United States
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park NC 27711
EPA-450/5-80-008a
May 1980
Air
oEPA
A Statistical Study
of Coal Sulfur Variability
and Related Factors
-------
EPA-450/5-80-008a
A Statistical Study of Coal
Sulfur Variability and Related Factors
by
George R. Warholic, John E. Morton, Yimin Ngan,
James E. Spearman, and Yvonne Harris
Foster Associates, Inc.
1101 Seventeenth Street, N.W.
Washington, D.C. 20036
Contract No. 68-02-2592
EPA Project Officer: Rayburn Morrison
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air, Noise, and Radiation
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
May 1980
-------
This report is issued by the Environmental Protection Agency
to report technical data of interest to a limited number of
readers. Copies are available free of charge to Federal
employees^ current contractors and grantees, and nonprofit
organizations - in limited quantities - from the Library
Services Office (MD35), Research Triangle Park, North
Carolina 27711; or, for a fee, from the National Technical
Information Service, 5285 Port Royal Road, Springfield,
Virginia 22161.
This report was furnished to the Environmental Protection
Agency by Foster Associates, Inc., Washington, D.C. in ful-
fillment of Contract Nos. 68-02-2592 and 68-01-5845. The
contents of this report are reproduced herein as received
from Foster Associates, Inc. The opinions, findings, and
conclusions expressed are those of the authors and not
necessarily those of the Environmental Protection Agency.
Mention of company or product names is not to be considered
as an endorsement by the Environmental Protection Agency.
11
-------
TABLE OF CONTENTS
Summary . .
1.0 Introduction
2.0 Coal Sulfur Variability 2-1
2.1 Background 2-1
2.2 Types of Sulfur Dioxide Emission-Limiting
Regulations 2-5
2.2.1 New Source Performance Standards
(NSPS) 2-5
2.2.2 State Implementation Plans (SIPs). . 2-7
2.2.3 Impact of the Length of the
Averaging Time Interval on Compli-
ance with Emission Regulations . . . 2-8
2.2.4 Impact of the Averaging Time Frame
and Allowable Exceedances on Compli-
ance with Emissions Regulations . . 2-12
2.3 Conversion of Coal Sulfur to Sulfur
Dioxide 2-22
2.4 Effect of the Choice of Statistical
Distribution on Compliance with Sulfur
Dioxide Emission Regulations 2-24
2.5 Theoretical Effects of Measurement Error
on the Relative Standard Deviation of
Lbs S/MMBtu 2-30
2.5.1 Estimates of Sampling and Analytical
Error 2-31
2.5.2 Analysis of the Impact of Measure-
ment Error on the RSD of Lbs
S/MMBtu 2-33
3.0 Methodology 3-1
3.1 Data Collection 3-1
3.2 Data Base 3-5
3.2.1 Coal Analysis Data File 3-5
3.2.2 Stack Monitoring Data File 3-5
3.2.3 Index to Mine Locations and Seams
Produced 3-6
3.2.4 Maps of Mine Locations by Producing
District, State, and County .... 3-7
3.3 Analysis of Data 3-8
3.3.1 Plot of Variable Vs. Time 3-9
3.3.2 Sample Statistics 3-10
3.3.3 Frequency Distribution of Observed
Data 3-10
3.3.4 Comparison of Observed Distribution
to Expected Distribution 3-10
3.3.5 Goodness of Fit Between the
Observed and Expected Frequency
Distributions 3-12
111
-------
4.0 Analysis of Coal Data Received from
Respondents 4-1
4.1 Sample Statistics 4-2
4.1.1 Individual Data Sets 4-3
4.1.1.1 RSD Versus Lot-Size .... 4-4
4.1.1.2 Multiple Data Sets for
Individual Mines 4-7
4.1.1.3 Mine 10006 4-9
4.1.2 Lot-Size Interval Analysis of Data . 4-11
4.1.2.1 Theoretical Relationship
Between RSD of Lbs
S/MMBtu and Lot-Size . . . 4-12
4.1.2.2 Observed Relationship
Between RSD of Lbs
S/MMBtu and Lot-Size . . . 4-16
4.1.3 Analysis of Data on an Aggregate
Basis 4-21
4.2 Predictive Ability of Producing District
and Seam Data for Individual Mines .... 4-23
4.2.1 Methodology 4-24
4.2.2 Analysis of Predictive Capabilities
of Seam Data 4-24
4.2.3 Analysis of the Predictive Capa-
bilities of Producing District
Data 4-27
4.3 Analysis of the Statistical Distributions
of Coal Characteristics 4-27
4.3.1 Methods of Analysis 4-29
4.3.2 Results of Best Fit Analysis .... 4-33
5.0 Analysis of Bureau of Mines Data 5-1
5.1 Analysis of Selected Coal Seams 5-2
5.1.1 Northern Appalachian 5-2
5.1.2 Southern Appalachian 5-4
5.1.3 Mid-Continent 5-4
5.1.4 Western 5-4
5.1.5 Comparison of Analysis of Seam Data
from Bureau of Mines Data with Data
Received from Respondents 5-5
5.2 Analysis of Selected Mines 5-5
5.2.1 Mine 01200, Upper Freeport Seam . . 5-7
5.2.2 Mine 00950, Pittsburgh Seam .... 5-7
5.2.3 Mine 07290, Middle Kittanning Seam . 5-7
5.2.4 Mine 00614, Various Seams 5-10
5.2.5 Mine 00637, Upper Elkhorn #3 and
Hazard #4 Seams 5-10
5.2.6 Mine 02557, Upper Elkhorn #3 Seam . 5-10
5.2.7 Mines 04184 and 04204, Hazard #5-A
Seam 5-11
IV
-------
5.2.8 Mine 04978, Herrin #6 Seam . . 5-11
5.2.9 Mine 07310, Morris #2 Seam . . 5-12
5.2.10 Mine 04730, Indiana Seams #5,
#6, and #7 5-12
5.2.11 Mines 00305 and 07243, Black
Creek Seam 5-12
5.2.12 Mine 00750, Various Seams . . . 5-13
5.2.13 Mine 02020, F Seam 5-13
5.2.14 Mine 07235, Hiawatha Seam . . . 5-13
5.2.15 Summary of Analysis of
Individual Mine Data from
"Detail Tape" 5-14
6.0 Analysis of Continuous Monitoring Data for
Sulfur Dioxide Emissions 6-1
6.1 Description of Data 6-1
6.2 Analysis of Data 6-2
6.3 Implications of Emissions Analysis .... 6-6
7.0 Coal Sulfur Regression Analysis 7-1
7.1 Objectives 7-1
7.2 Background 7-1
7.3 Average Sulfur Content Regression Results . 7-5
7.3.1 Producing District 10 7-6
7.3.2 Producing District 8 7-10
7.3.3 Producing District 4 7-14
7.3.4 Mine #3 7-16
7.3.5 Summary of Average Sulfur Content
Regression Results 7-19
7.4 Sulfur Variability Regression Results . . . 7-22
7.4.1 Relative Standard Deviation
Regressions 7-22
7.4.2 Variance Regressions 7-26
7.4.3 Summary of Sulfur Variability
Regression Results 7-28
8.0 Conclusions and Recommendations 8-1
8.1 Conclusions 8-1
8.2 Recommendations 8-5
Appendix A: Derivation of Equations Used for the
Development of Tables 10, 11, and 12 . . A-l
Appendix B: Derivation of Formula for True and
Measured RSD B-l
Appendix C: Coal Sulfur Analyses Data Base Format . C-l
Appendix D: Stack Monitoring Data Base Format . . . D-l
Appendix E: Index to Mine Locations and Seams
Produced E-l
Appendipt F: Maps of Mine Locations by Producing
District, State and County F-l
v
-------
Appendix G: Sample Output of Analytical Program . . G-l
Appendix H: Summary of Salient Characteristics of
Data Sets Analyzed H-l
Appendix J: Analysis of Data on an Aggregate Basis
by Producing District J-l
Appendix K: Analysis of the Statistical Distribu-
tions of Coal Characteristics K-l
VI
-------
SUMMARY
Coal analysis data and limited continuous monitoring
data were obtained from various electric utilities, coal
companies, Government agencies, and other organizations.
These data were documented, edited, and placed into data-
bases for the analysis of coal sulfur variability. Statis-
tical programs were used to analyze coal analyses as well as
continuous monitoring data to assist EPA in its assessment
of the impacts of coal characteristics on compliance strate-
gies and emissions regulations.
Although the data gathered and analyzed in this study
appear to be the best currently available, it should be
noted that the data have certain limitations for use in
statistical analysis. In particular, the coal data were not
the result of controlled experiments, but rather historical
data, generally used for establishing coal prices and moni-
toring overall coal quality. In most instances the sources
of data reported whether ASTM sampling and analysis pro-
cedures were used. However, it could not be ascertained how
rigorously ASTM procedures were followed. Finally, it was
not possible to isolate the effects of factors such as mining
techniques and coal handling operations.
Coal analysis data, on a raw and washed basis, were
analyzed by individual mine, composite coal seams, and USBM
Producing Districts. Variables analyzed included volume
(tons), heat content (Btu/lb), sulfur content (weight per-
cent) and pounds of sulfur per million Btu (Ibs S/MMBtu).
The sample statistics from these analyses failed to identify
any consistent, predictable relationships which would explain
coal sulfur variabilities. This study concluded that com-
posite coal seam or Producing District data cannot be used
VII
-------
to accurately predict sulfur variabilities for individual
mines.
The statistical analyses conducted in this study were
based on a simple model assuming independent variance. More
sophisticated autocorrelative models were not investigated.
However, given the limitations of available data, it is
questionable whether such models would yield better results.
Visual examination of the time plots used in this study sug-
gests that the data sets contain little, if any, autocorrela-
tion. Nevertheless, these preliminary findings should not
preclude the investigation of autocorrelative models in
future studies.
Results of a simulation study, which assumed independ-
ence, indicated that theoretically coal sulfur variability
should decrease with increasing lot-sizes. However, in con-
trast to the theoretical results and some previous studies,
the coal data analyzed by sorting the data sets by lot-size
groupings and comparing the variabilities between lot-sizes
failed to provide strong support for an inverse relationship
at any level of aggregation. These results may be due, in
part, to the lack of statistical control.
Various regression analyses performed by mine, seam,
and Producing District failed to provide any good explana-
tory variables for coal sulfur variability. The results of
these analyses tend to support the hypothesis that the pri-
mary factors affecting coal sulfur distributions are geologic
factors, mining techniques, and coal handling procedures,
while chemical and physical properties of coal are secondary
factors.
viii
-------
Comparisons of the observed frequency distributions to
the normal, lognormal, and inverted gamma distributions
indicated that coal heat contents are best approximated by
the normal distribution, while sulfur contents and Ibs
S/MMBtu are best represented by the inverted gamma which
appears to be slightly superior to the lognormal distribu-
tion.
Analysis of the limited continuous monitoring data
indicated that significant reductions in the relative varia-
bility of emissions can be achieved by increasing the averag-
ing time interval from one-hour, to three-hours, to 24-
hours, to 30-days. These findings support the theoretical,
inverse relationship between coal sulfur variability and
lot-size, since increasing the averaging interval is equiva-
lent to increasing the lot-size of coal burned. However, it
should be noted that the reductions in relative variability
were less than would be expected based on statistical
independence.
The analysis of continuous monitoring data also indi-
cated that, while flue gas desulfurization (FGD) units reduce
the mean level of emissions, the relative variabilities of
FGD outlet SC>2 concentrations are substantially greater than
the FGD inlet SC>2 concentrations.
The various analyses of coal sulfur variability identi-
fied no reliable method for coal suppliers or consumers to
predict variability, which is often critical for compliance
with the existing sulfur emission-limiting regulations.
Coal sulfur variability is especially critical in the case
of small coal-fired boilers and regulations which stipulate
short averaging periods. This suggests that the language of
many current regulations is not consistent with the state of
knowledge concerning coal sulfur variability.
IX
-------
Included in this report are the results of related
studies which examine the theoretical impacts of measurement
error, choice of frequency distribution, and emission regula-
tions which require a very low probability of excess emis-
sions. Included in the Appendices of this report are sum-
maries of the sample statistics at various levels of aggre-
gation, sample output of the analytical programs, and
expanded discussions of several topics examined in the
report.
Within the overall objective of understanding the prob-
lem of coal sulfur variability and how coal producers, coal
consumers, and pollution control agencies can cope with this
problem, this study failed to provide simple explanations or
solutions. Instead, the results illustrated the complexi-
ties of the problems and indicated the need for additional
studies.
The data base and analyses in this report are not viewed
as exhaustive, but rather serve to establish a base from
which further studies can build in order to provide the
inputs necessary to understand the consequences of sulfur
variability vis-a-vis current sulfur dioxide emission regula-
tions. Ideally, future studies will develop explanatory
relationships which can be used in a comprehensive model to
assess the impact on air quality, given the parameters for
coal characteristics, mining and handling methods, combus-
tion and control equipment, meteorological data, and other
variables. Alternatively, these studies may provide data to
develop new regulations which would mitigate the impact of
coal sulfur variability yet achieve the objectives of existing
sulfur dioxide regulations.
-------
1.0 Introduction
The purpose of this report is to assist EPA in its
analysis of the impact of coal sulfur content variability on
the ability of both utility and industrial boilers to comply
with the sulfur emission regulations of the current State
Implementation Plans (SIP) and the existing Federal New
Source Performance Standards (NSPS).
Special objectives of this study were to:
(1) Collect and consolidate all existing data pertain-
ing to the variability of coal sulfur contents and
enter these data into a computer data base.
(2) Classify and document each data set collected with
respect to those factors which may influence vari-
ability in sulfur content.
* *
(3) Physically locate the sources of the coal data
sets on a map of the United States.
(4) Use the data base to prepare a report providing
information on coal sulfur variability for low,
medium, and high sulfur coals throughout the
United States.
(5) Analyze the impact of coal sulfur variability on
the ability of coal-fired boilers to comply with
sulfur emission regulations.
Data were obtained from coal companies, electric utili-
ties, EPA files, Bureau of Mines, and previous studies of
coal sulfur variability.
1-1
-------
This study is basically an extension of an earlier EPA
study!/ concerning coal sulfur variability which focused
primarily on low-sulfur coals. This study examines the
findings of the earlier study through the use of additional
statistical techniques and a more extensive data base for
low-sulfur coals as well as medium- and high-sulfur coals.
In recent years air pollution control agencies and
facilities affected by SIP's and NSPS have been increasingly
aware of the problems associated with the impact of coal
sulfur variability on meeting sulfur emission regulations
which specify an emission ceiling never to be exceeded during
short-term averaging periods such as one hour, three hours,
24 hours or one month. Germane to the understanding of this
problem is an examination of how the short-term sulfur emis-
sions from a coal relate to the nominal or long-term average
sulfur and heat contents.
Coal is not a homogeneous commodity and is subject to
variations in physical characteristics. The degree of vari-
ation in the sulfur and heat contents has a substantial
impact on the ability of a coal to comply with sulfur emis-
sion regulations. This report attempts to quantify these
variations for various coal mines, seams, and producing
areas and to identify those factors which may contribute to
these variations.
Some insight into the impact of coal sulfur variability
can be gained from a review of the results from an EPA-
sponsored study on Louisville Gas and Electric Company's
I/ PEDCo Environmental, "Preliminary Evaluation of Sulfur
Variability in Low-Sulfur Coals From Selected Mines", EPA
Publication No. EPA-450/3-77-044, July 1977.
1-2
-------
Cane Run Unit No. 4.i/ The results of this study exhibited
an average emission rate of 0.95 Ibs SC>2/MMBtu for the entire
test period, during August-December, 1977. However, because
of the inherent coal sulfur variability and variability associ-
ated with the desulfurization processes there was a substan-
tial variation of emissions about the mean.
An analysis of the data indicated that the 24-hour aver-
ages were lognormally distributed. The geometric means,
geometric standard deviations, and implied exceedance rates
are set out in Table 1. From Table 1 the reduction in varia-
bility is readily apparent from the declining geometric
standard deviations and exceedance rates as the averaging
interval, and consequently the volume or lot-size of coal
burned, is increased.
TABLE 1
COMPARISON OF GEOMETRIC MEANS, GEOMETRIC
STANDARD DEVIATIONS, AND IMPLIED EXCEEDANCE
RATES FOR SELECTED AVERAGING INTERVALS
Averaging Interval
3-Hour 24-Hour 7-Day 14-Day 30-DaY
No. of Observations 678 89 12 6 3
Geometric Mean 0.885 0.908 0.95 0.95 0.945
Geometric Standard Deviation 1.462 1.352 1.26 1.17 1.111
Exceedance Rate (Percent) 21 18 16 7 1
I/ Based on the results reported in the publication "Air
Pollution Emission Test", Vol. I, as well as additional
analyses by the Energy Strategies Branch of the Environ-
mental Protection Agency. USEPA, Office of Air Quality
Planning and Standards, Emission Measurement Branch, "Air
Pollution Emmission Test, Volume I, First Interim Report:
Continuous Sulfur Dioxide Monitoring at Steam Generators",
EMB Report No. 77SPP23A, August, 1978.
1-3
-------
2.0 Coal Sulfur Variability
2.1 Background
The initial New Source Performance Standards (NSPS)
applicable to S02 emissions from fossil fuel-fired steam
generating units of more than 250 MMBtu/hr heat input pro-
hibit S02 emissions in excess of 1.2 Ibs/MMBtu when solid
fossil fuel is burned.I/ This standard was promulgated on
December 23, 1971. Since 1971 several studies have indi-
cated that the definition of NSPS "complying coal" required
to meet this standard is more complex than was originally
envisioned in the background information for the standard
due to the variability of sulfur and heat contents of coal
burned during short time periods, e.g., one hour or 24 hours,
Based on available data, it appears that much of the previ-
ously identified "complying coal" would result in excess SO2
emissions when burned.
In addition to the NSPS, many State Implementation
Plans (SIPs) also specify a sulfur or sulfur dioxide emis-
sions ceiling for coal-fired generating units. Although in
many cases these regulations are less stringent than the
NSPS, the definition of a complying coal, based on the coal
sulfur content, is equally complex.
The basic problem in defining a compliance coal is that
coal, even in a narrowly defined producing area, is not a
homogeneous commodity, but is subject to variations in physi-
cal characteristics. The variations in coal sulfur and heat
I/ Applicable to plants for which construction commenced
between.August 17, 1971 and September 18, 1978. Plants for
which construction commenced after September 18, 1978 are
subject to a revised NSPS.
2-1
-------
contents jointly contribute to variations in the emissions
(pounds of SC>2/MMBtu or pounds of S/MMBtu) resulting from
coal combustion.
Sulfur in coal occurs in three forms: organic, sulfate,
and pyritic. Organic sulfur is an integral part of the coal
and generally cannot be removed by existing coal cleaning
techniques. Organic sulfur generally comprises about 30 to
70 percent of the total sulfur content of most coals. Sulfur
in the sulfate form is a water soluble oxidation product
formed from the weathering of the iron sulfide in coal and
can be readily'removed through coal cleaning. Sulfate sulfur
contents are usually less then 0.05 percent. Pyritic sulfur
occurs in coal as pyrite and/or marcasite, which are iron
sulfides. Pyrite is a relatively heavy mineral with a speci-
fic gravity of 5.0 compared to coal which has a specific
gravity of 1.7 or less. The pyrite content of most coals
can be significantly reduced through current coal prepara-
tion processes.
The total sulfur content (organic, sulfate, and pyritic)
of coals in the United States generally ranges from about
0.2 percent to 7.0 percent. The total sulfur content, as
well as the ratio of organic to inorganic sulfur, varies
widely among coal seams, geographical locations, and fre-
quently among mines operating in the same coal seam in the
same geographical location. These natural variations in
sulfur content, as well as variations in other physical
characteristics, are attributed to many factors which include:
1. Mode of accumulation and burial of coal-forming
vegetal matter.
2. Structure and chemical composition of the coal
forming vegetal matter.
2-2
-------
3. Extent of inert material washed into the coal
swamp at the time of accumulation.
4. Age of the coal deposit and its geographical loca-
tion.
5. Subsequent geologic history of the deposit, such
as permeation of ground water.
The variability of sulfur contents, as observed in coal
analyses, are a result of the above natural events plus num-
erous other factors resulting from various mining, process-
ing, transportation, sampling, and utilization techniques.
A summary of the pertinent factors which have been
identified as potential sources of coal sulfur variability
is set out in Table 2. Since many of these factors are
interrelated, a study of coal sulfur variability requires
well documented data in order to isolate and individually
examine those factors which may contribute to variability.
A detailed analysis of each of these factors is not possible
due to the large number of factors and their interdependence
and the lack of adequate data.
Although many electric utilities, coal companies, and
research organizations have suitable data, the cost of
assembling and analyzing these data in many cases is pro-
hibitive. In addition, several companies reported that they
have assembled and/or analyzed coal sulfur variability data
but the information is considered proprietary due to pending
legal actions or company policies.
2-3
-------
TABLE 2
PERTINENT FACTORS IN STUDYING SULFUR VARIABILITY
1. Type- of coal
Organic and inorganic sulfur content.
Distribution of sulfur in coal (coarse pyrite or finely
disseminated throughout the coal). The form in which
the sulfur occurs is significant when the coal is
washed.
2. Stage of sampling
Core drilling (or channel samples after operation).
Run-of-mine production.
After preparation, cleaning.
As received at utility plant/consumer.
As burned.
3. Coal blending and processing
4. Mining plan (selective)
5. Mining technique
Number and location of machines, type of mining
6. Location of coal
Seam.
Mine.
Region or district.
7. Averaging times/tonnages
Daily.
Weekly.
Monthly.
Other.
8. Sampling procedure
Amount of coal sampled.
Method of collecting increments for gross sample.
Sample variation.
9. Analytical method
10. Cleaning technique
Source: PEDCo Environmental, "Preliminary Evaluation of
Sulfur Variability in Low-Sulfur Coals for Selected
Mines", EPA Publication No. EPA-450/3-77-044, July
1977.
2-4
-------
The impact of the variations in the physical character-
istics of coal is largely dependent upon the specific require-
ments of a given regulation. One of the most pertinent
factors concerns the specified or implied averaging period
associated with the regulation. Previous studies based on
statistical theory indicate that coal variability increases
when coal samples and analyses are based on successively
smaller volumes of coal or shorter averaging periods. How-
ever, studies of individual data sets have produced incon-
sistent results.
2.2 Types of Sulfur Dioxide Emission-Limiting Regulations
A review of the existing sulfur dioxide emission-limiting
regulations for the electric utility industry clearly shows
the complexity and diversity of the regulations applicable
to coal-fired generating units. These regulations vary with
respect to the units of measure in which the limitation is
expressed as well as the scope of equipment (boiler, stack,
or entire plant) to which the regulations apply. In addi-
tion, the regulations may or may not specify an averaging
time. When averaging periods are specified in the regula-
tions, they vary among the individual regulations, while
unspecified averaging times have generally resulted in some
assumed averaging time by the plant operators or the pollu-
tion control agencies. The following discussion examines
the diversity of sulfur dioxide emission-limiting regula-
tions and provides some perspective as to the number of
plants and annual coal requirements controlled by specific
types of regulations.
2.2.1 New Source Performance Standards (NSPS)
The NSPS limits the emissions of sulfur dioxide (SC>2)
to 1.2 Ibs SC>2/MMBtu of heat input. This standard applies
2-5
-------
to all coal-fired steam-generating units rated at more than
250 MMBtu/hr heat input. Although the NSPS specifies test
methods for determining excess emissions, the NSPS is
ambiguous with respect to the averaging time required for
compliance with the 1.2 Ibs SC>2/MMBtu limit and has been
subject to a variety of interpretations.
A survey conducted in early 1978 identified 23 coal-
fired electric utility generating units operating under the
NSPS.I/ Based on a total installed capacity of 9,078 MW,
these 23 units would require approximately 32 million tons
of coal per year.^/
The survey revealed that a variety of averaging inter-
vals were being used for reporting S02 emissions. A summary
of the reported averaging intervals is set out in Table 3.
TABLE 3
SUMMARY OF AVERAGING TIMES FOR S02 EMISSIONS
AS REPORTED BY NSPS UNITS
Averaging Interval Number of Units
24-hour 1
3-hour 4
3-hour moving 4
2-hour 1
1-hour 2
-------
2.2.2 State Implementation Plans (SIPs)
A similar survey of the distribution and applicability
of the various SIP sulfur dioxide emission-limiting regula-
tions was conducted by EPA.A/ This study identified 373
power plants operating under SIP sulfur regulations. The
annual coal consumption of these plants was estimated at 446
million tons.
As shown in Table 4, the regulations vary with respect
to the units of measure in which the limitations are expressed
The regulations are most frequently expressed in terms of
Ibs S02/MMBtu which account for 61.2 percent of annual coal
consumption. Regulations limiting the percent sulfur of the
coal and the Ibs SC>2/hr control the second largest amount of
coal at 10.1 percent each. Limitations expressed in ppm of
S02, and Ibs S/MMBtu occur less frequently and control a
total of about 13.4 percent of annual coal consumption. In
addition, the SIPs applicable to 28 plants, which account
for approximately 5.2 percent of annual coal consumption,
specify no limit for sulfur dioxide emissions.
I/ "The Types of SIP SO2 Emission-Limiting Regulations:
Their Distribution and Applicability", Memorandum from R. D.
Bauman, .Chief, Energy Strategies Branch to W. C. Barber,
Director, Office of Air Quality Planning and Standards,
November 3, 1978.
2-7
-------
TABLE 4
SUMMARY OF SIP SULFUR DIOXIDE CONTROL REGULATIONS
Amount Controlled
Number of Annual Coal Consumption
Type of Regulation Plants (Million Tons) (Percent)
Lbs S02/MMBtu 215 273
Lbs S/MMBtu 19 15
Percent Sulfur 54 45
Lbs S02/hr 23 45
ppm S02
- at stack (exhaust gas) 3 6
- at ground (AAQ) 31 39
No emission limit 28 J3
Total 373 446 100.0
Source: EPA.
The previous discussion has focused on electric utility
plants due to the lack of publicly available data for the
industrial sector. Preliminary data indicate that coal con-
sumption in the industrial sector during 1978 was approxi-
mately 59 million tons.I/ Although coal consumption in the
industrial sector is substantially less than in the utility
sector, most large industrial plants are subject to sulfur
dioxide emissions-limiting regulations of the NSPS and SIPs.
However, available data do not permit a tabulation of the
number of plants and annual volumes of coal controlled by
specific regulations.
2.2.3 Impact of the Length of the Averaging Time Interval
on Compliance with Emission Regulations
Some insight into the impact of the length of the aver-
aging time interval can be gained from an examination of the
results of the EPA-sponsored study on the Cane Run Unit
I/ DOE, Weekly Coal Report, No. 77, March 23, 1979.
2-8
-------
No. 4, noted in the introduction of this report. Cane Run
Unit No. 4, operated by Louisville Gas and Electric Company,
has a rated capacity of 178 MW of electricity and is equipped
with a flue gas treatment system designed to achieve an 85
percent reduction of sulfur dioxide emissions, in order to
comply with a SIP regulation of 1.2 Ibs S02/MMBtu. The
plant is an intermediate-load facility which tracks the
system load. Estimated coal consumption for Unit No. 4
assuming operation at full capacity as well as at a 57 per-
cent capacity factor, at which the unit operated in 1976, is
set out in Table 5.
TABLE 5
ESTIMATED COAL CONSUMPTION, CANE RUN UNIT NO. 4
Volume (Tons)
Interval
1-hour
3-hour
24-hour
30-day
1-year ftc
Full Capacity
75
225
1,800
54,000
1^^657,000
57 Percent of Capacity
43
128
1,026
30,780
374,490
The nominal properties of the raw, Western Kentucky
coal burned by Cane Run Unit No. 4 are: 3.7 percent sul-
fur; 11,200 Btu/lb; and 12.6 percent ash.
The results of this study exhibited an average emission
rate of 0.95 Ibs S02/MMBtu for the entire test period, during
August-December, 1977. However, because of the inherent coal
sulfur variability and variability associated with the desul-
furization processes there was a substantial variation of
emissions about the mean. Set out in Table 6 are the mini-
mum and maximum values of emissions as well as the rates at
which emissions exceeded a limit of 1.2 Ibs SO2/MMBtu for
3-hour and 24-hour averaging intervals.
2-9
-------
TABLE 6
MAXIMUM AND MINIMUM EMISSIONS AND EXCEEDANCES
CANE RUN UNIT NO. 4
Percentage of
Emissions, Observations
Averaging Ib SO2/MMBtu No. of Exceeding,
Interval Minimum~ Maximum Exceedances 1.2 Ib SQ2/MMBtu
3-hour 0.19 2.45 184 24
24-hour 0.37 1.74 18 20
An analysis of the data indicated that the 24-hour aver-
ages were lognormally distributed. The geometric means and
geometric standard deviations for various averaging intervals
are set out in Table 7. The implied exceedance rates, calcu-
lated through the use of normal distribution theory for
logarithms, are also set out in Table 7.
TABLE 7
COMPARISON OF GEOMETRIC MEANS, GEOMETRIC
STANDARD DEVIATIONS, AND IMPLIED EXCEEDANCE
RATES FOR SELECTED AVERAGING INTERVALS
Averaging Interval
3-Hour 24-Hour 7-Day 14-Day 30-Day
No. of Observations 678 89 12 6 3
Geometric Mean 0.885 0.908 0.95 0.95 0.945
Geometric Standard Deviation 1.462 1.352 1.26 1.17 1.111
Exceedance Rate (Percent) 21 18 16 7 1
From Table 7 the reduction in the exceedance rate is
readily apparent as the averaging interval, and consequently
the volume of coal burned, is increased.
Figure 1 provides a graphical display of the variability
of emission rates over successive 3-hour and 24-hour intervals
for the period October 21 to November 4, 1977. Emissions for
this 15-day period averaged 0.92 Ibs S02/MMBtu. The 3-hour
2-10
-------
FIGURE 1
EMISSIONS FROM FLUE GAS DESULFURIZATION UNIT
Louisville Gas and Electrlc's Cane Run Unit No. 4
High-sulfur coal; maximum sustainable output = 178 MW
to
I
1.5
CXI
OJ
i.o
JQ
0)
4->
rO
O
0.5
Emission limit = 1.20
Mean =0.92
24-hour averages
3-hour averages
20 21 22 23 24 25 26 27 28 29
OCTOBER 1977
1
30 31 1 2 3
NOVEMBER 1977
Source: Based on data from "Air Pollution Emission Test", Vol. 1, analyzed by Energy Strategies
Branch, Office of Air quality Planning and Standards, USEPA.
-------
averages ranged from a low of 0.38 to a high of 1.71, while
the 24-hour averages ranged from 0.50 to 1.34. Based on a
standard of 1.2, the 3-hour averages exceeded the standard
20 times, while the 24-hour averages exceeded the standard
two times.
The Cane Run Unit No. 4 data are certainly not repre-
sentative of all coals or all flue gas desulfurization
systems. However, the data demonstrate that due to emis-
sions variability, the averaging basis associated with an
emission limit is critical to the determination of what con-
stitutes compliance with that limit. It follows from this
analysis that the smaller the amount of coal burned per unit
time, the more difficult it is to comply with the standard
if the averaging basis is time.
2.2.4 Impact of Averaging Time Frame and Allowable
Exceedances on Compliance with Emissions Regulations
In order to avoid ambiguity.- sulfur emission-limiting
regulations must specify the averaging time frame for the
emissions measurement and the number of exceedances permitted,
These exceedance restrictions are generally stated in terms
of the number of days per year or days per month for which
excess emissions are permitted. For example, a sulfur
dioxide emission regulation may require that the sulfur
dioxide emission level is 1.2 Ibs S02/MMBtu based on a 24-
hour average never to be exceeded or not to be exceeded more
than 2 days per year. Such a sulfur dioxide emission regu-
lation forces the coal user to consider not only the proba-
bility of exceeding the 24-hour average sulfur dioxide limit
but also the probability of exceeding the yearly allowable
exceedances.
2-12
-------
Suppose the probability of a source being in violation
of the sulfur dioxide emission level on any particular day
is known. The question then arises as to what is the expected
number of days, for any specified time frame, in which the
source would be in violation of the sulfur dioxide emission
regulation. The binomial distribution, for example, can be
used to illustrate the impact of the averaging time frame on
compliance.!./
The binomial distribution is a probability distribution
which describes independent, identically distributed random
samples. The number of days in violation of a sulfur stan-
dard over a 365-day period can be considered to be binomially
distributed if the probability of violation on the first day
equals the probability of violation on the second or any
other day. The equation which describes this distribution
is as follows:
P(X = k)
k! (n - k) !
where P(X = k) = probability of k violations in n days
p = probability of violation on any particu
lar day
n = number of days being considered
k = number of violations being considered
! = factorial
I/ Determining the probability of a source being in viola-
tion on a particular day is a complex statistical problem.
The probability is not likely to be constant and is a func-
tion of the emission level on previous days (autocorrelative
process). To illustrate the implications of alternative
averaging intervals a binomial distribution was assumed.
However/ this example i« not- f-^ auqgest i-hat the process^
should be modeled via the binomial distribution.
2-13
-------
A simple way to calculate these probabilities for the series
k=0, k=l, etc., is to use the following pair of equations:
P(X = 0) = (1 - p)"
p
-------
FIGURE 2
.8
FREQUENCY DISTRIBUTION OF EXCEEDANCES
n = 365
p = 0.001, 0.005, 0.020
1
1
**
>*
1
(
p=.001
1
1
p. -.
P-.005
\/
\
,.,
^ ^v ^v «
;
'
'
r~ ~"
i
J
M M
1
I
1
I
1
1
p=.020
i
I
j
i
i
1
t
'
1
1
I- -,
1
.7
a
0
10
>>
03
,5
£
3
V
!».<
X)
!;
.1
3456
Number of days in violation
10
2-15
-------
The probability of five or more exceedances is so low that
it can be ignored in this calculation.
Tables 8 and 9 compare the expected exceedances with
the most likely number of exceedances, for a 365-day and 30-
day period, respectively. Note that the most likely number
of exceedances is usually lower. Phrases such as "one-day-
per-year standard" or "two-day-per-year standard" must be
used with caution because these phrases do not distinguish
between the expected exceedances and the most likely number
of exceedances.
Figure 3 shows the probability P (X > k) on the vertical
axis, as a function of p on the horizontal axis, assuming
n = 365 and k = 3 or k = 6. Figure 4 shows the same informa-
tion for k = 0 through k = 6. Figure 5 is similar to
Figure 4 except that n = 30.
Note that the binomial distribution assumes that the
probability of violation is the same on each day. This con-
dition may not be satisfied by a coal burning plant. Many
factors such as changes in coal quality, changes in the load
or burn ratio, and technical problems can affect the proba-
bility of violation on any particular day. The use of moving
averages also introduces autocorrelation which changes the
probability on a day-to-day basis.
2-16
-------
TABLE 8
Comparison of expected exceedances with the most likely
number of exceedances, over 365 days.
Probability of
Exceedance
on a Given Day
.001
.002
.003
.004
.005
.006
.007
.008
.009
.010
.011
.012
.013
.014
.015
.016
.020
Expected
Exceedances
.4
.7
1.1
1.5
1.8
2.2
2.6
2.9
3.3
3.7
4.0
4.4
4.7
5.1
5.5
5.8
7.3
Most Likely
Number of
Exceedances
0
0
1
1
1
2
2
2
3
3
4
4
4
5
5
5
7
TABLE 9
Comparison of expected exceedances with the most likely
number of exceedances, over 30 days.
Probability of
Exceedance
on a Given Day
.01
.02
.03
.04
.05
.06
.07
.08
.09
.10
.11
.12
.13
.14
.15
Expected
Exceedances
.3
.6
.9
1.2
1.5
1.8
2.1
2.4
2.7
3.0
3.3
3.6
3.9
4.2
4.5
Most Likely
Number of
Exceedances
0
0
0
1
1
1
2
2
2
3
3
3
4
4
4
2-17
-------
to
I
M
00
.5
8
I*
§
.o
I
.5
IA
f
X
e -
o
E.2
I
.1
0
FIGURE 3
PROBABILITY OF MORE THAN X DAYS IN VIOLATION PER YEAR AS FUNCTION OF THE
PROBABILITY OF VIOLATION OF AN INDIVIDUAL DAY
(n = 365, k = 3 and 6)
Probability of more than 3
in violation during a year
Probability of more than 6
days in violation during a year
.005 .010
Probability of violation on an individual day
.015
-------
I
M
VO
FIGURE 4
PROBABILITY OF MORE THAN X DAYS IN VIOLATION PER YEAR AS A FUNCTION OF THE
PROBABILITY OF VIOLATION ON AN INDIVIDUAL DAY
(n = 365, k = 0 to 6)
.005 .010
Probability of violation on an individual day
.015
-------
FIGURE 5
'RUBABILITY OF MORE THAN X DAYS IN VIOLATION PLR JO DAYS AS A FUNCTION
OF THE PROBABILITY OF VIOLATION ON AN INDIVIDUAL DAY
(n = JU, k = 0 to 6)
0
.05 .10
Probability of violation on an individual day
.15
-------
The results of this analysis indicate two important
statistical characteristics.
The first is that the probability of violation in any
individual day must be much lower to limit the probability
of exceeding a given number of violations than if one was
content to simply average the same number of violations.
For example, the probability of violation in any individual
day in order to have an expected number of 2 violations per
year is 2/365 or .0055. From Figure 4 it can be shown that
a .0055 probability of violation on an individual day pro-
duces a probability of exceeding more than two days per year
of approximately .33. In order to reduce the probability of
more than 2 violations in a year to approximately 1 percent,
it would require a probability of violation on any individual
day of .0016, again from Figure 4. At a probability of
.0016, the expected number of violations in a year, over a
long-term average, is .0016 x 365 or 0.58. Obviously, in
order to generate a very low probability of exceeding a
given number of days within a particular year, the average,
number of violations must be reduced far below that number.
The second statistical characteristic is that a binomial
distribution is very sensitive to the number of possible
violations within the time frame of the regulatory standard.
Consider for example the effect of using an hourly standard
as opposed to a daily one. For an hourly measurement, the
standard is 365 x 24 or 8760 possible chances for violation
within a year. Suppose the standard were written in a way
that the probability of exceeding 2 or more days within a
year had to be less than 10 percent. Under these conditions
Figure 4 shows the expected probability of violation on any
day would have to be .003. If hourly measurements were
taken, then the probability of violation within any given
2-21
-------
hour would be .000043. Thus, the probability of violation
within any given time frame is extremely sensitive to the
number of possible violations.
2.3 Conversion of Coal Sulfur to Sulfur Dioxide
As previously stated, the principal objective of this
report is to provide data and information to assist EPA in
its analysis of the impact of coal sulfur variability on
compliance with sulfur dioxide emissions-limiting regula-
tions. However, translating coal quality analyses to poten-
tial emissions presents several problems.
First, as discussed in this section of the report,
sulfur dioxide emission-limiting regulations vary with
respect to the units of measure (Ibs SC^/MMBtu, ppm SC>2,
percent sulfur, etc.) in which the limitations are
expressed. As such, comparisons with a specific regulation
require conversion to the appropriate measurement units.
Second, translating coal analysis data to potential or
theoretical emissions requires some assumptions concerning
the combustion characteristics of the coal. One of the most
important assumptions concerns the conversion of coal sulfur
to flue gas sulfur dioxide (SC>2) , which may subsequently
react with more oxygen, forming sulfur trioxide (803) or
sulfate radicals in a complex equilibrium. These reactions,
combined with the fact that coal sulfur exists in various
forms (pyritic, sulfate, and organic), result in only a por-
tion of the total sulfur in coal being emitted as SO2-
Various studies reviewed by EPA have indicated that for
bituminous and subbituminous coals approximately two percent
of the coals' sulfur is retained in the fly ash, about two
2-22
-------
percent is converted to 803, and about one percent is
retained in the slag or bottom ash. Thus, as an approxima-
tion, about 95 percent of the total sulfur in coal is emitted
as SC>2 from an uncontrolled boiler. It should be noted,
however, that in specific cases, conversion factors of less
than 90 percent were observed.!./
In the case of lignite coals, the various studies reviewed
by EPA have indicated conversion factors ranging from 98
percent to as low as 50 percent. The wide range of conver-
sion rates is related to the presence of reactive alkali
substances (sodium, calcium, magnesium, and potassium) and/or
clay and silica contents. In general, the alkali substances
decrease the conversion rate while increased clay and silica
contents increase the conversion rates.
The above discussion indicates that any assumption
which attempts to generalize the conversion of coal sulfur
to SO2 emissions may be tenuous. This problem of estimating
exhaust stack sulfur dioxide emissions from coal character-
istics is further compounded by the incidental effects of
coal processing equipment and control equipment not specifi-
cally designed for sulfur removal.
The first situation occurs when a plant utilizes a type
of coal pulverizer which rejects a portion of the pyritic
sulfur contained in the coal. Although the sulfur removal
efficiency of such pulverizers is low compared to conven-
tional coal cleaning facilities, in some cases, involving
I/ USEPA, Office of Air Quality Planning and Standards,
Emission Standards and Engineering Division, "Background
Information: Fuel Analysis Provisions for Performance
Testing and Emission Monitoring of Sulfur Dioxide Emissions
from Fossil Fuel Fired-Steam Generators", January 1977,
Draft.
2-23
-------
coals with high pyritic sulfur contents, the results may be
significant.
The second situation relates to control equipment
designed for removal of particulates. Although not specifi-
cally designed to control sulfur dioxide emissions, particu-
late control equipment frequently removes significant por-
tions of the sulfur retained in the fly ash.
In view of these various problems relating to the esti-
mation of potential sulfur dioxide emissions, the calcula-
tion of emissions in this report is based on the ratio of
sulfur to heat content (Ibs S/MMBtu) indicated by the
laboratory analyses of coal samples.
2.4 Effect of the Choice of Statistical Distribution on
Compliance with Sulfur Dioxide Emissions Regulations
In this report frequency distributions were analyzed to.
determine which type of distribution normal, lognormal,
or inverted gamma most closely fit the observed distribu-
tions. Some of the previous studies on coal sulfur varia-
bility assumed a normal distribution, primarily due to the
lack of adequate data and the statistical simplicity of the
normal distribution. This section addresses the consequences
of the choice of a particular distribution, as they relate
to compliance with emissions regulations.
The methodology of this analysis included the calcula-
tion of the mean or average Ibs SC>2/MMBtu required to meet
sulfur dioxide emissions standards ranging from 1.0 to 6.0
Ibs SC>2/MMBtu, based on relative standard deviations (RSD's)
2-24
-------
of the Ibs S02/MMBtu ranging from 5 to 30 percent.i/ In
addition, it was assumed that the stringency of the regula-
tions required only a 0.5 percent probability of exceeding
the standard on a given day. These calculations are set out
in Tables 10, 11, and 12 for the respective normal,
lognormal, and inverted gamma distributions.2/
Table 10, for example, shows that under the assumption
of a normal distribution an emissions regulation of 1.2 Ibs
S02/MMBtu would require a mean emissions of 1.063 Ibs
SO2/MMBtu or less, if the emissions had an RSD of 5 percent.
As the RSD of the emissions increases to 30 percent, and
beyond, successively lower mean Ibs SO2/MMBtu are required
for compliance.
I/ The relative standard deviation (RSD) is the ratio of
the standard deviation to the mean, expressed as a percent-
age. The RSD provides a measure of the relative dispersion
about the mean and is also called the coefficient of varia-
tion or coefficient of dispersion.
2/ The derivation of the equations used for the development
of Tables 10, 11, and 12 is set out in Appendix A.
2-25
-------
TABLE 10
NORMAL DISTRIBUTION
Calculation of the mean Ibs SO2/MMBtu required to meet
a given standard, assuming a normal distribution, a
0.5% probability of exceeding the standard on an
individual day, and a given relative standard deviation
for the averaging period.
Emissions
Standard Relative Standard Deviation of Lbs SO?/MMBtu (%) _<_
(Lbs SO?/MMBtu) 5 10 15 20 75 30
(Mean Lbs SC>2/MMBtu Required to Meet Emissions Standards)
1.0
1.2
1.4
1.6
1.8
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
Source: Foster Associates, Inc.
.886
1.063
1.240
1.417
1.595
1.772
2.215
2.658
3.101
3.544
3.987
4.430
4.872
5.315
.795
.954
1.113
1.272
1.431
1.590
1.988
2.386
2.783
3.181
3.578
3.976
4.373
4.771
.721
.866
1.010
1.154
1.298
1.443
1.803
2.164
2.525
2.885
3.246
3.607
3.967
4.328
.660
.792
.924
1.056
1.188
1.320
1.650
1.980
2.310
2.640
2.970
3.300
3.630
3.960
.608
.730
.852
.973
1.095
1.217
1.521
1.825
2.129
2.433
2.737
3.041
3.346
3.650
.564
.677
.790
.903
1.015
1.128
1.410
1.692
1.974
2.256
2.538
2.820
3.103
3.385
2-26
-------
TABLE 11
LOGNORMAL DISTRIBUTION
Calculation of the mean Ibs SO2/MMBtu required to meet
a given standard, assuming a lognormal distribution, a
0.5% probability of exceeding the standard on an indi-
vidual day, and a given relative standard deviation for
the averaging period.
Emissions
Standard Relative Standard Deviation of Lbs SO?/MMBtu (%)
(Lbs SO-?/MMBtu) 5 10 15 20 ~25 30
(Mean Lbs S(>2/MMBtu Required to Meet Emissions Standard)
1.0
1.2
1.4
1.6
1.8
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
Source: Foster Associates, Inc.
.880
1.056
1.232
1.409
1.585
1.761
2.201
2.641
3.081
3.521
3.961
4.402
4.842
5.282
.777
.933
1.088
1.244
1.399
1.555
1.943
2.332
2.720
3.109
3.498
3.886
4.275
4.664
.689
.826
.964
1.102
1.239
1.377
1.721
2.066
2.410
2.754
3.099
3.443
3.787
4.132
.612
.735
.857
.980
1.102
1.225
1.531
1.837
2.143
2.449
2.755
3.062
3.368
3.674
.547
.656
.765
.875
.984
1.093
1.367
1.640
1.913
2.187
2.460
2.733
3.007
3.280
.490
.588
.686
.784
.882
.980
1.225
1.470
1.715
1.961
2.206
2.451
2.696
2.941
2-27
-------
TABLE 12
INVERTED GAMMA DISTRIBUTION
Calculation of the mean Ibs S02/MMBtu required to meet a
given standard, assuming an inverted gamma distribution,
a 0.5% probability of exceeding the standard on an indi-
vidual day/ and a given relative standard deviation for
the averaging period.
Emissions
Standard Relative Standard Deviation of Lbs SO?/MMBtu (%)
(Lbs SO->/MMBtu) 5 10 15 20 ~25 30
(Mean Lbs SC>2/MMBtu Required to Meet Emissions Standard)
1.0
1.2
1.4
1-6
1.8
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
Source: Foster Associates, Inc.
.877
1.053
1.228
1.404
1.579
1.755
2.193
2.632
3.070
3.509
3.948
4.386
4.825
5.264
.767
.920
1.073
1.227
1.380
1.533
1.916
2.300
2.683
3.066
3.450
3.833
4.216
4.600
.678
.813
.949
1.084
1.220
1.356
1.694
2.033
2.372
2.711
3.050
3.389
3.728
4.067
.596
.716
.835
.954
1.073
1.193
1.491
1.789
2.087
2.385
2.683
2.981
3.279
3.578
.528
.633
.739
.844
.950
1.055
1.319
1.583
1.847
2.111
2.375
2.639
2.902
3.166
.461
.553
.645
.737
.829
.921
1.152
1.382
1.613
1.843
2.073
2.304
2.534
2.764
2-28
-------
The impact of the choice of distributions can be assessed
by a comparison of Tables 10, 11 and 12. For example, assume
a coal determined to have an RSD of 5 percent and an emis-
sions standard of 1.0 Ibs S02/MMBtu. If the emissions were
normally distributed, a mean of 0.886 Ibs SC>2/MMBtu would be
required for compliance. Alternatively, lognormal and
inverted gamma distributions would require means of 0.880
and 0.877, respectively.
Additional comparisons of Tables 10, 11, and 12 show
that the impact of the choice of distribution is almost
insignificant when stringent emissions standards and coals
with low variability are considered. However, as the
stringency of the emissions standard declines and more
importantly, as the variability increases, the difference
becomes more significant. For example, based on a 5.0 Ibs
SC>2/MMBtu emission standard and a coal with an RSD of 30 per-
cent the normal distribution would require a mean of 2.82
Ibs SC>2/MMBtu for compliance, while the inverted gamma dis-
tribution indicates 2.30 Ibs S02/MMBtu.
In summary, when highly variable coals and less restric-
tive emissions standards are considered, the choice of the
frequency distribution of Ibs SO2/MMBtu becomes increasingly
important. This analysis indicates that when dealing with
the NSPS the choice among these three distributions is
almost insignificant, since the differences among the means
required for compliance in most cases are within the accepta-
ble limits for error in the ASTM coal sampling and analysis
procedures. However, in the case of SIPs which frequently
permit a higher level of emissions, the choice of distribu-
tion cannot be ignored when a highly variable source of coal
is utilized.
2-29
-------
2.5 Theoretical Effects of Measurement Error on the Relative
Standard Deviation of Pounds of Sulfur per Million Btu
As discussed in Section 2.3, sulfur emissions in this
study were calculated as the ratio of sulfur content to heat
content (Ibs S/MMBtu), due to problems in translating coal
sulfur contents to sulfur dioxide emissions. The calculation
of Ibs S/MMBtu requires separate measurements for the sulfur
and heat contents of coal. Each of these measurements is
subject to two sources of error. The first source of error
is sampling error which may result in coal samples that are
not representative of the true coal population. The second
source of error arises from the analytical or laboratory
techniques used to chemically analyze the coal samples.
This analysis is not based on observed data but instead
examines the theoretical effects of measurement error in
coal sampling and analysis resulting from ASTM standards and
procedures. The analysis of the impact of measurement error
was performed in a three-step process. First, estimates of
the measurement and sampling error were developed for both
sulfur and heat contents. Second, the mathematics required
to determine how these errors affect the Ibs S/MMBtu were
derived. Finally, the results of the first two steps were
used to calculate the impact of measurement error on coal
sulfur variability, defined as the RSD of Ibs S/MMBtu.
In the next two sections, measurement and sampling
errors are derived and their effects upon the RSD of Ibs
S/MMBtu are examined. The derivation of the mathematical
formula which describes the impact of measurement and
sampling errors upon sulfur variability is set out in
Appendix B.
2-30
-------
2.5.1 Estimates of Sampling and Analytical Error
With respect to the analytical error for percent sulfur,
the ASTM standards state:
"16.2 Reproducibility The means of results of dupli-
cate determinations carried out by different laboratories on
representative samples taken from the same bulk sample after
the last stage of reduction should not differ by more than
the following:
Coal containing less than 2% sulfur 0.10%
Coal containing 2% sulfur or more 0.20%!/"
RSD, as previously defined, is the standard deviation
divided by the mean. Under the assumption that the error in
percent sulfur is normally distributed and the standard
deviation estimates represent approximately the 99th per-
centile (3 standard deviations), the RSD of the error intro-
duced by analytical techniques for a coal with a sulfur con-
tent of 1.5 percent is calculated as:
0>:L = .022 or 2.2 percent
X J X 3
Applying these assumptions to various coal sulfur contents
yields the following estimates of the RSD of the error intro-
duced by analytical techniques:
I/ ASTM.D 3177-75, "Standard Test Methods for Total Sulfur
in the Analysis Sample of Coal and Coke", 1978 Annual Book
of ASTM Standards, Part 26, p. 399.
2-31
-------
RSD of the Error
Coal Sulfur Introduced by
Content (%) Analytical Techniques (%)
0.5 6.7
1.5 2.2
2.5 2.7
3.5 1.9
4.5 1.5
With respect to the measurement error in coal heat con-
tents, the relevant ASTM standard says:
"13.1.2 ' Reproducibility The results submitted by two
or more laboratories (different equipment, operators, date
of test and different portions of the same pulp) should not
be considered suspect unless the two results differ by more
than 100 Btu/lb, dry basis."!/
Based on a coal with an average heat content of 13,000
Btu per Ib, a difference of 0.77 percent would not be sus-
pect. Assuming a normal distribution and three standard
deviations, the acceptable analytical error for heat content
translates to an RSD of approximately 0.26 percent.
With respect to sampling error, the general purpose
sampling procedure is intended to provide a precision of
plus or minus one-tenth of the ash content of the coal sampled
in 95 out of 100 cases.^/ Based on the general purpose
sampling procedures, the precision intended for ash is also
applicable to sulfur and heat contents. Thus, the RSD of
I/ ANSI/ASTM D 2015-77, "Standard Test Method for Gross
Calorific Value of Solid Fuel by the Adiabatic Bomb Calori-
meter", 1978 Annual Book of ASTM Standards, Part 26, p. 307,
2/ ASTM D 2234-76, "Standard Methods for Collection of a
Gross Sample of Coal", 1978 Annual Book of ASTM Standards,
Part 26, p. 310.
2-32
-------
the sampling error for both sulfur and heat contents would
result in a value of 5 percent.I/
As this discussion shows, the major source of variance
is in the collection of gross samples. Note that these
estimates assume the ASTM measurements are followed exactly.
If they are not, the RSD estimates are likely to be larger.
2.5.2 Analysis of the Impact of Measurement Error on the
RSD of Lbs S/MMbtu
The previous discussion shows that an evaluation of the
impact of measurement error on the RSD of Ibs S/MMBtu requires
an assessment of the impact of four separate error terms:
RSD of error in sulfur measurement due to sampling
RSD of error in sulfur measurement due to analysis
RSD of error in heat content measurement due to
sampling
RSD of error in heat content measurement due to
analysis
The mathematical formula which describes the relation-
ship of these error terms to measured and true RSD values
was derived and was used to construct Table 13.
Table 13 sets out the estimates of the RSD of Ibs S/MMBtu
which would be calculated from coal sample analyses along
with the true RSD after allowing for measurement error.
Table 13 shows, for example, that if a coal with an average
sulfur content of 2.5 percent exhibited an RSD of 10 percent
based on ASTM samples and analyses, the true RSD is only
6.6 percent. The difference between the measured and true
I/ Based on the most precise ASTM sampling classification
which requires unbiased, stopped belt cross-section incre-
ments, spaced evenly in time.
2-33
-------
RSD's (10.0-6.6 = 3.4) is attributed to measurement error.
The relationship between the measured RSD and the individual
error terms is such that the measured RSD is biased, result-
ing in overestimates of the true RSD. Note that there is
little difference between high- and low-sulfur coals. Also,
the difference between the measured and true RSD gets pro-
gressively larger as the measured RSD gets smaller. Calcu-
lations indicate that at a measured RSD of 7 to 8 percent,
the true RSD approaches zero.
TABLE 13
COMPARISON OF MEASURED RSD WITH TRUE RSD
(Lbs S/MMBtu)
Measured
RSD (%)
.8
10
12
14
16
18
20
22
24
26
28
30
32
34
36
38
40
42
44
46
True RSD (%) by Level of
Coal Sulfur Content (%)
0.5
2.5
7.1
10.1
12.7
15.1
17
19
,5
,7
21.9
24.1
26.3
28.4
30.5
32.6
34.7
36.7
38.8
40.9
42.9
45.0
1.5
3.0
6.7
9.4
11.8
14.2
16.4
18.5
20.7
22.8
24.9
27.0
29.0
31.1
33.1
35.2
37.2
39.3
41.3
43.4
45.4
2.5
2.7
6.6
9.3
11.7
14
16
18
20.6
22.7
24.8
26.9
29.0
31.1
33.1
35.2
37.2
39.2
41.3
43.3
45.4
3.5
3.2
6.8
9.5
11.9
14.2
16.4
18.6
20.7
22.8
24.9
27.0
29.1
31.1
33.2
35.2
37.2
39.3
41.3
43.3
45.4
4.5
<*^
3.4
6.9
9.6
12.0
14.3
16.5
18.7
20.8
22.9
25.0
27.0
29.1
31.2
33.2
35.3
37.3
39.3
41.4
43.4
45.4
Source: Foster Associates, Inc.
2-34
-------
3.0 Methodology
3.1 Data Collection
The approach used in this study consisted of two phases,
data collection and data analysis. In the first phase, all
relevant data pertaining to coal sulfur variability were
collected and consolidated in a computer data base. The
majority of the data sets in the data base reflect data
acquired with the assistance of the Edison Electric Institute
and have not been previously analyzed for coal sulfur varia-
bility.
Electric utilities, coal companies, and the Bureau of
Mines were the primary sources of the data included in this
study. Separate meetings were held with representatives of
the electric utility and coal industries to solicit data and
to provide comments and recommendations for the proposed
study. Based on these discussions, potential sources of
data were identified and contacted.
In the data collection phase of the study, electric
utilities, coal companies, and other organizations.!/ were
contacted. As set out below, 26 electric utilities, five
coal companies, and five other organizations responded with
data.
Number of Companies
Type of Companies Contacted Responded
(1) (2) (3)
Electric Utilities 69 26
Coal Companies 17 5
Other I/ JL .J.
Total 94 36
I/ Other organizations include research organizations,
Federal agencies, and industrial companies.
3-1
-------
Data collected in this study include stack monitoring
data and coal analysis data. Stack monitoring data relate
to the sulfur dioxide emissions (ppm or Ibs SC>2/MMBtu)
present in the exhaust gases resulting from combustion.
Coal analysis data reflect the measured physical and
chemical coal characteristics, determined from coal samples
sent to analytical laboratories.
A problem frequently encountered in the collection of
data, especially for coal analyses, was that companies did
not maintain data of the quality desired for this study. In
order to isolate those factors which may contribute to coal
variability, the following information was requested for
each data set:
A. Per coal source of supply represented by analyses
1. Coal source
a. Bureau of Mines Producing District
b. State
c. County
d. Seam(s)
e. Mine
2. Method of sampling (automatic or hand, ASTM
or non-ASTM)
3. Type of sample (core, as mined, as delivered,
as burned)
4. Analysis method (ASTM or other)
5. Degree of processing (run-of-mine, washed,
stoker, etc.)
6. Mining method (surface or underground)
3-2
-------
B. Per coal cample analysis
1. Date of sample or date of coal delivery
2. Lot-size (tons)
3. Sulfur content (percent)
4. Heat content (Btu/lb)
5. Ash content (percent)
6. Moisture content (percent)
Additionally, it was requested that each coal analysis
data set be restricted to coal analyses from one mine or one
coal seam and that ASTM sampling and analysis procedures
were preferred.
Particularly in the Appalachian producing areas, it was
found that the availability of data which satisfied these
criteria was extremely limited. In general, coal production
in this area is based on relatively small, multiple seam
mines and the coal is frequently shipped by truck directly
to the consumer, or is shipped to a central loading point
for volume or unit train shipments. In the case of truck
shipments, the coal is generally only spot checked by hand
sampling, and in the case of volume or unit train shipments,
the analyses represent a mixture of various unidentified
seams and mines.
One area of special interest in this study was an exam-
ination of coal sulfur variability with respect to lot-size.
Considerable effort was directed toward the acquisition of
smaller lot-size samples, such as for individual railroad
cars (about 80 to 100 tons) or for the composite of several
railroad cars. In practice, however, it was found that few
companies perform routine sampling and analysis according to
ASTM procedures for such small lot-sizes. The smallest lot-
sizes for which routine ASTM sampling and analysis were
3-3
-------
found available from electric utilities and coal companies
were approximately 750 or 1,500 tons, which correspond to
the capacities of regular- and jumbo-size barges.
The majority of the data sets based on ASTM sampling
and analysis procedures reflect unit-train size shipments
(approximately 10,000 tons) from the Mid-Continent and
Western producing areas. These data sets preclude an analysis
of small lot-size or short-term coal sulfur variabilities
since a typical plant of 500 MW capacity will burn about 200
tons per hour or approximately one unit-train over a period
of two days.
Coal analysis data obtained from the Bureau of Mines
provided data necessary to address the problem of short-term
coal sulfur variability. These data were obtained from the
detail records of the Bureau of Mines "Current Coal History"
data tape, which contained analyses for the period 1966 to
1978. Coal analyses on this data tape reflect coals purchased
by Federal installations and the gross samples of these coals
represent volumes of 1,000 tons or less, as specified by the
Bureau of Mines sampling procedures. It should be noted
that these coals are not necessarily comparable to the coals
consumed by electric utilities. A significant portion of
the Bureau of Mines data are based on washed, double-screened
coals which are more representative of coals consumed by
industrial plants.
No attempt was made to obtain random samples of coals
from different producing regions, seams, or mines. However,
in the data collection phase of the project, special emphasis
was directed toward obtaining representative data sets from
each of the Bureau of Mines Producing Districts.
3-4
-------
3.2 Data Base
The data base developed in this study consists of
approximately 116,000 records/ each of which represents an
individual coal analysis or stack monitoring observation.
The data base is divided into two sub-categories the coal
analysis file and the stack monitoring file.
3.2.1 Coal Analysis Data File
The coal analysis portion of the data base consists of
approximately 94,700 records and represents coal samples
from more than 1,200 mines or combinations of mines. The
ranks of coals included in the data base are lignite, sub-
bituminous, and bituminous. The analyses represent produc-
tion samples and core analyses collected in recent years,
primarily between 1966 and 1978. The coals from which these
samples were obtained were primarily steam coals delivered
to electric utility steam generating plants and Federal
installations. A relatively minor portion of this data base
contains coal analyses from shipments to industrial plants.
Appendix C sets out the data base format for the coal
sulfur analysis data. Included in Appendix C are the field
descriptions of the data base and a brief description of the
relevant parameters used in classifying the data.
3.2.2 Stack Monitoring Data File
The second portion of the data base consists of the
stack monitoring data file. This portion consists of approxi-
mately 21,500 records for continuous monitoring data from
six electric generating units and one industrial unit. The
general format for the stack monitoring data is set out in
3-5
-------
Appendix D. It should be noted that the available data sets
did not permit the documentation of all the factors set out
in the stack monitoring format. For example, some of the
data sets had no record of the gross load (MWH) or the coal
flow (tons/hr) to relate to the emissions data.
3.2.3 Index to Mine Locations and Seams Produced
Set out in Appendix E is an index to mines and seams
included in the data base developed in this study.!/ Each
mine or source of coal has been assigned a five-digit mine
code number as set out in Column (1) of Appendix E. When
the analyses were reported as a composite of multiple mines,
the composite source was assigned a mine code and the
individual mines included in the composite were set out in
Column (2), "Blend of Mines." Column (2) generally identi-
fies the multiple sources of coal in a stockpile which
relate to stack monitoring data or coal analyses based on
"as burned" or "as fired" samples.
Columns (3) and (4) identify the respective state and
county locations of the various mines for which analyses
were obtained.
Columns (5) through (7) in Appendix E identify the
seams associated with each of the individual mines or
sources. The "Reference Code" in Column (5) is the seam
code number identified in the computer data base. These
six-digit codes are interpreted as follows:
I/ Appendix E is applicable only to data collected by Foster
Associates. Bureau of Mines data were not tabulated due to
the large number of mines (approximately 1,000), many of
which contain only a few coal analyses.
3-6
-------
First digit;
Digits 2-4:
Digits 5-6:
Digits 2-6:
A zero indicates that there exists an
equivalent Bureau of Mines' code for the
seam or combination of seams identified.
Codes beginning with the numeral "1"
indicate that a Bureau of Mines' code
for the seam(s) does not exist.
In the case of codes beginning with a
zero, the second through fourth digits
reflect the three-digit Bureau of Mines'
codes assigned to specific seams or com-
binations of seams.
In the case of codes beginning with a
zero, the fifth and sixth digits are
identifiers for various local names
assigned to the same coal seam such as
No. 8, Pittsburgh/ Big Vein, etc.
For codes beginning with the numeral
"1", the second through sixth digits
represent unique codes assigned to the
seams or combination of seams for which
no Bureau of Mines' codes exist.
Column (6) identifies the name of the seam or seams
represented by the code in Column (5). When no Bureau of
Mines' code existed for a combination of seams, the Bureau
of Mines' codes for each of the individual seams were pro-
vided in Column (7) as supplemental information.
3.2.4 Maps of Mine Locations by Producing District, State,
and County
The geographical location of each of the mines or sources
of coal is set out in Appendix F. The maps identify the
Bureau of Mines Producing District by State and County. For
each of the Producing Districts, the individual mines and
sources of coal are identified by county of origin. For
each county, the total number of mines as well as the codes
of individual mines are identified.
3-7
-------
The relevant seam information for each of the mines
located on the maps may be obtained by cross-reference to
Appendix E.
3.3 Analysis of Data
The second phase of this study analyzed the data received
from coal companies, electric utilities. Bureau of Mines and
other sources. The objectives of this study and the magni-
tude of the data bases dictated the use of computerized
statistical programs. The output of these programs was not
a final product but rather provided the necessary statisti-
cal data, in a summarized format, for further comparisons
and analyses.
The analytical program, which consists of various sub-
programs, is discussed in the next section. Although the
discussion relates to coal analysis data, minor modifica-
tions to the program permitted a similar analysis of stack
monitoring data.
Set out in Appendix G is an example of the computer
output of the programs developed for the analysis of coal
data. A separate analysis was performed for each of the
following variables in the data sets analyzed: volume,
sulfur content, heat content, and pounds of sulfur per MMBtu
(Ibs S/MMBtu).
The volume variable represents the size (lot-size) of
the shipment from which the coal samples and analyses were
obtained and is expressed in tons. Sulfur contents and heat
contents were analyzed on an "as received" basis, with the
exception of several data sets for which only "dry" basis
3-8
-------
analyses were available. The variable Ibs S/MMBtu was cal-
culated based on the sulfur and heat contents reported in
the coal analysis data. As previously indicated, no attempt
was made to estimate potential exhaust stack emissions
resulting from coal combustion due to the inherent diffi-
culties and variables associated with plant-specific coals
and equipment.
Each of the four variables (volume, sulfur content,
heat content, and Ibs S/MMBtu) was analyzed in a computer
routine which: (1) plotted the variable as a function of
time and provided the mean, standard deviation, and relative
standard deviation (RSD); (2) plotted frequency distributions
for the observed data; (3) compared the observed frequency
distributions to the expected frequency distributions based
on normal, inverted gamma, and lognormal distributions, and
(4) compared the goodness of fit between the observed and
the expected distributions. Each of these analyses is
explained below and is illustrated in Appendix G.i/
3.3.1 Plot of Variable vs. Time
A time plot of each of the variables was constructed to
visually examine possible variations during the chronological
sequence of the data. With respect to volumes, this routine
chronologically plots the volume of coal represented by each
analysis and permits a visual examination for consistency in
lot-sizes. For sulfur content, heat content, and Ibs
S/MMBtu, this plot permits a visual examination for possible
I/ The analyses performed in this study were based on a
simple model assuming independent variance. Budget con-
straints did not permit an investigation of the merits of
more sophisticated autocorrelative models.
3-9
-------
autocorrelation in the data or indications of some external
factor, such as a change in raining method or coal prepara-
tion, which has produced shifts in the data population.
3.3.2 Sample Statistics
The mean, standard deviation, and relative standard
deviation were calculated for each of the variables analyzed.
Included in the statistical program was a routine which
flagged any analysis which had a value which exceeded plus
or minue five standard deviations from the mean. Each of
these flagged records was manually compared to the raw data
to check for possible keypunch or transcription errors.
3.3.3 Frequency Distribution of Observed Data
In the frequency distribution routine of the analytical
program, nine equal intervals or cells were defined, based
on the sample statistics (mean and standard deviation) of
the data set. A histogram of each frequency distribution
was constructed and the number of observations contained
within each of the nine cells was recorded.
3.3.4 Comparison of Observed Distribution to Expected
Distribution
The next step in the analytical program was to compare
the observed frequency distributions to the expected frequency
distributions of the normal, inverted gamma, and lognormal
distributions. The selection of these three distributions
for comparative purposes, was based upon findings of previ-
ous studies on coal sulfur variability and discussions with
personnel from EPA and the electric utility and coal indus-
tries. Although there appeared to be no consensus of opinion
3-10
-------
on the distribution of sulfur in coal or the distribution of
Ibs S/MMBtu, these three distributions were the ones most
frequently encountered in discussions of coal sulfur varia-
bility and represent a reasonable range of alternative dis-
tributions.
Set out in Table 14 is a sample comparison for a data
set consisting of 1,537 analyses for a Wyoming subbituminous
coal. 'Column (1) sets out the cell number for the nine
TABLE 14
COMPARISON OF OBSERVED AND EXPECTED FREQUENCY
DISTRIBUTIONS OF LBS S/MMBTU FOR A
WYOMING SUBBITUMINOUS COAL (CORE ANALYSES)
Number of Observations
(N = 1,537)
0.00-0.20
0.20-0.28
0.28-0.36
0.36-0.44
0.44-0.53
0.53-0.61
0.61-0.69
0.69-0.77
0.77
0
0
57
502
498
387
77
12
4
0.4
9.2
93.1
371.5
588.6
371.5
93.1
9.2
0.4
0.0
0.3
62.8
450.5
598.1
306.1
92.7
21.2
4.2
0.0
20.0
54.6
417.3
620.2
310.0
95.3
16.7
2.9
Expected Distributions
Cell Cell Limits Inverted Log-
Number (Lbs S/MMBtu) Observed Normal Gamma normal
(1) (2) (3) (4) (5) (6)
1
2
3
4
5
6
7
8
9
cells generated in the histogram, while the limits of each
of these cells are identified in Column (2). Column (3)
summarizes the number of observations occurring in each of
the cells and Columns (4), (5), and (6) reflect the expected
number of observations within each cell based upon normal,
inverted gamma, and lognormal distributions, respectively.
Figure 6 graphically displays the relationship between the
3-11
-------
observed data and the expected values based on the normal,
inverted gamma, and lognormal distributions.
As shown on Figure 6, visually there appears to be no
significant differences among the three expected distribu-
tions. However, given the requirement of a very low proba-
bility of exceeding an emissions level combined with highly
variable coal sulfur contents, the type of distribution can
have a significant impact on compliance as discussed in pre-
vious sections of this report.
3.3.5 Goodness of Fit Between the Observed and Expected
Frequency Distributions
The final routine in the analytical program addressed
the problem of goodness of fit between the observed distri-
bution and the expected values based on the normal, inverted
gamma, and lognormal distributions. The chi-square test of
statistical significance was used to determine which, if
any. of the expected distributions approximated the distri-
bution of the observed data.
The chi-square test was used to analyze the goodness of
fit for cells one through nine of the histogram (six degrees
of freedom), cells two through eight (four degrees of free-
dom) , and cells three through seven (two degrees of freedom),
These last two tests were used to exclude the end cells of
the distributions, which frequently had less than the five
observations per cell necessary for a valid chi-square test.
Under the assumptions of a normal distribution, approxi-
mately 20,000 observations would be required to obtain an
expected value of 5 observations each in cells one and nine.
The area under the normal distribution contained in cells
one and nine is equal to approximately 0.05 percent of the
3-12
-------
FIGURE 6
RELATIONSHIP BETWEEN OBSERVED DATA AND EXPECTED VALUES
BASED ON NORMAL. INVERTED GAMMA, AND LOGNORMAL DISTRIBUTIONS
Wyoming
Subbituminous Coal
Core Analyses,
N ' 1537
0.44
Lbs S/MMBtU
0.69
0.77
3-13
-------
total area. The"area contained in cells two plus eight is
approximately 1.2 percent of the total area, indicating that
approximately 800 observations would be required for valid
chi-square tests for these cells. These figures provide an
indication of the data required for valid goodness of fit
tests in the extreme right-tail of the distributions, which
becomes increasingly important when a high probability of
not exceeding an upper limit is required.
An example of the results of the goodness of fit tests
is set out in Table 15. The data used in this example are
those observed values graphically displayed in Figure 6.
From Table 15 it can be seen that, at the 95 percent level
of significance and six degrees of freedom, the rejection
region for the hypothesis that the observed distribution is
i
statistically the same as the expected distribution is when
the calculated chi-square is greater than or equal to 12.6.
The calculated chi-squares for the normal, inverted gamma,
and lognormal comparisons are 113.4, 50.4, and 72.6, respec-
tively. Thus, in each case the hypothesis that the observed
distribution is the same as the expected distribution is
rejected. However, as noted on Table 15, the chi-square
test of these data at six degrees of freedom is of question-
able value due to less than 5 observations in cells one and
nine. In fact, the only conclusive chi-square tests in
these comparisons were those performed for cells three through
seven (two degrees of freedom) and for cells two through
eight (four degrees of freedom) for the lognormal distribu-
tion. An examination of these tests also indicates that, in
all cases, the hypothesis that the observed distribution is
the same as the expected distribution is rejected.
3-14
-------
TABLE 15
COMPARISON OF GOODNESS OF FIT OF OBSERVED DATA
WITH THE EXPECTED NORMAL, INVERTED GAMMA AND
LOGNORMAL DISTRIBUTIONS
Type of Distribution
Normal - Observed
Normal - Expected
Inverted Gamma-Observed
Inverted Gamma-Expected
Lognormal - Observed
Lognorraal - Expected
Number of Observations in Cell Number:
1
0
0.4
0
0
0
0.4
2
0
9.2
0
0.3
9
9.2
3
57
93.1
57
62.8
70
93.1
4
502
371.5
502
450.5
448
371.5
5
498
588
498
598
514
588
.6
.1
.6
6
387
371
387
306
419
371
.5
.1
.5
7
77
93
77
92
67
93
.1
.7
.1
8
12
9.2
12
21.2
6
9.2
9
4
0.4
4
4.2
4
0.4
6 Degrees
of Freedom
113. 4£/
50. 4£/
72. 6l/
4 Degrees
of Freedom
84. 9l/
50. 3£/
44.1
2 Degrees
of Freedom
76.1
46.4
43.3
Chi-Square Test
Calculated Chi-Square*
Hypothetical
Distribution
Normal
Inverted Gamma
Lognormal
* Rejection Regions, 0.95 level of significance:
6 Degrees of Freedom, Chi-Square > 12.6
4 Degrees of Freedom, Chi-Square > 9.5
2 Degrees of Freedom, Chi-Square > 6.0
a/ Chi-Square test is of questionable value with less
than five observations in certain cells.
Source: Foster Associates, Inc.
Although all the calculated chi-square values in Table 15
indicate that the observed distributions are statistically
different from the three distributions used for comparison,
it is still possible to make a general statement about the
3-15
-------
goodness of fit.i/ Based on the chi-square values calculated
at two degrees of freedom, the distribution of the observed
data can be ranked according to goodness of fit as:
TABLE 16
RANKING OF GOODNESS OF FIT BY CHI-SQUARE TEST
Rank Distribution Chi-Squarel/
1. lognormal (43.3)
2. inverted Gamma (46.4)
3. normal (76.1)
I/ The lower the chi-square, the better
the fit.
At two degrees of freedom,, the two end cells for each
tail of the distribution were not considered, thus the dis-
tributions were compared or ranked only for the central por-
tion of the distributions accounting for approximately 99.15
percent of the total area under the curve in the case of a
normal distribution.
I/ This general statement about the goodness of fit is
based on the assumption that for each cell the square of the
absolute difference between the observed and expected fre-
quencies divided by the expected frequency is a valid measure
of goodness of fit. In later sections of this report the
goodness of fit analysis is used to examine the fit for
extreme values, or the tails of the frequency distributions.
3-16
-------
4.0 Analysis of Coal Data Received from Respondents
This section relates to the analysis of the coal data
provided by respondents and specifically excludes data
obtained from the Bureau of Mines. The analysis of Bureau
of Mines coal data is discussed in Section 5.0 of this
report.
In order to better understand the structure of this
section of the report, a brief discussion of the topics con-
tained herein and an overview of the methodology are helpful.
Three general topics are included in this section: (1)
sample statistics, (2) predictability of mine variability
based on composite coal seam and Producing District data,
and (3) choice of statistical distribution.
The section on sample statistics discusses the varia-
bilities of Ibs S/MMBtu exhibited by the coal data sets at
various levels of aggregation. The methodology consisted of
three general levels of analysis. First, the individual
data sets received from the respondents were analyzed.
These data sets generally reflect coal analyses from
individual mines, and in many cases, mines producing from a
single coal seam. With respect to sample statistics (mean,
standard deviation, and RSD), this analysis summarized the
entire data set with a single RSD of Ibs S/MMBtu and encom-
passed a wide range of lot-sizes. Second, the individual
data sets were examined according to lot-sizes. This
involved the generation of sub-data sets, sorted by specific
lot-size intervals. For example, a data set may have con-
tained coal analyses reflecting shipments ranging from 1,000
to 10,000 tons. Sorting by lot-size provided new data sets
representing smaller intervals such as 1,000 to 2,000 tons,
2,000 to 3,000 tons, etc. This permitted an assessment of
4-1
-------
the influence of lot-size on the variability (RSD of Ibs
S/MMBtu) within individual mines. Finally, a series of
special aggregate analyses was performed. This included
analyses of data by Producing District, by coal seam, by
degree of preparation, and other factors which may con-
tribute to variability.
The second topic, which examines the predictability of
mine variability from coal seam and Producing District data,
utilizes the results from the statistical analyses of indi-
vidual mines, aggregate seams and Producing Districts. These
results are analyzed and compared to determine to what extent
coal sulfur variabilities can be predicted or generalized.
The third topic examines the choice of statistical dis-
tributions based on the results of the chi-square routine of
the analytical program, comparisons of relative distribution
error and comparative plots of the observed and expected
data for the right-tail of the frequency distributions.
This analysis examines various coals a'hd discusses which
frequency distributions best fit the observed data.
4.1 Sample Statistics
In general, each of the data sets gathered by Foster
Associates, PEDCo, and EPA were analyzed through the use of
the analytical program. Appendix H sets out the salient
characteristics of these data sets which, as previously
indicated, reflect coal analyses from individual mines and
in many cases, mines producing a single coal seam. In cer-
tain cases, due to an inadequate number of observations in
the data set, the individual data sets were not separately
analyzed but used only for aggregate analyses for specific
seams or producing districts.
4-2
-------
A series of analyses was also performed on the data at
various levels of aggregation to examine the impact of the
variables which have been identified as possible factors
contributing to coal sulfur variability.
4.1.1 Individual Data Sets
The first step in the analysis of the data gathered in
this study focused on the data as supplied by the respondents.
The results of this analysis by USBM Producing District are
set out in Appendix I. Cross-reference of the mine codes in
Appendix I to the information in Appendix H provides a detailed
description of the documented parameters (rank, mining method,
preparation, etc.).
In this analysis 205 separate data sets, representing
18 Producing Districts, were analyzed. Since these data
sets reflect sequential deliveries from the same mine or
source of supply, many of the factors which may influence
coal sulfur variability have been isolated and may be assumed
constant. These factors include geographical location, seam,
rank and type of sample. Other factors, such as lot-size,
mining method, degree of preparation, and method of sampling
may vary within the individual data sets but in many cases
may exhibit sufficient consistency to be assumed constant.
The coals in these data sets are almost entirely steam
coals, produced for the electric utility market. In general,
these coals are crushed run-of-mine coals but in some cases,
such as in the Mid-Continent area (Illinois, Indiana, and
western Kentucky), the coals are predominantly washed.
Appendix I illustrates the variations in coal character-
istics actually experienced by electric utilities. Most of
4-3
-------
these data sets represent coal deliveries under term con-
tracts. Columns (4) and (5), for example, illustrate the
mean and the variation in the size of individual coal ship-
ments delivered. In general, the RSD of the shipment size
ranges from 20 to 60 percent. Many of the factors contri-
buting to these variations, such as strikes, transportation
availability, and bad weather, are beyond the control of the
consumer. One exception to the large variations in lot-size
occurs in Producing District 19 where uniform shipments of
approximately 10,000 tons (unit train) are predominant.
Here the RSD's of the lot-sizes are generally less than 10
percent.
A summary of the ranges of Ibs S/MMBtu based on the
means and RSD's is set out in Table 17- In this summary
only 140 data sets, each containing 30 or more observations,
were considered.!./ The average Ibs S/MMBtu ranged from 0.19
for a source in Producing District 16 (Colorado) to 4.99 for
a source in Producing District 4 (Ohio). The RSD's of the
Ibs S/MMBtu ranged from 2.1 percent for a source in Produc-
ing District 10 (Illinois) to 67.2 percent for a source in
Producing District 4 (Ohio) .
4.1.1.1 RSD Versus Lot-Size
The results of this first analysis of data were examined
for indications of any relationship between the RSD of the
Ibs S/MMBtu and the average lot-size of the data sets.2/
I/ RSD and other statistics computed with less than 30
observations are subject to a large sampling error and may
not be representative of the true population.
"if In this analysis, the average lot-size of each data set
was plotted against the RSD of the Ibs S/MMBtu. In later
analyses, data sets exhibiting large variations in lot-sizes
were aggregated by specific lot-size intervals within the
data set.
4-4
-------
:The expected inverse relationship between RSD and lot-size^)
was not exhibited in this analysis J^ An example of the results
of this analysis is set out in Figure 7, which provides a
comparison of the RSD's and average lot-sizes for data sets
from the Mid-Continent producing area (western Kentucky,
Illinois, and Indiana).
TABLE 17
SUMMARY OF THE AVERAGE LBS S/MMBTU
AND RSD FOR INDIVIDUAL DATA SETS BY PRODUCING DISTRICT
USBM Number of Range of Lbs S/MMBtu
District Data Setsl/
(1) (2)
1 7
2 0
3 2
4 35
6 0
7 0
8 21
9 13
10 24
11 4
12 2
13 0
14 0
15 1
16 1
17 4
18 3
19 17
20 1
21 3
22 2
23 0
Total 140 0.19-4.99 2.1-67.2
I/ Containing 30 or more observations.
Source: Appendix I.
Average
(3)
1.45-2.18
1.90-2.14
2.09-4.99
0.49-2.31
1.60-4.31
1.02-3.52
1.57-3.71
3.18-3.71
2.63
0.19
0.35-0.57
0.38-0.73
0.38-1.08
0.90
0.82-1.31
0.42-0.85
RSD (%)
(4)
8.8-32.6
8.3-11.6
6.6-67.2
4.9-62.3
4.9-47.2
2.1-51.0
8.3-34.3
18.1-21.5
22.7
20.8
6.5-20.7
11.7-45.8
6.9-31.8
27.6
25.2-33.3
17.8-30.1
4-5
-------
FIGURE 7
RSD OF LBS. S/MMBtu VS.
AVERAGE LOT-SIZE OF INDIVIDUAL DATA SETS FOR
PRODUCING DISTRICTS 9 (Western Kentucky), 10 (Illinios), and 11 (Indiana)
50
45
40
35
m
cr
z>
u.
_i
13
co
to
00
u.
O
O
CO
cc
30
20
15
10
6 8 10 12
AVERAGE LOT-SIZE (thousand tons)
14
16
4-6
-------
The 41 data sets included in Figure 7 each have 30 or
more observations and generally represent washed coals of
seams No. 5 and No. 6, which are common to all three states
in this producing area. From Figure 7 no relationship
between RSD and lot-size is observed. The majority of the
RSD's fall in the range of 5 to 18 percent, but appear to be
independent of the lot-size.
Additional analyses of these data were performed to
examine the relationship of RSD of Ibs S/MMBtu versus
average lot-size for:
individual Producing Districts
. individual coal seams
raw and washed coals
The results of these analyses were also inconclusive in
demonstrating any relationship between RSD and lot-size.
4.1.1.2 Multiple Data Sets for Individual Mines
In some cases the data base contained two or more data
sets for the same mine. These data sets were examined and
are summarized in Table 18. The circumstances resulting in
multiple data sets include: (1) deliveries to the same con-
sumer over different time periods, (2) deliveries to differ-
ent consumers, and (3) analyses of the same coal reported by
two different laboratories.
The one case, mine 10045, reflecting the results of
a "split-sample" analyzed by two different laboratories,
exhibits average Ibs S/MMBtu at 3.13 and 3.15 with
RSD's of 6.2 and 5.9 percent, respectively. The differ-
ence between the average HDS S/MMBtu is less than one
4-7
-------
percent, while the difference in the RSD's is approximately
five percent.
TABLE 18
COMPARISON OF LBS S/MMBTU FOR
MINES REPRESENTED BY MULTIPLE DATA SETS
Mine
Code
(D .
1003L§./
10019b/
10036^7
10042b/
10045£/
10097k/
10062^/
10 07 7 y
Lbs
S/MMBtu
Average RSD (%)
(2)
1.78
1.80
3.39
2.71
3.48
2.96
3.49
3.34
3.13
3.15
2.78
3.00
3.18
3.71
0.88
1.08
1.02
(3)
14.5
8.8
4.9
13.4
10.1
11.8
2.1
11.9
6.2
5.9
11.0
6.7
18.1
21.5
17.0
17.2
24.6
Number of
Observa-
tions
(4)
268
220
114
31
107
61
147
77
214
213
339
269
55
53
169
74
51
Mine Lbs
S/MMBtu
Code Average RSD (%)
(1) (2)
10098Jb/ 0.43
0.73
10006.b/ 0.48
0.46
0.43
0.44
0.43
0.42
10038&/ 0.57
0.79
10039l>/ 0.65
0.98
looeib/ 0.37
0.38
(3)
11.7
45.8
16.9
14.2
11.6
14.4
11.9
12.4
16.0
46.1
31.8
42.6
14.2
12.0
Number of
Observa-
tions
(4)
41
55
1,537*
33
49
213
64
1,780
140
29
30
28
47
52
a/ Same consumer, different time periods.
b/ Different consumers.
c/ Different laboratories.
* Core Analyses.
Source: Appendix I.
4-8
-------
The data sets for mine 10031 reflect deliveries during
two different time periods to the same consumer. The first
data set, covering the period January to December 1975,
exhibits an average Ibs S/MMBtu of 1.78 with an RSD of 14.5
percent. The second data set, for the period November 1977
to September 1978, reveals an average of 1.80 Ibs S/MMBtu
and an RSD of 8.8 percent. The more recent data set indi-
cates an increase of 1.1 percent in the average and a
decrease of 39 percent in the RSD.
The multiple data sets for the 11 remaining mines are
based on coal deliveries from the same mine to different
consumers. The time periods of the multiple data sets for
each mine are not exactly comparable, although most data
sets are based on coals analyzed between 1975 and 1978. The
differences between the low and high value of the average
Ibs S/MMBtu for each mine ranged from 2.7 to 69.8 percent,
while a similar comparison for the RSD's showed differences
ranging from 15.5 to over 466 percent.
A comparison of the RSD's to average lot-sizes for the
individual mines containing multiple data sets again failed
to demonstrate any consistent relationship.
4.1.1.3 Mine 10006
The data in Table 18 for mine 10006 permit additional
comparisons. The data set containing 1,537 analyses repre-
sents core analyses of the coal reserves. The four data
sets, containing from 33 to 213 observations, are based on
deliveries to four separate utilities, while the data set
containing 1,780 observations is based on all shipments from
the mine during the period November 1977 to October 1978.
All data sets, except for the core analyses, are based on
4-9
-------
unit-train size shipments of approximately 10,000 tons,
sampled by an automatic ASTM belt sampler. For convenience,
a summary of these data sets is set out in Table 19.
TABLE 19
COMPARISON OF DATA SETS FOR MINE 10006
Lbs S/MMBtu Number of
Type of Data Date Range Average RSD (%) Observations
(1). (2) (3) (4) (5)
Core N/A 0.48 16.9 1,537
As shipped 6/76-12/77 0.44 14.4 213
As shipped 12/76- 3/77 0.46 14.2 33
As shipped 1/78-10/78 0.43 11.9 64
As shipped 1/78- 8/78 0.43 11.6 49
As shipped 11/77-10/78 0.42 12.4 1,780
As shipped!/ 6/76-10/78 0.43 13.6 2,199
I/ Includes one data set of 60 observations which was not
analyzed separately.
A comparison of the core analysis data (1,537 observa-
tions) with all the available data for coal shipments (2,199
observations) shows that the average Ibs S/MMBtu as well as
the RSD is lower when calculated based on coal shipment
data. The average Ibs S/MMBtu for the coal shipments is
10.4 percent less than the average indicated by the core
analyses, while the RSD is 19.5 percent less.!/
A comparison of the data by date ranges indicates a
decline in the average Ibs S/MMBtu and RSD. Coals analyzed
during the period June 1976 to December 1977 exhibit an
average of 0.44 Ibs S/MMBtu and an RSD of 14.4 percent.
I/ Similar results have been reported for other Western
coals. For example, a study conducted at the Navajo Plant
by the Salt River Project indicated the mean sulfur content
of coal shipments was 10 to 20 percent less than the mean
indicated by core analyses.
4-10
-------
while comparable data for the period January 1978 to October
1978 show an average of 0.43 Ibs S/MMBtu with a RSD of 11.9
percent. Thus, during the period from June 1976 to October
1978, it appears that a slight decline in the Ibs S/MMBtu
and a reduction in the relative variability of the coal has
occurred.
4.1.2 Lot-Size Interval Analysis of Data
The next step in the analysis of data examined the
relationship between lot-size and the RSD of Ibs S/MMBtu
within individual data sets. As noted in the previous dis-
cussion, the expected inverse relationship was not observed.
It was hypothesized that variations in the size of individual
shipments, which in some cases was substantial as indicated
in the RSD of the volumes in Appendix I, were masking the
relationship between the RSD and lot-size. To test this
hypothesis, the individual data sets were analyzed with
respect to lot-size intervals within the individual data
sets. For example, a data set may have contained analyses
based on shipments ranging from 1,000 to 11,000 tons, while
the majority of the shipments were in the ranges of 1,000 to
2,000 tons, and 9,000 to 11,000 tons. The tonnage intervals
examined within each data set were based upon a visual exam-
ination of the frequency distributions and plots. Gener-
ally, intervals were selected to include a minimum of 30
observations. These selected lot-size intervals were then
analyzed with the computer analytical program to examine the
relationship of RSD to lot-size.
Before the results of this analysis are discussed, it
is useful to examine the relationship between RSD and lot-
size suggested by a theoretical model.
4-11
-------
4.1.2.1 Theoretical Relationship Between RSD of Lbs
S/MMBtu and Lot-Size
Among the various researchers who have examined the
problem of coal sulfur variability.- there has been consider-
able discussion on the theoretical versus the observed rela-
tionship between the RSD of Ibs S/MMBtu and the lot-size for
any particular type of coal. The EPA study "Preliminary
Evaluation of Sulfur Variability in Low Sulfur Coals from
Selected Mines", while based on limited data, suggests that
the RSD increases at a relatively constant rate as the lot-
size is successively decreased. Plotted in semilogarithmic
form this relationship would approximate a straight line as
shown in Figure 8. In contrast, some of the data sets
analyzed in Appendix C of the above study displayed a
horizontal slope or an upward slope with increasing lot-
sizes.
This discussion summarizes the results of a Monte Carlo
simulation study which examined the theoretical relationship
between lot-size and relative standard deviation. The study
concludes that the relationship is approximately linear in
semilogarithmic form and the curve has the general shape as
shown in Figure 8.
In order to develop this Monte Carlo simulation it was
first necessary to postulate a model of the physical process
which would determine the relationship between RSD and lot-
size. This model assumes that for a small amount of coal,
say one ton, there exists a frequency distribution which
describes the Ibs S/MMBtu as if each ton were used as
the basic sampling unit. Under this assumption each time
the lot-size is increased, for example to five tons, the
value of Ibs S/MMBtu is a simple average of five, one-
4-12
-------
I
M
CO
FIGURE 8
0.24
48
480
0.20 -
0.16
0.12
0.08
0.04
TONS OF COAL
4800
1
48.000
480.000
1
4.8 x 10s
33 TONS
(3 HOURS FOR
THE 25 HW PLANT)
600 TONS (3 HOURS FOR SOO HW PLANT)
REGION OF EXTRAPOLATION
-4800 TONS (1 DAY FOR 500 HW PLANT)
-8400 (1 UNIT TRAIN FOR 500 HU PLANT)
AVERAGING PERIOD
(500 HW)
_L
3 HOURS
1 DAY
30
J L
J_
90 180 360
AVERAGING PERIOD/TONS OF COAL (DAYS/HOURS/TONS}
RSD versus averaging period/tons of coal (days/hours/tons).
Source: PEDCo Environmental, "Preliminary Evaluation of Sulfur Variability in Low-Sulfur
Coals from Selected Mines", EPA Publication No. EPA-450/3-77-044, July 1977, p.5-8
-------
ton lot samples drawn at random from this frequency dis-
tribution. Correspondingly, a lot-size of ten tons would be
the average of ten individual tons drawn at random from the
frequency distribution, a lot-size of twenty tons would be
the average of twenty one-ton samples drawn from this fre-
quency distribution, etc.
Given any distribution, it can be argued on intuitive
grounds that the standard deviation should decrease as the
square root of the number of sample points in one ensemble
increases. This is given by the formula relating the sample
standard deviation to the population standard deviation as
the sample size increases as shown in Figure 9.
X/N
where °p = population SD
QS = sample SD
N = sample size
FIGURE 9
RELATIONSHIP BETWEEN SAMPLE STANDARD
DEVIATION AND SAMPLE SIZE
N
4-14
-------
By definition, the standard deviation of an ensemble is
£(Xj-X)2 - L
- where X = -
Increasing N, i.e., the lot-size, results in
N
YN - £ Xj
N
M
and
M
The overall variance is given by
M
_
(YN-YN) 2
M
If M is large enough, then
The objective is to find the individual value of °s/X, i.e.,
RSD, of one lot-size, as N increases. Because this problem
is very difficult to solve analytically, a simulation
approach was used.
The simulation procedure was based on an assumed log-
normal distribution, using actual coal analysis data. The
data used in the model are based on mine 02020 in Producing
District 16 (Colorado). The data for this mine exhibited an
average lot-size of 237 tons and a mean of 0.2558 Ibs
S/MMBtu with, a standard deviation of 0.048. The simu-
lation model assumed a lognormal distribution, generated
» ' *
random numbers, and calculated the new means and standard
4-15
-------
deviations as the lot-size was increased one-fold, two-fold,
and so on. The RSD's were then plotted against the result-
ing lot-sizes.
The results of the simulation study are shown in Figure
10. The horizontal axis on Figure 10 shows the successive
lot-sizes, while the vertical axis indicates the relative
standard deviation. The figure shows a curve which suggests
that the relationship between RSD of Ibs S/MMBtu and lot-
size is approximately linear in semilogarithmic form, and
the relative standard deviation increases greatly as lot-
size decreases.
It should be pointed out that this discussion is based
on a theoretical model. Although the assumptions used in
this model appear to be reasonable, sufficient research has
not been conducted to determine whether coal sulfur varia-
bility follows the simple physical process model developed
in this discussion.
4.1.2.2 Observed Relationship Between RSD of Lbs
S/MMBtu and Lot-Size
The purpose of the analysis presented in this section
is to determine whether or not the expected theoretical
relationship between RSD of Ibs S/MMBtu and lot-size is
confirmed by coal analysis data.
To begin the analysis, one must first recognize that
RSD is a statistical estimate and is subject to sampling
error. In order to allow for this variation, the first step
in the analysis is to calculate a relative error, the
standard deviation, of the RSD estimates. Given this num-
ber, one can then discuss not only the absolute value of the
4-16
-------
RESULTS OF SIMULATION MODEL FOR
RELATIONSHIP OF RSD OF LBS S/MMBTU
AND LOT-SIZE
r M II 1 1 1 1 IIHU-IIII l-l M 1 1--IJ 1 1 Illlllllll I-IIIIHIII IIIIHIIIimilHIHIIIIHII'lHI
; -,-.-. .. I | |
. . . .... .. 1
1
~ '
.... . . . . .,-,-., J
! |
il_ 1 1 i J
I I p
1' . :::~r; .. . .". " .". ', III ]
* ' ' i jl
- - -- i - i j
' t
1
_^_ r ;x ''-' : ;": ,ll Li
;-TT j. -;-:--:;-- : || I
- i _i_ - - . -. - . . 1 1 i 1 1 in iij! ii
.... . . _ ^ . . ! |
~H~T '. '. 1 f T f' ; '| ] p i T
I I Mil
II II 1 1 i| i i|
I'l ill 1 1 : 1 ill '
^r q- f f T rnn jlli!
Ilk ilL 1 i nl liiUr
V]pt"~"":7ti ^rni'iiif"
, ! i H i i ' M|! i i il
! ! I l!i li!
I , X J J- - r ' -1- rt|iJJ«J +
- -1- r-'- :f-T -J --H - -- - , H if llnO t
> I 1 ill 1
1 I Ill
i |
1 ill
1 1 1 J_ 'I ; 1 1 1 1 HI
li :::::::::::: :::: : T T inr ^\T~ .%*
\\ 1 1 M ''> '
: 1 i " " J 1 i . ' :
] 1 1 " 1 I 1 '' ' :
illr :::::-::-:-::::.. i":::-::.:::::.. il.ma^.^ vi^ .- ^
1 j Based on Mine 02020, Producing
1 Di *?1"r i r»-l- 1 fi
f ' ' [ . i
' !i| | " " Initial Conditions:
lj|| n Lot-Size = 237 Tons
|l ii Mean (Lbs g/MMBtu) = 0.4862
11 Standard Deviation (Lbs S/MMBtu)
! j! ------- --_ = 0.0912
iii i T ! TJ I jT:i M : ,;
1 " . ... ._ | | ^
i ii ! i ij'i
!l _"" . .._: :: :~:..
H !i 1 i j Li 1 hi1"' ;ll|i
I'M " ' M TijTfi i ;, . r^
i i ' '"'
! !' N ~.'-_~~i'-' "I" '.'./. '""." I~~^. - ; '' ' ! :! ili1'!' ;:
M:":=:--:^":"-"-=::">-^:|| .j '- ' if ^§1^
1 1 i Jl I. 1 i 1 1 1 | ' ! " '' !i '
v Y\ . ^r j^ T" " I t'T ' T iti! [jrr|ii;;,r ;;;
iiii; ' i " ~ ' " ~ r i i |i j_i i H ! ! 'r !' ' '! ', '
"I 1 ' II i ! " i ! ! i:!':
:!! J! :. . " . _".".. 1 l| i jiiii i IM'
' 'lij li " i ' ' ' I 1 llll i 1 i ,' ' 1
:j, ] !i - ...... .S . . ' """ i ! i1 ' ! "
J! ----- | - - - j j ' ' ' i ' I 1 'j '' 1
i!l|j"-:: 1 " rl " l-'ii " L I i I, ! :i;i! '= '" ! :-':
ji|!|-_T"":----t-r :-t -:t:"fTj-jTi|]-h4Tf- ;;rii:i'! -^ - ~
V+ -i- " " ~ :;" " ; .iL ^ ! ' : 1 : '
ili-': '^ : '-' '"- ^ T 'T^^ I'? li:'""
1 llll1 " " " ' ------ .I ]j j| | i||| ',! :;|i'. ,.
i,| - -] -HT- :-- T . -T -+ -VriT ^ril IT f ::;-"
!!| f - " i j !!| | ^i!C";
III !-:_:- i 1 ±L L iH LjllLlJiiL.llMi-!1-:'!1
20
sK>
15
4-1
PQ
(0
I
I
^J
O
Q
Oi
10-
100
1000
T .^_c i -70
10,000
-------
relative standard deviation, but also limits in terms of
plus or minus two, or plus or minus three, standard errors.
In this manner the analysis can explicitly recognize the
uncertainty in the calculations of RSD's.
Figure 11 illustrates this method of analysis for a
particular mine. The horizontal axis of the figure shows
the natural logarithm of the lot-size in tons. The vertical
axis shows the RSD of Ibs S/MMBtu. The actual RSD's
are shown by the dots, while the plus and minus two standard
errors are represented by the x's and the plus and minus
three standard errors are represented by the circles. Each
of the four sets of plots on Figure 11 represents a distinct
lot-size interval, the means of which are designated by the
dots. Note in this example that the RSD is an increasing
function of lot-size for all data-points except one, and
shows a substantial decrease in RSD with a decreasing lot-
size.
A total of 82 mines were analyzed as described in the
previous paragraph. Of these, 53 had three or more RSD
measurements and 29 had two RSD measurements. Again, each
RSD measurement represents a distinct lot-size interval
within the individual mine data.
Of the mines with three or more RSD measurements, eight
showed a log-linear relationship to within plus or minus
three standard errors on all data points, and ten showed a
log-linear relationship within plus or minus two standard
errors. Of these eighteen mines, RSD was a decreasing
function of increasing lot-size (negative slope) in nine
cases and an increasing function of increasing lot-size
(positive slope) in seven cases. In two cases, the RSD
remained constant with increasing lot-sizes.
4-18
-------
FIGURE 11
RSD Lbs S/MMBtu Versus Lot-Size
for Mine 10142, Producing District
25
I
S
- 20
D
4J
OQ
15
Q
CO
10
\
\\
I
I
to i mm
4-1
i
fi
II
1 S-
"T
rors
rors
4 6 8 10 20
LOT-SIZE (Thousand Tons)
40
-------
For 35 out of the 53 mines with three or more data
points, no consistent log-linear relationship could be shown
within plus or minus three errors. However, a nonlinear
relationship between RSD and lot-size could be estimated for
20 of these 35 mines. Of these 20 mines, 14 showed RSD to
decrease as a function of increasing lot-size, while six
cases showed RSD to be an increasing function of lot-size.
The remaining 15 mines exhibited a U-shaped relationship
between RSD and lot-size.
For 29 of the 83 mines analyzed, only two RSD measure-
ments or lot-size intervals were available. Naturally, a
log-linear line wil] pass through the two points in every
case. Twelve of the mines showed RSD decreasing with
increasing lot-size, 16 mines showing RSD increasing with
increasing lot-size, and 1 mine indicated a relatively
constant relationship.
In conclusion, there is no apparent consistent rela-
tionship between the RSD of Ibs S/MMBtu and lot-size
for any individual mine. The coal analysis data examined in
this exercise fail to confirm the theoretical relationship
which indicated that the RSD of Ibs S/MMBtu decreases
with increasing lot-sizes. However, other variables, such
as geological factors and mining methods, may have a rela-
tively greater impact on coal sulfur variability than lot-
size and may have distorted the expected relationship
between RSD and lot-size. Further, it should be noted that
this analysis is based primarily upon samples representing
from 1,000 to 10,000 tons with relatively few points repre-
senting volumes of less than 1,000 tons. However, because
of the wide dispersion of RSD measurements examined in this
analysis, it is unlikely that any simple relationship exists
between RSD and lot-size which could be used to accurately
4-20
-------
predict the RSD for a corresponding lot-size within an
individual mine.
These findings are particularly troublesome for coal
producers and/or coal consumers who must guarantee that
coals will comply with sulfur emission standards. Coal
sulfur variability is often critical in the cases of small
coal-fired boilers and regulations which specify short averaging
periods, both of which would correspond to small lot-sizes.
The results of the previous analysis indicate that the short-
term sulfur variability of coal from an individual mine
cannot be accurately predicted, even if substantial histori-
cal data are available. These findings are even more
onerous with respect to the development of new mines which
must rely on core analysis data. Individual core analyses
may represent in excess of 500,000 tons of coal reserves,
while potential customers frequently need to know the sulfur
variability on increments of 10,000 tons or less. This
suggests that the language and requirements of many sulfur
dioxide emission-limiting regulations are not consistent
with the state of knowledge concerning coal sulfur varia-
bility.
4.1.3 Analysis of Data on an Aggregate Basis
The objective of this analysis was to examine the prob-
lem of coal sulfur variability within individual Producing
Districts. This was accomplished by various aggregations of
the data available for each Producing District. In general,
and subject to the availability of data, the following
analyses were performed on a Producing District basis:
4-21
-------
All coals; with and without lot-size intervals.
Raw coals; with and without lot-size intervals.
Washed coals; with and without lot-size intervals.
Selected coal seams; with and without lot-size
intervals.
In addition, data for some Producing Districts per-
mitted more detailed analysis of other factors which have
been identified as possible sources of variability in coal
analysis data.
The analyses performed and the results obtained from
the aggregate analysis are discussed in detail in Appendix J.
Although no indisputable conclusions resulted from these
aggregate analyses, it was possible to make some general
statements about the relationships exhibited in the various
Producing Districts.
The results pertaining to the relationship between RSD
of Ibs sulfur/MMBtu were inconclusive. However, it appears
that as the level of aggregation is increased from mine, to
seam, to Producing District, the data tend to exhibit a
sharp increase in RSD at small lot-sizes, especially below
2,000 tons. This suggests that variables that could not be
controlled or were not analyzed may have masked the relation-
ship between RSD and lot-size. In the case of the composite
data, it is possible that the effects of these variables
tended to cancel each other.
Comparisons of washed and raw coals within the individ-
ual Producing Districts consistently indicated a lower Ibs
S/MMBtu and a lower RSD for the washed coals. These find-
ings tend to support the hypothesis that coal washing would,
in general, reduce the level of sulfur emissions as well as
the relative variability of the emissions.
4-22
-------
In the limited number of cases analyzed, it appeais
that significant differences in RSD's can exist among seams
within the same Producing District. As reported in previous
sections of this report, substantial, inconsistent differ-
ences in RSD's were also observed among individual mines and
among lot-sizes within mines. These observations raise
serious doubts about the extent of the relationship between
RSD and lot-size and the existence of a simple relationship
which can accurately generalize coal sulfur variabilities.
Finally.- some of the data on an aggregate basis
exhibited large RSD's, frequently in excess of 40 percent.
RSD's of this magnitude could have a substantial impact on
compliance. This suggests that coal consumers subject to a
given emission limit with only marginally acceptable coals,
must selectively evaluate the various sources of supply and
may consequently find it necessary to exclude those sources
which exhibit large variabilities.
4.2 Predictive Ability of Producing District and Seam Data
for Individual Mines
This section shows the results of a study performed to
determine whether composite seam or Producing District data
can be used to predict the relative variability of Ibs
S/MMBtu for an individual mine. If combined data can
predict mine data, the derived relationships between RSD and
lot-size for composite seams or Producing Districts would
permit the estimation of RSD's for an individual mine over a
range of lot-sizes.
Generally, the individual mine data sets permitted the
calculation of the relative standard deviation for only
several-lot-sizes. However, by combining the data within
4-23
-------
Producing Districts or seams, it was possible to calculate
RSD's for a wide range of lot-sizes.
4.2.1 Methodology
Data from all mines within a seam or district were com-
bined and analyzed by lot-size in terms of the relative
variability of Ibs S/MMBtu. A line of regression was
then fitted between the log of the lot-size and RSD and used
as the predictor of the RSD of Ibs S/MMBtu within indi-
vidual mines. -Standard errors of the RSD estimates for
individual mines were then calculated. The individual mine
RSD estimates, as well as the composite seam data, were then
plotted on the same graph. Two tests were used to determine
the degree to which the seam data could predict mine RSD.
First, mine RSD estimates were checked against the RSD as
predicted from the composite data to determine if the
regression line fell within plus or minus three standard
errors of the individual mine RSD's. Second, the absolute
average error was analyzed by assuming the composite data
were predictors of mine RSD's.
4.2.2 Analysis of Predictive Capabilities of Seam Data
Figure 12 shows the results of this analysis for the
composite seam data for the Pittsburgh seam (036) , Producing
District 4 (Ohio). The regression line intersected only two
of the eleven mine RSD's within plus or minus three standard
errors. Of the remaining nine mine RSD estimates, the error
ranged from a high of 8.75 percentage points to a low of 3
percentage points with an average of 5.6. This is a sub-
stantial error since the RSD's for individual mines
generally ranged from 15 to 25 percent. Furthermore, note
4-24
-------
Analysis of Composite Seam Data
For the Pittsburgh Seam (036)
in Producing District 4
© RSD Composite Data
RSD Mine Data
X + Two Standard Errors
0 4- Three Standard Errors
1000
10,000
TDTJTOW
-------
that the composite data were biased estimators of mine RSD's,
All except one of the mine RSD's lie below the composite
regression line.
Similar analyses were performed for the Middle
Kittanning (080) and Lower Kittanning (084) seams in
Producing District 4. For the Middle Kittanning seam, there
were 27 individual mine RSD estimates, of which only 6 fell
within the plus or minus three standard errors. The average
error in this case ranged from 8.7 percentage points to a
low of 4 percentage points, with an absolute average error
of 8.7 percentage points. For the Lower Kittanning seam, 6
of the 14 individual mine RSD estimates were within plus or
minus three standard errors of the composite RSD. The
absolute average error for this seam ranged between a high
of 10 percentage points to a low of 1.75 percentage points,
with an average of 3.8. In both instances, the composite
data provided biased estimates which consistently over-
estimated mine RSD's.
In addition to the three seams discussed above, several
other seams were plotted and the composite mine data were
compared to the individual mine estimates. These cases
exhibited the same general results as previously discussed,
that is, the composite data were not a good predictor of
mine RSD's.
Similar analyses were attempted using variance rather
than RSD as a measure of variability. In these instances,
the results were the same. Composite data were not a good
predictor of mine variability.
4-26
-------
4.2.3 Analysis of Predictive Capabilities of Producing
District Data
Figure 13 shows the same type of analysis for composite
data for Producing District 4 (Ohio). Analysis of this data
provides the same general conclusion as the composite seam
data. The estimates are clearly biased in that the composite
Producing District data overestimate mine RSD's in nearly
all instances. The composite Producing District RSD estimate
falls within plus or minus three standard errors for only 6
individual mine RSD estimates. Furthermore, the absolute
error is, fairly large and because of the dispersion of mine
RSD's, there is no curve which will yield a reasonable esti-
mate of RSD.
Analyses with composite data were also performed for
Districts 01, 08, 10, and 11. The results were the same as
previously discussed. The composite Producing District data
were a biased and inaccurate estimate of mine RSD's.
As in the case of the composite seam analyses, separate
analyses were performed using variance instead of RSD.
Again, there were no significant differences in the results.
In conclusion, composite seam or composite Producing
District data cannot be used to accurately predict the vari-
ability of Ibs S/MMBtu for individual mines.
4.3 Analysis of the Statistical Distributions of Coal
Characteristics
This section examines the coal analysis data provided
by the respondents with respect to frequency distributions.
As indicated in Section 2.4, the differences between the
4-27
-------
Figure 13
ANALYSIS OF COMPOSITE PRODUCING
DISTRICT 4 DATA AND DATA FOR
INDIVIDUAL MINES (RAW COALS)
0 Composite Data
Mine Data
100
1000
10,000
-------
three distributions examined normal, lognormal, and
inverted gamma can have a significant impact on compli-
ance with sulfur emission regulations.
Most statistical analyses of coal variability assume
drawing at random from a particular type of frequency dis-
tribution. This section attempts to provide recommendations
as to the most appropriate frequency distributions, based on
the observed frequency distributions of actual coal data.
In this analysis the frequency distributions of sulfur and
heat contents, as well as Ibs S/MMBtu, were examined.
The "best" distribution depends upon the specific pur-
poses or requirements of the analysis. This section analyzes
three possible alternative requirements. The first is to
most accurately predict the top 1.5 percent, or the extreme
right tail, of the distribution. The second requirement
examined is to predict the top 15 percent of the right tail
of the distribution. The final requirement examined is for
the prediction of the total distribution. Separate methods
of analysis were chosen for each of these three criteria.
4.3.1 Methods of Analysis
Table 20 presents an example of the analysis performed
to determine which frequency distribution best fits the
observed data in the top 15 percent of the distribution.
The cell information in Table 20 refers to the cells used in
the computer analysis program, described in Section 3 of
this report. The sum of cells 7, 8, and 9 is approximately
equal to the top 15 percent of the distribution. The rela-
tive error is the observed frequency minus the expected
frequency divided by the expected frequency. Since the
number of observations in the top three cells can be rather
4-29
-------
TABLE 20
ANALYSIS OF THE TOP 15 PERCENT OF THE FREQUENCY DISTRIBUTIONS
BY COMPUTATION OF RELATIVE DISTRIBUTION ERROR!/
Cell
Number
(1)
9
8
7
9,8,&7
9
8
7
9,8&7
9
8
7
9,8,&7
Number of Data Points
Observed
(2)
Expected
(3)
Analysis of Heat Content
Normal
.1
0
2
10
12
0
2
10
12
0
1
9
10
2.4
24.1
26.6
Inverted Gamma
Lognormal
.2
3.0
24.5
27.7
.1
2.4
24.1
26.6
Relative
Error (%)
(4)
-16.66
-58.51
-54.89
-33.33
-59.18
-56.68
-62.66
-62.41
9
8
7
9,8,&7
9
8
7
9,8&7
9
8
7
9,8,&7
Analysis of Lbs Sulfur/MMBtu
Normal
3 .1
8 2.4
13 24.1
24 26.6
3
8
13
24
3
6
15
24
Inverted Gamma
Lognormal
1.0
5.4
24.3
30.7
.1
2.4
24.1
26.6
2900.00
233.33
-46.06
- 9.77
200.00
48.15
-46.50
-21.82
2900.00
150.00
-37.76
- 9.77
I/ For mines with more than 200 observations and cells
containing more than 2 observations.
Relative Error = observed - expected
expected
Source: Foster Associates, Inc.
4-30
-------
small, the relative error can be fairly large without indi-
cating a poor fit.
The method for analyzing the best fit for the top 1.5
percent of the distribution is shown in Figure 14. This
figure is based on the same data presented in Table 20. If
the observed and expected frequencies agreed exactly, the
lines for all three distributions would lie along the 45
degree axis. In this example the inverted gamma distribu-
tion gives a much more accurate fit than either the lognormal
or the normal. i/
The method used to analyze the overall goodness of fit
was the chi-square test. The chi-square statistic is deter-
mined by the following formula:
Where 0^ = observed frequency in cell i
EI = expected frequency in cell i
The calculated chi-square statistic is compared with a
chi-square value for the appropriate degrees of freedom. If
the calculated chi-square statistic is greater than the chi-
square value from the appropriate tables, it can be assumed
that the actual distribution of the coal variable was not drawn
from the assumed distribution. To the extent that the square
of the observed minus the expected frequency divided by the
expected represents a measure of what constitutes a good fit,
for a given number of degrees of freedom, a lower chi-square
value indicates a better fit than a high chi-square value.
I/ The data sets did not contain enough extreme observations
to allow mdre sophisticated tests of goodness of fit.
4-31
-------
I
CO
NJ
FIGURE 14
ANALYSIS OF THE TOP 1.5 PERCENT OF THE FREQUENCY DISTRIBUTIONS
BY COMPARISON OF THE EXPECTED VS OBSERVED FREQUENCIES (%)
1.0
OBSERVED
-------
4.3.2 Results of Best Fit Analysis
The results of the analysis of the statistical distribu-
tions of coal characteristics are discussed in detail in
Appendix K of this report and are summarized in Table 21.
Column (1) in the table identifies the coal characteristic
analyzed, while Column (2) differentiates the raw and washed
coals. Column (3) identifies the portion of the frequency
distribution for which the analysis was performed top 1.5
percent, top 15 percent, or the total distribution. Columns
(4), (5), and (6) note the number of data sets which were
best represented by the respective normal, inverted gamma,
and lognormal distributions. It should be noted that the
data sets represented in Table 21 were not sorted by lot-
size. Lot-size analyses of these data sets are discussed in
Appendix K.
In general, no firm conclusions could be made with
respect to the best distribution for the coal characteristics
analyzed. The distributions exhibiting the best fit varied
considerably from data set to data set. However, the inverted
gamma distribution appears to provide the best fit for Ibs
S/MMBtu, while heat content is best represented by the
normal distribution. In the limited number of data sets
analyzed for sulfur content, supplemented by visual examina-
tion of numerous distributions, the inverted gamma distribu-
tion provided the best fit.
A comparison of the results for raw and washed coals
indicated that coal washing does not alter the shape of the
distribution for Ibs S/MMBtu or heat content. However,
as previously reported, coal washing appears to alter the
mean of the.distribution and reduce the relative variability
of the characteristics investigated.
4-33
-------
In view of the results of these analyses it seems appro-
priate to offer the following recommendations for both raw
and washed coals:
Coal Characteristic
Lbs S/MMBtu
Heat Content (Btu/Lb)
Sulfur Content
Best Fit Distribution
Inverted Gamma
Normal
Inverted Gamma
TABLE 21
SUMMARY OF BEST FIT ANALYSIS FOR THE OBSERVED
FREQUENCY DISTRIBUTIONS OF COAL CHARACTERISTICS.*/
Coal
Characteristic
(1)
Lbs S/MMBtu
Heat Content
Sulfur Content
Type of
Coal
(2)
Raw
Raw
Raw
Washed
Washed
Washed
Raw
Raw
Raw
Washed
Washed
Washed
Raw
Washed
Portion of
Distribution
(3)
Top 1.5 Percent
Top 15 Percent
Total
Top 1.5 Percent
Top 15 Percent
Total
Top 1.5 Percent
Top 15 Percent
Total
Top 1.5 Percent
Top 15 Percent
Total
Total
Total
_!/ Data sets not sorted by lot-size.
Source: Appendix K.
Number of Data Sets
Best Fit By:
Normal
(4)
4
23.5
14
3
9.5
1
13
33
17
9
12
5
15
1
Inverted
Gamma
(5)
16
10
15
7
4
5
10
5
9
2
4
4
17
3
Log-
normal
(6)
19
7.5
14
5
3.5
4
5
2
2
3
1
4
8
5
4-34
-------
5.0 Analysis of Bureau of Mines Data
This section summarizes the results of the analyses
performed on data contained on the Bureau of Mines Coal
History Data Tape ("detail tape"). The coal samples and
analyses contained on this tape were collected by the Bureau
of Mines during the period from 1966 to 1977 and reflect
coals purchased by various Government agencies. The purpose
of these analyses is to determine whether coal suppliers are
providing coal of the specifications defined in coal purchase
contracts. The samples are analyzed by the Bureau of Mines
laboratory and if the coal is not of the quality guaranteed
by the contractor, price adjustments are made.
Although the "detail tape" contains both tipple samples,
collected after the coal has received final treatment at the
tipple or cleaning plant, and delivered samples, collected
during coal unloading at the destination, only the delivered
samples were analyzed for variability.
All samples were collected in accordance with instruc-
tions issued by the Bureau of Mines.A/ The individual incre-
ments or "cuts" represent complete cross-sections of the
entire stream of coal, taken regularly throughout the period
of unloading, so that all parts of the shipment are equally
represented. These individual increments comprise the gross
sample, which weighed not less than 1,000 pounds. In addi-
tion, the maximum tonnage represented by one gross sample
was generally limited to 1,000 tons.
Time and cost constraints prohibited an analysis of all
the data contained on the "detail tape." Only selected seams
I/ Snyder, N. H. (Rev. by S. J. Aresco), Coal Sampling
Revision to Technical Paper 133. Bureau of Mines Handbook,
1957.
5-1
-------
and mines were analyzed, as discussed below.
5.1 Analysis of Selected Coal Seams
The criteria for selection of the coals seams analyzed
were primarily based on the major commercial steam-coal seams
in the United States as identified by Averitt.i/ These are
the few, thick seams, which are continuous over large areas
and possess special properties which make them commercially
desirable. The'se seams contain a substantial portion of the
domestic coal reserve base, and they have yielded the bulk
of past production.
Although data contained in the "detail tape" were not
sufficient to analyze all of the major commercial steam-coal
seams, it was possible to analyze representative seams for
the Northern and Southern Appalachian, Mid-Continent and
Western producing areas. The results of these analyses are
set out in Table 22 and are summarized below by producing
area. It should be noted that all of these analyses were
performed on a lot-size basis. The intervals examined were
0 to 300 tons, 300 to 600 tons, and 600 to 1,000 tons, the
midpoints of which are set out in Column (4) of Table 22.
5.1.1 Northern Appalachian
Two coal seams were analyzed in the Northern Appalachian
area the Pittsburgh seam (036) and Lower Kittanning seam
(084) . Data for the Pittsburgh seam indicated an inverse
relationship between RSD and lot-size; however, data for
this seam were limited to less than 30 observations per lot-
size interval.
I/ Averitt, Paul, Coal Resources of the United States,
January 1, 1974, U.S. Geological Survey Bulletin 1412, 1975.
5-2
-------
TABLE 22
ANALYSIS OF COAL SEAMS BY LOT-SIZE (USBM "DETAIL TAPE" DATA)
CO
Producing Area
(1)
Northern Appalachian
Northern Appalachian
Southern Appalachian
Southern Appalachian
Mid-Continent
Western
Western
Seam Code/Name Preparation
(2) (3)
036 Pittsburgh Raw
084 Lower Kittanning Raw
151 Upper Elkhorn #3 Raw
299 Black Creek Raw
484 Her r in 16 Washed
846 Hiawatha Raw
750 Wadge Raw
Midpoint of
Lot-Size
(Tons)i/
(4)
128
427
971
186
449
857
172
443
901
167
443
970
146
464
909
89
431
834
165
428
844
Lbs
S/MMBtu
Average RSD (%)
(5)
2.38
2.40
2.71
1.96
1.64
1.68
0.67
0.59
0.68
0.76
0.79
1.00
2.29
2.35
2.24
0.47
0.43
0.41
0.63
0.58
0.47
(6)
39.56
34.18
13.73
48.74
29.51
13.56
40.97
27.47
35.28
47.43
53.37
48.93
25.27
23.45
19.49
22.19
15.91
24.41
19.50
14.89
13.07
Number of
Analyses
(7)
19
18
8
96
94
229
102
96
431
53
41
619
493
396
2444
345
24
48
108
57
38
I/ The intervals specified for each seam were:
0 - 300 tons
300 - 600 tons
600 - 1000 tons
Column (4) identifies the midpoint of the intervals for each data set.
-------
The Lower Kittanning seam (084) also exhibited the
inverse relationship between RSD of Ibs S/MMBtu and
lot-size. In this case, the RSD declines from 48.7 percent
at the 0 to 300 ton interval to 13.6 percent at the 600 to
1,000 ton interval.
5.1.2 Southern Appalachian
The Upper Elkhorn #3 (151) and Black Creek (299) seams
were analyzed i.n the Southern Appalachian producing area.
Data for these seams do not exhibit an inverse relationship
between RSD and lot-size. As in the case of the Northern
Appalachian coals, these coals appear to have rather large
RSD's for the 0 to 300 ton lot-size. However, the Northern
Appalachian seams exhibited a progressive decline in RSD
with an increase in lot-size, while the RSD's of the Southern
Appalachian coals remained relatively constant with respect
to lot-size.
5.1.3 Mid-Continent
One coal seam, the Herrin t6 (484), was analyzed in the
Mid-Continent producing area. More than 3,000 observations
for this seam were available in Bureau of Mines "detail
tape." The analysis shows a decline in RSD from 25.3 per-
cent at the 0 to 300 ton interval to 19.5 percent at the 600
to 1,000 ton interval.
5.1.4 Western
Two coal seams, Hiawatha (846) and Wadge (750), were
analyzed in the Western producing area. Although the Wadge
seam exhibited a moderate decline in RSD with respect to
increasing lot-sizes, no such relationship was observed for
5-4
-------
the Hiawatha seam. In order of magnitude, the RSD's of the
smallest lot-sizes (0-300 tons) for the Western producing
area seams were slightly less than the RSD for the Herrin #6
seam in the Mid-Continent producing area and substantially
less than the RSD's for the Northern and Southern Appalachian
areas.
5.1.5 Comparison of Analysis of Seam Data from Bureau of
Mines Data with Data Received from Respondents
As was previously discussed, the Bureau of Mines data
were based exclusively on lot-sizes less than 1,000 tons
each, while data from the respondents generally reflected
larger lot-sizes. Data for four seams Pittsburgh, Lower
Kittanning, Upper Elkhorn #3, and Illinois #6 (Herrin)
were available from the Bureau of Mines "detail tape" as
well as from data received from respondents. A comparison
of these data on a lot-size basis is set out in Table 23.
In general, Table 23 indicates that when the data from the
Bureau of Mines and respondents are compared as a composite,
the inverse relationships between RSD and lot-size which may
have previously existed, are no longer readily apparent.
5.2 Analysis of Selected Mines
In this analysis 16 individual mines contained within
the Bureau of Mines "detail tape" were analyzed. The criteria
for the selection of these mines included: (1) mines produc-
ing individual coal seams for which a large number of analyses
were available, (2) mines for Producing Districts not repre-
sented in the data base assembled from respondents' data,
and (3) mines for which corresponding data were available
from respondents' data, which would be of interest for com-
parative purposes. The results of this analysis on a
5-5
-------
TABLE 23
COMPARISON OF ANALYSIS OF SEAM DATA FROM
BUREAU OF MINES WITH DATA RECEIVED FROM RESPONDENTS
Seam (Code/Name)
Midpoint of
Lot-Size
(Tons)
(1)
036/Pittsburgh
084/Lower
Kittanning
15 1 /Upper
Elkhorn 13
484/Illinois #6
(Herrin)
(2)
128
427
821
971
1,170
1,448
1,782
186
440
449
857
1,015
1,208
1,407
1,582
2,470
173
443
901
1,110
1,737
2,812
146
464
909
1,388
2,354
3,971
5,897
7,262
8,369
9,581
RSD of Lbs
S/MMBtu (%)
USBM Data from
Data Respondents
(3) (4)
39.6
34.2
13.7
48.7
29.5
13.6
41.0
27.5
35.3
25.3
23.5
19.5
22.0
16.1
23.3
17.8
20.7
20.2
20.4
19.9
17.2
18.3
60.7
30.9
22.9
28.8
13.6
16.6
14.3
43.1
45.6
45.3
Number of
Observations
(5)
19
18
15
8
42
78
47
96
29
94
229
105
188
517
91
22
102
96
431
64
32
62
493
396
2,444
993
172
72
186
155
461
204
Source: Foster Associates, Inc.
5-6
-------
lot-size basis are set out in Table 24 and are briefly dis-
cussed below.
5.2.1 Mine 01200, Upper Freeport Seam
Coals analyzed for this mine were washed coals of two
distinct sizes double screened stoker and single screened
slack. The average Ibs S/MMBtu for the single screened
slack coals were consistently greater than the double
screened stoker coals. The RSD's of the Ibs S/MMBtu
for the single screened coals exhibited a decline for each
successive increase in lot-szie. However, this relationship
was not observed for the double screened coals.
5.2.2 Mine 00950, Pittsburgh Seam
All coals analyzed were washed, double screened stoker
coals. Data for this mine exhibited a direct relationship
between RSD and lot-size.
5.2.3 Mine 07290, Middle Kittanning Seam
Data were available only for the 600 to 1,000 ton lot-
size for washed coals. At this lot-size the double-screened
stoker coals exhibited an RSD of 12.3 percent compared to
10.9 percent for the single screened coals.
Limited data for this mine obtained from an electric
utility indicated an RSD of 12.1 percent for shipments aver-
aging approximately 1,000 tons each.
5-7
-------
TABLE 24
ANALYSIS OF INDIVIDUAL MINE DATA FROM USBM "DETAIL" COAL DATA TAPE
Sheet 1 of 2
en
USBM Mine
District Code
(1) (2)
1 01200
2 00950
4 07290
7 00614
8 00637
8 02557
8 04184
Seam
(Code/Name)
(3)
071/U. Freeport
036/Pittsburgh
081/M.
Kittanning
100/Various
956/Elkhorn #3
and Hazard #4
15 I/Upper
Elkhorn #3
Ill/Hazard #5-A
Prepara-
tion
(4)
Washed
Hashed
Washed
Washed
Washed
Washed
Washed
Washed
Washed
Washed
Washed
Raw
Washed
Washed
Washed
Washed
Washed
Washed
Washed
Washed
Washed
Washed
Washed
Washed
Washed
Raw
Raw
Average of
Lot-Size!/
(5)
188
425
869
196
424
924
192
425
789
962
921
950
957
172
457
868
191
451
893
140
472
814
93
398
902
471
817
Lbs
S/MMBtu
Average RSD ( % )
(6)
0.89
0.90
0.90
0.96
0.97
1.01
1.37
1.38
1.25
2.12
2.12
0.87
0.76
0.58
0.57
0.55
0.57
0.59
0.56
0.73
0.72
0.69
0.82
0.84
0.78
0.49
0.47
(7)
11.60
15.21
12.91
13.98
13.11
12.63
15.59
16.84
18.29
12.31
10.87
18.65
18.01
14.86
15.24
13.52
13.50
12.34
13.76
13.96
19.69
21.88
11.89
23.63
18.79
18.28
17.14
Number of
Analyses
(8)
83
150
133
64
73
247
29
41
55
136
371
106
532
61
82
161
61
91
355
84
84
59
53
9
113
396
86
Coal Size
(9)
Double Screened Stoker
Double Screened Stoker
Double Screened Stoker
Single Screened Slack
Single Screened Slack.
Single Screened Slack
Double Screened Stoker
Double Screened Stoker
Double Screened Stoker
Double Screened Stoker
Single Screened Slack
Crushed Run-of-Mine
Single Screened Slack
Double Screened Stoker
Double Screened Stoker
Double Screened Stoker
Single Screened Slack
Single Screened Slack
Single Screened Slack
Double Screened Stoker
Double Screened Stoker
Double Screened Stoker
Single Screened Slack
Single Screened Slack
Single Screened Slack
Crushed Run-of-Mine
Crushed Run-of-Mine
-------
TABLE 24
Sheet 2 of 2
I
<£>
USBM Mine Seam Prepara- Average of Lbs S/MMBtu
District Code (Code/Name)
(1) (2) (3)
8 04204 Ill/Hazard #5-A
fc
10 04978 484/Herrin #6
10 07310 496/Morris #2
11 04730 001/#5, #6, 17
13 00305 299/Black Creek
13 00750 100/Various
13 07243 299/Black Creek
16 02020 799/F
20 07235 846/Hiawatha
tion Lot-Sizei/ Average
(4)
Washed
Washed
Washed
Washed
Washed
Washed
Raw
Washed
Washed
Washed
Washed
Washed
Raw
Raw
Raw
Raw
Raw
Washed
Washed
Raw
Raw
Raw
Raw
Raw
Raw
Raw
(5)
166
446
886
236
480
822
869
896
959
239
458
863
171
465
951
979
976
442
897
136
237
452
861
140
429
76
(6)
0.48
0.48
0.44
0.47
0.55
0.59
1.77
1.64
2.36
0.99
1.25
1.70
0.59
0.53
0.64
0.73
0.87
1.21
1.15
0.25
0.26
0.26
0.26
0.48
0.47
0.53
RSD (%)
(7)
10.73
8.60
12.73
14.09
10.23
13.11
13.17
9.32
10.48
32.35
34.53
34.23
46.22
34.95
32.61
46.33
53.17
14.75
17.96
21.33
18.76
16.13
17.80
17.98
8.52
11.37
Number of
Analyses
(8)
51
39
211
5
138
126
87
301
162
14
41
89
23
15
68
78
97
28
61
68
13
45
77
67
20
126
Coal Size
(9)
Double Screened Stoker
Double Screened Stoker
Double Screened Stoker
Single Screened Slack
Single Screened Slack
Single Screened Slack
Crushed Run-of-Mine
Single Screened Slack
Double Screened Stoker
Double Screened Stoker
Double Screened Stoker
Double Screened Stoker
Double Screened Stoker
Double Screened Stoker
Double Screened Stoker
Crushed Run-of-Mine
Crushed Run-of-Mine
Double Screened Stoker
Double Screened Stoker
Double Screened Egg
Single Screened Slack
Single Screened Slack
Single Screened Slack
Double Screened Stoker
Double Screened Smoker
Single Screened Slack
I/ Intervals specified for each mine were: 0-300 tons, 300-600 tons, and 600-1000 tons.
-------
5.2.4 Mine 00614, Various Seams
Coal analyses were available for this mine on a raw and
washed basis for the 600 to 1,000 ton interval. The average
Ibs S/MMBtu for the raw coal was calculated at 0.87 com-
pared to 0.76 for the washed coal. However, the RSD's of
the Ibs S/MMBtu show no significant difference at 18.6
and 18.0 for the raw and washed coals, respectively.
5.2.5 Mine 00637, Upper Elkhorn #3 and Hazard #4 Seams
Analyses for this multiple seam operation were available
for double screened and single screened washed coals. As
shown in Table 24, there appears to be no significant differ-
ence between the average Ibs S/MMBtu or the RSD of
these two coals. In addition, no inverse relationship
between RSD and lot-size is apparent.
5.2.6 Mine 02557, Upper Elkhorn #3 Seam
Data were available for both double screened and single
screened washed coals. The double screened coals generally
appear to have a lower average Ibs S/MMBtu. On a lot-
size basis, the data exhibit a direct relationship between
lot-size and RSD.
Set out in Table 25 is a comparison of the washed coals
from Mine 02557 to raw coals from the Upper Elkhorn seam.
From Table 25 it appears that the raw coals have a lower
average Ibs S/MMBtu and a higher RSD than the washed
coals.
Analyses for Mine 02557 were also available from data
received from an electric utility. These analyses indicated
5-10
-------
Lot-Size
140
472
814
93
398
902
172
443
901
Average
0.73
0.72
0.69
0.82
0.84
0.78
0.67
0.59
0.68
RSD (%)
14.0
19.7
21.9
11.9
23.6
18.8
41.0
27.5
35.3
Observations
84
84
59
53
9
113
102
96
431
TABLE 25
COMPARISON OF DATA FROM MINE 02557 TO
RAW COAL DATA FOR UPPER ELKHORN SEAM
Midpoint of Lbs S/MMBtu Number of
Source Preparation
Mine 02557 Washed,D.S.i/
Mine 02557 Washed,S.S.I/
Var ious Raw
Mine si/
_!/ Double screened.
2/ Single screened.
V From Table 21.
Source: Foster Associates, Inc.
an average of 0.64 Ibs S/MMBtu with an RSD of 8.9 percent,
based on shipments of approximately 2,600 tons each.
5.2.7 Mines 04184 and 04204, Hazard #5-A Seam
Mines 04184 and 04204 both produce coal from the Hazard
#5-A seam. Mine 04184 is an auger operation producing a
crushed run-of-mine product, while Mine 04204 is a strip
mine producing double and single screened coals. In general
there appears to be no significant difference in the average
Ibs S/MMBtu for these coals, although the RSD's for the
raw coals appear to be slightly higher.
5.2.8 Mine 04978, Herrin #6 Seam
Analyses of this mine were based on raw and washed coals
in the 600 to 1,000 ton lot-size. The raw coals exhibited
5-11
-------
an average of 1.8 Ibs S/MMBtu with an RSD of 13.2 per-
cent, while the washed coals had an average of 1.6 Ibs
S/MMBtu with an RSD of 9.32 percent.
Data for this mine received from an electric utility
indicated an average of 2.8 Ibs S/MMBtu with an RSD of
13.1 percent, based on washed coals with an average lot-size
of approximately 2,000 tons.
5.2.9 Mine 0-7310, Morris #2 Seam
Bureau of Mines data for this mine indicated an average
of 2.4 Ibs S/MMBtu with an RSD of 10.5 percent, for the
600 to 1,000 ton lot-size. Similar data received from an
electric utility provided an average of 2.5 Ibs S/MMBtu
with an RSD of 10.9 percent, based on an average lot-size of
4,700 tons.
5.2.10 Mine 04730, Indiana Seams #5, #6, and #7
Coal analyses for this multiple seam mine were based on
washed, double screened stoker coals. By lot-size the aver-
age Ibs S/MMBtu ranged from 1.0 to 1.7, while the RSD's
ranged from 32.4 to 34.5 percent. Comparable data received
from a coal producer indicated an average of 1.6 Ibs
S/MMBtu with an RSD of 34.3 percent, based on an aver-
age lot-size of 8,000 tons.
5.2.11 Mines 00305 and 07243, Black Creek Seam
Data for Mine 00305 were based on raw, double screened
and crushed run-of-mine coals, while data for Mine 07243
were based on washed, double screened coals. The raw coals
from Mine 00305 exhibited a lower average Ibs S/MMBtu
5-12
-------
and higher RSD's compared to the washed coals from Mine 07243
The limited data for these mines did not permit an analysis
of the relationship of RSD and lot-size.
5.2.12 Mine 00750, Various Seams
Based on crushed run-of-mine coals from a multiple seam
operation, this mine exhibited an RSD of 53.2 percent for
coal volumes ranging from 600 to 1,000 tons.
5.2.13 Mine 02020, F Seam
Analyses for this mine were based on raw, double and
single screened coals. No significant difference was
observed between the average Ibs S/MMBtu for these
coals, although the data indicated a slightly higher RSD for
the double screened coal.
Comparable data for this mine received from an electric
utility indicated an average of 0.54 Ibs S/MMBtu, com-
pared to 0.25 to 0.26 from the Bureau of Mines data. The
RSD of the Ibs S/MMBtu was calculated at 5.0 percent
from the utility data, while the Bureau of Mines data indi-
cated RSD's from 16.1 to 21.3 percent.
5.2.14 Mine 07235, Hiawatha Seam
Coals analyzed for Mine 07235 from the Bureau of Mines
data were raw, double and single screened coals. By lot-
size intervals the average Ibs S/MMBtu ranged from 0.47
to 0.53 while the RSD's ranged from 8.5 to 18.0 percent. In
comparison, data from an electric utility exhibited an aver-
age of 0.90. Ibs S/MMBtu with an RSD of 27.6 percent,
based on lot-size shipments of approximately 8,400 tons each.
5-13
-------
5.2.15 Summmary of Analysis of Individual Mine Data from
"Detail Tape"
In the analysis of the 16 selected mines from the U.S.B.M.
"detail tape" the RSD's of Ibs S/MMBtu generally ranged
from 10 to 20 percent with two exceptions. First, Mine
04730 in Producing District 11 (Indiana) exhibited RSD's
ranging from 32 to 34 percent. Second, Mines 00305 and
00750 in Producing District 13 (Alabama) exhibited RSD's
ranging from 32 to 53 percent.
Based on the theoretical relationship between RSD and
lot-size it was expected that rather large RSD's would be
observed in the Bureau of Mines data, since all lot-sizes
were 1,000 tons or less. As indicated above, relatively
large RSD's were not observed and the results in general
were similar to the results obtained from the respondents'
data, which generally represented substantially larger lot-
sizes.
In the 15 cases where two or more lot-size intervals
were available for the same coal (same mine, same prepara-
tion, and same size) only three cases exhibited a consis-
tent decline in RSD with lot-size.
The two mines for which raw and washed analyses were
available indicated that the average Ibs S/MMBtu as
well as the RSD's were lower for washed coals.
In the seven cases where analyses were available for
double screened and single screened coals from the same
mine, there was no apparent consistent difference in the
average Ibs S/MMBtu or the RSD's.
5-14
-------
Finally, the comparisons of Bureau of Mines data with
data received from respondents for the same mines yield
inconsistent results. Since the data from the respondents
generally reflected larger lot-sizes, it was expected these
data would exhibit lower RSD's than those observed in the
Bureau of Mines data. For the seven mines compared, the
respondents' data showed lower RSD's in two cases and higher
RSD's in two cases. For the three remaining mines, the RSD's
for the respondents' data were not significantly different
from the Bureau of Mines data.
The results of these various comparisons must be viewed
with caution, since the coal sampling and analyses were not
performed by the same samplers or laboratories under con-
trolled conditions. Moreover, the sporadic nature of the
Bureau of Mines coal samples presents problems in statisti-
cal analysis. For this reason, it is questionable whether
more sophisticated models of variance, including autocorrela-
tive models, would be useful alternatives to the simple model
used in this study.
5-15
-------
6.0 Analysis of Continuous Monitoring Data for Sulfur
Dioxide Emissions
The objective of this chapter was to analyze existing
data for sulfur dioxide (S02) emissions and the removal
efficiencies of flue gas desulfurization (FGD) systems. The
analysis focused on the behavior of the variabilities of 502
emissions and FGD efficiencies as a function of averaging
time. The behavior of the observed data was then compared
to the expected behavior, based on statistical theory-
6.1 Description of Data
The data used in this study were collected by EPA and
reflect the results of continuous monitoring test programs
conducted at the Cane Run Unit No. 4, Bruce Mansfield Unit
No. 1, Eddystone Unit No. 1, and Mitchell electric generat-
ing units. The data for the Mitchell unit were based on
one-hour averages while the data for the three remaining
units were based on 15-minute averages.
One advantage of using these data was that they were
previously reduced, edited and reviewed. !_/ Since these
data were collected under controlled conditions, it was
possible to delete anomalous observations resulting from
factors such as instrument and equipment malfunctions.
I/ A complete description of these data may be obtained
from Air Pollution Emission Test, Volume I: First Interim
ReporF:Continuous Sulfur Dioxide Monitoring at Steam
Generators, U.S. EPA, Emissions Measurement Branch, EMB
Report NO. 77SPP23A, August 1978.
6-1
-------
6.2 Analysis of Data
The data analyses in this section were performed with a
modified version of the analytical program developed for the
analysis of coal data.
The results of the analysis of sulfur dioxide emissions
and FGD efficiencies are set out in Table 26. AS the table
shows, the analysis consisted of calculating one-hour, three-
hour, and twenty-four-hour simple averages as well as the
corresponding moving averages. For the Mitchell and Cane
Run Units the data also permitted the calculation of a 30-
day moving average. The analysis is limited in its general
applicability due to the limited number of data sets. How-
ever, even given the limited amount of data, some general
observations can be made.
The first observation is that, while the FGD efficiencies
are not subject to large relative variations, the relative
variations exhibited in the FGD inefficiencies are substan-
tial. This can be seen by comparing Columns (10) and (11)
in Table 26. Based on a 24-hour averaging period, the RSD's
of the FGD sulfur removal efficiencies ranged from 1.2 to
6.2 percent. In contrast, the RSD's of the FGD inefficiencies
ranged from 10.5 to 70.8 percent for the same averaging period.
Since emissions into the atmosphere are the product of inlet
flue gas concentrations times FGD inefficiency, it appears
that the relative variability in FGD performance is more
important than the combined relative variabilities of the
coal and combustion processes.i/
I/ It is also interesting to note that in a correlation
analysis performed on the Cane Run Unit No. 4 data by the
Energy Strategies Branch of EPA a correlation coefficient of
0.35 was obtained for inlet versus outlet emissions, based
on three-hour averages. This correlation coeffient indicates
that inlet and outlet emissions tend to vary independently
about their respective means. This independence can be explained
only by the variability in the performance of the FGD unit.
6-2
-------
Table 26
ANALYSIS OF VARIABILITIES OF SULFUR DIOXIDE EMISSIONS AND ?OT
REMOVAL EFFICIENCIES AS A FUNCTION OF AVERAGING TIME PERIOD
Flue Gas Concentrations (Lbs SO-j/MMBtu)
PGD Inecfi-
Unit and Aver-
aging Period
(1)
Eddy stone No. 1
1-Hour
3-Hour
3 -Hour Moving
24-Hour
24-Hour Moving
Mitchell
1-Hour
3-Hour
3-Hour Moving
2 4 -Hour
24-Hour Movina
30-Day Moving!/
Mansfield No. 1
1-Hour
3 -Hour
3-Hour Moving
2 4 -Hour
24-Hour Moving
Cane Run No. 4
1-Hour
3-Hour
3 -Hour Moving
2 4 -Hour
2 4 -Hour Moving
30-Day Moving!/
I/ FGD Sfficien
Mean
(2)
5.1230
5.1232
5.1213
5.1161
5.1091
6.5482
6.7216
6.7178
6.7150
6.7194
6.7220
6.6198
6.6216
6.6212
6.6187
6.6354
5.6435
5.6423
5.6425
5.6398
5.6433
5.6503
Lbs
cv »
Inlet
S.D..i'
(3)
.4190
.4100
.4091
.3656
.3617
1.1886
.9284
.9289
.6988
.6827
.0099
.7948
.7533
.7566
.4486
.4814
.5412
.5255
.5236
.4203
.4108
.0318
S02/MMBt
RSD (%)
(4)
08.18
08.00
07.99
07.15
07.08
18.15
13.81
13.83
10.41
10.16
00.15
12.01
11.38
11.43
06.78
07.26
09.59
09.31
09.28
07.45
07.28
00.56
LU Inlet
Mean
(5)
.2607
.2597
.2584
.2422
.2456
.6737
.6699
.6697
.6697
.6702
.6742
1.2699
1.2692
1.2702
1.2722
1.2867
.9100
.9084
.9090
.9059
.9096
.9318
Outlet
S.D.-i'
(6)
.2799
.2578
.2583
.1837
.1702
.1832
.1606
.1604
.1089
.1094
.0050
,5433
.5137
.5146
.3625
.3606
.3669
.3533
.3533
.2892
.2859
.0978
- Lbs 502/MMBtu
FGD Efficiency
RSD m
(7)
107.38
99.28
99.97
75.83
69.30
27.20
23.98
23.95
16.26
16.33
00.74
42.79
40.47
40.52
23.49
28.03
40.32
38.89
38.87
31.92
31.43
10.50
Outlet
Mean
(8)
94.6551
94.6378
94.6637
94.9278
94.3830
89.9198
89.3779
89.8785
89.3406
39.3452
89.3043
30.8948
80.9864
80.9704
30.9522
90.7625
83.6626
33.6663
83.6582
83.6907
83.6310
83.2279
S.O.i/
(9)
5.4488
5.0120
5.0259
3.5912
3.3816
2.2469
1.8771
1.8792
1.0779
1.0696
0.0691
7.0020
6.5540
6.5469
4.7538
4.6223
6.4201
6.1958
6.2007
5.1752
5.1775
1.7603
(%)i/
aso (%)
(10)
5
5
5
3
3
2
2
2
1
1
0
8
8
3
5
5
7
7
7
6
6
2
.756
.295
.309
.783
.563
.498
.088
.090
.199
.190
.076
.655
.092
.085
.372
.723
.673
.405
.411
.183
.190
.115
T /
siencyi'
RSD (%)
(11)
101.85
93.47
94.18
70.30
66.09
22.29
18.54
18.57
10.61
10.53
0.68
36.31
34.48
34.40
24.96
24.03
39.30
37.93
37.94
31.73
31.63
10.50
Lbs SO2/MMBtu Inlet
2/ FGD Inefficiency * 1-Efficiency
V S.D. = Standard Deviation
4/ Analyses for 30-day moving average are based on very limited data.
Source: Foster Associates, Inc.
6-3
-------
The RSD's of the inlet SC>2 concentrations ranged from
8.2 to 18.2 percent based on a one-hour averaging period.
Although there are inherent difficulties in comparing coal
analysis data to emissions data, as previously discussed,
these RSD's are comparable to the RSD's of coal in the 0 to
300 ton range examined in the Bureau of Mines data.I/
A comparison of the FGD inlet data to outlet data shows
that the outlet emissions are much more variable than the
inlet emissions. Column (2) of Table 27 sets out the ratio
of the RSD of inlet emissions to the RSD of outlet emissions
for the various averaging periods examined for the four gen-
erating units. Based on the one-hour averaging periods,
this ratio varies from 1.5 for the Mitchell unit to more
than 13 for the Eddystone Unit No. 1. In other words, at
the Eddystone unit, the relative variability of the outlet
emissions is more than thirteen times greater than that of
the inlet emissions. The implication of these results is
that although the FGD unit reduces the mean emission rate,
which aids compliance, the FGD unit also increases the rela-
tive variability of outlet emissions, which increases the
difficulty of compliance.
A final observation is that the reduction in emissions
variability resulting from increasing the averaging period
is less than would be expected from statistical approxima-
tions, if independence is assumed. It can be shown that if
moving or simple averages are calculated from one-hour aver-
ages with a standard deviation of a , the standard deviation
for a moving or a simple average of n data points would have
a standard deviation of- This relationship assumes that
I/ A 500 MW generating unit would consume coal at a rate of
about 200 tons/hour.
6-4
-------
TABLE 27
COMPARISON OF THE VARIABILITY OF FGD INLET AND OUTLET
EMISSIONS, OBSERVED AND EXPECTED
Sulfur Dioxide Emissions (Lbs SOo/MMBtu)
Unit and Aver-
aging Period
(1)
Eddystone No. ]
1-Hour
3-Hour
3-Hour Moving
24-Hour
24-Hour Moving
Mitchell
1-Hour
3-Hour
3-Hour Moving
24-Hour
24-Hour Moving
30-Day Moving
Mansfield No. 1
1-Hour
3-Hour
3-Hour Moving
24-Hour
24-Hour Moving
Cane Run No. 4
1-Hour
3-Hour
3-Hour Moving
24-Hour
24-Hour Moving
30-Day Moving
Inlet
RSD Outlet
RSD Inlet
(2)
13.13
12.41
12.51
10.61
9.79
1.50
1.74
1.73
1.56
1.61
4.93
3.56
3.56
3.55
4.20
3.86
4.20
4.18
4.19
4.28
4.32
18.75
Expected
S.D.I/
(3)
._
.2419
.2419
.0855
.0855
__
.6862
.6862
.2426
.2426
.1276
__
.4589
.4589
.1622
.1622
_^ _JB
.3125
.3125
.1105
.1105
.0767
Observed S.D.
Expected S.D.
(4)
1.6949
1.6912
4.2760
4.2304
1.3530
1.3537
2.8805
2.8141
.0776
1.6415
1.6487
2.7657
2.9679
__
1.6816
1.6755
3.8036
3.7176
.4146
Outlet
Expected
S.D.I/
(5)
.1616
.1616
.0571
.0571
.1058
.1058
.0374
.0374
.0199
__
.3137
.3137
.1109
.1109
__
.2118
.2118
.0749
.0749
.0528
Observed S.D.
Expected S.D.
(61
1.5953
1.5984
3.2172
2.9807
1.5180
1.5161
2.9118
2.9251
.2513
1.6376
1.6404
3.2687
3.2516
1.6681
1.6681
3.8611
3.8171
1.8523
'L/ S.D. = Standard Deviation
Source: Foster Associates, Inc.
6-5
-------
the distribution of the one-hour averages is not serially
correlated.!/
Columns (4) and (6) of Table 27 compare the standard
deviations of the observed moving and simple averages with
the standard deviations which would be expected given the
relationship and statistical assumption presented in the
previous paragraph. From Table 27 it can be seen that in
almost every case the expected standard deviation is less
than the observed standard deviation. Thus, it appears that
substantial reductions in emissions variability result from
longer averaging periods, but these reductions are less than
what would be expected under assumptions of statistical
independence.
6.3 Implications of Emissions Analysis
To the extent that these four FGD units are representa-
tive of FGD units in general, the results of these analyses
identify several factors which have an impact on compliance
with sulfur dioxide emission regulations.
First, the reduction in emissions variability is readily
apparent as the averaging interval is increased consecutively
from one hour, to three hours, to 24 hours, to 30 days.
These findings support the theoretical inverse relationship
between coal sulfur variability and lot-size, since increasing
the averaging interval is equivalent to increasing the incre-
ment or lot-size of coal burned.
I/ Yamane, Taro, Statistics, An Introductory Analysis,
Harper and Row, Third Edition, p. 1072.
6-6
-------
It follows from this analysis that, the shorter the
averaging interval or the smaller the amount of coal burned
per unit time, the more difficult it is to comply with an
emission standard if the averaging basis is time. Moreover,
the actual air pollution decreases as the difficulty of
compliance increases. These findings are particularly rele-
vant to small coal-fired plants and plants operating at
coal-burn rates lower than the rates used to develop its
compliance strategy. With respect to coal purchasing these
plants would experience greater problems, especially the
smaller plants with compliance strategies based on low-sulfur
coal. While large sources might find it necessary to ensure
that unit train loads (100 cars or approximately 10,000 tons)
meet the standard, small sources would have to ensure compli-
ance for perhaps two or three carloads. Due to the natural
variability of coal, it is possible that, given the same
source of coal supply, the coal would comply when burned in
a large plant but would result in excess emissions in a
small plant. Alternatively, the source of supply may be
acceptable for the small plant, but the increased number of
coal analyses and the selectivity required for quality con-
trol would certainly increase coal costs.
The second implication of the result of this analysis
concerns those plants utilitizing FGD control strategies.
Based on current technology, FGD is in most cases the only
method available to meet stringent sulfur dioxide emission
regulations. Although FGD units reduce the mean or average
emission rate, it appears that they greatly increase the
relative variability of the outlet emissions, which increases
the difficulty of compliance. In the four units analyzed,
it was found that the relative variability in emissions
induced by F.GD performance is significantly more important
than the relative variabilities of the coal and the combustion
6-7
-------
process combined. As a consequence of these findings, the
evaluation of a source of coal supply for a coal-fired plant
equipped with an FGD unit must address not only the problem
of natural coal sulfur variability, but also the variability
in FGD performance. In general, this would require coal with
a lower mean sulfur content than if it were assumed that the
FGD unit operated at a constant rate of efficiency.
Finally, this analysis indicated that reductions in
variability obtained from longer averaging intervals, although
singificant, are less than would be expected based on assump-
tions of independence. These results suggested that an auto-
correlative model may be more appropriate for outlet emissions,
In addition, future studies of emissions analysis should
address the effect of FGD units on extreme values as well as
the relative variability.
6-8
-------
7.0 Coal Sulfur Regression Analysis
7.1 Objectives
The purpose of the coal sulfur regression analysis is
to examine the relationship between the sulfur content of
coals and other coal characteristics. Because of the data-
base limitations, these other coal characteristics are
restricted to the following: ash content (AS), heat content
(BT), moisture content (MO), lot-size (TON), mining method
(MM), sampling method (SM), and the level of coal prepara-
tion (PC). The relationships between coal sulfur content and
these other coal characteristics are examined for their
statistical significance, consistency, and their explanatory
or predictive power. Wherever data permit, the regression
analysis is disaggregated to three levels: a Bureau of Mines
Producing District, a seam in that producing district, and a
mine in that seam. In addition to examining the relationship
between the average sulfur content (SU) of coals and the
other coal characteristics, the relationship between the coal
sulfur variability, measured by the relative standard devia-
tion or the variance (o^), and the other coal characteristics
is examined.
7.2 Background
Sulfur does not occur as an element in coal, but as
chemical combinations with other substances. Organic sulfur
is combined with the organic coal substance and is part of
the coal. Pyritic sulfur is combined with iron as either
pyrite or marcasite. Sulfate sulfur is combined with either
calcium or iron and is generally less than 0.05 percent of
the coal. I/
I/ F. E. Walker and F. E. Hartner, Forms of Sulfur in U.S.
Coals, U.S. Bureau of Mines, 1C 8301, 1966, p. 2.~
7-1
-------
Large differences in local and regional variability of
sulfur frequently occur. Gomez, Donaven, and Hazen have per-
formed a statistical and spatial evaluation of sulfur in
coal seams and found very little correlation between the
chemical and physical properties of the coal and sulfur con-
centration.!/ Figures 15 and 16 show the relationship between
the average sulfur content and the various chemical and
physical coal properties examined by Gomez, Donaven, and
Hazen. No strong correlation between the sulfur content
and ash content, moisture content, volatile matter, heat
content, grindability, fixed carbon, free sweeling index, or
ash softening temperature appears to exist. The results of
their study led Gomez, Donaven, and Hazen to conclude, "...
the chemical and physical properties of the coal are secondary
variables influencing sulfur and mineral matter distribution
in coal beds. The primary variables affecting sulfur dis-
tribution, quite likely, are geologic factors related to the
depositional history of the seam."?./
Discussions with Professor Joseph Leonard, Dean of the
College of Mineral and Energy Resources at West Virginia
University and Director of the Coal Research Bureau, and
Dr. Francis Ting, professor of coal geology at West Virginia
University, support the conclusions reached by Gomez, Donaven,
and Hazen.3/ Both Professor Leonard and Dr. Ting agree that
it is the geologic factors which determine the sulfur content
of coals and not the chemical or physical properties of the
coals themselves. Therefore, characteristics such as seam
I./ Manual Gomez, Donald J. Donaven, and Kathleen Hazen, The
Statistical and Spatial Evaluation of Sulfur and Ash in Coal
Seams, U.S. Bureau of Mines, RI 7679, 1972.
2/ Ibid., p. 3.
V Conversations with Professor Joseph Leonard and Dr. Francis
Ting, West Virginia University, May 14, 1979.
7-2
-------
i
U)
12.0
0)
U
0)
Q.
> 10.0
8.0
UJ
^y ^
1.3
1.5 1.7 1.9 2.1
SULFUR MEAN, DRY BASIS, percent
2.3
2.5
1.3
1.5 1.7 1.9 2.1 2.3
SULFUR MEAN, DRY BASIS, percent
40.0 a
20.0
0
2.7
UJ
O
UJ
UJ
ct
Z)
t-
(/2
o
5
O)
000
a:
uj .a
2
- 13,000 »-
ffi
a:
9,000 < UJ
> i-
M
<
5,000 i
2.7
FIGURE 15 - Relationship Between Average Values for Volatile Matter, Ash, Moisture, Heating Value,and Sulfur.
Source: U.S. Bureau of Mines, RI 7679.
-------
500
LJ
o:
- 2,300
5
LJ
t-
<£>
Z
- 2, 100
LJ
I-
L.
O
(ft
I
(O
1.5 1.7 1.9 2.1 2.3
SULFUR MEAN, DRY BASIS, percent
2.5
2.7
900 <
Fixed carbon
~~~ "
/ Grindability index
1.5 1.7 1.9 2.1 2.3
SULFUR MEAN, DRY BASIS, percent
FIGURE 16 - Relationship Between Average Values for Ash Softening Temperature, Free-
Swelling Index, Fixed Carbon, Hardgrove Grindability Index, and Sulfur.
0
Source: U.S. Bureau of Mines, RI 7679.
-------
thickness, elevation, roof material, and type of deposition
which are associated with geologic occurrences may have a
stronger relationship to sulfur content than the chemical
and physical properties of coal. For example, in the Eastern
coal producing areas, lower sulfur coals are generally associ-
ated with thin seams, high elevation, clay roof, and fresh-
water deposition. Thin seams, high elevations, and freshwater
are inferences of moving water environments which removed
impurities from the coal. Clay roof materials are thought
to have acted as a filter which prohibited impurities from
penetrating the coalbed. Higher sulfur Eastern coals are
generally associated with thick seams, low elevations, and
marine deposition which are inferences of stagnant water
environments where impurities were not removed from the
coalbeds. However, the coal database developed for this
study does not contain geologic factors and cannot be used
to test the hypothesis of a greater correlation between
sulfur content and geologic factors than between sulfur
content and the chemical and physical properties of the coal.
7.3 Average Sulfur Content Regression Results
Linear regression analysis was used to examine the
relationship between the average sulfur content of the coals
and the other coal properties contained in the coal database.
Where sufficient data were available to provide meaningful
results, regressions were performed for individual coal pro-
ducing districts, particular seams, and single mines. Coal
Producing Districts 10 and 8 were disaggregated to the seam
and mine levels, while Producing District 4 data did not
permit analysis at the seam or mine levels. In Producing
District 19 the regression analysis at mine level was based
on core sample data and "as shipped" data.
7-5
-------
7.3.1 Producing District 10
The coal analysis data for Producing District 10 were
sufficient for regression analysis at the district level,
the seam level, and the mine level. Regression equations
for Producing District 10, the Illinois #6 Seam, and Mine #1
are shown below:
Producing District 10:
SU = 40.11608 + 0.60056 PC - 0.00270 Btu - 0.00775 SM
(15.53) (30.00) (1.48)
-0.00004 TON + 0.21590 MM - 0.37498 MO
(undefined)* (14.18) (28.03)
-0.38478 AS
(27.52)
R2 = 0.34258 F = 335.58687 S.E. = 0.65832
Illinois #6 Seam:
SD = 29.7889 + 1.04531 MM - 0.47590 PC
(44.83) (22.16)
-0.00194 BT + 0.07689 SM - 0.24858 AS
(17.54) (17.88) (14.86)
-0.22782 MO
(12.52)
R2 = 0.55424 F = 665.20574 S.E. = 0.58117
Mine #1:
SU = 5.05320 - 0.00006 TON + 0.05756 AS
(6.00) (3.08)
+ 0.60118 PC - 0.06762 SM - 0.05224 MO
(7.58) (3.30) (2.85)
- 0.00024 BT - 0.08957 MM
(2.00) (1.28)
R2 = 0.40888 F = 69.17062 S.E. = 0.22051
*Note: T-statistics were estimated by dividing the regression
coefficient of each variable by its corresponding
standard error. By defintion the t-statistic was
undefined when the standard error was zero.
7-6
-------
where: SU = sulfur content
AS = ash content
BT = heat content (Btu)
MO = moisture content
TON = lot-size
MM = mining method
SM = sampling method
PC = preparation code (Raw or Washed)
T-statistics for each regression coefficient are in paren-
theses. In most cases the t-statistics are significant at
the 95 percent confidence level which indicates that the
regression coefficients are statistically different from
zero. Further information about each regression equation is
presented in Table 28 which shows correlation coefficients,
regression equation R^, and variable rankings determined by
a stepwise regression which orders each regression variable
by its contribution to the explained variance of the equa-
tion.
Although the t-statistics of the regression coefficients
and the F-statistics for the equations are generally signifi-
cant at the 95 percent confidence level (t = 1.96, F > 2.00),
the equations have very poor explanatory or predictive power
as indicated by the low R^ and high standard errors. Only
about 34 percent of the total variance is accounted for by
the regression equation for Producing District 10 as a whole.
The regression equations account for about 55 percent and 41
percent of the total variance for the Illinois #6 Seam and
Mine #1, respectively. High standard errors which range from
about 0.66 to 0.22 also indicate weak explanatory or predic-
tive powers for these equations. However, there is generally
a trend toward higher R^ and lower standard errors as one
disaggregates from the Producing District level to the seam
level to the mine level. These trends tend to support the
7-7
-------
TABLE 28
SULFUR REGRESSION ANALYSIS SUMMARY FOR PRODUCING DISTRICT 10
(PRODUCING DISTRICT 10, ILLINOIS #6 SEAM, MINE #1)
Correlation Coefficients
Variable
SU
AS
BT
MO
TON
MM
SM
PC
Number of
Variables
1
2
3
4
5
6
7
Variable
Rank
1
2
3
4
5
6
7
Producing
District 10
Illinois
#6 Seam
SU SU
1.00000 1.00000
-0.01498 0.26644
-0.26196 -0.41158
0.17751 0.11881
-0.16336 -0.24973
0.20635 0.65276
-0.20061 0.14185
0.31943 -0.06650
Regression Equation
Producing
District 10
0.10203
0.16615
0.19855
0.21783
0.22517
0.23213
0.34258
Regression
Producing
District 10
PC
BT
SM
TON
MM
MO
AS
Illinois
#6 Seam
0.42069
0.46088
0.50120
0.51899
0.53247
0.55424
0.55437
Equation Var
Illinois
#6 Seam
MM
PC
BT
SM
AS
MO
TON
Mine #1
SU
1.00000
0.17981
-0.36402
0.22398
-0.46582
0.44043
-0.31045
0.34396
R2
Mine #1
0 .'21699
0.32448
0.37546
0.39796
0.40462
0.40749
0.40888
iables
Mine #1
TON
AS
PC
SM
MO
BT
MM
SU = sulfur (as received)
AS = ash (as received)
BT = heat content (Btu)
MO = moisture content
TON = lot-size
MM = mining method
SM = sampling method
PC = preparation code (raw or washed)
7-8
-------
a priori hypothesis that the fit of the regression equations
should improve as the data are disaggregated and are more
likely to be from similar populations at the seam and mine
level than at the Producing District level. However, it
should be noted that the highest R^ occurs in the regression
equation for the Illinois #6 Seam, while the lowest standard
error occurs in the equation for Mine #1.
The regression equations and Table 28 indicate many
inconsistencies in the values and signs of the correlation
coefficients, the values and signs of the regression coeffi-
cients, and the order of the independent variables in the
various levels of disaggregation. Only heat content (BT),
moisture content (MO), lot-size (TON), and mining method
(MM) have the same sign in each regression equation if the
statistically insignificant values are ignored. At each
level of disaggregation the sulfur content of the coal shows
an inverse relationship to the heat content, moisture content,
and lot-size and a positive relationship to the mining method.
The moisture content parameter is somewhat of an oddity in
that the moisture regression coefficient in each equation is
negative, while the moisture correlation coefficient is
positive. An explanation for this oddity may be that inclu-
sion of other variables in the regression equations which
exhibit stronger influences on the sulfur content than does
the moisture content suppresses the positive relationship
which the moisture content individually exhibits with sulfur
content. Multicollinearity, which is the relationship
between the explanatory variables, may also contribute to
this oddity. No consistent relationship appears to exist at
each level of disaggregation between the sulfur content and
the remaining independent variables (ash, sampling method,
preparation code).
7-9
-------
Large differences also occur in the relative importance
of each explanatory variable in each regression equation.
Only one variable, preparation code (PC), appears in each
equation as one of the three most important variables. How-
ever, each explanatory variable except moisture content (MO)
appears at least once as one of the three most important
variables in one of the equations. These differences in the
relative importance of each explanatory variable in each
regression equation is another indication that no strong and
consistent relationship between sulfur content and the coal
properties contained in the database exists for coals in
Producing District 10.
7.3.2 Producing District 8
Sufficient data were available for Producing District 8
to disaggregate the data to a seam and a mine level. Regres-
sion equations for Producing District 8, the Alma (Blue Gem)
Seam, and Mine #2 are shown below:
Producing District 8:
SU = -2.22768 - 0.15875 PC + 0.08315 MM
(32.14) (21.94)
+ 0.00007 TON + 0.05949 AS + 0.00018 BT
(undefined)* (7.87) (3.60)
+ 0.01521 MO
(1.49)
R2 = 0.43598 F = 260.62299 S.E. = 0.38109
*Note: T-statistics were estimated by dividing the regression
coefficient of each variable by its corresponding
standard error. By definition the t-statistic was
undefined when the standard error was zero.
7-10
-------
Alma Seam:
(Blue Gem)
SU = -2.58155 + 0.07662 AS + 0.00024 BT
(7.50) (4.80)
R2 = 0.16564 F = 25.67525 S.E. = 0.28988
Mine #2:
(1)
(2)
SU = 0.11525 - 0.09569 MM + 0.04916 AS
(6.16) (1.93)
-0.92574 MO + 0.00002 TON + 0.00008 BT
(1.02) (2.00) (0.53)
R2 = 0.28895
SU
F = 27.06459
S.E. = 0.24026
1.26368 - 0.09319 MM + 0.03656 AS
(6.33) (5.82)
-0.03751 MO + 0.00002 TON
(3.62) (2.00)
R2 = 0.28839 F = 33.84014 S.E. = 0.23999
where: SU = sulfur content
AS = ash content
BT = heat content (Btu)
MO = moisture content
TON = lot-size
MM = mining method
The t-statistics for each regression coefficient are in paren-
theses. For the Producing District equation and the seam
equation the t-statistics are generally significant at the
95 percent confidence level (t =» 1.96) for each regression
coefficient. However, only the mining method (MM) and the
lot-size (TON) regression coefficients are significant in
equation 1 for Mine #2. This is probably caused by the high
collinearity primarily between the ash (AS) and heat content
(BT) variables which have a correlation coefficient of
-0.89155. Dropping the heat content variable (BT) from the
equation doe's not significantly affect the value of R2, F,
7-11
-------
or the standard error and results in significant regression
coefficients for each variable (see mine #2, equation 2).
Table 29 shows correlation coefficients, regression equation
R2, and variable rankings as determined by a forward step-
wise regression.
At each level of disaggregation the regression equations
show very poor explanatory or predictive power even though
the regression coefficients and the equations themselves are
significant as- based on the t-tests and F-tests, respectively.
Only about 44 percent of the total variance for Producing
District 8 as a whole is accounted for by the regression
equation. The explanatory or predictive power of the regres-
sion equations for the Alma (Blue Gem) Seam and for Mine #2
is even less, accounting for approximately 17 and 29 percent,
respectively.- of the total variance. Standard errors for the
equations are fairly high ranging from about 0.43 to 0.29
and are further indications of the weakness of the explana-
tory or predictive power of these regression equations.
Unlike the regression equations for Producing District 10,
the goodness of fit of the regression equations for Producing
District 8 do not tend to improve as one disaggregates the
data to the seam and mine levels. The highest R2 occurs at
the Producing District level, although the standard errors
decline with each level of disaggregation. Therefore, these
regression results for Producing District 8 do not support
the a priori hypothesis that coal characteristics become more
homogeneous and more easily predictable as the level of
disaggregation increases.
Table 29 and the regression equations indicate many
inconsistencies in the values and signs of the correlation
coefficients and regression coefficients and in the order of
7-12
-------
TABLE 29
SULFUR REGRESSION ANALYSIS SUMMARY FOR PRODUCING DISTRICT 8
(PRODUCING DISTRICT 8, ALMA (BLUE GEM) SEAM, MINE #2)
Correlation Coefficients
Variable
SU
AS
BT
MO
TON
MM
PC
Number of
Variables
1
2
3
4
5
6
Variable
Rank
1
2
3
4
5
6
Producing
District 8
SU
1.00000
-0.09804
-0.00158
-0.04998
0.10708
-0.04654
-0.42106
Regress
Producing
District 8
0.17729
0.28202
0.37631
0.42898
0.43536
0.43598
Regression
Producing
District 8
PC
MM
TON
AS
BT
MO
Alma Seam
SU
1.00000
0.33760
-0.18080
-0.20457
0.18289
ion Equation
Alma Seam
0.11398
0.16518
0.16564
Equation Var
Alma Seam
AS
BT
TON
Mine #2
SU
1.00000
0.37465
-0.27160
-0.12180
0.14291
-0.43876
R2
Mine #2
0.19251
0.24918
0.27194
0.28839
0.28895
iables
Mine #2
MM
AS
MO
TON
BT
SU = sulfur (as received)
AS = ash (as received)
BT = heat content (Btu)
MO = moisture content
TON = lot-size
MM = mining method
PC = preparation code
7-13
-------
importance of the explanatory variables. The signs of the
regression coefficients for ash (AS), heat content (BT), and
lot-size (TON) are generally positive in the regression equa-
tions for Producing District 8, while they are negative in
the Producing District 10 regression equations. Given the
weak relationship which exists between sulfur content and
these explanatory variables and the possible existence of
multicollinearity, differences in the -signs of the regression
coefficients from one Producing District to another or from
one equation to another within the same Producing District
are not toally unexpected.
Inconsistencies also occur in the order of importance
of each explanatory variable in these equations. No single
variable appears in each equation as-one of the three most
important variables, as measured by each variable's contribu-
tion to the explained variance. However, every variable
appears at least once as one of the three most important
variables. The differences in the importance of each explana-
tory variable is another indication that the relationship
between the sulfur content of coal and these other coal
properties is weak and inconsistent and that these other
coal properties are poor estimators of the sulfur content.
7.3.3 Producing District 4
For Producing District 4 the regression analysis was
performed for the entire Producing District with no dis-
aggregation to the seam or mine levels. The regression
equation for Producing District 4 is shown below:
7-14
-------
Producing District 4:
SU = 9.53445 - 0.13534 MO - 0.00029 BT
(23.62) (9.67)
-1.14779 PC + 0.00008 TON - 0.23392 MM
(17.92) (8.00) (6.02)
+ 0.00926 AS
(2.93)
R2 = 0.24987 F = 352.15370 S.E. = 0.76893
where: SU = sulfur content
AS = ash content
BT = heat content (Btu)
MO = moisture content
TON = lot-size
MM = mining method
PC = preparation code (raw or washed)
T-statistic values for each regression coefficient are in
parentheses. Each regression coefficient is statistically
significant at the 95 percent confidence level. The regres-
sion equation R2 is low, indicating that the regression
equation accounts for only about 25 percent of the total
variance, while the standard error is quite large at
approximately 0.77.
Variable rankings and the respective regression equation
R2 for Producing District 4 are shown in the following table.
TABLE 30
Variable Explanatory Regression
Order Variable Equation R2
1 MO 0.10316
2 BT 0.19615
3 PC 0.23768
4 TON ^ 0.24934
5 ' MM 0.24886
6 AS 0.24987
7-15
-------
The three most important variables, moisture content (MO),
heat content (BT), and the preparation code (PC) have never
appeared together as the three most important variables in
any regression equation thus far. Heat content and prepara-
tion code variables have frequently been one of the three
most important explanatory variables in other equations.
However, the moisture content variable appeared as one of
the three most important variables in only one other equation
Mine #2 in Producing District 8.
7.3.4 Mine #3
The data for Mine #3 in Campbell County, Wyoming con-
sists of core sample analyses and "as shipped" analyses with
the core analyses generally having a-higher average sulfur
content than the "as shipped" analyses. Regression equations
for the core sample data and the "as shipped" data are shown
below:
Mine #3:
(Core data)
SU = -0.88979 - 0.02832 MO + 0.00024 BT
(14.23) (12.00)
+ 0.02889 AS
(8.60)
R2 = 0.23213 F = 154.47654 S.E. = 0.06152
Mine #3:
("as shipped" data)
SU = -1.82902 + 0.00029 BT + 0.02451 AS - 0.0000 TON
(24.00) (8.17) (undefined)*
+0.00162 MO
(3.00)
*Note: T-statistics were estimated by dividing the regres-
sion coefficient of each variable by its corres-
ponding standard error. By definition the t-statistic
is undefined when the standard error is zero.
7-16
-------
R2 = 0.24992 F = 152.85478 S.E. = 0.04323
where: SU = sulfur content
AS = ash content
BT = heat content
MO = moisture content
TON = lot-size
The value of the t-statistic for each regression coeffficent
is in parentheses. Each regression coefficient is signifi-
cant at the 95 percent confidence level. Sulfur correlation
coefficients, regression equation R2, and variable rankings
by order of importance are shown in Table 31.
Similar to the other regression equations, the Mine #3
regression equations have low R2 values, indicating very
weak explanatory or predictive power. The regression aqua-
tion based on "as shipped" analyses does have slightly more
explanatory power than the regression based on core analyses.
Although the Mine #3 regression equations have very weak
explanatory power, the low R2 values are not accompanied by
the high standard errors which have generally occurred in
the other regression equations. The standard errors in the
Mine 13 equations are the lowest for any equation. Another
interesting inconsistency is the ranking of the moisture
content variable (MO) in each equation. Moisture is the
most important explanatory variable in the core sample
regression equation and the least important in the "as
shipped" sample regression equation. The sign of the mois-
ture regression coefficient is different in each equation,
being negative in the core sample equation and positive in
the "as shipped" sample equation. The existence of multi-
collinearity between the explanatory variables may be at
least partially responsible for these inconsistencies.
7-17
-------
TABLE 31
SULFUR REGRESSION ANALYSIS SUMMARY FOR MINE #3
(MINE #3, CAMPBELL COUNTY, WYOMING)
Variable
SU
AS
BT
MO
TON
MM
PC
Sulfur Correlation
Coefficients
Core
Samples
SU
1.00000
-0.13969
0.30040
-0.39697
0.00000
"As Shipped"
Samples
SU
1.00000
-0.12402
0.45812
-0.07688
-0.17122
Number of
Variables
1
2
3
4
Regression Equation
Core
Samples
0.15769
0.19516
0.23213
"As Shipped"
Samples
0.20987
0.23202
0.24524
0.24992
Variable
Rank
1
2
3
4
Regression Equation Variables
Core "As Shipped"
Samples Samples
MO
BT
AS
BT
AS
TON
MO
SU = sulfur content
AS = ash content
BT = heat content (Btu)
MO = moisture content
TON = lot-size
MM = mining method
PC = preparation code
7-18
-------
7.3.5 Summary of Average Sulfur Content Regression Results
Tables 32 and 33 show the sign of each regression
coefficient and the order of importance of each explanatory
variable for each regression equation. In general there is
very little consistency in the sign of the regression coeffi-
cients either between each coal Producing District or within
each coal Producing District. No explanatory variable has
the same sign in each equation. The regression coefficients
for the ash content variables (AS) and the moisture content
variables (MO) show the most consistency. In seven of the
nine equations the regression coefficient of the ash content
is positive while the regression coefficient of the moisture
content is negative in six of the eight equations in which
the moisture variable appears. The other variable regression
coefficients are generally about equally divided between
positive and negative values. Similar inconsistencies also
occur in the order of importance of each variable. No single
variable appears among the three most important variables in
each quation. The variable that appears most often among
the three most important explanatory variables is heat con-
tent (BT), but this variable appears only six times.
Considering these inconsistencies in the sign of the
regression coefficients and in the rank of the explanatory
variables along with the generally low R2 values and high
standard errors leads to the conclusion that the sulfur con-
tent of coals cannot be explained or predicted with any
degree of confidence by regression equations in which the
physical and chemical properties of coal are the explanatory
variables. Based on these regression results, the sulfur
content appears to be largely uncorrelated with the other
physical and chemical coal properties. If the sulfur
7-19
-------
J
I
o
TABLE 32
SIGN OF THE REGRESSION COEFFICIENTS
Producing District 10
Producing District 8
Mine #3
Explanatory Dist.
Variable 10
AS
BT
MO
TON
MM +
SM
PC +
Illinois Mine Dist. Alma (Blue
#6 Seam #1 8 Gem) Seam
+ + +
+ +
+ NA
+ NA
+ NA + NA
f - NA NA
+ NA
Mine Dist.
#2 4
+
+
-
+ +
-
NA NA
NA
Core
Analyses
+
+
-
NA
NA
NA
NA
"As Shipped"
Analyses
+
+
+
NA
NA
NA
NA
NA = not applicable.
AS = ash content
BT = heat content
MO = moisture content
TON = lot-size
MM = mining method
SM = sampling method
PC = preparation code
-------
Producing District 10
TABLE 33
EXPLANATORY VARIABLE
Producing District 8
Mine #3
Variable
Rank
1
2
3
4
5
6
7
Dist.
10
PC
BT
SM
TON
MM
MO
AS
Illinois
#6 Seam
MM
PC
BT
SM
AS
MO
TON
Mine
#1
TON
AS
PC
SM
MO
BT
MM
Dist.
8
PC
MM
TOM
AS
BT
MO
Alma (Blue
Gem) Seam
AS
BT
TON
Mine
#2
MM
AS
MO
TON
BT
Dist.
4
MO
BT
PC
TON
MM
AS
Core
Analyses
MO
BT
AS
"As Shipped"
Analyses
BT
AS
TON
MO
to
AS = ash content
BT = heat content
MO = moisture content
TON = lot-size
MM = mining method
SM = sampling method
PC = preparation code
-------
content can in fact be explained or predicted from other
variables, these other variables are not the coal character-
istics generally determined from coal analyses. These other
variables may be geologic variables as hypothesized by Gomez,
Donaven, and Hazen and others. However, coal analyses which
contain geologic data are very limited at this time and are
not contained in the coal database assembled for this study.
7.4 Sulfur Variability Regression Results
Since the primary objective of this study was to examine
the variability of sulfur in coal, regression analysis was
used to examine the relationship between the sulfur varia-
bility and the other physical and chemical coal properties
contained in the coal database. Two measures of sulfur
variability are used in this regression analysis. One is
the relative standard deviation of sulfur and the other is
the variance (02) of sulfur. All regressions were performed
for the Illinois #6 seam.
7.4.1 Relative Standard Deviation Regressions
Two different regression analyses were examined for the
RSD of sulfur. First, the RSD was considered to be a linear
function of all coal properties except sulfur content (SU).
In the second regression analysis sulfur content was included
as an explanatory variable. The regression equations are
shown below:
Illinois #6 Seam:
(excluding sulfur)
RSD = -37.73231 + 5.65541 SM + 3.05614 AS
(2.80) (2.18)
- 0.00100 TON + 6.43196 MM
(1.16) (0.98)
R2 = 0.60573 F = 3.84077 S.E. = 8.92782
7-22
-------
Illinois 16 Seam:
(including sulfur)
RSD = -54.92410 - 13.88940 SU + 4.63748 AS
(4.54) (3.62)
-0.00106 TON - 0.00037 BT + 3.92511 MO
(1.03) (0.56) (2.64)
+ 5.87215 SM + 5.40324 MM - 2.85861 PC
(2.40) (1.19) (0.51)
R2 = 0.91745 F = 8.33555 S.E. = 5.27383
where: RSD = relative standard deviation of sulfur
AS = ash content
BT = heat content
MO = moisture content
TON = lot-size
MM = mining method
SM = sampling method
PC = preparation code
Regression coefficient t-statistics are in parentheses.
When the sulfur content was excluded, the regression coeffi-
cients for the lot-size (TON) and mining method (MM) vetria-
bles were insignificant at the 95 percent confidence level.
From Table 34 it can be seen that the regression coefficients
of the explanatory variables become insignificant as more
explanatory variables are included. Multicollinearity between
the independent variables may be the cause of the change in
the significance of the regression coefficients as more
explanatory variables are added. The addition of explanatory
variables does not have a significant effect on the regres-
sion equation R2 or standard errors. If the only explanatory
variables considered are sampling method (SM) and ash content
(AS), the equation R2 is about 0.52 and the standard error
is about 8.95 compared to an R2 of approximately 0.63 and a
standard error of 10.28 when all variables are included.
7-23
-------
TABLE 34
REGRESSION RESULTS SUMMARY FOR THE ILLINOIS #6 SEAM
Regression Analysis Excluding Sulfur
Variable
Order
1
2
3
4
5
6
7
Explanatory
Variable
SM
AS
TON
MM
MO
BT
PC
Regression Analysis
Regression
Equation R2
0.30198
0.52488
0.56806
0.60573
0.62701
0.63344
0.63399
Including Sulfur
Standard
Errors
10.41861
8.94659
8.90969
8.92782
9.15319
9.62436
10.28122
Content as an Explanatory Variable
Variable
Orcer
1
2
3
4
5
6
7
8
Explanatory
Variable
SU
AS
TON
BT
MO
SM
MM
PC
Regression
Equation R2
0.48643
0.68482
0.76348
0.79831
0.81879
0.89794
0.91392
0.91745
Standard
Errors
8.93667
7.28679
6.59301
6.38549
6.37995
5.07845
4.98589
5.27383
SU = sulfur content
AS = ash content
BT = heat content
MO = moisture content
TON = lot-size
MM = mining method
SM = sampling method
PC = preparation code
7-24
-------
The addition of sulfur content as an explanatory vari-
able improves the regression equation R^ almost 50 percent
to nearly 0.92 and reduces the standard error by about the
same percent to 5.28. Such an improvement in the regression
equation is expected since a fairly strong relationship
should exist between the RSD of sulfur and the sulfur content
because the average sulfur content is a component of the RSD
(RSD = o/x) . However, the inclusion of the sulfur content
variable presents problems of a statistical nature. First,
the existence of multicollinearity may be increased because
the previous regression analyses showed that sulfur content
has some relationship, although weak, to the other physical
and chemical coal properties. As a consequence of this
multicollinearity the t-statistics and the signs of the
regression coefficients may be unreliable, and the regression
coefficients may be highly sensitive to the particular sets
of data and the number of observations. Inferences about
the degree or seriousness of the multicollinearity can be
obtained by using the Farrar-Glauber test for multicol-
linearity.I/
Another problem created by the inclusion of sulfur con-
tent as an explanatory variable is the possible existence of
heteroskedasticity. Heteroskedasticity exists when the
variance of the disturbance term is not constant. Since the
RSD of sulfur is inversely related to the average sulfur
content, the variance of the disturbance term may not be
constant for all observations. Heteroskedasticity results
in the least squares estimators not having the smallest
variance and, therefore, provides least squares estimators
of the regression coefficients which are inefficient not
the best linear unbiased estimates. One method of testing
I/ Johnston, Econometric Methods, 2nd Ed., McGraw-Hill Book
Company, 1972, pp. 163-164.
7-25
-------
for the existence of heteroskedasticity is a nonparametrie
test proposed by Goldfield and Quant.!/ Although the possi-
ble presence of multicollinearity and heteroskedasticity
make the actual contribution of each explanatory variable
unreliable, the high R2 value indicates that the regression
equation containing sulfur content as an explanatory varia-
ble has strong predictive powers accounting for almost 92
percent of the total variance. However, with a standard
error of about 5.3 the predicted RSD would be in the range
of RSD + 10.3 at the 95 percent confidence level. This is a
large range since the mean RSD for the Illinois #6 seam is
about 13.7 percent.2./
7.4.2 Variance (o2) Regressions
In an attempt to reduce the possible existence of hetero-
skedasticity a regression analysis was performed using the
variance of sulfur as the dependent variable. The regression
equation for the Illinois #6 Seam is shown below:
Illinois 16 Seam:
02 = -0.89731 + 0.09950 SM + 0.06120 AS
(3.21) 3.77)
- 0.07317 SU + 0.04069 MO - 0.00002 TON
(1.88) (2.16) (2.00)
- 0.00001 BT - 0.0160 PC + 0.04493 MM
(1.00) (0.86) (0.78)
I/ Johnston, Econometric Methods, 2nd Ed., McGraw-Hill Book
Company, 1972, pp. 218-219.
2/ For example, based on a 1.2 Ibs S02/MMBtu standard,
Table 10 in Chapter 2 shows that the mean level of emissions
required to meet the standard would range from 1.06 to 0.73
Ibs S02/MMBtu for respective RSD's of 5 and 25 percent,
assuming a normal distribution.
7-26
-------
R2 = 0.08834 F = 5.68211 S.E. = 0.06696
where: o2 = variance of sulfur
SU = sulfur content
AS = ash content
BT = heat content
MO = moisture content
TON = lot-size
MM = mining method
SM = sampling method
PC = preparation code
Values of the t-statistic for each regression coefficient
are in parentheses. Many of the regression coefficients are
insignificant at the 95 percent confidence level with only
the sampling method (SM), ash content (AS), and moisture
content (MO) variables having significant t-statistics.
However, because of the possible existence of multicolline-
arity, especially with the inclusion of sulfur content (SU)
as an explanatory variable, the regression coefficients are
suspect as to their values and significance. Removal of the
insignificant variables, as indicated in the following table,
does not cause major changes in the R2 values or the standard
errors. The R2 is about 0.73 and the standard error about
0.072 when only sampling method (SM) and ash content (AS)
TABLE 35
Variable Explanatory Regression Standard
Order Variable Equation R2 Error
1 SM 0.31056 0.11062
2 AS 0.73325 0.07162
3 SU 0.74853 0.07263
4 MO 0.83410 0.06187
5 TON 0.85464 0.06104
6 BT 0.86562 0.06225
7 PC 0.87161 0.06505
8 MM 0.88340 0.06696
7-27
-------
are the explanatory variables, compared to an R2 of 0.88 and
a standard error of 0.067 when all variables are included.
Although the contribution of each explanatory variable is
suspect, the high R2 values indicate that this regression
equation has good predictive power because it accounts for
over 88 percent of the total variance. However, given a
standard error of about 0.067, the predicted variance would
be between a2 + 0.13 at the 95 percent confidence level.
This is a very large range considering that the mean variance
in the coal samples for the Illinois #6 Seam is approximately
0.13.
7.4.3 Summary of Sulfur Variability Regression Results
Table 36 shows the sign of each regression coefficient
and the order of importance of each explanatory variable.
Much more consistency appears in the signs of the regression
coefficients in these equations than in the equations where
sulfur content was the dependent variable. The signs of the
regression coefficients are the same in each equation for
six of the eight explanatory variables. Only the heat con-
tent (BT) and preparation code (PC) regression coefficients
do not have the same sign in each equation. More consistency
also occurs in the relative importance of the explanatory
variables. Ash content (AS) is the second most important
variable in each equation. Three other variables sampling
method (SM), lot-size (TON), and sulfur content (SU)
appear as one of the three most important variables in two
of the three equations. The frequent occurrence of the lot-
size variable among the three most important explanatory
variables and the negative sign of the lot-size regression
coefficient lends support to the hypothesis that the relative
standard deviation of sulfur tends to decrease as the lot-
size increases.
7-28
-------
TABLE 36
REGRESSION EQUATION SUMMARIES FOR SULFUR
VARIANCE IN ILLINOIS #6 SEAM
Sign of the Regression Coefficients
Relative Standard Deviation
(RSD) Equations
Explanatory Excluding Including Variance (o2)
Variable Sulfur Content Sulfur Content Equation
SU NA
AS 4- + +
BT +
MO + + +
TON -
MM + 4- +
SM + + +
PC -
Explanatory Variables
Relative Standard Deviation
(RSD) Equations
Variable Excluding Including Variance (o2)
Rank Sulfur Content Sulfur Content Equation
1 SM SU SM
2 AS AS AS
3 TON TON SU
4 MM BT MO
5 MO MO TON
6 BT SM BT
7 PC MM PC
8 PC MM
Note: The statistical significance of the variables is not
considered.
NA = not applicable.
SU = sulfur content TON = lot-size
AS - ash content MM = mining method
BT = heat content SM = sampling method
MO = moisture content PC = prepartion code
7-29
-------
Although the problems of multicollinearity and hetero-
skedasticity are more likely to occur in the regression
analysis of the relative standard deviation of sulfur and
the variance of sulfur, the R2 values are much higher and
the standard errors are much lower in the RSD and o2 regres-
sion equations than in the sulfur content (SU) regression
equations. The RSD and o2 regression equations account for
approximately 90 percent of the total variance when all
explanatory variables are included and account for about
70 percent of the total variance when only the two most
important explanatory variables are included. Although the
presence of multicollinearity and heteroskedasticity make
the individual regression coefficients unreliable and suspect
for explanatory purposes, the equations as a whole have high
predictive power as measured by the equation R2. However,
the large ranges in which the predicted RSD and o2 would
fall at the 95 percent confidence level greatly reduce the
usefulness at these equations for predicting either the RSD
or the o2 of sulfur.
7-30
-------
8.0 Conclusions and Recommendations
8.1 Conclusions
1. The coal data collected for this report was analyzed by
sorting by lot-size intervals and comparing the RSD of
Ibs S/MMBtu versus lot-size. The data were analyzed by
U.S. Bureau of Mines Producing District, coal seam, and
individual mine on a raw and washed basis. For individual
mines, which generally had data available for only a few
lot-size intervals, the results exhibited no consistent
relationship between RSD and lot-size. As the data were
aggregated to seam and Producing District, the results
were still inconsistent, but in the majority of the
cases the results exhibited an increase in the RSD of
Ibs S/MMBtu for successively smaller lot-sizes. These
results provide limited support for an inverse relation-
ship between RSD and lot-size.
2. Various regression analyses of the coal data provided
limited support for an inverse relationship between the
RSD of Ibs S/MMBtu and lot-size.
3. The results of a simulation model, which was developed
to examine coal sulfur variability, indicated that
theoretically, coal sulfur variability should decrease
with increasing lot-sizes.
4. An analysis of stack monitoring data from four electric
generating units indicated that significant reductions
in the relative variability of sulfur dioxide emissions
can be achieved by using longer averaging intervals.
It follows from this analysis that the smaller the
8-1
-------
amount of coal burned per unit time, the more difficult
it is to comply with an emission standard if the averag-
ing basis is time. In addition, these results support
the existence of an inverse relationship between the
RSD of Ibs S/MMBtu and lot-size, since increasing the
averaging interval is equivalent to increasing the
volume or lot-size of coal burned.
5. Sulfur dioxide emission regulations that require a
probability of a very low number of days of excess
emissions per year (for example, one or two days per
year), require extremely high probabilities of compli-
ance on the individual days of the year and substan-
tially reduce the average level of emissions required
for compliance.
6. Although flue gas desulfurization (FGD) decreases the
mean level of sulfur emissions, the limited data analyzed
showed that the relative variability of the emissions
increases. In one case examined the relative variability
of outlet emissions was more than thirteen times greater
than the inlet emissions.
7. Composite coal seam or Producing District data cannot
be used to accurately predict the variability of Ibs
S/MMBtu for individual mines within the coal seams or
Producing Districts. Both seam and Producing District
data provide biased estimates which consistently over-
estimate the RSD of Ibs S/MMBtu for individual mines.
Even if a scaling factor were used, the composite esti-
mates would not reasonably predict mine variabilities.
8. The overall frequency distributions of Ibs S/MMBtu and
coal sulfur contents (weight percent) are skewed to the
8-2
-------
right and are best represented by the inverted gamma
distribution which appeared to be slightly superior to
the lognormal distribution, and definitely superior to
the normal distribution.
9. In the extreme right tail of the frequency distribution
for Ibs S/MMBtu (top 1.5 percent of the distribution)
the data provided ambiguous results with respect to the
best choice between the lognormal and the inverted
gamma distributions.
10. The overall frequency distributions of coal heat contents
(Btu/lb) appeared to be reasonably symmetrical and are
closely approximated by the normal distribution.
11. The normal, lognormal, and inverted gamma distributions
provided similar estimates for the mean Ibs SC>2/MMBtu
required for compliance with stringent sulfur dioxide
emmission regulations. However, these distributions
provided significantly different estimates under the
assumptions of less stringent emission limits and coals
with large RSD's for Ibs S/MMBtu.
12. Comparisons of raw and washed coals on a mine, seam,
and Producing District basis consistently indicated
lower average Ibs S/MMBtu as well as lower RSD's for
the washed coals.
13. Based on the data analyzed, the type of frequency dis-
tributions for washed coal characteristics are not
significantly different from the type of frequency
distributions for raw coal characteristics.
8-3
-------
14. Within individual mines, no significant differences
were observed in the Ibs S/MMBtu and RSD's for double
screened and single screened coals.
15. Measurement errors in ASTM sampling and analysis pro-
cedures resulted in biased estimates which consistently
overestimated the true RSD of Ibs S/MMBtu. The differ-
ences between the measured and true RSD's are most
significant for low measured RSD's. Theoretical calcu-
lations indicated that at a measured RSD of 7 to 8 per-
cent, the true RSD was approximately zero.
16. Coal sulfur variability is a result of many interrelated
factors. Statistical analysis of several of the factors
believed to contribute to sulfur variability failed to
identify any consistent, predictable relationship.
17. Various regression analyses based on mines, seams, and
Producing Districts indicated that neither coal sulfur
contents (weight percent) nor coal sulfur variabilities
can be accurately predicted from the database developed
in this study. The results tend to support the hypo-
thesis that the primary factors affecting coal sulfur
distributions are geologic factors related to the deposi-
tional history of the coal, while chemical and physical
properties of coal are secondary factors influencing
coal sulfur distributions.
18. Discussions with coal companies, Federal agencies, and
research organizations did not reveal the existence of
data which would permit an examination of the relation-
ship between coal sulfur variability and geological
factors or mining techniques. Further, these discussions
did not reveal the existence of reliable data which
8-4
-------
would permit an accurate assessment of the correlation
between stack emissions and coal analyses, or between
raw and washed coals.
19. The various analyses performed in this study identified
no reliable method for coal suppliers, coal consumers,
or air pollution control agencies to predict coal sulfur
variability, which is often critical for compliance
with existing sulfur dioxide regulations. Coal sulfur
variability is especially critical for small coal-fired
boilers subject to regulations that specify short aver-
aging time intervals. The findings of this report sug-
gested that the requirements of many current sulfur
dioxide regulations are not consistent with the state
of knowledge concerning coal sulfur variability.
8.2 Recommendations
1. Additional studies should be performed using more sophis-
ticated models, such as autocorrelative models, which
may yield more useful results than classical statistical
models assuming independence.
2. This study investigated the goodness of fit between the
observed distributions of coal characteristics and the
normal, lognormal, and inverted gamma distributions.
Additional studies should be performed to examine the
goodness of fit for other skewed distributions, especi-
ally in the extreme right tail which becomes increasingly
important when a high probability of not exceeding an
upper limit is required.
3. Geosta-tistical methods, which would take into account
both structure and randomness, should be used to
8-5
-------
investigate the individual processes which influenced
the coal from the time it was deposited to the time it
was burned. These processes include depositional environ-
ment, in situ variability, mining methods, blending,
cleaning, burning, and desulfurization.
4. Controlled experiments, although a major undertaking,
might be performed to obtain the quality of data required
to investigate the various processes which influence
coal sulfur and sulfur dioxide emissions variability.
The limited data currently available are observational
data used for establishing coal prices and monitoring
overall coal quality.
5. A comprehensive model would be useful to assess the
impact on air quality. Inputs to this model would
include parameters for coal characteristics, mining and
handling methods, combustion and control equipment,
meteorological data, and other variables.
6. Alternative sulfur dioxide emission regulations, which
would mitigate the impact of coal sulfur variability
yet achieve the objectives of existing regulations,
should be investigated.
8-6
-------
APPENDIX A
DERIVATION OF EQUATIONS USED FOR THE
DEVELOPMENT OF TABLES 10, 11, AND 12
This technical appendix sets out the formulas and approxi-
mations used to derive Tables 10, 11, and 12 in this study.
TABLE 10, NORMAL DISTRIBUTION
(1) Z = fSSX
where: Smax = Emission standard, Ibs S02/MMBtu
/ix = Required mean, Ibs S02/MMBtu
ox = Standard deviation
(2) RSD =
Let Z = 2.57583, which corresponds to a 0.005 probability
of values higher than Smax.
(3) Therefore: 2.57583 = Sraax ~ ^x
(RSD)
(4) And:
1 + 2.57583 (RSD)
TABLE 11, LOGNORMAL DISTRIBUTION
From Naylor, e_t al. ;!./
(1) Ex = e y 2
(2) Vx = (Ex)2 |_e °y -1J
I/ Naylor, Tl, Balintfy. J., Burdick, D., and Chu, K.,
Computer Simulation Technique, Wiley & Sons, 1968, p. 100
A-l
-------
Let Z = 2.57583, where Z is the standard normal variate
corresponding to a 0.005 probability of higher values.
(3) In Smax = Jiy + Zoy = fly + 2.57583 oy
Given Smax and R = , one can find Ex
(4) ay = >/ln (R2 + 1)
(5) /JLy = In Smax - Z oy = In Smax ~ 2.57583 oy
(6) Ex = e (^Y +
(7) Ex = e ln Sinax - 2-57583 *y + ± o2
r 1 f I CT2 - 2.57583 o,
(8) Ex = SmaJ e 2 y
]
J
TABLE 12, INVERTED GAMMA DISTRIBUTION
For the inverted gamma distribution the probability of
a value being greater than the emission standard can be
estimated by:
(1) Pr (X>Smax) = Pr (X< I f = 2B)
First, determine B to the nearest integer by:
(2) B = + 2
RSD2
Next, determine the X2 (chi-square) value at the 0.005 con
fidence level for f = 2B, then
A-2
-------
(3)
(4)
a = (X2) (Smax)
Next determine MX by
(5) MX = g-S-T-
A-3
-------
APPENDIX B
DERIVATION OF FORMULA FOR TRUE AND MEASURED RSD
Definition of Terms:
Cm = measured sulfur content, Ibs S/MMBtu
Sm = measured sulfur content, percent
Hm = measured heat content, Btu/lb
Ct = true sulfur content, Ibs S/MMBtu
St = true sulfur content, percent
Hf- = true heat content, Btu/lb
ess = error in sulfur measurement due to sampling
esa = error in sulfur measurement due to analysis
ehs = error in heat measurement due to sampling
eha = error in heat measurement due to analysis
<»2 = variance
o = standard deviation
M = mean
cov = covariance
Assume: Sm = St + ess + esa
Hm = Ht + ehs + eha
Mess = Mesa = Pehs = "eha = 0
222 2 «
OG ~ °c. + °e + °e if it is assumed;
Ojji "t ss sa ~~^'^'^^^"^^^^~'^^^^^"~~^
cov (St/ ess) = cov (Str esa) = cov (ess, esa) = 0
922 2
a =or, +-0 =o ifitis assumed:
Hnj Ht ehs eha
cov (Ht, ehs) = cov (Ht> eha) = cov
-------
Then:
/st
VARl=i
\Ht
r^JV2
St
Q
St
Ht 2 cov (St,Ht)
Q
St
Me MH
St Ht
VAR
H
m,
'22
°Sm Hm _ 2 cov (Sm, Hm)
H
m
m
'Pst\2
MHt ]
' /
2
°st
2
2
+ °ess ^
2
st
cov (Sm,
222
°esa + °Ht + °ehs +
U2
Hm)
Ht
2
°eha
c TT
St Ht
Because Ex
= Ex
and'
Ex (PHt) = Ex (MHm)
Assume: Cov (St, Ht)
Then:
= Cov (Sm, Hm)
Ex VAR|=| ~ VARl=± =
fM
-------
where:
Rss = RSD of error in sulfur measurement due to sampling
Rsa = RSD of error in sulfur measurement due to analysis
= RSD of error in heat measurement due to sampling
= RSD of error in heat measurement due to analysis
Dividing by[-^j yields
where RSDm = RSD as measured
= True RSD
B-3
-------
APPENDIX C
COAL SULFUR ANALYSES DATA
DATA BASE TAPE FORMAT
Field
Number
(1)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Field
Position
From - To
(2)
1- 2
3- 3
4- 6
7- 8
9- 13
14- 18
19- 23
24- 28
29- 34
35- 35
36- 36
37- 37
38- 43
44- 49
50- 55
56- 56
57- 57
58- 64
65- 65
66- 69
70- 73
74- 77
78- 81
82- 85
86- 90
91- 94
95- 98
Field Size^/
(3)
XX
X
XXX
XX
XXXXX
xxxxx
XXXXX
xxxxx
xxxxxx
X
X
X
xxxxxx
xxxxxx
xxxxxx
X
X
xxxxxxx
X
xx.xx
xx.xx
xx.xx
xx.xx
xx.xx
xxxxx
xxxx
xx.xx
Field Descriptions
(4)
FIPS State Code-'
State Modifier-
FIPS County Code
USBM Producing District
Company Code
Supplier Code
Mine Code Number
47
Town Code Number-
Bed Code Number-
Rank of Coal
Mining Method
8/
Preparation Code
9/
Date of Sample
107
Sequence Number
Core Hole Number
117
Method of Sampling
Type of Sample
Tonnage Sampled
Flag for Estimated Tonnag
Moisture Content, Percent
Volatile Matter, Percent,
Fixed Carbon, Percent, as
Ash Content, Percent, as
Sulfur Content, Percent,
137
-k-*/
, as Received
as Received
Received
Received
as Received
Heat Content, Btu per Pound, as Received
Ash Fusion Temperature , °
F.
SO Emissions, Ibs SO /MMBtu
C-l
-------
Field
Number
(1)
28
29
30
31
32
33
34
Field
Position
From - To
(2)
99-102
103-106
107-110
111-114
115-119
120-123
124-125
Field Size
(3)
XX.XX
XX.XX
XX.XX
XX.XX
xxxxx
XX.XX
XX
Field Descriptions
(4)
Volatile Matter, Percent, Dry
Fixed Carbon, Percent, Dry
Ash Content, Percent, Dry
Sulfur Content, Percent, Dry
Heat Content, Btu per Pound, Dry
SO Emissions, As Reported, Ibs
Size Codeli/
SO /MMBtu /
I/ A = implied decimal.
2/ U.S. Department of Commerce, National Bureau of Standards, Federal Informa-
tion Processing Standards Publication 601, November 1, 1968 (34 pages).
3/ For Kentucky 1 = Eastern, 2 = Western; For Pennsylvania 1 = Anthracite,
2 = Bituminous; all others = 0.
4_/ CODES beginning with 0 are equivalent to four-digit codes used by USBM
in attachment to analytical Data Tape documentation. Codes beginning with
1 represent towns not listed by USBM.
5_/ First Digit: 0 means an equivalent USBM code exists, 1 means no USBM code
exists and a special code was assigned.
Digits 2-4: equivalent to USBM bed codes.
Digits 5-6: identifiers for local bed names used for the same bed.
6_/ 1 = lignite, 2 = subbituminous, 3 = bituminous, 4 = anthracite.
7/ 1 = underground, 2 = surface, 3 = surface-auger, 4 = underground-surface,
9 = unknown.
8/ 1 = Raw, 2 = washed or cleaned, 3 = partially washed or cleaned, 9 = unknown.
9_/ Year-month-day.
10/ Railroad car, train, barge, or sample number.
ll/ 1 = automatic sampleASTM; 2 = hand sampleASTM; 3 = automatic sample
non-ASTM; 4 = hand samplenon-ASTM; 9 = unknown.
12/ 1 = core sample, 2 = as mined, 3 = as shipped, 4 = as delivered, 5 = as burned,,
9 = unknown.
13/ 0 = measured tonnage, 1 = estimated based on number of railroad cars, barges, etc
14/ Calculated based on sulfur and heat contents and assuming that 95 percent by
weight of the sulfur present in the coal is released as sulfur dioxide.
15/ Calculated and reported by coal consuming company.
16/ Bureau of Mines data only.
C-2
-------
APPENDIX D
STACK MONITORING DATA
DATA BASE TAPE FORMAT
Field
Number
(1)
1
2
3
4
5
6
7
3
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Field
Position
From - To
(2)
1- 5
6- 8
9- 12
13- 14
15- 20
21- 24
25- 28
29- 32
33- 37
38- 40
41- 43
44- 47
48- 52
53- 55
56- 58
59- 62
63- 65
66- 71
72- 77
78- 82
83- 88
89- 91
92- 94
95- 97
98-100
101-103
104-106
Field Size-/
(3)
XXXXX
XXX
xxxx
XX
xxxxxx
xxxx
xxxx
xxx.x
xxxx.x
XXAX
XXAX
XX. XX
xxxx.x
XXAX
XXAX
XX..XX
XXAX
XXXX..XX
xxxx.xx
x.xxxx
xxxx.xx
XAXX
XAXX
XAXX
XXAX
XXAX
XXAX
Field Descriptions
(4)
Company Code
Generating Unit Number
Unit Size (MW)
FGD Unit Nmnber
Date*/
Time
Gross Load (MWHr. )
Coal Flow (Thousand Pounds/Time Period)
Inlet SO (ppm Wet)
Inlet 0 (percent)
Inlet HO (percent)
Inlet SO (Ibs/MMBtu)
Outlet SO (ppm Wet)
Outlet O (percent)
Outlet HO (percent)
Outlet SO2 (Ibs/MMBtu)
Efficiency-
Outlet SO ; 1 Hr Avg. (Kg./Hr)
Outlet SO ; 3 Hr Avg. (Kg./Hr)
Outlet Wt. Sulfur; 3 Hr Running Avg. (percenl
Outlet Wt. Sulfur; 3 Hr Running Avg. (Ibs )
Outlet SO2; 1 Hr Avg. (Ibs/MMBtu)
Outlet SO ; 3 Hr Avg. (Ibs/MMBtu)
Outlet SO2; 24 Hr AVg. (Ibs/MMBtu)
Outlet O ; 1 Hr Avg. (percent)
Outlet O ; 3 Hr Avg. (percent)
.Outlet O2; 24 Hr Avg. (percent)
D-l
-------
Field
Number
(1)
28
29
30
31
Field
Position
From - To Field Size
(2) (3)
107-111 XXXXX
112-116 XXXXX
117-121 XXXXX
122-126 XXXXX
Field Descriptions
(4)
Coal Source No . 1 , Mine Code Number
Coal Source No . 2 , Mine Code Number
Coal Source No. 3, Mine Code Number
Coal Source No . 4 , Mine Code Number
I/ A = implied decimal.
2/ Year-Month-Day
3/ Efficiency = Inlet SO - Outlet SO
Inlet SO.,
D-2
-------
APPENDIX E
INDEX TO MINE LOCATIONS AND SEAMS PRODUCED
Mine
Code
(1)
D-LCIIU
Of
Mines
(2)
State
(3)
County
(4) '
Reference
Code
(5)
Name
(6)
Composite
Of Seams
(7)
10001
10002
10003
10004
10005
10006
10007
10008
10009
10010
10011
10012
10013
10014
10015
10016
10017
10018
10019
10020
10021
10022
10023
10024
10025
10026
10027
10028
10029
10030
10031
10032
(10006
\10007
State
(3)
KY E
KY E
KY E
IL
MT
WY
WY
WY
WY
KY W
KS
MO
MO
MO
MO
MO
KY E
MD
KY W
KY W
KY W
KY W
KY E
WV
IL
KY E
KY E
KY E
KY E
WV
OH
WV
TN
County
(4) '
Knott
McCreary
(Clay
^Laurel
Peoria
Rosebud
Campbell
Campbell
Campbell
Campbell
Muhlenburg
Crawford
Howard
Audrain
Howard
Randolph
Randolph
Pike
Allegany
Webster
Hopkins
Union
Union
Martin
Logan
Perry
_ Perry
Rockcastle
Clay
Breathitt
Monongalia
Harrison
Grant
J Marion
\ Sequatchie
Roland
010401 Hazard No. 7
015109 Jellico
100001 ( Hazard No. 4
< Horse Creek
(Lily
090002 No. 6
080800 Rosebud
095100 Smith &
100003 Wyodak
100002 jSmith & Roland
]Wyodak
048913 No. 9*
049202 Bevier
049202 Bevier
049008 Mulky
049202 Bevier
049202 Bevier
098700 Bevier and Wheeler
100004 fDorothy
(Thacker
100032 (Pittsburgh
\Sewickley
012704 No. 9
012704 No. 9
048415 No. 11*
012704 No. 9
100005 ( Stockton
No. 5 Block
Clarion
015104 Cedar Grove
090002 No. 5 and 6
100006 Hazard No.
Hazard No.
Hazard No.
Hazard No.
015108 Elkhorn No. 3
021202 Horse Creek
011108 Hazard No. 5A
003604 Pittsburgh
007402 Lower Freeport
007102 Upper Freeport
028601 Sewanee
013520
021202
021203
095100
012105
015119
003608
002902
010301
008402
008701
5
5A
7
9
012103
011108
010401
009601
E-l
-------
Mine
Code
(1)
10033
10034
10035
10036
10037
10038
10039
10040
10041
10042
10043
10044
10045
10046
10047
10048
10049
10050
10051
10052
10053
10054
10055
10056
10057
10058
10059
10060
10061
10062
10063
10064
10065
10066
10067
10068
10069
10070
Blend
Of
Mines
-(2)
State
(3)
KY E
KY E
IL
IL
MT
WY
WY
WY
WY
IL
IL
ND
IL
IL
WV
KY E
KY W
KY E
KY E
KY W
PA
PA
PA
KY W
CO
IL
ND
ND
WY
IA
MO
MO
IA
MO
IA
IA
IA
IA
County
(4)
Johnson
Elliot
Lawrence
Martin
Christian
Perry
Bighorn
Carbon
Carbon
Sheridan
Sweetwater
Macoupin
Fulton
Mercer
St. Clair
Randolph
Grant
Bell
Ohio
Whitley
Knox
Muhlenburg
Armstrong
Indiana
Indiana
Unknown
Routt
Douglas
Mercer
Bowman
Campbell
Monroe
Randolph
Macon
Lucas
Putnam
Mahaska
Mahaska
Mahaska
Marion
S E
Reference
Code
(5)
008402
016200
048408
090002
069802
036555
081700
078300
100007
048408
048905
056100
048408
048408
100008
100009
090000
015703
015703
100010
095203
007102
007102
048913
009900
048408
056901
056400
092600
051700
049202
049202
051700
048414
053004
AM(S) PRODUCE
Name
(6)
No. 5 Block
Mudslip
No. 6
No . 5 and 6
Dietz No. 1
No. 25
Hanna No. 2
Monarch
Deadman Bed
No. 6
No. 5
Scranton
No. 6
No. 6
/ Bakerston
.) Upper Freeport
r Hignite
j Red Spring
( + 9 others
No. 9* and 11*
Blue Gem
Blue Gem
, No. 11
j No. 12
| No. 9
Lower & Upper Freeport
Upper Freeport
Upper Freeport
No. 9*
Fish Creek
No. 6
Beulah-Zap
Harmon
Wyodak-Anderson
Lucas County No. 5
Bevier
Bevier
Lucas County No. 5
Lexington
Lower Ford
Unknown
Unknown
Unknown
D
Composite
Of Seams
(7)
006305
007102
012601
010406
-
048415
048306
048913
E-2
-------
10071
10072
10073
10074
10075
10076
10077
10078
10079
10080
10081
10082
10083
10084
10085
10086
10087
10088
10089
10090
(10072
(10073
AZ Navajo
KY E Perry
KY E Floyd
KY E
PA
PA
WV
WV
PA
PA
Perry
Floyd
KY W
WY
WY
CO
CO
PA
Henderson
Campbell
Carbon
Weld
Jackson
Clearfield
048913
095100
100017
076800
004900
100015
Clearfield
Armstrong
Marion
Marion
CJearfield
Fayette
KY E.
CO
CO
CO
Martin
Routt
Moffat
Mo f fat
016811
075000
076900
075701
Reference
Code Name
(5) (6)
100011 Green
Red
Blue
100012 Hazard No. 5 A
100013
Hazard No. 7
and others
Elkhorn No. 1
Elkhorn No. 2
Fire Clay
Van Lear
100014 Hazard No. 5A
Hazard No. 7
Elkhorn No . 1
Elkhorn No. 2
Fire Clay
Van Lear
and others
048913 No. 9*
095100 Smith & Roland
100017 ( No. 80
| No. 82
076800 Laramie No. 3
004900 Sudduth
100015
100015
Lower Kittanning
Middle Kittanning
Upper Kittanning
Lower Freeport
Upper Freeport
Lower Kittanning
Middle Kittanning
Upper Kittanning
Lower Freeport
Upper Freeport
100016 ( Lower Kittanning
< Middle Kittanning
(upper Kittanning
003604 Pittsburgh
003604 Pittsburgh
100016 /Lower Kittanning
/ Middle Kittanning
f Upper Kittanning
100016 /Lower Freeport
/ Upper Freeport
( Wayne sburg
016811 Pond Creek
075000 Wadge
076900 Collom
075701 F
Composite
Of Seams
(7)
050100
050200
050300
011108
010401
-
015704
015402
013504
015126
011108
010401
015704
015402
013504
015126
-
039900
040100
008419
008002
007603
007402
007102
008419
008002
007603
007402
007102
008419
008002
007603
008419
008002
007603
007402
007102
002302
E-3
-------
SEAM
PRODUCED
Biena
Mine Of
Code Mines
(1) (2)
10091
10092
10093
10094
10095
10096
10097
10098
10099
10100
10101
10102
10103
10104
10105
10106
10107
10108
10109
10110
10111
10112
10113
10114
10115
10116
10117
10118
10119
10120
10121
10122
10123
10124
10125
10126
10127
10128
State
(3)
CO
IL
IL
IL
IL
KY W.
IL
NM
WY
IN
IL
IL
IL
IL
IL
CO
UT
VA
IL
KY E
KY E
KY E
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
County
(4)
Las Animas
Perry
Perry
Jackson
Macoupin
Ohio
Perry
McKinley
Carbon
Pike
. Franklin
Franklin
Jefferson
Jefferson
St. Clair
Mo f fat
Carbon
Buchanan
Randolph
Unknown
Perry
Harlan
Muskingum
Muskingum
Coshocton
. Coshocton
Coshocton
( Guernsey
( Belmont
Perry
Coshocton
Tuscarawas
Tuscarawas
Coshocton
Coshocton
Tuscarawas
Coshocton
/Jackson
(Vinton
Tuscarawas
Reference
Code
(5)
074502
048408
090002
048408
048408
050609
090002
047800
039800
100018
048408
048408
048408
048408
048408
100021
100022
100023
048408
100024
100025
100026
003607
008013
008013
008013
008002
002302
100027
008404
008424
008424
008404
100028
100029
Name
(6)
Robinson
No. 6
No. 5 and 6
No. 6
No. 6
No. 6*
No. 5 and 6
Green
No. 65
1V
\ Lower Millersburg
No. 6
No. 6
No. 6
No. 6
No. 6
Yampa Field
( Hiawatha
\ Wattis
Glamorgan!
Splash Dam
Blair
Hagy
No. 6
Unknown
Unknown
Unknown
Unknown
No. 8
No. 6
No. 6
No. 6
Unknown
Middle Kittanning
Waynesburg
I Lower Kittanning
(Middle Kittanning
Lower Kittanning
No. 5
No. 5
Lower Kittanning
Unknown
Brookville
No. 4A
No. 5
No. 6
( No. 7
\No. 7A
Composite
Of Seams
(7)
048911
048304
084601
023600
018505
021002
017701
019502
008404
008010
009501
008705
008424
008013
007117
007007
E-4
-------
Blend
Of
Mines
(2)
10129
10130
10131
10132
10133
10134
10135
10136
10137
10138
10139
10140
10141
10142
1D143
10144
10145
10146
10147
10148
10149
10150
10151
10152
10153
10154
10155
10156
10157
10158
10159
10160
10161
10162
10163
10164
10165
10166
10167
10168
10169
10170
10171
10172
State
(3)
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
County
(4)
( Harrison
| Belmont
Jefferson
Coshocton
Unknown
Muskingum
Tuscarawas
Unknown
Unknown
Coshocton
Perry
Unknown
Unknown
Unknown
Vinton
Vinton
Hocking
Vinton
Hocking
Perry
Vinton
Hocking
i Hocking
\ Vinton
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
( Guernsey
^ Belmont
Muskingum
Perry
Coshocton
Perry
Tuscarawas
Morgan
Jackson
Unknown
Vinton
Unknown
Unknown
Unknown
Muskingum
Unknown
SEAM (S)
Reference
Code
(5)
PRO
Name
(6)
D U C E D
Composite
Of Seams
(7)
Unknown
003604 Pittsburgh
Unknown
Unknown
008002 Middle Kittanning
008002 Middle Kittanning
Unknown
Unknown
008013 No. 6
Unknown
Unknown
Unknown
Unknown
008701 Clarion
Unknown
007406 No. 6A
Unknown
008013 No. 6
008013 No. 6
088501 Clarion & L. Kittanning
Unknown
088701 Brookville S
M. Kittanning
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
085801 Meigs Creek S
Waynesburg
Unknown
Unkno'.m
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
008002 Middle Kittanning
Unknown
E-5
-------
SEAM(S) PRODUCED
Mine
Code
(1)
10173
10174
10175
Biena
Of
Mines
(2)
10169
10138
10139
10113
10114
10118
10157
10130
10161
10119
10131
10132
10120
10122
10140
10170
10133
10123
10124
10135
10159
10158
10125
10141
10160
10134
10126
10168
10127
10128
10162
10116
10117
10115
10137
10163
10129
10136
10164
State
(3)
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
Reference
County Code
(4) (5)
Perry
Coshocton 092801
Unknown 100031
Perry
Unknown
Muskingum
Muskingum
/ Guernsey
| Belmont
Unknown
Jefferson
' Coshocton
Perry
Coshocton
Unknown
Coshocton
Tuscarawas
Unknown
Unknown
Muskingum
Coshocton
Coshocton
Unknown
Muskingum
i Guernsey
\ Belmont
Tuscarawas
Unknown
Perry
Tuscarawas
Coshocton
Unknown
( Jackson
( Vinton
Tuscarawas /
1
Perry
Coshocton
Coshocton
Coshocton
Coshocton
Tuscarawas
/ Harrison
\ Belmont
Unknown
Morgan
Composite
Name Of Seams
(6) (7)
Unknown
L. & M. Kittanning
Unknown
Unknown
Unknown
Unknown
No. 8 003607
Unknown
Unknown
Pittsburgh 003604
Unknown
Middle Kittanning 008002
Unknown
Unknown
Waynesburg 002302
Lower Kittanning 008404
Unknown
Unknown
Middle Kittanning 008002
No. 5 008424
No. 5 008424
Unknown
Unknown
Meigs Creek &
Waynesburg 085801
Lower Kittanning 008404
Unknown
Unknown
Middle Kittanning 008002
Unknown
Unknown
Brookville 009501
No. 4A 008705
No. 5 008424
No. 6 008013
No. 7 007117
No. 7A 007007
Unknown
No. 6 008013
No. 6 008013
No. 8 003607
No. 6
Unknown
Unknown
Unknown
Unknown
E-6
-------
10176
10177
Blend
Of
Mines
(2)
10113
10142
10157
10143
10144
10119
10156
10145
10159
10155
10152
10160
10171
10146
10151
10127
10154
10148
10153
10157
10142
10143
10144
10174
10165
10119
10173
10149
10145
10167
10172
10160
10146
10150
10147
10166
10148
State
(3)
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
County
(4)
Muskingum
Vinton
Unknown
Vinton
Hocking
Perry
Unknown
Vinton
Muskingum
Unknown
Unknown
Perry
Muskingum
Hocking
Unknown
I Jackson
I Vinton
Unknown
Vinton
Unknown
Unknown
Vinton
Vinton
Hocking
Coshocton
Jackson
Perry
Perry
Hocking
Vinton
Vinton
Unknown
Perry
Hocking
( Hocking
\ Vinton
Perry
Unknown
Vinton
SEAM(S) PRODUCED
Reference
Code
(5)
100031
Name
100031
(6)
Unknown
Clarion
Unknown
Unknown
No. 6A
Middle Kittanning
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Middle Kittanning
No. 6
Unknown
Brookville
No. 4A
No. 5
No. 6
Unknown
Clarion S
L. Kittanning
Unknown
Unknown
Clarion
Unknown
No. 6A
L. & M. Kittanning
Unknown
Middle Kittanning
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
No. 6
Brookville &
M. Kittanning
No. 6
Unknown
Clarion fi
L. Kittanning
Composite
Of Seams
(7)
008701
007406
008002
008002
008013
009501
008705
008424
008013
088501
008701
007406
092801
008002
008013
088701
008013
088501
E-7
-------
Mine
Code
(1)
10178
10179
10180
10181
10182
10183
10184
10185
10186
10187
10188
10189
10190
10191
10192
10193
10194
10195
10196
10197
10198
10199
10200
10201
10202
10203
10204
10205
10206
10207
10208
10209
10210
10211
10212
Blend
of
Mines State
(2) (3)
MT
ND
KY E
KY W
IN
IN
KY W
IN
IL
IN
IL
KY E
OH
WV
MT
MT
MT
MT
MT
MT
MT
MT
MT
MT
PA
PA
WV
PA
PA
PA
MD
PA
PA
County
(4)
Big Horn
Oliver
Unknown
Unknown
Unknown
Muhlenburg
Warrick
Clay
Muhlenburg
Sullivan
Wabash
Warrick
Fulton
Knott
Belmont
Kanawha
Wibaux
Rich land
Rich land
Powder River
Custer
Custer
Custer
Rosebud
Rosebud
Big Horn
Somerset
Jefferson
Harbour
Clearfield
Cambria
Indiana
Somerset
Garrett
Somerset
Armstrong
Reference
Code
(5)
100033
100034
100010
048305
050202
100035
100036
048905
048905
049608
011108
085802
008402
034500
093100
093100
Seams (s) Produced
Name
(6)
Rosebud & McKay
Robinson
Hagel
Top
Unknown
Unknown
Unknown
No. 11*
No. 12
No. 9
No. VI
No. Ill
No. 11*
No. 12
No. 13
No. V
No. VI
No. VII
No. 5
No. V
No. 2
Hazard No. 5-A
No. 9 & No. 11
No. 5 Block
Unknown
Unknown
Pust
Unknown
Unknown
Unknown
Unknown
Unknown
Rosebud & McKay
Rosebud & McKay
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Composite
of Seams
(7)
093100
051700
048415
048306
048913
048415
048306
048209
048905
048305
048006
E-8
-------
Mine
Code
ITT
State
County
(4)
Seams(s) Produced
Reference
Code
Name
Composite
of Seams
(7)
10213
10214
10215
10216
10217
10218
10219
10220
10221
10222
10223
10224
10225
10226
10227
10228
10229
10230
10231
10232
10233
10234
10235
10236
10237
10238
10239
10240
10241
10242
10243
10244
10245
10246
10247
10248
10249
10250
10251
10252
10253
10254
PA
PA
PA
PA
PA
PA
PA
WV
PA
PA
PA
PA
PA
PA
PA
WV
PA
WV
PA
PA
PA
PA
PA
MD
PA
PA
PA
WV
PA
PA
PA
PA
PA
PA
PA
PA
PA
Clarion
Clarion
Clearfield
Somerset
Cambria
Clearfield
Centre
Mineral
Clearfield
Somerset
Somerset
Somerset
Jefferson
Clearfield
Indiana
Boone
Somerset
Mineral
Indiana
Unknown
Unknown
Unknown
Clearfield
Jefferson
Armstrong
Somerset
Allegheny
Clearfield
Jefferson
Clearfield
Harrison
Indiana
Unknown
Somerset
Unknown
Clarion
Centre
Somerset
Elk
Somerset
Clearfield
Clearfield
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
E-9
-------
Blend
Mine of
Code Mines
(1) (2)
10255
10256
10257
10258
10259
10260
10261
10262
10263
10264
10265
10266
10267
10268
10269
10270
10271
10272
10273
10274
10275
10276
10277
10278
10279
10280
10281
10282
10283
10284
10285
10286
10287
10288
10289
10290
10291
10292
10293
10294
10295
10296
10297
10298
State
(3)
PA
PA
PA
PA
PA
PA
WV
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
WV
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
WV
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
Clearfield
Jefferson.
Somerset
Cambria
Clearfield
Westmoreland
Randolph
Armstrong
Indiana
Armstrong
Indiana
Clearfield
Indiana
Somerset
Armstrong
Armstrong
Armstrong
Armstrong
Clearfield
Fayette
Somerset
Barbour
Somerset
Cambria
Unknown
Armstrong
Indiana
Cambria
Clearfield
Clearfield
Somerset
Jefferson
Jefferson
Jefferson
Preston
Somerset
Clearfield
Clearfield
Clearfield
Clearfield
Unknown
Armstrong
Armstrong
Armstrong
Clearfield
Clearfield
Somerset
Reference
Code Name
(5) (6)
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Composite
of Seams
(7)
E-10
-------
Mine
Code
TIT
Blend
Of
Mines
Seams (s) Produced
State
(3)
County
(4)
Reference
Code
(5)
Name
(6)
Composite
of Seams
(7)
10299
10300
10301
10302
10303
10304
10305
10306
10307
10308
10309
10310
10311
10312
10313
10314
10315
10316
10317.
10318
10319
10320
10321
10322
10323
10324
10325
10326
10327
10328
10329
10330
10331
PA
PA
PA
WV
PA
PA
WV
PA
PA
WV
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
PA
WV
PA
PA
PA
PA
PA
PA
PA
KY E
WV
Clearfield
Clarion
Clearfield
Harrison
Clearfield
Somerset
Preston
Cambria
Jefferson
Clearfield
Barbour
Unknown
Clearfield
Jefferson
Jefferson
Clearfield
Somerset
Clearfield
Jefferson
Jefferson
Jefferson
Clearfield
Clearfield
Indiana
Monongalia
Clearfield
Clearfield
Indiana
Somerset
Clearfield
Indiana
Clearfield
Martin
Unknown
Unknown
Unknown
Unknown
Unkn :>wn
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
E-ll
-------
APPENDIX F
MAPS OF MINE LOCATIONS BY PRODUCING
DISTRICT, STATE AND COUNTY
F-l
-------
Definition of Bureau of Mines Bituminous Coal and
Lignite Producing Districts
K)
DISTRICT 1. EASTERN PENNSYLVANIA
Peaaeyleanla
Armstrong County (part). All mlnu eaat of Allegheny
River, end thuee mine* eerved by the Pittsburgh A
Shawniirt Railruad located on the wen bank of the river.
P»>rtl< County (pin). All mines tooled on and e>et of
ll.i line ( Indian Cmk Valley branch of the Haltimora
A llhiu Railroad.
In>tisna County (part) All mmee not lervod by the
Nallsbtirg brsnth of the Pennsylvania Railroad.
Wr>lin..irlsn>l Cuunly (fieri). All minee eerved by the
I'rtiiisylvama Hailrtiao from Torrance, east.
All mine* in the following counties:
Hrdfoid Onlre Korea! McKuD
Illair Clsnon Fullon Mifflin
llradford ricerneld Huntingdon Toller
Csmbne Clinton Jelfereon Somereal
Csnit run Elk I.yrominf Tiof K
Merylsnd. All rt.mrs In the Slat*.
Weil Virginia. All mini. In the following counliae:
Grant Mineral Tucker
DISTHIlT 2 WESTERN PENNSYLVANIA
PennayUsnla
Armstrong County (pert). JU1 mlnee weet of the Alla-
K' *ny River escc|it those minee eerved by the Pllla-
rgh A Sbewmut Railroad.
Feyelle County (parti. All mine* eicept thoee on and
reel of the line of Indian Craek Valley branch of the
lUllimore A Ohio Railroad.
Indiana County (part). All mlnu aerved by the Salis-
bury branch of the Pennsylvania Itailroad.
Wrslmorclsnd County (nait). AM inlnea eicept thoee
e«rvcd by Ihe Pennsylvania Railroad from Torranco,
eael.
All rninee In UM following counties:
Allegheny Butler Lawrence Vanango
Beet-T Gracna Mercer Waahin(toB
DISTRICT > NORTHERN WEST VIRGINIA
We«l Virginia
Nirhulee County (part) All mlnee aervcd by or n«rth
of UK Dallimora A Ohio Hailcuad
AH minra in the following eoonliee:
llaiUiur Jackson Randolph Webster
Rraston Ijcwia Ritchie Wetiel
C.lh..un Uamm R*>ana Wlrt
IV.Miidg< Munungelle, Taylor Wood
Gilmcr 1'lcaeanla Tyler
Hamsun Preston U|>thur
DISTRICT 4. OHIO. All nines In lie Bute.
IllSTKICT a. yiCHICAN. All mlnea In the Stala.
DISTRICT «, PANHANDLE
a following coui
Hancock ManhalT Ohio
Weal VireuOa. All minea u Ua following couitlie:
ManhalT
UrooVe
DISTRICT 7.SOUTHERN NO. 1
Weal Vlrftnla
Payelta County (part).All mines eeet of Gaulev River
and all mines served by the Cauley River branch of the
Chesapeake aj Ohio Railroad and mines served by the
Vtrfinian Railway.
U.-Dowell County (pan).All mines In that portion of
the county served by the Dry Fork branch of the Nor-
folk A Western Railroad >nd east thereof.
Ualeifh County (pert).All minri except Ihoac on the
Coal River branch of the Chesapeake a\ Ohio Railroad
and north thereof.
Wyoimni County (part). All minee In that portion
served by the liilberl branch uf the Virginian lUilway
lyin| east of the mouth of Skin Fork of Guyendol
River and in that portion served by the m«in line and
the Glen Rogers branch of the Virginian Railway.
All mines in the followinK counties:
Grcenbner Mercer Monroe Pocahontju Summers
Virginia
Buchanan County (part).All mines In that portion of
lha county served by the Richlands-Jewell Hidgt
branch of the Norfolk * Western Railroad and in that
portion on the headwaters of Duma) Creek eaat of
Lynn Camp Creek (a tributary of Dismal Creek).
Taitwell County (part).All mines in those portions of
the county served by the Dry Pork branch to C«dar
UlulT anil from Ulueston» Junction to Uoisjcvaln
branch of the Norfolk A Western Railroad and Rlrh-
lands-Jenll Ridge branch of the Norfolk A Western
Railroad.
All minee In the following counties:
Montgomery Pulaaki Wythe Cllaa Craig
DISTRICT 1.SOUTHERN NO. 2
Weal Virginia
Fayetta County (part).All mince west of tht Gaulev
River eirepl mines served by the Gauley River branch
uf lha Chcespeake * Ohio Railroad. .
ifcllowell County (part)All mines west of and not
aerved by lha Dry Kork branch of the Norfolk *
WesUrn Railroad.
Nichols! County (part).All minee In that part of lha
county aouth of and not acrvad by lha Baltimore A
Ohio Railroad.
Kalelgh County (part).All mlnee oa lha Coal River
branch of the Chasapeaka A Ohio Railroad and north
thereof.
Wyoming County (part).All mines In that portion
aerved by the Gilbert branch of the Virginian Railway
and lying west of the mouth of SUn Fork of Guyandot
River.
All mfnee In the following counties:
Boona Kanawha Mason Wayne
Csb.ll Lincoln Mmgo
Clay Logan Putnam
Virginia
Buchanan County (part).All mines In the county, al-
cept in that portion on the headwaters of Dismal
Craek. eaat of Lynn Camp Craek (a tributary of Dis-
mal Creek) and in that portion served by the Richlands-
Jewell Ridge branch of the Norfolk A WesUrn Rail-
road.
Tasewell County (part).All mines In the county ex-
cept in thuse portions served by Ihe Dry Fork branch
of Ihe Norfolk A Western Railroad and branch from
Blueslonr Junction to Roissevain of Norfolk A West-
em Kailrosil and Hichlands Jewell Ridge branch of
Ihe Norfolk A Western Railroad.
All mines in the following counties.
Dickinson Russell Wise
Lee Scott
Kentucky All mines in the following counties In esat-
eni Kentucky.
Nrll Creenup I awrence Morgan
Boyd llsrlan I
-------
PRODUCING DISTRICTS 1 AND 2
MARYLAND (District 1) 3 Mines WEST VIRGINIA (District 1) 4 Mines
PENNSYLVANIA (District 1) 109 Mines PENNSYLVANIA (District 2) 3 Mines
Clarion (4 Mines)
10213 10248
10214 10300
Armstrong
10212
10237
10262
10263
10267
10268
10053
(14 Mines),
10269
10270
10278
10293
10294
10295
10082
Jefferson
10205
10225
10236
10241
10256
10283
10284
(14 Mines)
10285
10307
10311
10312
10316
10317
10318
Elk (1 Mine)
10251
10321'
10325
10328
Indiana (9 Mines)
10054 10265
10055
10227
10231
10244*
Westmoreland (1 Mine)
10260
Fayette (2 Mines)
10086 10272*
Somerset
10204
10209
10211
10216
10222
10223
10224
10229
10238
10246
10250
(22 Mines)
10252
10257
10266
10273
10275
10282
10287
10298
10304
10314-
10326
Garrett (1 Mine)
10210
Grant (2 Mines)
10031 10047
Unknown (9 Mines)
10232 10277
10233 10292
10234 10309
10245 10331
10247
Clearfield (37 Mines)
10080
10081
10085
10207
10215
10218
10221
10226
10240
10242
10253
10254
10235
10255
10259
10264
10271
10280
10281
10288
10289
10291
10296
10297
10290
10299
10301
10303
10310
10313
10315
10319
10320
10324
10327
10329
10323
Centre (2 Mines)
10219 10249
Cambria (6 Mines)
10208 10276
10217 10279
10258 10306
Allegany (2 Mines)
10017 10239
Mineral (2 Mines)
10220 10230
* District 2
F-3
-------
PRODUCING DISTRICTS 3 AND 6
WEST VIRGINIA
District 3 12 Mines
District 6 0 Mines
Marion (2 Mines)
10083 10084
Harrison (2 Mines)
10243 10302
Mononqalia (2 Mines)
10029 10322
Preston (2 Mines)
10286 10305
Barbour (3 Mines)
10206 10308
10274
Randolph (1 Mine)
10261
F-4
-------
PRODUCING DISTRICT 4
OHIO 64 Mines
Coshocton (11 Mines)
10115
10116
10117
10120
10123
10124
Muskingum
10113
10114
10133
10126
10131
10137
10161
10174
(5 Mines)
10159
10171
Perry (6 Mines)
10119 10160
10138 10162
10147 10173
Hocking (4 Mines)
10144 10149
10146 10150
Vinton (5 Mines)
10142 10148
10143 10167
10145
Jackson
10127
(2 Mines)
10165
Tuscarawas (6 Mines)
10121 10128
10122 10134
10125 10163
Guernsey (2 Mines)
10118 10158
Jefferson (1 Mine)
10130
Harrison
10030
(2 Mines)
10129
Belmont (1 Mine)
10192
Morgan (1 Mine)
10164
Unknown County
10132 10151
10135 10152
10136 10153
10139 10154
10140 10155
10141 10156
(18 Mines)
10157
10166
10168
10169
10170
10172
F-5
-------
PRODUCING DISTRICTS 7 AND 8
VIRGINIA (District 7) 0 Mines KENTUCKY (District 8) 2t Mines
WEST VIRGINIA (District 7J 0 Mines TENNESSEE (District 8) 0 Mines
VIRGINIA (District 8) 1 Mine
WEST VIRGINIA (Districts) 3 Mines
Breathitt (1 Mine)
10028
Perry (3 Mines)
10025 10111
10072
Rockcastle {1 Mine)
10026
Elliot (1 Mine)
10034
Johnson {1 Mine)
10033
Clay (2 Mines)
10003 10027
McCreary (1 Mine)
10002
Whittey (1 Mine)
10050
Kanawha (1 Mine)
10193
Martin (3 Mines)
10022 10330
10087
Boone {1 Mine)
10228
Logan (1 Mine)
' 10023
Pike (1 Mine)
10016
Buchanan (1 Mine)
10108
Floyd (1 Mine)
10073
Knott (2 Mines)
10001 10191
Knox (1 Mine)
10051
Bell (1 Mine)
10048
Harlan (1 Mine)
10112
Unknown, Kentucky (1 Mine)
10110
Unknown, (District 8) (3 Mines)
10180 10182
10181
P-6
-------
PRODUCING DISTRICT 9
WESTERN KENTUCKY 12 Mines
Union (2 Mines)
10020 10021
Henderson (1 Mine)
10075
Webster (1 Mine)
10018
Ohio (2 Mines)
10049 10096
Hopkins (1 Mine)
10019
Muhlenburg (4 Mines)
10009 10052
10183 10186
Unknown County (1 Mine)
10056
F-7
-------
PRODUCING DISTRICTS 10 AND 11
ILLINOIS {District 10) 22 Mines
INDIANA (District 11) 5 Mines
Peoria (1 Mine)
10004
Douglas (1 Mine)
10058
Fulton (2 Mines)
10043 10190
Christian (1 Mine)
10035
Macoupin (2 Mines)
10042 10095
St. Clair (2 Mines)
10045 10105
Randolph (2 Mines)
10046 10109
Perry (5 Mines)
10024 10093
10092 10097
10036
Clay (1 Mine)
10185
Sullivan (1 Mine)
10187
Wabash (1 Mine)
10188
Pike (1 Mine)
10100
Warrick (2 Mines)
10184 10189
Jackson {1 Mine)
10094
Franklin (2 Mines)
10101 10102
Jefferson (2 Mines)
10103 10104
F-8
-------
PRODUCING DISTRICT 12
IOWA 6 Mines
Marion (1 Mine)
10070
Lucas (1 Mine)
10065
Mahaska (3 Mines)
10067 10069
10068
Monroe (1 Mine)
10062
F-9
-------
PRODUCING DISTRICT 13
ALABAMA 0 Mines
TENNESSEE 1 Mine
Marion (1 Mine)
10032
P-10
-------
PRODUCING DISTRICT 15
KANSAS 1 Mine
MISSOURI 8 Mines
OKLAHOMA 0 Mines
Putnam (1 Mine)
10066
Macon ( 1 Mine)-
10064
Crawford (1 Mine )
10010
Randolph (3 Mines)
10014 10063
10015
Howard (2 Mines)
10011 10013
Audrain (1 Mine)
10012
-------
PRODUCING DISTRICTS 16 AND 17
COLORADO
District 16 2 Mines
District 17 6 Mines
Moffat (3 Mines)>
10089 10106
10090
Routt (2 Mines)
10057 10088
Jackson (1 Mine)
10079
.Weld (1 Mine)
10078
Las Animas (1 Mine)
10091
P-12
-------
PRODUCING DISTRICT 18
ARIZONA 1 Mine
NEW MEXICO 1 Mine
Navajo (1 Mine
10071
McKinley (1 Mine)
10098
F-13
-------
PRODUCING DISTRICT 19
WYOMING 10 Mines
Sheridan (1 Mine)
10040
Campbell (4 Mines)
10006 10061
10007 10076
Sweetwater (1 Miner,
10041
Carbon (4 Mines)
10038 10077
10039 10099
F-14
-------
PRODUCING DISTRICT 20
UTAH 1 Mine
Carbon (1 Mine)
10107
F-15
-------
PRODUCING DISTRICT 21
NORTH DAKOTA 4 Mines
Mercer (2 Mines)
10044 10059
Bowman (1 Mine)
10060
Oliver (1 Mine)
10179
P-16
-------
PRODUCING DISTRICT 22
MONTANA t3 Mines
Rosebud (3 Mines)
10005 10202
10201*
Richland (2 Mines)
10195* 10196
ibaux (1 Mine)
10194*
Big Horn (3 Mines)
10037 10178
10203*
Glister (3 Mines)
10198 10200
10199
Powder River (1 Mine)
10197*
Core Samples, No production.
F-17
-------
l.'F T')fJ3
212177 01STKICII H'»
I7C
51 tr
IIU.H If
1111 'If
i inrt
I "2
1 1121;
SA'l 22i">
111 I I Sf I
1211 1121
I I \»
I « I
i.3
//S.7
1 ? 1 7C
1
I2C
I 1 H2
t I M
1 IIIISF
M I 121
211 Hi
II MI.
'JL
If I
12 I 10
I < C
12 II 2F2
22 I2J
t 22I21IL
I 1
1 ^
I I
2 I SM
J "I
221 ''« I
21 I ^
MINI-DIM nAlt IS /U0109
- N= 1027
.
1 H009 CIJI OH<»20 8MM| 04H9IJ M-0-SAMPL JNU I
-------
o
10
CMK.KB.M "19 009^0 10009 252 92600. 3.67 11299. 6.17
NitH-HAL, DISlWIH'lT ION - IIHSEHVH;
*(.'< *
11140
ruu *
F H I >4i !( NL Y
FMll'll
till
1
ft
.00 -52$«.6:
-S?»H.6$ -1923. H
2
0
s
1
3
3<>
-1923. «J
1391.76
u
120
1391.76
4706.96
5
Bfc«
U706.96
8022. Ih
6
2
6022.16
11337.35
7
(I
11 1S7.3S
|4e>52,55
8 <*
1 <;
I46S2.SS 1796/,/a
17967. 7u *»***«**
SbO
S2S
'I'JO
JlS
I/S
HO
lO'i
/O
is
Crt
s
1
-------
HSlHlniM IIJN MHKC1FI)
CELL 1
Hlf.u.lt "I Y 0
Ft- 'I'"! .00 -52
lilt -b,MH.6* -19
F»CM « tUtlALH 16 HMlNlS
J6M
452
2S6
o r*2
i, 1»6
I '4 1
JO
O'J
n (4
,,
16
CFLI- l'HSH*./fl> F«HEC1tl»
1 II .2
2 8 AMF. '»S
* > t»
* * * S IS
: : : 1 1
. ...... . . ........... ...>._. H
o x
njFFfHlNCF S
.* ^5?
a5:5 a
«!:5 -
?U5.8 >
C.T » 25
57.7 ^
-------
CM L «
i
U 0
.00 -S2JR.M
11V1.76
J4|
IJ91.7*.
47116.94
070
6
11147. iS
7
4}
l«
146S2.SS
17967.7« »*»«*
EACH tUU«i<| |>t Hi||fO3
ill
in
tit
ti-t
s
11
Cl.tt
I
n
0
14
*! cir«
MI.
!%.«
i«,«
(*.«
*.»
-------
Lili; nUKMAL OISIUIHUTILM
CU.L « I
22
.00
O MO
I MS
Wl ,,.
SbO
US
I /S
I'Hi
I os
19
8.91
9. *d
3 >d
ft 2
W O
M
O X
^3 O
3"
P
50
8
-------
i>Tsifii;
1(1. SO
**»**«*»
J'>'»M
!»»»
FO
HII
o't
MM (,
i.» *
1" *
CHL UH.StMVFO f.XMfcCTKl) DIFffUtNCF
1 i? .«? 21. S
? If. h.t 9. U
< <*i ha.? ^o.?
'» «>S ?'lrt.J Ifl^.B
S HS9 JH.4 'IhS.a
<> l«* ?2.2 58.7
» 1 6.1 « . h
' 1 .«? .i
W f>
2 *o
s: »ti
£3 M
A "U PI
tr1 Z
* WO
* l-l
ox
Id 0
r*
~ * 4
^
O
IT!
*J
g
r1
a
M
O
(Hi S'JIIAMfl'S IHF
§
-------
o
L»31S i
JJ />li>1
I in. «
221. b
M2.4
Ml. 3
bSa.l
«*»'*, 9
1 1 '> . 7
ttrtb.S
<*'»/. 3
77 iM.sfMicii nv K|Nf.i 10009 cm o»'»2o SMMI 040913 M-rj-3*MHi.|Nni u t-n-s*Mpt.iNui 4 HAMb
1
en
S
a
cvi
§
c
o
§
>
K
>-3
n
i n
^
t)
50
a
5.14 *
>
xJ
M
Q
H
X
0
"
-------
I'ISlH JHUI Tl.if - HHSE
CM L » J
.00
s.l*
2
3.67
J.«»7
3.97
«.BH
**»*»***
> rtC'i * t»li»l S t 7 HHI«t8
1? / *?
^S'i
M *
1 7 *
K %
* 2 *X)
?§
WO
. w
: §x
a«
* a
* ^
: s
* * s
' v
i
H
o
p
d
1
-------
t>isiM|piit ION - E
r.u L «
t25«5fc7-2 o 0
(HIl'U
I'll
EACH *
JH'I
(on
V>-4
."1 )
I/O
o ""
1 1 f">
vo t lf.
HO
tvl
1M
4«>
lo
CH L
I
f
4
S
i,
7
H
^
cm 3'j
.<»0 2.76 5.06 5.56 5.67 5.97 «.2H tt.SB U.Hft
^.76 i.Oh 5.56 5.67 5,97 *!.<>« 'I.S8 «.fi« »**«***«*
tUIIAL<4 \H HIII'-tlH
*
*
*
* en >
* K S
A ^13 SJ
_ _ ..._.._.__>___._« ^ o
M
UMRI-MVtl' FXt'tCTtO DJFFFWtNCt O X
? .:f .:£ if?
«7 62.? H.7 §
^qi 2^H( 12.2
'i(>? I'M [5 8.2 9
?07 ?'»H.i U0.8
64 62. ? 1.5 §
ii 6.1 u.a >
" .? 5.5 ^
*"**'* o
f
50
8
-------
fiAMH4 DtSFHIHUl IUN
CtlC « I
a
..)<»
f.. i*
F^^^ out u v
FrtOlf
M:
6
226
j.06
J.J6
1.67
H.2«
h«l.i tt.'UAi s I h I'MINTS
JHU
\«>H
JS2
J^
i2«>
JIM
o
1
l-»
o
l" It
^s»>
^'tl)
Sli-t
f(IH
I'M
|7h
160
1 M'(
|2«
lie'
Jh
F »»
no
oa
tH *
l<> *
In *
CFLL UHSI-Ktffcn FXMfCrFl) DIFFfKENCE
1 1 ." .0
» 1 2.2 .7
? '17 SS.6 B.J
" ^«»l ^72. S |6.0
S 'JO? $0<4.<) fc.6
t> ?0 7 226. 1 1 H.B
/ j»u AIR M^
' O* f«».n .C
» It H>.s .0
" a I.I 2.1
CM) SUMACH'S AWf 9.I04 4.27H d.flSl Pi
SA-^'L^ SIAlfSTfCSl MMN, 30, H8i) 1.5172 .07ftJ
H 6,4,2 OFDHI;
.OSflb
t
I
tS ilf fKtfcOD1'
* Cft >
a* x)
* S *5
A *D M
M 25
* WO
0 X
a
H 0
3"
o
«i
>
>
f
3
M
O
1 >
f
ti
»
o
n
-------
l.in. .'IMWriAl IIISTUMilT HH - MM5JFI«VEn
4
CM 1. i \ ? i M
F HI- -jut if. V 0 ? 5) 26«
fi"MJ .00 i.n* |.i« 1,22
Ml !.'"> l.M 1.22 1.30
t «I.M HJUALI ) PHlNia
»HM
5f)M
i'i^
iS6
^/?n
0 I'1"
i ><\n
*~* > 1 1
i_i <* /«
-»S6
f II)
^^a
,>ii/i
f^
1 /6
1 1>0
1 'l'4
1
*M ^p«
S »d
d M
tr' ^^
td c)
M
0 X
§5?
^
a
* ^
* o
* ^
* * E
P
<
K|
H
n
>
f
tj
»
8
-------
LIU; 'JuHMAi
Ct i L a
104
I
0
.00
-
1.61
'-hl
«****«***
o
FAC" H4li*l S 16
,91* H
t1 <)
I /b
I Ml
1 1«?
'*(>
HO
n I
Id
*
*
*
g$
S tJ
tr1 z
BO
o x
35?
H)
a
H
3
g
>
5
i-3
*-«
CH.L
i
0
?
s*
FXMFf. IH)
i.
»fl(> i'M.j
in
fc.J
.i?
3.6
3^8
4."
«.3
'o.d/i;
1J.H17
9.0<>0 FOH 6,4.2 UEIiHfKS Up
!
-------
o
I-1
to
ANALYSIS 1>F HIM N= l<>27
f- !>>« r*l2i77 DI31HICI: ()» MINJ i inoov cut o««»2P SI-AMI 0/1*913 M-(l-SA"Pl. INT, i u T-IJ-SAKPLINGI s22 1 3 11 1
t 1 1 1 1 1 2 11 M 1 1 311 3111 1 1 r> 1 it
2 I 212 3 2111 i 1 12 1 1 M 21 1
1 1 1 1 1 2 1 1 1
1
1
w >
S 2
M td
M D
M
0 X
H O
g"
o
is
>
K
1-3
H
Jj*
f
HJ
s
.... 8
.0211
M'lU'j.43
-------
Hiii. UL iiisii>|»)iji i
c» i.u «
.00 IOOIO.B-J
231
tUMO.JJ
ttJ6
10«75».08
7 l< *
««> U 2
6 11669.35 ll»S G
» ^
s
A nr|
' * ?*
^
r
t<
n
M
n
>
f
"0
a
^
w
»d
§
O
X
CD
8
-------
OlSlH Irtllfllli'J -
Cll L * \ j \ l
(HI i|iif4LV n 6 6(> 201
FniHl .00 100RO.PJ 10J<4b.58 10610. J4
MI'll t"U*(.S 1h HHl^lS
Joh
^<>il
V'l
Q ,MIH
i rv
t^ I'n
in"
| '('l
4n
5"
O'l
IM *
J** *
1 h *
1 ? ./? 1.5
^ A 6.1 .1
I S'» »>?..S 16.H
S 'IJS VM.H «i.^
h . '*
> f ,f 1.1
i «
\ 591
10875.08
i 6 7 B «>
I ?4ft t>2 6 U
11139.83 I1U04.SB 11669.33 livi'l.n?
11004. SA 11hh9.]J 1193/1.07 *********
|S
* 13 M
* MO
* M
i "
o
2
IP
K
H
n
SUUAHIus
17.760
FHH 6,0,? CtliKEt S (IF
-------
MV/KKIUI I;AM«H IMSIIMHMT HIM -
CHL » I
FrtCrfUfitY 0 S 61 255 39
Fw""« ."» KtORo.HJ lu5'*5.bfl 10610.53 I0a7b.0fl
r"! 100HO.H3 IDS'IS.SH 1D610.33 10P75.0P 111.19.83
FAI..I * tUIJALS 1», PillNIS
< <<
\H*
*>"T>
\\H
;M«
P/
.»4ll ,
ft* t
^'t *
IV *
1 /" *
O too
M I'"' *
0\ |^t< ,
(^ *
1» .
H.) .
'>'! t
\h »94.« 02.1
» ->3J 2«l.3 7.8
' «S 63.1 17.6
M n ;.s s.o
' ? .'| t.i
TIM H-JIIAHH)^ AKF. <-5.t79 lb.3«l 11.9U/1 f-l)H 6./I,? DfUHt
SAMI'LF SUTIMICSI '^AM, til), rtSP --- 9.3060 .02HO ,OH3«
3 2U\ 63 7 0
11139. B3 lldOM.btt 11669.33 llVi«.ll/
ll«04.Sfl 11669.33 11931.07 ***»«***
*
* W 3*
IS
W O
M
0 X
° 0
^ ..
G
i-3
§
§
>
g
tS IIF l-Hfil-UM ||
M
n
P
T3
§
-------
riisii
rfl.i. 1 ? J 'I «
f-Wh'J"fc"CY f 2 J9 222 50<
F^n-'l .00 9,21 9.2,.12
II M 4
O
1 t«M>
h- x l'i
*^j
f-iH
?^£
/? 4 1
/»l'l
1 i> M
1 II
IOS
5i
t \ *
> 6 7 « 9
1 211 Ifl 7 u
9.32 9.35 9.?>« 9. /ID
9.3S 9.i8 9.00 ***»»
W >
s ^
t^ «Z
td o
M
0 X
G
^d "
a
j
o
^
* * !z!
f
»<
H
I 1
n
>
f
d
§
-------
UK;
CH L «
MI'J.ltnCY
f HIM:
fi'j
.00
''.'I
1 3
1 h
''.31
9^3't
1 U
63 3«8
9.3'' 9.36 9»
9.36 9.?9 9.
5
3«>3
39
32
3<
<».33
9.35
t
16
7
63
9.35
9.16
9
9
.3*
.>
InO
1 *4 '4
131
1 I/1
'in
'10
o'l
IH *
tf *
1 n *
Ctl.L IHISfKVfU t-XPFCTFI) (if FFI-HfcNCF
I ?. .3 1.3
,' 3 6.1 3.6
< 5* 63.3 33. /
'1 />?/> 3'4M.3 3S.H
- SOI J'M.J IJO. 3
f, p| ^ 3'4M. J JU.H
' J1* ^>3.3 33.'
'» 7 h , ( . >
'' f* .^ .3
*
*
*
*
*
*
A
*
*
*
*
* *
* *
* i
2$
3jj Jd
M Z
n O
H
0 X
a
tti «
a
o
^
>
3s
tr1
K
H3
M
0
tr1
TJ
w
mi
h«.
s/.
r.s iif Fwn I'd-
-------
O
I
h-«
VO
AUALT31S l\t LBS SULFUR/MMBtu N= 1027
Flt'Sl 2I?177 DISIUICII O'/ MlNfj 10009 CHI 08920 8EAMI 0«H9|3 M-0-SAMRL 1 NG| U T-O-SAMPl J N(J 1 U HANH| J M.O-HIMINui 1
.0
1 1 tl . «
«.'«M ,f>
!>.>.«
Vi l. 1
ui>;i.9
,,s.,
MMf.S
>'> 7. i
1 11111 1 11
1 1 1 t II 1 1 II 1 1 2 1 111 1
11 12 1 21111 1113 1
1 11 U 12 1 1 1 1 21 1 1 t
1 1 1 111 1 3 11 11 2 1 1 11 1
1 1 1121 1 II 1 1 11 1 1 1
1 t 21 1 1 1212 1211 1 2 1
1 1 3 II 1 2122 11 11 1
111 1 512 1 1 2 1 1
t 112 1 112 11 1 t II II 1
II 1 1 1 1 12 2 1 1 t 1 1
111 31 23 1 11 11 11 IS t
1 1 II II 1 1 1 t 13131 21 112 21 it 3 1 I 1 1 1
1 1 1112 1121 Illl 1211 It
11 till 1 til
1 t i 1 1 II
Illl i 21 1 III 1 1 II
11 13 I 121 2 1 1 2111 till
i 11 1 I 2 1 1 2 1 23 11 12
It 121 1 1 II 11 1 1 t 1
12 11 1 121111 2 1 13 1 1 11 1
t 11 11 31 2 tl 22 1 1 1
II 1 211 111 1 111 12 Illl t 1
1 112 2132 21111 2111 1 I 1
1 1 21 1 Illlll 1 It? 21 1
t 1 11 2 221 21 1 1 3 1 II
1 1 2111 13 11 II 1 1 1 III 1
1113 211111 1 1 12 1 22 1 1
1 1 3 1 1 2 1 1 1 1 1 2 1 1 131 11
1 1 3? t 1 15 1 Illl II
1 1 1 1 2 III 1 2 II 1 1 1 1
1 1 1 211 1 1 1 1 1 11 1 1
1 t 11 1 1 1 131 1 t 121? 21 11
t II II 11 121 21 11 1 1 2 1
1 Illl 1121 1 II 11 It III t 11 1 11
t 2 1 t 1 12 31122 21 ? 1 2 1 II 1
1 1 Illl 12122 II 212 1 1 2 i
1 111 22211 21 IIII2I 11 11 | t
1 11 i 1 11
1 1 f\? « 11 1 1 2 1 11 1
1 11 1 121113 2 Illl 3 1
111 2111 1 1 1 2 12 311 \\ t l
11 112 1 1 21 I 1 12? 1 1 t 1 II 111 t
1 II I 2 1 11 1 2 ?! 11211 1 / \ \
? 1 32 21 121? 1? llltll? 1 III
1 111 13211 212 f 1 PI 1 113 1 II
t OM>,.0
i i i i i i 12 i
.
2-54 2.75 2.96 3.17 3.37 1.5ft 1.79 4.00 4 21 4 52
SA'in^ SUTiuiCS: '-if AN, Hh, f-,i> 3.4724 .3151 ,i'907
PHfP.I 1
1
1
Cfi >
3 *ti
*d W
r1 2
M a
M
O X
a
i?
1 "3
O
z
r1
i ""^
o
^
ti
50
: : 8
4.61 4.81 g
-------
Nil" ii»L I'l
OIL «
ftf\ 'JIIK II Y
(III
STH1HUU"'I -
5
19
.00
i.sn
5.10
s. in
?6<4
>.7
O
I- HT'I tUllftl.S 1 I PlMNfS
llltH
f.)f
.»ss
/o
in
OH
SI
i'l
M
W
o
n
o
I
-------
fllll/.-ML I'l
CM.L «
FHM»f'CV
F-'liI"!
I'll
1- AIM * ( u
?<\n
C^ ?ll«
I
£ I/*
|-l<4
HO
11
4,>
1*
CM U U»
,>
\
'1
S
»
Cil| HuHArt
SlK|IIHT|ll'l - EXPtCtH>
1 ? 3 » '.
0 6 62 2aft 31"
.00 a. 5n 5.10 5.70 6.30
'J.5n S.lii 5.70 6.^0 6.90
*
* *
* *
:>t'wv/F» f- h!l 2i6
'10 6->.2 12.7
Pb'i 2'IH.$ 15.2
\?<' I'M.* *0.2
/O »i2.^ 7. S
H o.i i.a
i .-» J.I
HIS Al'f hi. IHS IS.SIu l<4.07*< F(IK 6,0,? [>f IjH
5 ft 7 » V
i 2UH 6? 6 U
6.00 7.50 P. 00 H.f>9
7.5(1 8,09 6.6O *»**»«
cn >
g HJ
S 0
> « M
§x
' * w °
O
2
>
5
H
M
n
>
f
v
*&
es u» FHitmin g
-------
(,AM IA
rut «
FHl
FNIMJ
rui
S.IO
S5
5,10
5.70
276
6.10
6.90
6.90
7.50
H.09
6.09
fi.69
*********
K4t il * t "7f>.() 11.5 O
S itfH 4«»'l.5 ?"».0 "^
» i»OS ?2J.l 17. » >
7 /o fc'4. o 5.S 1
h n \\.<> i.T f.
i i.i a.i j|
M
CHI SiJUAl-tPS A^F. '».?i9 S.Bhh t.99'1 Fl'R 6,4,2 OFGIifFS (IF FMtFI'DM O
XJ
»
8
^j
»0
M
X
O
-------
l.l II, NUWMAi 01 SIKIUHT ID'J - MKSFHVFl)
CH.L * 12$<
f lu iJ-u NLY o 6 56 241
FHIMI .00 1.S7 1.66 1.75
'US I.S7 \.bh 1.7% 1.84
Ml M * FUllAuM 1 7 HlllNfS
|U<<
t 11 1
T" 1
i/«
Vi/
O t"C
to 4*4
W «0o
?-»9
^/^
^s->
P i«
^ 6 7 « 9
1 226 h9 ft i.
1,91 2.02 2.11
.............. . ... ...... ,_,
3
M
O
f
-------
8
O
cn
cn
8
I
cn
M
Z
0
i
1 iii; IMIHMAL 11 [3I'<| ilill liN - MfFllfl)
rlt-L " \ i $ U S * / 8 V
fMlM| .''» 1.57 t.hh I.7S l.fll l.»i 2.02 2.11 2.20
Tll« I.V l.6h 1.7S 1.84 l.«*i 2.02 . 2.U ?.20 ****.****
HAf.i1 * b"UAL^ \H HUlNlS
!»M
$2')
r»2
Ion
l-l'l
120
1 12
*».
b'l
UH *
4
3. *tJ
M Z
w a
* M
* ox
1?
I'tLL UMStl^tH FXl»FtlH> OlFFRHtNCf O
1 0 .2 .2 ^
».' h 6 . 1 . 1 Jjy
" ^»l T ??h 2UH.) 21. H 3
7 (i<> h2.2 6.J 0
H ». -S.t .1 fl
fill SOUAltl'.S A'
-------
SUTtlAHY OF SALIENT CIIAHACTEK1ST1CS Of DAI A SITIS ANAI.Y/.KU
(Statlatlcal analyala baaed on data aa received from respondents, disregarding lot-size)
AVERAGE CHARACTERISTICS
Mine
Code
(1)
10001
10002
10003
IOOC4 /
iimosj}
10005 -
1000 5
10006
10006
10006
10006
10006
10006
10007
10007
IOOOR
10009
10010
1001 1
10012
10013
10014
10015
10016
10017
100 18
inn io
HJ I0nl9
I 10020
M 10021
10022
10023
10024
10025
10026
10(127
10028
10029
1003d
10031
10031
10032
10033
10034
10035
10036
10036
10037
10037
10038
I003B
10031
1 00 1';
10040
IOD',1
Klll'i?
1 00'. 7
inn'ij
loiy.'.
100'. 5
10041
F1PS
Code
(2)
711119
211109
211051
170143
30008 /
300087
300087
560005
560005 .
560005
560005
560005
560005
560005
560005
560005
212177
200037
290089
200007
290089
290175
290175
211195
240001
212733
212107
212107
212226
212225
211)59
540045
170157
211193
211203
21)051
211025
540061
390067
540023
540023
470115
211115
211063
170021
170145
170145
300003
300003
560007
560007
560007
560007
560003
190017
1 7IH 1 7
1701 17
1 70057
380057
1 70lf.J
170161
USRH
Dlat.
(3)
a
a
8
10
22
22
22
19
19
19
19
19
19
19
19
19
9
15
15
15
15
15
15
a
i
9
9
9
9
9
8
8
10
8
6
a
8
3
4
I
I
13
8
8
10
10
10
22
22
19
19
19
19
19
10
III
10
III
21
10
10
Red
Code
(4)
010401
015109
100001
090002
080800
080800
080800
095100
095100
095100
095100
095100
095100
100003
100003
100002
048913
049202
049202
049008
049202
049202
098700
100004
001000
048415
048415
04P415
048415
048415
100005
015104
090002
100006
015108
021202
011108
003604
007402
007102
007102
028601
008402
016200
048408
090002
090002
069802
069802
036555
036555
081700
081700
078300
10(1007
O/i H'. OH
O'.ft'iOR
(I'.flnO1)
U56100
(I'.R'illR
O'iB'iOII
Rank
(5)
3
3
3
3
2
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
2
2
2
2
2
2
2
?
:i
3
1
1
3
3
Mining Preparation Method of
Method Code Sampling
(6) (7) (8)
9 9
9 9
9 9
2 4
2
2
2
2
2
2
2
2
2
2
2
2
1
2
2
2
2 ;
2
2
1
2
1
9
9
1
1
2
9
M
.2 4
,2 4
! 4
! 4
1 4
1 4
! 9
I 1.3,9
I 9
9
2 1.4
3 9
3 9
42 9
2 9
2 1
9 1 4
9 1 4
9 1 4
9 1 4
1 1 1
1 1 1
1 3 1
' 2 1.4,9
9 1 2
9 1 2
9 1 2
1 2 2
2 2 1
2 2 1,4.9
2 1
2 9
2
T- ,-
2 .4
2
2 1
1 7
1 7 ,/.
2 2
2 1
1 2
I 2
Type of
Samp 1 in
(9)
9
9
9
4
4
1
2
2
3
1
2
3
3
2
3
5
4
4
4
4
4
4
4
4
4
4
4
4
4
4
2
4
4
it
4
4
4
2
2
4
4
3
3
3
3
3
4
3
2
3
4
3
I:
3
3
)
/,
1
4
4
3
Volume
(Tonal
(10)
5318
3150
5688
9110
8423
10147
10513
..
11035
10798
10402
10800
10625
11692
6254
2496
706
505
712
904
580
10024
2055
4548
3739
5643
4031
4354
5060
2594
1470
1800
1100
1600
300
11870
10360
2824
3291
1500
1500
1500
10000
10000
8332
10000
10000
8746
10000
9056
10000
luoon
i. (Kin
HI 1"
;uon
jooooii
1752
IH'.fi
Sulfur
Content
(7. A. R.)
(11)
2.69
1.34
2.18
2.24
0.74
0.36
0.36
0.41
0.39
0.37
0.36
0.40
0.38
0.41
3.82
2.99
3.87
3.77
3.68
4.03
4.74
1.48
1.61
4.19
3.65
3.03
2.99
3.46
0.61
0.73
3.49
0.85
1.13
0.98
0.68 d/
2 RO^
2.bO~
2.05
2.12
0.71
0.71
1.41
3.69
3.67
3.17
0.41
0-*£
Jo
0.56
0.82
0.60
1.04
0.64
0.73
I.I.I
3.51
?.'.(.
0.55
3.25
3. 27
llo,at
Con rent
(Btu/lb A.R.)
(12)
11650
12423
11440
10588
8639
8400
B40'.
8376
8308
8360
8389
8298
8406
8440
11008
10905
10056
108 34
10752
9615
9829
11836
11078
11054
10760
11197
10965
10577
12301
11329
10990
12404
12195
12936
12620 ,
1 3054 -'.
12481-'
11478
11781
13052
12463
12270
10495
10535
10718
9653
9444
9758
10308
10f>(,5
10585
9314
I5R2
!"'.(. 5
I'ISI'I
111 m
(.b(.5
10 '.(I',
lO'.Ofl
Lha Sulfur/
MIIUu
03)"
2.31
l.OH
1.91
2.11
.85
.42
.43
.46
,46
.44
.43
.47
.45
.48
3.47
2.75
3.89
3.48
3.42
4.20
4.85
1.25
1.45
3.80
3 31
2 '. 70
2.73
3.27
.49
.64
3.17
.69
.04
. 76
.5'.
2. 14
2.09
1. 78
1.80
.54
.56
1. 15
3.52
3.48
2.96
.42
3 1
.57
. 79
.65
.98
.('8
. "'
*. '< "
1. i'.
2.'. 7
.112
3. 13
3. 15
RSI) of
l.bn Sulfur/
HMRtu
21.55
21. 76
23.92
18.49
30.07
12. 1(.
I 1.9'.
16.B'I
14. 17
14.36
11.55
9.90
6.51
6.H6
9.08
12. 19
I6.6R
12.23
5.98
14.37
14.46
10.52
32.65
5.67
'..8R
13.'. )
'.. 'Hi
5.30
'.. 0'.
8.89
11.20
20. 7'>
1.2. 14
45.19
14. 36
B. in
(.. 5h
14.47
8. 78
4. >I9
12.75
21.11
4.00
10.01.
1 1. 84
17. 77
IC.5H
15.'IK
4(,. I?
31. 7B
'. 2 . 5 7
15. 37
I !.'> i
1.5'.
1 i 2f
(.. l(.
5. 911
-------
2 ol >
a
i
to
HI IIP
Code
(1)
10046
10047
10048
10049
10050
10051
10052
10053
10054
10055
10056
10057
10058
10059
10060
10061
ionr, i
10062
10062
10063
100*4
10065
10065
10066
10067
IOOAR
11)069
10070
10071
10072
10073
10074
10075
10076
10076
10077
10077
10077e,
10078-1
100/9-
10079
10080
Hum i
10082
IODR3
10084
10085
10086
nrs
Code
(2)
170157
540023
211013
212183
211235
211121
212177
422005
422063
422063
212---
080107
170041
380057
380011
560005
560005
190135
100135
290175
290121
190117
190117
290171
190179
190123
190123
190125
O'i0017
211193
211071
211---
212101
560005
560005
51.0007
500007
560007
080123
080057
080057
422033
422031
422005
540049
540049
422033
422051
10
I
8
9
8
6
9
I
I
9
17
10
21
21
19
19
12
12
15
15
12
12
13
12
12
12
12
18
8
9
19
19
19
19
19
16
16
16
1
1
I
3
3
I
2
0/18408
100008
100009
090000
015703
015703
100010
095203
007102
007102
048913
009900
048408
056901
056400
092600
092600
051700
051700
049202
049202
05)700
051700
048414
053004
100011
100012
100013
100014
046913
095100
095100
100017
100017
100017
076800
004900
004900
100015
1(10015
100016
003004
003604
100016
100031
Conl
Rank
(5)
3
3
3
3
3
3
3
3
3
3
3
3
3
1
1
2
2
3
3
3
3
3
3
3
3
3
3
3
2
3
3
3
3
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
Mining
Method
(6)
I
2
2
4
4
It
/,
I
1
1
2
2
1
2
2
2
2
1
1
2
2
1
1
2
2
2
2
2
2
4
6
It
2
1
1
1
2
2
1
2
2
2
2
2
1
1
2
2
Preparation
Code
(7)
2
9
9
2
1
1
1
1
1
1
1
1
1
1
2
1
1
1
1
1
1
1
1
I
1
1
1
1
1
1
1
1
2
2
2
2
2
1
1
Method of
Sampling
(8)
1
1,4
9
9
9
9
9
9
9
9
9
9
9
1
1
1
1
4
9
9
9
4
9
9
9
9
9
9
1
4
4
4
1
1
4
1
4
1(4
--
--
4
Type of
_&omal£
(9)
4
4
4
3
3
3
3
4
4
4
9
4
4
4
4
3
3
4
9
9
9
4
9
9
9
9
9
9
5
3
3
3
3
3
4
3
4
4
--
--
3
4
4
4
4
4
4
4
AVERAGE CHARACTERISTICS
Volume
(Tonsj
(10)
1378
4958
6251
1390
2133
5964
16916
11370
5306
2685
1343
6838
1967
2254
10227
10467
10715
1377
8627
51280
57120
788
3646
6619
4083
1313
3596
450
12267
1500
1500
100
1422
10000
8632
2959
3442
9342
741
1279
741
907
854
1032
571
1083
Sul fur
Content
(* A. R. )
(11)
3.48
1.99
1.35
2.96
0.97
1.09
3.11 -
2.27
2,39
2.52
3.37
0.59
3.03
0.79
0.80
0.33
0.34
3.44
3.04
5.13
3,87
2.25
3.07
2.72
5.76
4.21
5.55
4.18
0.41
0.74
0.78
0.74
4.25
0.40
0.53
0.92
1.08
1.06
0.21
2.04
1.70
1.19
2.43
2.49
1.54
2.04
llcnt
Com IMI t
(Btu/ll. A.R.)
(12)
10567
11(160
12456
11623
13466
12763
11480
11916
11574
11594
10344
11035
10793
70V. 5
6115
8764
8797
9342
9539
8846
10269
8899
8969
10252
9666
9501
9181
8933
10574
12337
12126
12242
9890
82'.4
8282
10356
10017
10362
10803
12405
12354
12776
13157
13111
12623
12666
l.hs Sulfur/
mill u
O~3)
3.29
1.80
1.08
2.56
.72
.85
2,70
1.90
2.07
2. IP
3.27
.53
2. 81
1. 12
1.10
.37
.38
3.70
3.18
5.80
3.7f,
2.53
3.42
2.66
5.99
4.44
6.12
4.68
.38
.60
.65
.61
4.30
.'18
.f.4
.88
1.08
1.02
"~~
1 . 1.4
1.3B
.93
1.85
1.90
1.23
1.61
KSI> o(
I.I.-. Sul I in/
__ MIIHliiJ : )_
"
7.20
I'-. Hi)
2'>. Ill
2H.01)
/.2. "2
19.50
7. 11
10.9'.
ll.DI'
2". 01
15. m
0.4')
13.01,
2 'i.20
?vn
14.1?
1 2 . IV.
21. M
18.08
3. Id
4.81
?<>. V.
19. 13
20.12
16. H6
13.75
IR.l'l
16.1*8
15.55
.V..1I'
48.30
11. 7(.
10. II
').('.
15.5H
1 7 . no
17.21
24 . 1.'.
20. HI
21.11
26.24
in.no
6.29
II. hS
25.'"'
14. n
-------
Shret 3 of 5
AVEUAGE CIIARACTER1STICS
Mine
Code
(1)
10087
10088
IOOB9
10090
10091
10092
10093
10094
10095
10096
10097
10097
10098
10098
10099
10100
10101
10102
10103
10104
10105
10106
10107
a uiiofl
1 10109
W 101 10
10111
10112
10113
10114
10115
10116
10117
I01IP
10119
10120
10121
10122
10123
1 0 1 24
KU25
101 2(>
10127
10I2R
10129
10130
10131
10132
10133
10134
10115
101 Jd
10117
nniH
1 0 1 39
I01/.0
FITS
Code
(2)
211159
060107
080081
060071
08008 t
170145
170145
170077
170117
212183
170145
170145
350031
350031
560007
180125
170055
170055
170081
170061
170163
060081
4 9000 7
510027
170057
211000
211193
211095
390119
390119
390031
390031
390031
300059
390127
390031
390157
390157
390031
390031
3'.l()157
390031
390079
390157
390067
3900RI
390031
390000
390119
390 r, 7
390000
3000(10
390031
39(1177
3900110
19000(1
USBM
Dlat.
(3)
8
17
17
17
17
10
10
10
10
9
10
10
18
18
19
U
10
10
10
10
10
17
20
8
10
8
6
8
4
4
4
it
4
4
4
4
4
4
4
4
4
4
4
4
4
4
/.
4
4
4
4
4
/,
',
4
t.
Bed
Code
(4)
016811
075000
076900
075701
074502
048408
090002
048406
048408
050609
09000]
090002
047800
047800
039800
100018
046408
048408
046408
048408
048408
100021
100022
100023
046408
100024
100025
100026
003607
008013
008013
008013
008002
002302
100027
008404
008424
0084 24
008424
100028
100029
003604
008002
OOB002
OOBOI3
Coal
Rank
(5)
3
3
3
3
3
3
3
3
3
3
3
3
2
2
2
3
a
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
.1
3
3
3
1
3
3
Mining Preparation Method of
Method Code Sampling
(6) (7) (8)
1 2 1
2 1 4
2 1 4
2 1 4
1 1 4
2 2 1.4,9
2 2 1.4
2 2 1.4,9
1 2 1.4.9
2 2 4.9
2 2 1.4.9
2 2 1
2 1 1
2 1 1.4
2 1 1.4
22 4
1 2 4
1 1 1,4.9
1 2 4
1 2 1.4,9
1 2 4
2 1 1,4
1 2 1,4
1 2 4
1 2 1,4
2
2
2
2
2
2
I
1
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
1
1
1
1,4
1.4
1,4
1,4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1,4
1,4
1,4
1,4
1,4
1.4
1.4
1.4
1,4
1,4
1,4
1,4
1.4
1 ,',
1.4
1,4
1 ',
Type of
S/imple
(9)
3
4
4
4
4
4
4
4
4
4
4
3
3
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
It
4
U
4
U
4
4
4
4
It
It
i,
t,
t,
4
4
it
Volime
(Tona)
(10)
9199
1983
4133
2232
4959
8568
9126
6128
8069
1386
8945
9288
6162
8716
9162
1379
1317
8103
1316
3384
1416
6364
8360
1446
8730
5959
6', 11
6274
1183
1192
3361
2373
947
13B2
1192
1294
1150
1426
1388
1044
121 J
959
1345
12', 8
1)11
1495
1181
11', 3
I56B
1 190
1056
1 1 70
1250
1 11,0
102',
Id'lli
Sulfur
Content
(7. A.R.)
(ID
0.67
O.bl
0.37
0.60
0.4B
2.87
2.92
1.33
3.19
1.87
3.03
3.30
0.44
0.75
0.74
1.22
1.21
2.73
1.90
1.51
3.29
0.56
1.01
1.12
2,90
1.43
1.37
1.15
4.04
3.88
4.67
4.62
5.00
J.52
3.76
3.95
4.44
4.29
4.27
4. 10
3.74
4. IB
4.10
3.91
3.58
3.52
4.11
3.99
3.59
3.92
3. 73
4.11
4. 52
4 . '< 0
2.82
3.9B
Ural
Conri'iit
(Blu/lb A.R.^
(12)
12985
10721
10479
11132
10101
11056
10939
10796
10689
11840
10900
10974
10)56
10402
9997
11095
11998
11076
11514
11604
10821
%28
11204
1318',
11039
11680
11003
11997
11020
11038
10767
10096
10073
11151
10989
11135
11273
11243
11340
11133
11151
11177
11414
JII65
11130
11081
I09B8
11 Jhd
1 1 04 9
1 1 J23
1 1 3B7
11219
1072 1
1 1 192
1 1372
1 ISbl
U>s Sulfur/
MMIItu
(13)
.51
.57
.35
.54
.47
2 . I'd
2.67
1 . 2->
2.99
1.60
2. 7ft
3.00
.43
.73
.75
1. 10
1.02
2. ',8
1.66
1.28
3.05
.58
.90
.85
2.1-3
1.20
1.25
.96
3.67
3.52
4.35
4.59
4.99
3. 16
3.42
3.55
3.95
3.82
3.76
3.69
3.35
3. 75
3.t,l
3.50
3. 22
3. 16
3. 75
3 <-9
'J.l<
3.47
3. 27
1. ('<
',.?(
3 9'«
2 . ', 7
3 . 4 >
RSI) of
1.1, s Sul fui/
MMBtu (7->
O4)
9.49
10. 3',
20.69
5.05
11.51
10.07
9. 11,
51.03
14.07
4 7 . 24
10.47
l>.74
11.70
45.RO
lfl.1'3
17.09
IS.hB
22.35
28.52
12.92
6.94
30. B2
27. hi
11.57
0. 7.B2
19.47
21 . 19
l(>. 99
23. 39
H> . (' 5
24.51
IB. 8h
20. i>2
1 9 . (,4
22. 1 1
1 ', . 7 7
12.2',
2 1 . o 7
22 2^
'. 2H
I ' 22
22 '» i1
/, ' i
l . ' i
H ( 0
5. 91
-------
She i-1
AVEPAGi: CHARACTERISTICS
Ml IIP
Code
10141*'
10142
10143
10144
10145
I01/i6
10147
IOI4H
10149
10150 .
10151*'
I015Z*'
10133
101 54
10155
KM56
10157
10I5H
10159
10160
10161
T 10162. .
A 10163*'
10164
10165. .
10166-'
10167
IOI6B
101692'
10170
10171
10172
101/3
10174
10175
10175
10175
10175
101 76
10177
101 78
10179
101 HO
IOIB1
IOIH2
IOIB3
IOIR4
IOIR5
10186
IOIR7
IOIRR
10IR9
10190
10191
1010?
FlI'S
Code
(2)
390000
390163
390163
390073
390163
390073
390127
390163
390073
390073
390000
390000
390000
390000
390000
390000
390127
390059
390119
390127
390031
390127
390157
390115
390079
390000
390163
390000
390000
390000
390119
390000
390127
390031
390000
390000
390000
390(100
390000
310000
300003
3R0065
211---
212177
1ROI73
IRII021
212177
1ROI53
170185
1ROI73
170057
211119
19001 '1
US DM
Dlst.
(3)
4
4
4
It
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
22
21
8
8
8
9
11
11
9
11
10
II
10
8
4
Bed
Code
(4)
008701
007406
008013
0080 IJ
008501
088701
085R01
008002
092801
100030
100030
100030
100030
100030
100030
100013
100034
100010
048305
050202
100035
100036
0/i8905
048905
04960R
011108
OR1R07
Coal
Rank
(5)
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
2
1
3
3
3
3
3
3
3
3
3
3
3
3
3
Mining Preparation Method of Type of Volume
He 1 hod Code Sampling Snmplc (Tons)
(6) (7
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
3
4
4
4
4
4
4
2
2
9 «
9 <
9 «
2
2 ;
2 ;
2 ;
2 ;
i
2
2 ;
2
2
) (8) (9)
1,4 4
1,4 4
1,4 4
,4 4
,4 4
,4 4
,4 4
.4 4
,4 4
,4 4
,4 4
,4 4
.4 4
.4 4
,4 4
,4 4
,4 4
.4 4
,4 4
.4 4
.4 4
.4 4
.4 4
.4 4
.4 4
,4 4
,4 4
,4 4
,4 4
.4 4
.4 4
,4 4
.4 4
,4 4
I
I 5
4 5
5
3
2 1
» 9 5
> 9 5
> 9 9
4
3
2
3
2
2
3
3
4
4
(10)
989
9R/i
919
960
779
992
849
876
920
1371
1273
1276
128
990
1148
1245
1046
1449
2025
1446
801
1040
883
1526
924
398
512
1016
3757
2443
1696
I0/i2l
5'. 9 5
26411
67B9
7326
5525
6425
1466
R050
6056
4543
4701
4350
31 IV!
Sul fur
Content
(JLA.H.)
(U)
4. 19
3.94
3.64
3.79
3.47
2.68
3.87
3.68
2.99
1.96
2.58
4.53
4.47
3.17
3.49
3.71
3.36
3.81
O.R5
4.00
3.64
3.15
4.49
4.60
3.57
4.35
3.67
3.54
4.05
3.40
3.H2
0.65
0.76
0.69
0.97
1.01
3.97
3.14
4.07
2.23
1.70
1.53
3.56
2.B4
1.20
? . %
llent
Conl PM t
(l)tu/lh A.R.)
(12)
11061
11089
11242
11109
11055
11215
11057
11449
11008
10774
10424
9525
9770
10937
11061
10703
11017
10593
10650
10962
10664
10580
11572
11296
11186
10472
11176
11772
10645
11329
10985
«625
biy,
11663
12219
11416
10239
11039
10989
1 1 006
IORIO
10657
10573
11163
11098
10565
Lbs Sulfur/
HMIItu
(1 3)
D.eo
3.56
3.25
3.43
3. 17
2. 39
3.13
3.20
2.72
1.86
2.49
4.75
4.57
2.">0
3. 16
3.4B
3.07
3.60
.7M
3.71
3.43
2.97
3.90
4.08
3. IB
4.24
3.29
3.00
3.80
3.00
3.48
. 75
1.23
.59
.79
.85
3. BO
2.B5
1. 70
2.03
1.57
1.43
3.3;
2 . 54
I.OH
2. 78
I'SD of
l.hs SuH.it/
IIHIUu (')
(14)
IV 11
IB. 85
i'i.57
24.66
22.71
12.07-
2 (..60
7.01.
14. 74
9.2H
20 . 4 1
0.00
0 (10
10. '.6
Zli.MI
IS. 04
2 1 . (>/.
7. 3''
0.00
20.62
17.81
0.00
15.02
14. 'id
H.fl?
36. 1 1
I/.. 72
1. 15
16. Ifl
IS. 71'
? 1 . 4 B
1 ? . 'i 2
' b . 0 1
IB 61
11.56
1..39
1 . 51
«.?''
10. 7d
!?.'.!
y. . v.
12. 72
1 1.1)6
10.91
28.25
15.41
-------
Sheer 5 of 5
AVERAGE CHARACTERISTICS
Mine
Code
(1)
10193
10194''
10191 '
10196. .
ini97^'
inisfti''
IIM1*^- .
10201-'
10202
1020^'
FII'S
Code
(2)
540039
540039
300083
300017
lonn 1 7
30001 7
300087
300087
300003
USBH
Olst.
(3)
8
22
22
71
))
??
79
37
22
22
Bed
Code
(4)
008402
008402
034500
093100
093100
Coal
Bank
(5)
3
3
1
2
2
2
0
2
2
2
Mining Preparation Method of
Method Code Sampling
(6) (7) (8)
2 1
2 1
2 9
q
q
q
q
9
2 9
2 9
Type of
Sample
(9)
4
4
2
2
1
Volume
(Tons)
(10)
3111
--
--
Sulfur
Content
(X A. R.)
(U)
0.94
0.55
0.75
Meat
Content
(Dtu/lb A.R. )
(12)
11188
6478
9042
Uia Sulfur/
MM II tu
(13)
.84
.8'.
.83
RSI) of
U.s Sulfur/
MMHtu ( )
(14)
18. 14
47. 71
""
19.87
»/ For Interpretation of codea In columns (l)-(9) refer to Appendix £,
li/ Data net not analyzed.
£/ Coal source represented by stack monitoring data, no fuel analyses available.
d/ Dry basin.
c/ Data gels 10204 through 10331 analyzed on USBH District basis only, due to lack of aeam Information and small number of analyses.
Source: Foster Associates, Inc.
-------
APPENDIX I
SAMPLE STATISTICS OF COAL CHARACTER! STICS AliAl.YZF.il IIY I ItlllVI IIHAI. DAIA SI.TS
(FIRST HUN: DATA AS RECEIVED FROM RF.SI'OHUF.NTS, DISREGARDING LOT-SIZE ANALYSIS)
I1 S
I'ROI
I'lF
(1
1
1
1
1
I
1
1
1
1
1
2
3
R.M.
PC INC SPAM
Kid COIif.
1 ~~ (31
001000
007102
007102
100008
095203
007102
007102
100015
100015
1000)6
100016
100031
003604
003604
007402
003607
008013
008013
008013
OOB002
002302
100027
008404
008424
008424
OOB404
100028
100029
003604
008002
008002
008013
008701
007406
0000)3
000013
4 08B101
4 OBR701
4
H 1 NF.
CflllE
"n~
10017
10031
10031
10047
10053
10054
10055
10080
10081
10082
10085
10086
100295''
looaib/
10084
100302/
10113
10114
10115
10116
10117
10118
10119
10120
10121
10122
10123
10124
10125
10126
10127
10128
10129
10130
10131
10132
10133
10134
10135
10136
10137
10138
10139
10140
1014)1'/
10142
10143
10144
10145
10146
10U7
10140
10149
10150
10151>y
I01521'/
10111
10114
mi <;«;
VOLUHF SAMPLED
(Ions)
Avi-mrc RSD(X)
(4)
2055
2824
3291
4958
11370
5306
26B5
1279
741
907
571
1083
11870
1032
10360
1183
1192
3361
2373
946
1382
1192
1294
1149
1426
1388
1044
1212
959
1345
1248
1311
1495
1100
1143
1568
1190
1056
1170
1250
1360
1024
1096
989
984
919
9r>9
779
99?
B49
B/6
920
1371
1271
1 ->7f.
(5)
64.4
50.6
61.7
88.2
41.0
31.7
36.0
89.9
38.7
65.0
47.8
78.4
31.6
66.0
10.0
32.8
33.5
30.9
59.8
47.5
21.7
29.2
22.7
38. 3
22.1
21.9
36.6
26.3
29.7
31.3
27.6
24.1
30.2
31.2
46.7
23.8
36.9
19. 8
18.2
60.1
30.1
45.0
20.0
28.0
20.0
28.3
23.5
40.4
17.1
J-..2
16.4
If. .5
n .<>
20.2
SULFUR CON1F.NT
(Prrconl, As Received)
Avernr.e_ RSH(l)
"~(6)~ ~ (7)
61
04
12
99
27
2.39
2.52
04
70
19
54
04
2.80
2.49
2.60
4.04
3.8B
4.67
4.62
5.00
3.52
.91
.95
.44
.29
.27
.10
3.74
16
10
91
se
51
10
99
se
92
3.73
4.11
4.12
4.40
2.82
3.96
4.19
3.94
3.64
3.79
3.47
2.68
3.117
3.68
I .06
2.17
4.11
34.0
14.2
8.0
15.7
10.0
10.5
23.9
22.0
23.8
30.0
24.9
14.3
7.9
10.1
5.0
31.
17,
20.9
17.0
16.
17.
70.
15.
17.
20.
16.
21.6
16.7
22.4
15.6
20.2
19.
23.
12.
14.9
21.5
19.6
21.0
13.9
19.0
14.B
II.1
5.2
13.6
17.2
17.B
21.3
18.1
11. B
23.0
9.5
11. (i
in.o
16.5
o.n
HEAT CONTENT
(Hlu/lh., As Rrquircd)
AvnrnBt RSD(I).
(B)
LRS SULFUR/NMht.i
Avrr.iiT
11078
11478
11781
11060
11916
11574
11594
12405
12354
12776
12623
12666
13054
13111
12481
11020
1103B
10763
10096
10073
11151
11003
11135
11273
11243
11340
11133
11151
11177
11414
11165
11131
11061
1(19011
11366
11049
11323
113B7
11210
10721
11192
11372
11563
11061
11069
11242
11109
11055
11215
11057
11269
11000
10774
10424
952S
(9)
4.4
2.2
2.1
3.9
2.5
2.0
4.3
2.0
3.B
2.9
4.0
3.2
1.7
2.0
5.6
6.5
5.3
5.7
4.8
6.3
3.9
5.9
4.1
4.0
4.3
4.1
4.0
4.1
4.9
5.3
4.6
4.1
5.3
4.9
3.0
5.0
2.0
2.6
3.2
9.2
5.1
4.5
2.5
3.9
J.fl
1.5
5.6
6.1
1.7
5. 1
2.5
2.0
1.4
5. 1
n n
(10)
1.45
1.78
1.80
1.80
1.91
2.07
2.18
1.64
1.38
0.93
1.23
1.61
2.14
1.90
2.09
3.67
3.51
4.35
4.59
4.99
3.15
3.55
3.55
3.95
3.82
3.76
3.69
3.35
3.70
3.61
3.51
3.22
1.16
3.75
3.49
3.24
3.47
J.27
3.66
4.26
1.94
2.47
3.4r>
3.00
3.56
3.25
3.43
1.17
2.39
3. S3
1.21
2.72
1 .112
2.41
4. l'<
(11)
32.6
14.5
B.B
16.0
10.9
11.1
26.0
21.1
26.2
30.0
25.9
14.7
8.3
11.6
6.6
31.3
17.0
21.7
19.1
18.9
16.8
67.2
15.8
19.5
21.2
17.0
23.4
16.6
24.5
18.9
20.6
19.6
22.1
14. B
12.2
21.1
22.3
20.3
14.2
22.6
16.5
8.6
5.9
15. 3
18. n
19.6
24.7
/2.J
12.1
2fi 6
7. I
M.7
9. 1
20.4
0.0
(H'HIIFR
OF
ANALYSIS
(12)
55
268
220
215
2039
1279
1033
11
17
28
19
21
704
42
275
127
94
417
427
130
495
278
516
63
351
251
102
322
29
164
165
2?3
110
31
5
461
42
19
9
37
10
5
4
2 on
?'il
lr)2
174
19
261
77
2
11
7
I )
I
-------
KAMI'I.F. SfATISllCS OF COAL CIIAKACI I.RISTI CS AHAIWP1) K\ I Illi I V I |I|IA|. DA1A SLIS
(FIRST RUN: DATA AS RECEIVED FIIOM RKSI'OHIIEHI S, IUSRU Alllil l«: I.OI-SI7F AIIAI.YS1S)
ll.S.B.H.
iiisrnci
I))
4
I
NJ
SF.AM
C"|iE
(2)
085001
008002
092801
100030
100030
100030
085802
010401
015109
100001
100004
100005
015104
100006
015108
021202
011108
008402
016200
100009
015703
015703
100012
100013
100014
016811
100023
100024
100025
100026
011108
008402
048913
048415
048415
048415
048415
048415
090000
HI HE
ronr
(3)
10158
10171
10174
10175
10176
10177
10192
10001
10002
10003
10016
10022
10023
10025
10026
10027
10028
10033
10034
10048
10050
10051
10072
10073
10074
10087
10108
101 10
10111
10112
10191
10193
10009
10018
10019
10019
10020
10021
10049
VOLUME
(Ton
Average
(M
1148
924
1016
3757
2443
1696
3104
5318
3150
5688
10024
5060
2594
1800
1100
1600
300
1500
1500
6251
2133
5964
1500
1500
100
9199
1446
5959
6411
6274
4350
3111
6254
4548
3739
5643
4031
4354
1390
SAMI'l.ED
s)
RsnU)
I'M
37.7
14.7
4.5
51.3
49.2
49.6
49.0
24.7
52.8
18.5
11.4
55.8
56.7
c/
c/
c/
c/
c/
c/
12.0
27.2
37.4
c/
c/
c/
13.1
44.0
26.5
6.9
5.5
35.3
51.4
27.7
12.8
14.3
37.0
14.6
17.7
7.1
SULFUR CON1F.MT
(PrrciMil, As Received)
RSDU)
Avprnp.r
(0)
4.47
0.85
4.00
3.64
3.15
4.49
4.60
3.57
4.35
3.67
3.54
4.05
3.40
3.82
2.94
2.69
1.34
2.18
1.48
0.61
0.73
0.85
1.13
0.98
0.68
0.71
1.41
1.35
0.97
1.09
0.74
0.78
0.74
0.67
1.11
1.43
1.37
1.15
0.69
0.97
1.01
1.20
0.94
3.82
4.19
3.65
03
,99
.46
(7)
0.0
10.1
25.8
11.7
17.7
4.1
0.0
10.6
16.3
0.0
12.3
12.9
10.7
29.7
13.7
3.4
15.9
15.0
21
15
22
22
23
10.9
5.0
9.9
19.3
58.1
42.9
12.5
11.2
20.6
24.
41.
18.
2.96
35.4
47.7
12.0
9.6
10.6
29.5
41.7
12.5
18.8
10.8
34.2
28.6
16.7
8.0
4.4
4.2
14.1
4.4
4.7
25.5
IIF.A7 COU1F.HT
(Rlu/lb.
Avr i am
IB)
9770
10937
11061
10703
11017
10593
10650
10962
10664
10580
11572
11296
11186
10472
11176
11772
10645
11329
10985
10565
11650
12423
11440
11636
12301
11329
12404
12195
12936'
12620
12463
12270
12456
13466
12763
12337
12126
12242
12985
13184
11880
11004
11997
11683
12219
11016
11090
11 188
11008
11054
10760
11197
10965
10577
11623
, As Rcqui red )
ItSUU)-
(9)
0.0
2.0
5.0
7.0
5.1
4.1
0.0
9.2
3.8
0.0
3.5
1.7
2.6
5.9
1.0
0.2
5.1
3.0
3.3
3.7
1.8
4.4
3.4
3.6
1.3
2.6
4.7
5.3
3.3
3.8
4.4
2.2
2.8
3.5
3.9
4.9
6.2
3.2
1.0
2.9
2.5
3.6
2.2
3.0
.2.7
4.4
4.1
4.5
2.4
2.3
2.5
1.5
1.4
1.7
3.7
l.BS SHI
A v r r .T £ r
(10)
4.57
2.91
3.16
3.48
3.07
3.60
0.79
3.71
3.43
2.97
3.90
4.08
3.10
4.24
3.29
3.01
3.81
3.01
3'. 4 8
2.78
2.31
1.08
1.91
1.25
0.49
0.64
0.69
0.94
0.76
0.54
0.56
1.15
1.08
0.72
0.85
0.60
0.65
0.61
0.51
0.85
1.20
1.25
0.96
0.59
0.79
0.05
1.08
0.84
3.47
3.80
3.39
2.71
2.73
3.27
2.56
.FUR/MMItiii
iisnm
(ID
0.0
10.5
26."
15.1
21.0
7.3
0.0
20.6
17.8
0.0
15.0
14.5
8.8
36.1
14.7
3.1
16.4
15.8
21.5
15.4
21.6
23.8
23.9
10.5
4.9
8.9
20.8
62.3
45.2
14.4
12.7
21.1
26.0
42.9
19.5
34.4
48.3
13.8
9.5
11.6
30.1
44.0
13.4
18.6
11.6
34.4
28.2
18.1
9.1
5.7
4.9
13.4
5.0
5.3
28.1
M'MHFR
OF
AKAI.VS1S
11?)
1
6
27
45
50
(,
1
5
28
1
in
2
5
5
2
2
1564
398
286
47
45
22
73
30
113
115
55
51
11
10
10
40
119
173
219
6B
50
6
57
26
29
5
-------
SAIU'I.F R1AMKIICS
049202
1149702
019000
049202
0. 1 9202
n'in/no
019702
IM1JII I
04114 14
101)10
10011
1001 2
1001)
10014
10015
lour, i
10064
lonsr,
2396
706
505
712
904
5011
4111,0
50640
6r, 19
58.6
59.6
12.6
64. 2
76.1
ni .0
12.9
79. f,
ni.o
2.99
1.R6
). 71
1.60
4.01
4. M
5.06
1.94
2.69
11.5
9. )
11.7
5.1
12.9
11.9
6.0
9.0
21 . 1
114110
10344
9690
11840
10219
1 101)6
ln50R
10990
10495
10515
11)710
IOS65
10501
10)71
10404
10400
10567
10791
11056
10919
10796
10609
10900
10974
11490
11076
11514
11R04
10021
110)9
10657
11161
11095
11019
in9R9
lunin
10571
95)9
9)42
OP99
09fi9
>(,(<(<
9501
9IHI
119 11
I1US2
1091)5
I005G
11)8)4
10752
9615
'1029
HU96
102)6
10252
2.9
4.2
2.R
4.4
4.5
2.3
2.5
3.3
1.7
3. 1
.1.1
n.7
3.1
1 .4
2.4
2.)
2.1
3.0
2.6
2.5
4.3
3.4
3.2
1.5
2.1
4.7
2.R
2.9
2.7
2.6
4.2
1.1
2.5
1.5
2.0
1.7
J.6
1.7
4.6
4.n
3.)
5. 1
2.2
9.9
4.6
1.0
) 7
B.7
I. \
2.5
4.9
5.6
1.7
2.11
4.0
2.71
1.27
4.31
1.60
3.H9
2.01
2.11
3.17
3.52
3.1B
7.96
3.49
3.34
2.47
3. 13
3.15
1.29
2.R1
2.59
2.67
1.24
2.99
2.78
3. DO
1.07.
2.48
1.66
1.2B
3.04
2.63
1.4)
2.51
l.ll
.'.05
1.71
1.57
1. 17
3. IB
3.71
2.51
1.47
5.9"
4.44
6.12
4.60
0.54
2.70
3.09
).1D
1.42
4. 20
4.05
5. m
J.06
2.6)
7.1
15.0
10.1
47.2
14.5
17.4
10.5
11.2
4.0
10. 1
11. B
2.1
11.9
3.6
6.7
5.9
7.2
13.1
10. 1
9.4
51.0
14.1
11.0
6.7
15.7
22.3
28.5
12.9
8.9
9.8
12.7
10.9
17.1
B.)
10.7
)4.3
14. n
in.i
21.5
70.5
19.1
16.9
13.2
1R.7
16.9
4.0
12.2
16.7
12.2
5.6
14.4
14.5
7. 1
12.11
22. 7
111
7IIB
41
41
254
1711
126
lf.6
fi)
in;
M
147
77
in
71 1
21 )
590
2)1
14
121)
31)
147
1)1
269
)B
161
in
170
64
19
71B
152
74
U,7
211
211
1)5
Vi
51
15
27
7)
17
n
5
Ii
r
i'
i<
2
2(
2'
2'
?'
ml
-------
SAMI'LF. STATISTICS OF COAI, CIIARACI F.RI STI CS ANAI.Y7.F.II BY I Kill l'| IMIAI. HAIA SMS
(F1KST HUH: UA1 A AS RECEIVED KHOH RF.SI'OHIil'.NTS, DIKKKCAKHIIIG LOT-SUE ANALYSIS)
n.s.n n.
I'KOIMK UK:
HIM Kin
HI
16
16
16
17
17
17
17
17
17
ia
18
ia
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
VOIUME SAMI'I.F.D
ST.AH
COPE
12)
004900
009900
075000
076900
075701
074502
100021
100011
047800
047800
095100
095100
095100
095100
095100
095100
100003
100003
100002
036555
036555
081700
081700
078300
100007
092600
092600
095100
095100
100017
100017
100017
039800
MINE
CODE
13)
10078^
10079^
10079
10057
1008B
10089
10090
10091
10106
10071
10098
10098
10006
10006
10006
10006
10006
10006
10007
10007
1000B
10038
10038
10019
10039
10040
10041
10061
10061
10076
10076
10077
10077
10077
10099
(Tons)
Av'ri Age
(4)
741
6838
1984
4133
2232
4959
6364
12267
6162
8716
CORE
11035
10402
10798
10513
10147
10800
10625
11692
10000
8746
10000
9056
10000
10000
10467
10710
10000
8632
2959
3442
9342
9162
KsnU)
(5)
95.5
IB. B
31.0
38.6
45.8
29.1
16.9
33.0
5.3
13.3
2.0
4.8
3.5
5.8
22.1
12.0
5.5
0.7
£/
7.3
£/
7.5
£/
c/
4.1
3.9
£/
9.9
42.9
40.0
4.B
8.1
SULFUR
(Percent
Average
16)
0.21
0.59
0.61
0.37
0.60
0.48
0.56
0.41
0.44
0.75
0.41
0.39
0.36
0.37
0.36
0.36
0.40
0.38
0.41
0.56
0.82
0.69
1.04
0.64
0.73
0.33
.33
0.40
0.53
0.92
1.08
1.05
0.74
COHTF.NT
, As Received)
RSUU)
(7)
19.7
6.4
10.7
21.4
5.3
11.3
41.9
12.0
11.3
46. B
17.2
14.5
12.0
14.9
12.2
12.9
10.3
6.8
7.6
14.6
47.9
31.0
43.0
15.0
6.5
14.1
12.0
11.2
36.1
16.8
16.2
26.8
17.7
20
100022
10107
8360
22.6
1.01
26.9
21
21
21
21
22
22
22
22
?2
056100
056901
056400
oflonoo
080800
069802
069B02
100033
10044
10059
10060
10179&/
10005
10005J?/
10037
10037*?/
10178
77192
2254
10227
CORE
B423
10000
10421
25.8
23.2
10.1
4.2
-'
0.9
0.55
0.79
0.80
0.74
0.41
0.65
33.3
24.9
25.6
29.8
17.4
11.8
IIF.AT CONTENT
(Hi ii/lb. , As Ri-qui red )
Avcrfi-r RSD(Z).
(8) (9~)
10803
11204
6665
7045
6115
2.3
2.a
1.)
1.9
3.0
i.ns sui.niR/iriBiu
Avri nr-c . RSI)( ! )
110) (11)
0. 19
0.90
0.82
1.12
1.31
20.8
27.6
33.3
25.2
25.9
IH'IIHF.R
OF
ANALYSIS
73
11035
10721
10479
J1132
10100
9627
10574
10156
10402
8376
8308
B388
8360
8404
B400
8298
8406
B440
975B
10308
10665
10585
9314
9582
8764
B797
B244
8282
10356
10017
10362
9997
1.2
3.0
3.1
5.0
2.3
4.0
3.5
2.7
6.4
1.7
1.2
1.1
1.9
1.4
1.4
1.8
1.4
3.0
2.7
4.4
2.1
3.8
1.0
1.6
1.0
1.2
2.9
3.0
2.B
2.1
3.4
4.2
0.53
0.57
0.35
0.54
0.47
0.58
0.38
0.43
0.73
0.48
0.46
0.43
0.44
0.43
0.42
0.47
0.45
0.4B
0.57
0.79
0.65
0.98
0.68
0.76
0.37
0.38
0.48
0.64
0.08
1.08
1.02
0.75
6.5
10.3
20.7
5.0
11.5
39.8
15.6
11.7
45.8
16.9
14.2
11.6
14.4
11.9
12.4
9.9
6.5
6.9
16.0
46.1
31.8
42.6
15.3
7.4
14.2
12.0
9.6
35.6
17.0
17.2
24.6
IB.fi
112
272
92
26
81
19
203
41
55
1537
33
49
213
64
1760
302
10
99
140
29
30
2B
206
42
47
52
10
15
169
74
51
23
44
1216
770
8639
9653
8625
2.4
1.0
1.5
O.B5
0.42
0.75
30.1
17.8
12.4
5R6
5S5
19
9/Analynpo on "dry" limits
iils not pel fanned due to limited number of observations.
volumor. were not piovldcd, volume Indicates approximate size at each shipment.
Kovitrpt Fontrr Ar.r.ocLotnR, Inc.
-------
APPENDIX J
ANALYSIS OF DATA ON AN AGGREGATE
BASIS BY PRODUCING DISTRICT
This Appendix sets out, in detail, the analyses performed
on an aggregate basis in order to examine coal sulfur varia-
bilities within individual Producing Districts. In general,
for each Producing District the following analyses, with and
without lot-size intervals, were performed:
. All coals
Raw coals
Washed coals
Selected coal seams
In addition, the quantity of data for some Producing
Districts permitted more detailed analysis of other factors
which may be possible sources of coal sulfur variability.
These factors are noted in the following discussion.
1.0 Producing District 1 (Pennsylvania)
1.1 Raw and Washed Coals
An aggregate comparison for all coals in Producing
District 1 indicated an average of 1.83 Ibs S/MMBtu
with an RSD of 26.18 percent. As expected, when the raw and
washed coals were analyzed separately, both the average Ibs
S/MMBtu and the RSD were lower for the washed coals.
These results are summarized in Table J-l.
J-l
-------
TABLE J-l
PRODUCING DISTRICT 1: ALL COALS, RAW, AND WASHED
_ Lbs S/MMBtu Number of
Type of Coal Average RSD (%) Observations
All Coals 1.83 26.18 6,619
Raw 1.86 26.22 5,738
Washed 1.65 22.78 881
Source: Schedule J-l, found at end of Appendix J.
As shown in Schedule J-l at the end of this Appendix,
analyses of these data by lot-size intervals exhibit a strong
relationship between RSD and lot-size, especially in the case
of all coals. An examination of the lot-size analyses for
all coals in Producing District 1 shows a steady decline in
RSD from 42.5 percent at the 200-500 ton interval to 9.3 per-
cent at the 9,000-10,000 ton interval, while slight increases
(to 10.4 percent) are observed in the intervals 11,000-13,000,
13,000-15,000, and 19,000-23,000 tons.
1.2 Upper Freeport Seam, Raw and Washed
Within Producing District 1, a comparison was also made
of the raw and washed analyses for the Upper Freeport seam
(USBM Code 071) . Again, and as expected, both the average
Ibs S/MMBtu and RSD were lower for the washed coals.
For the washed Upper Freeport seam there appears to be no
relationship between lot-size and RSD over the tonnage
ranges analyzed (0 to 5500 tons).
2.0 Producing District 4 (Ohio)
2.1 Raw and "As-Burned" Coals
In the case of Producing District 4, insufficient data
were available for an analysis of washed coals. The raw
J-2
-------
coals (as delivered) on an aggregate basis exhibit an average
of 3.58 Ibs S/MMBtu with an RSD of 24.8 percent. As a
comparison, analyses for these coals on an "as-burned" or
"as-fired" basis indicate an average of 3.62 Ibs S/MMBtu
with an RSD of 19.0 percent. With the exception of the 200
to 500 ton interval for the "as-burned" coals, both the "as-
delivered" and "as-burned" aggregate for raw coals exhibit a
general decline in RSD with increasing lot-sizes. As expected,
due to the mixing of coals in stockpiles, handling, feeding,
etc., the raw "as-burned" coals exhibit lower RSD's or less
variation than the raw "as-delivered" coa^s.
2.2 "As-Burned" Coals by Individual generating Units
The available data for the "as-burned" coals in Producing
District 4 permits a more detailed examination of variability.
These coals, from the same general sources, were burned in
one utility's six generating units. The analyses for units
1, 2, and 3 were reported on a composite basis, while separate
analyses were available for units 4, 5, and 6. In general,
each observation represents the volume of coal burned during
a 24-hour period. The statistics for these units are pro-
vided in Table J-2.
TABLE J-2
PRODUCING DISTRICT 4: "AS-BURNED"
Lbs S/MMBtu Number of
Unit Number Average RSD (%) Observations
1, 2, and 3 4.04 15.34 614
4 3.79 16.10 402
5 3.53 14.34 410
6 3.65 16.12 139
Source: Schedule J-l.
J-3
-------
The results for the various generating units indicate
relatively consistent RSD's ranging from 14 to 16 percent.
However, within the units there does not appear to be a rela-
tionship between RSD and lot-size, possibly due to the rather
limited ranges in lot-size (tons burned/day).
2.3 Pittsburgh, Middle Kittanning, and Lower Kittanning Seams
Within Producing District 4, three seams Pittsburgh
(036), Middle Kittanning (080) , and Lower Kittanning (084)
were examined on an aggregate basis. These seam data reflect
raw coals from various mines in Ohio.
The data for the Pittsburgh seam do. not exhibit an
inverse relationship between RSD and lot-size. This may be
due to the limited number of observations (182) and the rela-
tively narrow range of lot-sizes (600-2100 tons).
The Middle Kittanning and Lower Kittanning seams both
exhibit a general decline in RSD with increasing lot-size.
A comparison of the aggregate data for these seams is set
out in Table J-3. These data indicate that the relationship
between RSD and lot-size differs between the seams. The
slope of RSD as a function of lot-size is greater for the
Middle Kittanning seam, resulting in a relatively greater
decrease in RSD for an equal increase in lot-size.
3.0 Producing District 8 (Eastern Kentucky)
In Producing District 8, aggregate comparisons were per-
formed for raw and washed coals and for the Upper Elkhorn
No. 3 (151) and Blue Gem (157) seams.
J-4
-------
TABLE J-3
COMPARISON OF AGGREGATE DATA FOR THE
MIDDLE AND LOWER KITTANNING SEAMS IN OHIO
Lot-Size Interval
(Tons)
Mid-
Point
440
456
763
1,015
1,115
1,208
1,407
1,425
1,582
1,745
2,137
2,470
2,796
3,441
4,357
Range
250-600
300-600
600-900
900-1,100
1,000-1,300
1,100-1,300
1,300-1,500
1,300-1,600
1,500-1,800
1,600-1,900
2,000-2,300
2,000-3,000
2,600-3,000
3,000-4,000
4,000-5,000
RSD of Lbs S/MMBtu (%)
Middle Lower
Kittanningl/ Kittanningj/
33.68
31.57
31.96
26.07
28.40
22.61
18.21
20.56
21.09
20.74
20.23
19.87
18.28
Number of
Observa-
tions
29
90
120
105
426
188
517
394
91
198
101
22
89
229
142
I/ USBM Code 080
y USBM Code 084
Source: Schedule J-l.
3.1 Raw and Washed
The aggregate analyses for all coals in Producing
District 8 indicate an average of 0.83 Ibs S/MMBtu with
an RSD of 47.7 percent. A comparison of the raw and washed
coals indicates that the washed coals have a lower Ibs
S/MMBtu as well as a lower RSD. The raw coals averaged
1.00 Ibs S/MMBtu with an RSD of 51.0 percent, while the
washed coals averaged 0.65 Ibs S/MMBtu with an RSD of
38.5 percent.
An analysis of lot-size for the composite of all coals
and raw and washed coals individually failed to demonstrate
J-5
-------
any significant inverse relationship between RSD and lot-
size.
3.2 Upper Elkhorn No. 3 (151) and Blue Gem (157) Seams
Data for these seams show an average of 0.77 and 0.79
Ibs 5/MMBtu for the Upper Elkhorn No. 3 and Blue Gem
seams, respectively- An analysis by lot-size indicates an
inverse relationship between RSD and lot-size for both
seams. The RSD's for the smallest lot-sizes analyzed for
these seams are relatively high. The 900-1400 ton interval
for the Upper Elkhorn No. 3 seam indicated an RSD of 60.7
percent, while the 1000-2000 ton interval of the Blue Gem
seam indicated an RSD of 42.5.
4.0 Producing District 9 (Western Kentucky)
4.1 Raw and Washed Coals
Analysis of all coals in Producing District 9 yielded
an average of 2.84 Ibs ;3/MMBtu with ah RSD of 26.8 per-
cent. As expected, the washed coals exhibited both a lower
average Ibs 3/MMBtu and a lower RSD compared to raw
coals.
Set out in Table J-4 is a summary of the results obtained
from the lot-size analysis of the raw and washed coals in
Producing District 9. For both the raw and washed coals,
there appears to be a definite inverse relationship between
RSD and lot-size. At the smallest lot-size examined (1000-
1500 tons), there appears to be a significant difference in
the RSD's of the raw and washed coals. The raw coals in
this interval exhibit an RSD of 30 percent compared to 15
percent for washed coals. However, at the larger lot-sizes,
J-6
-------
it appears that the difference between the RSD's of raw and
washed coals become less significant.
TABLE J-4
COMPARISON OF RSD VS. LOT-SIZE FOR ALL COALS, RAW
COALS, AND WASHED COALS FOR PRODUCING DISTRICT 9
Lot-Size Interval
(Tons)
RSD of Lbs
S/MMBtu (%)
Mid-
Point
1,359
1,390
1,422
1,549
1,562
1,634
2,479
2,521
3,451
3,454
4,281
4,436
4,516
5,508
6,921
6,924
7,615
14,894
24,047
Range
1,000-1,500
1,000-1.500
1,000-1,500
1,500-2,000
1,500-2,000
1,500-2,000
2,000-3,000
2,000-3,000
3,000-4,000
3,000-4,000
4,000-6,000
4,000-6,000
4,000-5,000
5,000-6,000
6,000-7,000
6,000-8,000
6,000-10,000
10,000-20,000
20,000-30,000
All
Coals
29.35
29.00
15.55
8.42
11.61
9.01
Raw
30.39
Washed
15.05
14.61
21.53
11.78
7.88
8.43
9.16
8.73
9.09
9.83
6.64
5.57
Number of
Observa-
tion s^./
412
1,553
861
427
516
81
46
58
143
140
398
660
225
45
840
865
37
59
49
I/ Excludes all intervals with less than 30 observations.
Source: Schedule J-l.
4.2 No. 11 Seam (484)
\
The No. 11 seam in western Kentucky, which is equivalent
to the No. 6 seam in Illinois, was also examined. A compari-
son of the raw and washed coals from this seam indicates a
lower Ibs S/MMBtu and a lower RSD for the washed coals.
Sufficierit data were not available for an examination of the
relationship between RSD and lot-size.
J-7
-------
5.0 Producing District 10 (Illinois)
5.1 All Coals: Washed
Data for Producing District 10 are based almost entirely
on washed coals. An analysis of all coals yielded an average
of 2.65 Ibs S/MMBtu with an RSD of 29.5 percent.
Although substantial data were available for a lot-size
analysis in the tonnage ranges from 1,000 to 12,000 tons/
the results of this analysis do not support the inverse rela-
tionship between RSD and lot-size. The lot-size intervals
and the calculated RSD's for these coals are set out in
Table J-5.
TABLE J-5
COMPARISON OF RSD AND LOT-SIZES FOR WASHED ILLINOIS COALS
Lot-Size Interval (Tons) RSD of Lbs Number of
Mid-Point Range S/MMBtu (%) Observations
1,345 1,000-1,500 28.60 994
1,715 1,500-2,000 14.52 264
2,143 2,000-2,500 14.29 187
2,870 2,500-3,500 21.17 158
4,220 3,500-5,000 32.84 274
6,060 5,000-7,000 24.17 436
8,224 7,000-9,000 37.74 1,040
9,812 9,000-12,000 24.81 1,203
Source: Schedule J-l.
5.2 Illinois Seam No. 6 (484) and Seams No. 5 and No. 6 (900)
In Producing District 10, separate analyses were performed
for the Illinois No. 6 seam and a composite of the No. 5 and
No. 6 seams. Data for the composite of the No. 5 and No. 6
seams were obtained from mines engaged in multiple seam opera-
tions producing a mixed product.
J-8
-------
Analysis of the data for the No. 6 seam indicated an
average of 2.67 Ibs S/MMBtu with an RSD of 34.23 percent.
The results of the lot-size analyses for this seam, which are
set out in Table J-6, did not demonstrate an inverse relation-
ship between RSD and lot-size. In fact, the larger lot-sizes
exhibited substantially greater RSD's than the smaller lot-
sizes. An examination of the data indicated that the inter-
vals exhibiting the larger RSD's also had relatively lower
means for the Ibs S/MMBtu.
TABLE J-6
COMPARISON OF RSD AND LOT-SIZES FOR ILLINOIS SEAM NO. 6
Lot-Size Interval (Tons) Lbs S/MMBtu Number of
Mid-Point Range Average RSD (%) Observations
1,388 1,000-2,000 2.85 28.76 993
2,354 2,000-3,000 2.89 13.57 172
3,971 3,000-5,000 2.93 16.61 72
5,897 5,000-6,200 3.33 14.31 186
7,262 6,200-7,800 2.20 43.08 155
8,369 7,800-9,000 2.17 45.62 461
9,581 9,000-10,000 2.40 45.28 204
Source: Schedule J-l.
The data sets for the Illinois No. 6 seam were analyzed
a second time, based on the average sulfur content of the
coals. Two categories of data sets were analyzed: (1)
average sulfur content less than two percent, and (2) aver-
age sulfur content greater than two percent. This analysis
provided an RSD of 40.8 percent for the data sets with an
average sulfur content of less than two percent and an RSD
of 13.3 percent for the data sets with an average sulfur
content of greater than two percent.
This, analysis of RSD versus average sulfur content was
extended to other data sets in the Mid-Continent producing
J-9
-------
area (Illinois, Indiana, and western Kentucky). The results
of this analysis are discussed in Section 13.0 of this Appendix.
The aggregate analysis for the composite of the Illinois
No. 5 and No. 6 seams yielded an average of 2.75 Ibs S/MMBtu
and an RSD of 18.2 percent. A comparison of the composite with
the Illinois No. 6 seam reveals that the composite has a
slightly higher average Ibs S/MMBtu but a significantly
lower RSD as set out in Table J-7.
TABLE J-7
COMPARISON OF AVERAGE LBS S/MMBTU AND RSD FOR
ILLINOIS SEAM NO. 6 AND A COMPOSITE OF
ILLINOIS SEAMS NO. 5 AND NO. 6
Lbs S/MMBtu
Seam
Illinois No. 6
Illinois No. 5 and No. 6
Source: Schedule J-l.
Average
2.67
2.75
RSD (%)
34.2
18.2
As in the case of the Illinois No. 6 seam, the composite
of the No. 5 and No. 6 seams failed to demonstrate an inverse
relationship between RSD and lot-size.
5.0 Producing District 11 (Indiana)
Aggregate analyses for Producing District 11 coals were
examined only with respect to sulfur contents. As in the
case of the Illinois No. 6 seam, the Indiana data sets were
individually analyzed for average sulfur contents less than
and greater than two percent.
Data sets with an average sulfur content less than two
percent had an average of 1.5 Ibs S/MMBtu with an RSD
J-10
-------
of 35 percent, while data sets with sulfur contents greater
than two percent exhibited an average of 3.25 Ibs S/MMBtu
with an RSD of 16 percent. Both cases failed to demonstrate
an inverse relationship between RSD and lot-size.
6.0 Producing district 12 (Iowa)
The data available for Producing District 12 limited
the analyses to an aggregate of all coals and the Lucas County
No. 5 seam (517). The average Ibs S/MMBtu for all coals
was 3.88 with an RSD of 32.5 percent. Data for the Lucas
County No. 5 seam indicated 3.35 Ibs S/MMBtu and an RSD
of 22.6 percent. In both cases insufficient data precluded
a lot-size analysis.
7.0 Producing District 15 (Kansas and Missouri)
In Producing District 15 only data for the Bevier seam
(492) were analyzed. Raw coals from this seam exhibited an
average of 4.29 Ibs S/MMBtu with an RSD of 25.8 percent,
while the washed coals contained 3.38 Ibs S/MMBtu with
an RSD of 8.5 percent. The available data did not permit a
lot-size analysis.
8.0 Producing District 17 (Colorado)
An aggregate analysis of all coals in Producing District 17
indicated an average of 0.51 Ibs S/MMBtu with an RSD of
21.9 percent. In the interval analysis of these coals, RSD's
from 11.7 to 25.8 percent were obtained. However, an inverse
relationship between RSD and lot-size was not observed.
J-ll
-------
9.0 Producing District 18 (Arizona and New Mexico)
9.1 All Coals
Data for Producing District 18 consisted of three data
sets representing two mines. An aggregate analysis of these
data yielded an average of 0.45 Ibs S/MMBtu with an
RSD of 44.1 percent.
9.2 Automatic ASTM Samples
Within these data, the samples that were collected by
automatic ASTM samplers were separately analyzed. These
samples indicated an average of 0.43 Ibs S/MMBtu with
an RSD of 35.7 percent. An analysis of the automatic, ASTM
samples by lot-size indicated RSD's ranging from 12.6 to 50.2
percent but failed to demonstrate an inverse relationship
between RSD and lot-size.
9.3 Automatic ASTM Samples and Analyses, by Laboratory
A special analysis of Producing District 18 coals
focused on the automatic, ASTM samples collected at the
Navajo Power Plant. These samples were analyzed by two
different laboratories Salt River Project (SRP) and
Commercial Testing and Engineering (CTE). The results of
the analysis of these data are set out in Table J-8. Suffi-
cient data were not available to perform a lot-size analysis.
J-12
-------
TABLE J-8
COMPARISON OF NAVAJO POWER PLANT AUTOMATIC, ASTM
ANALYSES PERFORMED BY DIFFERENT LABORATORIES
Lbs S/MMBtu Number of
Laboratory Average RSD (%) Observations
SRP 0.37 14.23 114
CTE 0.40 15.77 89
Source: Schedule J-l.
10.0 Producing District 19 (Wyoming)
10.1 All Coals
The aggregate analysis of all coals for Producing
District 19 indicated an average of 0.51 Ibs S/MMBtu
with an RSD of 37.7 percent. A lot-size analysis of these
data failed to indicate an inverse relationship between RSD
and lot-size.
10.2 Comparison of All Coals to Individual Mines
Set out in Table J-9 is a comparison of all Producing
District 19 coals to the individual mines from which the
coal analyses were obtained. From Table J-9 it can be seen
that there is substantial variation in the average Ibs
S/MMBtu and the RSD among the various mines. Although
the average Ibs S/MMBtu for Producing District 19 was
found to be 0.51, individual mines exhibit a range from 0.38
to 0.96 Ibs S/MMBtu. A similar comparison of the RSD's
shows 37.7 percent for all coals in Producing District 19,
while individual mines exhibit RSD's ranging from 9.9 to
45.4 percent.
J-13
-------
TABLE J-9
COMPARISON OF PRODUCING DISTRICT 19 COALS TO
INDIVIDUAL MINES WITHIN PRODUCING DISTRICT 19
Lbs S/MMBtu Number of
Data Source
Producing District 19, Total
Mine 10006
Mine 10007
Mine 10038
Mine 10039
Mine 10061
Mine 10076
Mine 10077
Average
0.51
0.43
0.47
0.61
0.81
0.38
0.57
0.96
RSD (%)
37.7
13.6
9.9
31.3
45.4
13.1
34.1
20.9
Observations
3,526
2,199
312
169
58
99
25
294
Source: Schedule J-l.
11.0 Producing District 21 (North Dakota)
An aggregate analysis of the lignite coals of North
Dakota indicated an average of 1.18 Ibs S/MMBtu with an
RSD of 27.4 percent. The limited data available for a lot-
size analysis failed to reveal an inverse relationship
between RSD and lot-size.
12.0 Producing District 22 (Montana)
An aggregate analysis for Montana coals indicated an
average of 0.65 Ibs S/MMBtu with an RSD of 44 percent.
Sufficient data were not available for a lot-size analysis.
13.0 Relationship of RSD of Lbs S/MMBtu to Average
Sulfur Content in the Mid-Continent Producing Area
In the analysis of the Illinois No. 6 seam, it was
observed that the data sets with lower average sulfur con-
tents exhibited relatively higher RSD's for Ibs S/MMBtu.
This type of analysis was extended to include all data sets
J-14
-------
in the Mid-Continent producing area (Illinois, Indiana,
and western Kentucky) .
A basic assumption in this analysis is that coal sulfur
contents and heat contents are independent. Numerous data
sets were analyzed and the results of these analyses support
this assumption.
The definition of RSD of Ibs S/MMBtu is the stand-
ard deviation divided by the mean.
RSD (Ibs Sulfur/MMBtu) = S/MMBtu)
X (Ibs S/MMBtu)
From this definition, it can be seen that if the stand-
ard deviation of coals were relatively constant, an increase
in the average Ibs S/MMBtu (or average sulfur content)
would result in a lower RSD, while a decrease in the average
Ibs S/MMBtu (or average sulfur content) would result in
a higher RSD.
The basic problem examined is the impact of the standard
deviation on the RSD for various levels of the average Ibs
S/MMBtu (or average sulfur contents) .
As shown on Figure J-l, the RSD of Ibs S/MMBtu was
plotted against the average sulfur content for each of 44
data sets representing coals from Illinois, Indiana, and
western Kentucy. These coals are primarily washed coals;
raw coals are designed by the solid points plotted on Figure
J-l. A linear regression of the plots for the 44 data sets
indicates that the RSD of Ibs S/MMBtu decreases with an
increase (in the average sulfur content. The particular slope
of the regression line indicates that the standard deviations
J-15
-------
Figure
55.Q
49.5
44.0-
38.5-
D
+J
33.0
CO
w
^ 27.5
m
O
P
S'22.0
16.5
11.0
5.0
COMPARISON OF RSD OF LBS. S/MMBtu AND AVERAGE SULFUR CONTENT FOR
44 DATA SETS IN ILLINOIS, INDIANA, AND WESTERN KENTUCKY
1.0
2.0 3.0
AVERAGE SULFUR CONTENT (%)
4.0
5.0
-------
are relatively greater for the lower average sulfur content
coals.
With respect to compliance with sulfur-limiting regula-
tions, it is interesting to examine the consequences of the
relationship in Figure J-l. First, for these Mid-Continent
coals, there appears to be a trade-off between average sulfur
content and variability among the mines. Given a specific
I
sulfur emissions level, the selection of a lower sulfur source
would require an assessment of the impact of increased varia-
bility. The attractiveness of a lower sulfur source may be
mitigated to some extent due to intermittent periods of
excess emissions, although the average level of emissions
would be lower than those resulting from a higher sulfur con-
tent coal.
A second and related consequence of Figure J-l concerns
alternative strategies for obtaining compliance coals through
washing. In general, comparisons of raw and washed coals have
shown that within individual mines and coal seams washing
reduces the average sulfur content and reduces the RSD of the
Ibs S/MMBtu. However, Figure J-l indicates that the net
benefits of the reduction in relative variability from coal
washing may be less than expected as progressively lower
sulfur coals are selected.
14.0 Summary of Analysis of Data on an Aggregate Basis
Although no defensible conclusions resulted from these
aggregate analyses, it is possible to make some observations
about relationships observed in the various Producing Districts,
On the basic question of the relationship of the RSD of
Ibs S/MMBtu to lot-size, the results were inconclusive.
J-17
-------
However, it appears that, in general, as the level of aggre-
gation increases from mine to seam to Producing District the
data increasingly tend to exhibit an inverse relationship
between RSD and lot-size.
Within the individual Producing Districts and coal
seams analyzed, the washed coals consistently exhibited both
a lower Ibs S/MMBtu and a lower RSD than the raw coals.
This tends to support the hypothesis that coal washing would
reduce the average level of sulfur emissions as well as the
relative variability of the emissions.
Based on the limited number of cases analyzed, it appears
that substantial differences in RSD's can exist among seams
within the same Producing District. Previous sections of
this report indicated that substantial, inconsistent differ-
ences were observed among individual mines and among lot-
sizes within mines, on a Producing District basis. These
observations are reason for serious concern with respect to
the extent of the relationship between RSD and lot-size and
the existence of a simple relationship which can accurately
generalize coal sulfur variability.
Finally, some of the data on an aggregate basis exhibited
rather large RSD's, some in excess of 60 percent. RSD's of
this magnitude could have a substantial impact on compliance
with sulfur dioxide emissions regulations. This suggests
that coal consumers subject to a given emissions limit with
only marginally acceptable coals must selectively evaluate
the alternative sources of supply and may find it necessary
to exclude those sources exhibiting large variabilities.
J-18
-------
Schedule j-i
SAMPLE STATISTICS OF COAL CHARACTERISTICS ANALYZED DY VARIOUS LEVELS OF AGGREGATION
(INCLUDES LOT-SIZE ANALYSIS)
Shoot 1 nl 7
C-l
I
M
VO
nata Dcncr Iptlon:
USDM District, Seam, Preparation
lat-Slzea, etc.
Diet.
Dlst.
Dint.
Dlst.
Dlst.
out.
Dlst.
Dlst.
Dint.
Dlst.
Dlst.
Dlsr.
Dlst.
Dlst.
Dlst.
Dlst.
Dlst.
DlRt.
Dlst.
Dlst.
Dlot.
Dlst.
Dint.
Dlst.
Olst.
Dlst.
Dlst.
Pint.
Dlst.
nisi .
Dlst.
Dint.
Dint.
DlRt.
Dint.
nint.
Dlst.
Dlat.
01 '.i.
nlnl .
Dint.
[)!".!.
III".! .
Dint.
01,
01,
01,
oik.
01,
01,
01,
01.
01,
01,
01,
01,
01,
01,
01,
01.
01,
01,
01,
01,
01,
01.
01,
01,
01,
01,
01,
01,
01,
01,
01,
ni,
01,
01,
01,
01,
01,
01.
01 ,
01,
01,
04,
04,
04,
(1)
All Coalfl
All Coals, 200-500 T
All Coals, 500-750 T
All Coals, 750-1000 T
All Coals, 1000-2000 T
All Coals, 2000-3000 T
All Coals, 4000-6000 T
All Coals, 7000-9000 T
All Coals, 9000-11000 T
All Coals, 11000-13000 T
All Coals, 13000-15000 T
All Coals, 19000-23000 T
Raw
Raw, 100-300 T
Raw, 300-500 T
Raw, 500-750 T
Raw, 1000-1200 T
Raw, 2000-3000 T
Raw, 3000-4000 T
Raw, 4000-5000 T
Raw, 5000-7000 T
Raw, 7000-9000 T
Raw, 9000-11000 T
Raw, 11000-14000 T
Washed
Hashed, 100-300 T
Wander), 300-500 T
Wnnliod, 500-1000 T
Washed, 1000-1500 T
Washer), 1500-2000 T
Washed, 2000-3000 T
Washcil, 3000-4000 T
Washed, 4000-6000 T
Wnnhrd, 6000-9000 T
Sean 071, Raw
Seam 071, Washed
Seam 07), Washed, 0-500 T
Seam 071, 500-1000 T
Sonm 071, 1000-2000 T
Senro 071, 1000-4000 T
Sf.im 071, 4000-5100 T
P.Tw/Ool Ivorod
n.iw/IVIvit. , 2(10-500 T
nnw/Dolvrl , 100-750 T
Volume Sampled
(tons)
Average RSD(t)
(2)
5677
339
612
861
1509
2506
4943
7832
10001
12036
14032
20088
6111
169
397
605
1097
3479
4496
5937
7833
9994
12576
2851
2C6
415
757
1228
1713
2506
34?4
4799
7200
4134
3014
401
756
154f,
3424
47)1
MSB
370
f|J4
<3)
88.51
26.48
11.02
8.70
18.87
11.52
11.65
7.24
6.08
4.79
4.07
3.99
84.85
32.46
14.45
11.10
5.38
8.17
6.40
9.26
7.29
6.11
6.91
81.25
15.92
14.88
16.28
11.74
8.85'
11.35
8.63
10.75
9.89
46.39
57.31
12.36
18.71
17.20
8.70
S.62
57.84
22. f,)
U.27
Sulfur Content
(Percent, As Received)
Average RSD(t)
(4)
2.16
1.43
1.69
1.89
2.00
2.28
2.33
2.32
2.28
2.29
2.28
2.22
2.19
1.36
1.52
1.66
1.93
2.37
2.41
2.40-
2.34
2.33
2.28
2.29
1.96
1.53
1.58
1.82
1.97
2.08
1.99
2.10
2.02
2.10
2.44
2.07
2.19
2.07
2.09
2. 11
2.04
3.92
4.02
3.14
(5)
24.61
41.25
38.54
31.25
28.55
25.38
15.88
11.46
8.94
9.15
9.05
9.77
24.60
38.08
40.93
40.78
29.09
25.11
22.55
16.70
11.12
10.78
8.87
9.24
22.12
22.29
39.75
28.23
22.64
15.39
20.14
15.05
17.53
17.57
ID. 31
11.88
11.12
12.60
10.64
11.50
13.42
22.60
24.29
20.96
Heat Content
(Dtu/Lb., As Received)
Average RSD(t)
(6)
11822
12276
12038
12035
11903
11729
1)637
11802
11827
11832
11872
11798
11804
12400
12292
12000
11682
11481
11553
11653
11777
11823
11850
11942
12109
11986
12178
11916
11816
118B1
11758
11873
12144
11583
11614
11530
11467
11617
11650
11607
mm
10994
1 1027
(7)
4.44
5.67
6.25
5.47
4.84
4.41
3.38
2.94
2.68
2.55
2.12
3.87
4.37
5.46
5.35
6.53
4.31
4.37
3.12
2.65'
2.75
2.62
2.43
4.73
5 74
.28
5.34
4.57
3.69
4.45
3.36
4.68
3.25
3.26
2.54
2.31
3.16
2.57
2.45
2.45
5.41
7. in
'>.14
Lbs. Sulfur/
HHBtu
Average RSD(t)
(8)
1.83
1.17
1.40
1.59
1.69
1.96
2.00
1.97
1.91
1.94
1.92
1.89
1.86
1.10
1.24
1.37
2.04
2.1)
2.09
2.01
1.98
1.93
1.93
1.65
1.26
1.32
1.50
1.66
1.76
1.68
1.79
1.71
1.71
2.12
1.79
1.10
1.81
1 .80
1.B2
1.76
3.5B
3.70
1 f, I
(9)
26.18
42.49
30.98
32.87
29.45
26.97
17.07
12.19
9.26
9.76
9.82
10.41
26.22
39.19
42.01
41.19
26.68
24.20
17.81
11.90
11.44
9.28
9.88
22.78
20.10
39.59
29.81
23.30
1 '1 . 9 3
20.67
15.71
18.05
17,56
19.70
12.22
13.14
12.45
11.01
12.16
11.71
24.81
21 47
?t . 49
Kurflbor
of
Analyses
(10)
6619
289
257
156
698
87)
956
413
287
359
428
50
5738
341
135
213
88
666
625
27
690
411
78)
561
881
25
15
92
89
75
205
126
155
46
2312
4BR
B
25
107
101
101
6 If, 2
227
271
-------
"t 2 i.l 7
SAMPLE STATISTICS OF COAL CHARACTERISTICS ANALYZED BY VARIOUS LEVELS OF AGGREGATION
(INCLUDES IAT-SIZE ANALYSIS)
Data noficrlptlon:
USnH District, Scam, Preparation
I,ot-Slzea, etc.
nlot.
Dlst.
out.
Dlst.
out.
Dlat.
Dlst.
Dlst.
Dlst.
Dlat.
Dtst.
Dlst.
Dlat.
Dlst.
04,
04,
04,
04,
04,
04,
04,
04,
04,
04,
04.
04,
04,
04,
(1)
Rnw/Delvd., 750-1000 T
Rnw/Delvd., 1000-1500 T
Row/Dc Ivd., 1500-2000 T
Raw/Del vd. , 2000-3000 T
Raw/Delvd., 3000-4000 T
Raw/Dc Ivd., 4000-5000 T
Riiw/As Durned
Raw/As Burned, 200-500 T
Raw/As Burned, 500-1000 T
Raw/As Burned, 1000-1500 T
Raw/Aa Burned, 1500-2500 T
Raw/As Burned, 2500-3500 T
Raw/Aa Burned, 3500-5500 T
Raw/As Burned, 6000-8000 T
Volume Sampled
(tons)
Average RSDU)
(2)
925
1257
1667
2447
3419
4367
3260
361
BOS
1266
2015
3049
406}
6974
(3)
7.75
12.46
8.23
12.17
8.40
5.80
57.67
23.77
15.02
11.21
14.36
9.43
11.68
6.75
Sulfur Content
(Percent, As Received)
Average RSD(t)
(4)
3.B4
3.80
3.95
4.33
4.49
4.64
3.90
3.92
3.93
3.85
3.92
3.87
3.84
3.99
(5)
23.19
21.96
21.23
20.87
20.83
21.15
17.68
17.04
20.63
20.00
16.78
17.69
17.53
15.70
Heat Content
(Dtu/Lb., As Received)
Average RSDU)
(6)
11062
11100
10992
10616
10582
10675
10809
10865
10860
10968
10941
10715
10748
10702
(7)
5.39
4.77
5.63
6.48
5.88
5.64
5.16
3.49
4.51
4.49
4.41
5.62
5.25'
5.50
LbB. Sulfur/
MMOtti
Average RSD(%)
(B)
3.50
3.44
3.61
4.10
4.25
4.17
3.62
3.63
3.64
3.53
3.60
3.63
3.60
3.74
(9)
26. 7B
22.98
23.38
22.61
21.62
22.43
19.00
18.02
22.72
21.98
18.77
19.00
17.95
16.14
Number
of
Analyses
(10)
701
3479
ono
341
244
I4B
.2251
26
124
232
492
563
467
294
Dlst. 04, Source 10175, Units 1,2 t 3
As Burned
2578
45.14
4.29
14.07
10634
3.36
4.04
15.34
614
I
ts>
o
Dlst. 04, Source 10175, Unit 4, As Burned
Dtst. 04, Source 10175, Unit 4, As Burned
2500-5000 T
5000-6000 T
6000-7000 T
7000-8000 T
Dlst. 04, Source 10175, Unit 5, As Burned
Dist. 04, Source 10175, Unit 5, As Burned
2000-3000 T
3000-3500 T
3500-4000 T
4000-4100 T
Dint. 04, Source 10175, Unit 6, As Burned
Dlr.t. 04, Source 10175, Unit 6, As Burned
1300-2000 T
2000-3000 T
1000-4000 T
6246
26.38
4.04
3918
5512
6594
7361
3382
2564
3263
3730
4157
2868
1628
2593
3493
13.95
4.46
4.22
3.48
20.23
11.60
4.42
3.66
2.99
40.78
12.08
10.52
8.45
3.92
4.27
3.99
4.03
3.74
3.69
3.71
3.74
3.89
3.87
3.67
3.90
3.91
15.45
15.57
18.24
14.71
14.21
15.25
11.92
41.28
16.77
14.29
16.66
24.90
13.89
15.42
10693
10660
10695
10728
10669
10616
10569
10618
I061B
10697
10632
1059B
10459
106BO
5.63
3.79
16.10
402
6.
5.
5.
5.
6.
6.
6.
5.
5.
6.
4.
6.
6.
74
62
33
61
10
67
37
76
18
26
81
84
13
3.68
4.01
3.7-3
3.79
3.53
3.50
3.50
3.52
1.64
3.65
3.46
3.73
3.67
14.
19.
15.
14.
14.
11.
13.
15.
13.
16.
23.
13.
15.
25
82
12
85
34
11
85
53
92
12
87
89
36
46
39
141
146
410
69
151
127
49
139
20
38
54
Dint. 04, Spam 016 Raw
600-1000 T
1000-1100 T
1300-1600 T
1600-2100 T
821
I 170
1448
1782
14.48
7.32
5.28
6.82
3.70
3.88
3.59
3.65
21.23
17.20
23.48
18.79
11098
iioin
11038
11108
4.75
5.50
5.48
4.42
3.33
3.51
3.25
3.28
21.96
16.07
23.33
17.77
15
42
78
47
nint. 04, r.cam 080, Raw
inn-600 T
600-900 T
456
763
17.49
12.00
4.18
4.26
20.05
26.22
10626
10553
8.67
7.49
4.00
4.10
33.68
31 .57
90
MO
-------
SAMPLE STATISTICS OF COW. CHARACTERISTICS ANALYZED BY VARIOUS LEVELS OF AGGREGATION
(INCLUDES LOT-SIZE ANALYSIS)
Sheet 3 <'C 7
C-4
\
Data Description!
OSDM District, Seam, Preparation
Lot-Sizes, etc.
(1)
Dlst. 04, Seam 080, Raw
1000-1100 T
1100-1600 T
1600-1900 T
2000-2300 T
2608-3000 T
3000-4000 T
4000-5000 T
Diet. 04, Seam 084 Raw
250-600 T
900-1100 T
iioo-uon T
1300-1500 T
1500-1000 T
2000-3000 T
Dint. 08, All Coals
Dlst. 08, All Coals, 200-500 T
Dlst. 08, All Coals, 500-1000 T
Dlst. 08, All Coals, 1000-1500 T
Dtst. .08, All Coals, 1500-2000 T
Dlst. 08, All Coals, 2000-3000 T
Dint. OB, All Coals, 3000-4000 T
Plst. 00, All Coals, 4000-5000 T
Plst. 08. Ml Coals, 5000-7500 T
Dlst. 08, All Coals, 7500-10000 T
[list. 08, All Coals, 15000-18000 T
Dist. 08, All Coals, 25000-35000 T
Dint. OB, Raw
Dint. 00, Raw, 900-1200 T
Ulnl. 00, Raw, 1200-1500 T
nlnl-. 00, Rnw, lr>00-2000 T
Dlst. 00, Raw, 2000-3000 T
Dint. OR, Paw, 3000-4000 T
Dlst. 08, Rnw, 4000-6000 T
Dlst. 08, Raw, 6000-0000 T
Dlst. On, Wnslieil
Dint. flO, W.ii:hetl, 1000-2000 T
nut. on, Hashed, 2000-3000 T
Dint. OR, Unshod, 3000-6000 T
nlnt. 0(1, Washed, 6000-12000 T
ni -.t . OB, sonm isi
nl-.l. OH, !>,im ill, 900-1400 T
Dim. 00, Sc.lln IS), MOO-2000 T
flint. OR, Sonm 111, 2000-1000 T
Volume Sampled
(tons)
Average RSD(»)
(2)
1115
1425
1745
2137
2796
3441
4357
440
1015
1208
1407
1582
2470
5732
339
751
1351
1637
2487
3571
4392
6140
8526
16583
2989B
4095
10B5
1409
HOI
2450
3546
5307
6474
5140
1535
2550
4295
0936
2253
11 10
1737
2812
O)
9.11
5.68
5.16
4.12
4.30
B.23
5.78
23.23
5.36
4.92
3.72
5.42
10.92
176.60
26.40
20.40
13.41
9. 88
10.72
8.74
6.14
8.54
7.56
6.14
8.61
58.68
4.10
2.56
9.51
9.67
9.10
14.46
7.73
65.61
19. B9
10.96
23.17
15.75
65.37
4.11
11.20
16.97
Sulfur Content
(Percent, As Received)
Average RSD(«)
(4)
3.41
3.84
4.00
4.39
4.61
4.S6
4.72
4.05
3.86
3.99
4.16
4.33
4.15
.99
.89
.96
.94
.92
.96
.91
.69
1.08
.97
1.08
.98
1.21
1.15
.91
.92
1.03
1.27
1.57
1.60
.78
.70
.72
.76
.83
.90
1.05
.86
.79
(5)
28.39
23.91
24.67
20.60
17.61
19.76
19.68
20.01
19.90
20.49
IB. 92
16.97
19.56
46.08
33.91
38.80
46.14
39.43
38.91
46.24
40.31
55.38
34.55
16.70
13.70
47.12
SB. 55
44.08
42.25
41.00
40.72
46.18
41.62-
17.74
30.94
2} -14
37.61
43.60
47.62
57.44
34.04
26. 17
Heat Content
(Dtu/l.b., As Received)
Average RSD(%)
(6)
10973
10910
10718
10431
10613
10570
10679
11064
niaa
11175
11235
11348
11159
12022
12151
12149
12197
12263
12412
11891
11958
11771
12412
11666
12136
12288
12392
12275
12434
13023
12291
11697
11851
12039
1 1675
1156/
11903
12452
11692
170r, ?
11554
1 1416
(7)
5.07
5.88
6.58
7.25
6.10
5.94
5.72
4.9B
4.07
3.89
4.21
4.17
4.74
5.81
5.03
4.98
5.55
5.77
8.06
5.24
6.13
4.54
4.94
4.64
3.42
J.06
6.30
5.74
5.50
7. 36
:.07
7.15
5.49
5.50
4.41
4.60
5.01
4.60
S.49
5.50
3.70
4.0'>
Lbs. Sulfur/
MMDtu
Average RSD(%)
(8)
.13
.54
.76
.24
.35
.33
.34
3.66
3.45
3.57
3.70
3.82
3.71
0.83
0.74
0.80
0.78
0.75
0.78
0.77
0.74
0.92
0.78
0.93
0.81
1.00
0.95
0.75
0.75
0.80
1.05
1.34
1.17
n.65
0.67
0.62
0.64
0.67
0.77
o.nn
0.74
0.69
19)
31.96
26.07
28.40
22.61
IB. 21
20 . 56
21.09
20.74
20.23
20.44
19.87
17.19
18.28
47.71
34.54
39.71
47.64
39.30
40.19
46.90
40.80
56.08
35.59
19.38
14.63
50.97
63.29
45.41
42.60
45.35
45.40
48.47
44.00
18.49
29.39
23.03
36 57
47.06
i ;. i?
60.73
30.90
22. B9
Kumlior
of
Analyses
110)
426
394
198
101
89
229
142
29
105
IBB
517
91
22
3299
140
156
377
389
316
205
309
996
220
27
32
1211
65
204
2(14
U.3
4B
303
1 If,
H(>
65
SI
70
136
ina
64
12
62
-------
SAMPLE STATISTICS OF COAL CHARACTERISTICS ANALYZED BY VARIOUS LEVELS OF AGGREGATION
(INCLUDES LOT-SIZE ANALYSIS)
Sl.r-t 4
Dnt.i Ocscciptloni
USDM District, Spam, Preparation
Lot-Sites, etc.
Volume Sampled
(tono)
BSD(«)
Sulfur Content
(Percent, As Received)
Average RSDO)
Heat Content l.bs. f.ulfm/
(Btu/Lb., As Received) MMBtu
_A«erage EfiPain 464 Har.hrd 4394 22.23 3.34
Dint. 09, Roam 464, Hushed, 4200-4800 T 4512 1.58 3.01
Dint. 09, Snnm 464, Hnnhcc), 5500-6500 T 5992 5.35 3.12
8.12
7.40
6.74
10666
1096)
11123
2.66
3.31
2.62
4)1
-------
SAMPLE STATISTICS OF COAL CHARACTERISTICS ANALYZED BY VARIOUS LEVELS OF AGGREGATION
(INCLUDES LOT-SIZE ANALYSIS)
Slmct 5 of 7
C4
I
ro
CO
Data Description:
USBH District, Seam, Preparation
Lot-Sizes, etc.
(1)
Dlst. 10, All Coals
Dint. 10, All Coala, 500-1000 T
Dlot. 10, All Coals, 1000-1500 T
Dlst. 10, Ml Coals, 1500-2000 T
Dlst. 10, All Coals, 2000-2500 T
Dlst. 10, All Coals, 2500-3500 T
Dlst. 10, All Coals, 3500-5000 T
Dlst. 10, All Coals, 5000-7000 T
Dlst. 10, All Coals, 7000-9000 T
Dlst. 10, All Coals, 9000-12000 T
Dint. 10. Sean 900
Dlst. 10, Seam 900, 1500-2200 T
DUt. 10, Seam 900, 7000-8000 T
Dlst. 10, Seam 900, 6000-9000 T
Dlst. 10, Seam 900, 9000-10000 T
DUt. 10, Scam 900, 10000-12000 T
Dlst. 10, Seam 404, Average Sulfur < 21
Dint. 10, Seam 484, Average Sulfur
< 2», 1000-1500 T
7000-8000 T
8000-9000 T
9000-10000 T
Dlst. 10, Scam 4S4, Average Sulfur y2\
Dint. 10, Scam 484, Average Sulfur
>2%, 6000-7000 T
DUt. 11, All Coals
DUt. 11, Average Sulfur <(2»,
DUt. 11, Avcraqo Sulfur < 2%, 6000-7000 T
DUt. 11, Average Sultur<;2t, 7000-8000 T
Dlst. 11, Average Sulfur <_2», 6000-9000 T
DUt. 11, Average Sulfur (21, 9000-10000 T
Dint. 11, Average Sulfur >2t
Dlst. 11, Avcrnqc Sulfur >21, 200-500 r
flUl. 11, Avr-r.vjo Sulfur )>2«, 500-1000 T
Illst. 11, Avir.iqv Sulfur !>2», 1000-1500 T
DUI. II, Avnrnqn Sulfur S2», 1500-2000 T
Pint. 11, Avcr.vje Sulfur >2\, 2000-3000 T
DUr. II, Avc?r,vjr Sulfur >2», 3000-5000 T
Dlst. 11, Avrrnqp Sull»r^2l, 5000-7000 T
nl-,1. 11, Avr-r.iqp Sulfur N>2», 7000-11000 T
Volume Sampled
(tons)
Average RSD(t)
(2)
5921
843
1345
1715
2143
2870
4220
6060
8224
9812
8340
1563
7470
8633
9575
10289
5932
1317
7498
8527
9341
4120
6149
5836
7426
6568
7575
8445
9398
5397
366
783
1205
1753
2443
4156
6096
8625
13)
58.74
16.88
6.44
11.57
7.17
8.88
9.12
8.50
6.27
5.01
31.99
2.79
4.09
3.18
3.30
3.00
58.02
7.59
4.14
3.15
2.35
75.55
5. IB
53.70
35.41
4.17
3.78
3.67
3.04
57.81
26.72
16.62
11.06
8.32
11.34
14.53
9.39
12. 32
Sulfur Content
(Percent, As Received)
Average nsi>(t)
(4)
2.84
3.15
3.06
3.19
2.90
2.95
2.49
3.10
2.45
2.87
2.98
3.48
3.07
2.75
2.89
3.20
1.40
1.47
1.41
1.31
1.32
3.30
3.57
3.12
1.65
1.72
1.68
1.65
1.76
3.52
3.36
3.41
3.35
3.47
3.45
3.51
3.64
1.50
(5)
28.51
18.12
26.47
15.11
13.04
20.22
32.15
23.32
37.14
24.58
18.51
12.38
12.82
20.59
20.89
13.48
39.58
19.99
50.30
46.07
47.72
11.99
8.61
29.97
34.45
38.09
32.42
34.32
33.39
14.81
in. 73
15.28
10.62
16.73
12.02
15.23
14.34
14.78
Heat Content
(Btu/Lb. , As Received)
Average RSD(t)
(6)
10788
10806
10842
10720
10652
10565
10746
10717
10842
10781
10834
10967
10908
10760
10752
10826
11219
11812
11142
10923
10666
10627
10586
10853
10836
10799
10009
10805
10847
lonsa
10939
10752
ion 26
lOBifl
1075'.
innsfl
10074
10802
(7)
4.11
1.97
5.27
3.25
3.71
3.13
3.66
3.35
4.20
3.39
3.11
3.36
2.82
3.09
3.21
3.14
5.84
3.40
4.24
4.65
5.42
3.28
1.76
2.70
1.99
1.92
1.62
1.74
1.71
2.86
2.54
3.29
2.65
3.02
3. 41
2.60
2.81
2.68
LbR. Sulfur/
MMBJtti
Average RSD(l)
18)
2.65
2.93
2.nr,
2.98
2.73
2.81
2.32
2.91
2.80
2.66
2.75
3.18
2.82
2.55
2.69
2.97
1.25
1.25
1.27
1.21
1.25
3.12
3.38
2.88
1.53
1.60
1.55
1.54
1.62
3.25
3. OB
3. 10
3. 10
3.22
3.22
3.26
3.35
3.22
(9)
29.46
20.25
28.60
14.52
14.29
21.17
32.84
24.17
37.74
24.81
18.16
11.88
12.44
19.77
21.12
14.84
40.81
21.86
49.26
47.25
48.33
13.26
9.16
30.45
34.98
38.49
32.03
35.08
33.71
)r>.74
20.07
16.65
11.17
1(1 .20
14.63
16 66
1 4 . 6 4
IS. ",9
Number
of
Analyses
(10)
45)6
5
994
264
187
158
274
436
1040
1203
1588
70
109
238
647
418
544
184
71
172
81
1946
1R2
1188
257
28
50
59
46
911
21
41
42
51
Rl
IBS
227
230
-------
SAMPLE STATISTICS OF COAL CHARACTERISTICS ANALYZED BY VARIOUS LEVELS OF AGGREGATION
(INCLUDES LOT-SIZE ANALYSIS)
Shc-ct 6 nl 7
Cl
I
to
*»
0;U« Description:
USPM District, Seam, Preparation
I.ol-Slzes, etc.
Dlat.
Dist.
Diat.
Dint.
Dist.
Dlst.
DLst.
Dlst.
Dlat.
Dlst.
Dlst.
Dlat.
Dlst,.
Olst.
Dlst.
Dlat.
Dist.
Dlst.
Dlst.
Dlst.
Dist.
Hist.
(1)
12, All Coals
12, All Coals, 200-1000 T
12, All Coals, 1000-1500 T
12, All Coals, 1500-2000 T
12, Seam 517, Raw
12, Seam 517, Raw, 200-500 T
12, Senm 517, Raw, 500-1000 T
12, Seam 517, Raw, 1000-2000 T
15, Seam 492, Raw
15, Seam 492, Raw, 1000-2000 T
15, Seam 492, Raw, 45000-75000 T
15, Seam 492, Washed
17, All Coals
17, All Coals, 1300-1800 T
17, All Coals, 2000-3000 T
17, All Coals, 3000-4000 T
17, All Coals, 6000-6900 T
IB, All Coals
10, Auto., ASTH Samples
18, Auto., ASTH Samples, 5000-7000 T
18, Auto., ASTH Samples, ROOO-10000 T
in, Aulo., ASTH Samples, 10000-
13000 T
Dist.
in, Auto., ASTM Samples, 13000-
16000 T
Pint.
Dlst.
IH, Hlno 10071 LAB - CTE
in. Hi IIP 10071, LAB - CTF.
13000-17000 T
Dlst.
1R, Mine IOO7] IJVB" SRP
Volume Sampled
(tons)
Average RSD(t)
(2)
3965
638
1322
1708
4384
353
759
1482
3011
1503
58714
753
3085
1576
2636
3108
6166
10776
11105
6182
8872
11462
14915
12200
15017
12318
13)
105.86
35.47
11.34
6.35
105.29
23.34
18.53
15.84
97.18
18.47
11.87
60.65
56.93
3.92
11.07
4.26
3.59
37.82
38.11
4.25
5.05
6.73
5.10
34.43
5.27
31.BR
Sulfur Content
(Percent, As Received)
Average RSD(%)
(4)
3.62
3.41
3.22
3.43
3.10
2.27
3.23
3.14
4.11
3.20
4.56
3.63
0.54
0.60
0.56
0.48
0.45
0.47
0.44
0.43
0.63
0.39
0.40
0.42
0.42
0.39
(5)
32.04
29.83
22.31
27.30
21.36
17.15
27.72
20.42
18.95
15.41
14.27
7.29
22.70
12.15
20.17
26.48
20.88
43.16
34.67
11.78
49.81
10.85
11.93
11.65
11.19
10.95
Heat Content
(Btu/Lb., 'As Rccelvf-d)
Average RSD(t)
(6)
9346
9125
9390
9410
9302
8674
9105
9377
9770
10374
9515
10771
10546
10733
10718
10473
10124
10485
10503
10203
10571
10592
10611
10577
10626
10571
5
5
3
2
4
5
6
3
8
7
7
2
4
3
4
3
3
4
3
2
4
2
3
4
3
3
(7)
.13
.97
.69
.92
.71
.21 .
.29
.53
.23
.22
.63
.61
.35
.14
.22
.52
.91
.34
.92
.94
.25
.86
.24
.00
.52
.11
Lbs. Sulfur/
MMBtu
Average RSUU)
(8)
3.88
3.74
3.43
3.65
3.35
2.64
3.57
3.35
4.29
3.13
4.87
3.38
0.51
0.56
0.53
0.46
0.45
0.45
0.43
0.42
0.60
0.37
0.38
0.40
0.40
0.37
(9)
32.47
30.11
22.47
27.19
22.61
19.88
31.55
20.59
25.78
21.55
21.13
8.49
21.88
11.75
19.04
25.79
22.56
44.06
35.65
12.61
50.20
12.75
15.33
15.77
15.03
14.21
Number
of
Analyses
(10)
203
36
41
27
150
9
15
57
92
12
35
33
495
1B2
72
121
91
299
261
47
11
31
111
09
54
114
-------
Sheet 7 of 7
SAMPLE STATISTICS OF COAL CHARACTERISTICS ANALYZED BY VARIOUS LEVELS OF AGGREGATION
(INCI.UDES LOT-SIZE ANALYSIS)
Data Description:
USHM District, Seam,
Lot-Sizes, etc.
Dlst. 1,9,
Dist. 19,
Dlgt. 19,
Dist. 19,
Dlst. 19,
Dist. 19,
Oist. 19,
Dint. 19,
Dist. 19,
Dist. 19,
C-( Dlst. 19,
1 ,,
10 Dlst. 19,
Dlst. 19,
Dist. 19,
Dlst. 21,
Dint. 21,
Dist. 21,
ID
All Coals
All Coals,
All Coals,
All Coals,
All Coals,
All Coals,
Seam 951
10006 (All
Mine 10007
Mine 10038
Mine 10039
Mine 10061
Mine 10076
Nine 10077
All Coals
All Coals,
All Coals,
Preparation
1000-2000 T
2000-3000 T
8000-10000 T
10000-11000 T
11000-12000 T
Analyses)
(All Analyses)
(All Analyses)
(All Analyses)
(All Analyses)
(All Analyses)
(All Analyses)
0-30000 T
200000-400000 T
Volume Sampled
(tons)
Average RSD(I)
(2)
9766
1605
2745
9737
10466
11308
10222
10234
10794
9784
9544
10594
9179
4187
13127
5345
336277
(3)
25.78
17.39
11.65
4.36
3.18
2.21
19.90
19.94
11.82
5.54
7.00
4.16
10.25
64.36
373.21
74.06
11.23
Sulfur Content Heat
(Percent, As Received) (Btu/Lb.,
Average RSD(%) Average
it]
0.45
1.01
0.92
0.58
0.41
0.36
0.36
0.36
0.39
0.60
0.85
0.33
0.47
0.98
0.78
0.79
0.55
(5)
47.30
19.82
18.15
43.88
31.07
13.06
15.09
14.01
10.23
33.63
45.22
13.09
34.55
20.65
25.80
25.16
28.83
(6)
8739
10298
10302
9208
8647
8391
8393
8394
8301
9BS2
10626
8781
8266
10271
6684
6684
6663
Content
As Received)
nSDU)
(7)
7.92
3.68
3.72
8.33
6.20
2.24
1.47
1.43
1.84
3.74
3.10
1.51
2.98
3.13
7.07
7.17
1 1;
Lbs. Sulfur/
MHBtu
Average RSO(t)
(B)
0.51
0.99
0.89
0.61
0.48
0.43
0.43
0.43
0.47
0.61
0.81
0.38
0.57
0.96
1.18
1.19
0.82
(9)
37.69
22.13
18.10
37.35
25.60
12.45
14.79
13.62
9.88
31.33
45.44
13.08
34.06
20.93
27.37
26.78
28.87
Nurabc r
ot
Analyses
(10)
3526
38
157
813
2023
738
2224
2199
312
169
58
99
25
294
2040
1986
42
Dist. 22, All Coals
9243
9.13
0.58
39.65
9107
5.80
0.65
4] 94
1196
-------
APPENDIX K
ANALYSIS OF THE STATISTICAL DISTRIBUTIONS
OF COAL CHARACTERISTICS
This Appendix provides a detailed discussion of the
analyses performed to determine which type of frequency
distribution-normal, inverted gamma, or lognormal-best fits
the observed frequency distributions of actual coal data.
Section 4.3 of this report discussed the methods used in
this analysis and provided a summary of the results. The
following discussion and tables provide t..e specific results
obtained for the various analyses performed for both raw and
washed coals.
1.0 Raw Coals
1.1 Raw Coals, Overall Goodness of Fit
The results of the chi-square goodness of fit test for
raw coals are presented in Table K-l. "Best fit" in this
table refers to the lowest chi-square value. The term
"significant" means that the observed distribution was not
significantly different from the assumed distribution. To
illustrate, consider the analysis of heat content values for
U.S. Bureau of Mines Producing District 1. For two of the
data sets analyzed, the hypothesis that the data came from a
normal distribution could not be rejected. Similarly, that
hypothesis cannot be rejected for the inverted gamma distribu-
tion for two data sets, and for the lognormal distribution for
one data set. The best fit in terms of the lowest chi-square
value for Btu occurred once for the normal distribution and
three times for the inverted gamma distribution. In this
example, no data sets indicated a best fit for the lognormal
distribution.
K-l
-------
TABLE K-l
Sheet 1 of 2
CHI-SQUARE ANALYSIS, RAW COAL
U.S.B.M.
Producing
District
Variable
Normal
Inverted
Gamma
Log-
normal
10
11
12
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu)
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu
(Best Fit) 4
(Significant) 1
(Best Fit) 1
(Significant) 2
(Best Fit) 2
(Significant) 1
(Best Fit) 4
(Significant) 3
(Best Fit) 7
(Significant) 5
(Best Fit) 3
(Significant) 2
(Best Fit)
(Significant)
(Best Fit)
(Significant) 1
(Best Fit)
(Significant) 1
(Best Fit) 1
(Significant) 1
(Best Fit)
(Significant) 1
(Best Fit) 1
(Significant) 1
(Best Fit) 2
(Significant)
(Best Fit) 2
(Significant)
(Best Fit) 2
(Significant)
(Best Fit) 1
(Significant) 1
(Best Fit) 3
(Significant) 2
(Best Fit)
(Significant) 1
(Best Fit)
(Significant)
(Best Fit)
(Significant) 1
(Best Fit)
(Significant)
Inverted
Gamma
1
3
2
-
1
4
3
-
5
4
4
2
3
1
1
2
3
1
1
2
1
1
1
2
2
1
2
1
2
2
1
1
1
1
Log-
normal
1
-
1
2
2
2
3
1
3
3
1
1
1
-
-
1
2
1
1
1
1
1
1
1
-
-
1
-
-
2
1
^
-
1
K-2
-------
Sheet 2 of 2
U.S.B.M.
Producing
District
16
Variable
17
18
19
21
22
Total
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
Normal
2
2
r
3
3
2
3
4
15
9
17
14
14
12
Inverted
Gamma
2
1
1
1
4
4
1
2
2
4
1
1
17
15
9
14
15
19
Log-
normal
1
1
2
2
5
2
1
1
1
1
1
8
9
2
10
14
15
K-3
-------
Based on the information in Table K-l, no firm conclu-
sions can be made with respect to the best distribution for
any of the three variables. The distributions exhibiting
the best fit vary considerably from data set to data set.
However, the normal distribution seems to be slightly better
for the Btu values, while the sulfur content and Ibs
S/MMBtu appear to be best represented by the inverted
gamma distribution.
1.2 Raw Coals, Top 1.5 Percent of the Distribution for Lbs
S/MMBtu
Table K-2 presents the analysis of the best fit for Ibs
S/MMBtu for the top 1.5 percent of the distribution for
raw coals. Part A of Table K-2 represents coals which were
not sorted by lot-size, while Part B shows the same informa-
tion for coals after the data were sorted by lot-size. In
general, the coals in Part A exhibit a wide range of volumes
(for example, 1,000 to 10,000 tons), while the coals in
Part B are based on narrowly defined volume intervals (such
as 1,000 to 2,000 tons). The choice for the best fit here
is the lognormal, although it appears only slightly better
than the inverted gamma. The normal distribution not only
does not fit well, but also appears to be biased because it
consistently underestimated the number of observations in
the top 1.5 percent of the distribution. Thus, if one were
interested in the top 1.5 percent, one should use either the
inverted gamma or lognormal distribution.
1.3 Raw Coals, Top 15 Percent of the Distribution for Lbs
S/MMBtu
Table K-3 shows the results of the analysis of raw
coals for the top 15 percent of the distribution. Part A of
the table is based on coals not sorted by lot-size, while
K-4
-------
TABLE K-2
BEST FIT ANALYSIS FOR LBS S/MMBTU
(Raw Coals, Top 1.5 Percent of Frequency Distribution)
Number of Data
Sets Best Fit By;
u.
A.
S.B.M. Producing District
(1)
Not Sorted by Lot-Size
1
4
8
9
10
11
17
18
19
21
22
Total
Normal
(2)
_
1
-
1
-
-
-
1
,1
4
Inverted
Gamma
(3)
2
6
1
1
2
1
1
-
1
1
,^^_
16
Log-
normal
(4)
4
5
2
2
-
-
-
1
3
1
_1
19
B. Sorted by Lot-Size
1
4
8
9
19
21
22
Total
1
1
1
1
3
1
2
2
-
2
4
1
1
1
Source: Foster Associates, Inc.
K-5
-------
TABLE K-3
BEST FIT ANALYSIS FOR LBS S/MMBTU
(Raw Coals, Top 15 Percent of Frequency Distribution)
Average Relative
U.S.B.M.
Producing
District
(1)
Number
of Mines
(2)
Normal
(3)
Error (%)
Inverted
Gamma
(4)
Log-
normal
(5)
Normal
(6)
Best Fit
Inverted
Gamma
(7)
Log-
normal
(8)
A. Not Sorted by Lot-Size
1
3
4
8
9
10
11
17
18
19
21
22
4
1
16
3
4
1
1
1
1
5
2
2
10.5
17.8
22.7
32.9
9.0
22.3
2.6
37.1
15.6
11.1
17.1
33.6
12.9
29.8
31.3
32.4
17.4
38.5
11.7
45.0
1.9
17.3
29.7
51.5
241.4
30.8
46.4
30.6
33.8
22.3
2.6
102.2
25.0
24.2
1.5
113.6
3
1
9.5
-
3
.5
.5
1
-
2.5
.5
2
1
-
3
2
1
-
-
-
1
2
-
-
_
-
3.5
1
-
.5
.5
-
-
.5
1.5
-
Total 41 23.5 10 7.5
B. Sorted by Lot-Size
21
3 -
- 1.5
1
3
4
8
9
19
21
22
5
1
6
3
3
2
1
1
10
42
11
12
10
1
23
.1
.7
.1
.9
.8
.9
.6
9
58
14
4
16
10
13
.7
.0
.1
.4
.0
.3
.2
28
78
19
16
20
30
6
.3
.3
.9
.2
.9
.9
.7
2
1
3
1
2
1
-
.5
-
Total 22 10.5 8 3.5
Source: Foster Associates, Inc.
K-3
-------
Part B shows the same analysis based on data sorted by lot-
size.I/ Columns (3), (4), and (5) on Table K-3 set out the
average relative errors for the relevant mines. Since the
average relative errors could be biased by the presence of
one particularly bad fit, the number of mines for which each
particular distribution provided the best fit is also indi-
cated in Columns (6), (7), and (8). The inverted gamma and
the normal distributions provide the best fit for the top 15
percent. The normal distribution appears to be slightly
better, but an insufficient number of data sets were analyzed
to make this statement with any degree of confidence. Based
on this analysis, the lognormal distribution clearly provides
the worst fit of the three distributions for the top 15
percent of the frequency distribution.
1.4 Raw Coals, Top 1.5 Percent of the distribution for Heat
Content
Set out in Table K-4 are the results of the best fit
analysis for the heat content of raw coals in the top 1.5
percent of the distribution. Part A of Table K-4 is based
on coals not sorted by lot-size, while Part B reflects data
which were sorted by lot-size. In both instances there
appears to be an indication that the normal distribution
provides the best fit in the top 1.5 percent. However,
because of the limited number of data sets, the evidence is
not conclusive.
I/ It should be noted that when the data were sorted by
lot-size (or interval analysis), the number of observations
contained within a specific lot-size were in effect a sub-
set of the data unsorted by lot-size. Frequently, the num-
ber of observations contained within the interval specified
were insufficient for an examination of the right tail of
the distribution.
K-7
-------
TABLE K-4
BEST FIT ANALYSIS FOR HEAT CONTENT (BTU/LB)
(Raw Coals, Top 1.5 Percent of Frequency Distribution)
Number of Data
Sets Best Fit By:
U.S.B.M.
A. Not
Producing District
(1)
Sorted by Lot-Size
1
3
4
9
10
11
19
21
22
Total
Normal
(2)
1
1
5
2
-
2
1
-
_1
13
Inverted
Gamma
(3)
2
-
2
1
1
-
3
1
10
Log-
normal
(4)
1
-
2
1
-
-
-
-
1
5
B. Sorted by Lot-Size
1
3
4
8
9
19
21
22
3
2
1
-
2
-
1
1
2
-
-
1
1
-
-
2
1
1
-
Total
Source: Foster Associates, Inc.
K-8
-------
1.5 Raw Coals, Top 15 Percent of the Distribution for neat
Content
Table K-5 shows the results of the analysis of the best
fit for the heat content of raw coals for the top 15 percent
of the distribution. Part A of Table K-5 is based on coals
not sorted by lot-size, while Part B is based on data which
were sorted by lot-size. From Table K-5 it can be seen that
the normal distribution has a consistently lower average
relative error than either the inverted gamma or lognormal
distributions. Also, the analysis clearly shows that the
normal distribution tends to fit the data better more often
than any of the two alternative distributions.
1.6 Raw Coals, Top 1.5 and 15 Percent of the Distribution
for Sulfur Content
Because sulfur content was not considered to be of as
much interest as the Ibs S/MMBtu variable, detailed
analyses of the top 1.5 and 15 percent were not undertaken.
However, literally hundreds of distributions were visually
examined to determine which of the three distributions appeared
to fit the top right tail the best. These visual examinations
indicated that the percent sulfur can best be described by
the inverted gamma distribution, which also provided the best
fit for Ibs S/MMBtu.
1.7 Summary of the Analyses for Raw Coals
In conclusion, the following recommendations are made
with respect to the most appropriate choice of distributions
for raw coals:
(1) .For heat content, the normal distribution
(2) For Ibs S/MMBtu, the inverted gamma distribution
(3) For sulfur content, the inverted gamma distribution
K-?
-------
TABLE K-5
BEST FIT ANALYSIS FOR HEAT CONTENT (BTU/LB)
(Raw Coals, Top 15 Percent of Frequency Distribution)
Average Relative
U.S.B.M.
Producing
District
(1)
Number
of Mines
(2)
Normal
(3)
Error (%)
Inverted
Gamma
(4)
Log-
normal
(5)
Normal
(6)
Best Fit
Inverted
Gamma
(7)
Log-
normal
(8)
A. Not Sorted by Lot-Size
1
3
4
8
9
10
11
17
18
19
21
22
4
1
15
3
4
1
1
1
1
5
2
2
66'. 9
47.2
47.3
46.8
25.2
32.5
103.6
264.0
575.0
107.3
31.5
8.4
74.7
51.3
55.1
49.8
28.7
40.0
110.9
280.0
610.0
110.8
39.6
9.5
181.9
147.9
91.4
153.9
42.9
32.5
148.9
506.7
-
846.0
123.1
306.1
Total 40
3
1
14.5
2
3
.5
1
1.
1
3
2
1
33
1
-
-
1
-
-
-
-
-
2
-
1
5
_
-
.5
-
1
.5
-
-
-
-
-
-
2
B. Sorted by Lot-Size
1
3
4
8
9
19
21
22
5
1
6
3
3
2
1
1
41.1
33.8
42.5
14.1
24.2
14.3
6.7
17.5
45.5
36.9
51.1
25.8
27.3
16.9
3.9
15.0
183.6
52.9
73.7
21.6
67.7
820.6
5.0
2.9
4
1
6
1
2
2
-
-
Total 22 16
Source: Foster Associates, Inc
2
1
K-10
-------
2.0 Washed Coals
2.1 Washed Coals, Overall Goodness of Fit
Table K-6 presents the results of the chi-square analysis
for washed or partially washed coals. The results are some-
what ambiguous but the best distribution for Ibs S/MMBtu
appears to be the inverted gamma or lognormal distribution.
Heat contents seems to be best fitted by a normal distribu-
tion, and sulfur contents by either an inverted gamma or
lognormal distribution.
2.2 Washed Coals, Top 1.5 Percent of the Distribution for
Lbs S/MMBtu
Set out in Table K-7 are the results of the analysis of
the distributions of Ibs S/MMBtu for washed coals.
Table K-7 indicates that for the top 1.5 percent of the dis-
tribution the best fit is provided by the inverted gamma,
which appears superior to the lognormal.
2.3 Washed Coals, Top 15 Percent of the Distribution for
Lbs S/MMBtu
Table K-8 presents the analysis of the top 15 percent
for Ibs S/MMBtu. The lognormal seems to be clearly
less preferable than the inverted gamma or the normal dis-
tribution. With respect to these last distributions, there
appears to be no significant differences in the case of the
coals sorted by lot-size but in the coals not sorted by lot-
size, the normal appears superior to the inverted gamma.
K-ll
-------
TABLE K-6
CHI-SQUARE ANALYSIS
WASHED OR PARTIALLY-WASHED COALS
U.S.B.M.
Producing
District
Variable
10
11
15
Total
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu
Sulfur
Btu
Lbs S/MMBtu
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
(Best Fit)
(Significant)
Normal
1
1
1
2
2
7
1
4
1
3
5
8
1
5
Inverted
Gamma
1
1
1
1
_
-
1
1
1
1
^
-
3
3
2
4
3
4
1
^m
Log-
normal
1
1
-
1
1
1
-
_
-
1
1
1
1
2
2
4
6
2
5
1
1
1
1
1
1
3
6
4
6
5
6
5
6
4
8
4
8
Source
Foster Associates/ Inc.
K-12
-------
TABLE K-7
BEST FIT ANALYSIS FOR LBS S/MMBTU
(Washed Coals, Top 1.5 Percent of the Frequency Distribution)
Best Fit
Inverted Log-
Normal Gamma normal
(3) (4) (5)
1 1
1
1
1 4
1
U.S.B.M.
Producing
District
(1)
A. Not Sor
1
4
9
10
11
Total
B. Sorted
1
4
7
8
9
10
Number
of Mines
(2)
ted by Lot-
2
1
1
9
_2
15
by Lot-Size
2
2
1
3
3
2
Total 13 0
Source: Foster Associates, Inc.
1 1
2
1
1 2
2 1
2
K-13
-------
TABLE K-8
BEST FIT ANALYSIS FOR LBS S/MMBtu
(Washed Coals, Top 15 Percent of the Frequency Distribution)
U.S.B.M.
Producing
District
Average Relative
Error (%)
Best Fit
Number
of Mines
(1)
(2)
Normal
(3)
Inverted
Gamma
Log-
normal
(5)
A. Not Sorted by Lot-Size
Total 17
Normal
(6)
9.5
Inverted
Gamma
(7)
1
4
9
10
11
2
1
2
9
3
9.4
23.9
14.3
42.2
11.3
12.1
14.8
17.0
51.3
6.6
30.8
23.9
10.7
70.1
20.7
1
-
.5
6
2
1
1
-
1
1
Log-
normal
(8)
1.5
2
3.5
B. Sorted by Lot-Size
1
4
9
10
7
8
2
2
3
2
1
3
17.0
13.6
13.0
78.9
1.4
18.7
7.0
14.5
17.2
75.5
12.8
10.8
17.0
18.1
15.7
133.3
18.3
16.4
1
1
1.5
1
1
-
1
1
1
1
-
2
Total 13
5.5
.5
1.5
Source: Foster Associates, Inc.
K-14
-------
2.4 Washed Coals, Top 1.5 Percent of the Distribution for
Heat Content
The results of the analysis for the top 1.5 percent of
the distribution for heat contents of washed coals are set
out in Table K-9. As in the case of raw coals, the normal
distribution appears to be most appropriate.
2.5 Washed Coals, Top 15 Percent of the Distribution for
Heat Content
The results of the analysis of the top 15 percent of
the distributions for heat contents of washed coals are sum-
marized in Table K-10. As in the case of all previous
analyses of heat content, the best fit appears to be provided
by the normal distribution.
2.6 Summary of the Analysis of Frequency Distributions for
Washed Coals
The analysis of the frequency distributions of washed
coals produced results which were not significantly different
from the raw coals. In previous sections of this report, it
was found that compared to raw coals, the washed coals exhibited
a lower mean Ibs S/MMBtu as well as a lower RSD of Ibs
S/MMBtu. However, based on this analysis of frequency
distributions, it appears that coal washing does not alter
the shape of the statistical distribution.
In view of the results for washed coals, the choice of
the type of distributions for raw coal characteristics is
equally applicable to washed coals. Thus, it seems appropri-
ate to make the following recommendations for both raw and
washed coals:
K-15
-------
TABLE K-9
BEST FIT ANALYSIS FOR HEAT CONTENT (BTU/LB)
(Washed Coals, Top 1.5 Percent of the Frequency Distribution)
U.S.B.M.
Producing
District
(1)
A. Not Sor
1
4
9
10
11
Total
B. Sorted
1
4
7
8
9
10
Number
of Mines
(2)
ted by Lot-Size
1
1
1
9
_2
14
by Lot-Size
1
2
1
3
3
2
Normal
(3)
1
-
-
6
2
9
1
2
1
2
1
Best Fit
Inverted
Gamma
(4)
_
-
1
1
_
2
_
-
-
1
1
-
Log-
normal
(5)
1
-
2
3
-
-
-
1
2
Total 12 7
Source: Foster Associates, Inc,
K-16
-------
TABLE K-10
BEST FIT ANALYSIS FOR HEAT CONTENT (BTU/LB)
(Washed Coals, Top 15 Percent of the Frequency Distribution)
Average Relative
Error (%)
Best Fit
U.S.B.M.
Producing Number Inverted Log- Inverted Log-
District of Mines Normal Gamma normal Normal Gamma normal
(1)
(2)
(3)
(4)
(5)
A. Not Sorted by Lot-Size
Total 17
(6)
12
(7)
(8)
1
4
9
10
11
2
1
2
9
3
33.9
9.4
6.1
54.1
34.1
38.4
11.9
6.1
58.4
36.1
116.0
75.0
119.4
80.2
83.2
2
1
1
6
2
-
-
1
2
1
B. Sorted by Lot-Size
1
4
9
10
7
8
1
1
3
2
1
3
27.1
13.4
8.8
8.45
108.8
106.4
30.0
17.1
8.1
8.9
117.1
110.6
84.4
41.3
94.7
44.7
787.5
229.1
1
1
1
1
1
2
Total 11
1
1
2
Source: Foster Associates, Inc.
K-17
-------
Coal Characteristic Best Fit Distribution
Lbs 3/MMBtu Inverted Gamma
Heat Content (Btu/Lb) Normal
Sulfur Content Inverted Gamma
K-13
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-450/5-80-OQ8a
2.
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
A Statistical Study of Coal Sulfur Variability
and Related Factors
5. REPORT DATE
May 1980
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
George R. Warholic, John E. Morton, Yimin Ngan
James E. Spearman, and Yvonne Harris
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
10. PROGRAM ELEMENT NO.
Foster Associates, Inc.
1101 Seveteenth Street, N.W.
Washington, D.C. 20036
O
11. CONTRACT/GRANT NO.
68-02-2592
12. SPONSORING AGENCY NAME AND ADDRESS
13. TYPE OF REPORT AND PERIOD COVERED
EPA, Office of Air Quality Planning and Standards
Pollutant Assessment Branch
Research Triangle Park, NC 27711
Final
14. SPONSORING AGENCY CODE
U.S. EPA
15. SUPPLEMENTARY NOTES
Project Officer - Rayburn Morrison, OAQPS/SASD, MD-12
16. ABSTRACT
Coal analysis data and power plant continuous monitoring data were gathered,
reviewed and analyzed to assess the impact of fuel coal characteristics on complianci
strategies and emission regulations. Coal analysis data, on a raw and washed basis,
were analyzed by individual mine, composite coal seams and USBM Producing Districts.
The results indicated that composite coal seam or Producing District data cannot
be used to accurately predict sulfur variabilities for individual mines. Analyses
indicated that the heat content (Btu/lb) was best approximated by the normal
distribution, which the sulfur content and pounds sulfur/MMBtu were best represented
by the inverted gamma distribution which was slightly superior to the lognormal
distribution. Analysis of available continuous monitoring data supported the
inverse relationship between coal sulfur variability and lot size, i.e.,
significant reductions in relative variability of emissions occur as the averaging
time increases. The continuous monitoring.data indicate that while FGD systems
reduce mean emission levels, the relative variabilities of outlet SO? concentra-
tions are substantially greater than those of inlet S02 concentrations. The
various analyses of coal sulfur variability identified no reliable method for coal
suppliers or sonsumers to predict variability which may be critical for compliance
by some coal-fired boilers to existing sulfur emission-limiting regulations.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COS AT I Field/Group
S02 Emissions
Coal Air pollution
Sulfur variability
Power plants
Fuel Standards
Emission standards
Coal sampling
Coal
Sulfur variability
Air pollution control
8. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
279
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
------- |