR 6215
WP 8788
MTR 6260
WP 10149
MTR 6288
WP 10179
WP 10459
WP 10480
WP 10445
MTR 6566
November 1971
May 1972
July 1972
15 July 1972
October 1972
14 Dec. 1972
December 1972
16 Feb. 1973
WP 10427 25 Oct. 1973
WP 7905 Vol. XXI 19 Nov. 1973
MTR 6546 November 1973
15 Jan. 1974
31 Jan. 1974
15 Feb. 1974
28 Feb. 1974
Management Plan
Overview of Measurements
Continuous Monitoring System
(Specifications COK-1 through 23)
Important Considerations Concerning
Sampling
Emissions Guide Systems
Design of an Optical Emissions
Measurement System for Coke Oven
Monitoring
Manual Sampling System
Manual Sampling and Analytical Re-
quirements for the Coke Oven Charg-
ing Emissions Test Program
Coke Oven Data Analysis Workbook
Test Plan for Coke Oven Emissions
Optical Emissions Measurement Pro-
gram for the Smokeless Coke Oven
Charging Demonstration
Monitoring Emissions from Leaking
Coke Oven Doors
Direct Impaction Particulate Collec-
tion Carrousel
Correction and Conversion Factors
of Coke Oven Emissions Data
A Continuous Monitoring System for
Coke Oven Emissions Due to Charging
163
-------
TECHNICAL REPORT DATA
(1 lease read Jnuniclions on the reverse before completing)
EPORT NO. 2 ~
PA-650/2-74-062
flTLE AND SUBTITLE
Poke Oven Charging Emission Control Test
Program — Volume I
AUTHORls)R.W.Bee, G.Erskine, R.B.Shaller,
R. W. Spewak , A. Wallo HI , and W. L . Wheaton
TERFORMING ORQANIZATION NAME AND ADDRESS —
The Mitre Corporation
Westgate Research Park
McLean, Virginia 22101
2. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
NERC-RTP, Control Systems Laboratory
Research Triangle Park, NC 27711
3. RECIPIENT'S ACCESSION-NO.
5. REPORT DATE
July 1974
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
M74-45
10. PROGRAM ELEMENT NO.
1AB013; ROAP 21AFF-004
1 1. CONTRACT/GRANT NO-EPA~IAG-
F192628-71-C-002 and
Contract 68-02-0650
73. TYPE OF REPORT AND PERIOD COVERED
Final- 4/71 Through 5/74
14. SPONSORING AGENCY CODE
5. SUPPLEMENTARY NOTES Although Volume H (Supporting Appendices) of this report was not
made available to NTIS , it is available through the Sponsoring Agency.
6. ABSTRACT
The report summarizes results of a coke oven charging emission control test progran
conducted at the P4 Battery of the Jones and Laughlin Pittsburgh Works between
April 1971 and May 1974; actual field testing was between May and August 1973.
Objectives of the test program were: to quantify atmospheric pollutants resulting from
the coking process charging operation; to provide a comparative evaluation of a
pollution abatement system (an improved design larry car versus an existing larry
car); and to determine the feasibility of a compliance monitoring system concept
based on optical measurement. All program objectives were accomplished:
emission characteristics of the charging operation have been defined in terms of both
gases and particulates released to the atmosphere. Emissions were also defined
from leaking seals on the pusher side doors of the oven. Several pertinent conclu-
sions were also developed relating to coke oven emissions measurement technology.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Air Pollution Emission
Iron and Steel Industry
Coke Measurement
Metallurgical Fuels Optical Measurement
Coking Monitors
Charging Sampling
Air Pollution Control
Stationary Sources
Manual Sampling
Larry Car
13B
11F
21D, 14B
20F
13H
^ DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
Form 2220-1 (9-73)
------- |