&EPA
Comprehensive Assessment of
the Specific Compounds Present
in Combustion Processes
Volume 2
Design for a National Survey
of Emission of Specific Compounds
from Coal Fired Utility Boiler Plants
-------
COMPREHENSIVE ASSESSMENT OF THE SPECIFIC COMPOUNDS
PRESENT IN COMBUSTION PROCESSES
VOLUME 2 - DESIGN FOR A NATIONAL SURVEY OF EMISSION OF
SPECIFIC COMPOUNDS FROM COAL FIRED UTILITY
BOILER PLANTS
by
Robert M. Lucas
Denise K. Melroy
Research Triangle Institute
SPECIAL REPORT
EPA Prime Contract No. 68-02-3938
Midwest Research Institute
MRI Project No. 8501-A(1)
August 8, 1985
Prepared for
U.S. Environmental Protection Agency
Office of Pesticides and Toxic Substances
Field Studies Branch
401 M Street, S.W.
Washington, DC 20460
Attn: Dr. Joseph J. Breen, Project Officer
Mr. Daniel T. Heggem, Work Assignment Manager
-------
DISCLAIMER
This document has been reviewed and approved for publication by
the Office of Toxic Substances, Office of Pesticides and Toxic Substances,
U.S. Environmental Protection Agency. The use of trade names or commercial
products does not constitute Agency endorsement or recommendation for use.
-------
PREFACE
The research and preparation of the draft of this report was per-
formed for the U.S. Environmental Protection Agency under Contract No. 68-01-
5848, Research Triangle Institute Project No. 1864-11. Mr. Joseph Carra was
the Contract Officer and Dr. John Smith was the Task Manager. The final re-
visions and preparation of this final report were completed by Research Triangle
Institute under subcontract to Midwest Research Institute under Prime Contract
No. 68-02-3938.
MIDWEST RESEARCH INSTITUTE
Clarence L. Haile
Deptrhy. Progrann Manager
)hn E. Going
"Program Manager
Approved:
I)
James L. Spigarelli, Director
Chemical and Biological Sciences
Department
-------
ACKNOWLEDGEMENTS
The authors greatly appreciate the cooperation of Mr. David Redford
of the U. S. Environmental Protection Agency, Washington, D.C.; Dr. Clarence
Haile of the Midwest Research Institute during the conduct of the study.
Dr. Haile directed the chemical analyses that provided the data for many of
the tables included in this report. The authors also thank Mr. Frank Potter
for his helpful comments in revising this report and Ms. Martha Clegg for
excellent work in typing this report.
-------
CONTENTS
I. Executive Summary 1
II. Introduction 2
A. Background 2
B. Objectives of the Pilot Study 2
C. Statistical Design for the Pilot Study 3
III. Results of the Pilot Study 10
A. Introduction 10
B. Summary Statistics 10
C. Compositing and Weighting 13
D. Chemical Analysis Measurement Errors 13
IV. National Survey Design Development 16
A. Introduction. 16
B. Variance Components 16
C. Cost Components 19
D. Optimum Allocation 19
E. Sampling Frame 23
F. Stratification of Coal Combustion 23
G. Sample Selection of Coal Combustion 23
References 26
Appendices
A. Ames, Iowa Plant Solid and Liquid Sampling Protocol. . 28
B. Chicago, NW Incinerator Solid and Liquid Sampling
Protocol 31
C. Theoretical Justification of Estimation Formulae Used
in Statistical Analysis of Composited Data 34
D. Coal Combustion Sampling Frame .- • • 39
-------
TABLES
Number Page
1 Design Sample Sizes and Number of Specimens Actually
Collected for Various Media at Ames, Iowa Power Plant. . 8
2 Design Sample Sizes and Number of Specimens Actually
Collected for Various Media at Chicago, NW Incinerator . 9
3 Summary Statistics for Total Organic Chlorine Concentra-
tion Data from the Ames, Iowa Plant 11
4 Summary Statistics for Total Organic Chlorine Concentration
Data from the Chicago, NW Incinerator 12
5 Summary of Surrogate Compounds Percent Recovery for Speci-
mens from the Ames, Iowa Plant 14
6 Summary of Surrogate Compounds Percent Recovery for
Specimens from the Chicago, NW Incinerator 15
7 Validity of Confidence Statements for Selected Levels of
Bias 17
8 Summary of Coefficient of Variation for Pilot Study. ... 18
9 Estimates of Variance Components 20
10 Recommended Sample Allocations for National Survey .... 22
11 Strata of the Coal Combustion Survey and Sum of Size
Measures (in millions of tons) 24
VI
-------
FIGURES
Number Page
1 General diagram of combustion site indicating possible sam-
pling points and their relationship to certain operations
in the process 5
2 Geographic strata of the coal combustion survey design. . . 25
VI 1
-------
I. EXECUTIVE SUMMARY
The emission of several toxic compounds in the polycyclic organic
group, specifically polychlorinated dibenzo-p-dioxins, dibenzofurans and bi-
phenyls have been reported from stationary combustion processes. It has been
claimed that these compounds are ubiquitous in air and are formed from many
combustion sources. It has been demonstrated that a theoretical potential
exists for the formation of these compounds as the results of combustion.
The most likely sources involve coal-refuse, wood, municipal refuse, waste
oil and coal.
To investigate this topic further, a pilot study was designed to
obtain data on which to base a national survey. The overall objectives of
the pilot study were to ascertain the number of combustion sites and the
number of days of sampling required at each site to adequately estimate the
level and prevalence of these toxic substances in the emissions from combus-
tion processes and to do so at a minimum cost.
The pilot included two sites. One was a coal-refuse burning elec-
tricity generating facility and the other a municipal incinerator. These two
types of facilties were selected because they were in categories that were
judged most likely to emit the substances of interest.
For each facility a complex, multimedia sampling design was devel-
oped for the collection of solid, liquid and gaseous influents and effluents.
In addition, measurements of process parameters were also taken. This design
allowed for the estimation of the inputs into the process, the efficiency of
the combustion process and the emissions from the process.
The level of total organic chlorine (TOC1) was used as a surrogate
for the levels of toxic substances being investigated. The use of a surro-
gate was necessitated because the large number of chemical analyses required
to retain sufficient statistical information to design a national study would
be beyond budgetary restraints. TOC1 was selected because it was believed
that its levels and the levels of the toxic substances of interest would be
correlated.
The TOC1 data was statistically analyzed taking into account the
design and compositing of specimens. The level of TOC1 was in general higher
for the municipal incinerator data and also tended to show more variability
among specimens. The TOC1 levels for the several media common to both fa-
cilities appeared to be significantly different.
The true levels of TOCL at both facilities may be underestimated
because of less than complete recovery of the TOC1 in the chemical analysis.
The underestimation may be more serious for the incinerator estimates.
Using the estimates of the variability of the data and cost esti-
mates based on the experience gained in the pilot, a national survey of two
combustion categories was developed. Sampling is planned for seven coal and
nine refuse combustion facilities for five days each. Estimates of the levels
of toxic substances are anticipated to have a precision of ±50 to ± 60%.
-------
II. INTRODUCTION
A. Background
The emission of several toxic compounds in the polycyclic organic
matter (POM) group, specifically polychlorinated dibenzo-p-dioxins (PCDDs),
dibenzofurans (PCDFs), and biphenyls (PCBs) have been reported from station-
ary conventional combustion processes (DC-USA 1978, 01ie et al. 1978 and Shin
1979). These compounds have been proposed to be ubiquitous in the aqueous
environment and it has been claimed that POMs are ubiquitous in air and are
being formed from many combustion sources. (DC-USA 1978).
A study conducted for the U. S. Environmental Protection Agency
(EPA) (Shin 1979) demonstrates that a theoretical potential exists for POMs
to be formed during and as a result of conventional combustion processes.
These POMs may include PCDDs, PCDFs, and PCBs. The major combustion sources
are suspected to involve such fuels as coal-refuse, wood, municipal refuse,
waste oil and coal.
Because the acquistion of field specimens and their chemical deter-
mination are very costly, it is important to carefully plan studies to as-
certain the level and prevalence of the substances being emitted into the en-
vironment. Because not enough appropriate data were available on which to base
a statistical design for a national survey, a pilot study was conducted spe-
cifically designed to generate the appropriate data.
B. Objectives of the Pilot Study
To achieve the overall objective of the pilot, several specific ob-
jectives were defined; these were:
1. Estimate the variability among sites and among days within sites
of the level and prevalence of selected toxic substances in the POM group;
2. Estimate the relationship (correlations) of the level and pre-
valence of selected toxic substances in the POM group among the various in-
fluent and effluent streams;
3. Estimate the fraction of the total varibility of the level and
prevalence of selected toxic substances in the POM group due to the chemical
analysis;
4. Test the feasibility of the sampling protocols used in the ac-
quisition of field specimens; and
5. To estimate the cost (in dollars) of the various aspects of the
study.
For example, the cost of overall project management; the cost of travel, trans-
portation, and shipping of personnel and materials to and from a site; and
the cost of the various stages of the chemical analysis from extraction through
quantitation and confirmation of specific toxic substances in the POM group
would be estimated.
-------
These objectives evolved from the necessity to acquire specific in-
formation based on the criteria below.
1. Knowledge of the variability is necessary to anticipate the pre-
cision of estimates of the level of toxic substances obtained in the survey
for specific sample sizes.
2. If a strong relationship exists among the various influent and
effluent streams, the cost efficiency of the study may be improved by collect-
ing more samples from less expensive sampling points. One may be able to use
this relationship to improve the precision of the estimates for the more ex-
pensive sampling point.
3. Knowledge of the fraction of the total variability of the level
of toxic substances due to the chemical analysis is necessary to develop cost
efficient compositing protocols which retain sufficient information for proper
statistical analysis of the data.
4. Sampling protocols based on statistical principles allows one
to assess the quality of the data.
5. The cost of various aspects of the study along with the varia-
bility is necessary to design a cost efficient (obtaining the most informa-
tion for a given cost) national survey.
C. Statistical Design for the Pilot Study
Two sites were selected for the pilot study: one, a coal-refuse
burning electrical generating facility Ames Municipal Power Plant, Ames, Iowa,
and the other, a municipal incinerator Chicago Northwest (NW) Incinerator,
Chicago, Illinois. These types of facilities were selected because they were
judged to be most likely to emit chlorinated POMs into the environment (Shin,
1979). Also, because only coal was burned on some days at the Ames, Iowa plant,
some information about coal fired plants could.be obtained.
Because only two sites were to be used to estimate the variability
among sites, they were purposely selected from different categories. This is
expected to result in, at worst, an overestimate of the variability among
sites within a category. Prudence motivated this conservative approach.
1. Number of Days Per Site
Sampling for 14 days at each site was scheduled. This sample is
adequate to detect large correlations among the various media (p >_ .5) at
least 60% of the time. For p > .9, the correlation would be detected at least
95% of the time. Hence, the sample size is adequate to detect correlation of
sufficient magnitude to provide substantial improvement in precision. For
example, a correlation of .866 is required to attain improvement in the pre-
cisions of estimates of 50%.
-------
2. Multimedia Sample Design
The combustion process can be described as consisting of three
phases: (i) input, (ii) combustion, and (iii) emission. The amounts of toxic
substances that are emitted into the environment from a given site depends on
the concentration of the substances and their precursors in the first phase,
the efficiency or completeness of combustion in the second phase and the type
of emission control devices used before the third phase. (Shin et al 1979
and DC-USA 1978)
To obtain the appropriate data to meet the objectives of the study,
the design necessarily involves a complex matrix of multimedia sampling at
different time periods and frequencies incorporating specimen and data col-
lection at all three process phases. The specific intervals used in the pi-
lot study were site specific and were based on such factors as type of fuel
feed or the frequency and mechanism by which effluents such as bottom ash are
removed.
The sampling sites were dispersed in a manner that was intended to
give estimates on the input into the combustion process, efficiency or com-
pleteness of combustion and the emission from the process. The following is
a list of sample points for each phase and the sample point's relative loca-
tions are given in Figure 1.
(i) Input
T! - Probability sampling of the fuel entering the combustion pro-
cess was conducted six times per day. The protocols were
developed after a site visit to each site.
I2 " High volume sampling of ambient air (or intake air) was con-
ducted for each 24 hour period during the pilot survey.
I3 - Intake water was sampled three times during the study.
(ii) Combustion
C1 - Real time monitoring of carbon monoxide (CO)
C2 - Real time monitoring of carbon dioxide (C02)
C3 - Real time monitoring of free oxygen (02)
C4 - Real time monitoring of temperature
C5 - Real time monitoring of hydrocarbons.
(iii) Emissions
E! - Effluent gas sampling before emission control devices was
conducted. More than one sampling interval (period of time)
was desired but was not practical.
-------
PLANTS WITHOUT
AIR POLLUTION CONTROL
AIR (\i J
1W
FUEL
L fc,
rV
WATER w
f
©
COMBUSTION
>
1
- — —
- — —
—
1
1
. AIR AND/OR WATER
GASES 1 1
AND 1 Y
ENTRAINED 1
SOLIDS 1
1 ^
0
GAS
COOLING
RESIDUE
GASES ^
AND
ENTRAINED
SOLIDS
0
^
--N
'3
^_X
DISPERSION f
>
) WATER
{
>„
"
k TREATED
COLLECTION
OF
EMISSIONS
FLY ASH
(WATER)
r
(FLY
ASH
>
r AND WATER)
RESIDUE
' QUENCHING
>
r
(WATER)
1
1
1
T
rV
™
V.
WATER
TREATMENT
1 1
T T
L b£
-N,
FLY ASH
(WATER)
r
(FLY ASH
AND WATER)
r
RESIDUE
^-^
11
WATER (FLY ASH)
O
LAND SEWER
DISPOSAL
SAMPLING LOCATIONS (SECTION 3.3)
Fi gure 1 . General diagram of combustion site indicating possible sampling points and their relationship
to certain operations in the process.
* Adapted from Gorden et al.,
-------
E2 - Effluent gas sampling after emission control devices was also
conducted.
E3 - A high volume sample of ambient air was collected.
E4 - Probability sampling of bottom ash and quenching water several
times daily was planned. The frequency and interval of samples
was determined after site visits were made.
E5 - Probability sampling of effluent water and sludge was planned.
The frequency was determined after site visits were made.
E6 - Probability sampling of collected fly ash from emission control
devices was conducted. The frequencies were determined after
site visits were made.
The choice and exact location of each sampling point depended on
the combustion site. For example, the location where the on-line detection
and measuring devices were located depended on the design and access to ap-
propriate locations near the combustion center.
Exact protocols for probability sampling of specific sites were de-
veloped. These protocols depended on the physical characteristics of the sam-
pling point, the medium to be sampled and the conventional methods normally
used in a similar situation (although not limited to use of "conventional"
methods).
Adequate sample volumes and mass to assure satisfactory chemical
analysis sensitivity were necessary. These volumes and mass were sufficient
to perform the two tier analysis and quality control/quality assurance method
of standard additions. Midwest Research Institute (MRI) recommended sample
volumes and masses for the different media and sample locations.
Sample handling requirements such as on-site stabilization proce-
dures, equipment cleaning specifications, and container requirements were
recommended by MRI. A "Chain of Custody Record" was kept for each specimen
collected.
3. Site Specific Sample Designs
One study site was unit number 7 at the coal-refuse fired electrical
generating plant in Ames, Iowa. The engineering details of the plant opera-
tions and flue gas sampling methods are given in TRWEED (1980). The important
statistical design features are summarized below with additional details in-
cluded in Appendix A. The other study site was boiler number 2 at the Chicago
Northwest municipal incinerator in Chicago, Illinois. The engineering details
of the plant operations and flue gas sampling methods are given in Bakshi et
al (1980). The important statistical design features are summarized below
with additional details included in Appendix B.
-------
At Ames, specimens were collected from ten locations. The three
gas sample locations were: (i) flue gas inlet (from duct before the electro-
static precipitator (ESP)), (ii) flue gas outlet (from stack after ESP), and
(iii) ambient air (located on roof of the plant). The four solid sample loca-
tions were: (i) fly ash (from the ESP ash hoppers), (ii) bottom ash (from
the base of the furnace), (iii) coal (from the feedline leading from the stor-
age bunkers into the gravimetric feeders) and (iv) refuse-derived fuel (RDF)
(from the feeders prior to being pneumatically conveyed to the boiler furnace).
The three liquid sample locations were: (i) bottom ash hopper quench water
overflow (OW) (from the overflow trough), (ii) quench water influent (cooling
tower blow down) (from transport pipes), and (iii) well water (from transport
pipe). The designed sample sizes and frequencies per day are summarized in
Table 1. Also included in the table are the number of specimens actually col-
lected. The collected number of specimens differs from the design because of
physical problems resulting from weather and the operations of the plants.
The sampling schedule varied depending on the medium and location.
The gaseous specimens were time integrated samples over 8-to 13-h periods.
Exact time durations are given in TRWEED (1980). The solid and OW specimens
were collected using a systematic time schedule (every 4 h) based on a random
starting time. Two random starting times were used; one for the first week
and another for the second. The other two liquid media were collected at ran-
domly selected times during the scheduled two week test period. (Appendix A
contains additional details of how the sampling schedule was developed.)
At the Chicago, NW incinerator, specimens were collected from seven
locations. The three gas sampling locations were: (i) flue gas inlet (ESP
inlet), (ii) flue gas outlet (duct leading from ESP to stack), and (iii) am-
bient air (located on roof of the plant). The solid samples were collected
from three locations (i) fly ash (from ESP ash hopper), (ii) combined ash
(from base of incinerator where bottom ash and fly are combined), and (iii)
refuse (from charge hopper at top of furnace). One liquid sample of city tap
water was collected from a pipe entering the building. The designed sampling
sizes and frequencies per day are summarized in Table 2. Also included in
the table are the number of specimens actually collected. The number of spe-
cimens collected differs from the design because of physical problems result-
ing from plant operations.
The sampling schedule varied depending on the medium. The gaseous
specimens were time integrated samples over 7- to 12-h periods. The exact
durations of the samples are given in Bakshi et aj (1980). The solid speci-
mens were collected using systematic time schedule (every 4 h) based on a ran-
dom starting time. Two random starting times were used, one for the first
week and another for the second. The liquid samples were collected at ran-
domly selected times during the study period. (Appendix B contains additional
details of how the sampling schedule was developed.)
At both combustion sites, several sampling locations had more than
one access point (for example, the fly ash could be collected from more than
one hopper). For these cases, an access point was randomly selected for each
sampling time.
-------
Table 1. Design Sample Sizes and Number of Specimens Actually Collected
for Various Media at Ames, Iowa Power Plant
Media
Designed
sample size
(number of specimens)
Number of
specimens
actually collected
Gaseous
Flue gas inlet
Flue gas outlet
Ambient air
Solid
Fly ash
Bottom ash
Coal
Refuse derived fuel
Liquid
OWb
Quench water influent
Well water
14 (1 per day)
14 (1 per day)
14 (1 per day)
84 (6 per day)
84 (6 per day)
84 (6 per day)
84 (6 per day)
84 (6 per day)
6
3
19
11
20
90
88.
llc
67
91
6
3
More specimens were collected but only eleven (11) were chemically analyzed
because of low levels of Total Organic Chlorine (TOC1).
Bottom ash hopper quench water overflow.
-------
Table 2. Design Sample Sizes and Number of Specimens Actually Collected
for Various Media at Chicago, NW Incinerator
Designed Number of
sample sizes specimens
Media (number of specimens) actually collected
Gaseous
Flue gas inlet 14 (1 per day) 11
Flue gas outlet 14 (1 per day) 11
Ambient air 14 (1 per day) 13
Solid
Fly ash 84 (6 per day) 72
Combined ash 84 (6 per day) 76
Refuse 84 (6 per day) 61
Liquid
City tap water 3 3
-------
Continuous monitoring of oxygen (02), carbon dioxide (C02), carbon
monoxide (CO) and total hydrocarbons (TH) was conducted during the gaseous
sampling periods at both sites. These parameters give an indication of the
efficiency of the combustion process.
III. RESULTS OF THE PILOT STUDY
A. Introduction
This chapter summarizes the total organic chlorine (TOC1) data ob-
tained in the pilot study. These data are presented in Haile et al (1984).
More specifically, this chapter includes summary statistics, descriptions of
the compositing protocols for the chemical analysis and statistical analysis
methods, and discussion of the chemical analysis and sampling error. No cor-
relations between the TOC1 levels of the different media are included in the
summary statistics because none were found to be significant.
B. Summary Statistics
To summarize the data, the arithmetic mean, coefficient of varia-
tion (CV) and nominal 95% confidence intervals were calculated for each sam-
pling location at both combustion sites. The arithmetic mean (X) can be cal-
culated for each sampling location by
n
X = I X. / n
1=1 n
where X. is the TOC1 concentration of the ith specimen and n is the number of
specimens. The CV can be calculated by first calculating the sample variance
(S2)
n - 2
S2 = I (X. - X) / (n - 1) .
1=1
The CV = S / X. The nominal 95% confidence intervals are calculated by
(X - t>05 (df) S / VH , X + t 05 (df) S / Vn )
where t>05(df) is obtained from tables of Student's t distribution [Snedecor
and Cochran p. 469 (1980)] and df denotes the appropriate degrees of freedom
which depend on number of independent chemical analyses.
Because some specimens were composited before chemical analysis,
the above formulas were not used to calculate X and S2 for all locations.
However, the calculations were adjusted to take into account the compositing.
These adjustments are discussed below in Section C. The summary statistics
are given in Tables 3 and 4.
10
-------
Table 3. Summary Statistics for Total Organic Chlorine Concentration
Data from the Ames, Iowa,
Plant3
Media (units)
Number of
specimens Mean
Coefficient
of
variation (%)
Degrees
of
freedom
Nominal 95%c
confidence
interval
Gaseous (ng/dscm)
Flue gas inlet 19 562
Flue gas outlet 11 632
Ambient air 20 d
Solid (ng/g)
Fly ash 90 8.3
e (89) (3.6)
Bottom ash 88 58.6
Coal 11 4.4
Refuse-derived fuel 67 11,900
Liquid (ng/L)
49
85
536
(81)
183
23
116
18 (426, 698)
10 (254, 1010') '
50 (-1.0, 17.6)
(49) ((2.9, 4.2))
50 (35.1, 82.1)
5 ( 3.5, 5.3)
36 (8,342, 15,470)
ow1
Quench water
Well water
91
influent 6
3
664
373
54
77
33
32
51
5
2
(570, 760)
(231, 514)
(1.4, 101)
?0riginal data from Haile et al (1984).
Number of independent chemical analyses minus one.
.Nominal value based on normal probability distribution theory.
Measured values in field specimens not significantly different from blanks.
Numbers in ( ) are estimates excluding the maximum value of 210 ng/g. This
value is 21 times larger than the next largest value. Both sets of summary
statistics are included to illustrate the impact of the one extreme value on
the estimates.
Bottom ash hopper quench water overflow.
11
-------
Table 4. Summary Statistics for Total Organic Chlorine Concentration
Data from the Chicago, NW Incinerator3
Number of
Media (Units) Specimens Mean
Coefficient
of
Variation (%)
Degrees
of
Freedom
Nominal 95%c
Confidence
Interval
Gaseous (ng/dscm)
Flue gas inlet
Flue gas outlet
d
Ambient air
Solid (ng/g)
Fly ash
Combined ash
Refuse
Liquids (ng/L)
City tap water
11 2200
11 3220
(10) (2190)
1 1.67
72
67
61
93.6
7.0
902
30
36
109
( 36)
64
85
162
251
10 (1698, 2702)
10 ( 862, 5578)
( 9) ((1330, 3040))
11 (-.68, 4.02)
52 (71.7, 115.6)
50 ( 5.8, 13.9)
50 (283.8, 1,520)
^Original data from Haile et al. (1984).
Number of independent chemical analyses minus one.
.Nominal value based on normal probability distribution theory.
Numbers in ( ) are estimates excluding the maximum value of 13,500 ng/dscm.
This value is 4 times larger than the next largest value. Both sets of sum-
mary statistics are included to illustrate the impact of the one extreme
value on the summary statistics.
Not calculated because there was no variability in the data.
12
-------
C. Compositing and Weighting
To minimize the cost of chemical analysis while retaining sufficient
statistical information, a complex compositing protocol was developed for the
sample locations where more than one specimen per day was collected. The com-
positing varied for the six samples collected each day. On some days all were
composited, on others, the two within a shift were composited, and on others
none were composited. These locations were fly ash, bottom ash, coal, RDF
and OW at the Ames plant and fly ash, combined ash and refuse at the Chicago,
NW incinerator. No compositing was done for the specimens collected at the
other sample locations.
To modify the calculations for X and S2 to compensate for the com-
positing, each chemical determination was assigned a weight equal to the number
of specimens composited. Then the weighted mean Y is calculated by
w
m
I
W. Y. / I
1 1 1=1
W.
where Y. is the ith chemical determination, W. is the number of specimens com-
posited for the ith chemical determination and m is the number of chemical
m
m
determinations. Because I W. = n and on
1=1 n
equals X on average.
average I W. Y. = I X., then Y
i=l n 1 1=1 1 w
To estimate S2 from the composited data, calculate
m 2 m
S5 = . wi
-------
Table 5. Summary of Surrogate Compounds Percent Recovery for Specimens from the Ames, Iowa, Plantc
dg-Naphthalene
Media
Gaseous
Flue gas inlet
Flue gas outlet
Solid
Fly ash
Bottom ash
Coal
Coal-derived fuel
Liquid
OWC
Quench water influent
Well water
Number
of
analyses
18
11
51
42
6
37
40
6
2
Mean
percent
recovery
56
47
44
55
90
64
51
69
66
Coefficient
of
variation (%)
45
25
56
36
18
22
54
25
1
Number
of
analyses
19
11
51
49
6
37
48
6
3
d]2~Chrysene
Mean
percent
recovery
71
86
96
85
90
111
88
111
88
Coefficient
of
variation (%)
27
14
24
37
19
25
29
16
29
JTlata from Haile et al. (1984).
Specimens that were inadvertantly evaporated to dryness were excluded.
-Bottom-astrquench water overflow.
-------
Table 6. Summary of Surrogate Compound Percent Recovery for Specimens from the Chicago, NW Incinerator'
d8-Naphthalene
Media
Gaseous
Flue gas inlet
Flue gas outlet
Ambient air
Solid
Fly ash
Combined ash
Refuse
Liquid
City tap water
Number
of
analyses
11
11
12
53
33
44
3
Mean
percent
recovery
37
27
31
26
35
9
27
Coefficient
of
variation (%)
84
98
75
68
57
51
131
Number
of
analyses
11
11
12
52
33
44
3
d12-Chrysene
Mean
percent
recovery
74
62
51
36
22
12
13
Coefficient
of
variation (%)
48
82
88
61
105
193
92
Data from Haile et al. (1984).
-------
If the percent recoveries in these tables are indicative of the re-
covery rate for TOC1, then the concentrations of TOC1 are under estimated.
This under estimation would be greater for the specimens from the Chicago, NW
incinerator, than those from the Ames plant. However, the summary statistics
reported in Table 3 and 4 above are not adjusted for percent recovery, as
recommended by MRI. Biases of this type can affect the true confidence of a
nominal 95% confidence interval. Table 7 illustrates the impact of various
levels of bias on the true confidence of a nominal 95 percent confidence in-
terval.
Table 8 summarizes the estimates of the CVs (S/X) for both the sam-
pling and measurement (as indicated by the surrogate recovery data) component.
One should note that the measurement CVs for the Ames plant are uniformly less
than those for the Chicago, NW incinerator. In fact, for some sampling loca-
tions at the Chicago, NW incinerator, the measurement component dominates the
total variability giving negative estimates of the sampling component. This
is not unexpected for the ambient air and city tap water because at these two
locations one would expect the media to be rather homogeneous. However, this
is unexpected at the flue gas inlet. Note that the measurement CV is larger
than the sampling CV at the inlet but the opposite is true at the outlet. In
this report, no physical explanations are attempted for this phenomena.
The design of the national survey of combustion sites is based on
the CV of the flue gas outlet. Two important factors motivated this decision;
(i) approximately 75 to 80% of the total mass of TOC1 is emitted through the
flue gas; and (ii) flue gas is, by far, the most expensive location to sample.
Even though the CVs for some of the other sampling locations are
much larger than the flue gas outlet CV, the precision for these locations
can be easily controlled by the number of specimens collected per day with
little effect on the total cost of the survey. Chapter IV discusses the de-
velopment of the national survey using the TOC1 data obtained in the pilot.
IV. NATIONAL SURVEY DESIGN DEVELOPMENT
A. Introduction
In this section the techniques used to determine the number of sites
to be sampled and the number of days to sample at each site are discussed.
This section also includes the discussions on how the sample is to be dis-
persed across the U.S. and what procedures are used to select the sampling
sites. The first subsection is a discussion of how these estimates of the
variability anticipated in the survey are formulated.
B. Variance Components
Because more than one combustion site will be sampled and each site
will be sampled for more than one day, the total variability in the data can
be partitioned into two components. One due to the differences in the average
emissions among the sites and the other due to the differences in the average
16
-------
Table 7. Validity of Confidence Statements for Selected Levels of Bias
BIAS/SEb
0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
True Confidence Level3
for the x ± 1.96 x SE Interval
0.95
0.92
0.83
0.68
0.48
0.29
0.15
0.06
0.02
Calculated according to the integral of the
1.96 + BIAS/SE
/
V 2 n
-1.96 + BIAS/SE
SE denotes the standard error of the estimate and is equal to the standard
deviation (a) divided by the square root of the sample size (Vn).
17
-------
Table 8. Summary of Coefficient of Variations for Pilot Study3
Ames Chicago
Media Sampling Measurement Sampling Measurement
Gaseous
Flue gas inlet 42 25 b 68
Flue gas outlet 83 13 85 68
Ambient air c c b 87
Solid
Fly ash 555 (78)c 24 164 64
Bottom ash 179 38
Combined ash 143 76
Coal 12 19
Refuse-derived fuel 114 18
Refuse 194 159
Liquid
OW 58 38
Quench 2ater
influent 17 28
City tap water b 132
aThe measurement CVs present above are a weightedjaverage of the CVs in Tables
5 and 6. They were calculated by CV = (Sg + S^)^/(X8 + X12). Where the
subscripts a and 12 denote d#-naphthalene and d12-chrysene respectively.
The estimates of these values were negative and were excluded because the
CV must be non-negative.
Not calculated because specimen amounts were not significantly different than
blanks.
Number in ( ) are estimates excluding the maximum value of 210 ng/g. This
value is 21 times larger than the next largest value. Both sets of summary
statistics are included to illustrate the impact of the one extreme value on
the estimates.
18
-------
emissions among days within each site. This concept can be formulated mathe-
matically using the model
°f • «X + «8-
where a? denotes a measure of variability among the average emissions of all
plants and c£ denotes a measure of the variability of the average emissions
among days within all plants. One purpose of the pilot was to gather data to
estimate these quantities. Motivated by the hypothesis that these quantities
would differ depending on the type of combustion process, two different types
of combustion processes were selected for the pilot.
Two major categories were considered, refuse and coal combustion.
The estimates of the variance components are given in Table 9 with statisti-
cal details included in Appendix C. Note that even though the absolute mea-
sures of variability for refuse combustion are much larger than those of coal
combustion, the relative measures (percent of average concentration) are only
slightly larger. This is a common occurence among data from many different
sources.
C. Cost Components
To determine the sample sizes for the two categories of combustion
processes, the costs of sampling and chemical analysis were considered. The
estimated costs of the survey were broken down into two components; (i) those
associated only with the number of sampling sites (such as travel) and (ii)
those associated with the number of days of sampling (such as per diem and
chemical analyses). The chemical analysis cost can be associated with the
number of days because samples will be composited for daily averages. This
concept can be formulated mathematically using the model
where C is the total cost, G! the cost associated with the number of sites,
nt the number of sites, C2 the cost per day and n2 the number of days of sam-
pling at each site. Hence, for each combination of nx and n2, the total cost
of the survey can be estimated.
The variance component and cost models above can be used to deter-
mine the best combination of nx and n2. This procedure is discussed below.
D. Optimum Allocation
One common measure of the precision of an estimate is the standard
error (SE). This quantity depends upon the inherent variability in the com-
bustion processes being studied and the number of sites and days of sampling
conducted. The SE is calculated by
a* 0-2
SE = A ' W
n^ nin2
where a?, of., nit and n2 are the same as above.
M W
19
-------
Table 9. Estimates of Variance Components
Refuse Coal
Parameter combustion combustion
02 3,340,000 99,000
o-J 12,200,000 289,000
CVAa 57% 50%
CVwb 109% 85%
CV^ denotes coefficient of variation among plants.
CVW denotes coefficient of variation among days within plants.
20
-------
Because the average concentrations differ greatly between the two
categories, the SE are not easily comparable. To compensate for the differ-
ence, the relative standard error (RSE) is used and allows for direct com-
parison of the two groups on an equal scale. The RSE is calculated by
CVA CVW
RSE = A ' W
where CV. and CVW are the coefficients of variation among sites and among
days witnin sites respectively.
The nl and n2 are determined so that the RSE is less than or equal
to 25% and the total cost (C) is minimized. The combination of nt and n2 is
refered to as the optimum allocation. The recommended sample size allocations
are given in Table 10.
To estimate the anticipated size of the 95% confidence intervals
for the estimates, one has to make assumptions about the relative variability
among sites and among days within sites. If the among days within sites var-
iation is large relative to the among sites variability, one can make the as-
sumption that the correlation among days within sites is close to zero. With
this assumption, an estimate for the number of degrees of freedom associated
with an estimate of the precision is approximated by the number of sites and
the number of days per site less 1 (or n1 x n2 - 1). That is, the sample is
tending to be like a simple random sample of days. On the other hand, if the
among days within sites variability is small relative to the among sites vari-
ability one assumes that the correlation among days within sites is high or
close to one. Under this assumption, an estimate of the number of degrees of
freedom for an estimate of the precision is approximated by the number of
sites less 1 (or nj - 1). Conceptually if the CV among days within sites is
small, the number of days per site offer no additional information to explain
the variability. That is, the sample is tending to be a random sample of
sites with information for only one day per site.
The estimate of the width of the 95 percent confidence interval for
the coal combustion sites are based on either 34 degrees of freedom or 6 de-
grees of freedom providing a lower and upper bound, respectively, for the an-
ticipated precision. Using the equation,
Relative 95% Confidence Interval Half Width = t 05(df) RSE,
the anticipated precision was calculated and presented in Table 10. The ap-
proximated degrees of freedom for estimates from the refuse combustion sites
would be either 44 or 8 assuming little or no correlation or high correlation
among days within sites, respectively. The anticipated 95% confidence inter-
vals for the refuse combustion sites are also given in Table 10.
, The next phase of design development involves constructing an in-
ventory of potential sampling sites (sampling frame), defining important
factors to be considered in partitioning the sampling frame into subsets
(stratification) and the methods of site selection. These are discussed be-
low.
21
-------
Table 10. Recommended Sample Allocations for National .Survey
Combustion
category
Number of
sites
Number of
days per
site
Anticipated
relative
standard
error (RSE)
Range for anticipated
Nominal 95%
confidence interval
Assumption 1 Assumption 2
Refuse
Coal
9
7
5
5
25%
25%
± 50% ± 58%
(t = 2.017) (t = 2.306)
± 513£ ±61%
(t = 2.034) (t = 2.447)
Assumptions for range of anticipated nominal 95% confidence intervals are as
follows:
Assumption 1:
Assumption 2:
essentially zero correlation between days within the same
plant implies that the number of degrees of freedom can be
approximated by one less than the product of number of plants
and the number of days per plant. (For coal plants 7 plants
x 5 days per plant -1 = 35, similarly for refuse plants.)
high correlation between days within the same plant implies
that the number of degrees can be approximated by one less
than the number of plants.
22
-------
E. Sampling Frame
1. Coal Combustion
The scope of this phase of the national study is narrowed to a par-
ticular type of coal combustion category. The frame includes only large
(greater than 108 BTU/h) coal burning electricity generating plants. The
frame is a subset of the National Emissions Data System (NEDS) computer file.
The sampling frame is given in Appendix D. Each sampling unit is a point
source (boiler or group of boilers associated with a particular stack) of
emissions.
2. Refuse Combustion
The refuse combustion category will include those facilities that
burn refuse for all or part of their fuel. Included will be municipal refuse
incinerators and coal-refuse burning sites. The frame will be compiled using
information in Gordon et al., the NEDS computer file and other supplementary
sources. Completion of this frame was postponed until immediately before the
refuse portion of the survey will be conducted.
F. Stratification of Coal Combustion
Because the total amount of emissions was felt to be highly corre-
lated with the amount of coal burned by each plant, the number of tons of coal
burned annually was chosen as the measure of size of each plant. In order to
distribute the sample geographically, the U.S. was partitioned into seven
groups (strata) of contiguous states. The groups were arranged so that the
sum of all the size measures (amount of coal burned annually in the stratum)
were approximately equal. Table 11 lists the states and the size measures
for each stratum. Figure 2 illustrates the strata.
G. Sample Selection of Coal Combustion
One point source was selected at random from each stratum; the prob-
ability of its selection was proportional to the size measure. This increased
the likelihood that the sample was weighted towards the large emission sources.
Alternates or supplements may be selected, if necessary, using the same methods.
Using a random selection technique allows for unbiased estimates of
the average emission and estimation of the precision of the estimates. Confi-
dence intervals based on the survey data will give ranges for the true emission
values. Even though it is anticipated that the confidence intervals will be
wide for the coal combustion category, by combining the estimates obtained
for other categories using similar methods, relatively precise (narrow confi-
dence intervals) estimates of the total emissions from combustion sources will
result.
23
-------
Table 11. Strata of the Coal Combustion Survey and Sum
of Size Measures (in millions of tons)
North East 79.5
Maine
New Hampshire
Vermont
Massachusetts
Rhode Island
Connecticut
New York
New Jersey
Pennsylvania
Delaware
District of Columbia
Maryland
West Virginia
South East 72.5
Virginia
North Carolina
South Carolina
Georgia
Florida
Alabama
Ohio Valley 74.9
Ohio
Kentucky
Great Lakes 64.2
Michigan
Indiana
Wisconsin
0.0
0.8
0.0
0.0
0.0
0.0
6.3
2.4
38.5
0.7
0.0
4.4
25.8
4.8
20.4
6.8
17.7
6.1
16.7
44.9
30.0
21.3
31.2
11.7
North Central 75.0
Minnesota
Iowa
Missouri
11 1 inois
South Central 63.6
Tennessee
Mississippi
Arkansas
Louisiana
Oklahoma
Texas
West 69.9
North Dakota
South Dakota
Nebraska
Kansas
Montana
Wyoming
Colorado
New Mexico
Idaho
Utah
Arizona
Washington
Oregon
Nevada
California
Alaska
Hawaii
12.1
8.9
21.2
32.8
22.0
1.6
7.0
4.5
2.1
26.1
7.5
2.4
1.9
7.1
3.2
16.2
8.8
8.0
0.0
2.5
1.6
4.2
0.0
4.0
0.0
0.5
0.0
24
-------
Figure 2. Geographic strata of the coal combustion survey design.
-------
REFERENCES
Bakshi PS, Sarro TL, et al. 1980. Pilot test program Chicago Northwest In-
cinerator Boiler No. 2 (Engineering Report). TRW Environmental Engineering
Division, TRW, Inc. Office of Research and Development, USEPA, Research
Triangle Park, NC.
Cochran WG. 1963. Sampling techniques. New York: John Wiley and Sons.
DC-USA. 1978. Dow Chemical U.S.A. The trace chemistries of fire - A source
of and routes for the entry of chlorinated dioxins into the environment. The
Chlorinated Dioxin Task Force, the Michigan Division.
Draper N, Smith H. 1966. Applied regression analysis. New York: John Wiley
and Sons, Inc.
Gordon J, Helfund R, Belew W. 1976. Preliminary study for PCB municipal in-
cinerator test study. Report prepared by Mitre Corporation for the U.S. En-
vironmental Protection Agency. MTR-7305.
Graybill FA.
Scituate, MA:
1976. Theory and application of the linear model.
Duxbury Press.
North
Haile CL, Stanley JS, Lucas RM, Melroy DK, Nulton CP, Yauger WL, Jr. 1984.
Comprehensive assessment of the specific compounds present in combustion pro-
cesses: Vol. 1. Pilot study of combustion emission variability. Final re-
port. Environmental Protection Agency. Contract 68-01-5915. EPA 560/5-83-
004, NTIS PB-84-140-870.
Hamersma JW, Reynolds SL, Maddalone RF. 1976. IERL-RTP Procedures manual:
level 1 environmental assessment. EPA-600/276-160a.
Kish L. 1965. Survey Sampling. New York: John Wiley and Sons.
Lentzen DE, Wagoner DE, Estes ED, Gutknecht WF. 1979. IERL-RTP procedures
manual: level 1 environmental assessment (Second Edition). EPA-600/7-78-201.
Lucas RM, Zweidinger RA, Westbrook W. 1980. Sample design for pilot study
of stationary conventional combustion processes. Draft report. Washington,
DC: Office of Toxic Substances, U.S. Environmental Protection Agency. Con-
tract 68-01-5848.
Lucas RM, Matthews BJ, Haile CL, Westbrook W, Zweidinger RA. 1980. Detailed
work plan: Pilot study for dioxin activity. Washington, DC: Office of Toxic
Substances, U.S. Environmental Protection Agency. Contract 68-01-5848.
Moore DR, Korner RW, et al. 1980. Pilot test program Ames Municipal Power
Plant Unit No. 7. Sampling plan. Research Triangle Park, NC: Office of
Research and Development, U.S. Environmental Protection Agency. Contract
68-02-2197.
26
-------
Moore DR, Korner RW, et al. 1980. Pilot test program Chicago Northwest In-
cinerator Boiler No. 1. Sampling plan. Research Triangle Park, NC: Office
of Research and Development, U.S. Environmental Protection Agency. Contract
68-02-2197.
01ie KP, Vermeulen L, Hutzinger 0. Undated. Chlorodibenzo-p-dioxins and
chlorodibenzofurans are trace components of fly ash and flue gas of some
municipal incinerators in the Netherlands. Chemosphere 2:105-172.
Shin C, Ackerman D, Scinto L, Moon E, Fishman E. 1979. POM emissions from
stationary conventional combustion processes with emphasis on polychlorinated
compounds of dibenzo-p-dioxin (PCDD), biphenyl (PCB) and ibenzofuran (PCDF).
Draft report. U.S. Environmental Protection Agency.
TRWEED. 1980. Pilot test program Ames Municiapl Power Plant Unit No. 7.
Engineering report. Research Triangle Park, NC: Office of Research and De-
velopment, U.S. Environmental Protection Agency. Contract No. 68-02-2197.
Williams RS, Bursey J, Erickson MD. 1979. Draft Review of TRW Report.
27
-------
APPENDIX A
AMES, IOWA PLANT SOLID AND LIQUID SAMPLING PROTOCOL
28
-------
At Ames, Iowa five non-gaseous locations were sampled six times per
day. These were bottom ash (BA), fly ash (FA), coal (CO), refuse-derived fuel
(RDF) and bottom ash quench water overflow (OW). A systematic time schedule
was developed using the following method.
Because plant cooperation was necessary to obtain BA samples, the
schedule was based on BA. Two cycles of the sampling order and time between
sampling is given below.
BA RDF CO FA OW BA RDF CO FA OW
1:00 0:30 1:30 0:30 0:30 1:00 0:30 1:30 0:30
The schedule was based on the times of the working shifts at the plant. They
were: (1) 11:00 PM to 7:00 AM, (2) 7:00 AM to 3:00 PM and (3) 3:00 PM to 11:00
PM. The plant restricted the BA sampling for operational reasons, from 1:30
after a shift began to 6:30 after the shift began. To be able to collect two
specimens of BA within that time period, little flexibility in the time of
the BA sampling was allowed. The allowed times on which to base the schedule
were 12:30 AM, 1:00 AM and 1:30 AM. Two times were selected with equal prob-
abilities and without replacement. 1:30 AM was selected for the first week
of sampling and 12:30 AM for the second. The resultant schedule is given in
Table A.I for these media.
In addition to the time schedule, the specific location where a
specimen would be collected had to be identified. OW had only one access point,
hence no randomization was required. However, for RDF, CO, and FA there were
four, two and two access points respectively. For these locations, an access
point was randomly selected with equal probabilities. The BA sampling was
more troublesome. Because of the physical restriction imposed on sampling by
the furnace structure, it was impossible to obtain cores of the BA, only
surface samples. The surface was divided into an imaginary grid and specimens
were selected from randomly selected areas of the grid with equal probabilities.
In addition to the above locations, bottom ash quench water influent
(cooling tower blowdown (CTW)) and well water were sampled. The CTW was sam-
pled once per day at a time selected with equal probabilities. The well water
was sampled three times during the study. First three days were randomly se-
lected with equal probabilities and then a time was selected for each day with
equal probabilities.
29
-------
Table A-l Sampling Schedule for Ames, Iowa Plant (Military Time)
Week One Week Two
FA OW BA RDF CO FA OW BA RDF CO
0030 0100 0130 0230 0300 2330 2400 0030 0130 0200
0430 0500 0530 0630 0700 0330 0400 0430 0530 0600
0830 0900 0930 1030 1100 0730 0800 0830 0930 1000
1230 1300 1330 1430 1500 1130 1200 1230 1330 1400
1630 1700 1730 1830 1900 1530 1600 1630 1730 1800
2030 2100 2130 2230 2300 1930 2000 2030 2130 2200
30
-------
APPENDIX B
CHICAGO. NW INCINERATOR SOLID AND LIQUID SAMPLING PROTOCOL
31
-------
At Chicago, NW incinerator three nongaseous locations were sampled
six times per day. These were fly ash (FA), combined ash (CA) and refuse (RF).
Combined ash results from the mixing of bottom ash and fly ash in the bottom
ash hopper before sampling is possible. A systematic time schedule was de-
veloped using the following method.
No plant cooperation was necessary to obtain specimens, hence there
was complete flexibility in arranging the schedule. The sampling schedule
was based on FA. Two cycles of the sampling order and times between sampling
are given below.
FA CA RF FA CA RF
1:00 2:00 1:00 1:00 1:00
The schedule was based on the times of the working shifts at the plant. They
were: (1) 11:00 PM to 7:00 AM, (ii) 7:00 AM to 3:00 PM and (iii) 3:00 PM to
11:00 PM. A random time to begin collecting FA was selected each week. This
time was selected with equal probabilities. The ordering above determined
the times for the other locations. The resulting schedule is given in table
B.I.
In addition to the above locations, city tap water (CTW) was also
sampled three times. Three days were randomly selected with equal probabil-
ities and random times within those days were selected with equal probabil-
ities.
More than one access point was available for RF and CA. The physi-
cal characteristics of the RF charge bin prohibited core sampling or even ran-
domly selected surface sampling. Specimens were obtained from one side or
the other, each side selected with equal probabilities. Because of the phys-
ical characteristics of the RF, several cubic feet of RF were collected,
homogenized and subsampled. The CA bin was partitioned into five equal areas.
The area to be sampled was then selected with equal probabilities. The FA
was collected from a single access point in a transport duct.
32
-------
Table B.I Sampling Schedule for Chicago, NW Incinerator
(Military Time)
Week One
CA RF FA
2300 0100 0200
0300 0500 0600
0700 0900 1000
1100 1300 1400
1500 1700 1800
1900 2100 2200
Week Two
RF FA CA
2300 2400 0100
0300 0400 0500
0700 0100 0900
1100 1200 1300
1500 1600 1700
1900 2000 2100
33
-------
APPENDIX C
THEORETICAL JUSTIFICATION OF ESTIMATION FORMULAE
USED IN STATISTICAL ANALYSIS OF COMPOSITED DATA
34
-------
Let us consider a random vector {X.}, i = 1 to n. For the purposes
of this discussion we consider the components of the vector to be independent
and identically distributed with mean u and variance a2. Hence
E [X.] = M i = 1,. . ., n
and
Var[X.] = a2 i = 1,. . . , n
Then for
n
X = Z X-./n
and
s2 = Z (X. - X)2/(n - 1)
we have E [X] = u and E [s2] = a2.
Two statistics based on composited data which are analogous to X and
s2 are:
m m
YW = Z Wi Y1 / Z W1
and
m m
s§ = Z W2 (Y. - YJ2 / Z W.,
1=1 W i=l 1
where Y. is the i-th chemical determination obtained from compositing W.
number of specimens (a subset of {X.}).
The mathematical expectations are now calculated for Y and s2 for
comparison with those of X and s2.
m m
E [Y ] = Z W. E [Y.] / Z W. = u.
w i=l 1 1 .1=1 n
because Y. is the average of a subset of X.. So Y and X are unbiased esti-
mators of the mean u.
35
-------
To compare the expected values of s2 and s2 we first partition a2
into two components, one due to the inherent variability of the media (denoted
by a2) and the other due to the measurement error (denoted by a2). Hence we
can write m
a2 = a? + a2 .
and
of
= w: + "m •
The term af is divided by W. because Y. is an estimate of the mean
of a subset of W. members of {X.}. n ""
To calculate E [s2] we first calculate
W
m
E [(n - 1) s2] = E [ I W2 (Y. - Y
w 1=1 i i w
where n = I W.. This expectation can be rewritten as
1=1 1
m m
E [ Z W2 (Y. - YJ2] = Z W? E [(Y. - u)2]
1=1 n W 1=1 1
m
+ 2 I W? E [(Y. - (j)(Y
1=1 n 1 W
m
+ I W2 E [(Yw - M)2] (C.I)
1=1
The expectation of the three terms in equation (C.I) are now derived. To cal-
culate the expectation of the first term, we have
m m
I W2 E[(Y. - |j)2) = I W2 Var[Y.]
1=1 n n 1=1 1 1
= Z W2 (a2/W. + a2)
1=1 1 n m
m
= n af + a2 I W2 . (C.2)
1=1
36
-------
For the second term in equation (C.I) we obtain
m
Z W? E[Y. - u) (Y -
1=1 n 1 W
m m
= Z W2 E[(Y. - M) 2 W. (Y. - u)/n]
1=1 1 1
m
= Z W2 (W. Var[Y.]/n
1=1 n
m m
= of Z W2/n + a2 I Wf/n . (C.3)
1=1 n m 1=1 1
No covariance terms appear in this expression because Y. and Y., i ^ j, are
functions of disjoint subsets of the {Y.. } and thus are independent.
For the third term we have
m
Z W? E[(YW -
i 1=1 W
m m
= Z W? E[( Z W. Y./ Z W. - u)2]
; 1=1 ! j=l J J j=l J
m
= Z W2 E[Z W? (Y. - u)2]/n2
1=1 n
m m
= Z W2 Z W2. Var[Y.]/n2
1=1 j=l J
m m
= of I W?/n + a2 ( Z W2)2/n2 (C.4)
1=1 1=1
Substituting the results in equations (C.2 - C.4) into equation (C.I) and di-
viding by (n-1) we obtain
m
E[S2] = n a\ (1 - Z W2/n2) / (n-1)
w 1=1 n
mm m
+ a2 ( Z W2 - 2 ( Z W2)2 / n2 + Z W? / n2) (n-1) (C.5)
m 1=1 1 1=1 n 1=1 n
37
-------
For the case of no compositing, w. = 1 and m = n, equation (C.5) reduces to
E[SJ] = a? + 0-2 = E[S2] = a2
Hence, the bias resulting from using S2 to estimate a2 depends on the w. and
a2 and a2. Using the surrogate recovery data, an independent estimate of a2,
denoted By S2 can be obtained. If E[S2] = a2 then it is not difficult to sRow
that
mm m
S| = {S2 - S2 ( I W2 - 2 ( I W2)2/n2 + I W2/n2)/)n-l)}
w m 1=1 n 1=1 1 1=1 n
m .1
(n (1 - I W2/n2)/n-l}
1=1 n
is an unbiased estimator of af. That is E[Sf] = af. Hence S2 = Sf + S2 is an
m
unbiased estimator of a2 = af + a2
38
-------
APPENDIX D
COAL COMBUSTION SAMPLING FRAME
39
-------
v, 1 „ | t n ! --
CPSPLN^AD •
1 COLP.ERT STEAM PLANT TVA CHATT TTfJN 77101
2 ALA FLFC COOPERATIVE US HWY ? <» P. AH T T 7-61 2
3 ALA. POWER GADSDFN GOODYEAR AVE. G40SDEN
1 ALA POWER-GREENE COUNTY P 0 QRAWP P?f732
5 WIOOWS CREEK STEAM PLANT
f- ;LA POWER CO J H r-'ILLFR STA;
7 US PIPE ?. FOUNDRY 7rOO 35TH AV !>.'
!> AI.APAMA POWER CO P A •< ,1 Y STEAM PLANT
* ALA POWER CO VIADUCT ROAD CHICKASAW
10 ALA POWER CO E.C.C-ASTON UILSC'-.'VILLF3Slfif.
11 ALAPAWA POWER, GOROAS5ROUTE ?tnOvll 35560
12 ALAF-AMA ELECTRIC COOPERATIVE I.EPCY 36518
0 f ' S P L K M A n
13 AAC/E)E,'-'UK ELMENOORF AFB AK 99506
11 HUM C I PAL UTILITIES SYS 615 5TH AVF99707
1? H°«US AP. 1Y AL^^KA a^Af;f,'-I AT-OSEA WA9H719
16 /GOLTEN VALLEY F. BSS.M EOX1.719 FBI lK^-9707
I7 ". CLEAR HEWS USiF 13 MWS ADC Sf. ATTLES-P 1 01
O
|
CP", PLNMAO
1ft NAVAJO ST'Af GEN PLiMT SFF PO FOX W PA
1- AM? "UP SER CO CHGLLt STA JOSEPH CITY
C',Z PLMXAD
?0 SCUTHWrSTERN ELECT . POVE P , GENTRY C1C107
21 ARKANSAS Ff.L, I NO EPE NOEN'CE , NEWAFK320012
22 A = K. POWER S LIGHT CO, WHITE CLUFF 3fOllO
COUMTY
POO
eeo
I 200
Id 00
1920
I9f 0
1900
2100
2100
3060
33PO
3100
COUNTY
IPO
1RO
1BO
620
620
COUNTY
200
52C
COUMTY
160
1200
1260
S T N t, H f
A L A R A !•' A
/. L A r A H A
A L A H A I-1, A
ALA?', t-'A
A L .A "? A M A
ALAMAHA
AL Art AHA
1- LARA M A
ALA6AC-A
ALABAMA
A L A (i A M A
ALABAMA
STMAME
ALASKA
ALASKA
ALASKA
k L 4 S K *
ALASKA
STNAME
ARIZONA
A R I 2 C .'•! A
STNAME
ARKANSAS
A P. K J N S A S
ARKANSAS
P L I. >, T
0010
C001
'JO OP
nooi
000 1-
Foni
0 3 ': 0
1001
1 002
o o n c
OGC1
0001
PLANT
OOC1
0002
OOCi
com
3005
F L 4 N T
0001
0001
PLANT
0001
0007
C011
PO It. K.
•-..
.';
1
P
i\
1
7
•=.
1
r
7
1
POINTS
1
1
1
1
1
POIMTS
7
1
POINTS
1
-
2
-------
OPS
<"3
24
25
26
21
?e
2=5
20
31
*2
-Or S
T 7
-• o
OrS
3."
3f
Or
56
?7
3C
39
40
1 1
oe-s
i.?
43
44
45
46
47
4?
49
eo
STATE=6 -
FLNMAD
PUFLIC SERVICE CO 619S FRANKLIN St PEK C
FUP.LIC SERVICE CO V/VLWONT 1792 M 63RO B'
PUrLIC SERVICE CO ARAPAHOE ?601 S PLATT
DEPARTMENT OF PU?LK UTILITIES DRAKE PL
SOUTHERN COLO POVf.R OIV U HWY 50 CANON C
^UFLIC SERVICE CO OF COLO CAMEO PLANT
COLORAOO-UTE ELEC A?SN NUCLA COLO
COLORADO-UTE JIK PULLOCK PLANT MONTFOSE •
FUPLIC SERVICE CO COMANCHE ?TH STA FUTR
COLCRADO-UTE ELECT ASSOC HAYDEN COLO
PL MM At)
PIERCE CF:NERATINO STATIIOO JOHM UALLIOFD
C T t T f — D
PLNK40
CELKARVA POWER INDIAN RIVER KILLSBCRO
E ! OUPOMT NYLON PLT SE-AFCor; 19 = 73
\
; PLNNAD
GULF POWER CO ST RD 391 LYNI,1 HAVEN
FLA POWER CORP RED LEVEL FLA.
GULF POWER CO CRIST PL«NT PEU!? ACOLf.
MG PF.N-0 TECC BOX Ttlll TC"P« FLOFIOA
GULF PP'JEP CO ST F:3 271 SNEAC'K
. CITY OF NEW SMYRNA PCH BOX 519 32069
PLNK&D
GA FOV.'ER CO-POUEN TAYLORSVILLE 3017F
GF.CPGIA KRAFT COMPANY POX 32-1S
GA POKER CC-ARKWRIGHTMACON 71208
G« POWER CO-MCOCNOUGH SMYRMA 300SP
G4 PWR-YATES SX71E N-IWNA.N 30263.
F& POWER CO-C.ITChFLL iL?AMY 31701
GREAT SOUTHERN PAPER CO
f:A POWER CC-HA.MfOMD CCOSA 30129
GA. PCUiTR CO-PLAMT U ANS LE Y, POOFV IL LE
COUNTY
20
220
600
760
SfcP
1520
If 10
]£<(0
1810
1920
COUNTY
70S
COUNTY
210
210
COUNTY
200
580
1 ir,0
1PQO
1910
1600
COUNTY
380
160
160
1160
1420
1760
I860
2140
2720
STNAN-;
COLOR 400
COLORADO
COLORADO
COLORADO
CCLORAHO
COLORADO
COLORADO
COLCSAOO
COLOnAHO
COLORADO
STNAME
CONNECTICUT
STNAME
DELAWARE
DELAWARE
STMAME
FLOR ICf
FLORIDA
FLORID 1
FLORIDA
FLOP IDA
FLOP IDA
STNAME
GEORGIA
GEORGIA
GEOP.GIA
GEORGIA
GEORGIA
GEO'.GIA
GEORGI i
GEORGIA
GEORGIA
P L £ N T POINTS
0 D 0 1 1 .
000 1 1
n n o '• i
0004 3
0002 2
OU02 • 1
COC1 3
0 0 0 S ?.
0003 2
0001 2
F L « M T POINTS
6 M 4 ?
PLANT PCIf.'TS
0001 3
C002 1
P L A \' T = 0 I N T 5
0014 2
0 0 0 « 1
004? 4
0071 1
OC 11 2
0003 I
PLAMT POINTS
0011 3
0001 2
OC02 4
0003 2
C001 7
ooo? ;
0001 2
0003 4
OOC1 1
-------
ro
CHS
51
= 2
OPS
53
51
55
56
57
50
59
60
61
f>2
6 3
61
65
66
67
6H
6.9
70
71
72
73
74
7=.
7i
77
7»
79
flO
81
e?
P3
C: S
P"
P^-
86
(-. 1
Pf
j- q
90
;, | IT | L - 1 1 -
p L N n A o
CA POWER CO-RRANCH Ml LLEOGE VI LLE 31061
CRISP CO POU'ER COMM WARWICK 31706
PLIIM.AD
COM -ID - KINCAID GEM HI. V. OF KINCAID
WIMNETKA ELECTRIC PL725-735 TOWER RO
COV EO - CRAWFORD ST3501 S. PULASKI FO«D
COVED - FISK STATIOllll W. CERMAK
CENTRAL ILLINOIS PURSIISAL
CENTRAL ILLINOIS LIGOUCK CREEK GEN
CENTRAL ILLINOIS PUBRURAL GRAND TOWER
CENTRAL ILLINOIS PUR
CC'1 En - WAUKEC-AN ST2POO NCRTHWE-STEF N AV
CITY OF PERU GEf:EP.ATl
-------
OS
51
9?
93
94
95
9ft
97
9 R
o q
100
101
102
103
104
105
106
107
10P
10°
110
111
112
> 113
** 114
1 15
116
CFG
1 17
IIP
115
120
121
122
123
124
I 2?
126
127
1 ?P
1 ?c:
130
1 31
133
13?
134
135
1 36
STATE =15
F L N MAO
FSI-K'OFLSVLLE GEN STA,R°2 P-OY 3-5A, 46060
MPSCO R.M.SCHAH.FER 5TA RRM1 EOX^.6 46392
CLIFTY CREEK IKEC BOX 97 HWY 56 R 6? MAD
EDwnSPOKT STA PS I RR 1 EOUAROSPORT 17528
C'CGTHEP-N INDIANA FURLIC SERVICE COMPANY-
COP EO STATELINE GEN 103RD I LK MICHIGAN
MICHIGAN CITY-NIPSCO
DELCC REKY PLAMT «1
DELCO REHY PLANT S3
IMDPLIS PRL-STQUT STA 3700S HARD If.'S462 06..
INTPLIS PUR f. L.FERRY-K STA. 336 KY AVE
INDPLS FWRRL = ERRY U STA, 744 WASHNGTN AV
FEP.U EUEC LIP 3P1 E CANAL ST 46^70
CR'JFOSVLLE tLRP PO 428 CRAWFDSVLLE 17933
INOFLS PSL CO PRITCHARD
^COSIER EN PIV-INO R£C PETERSBURG 17567
INO! ANAPOLIS P'.L CO PETERSBURG I7?f,7
BAILLY GENE RAT IMG STA R°3 PO* 216 1f301
KUSHVILLE STA PSI PC 311 RUSHVILLE 16173
ISM ELECT-BREED STA 3X568 SULLIVAN 17882
CSYUOA GEN STA PSI "3 188 CAYUCA 179?8
CPESSER STA FSI FO 359 TERPE HAUTE 17808
FUHLIC SERVICE WASfrSH STATION
SI GSE CULLEY STi W-:WRU"GH 17630 BX 218
ALCOA GENERATING CORP
RICHMOND POWER & LLGHT CO
P L N K 1 P
INT. POWER, 1000 «1AIN ST., OUFUQUE, 52001
CEDAR FALL? UTILITIES
It PUPLIC SERV CO f-AYNARC PLT WATERLOO
IOWA ELECTP.IC SOONE
IA PUF SERVICE CO HAWKEYE PLT STORM LAKE
IA PUP SERVICE CO CARROLL PLT CARROLL
CORN r
-------
1 ~* 1
1 .J I
1 38
110
111
112
113
111
H5
1*6
1*7
OFS
ne
119
150
151
152
i P;
151
155
156
157
PC.*'
159
160
161
162
163
164
165
1 66
167
1 68
1 6C
1 73
1 71
1 11
\ 7 *
\ 1"
1 7f
176
FLNMAO
MUSCAT I ME Pi.JR.700 1APLE G ".0 VE »I"U S C A T I NE
IA. POWER R LIGHT C0.» 339? SF 16. CES K
li POWER I LIGHT BOX 128 COUNCIL BLUFFS
IA-ILL GAS + ELEC RIVERSIDE PLT PETTNCRF
AI'ES MUNICIPAL ELECTRIC
IOWA STATE APES
CENTRAL IOWA POWER COOP
I«. SO. UTILITIES. PPIOGEPORT STA,EODYVILLF'
IOWA PUBLIC SERVICE co. SARGEANT FAUFF
I A. PUR. SERVICE, FAC-LiC GROVE PLTtEAGLE GRV
PLNHAO
F.KPIRF DISTRICT ELEC CO RIVrRTCN
KS POUERRLIC-HT LAWRENCE GENERATING STN
KANSAS GAS?ELECiME03HO PLANT
• KC- POV.'ER».LIGHT LACYC-NE
.K'S POUERS.LIGHT JEFFREY ENERGY CEN F.ELVUE
•KS POWERRLIGHT TECUMSEH GENERATING STM
COARD OF PUBLIC UTILITIES MEAFMAN CREEK
".CARD OF PU = LIC UT I L »QUI NO A» 0 V 3 66101
POARt) OF PUPLIC UTLiKAU STA 66101
PCARO OF PUBLIC UT!L»OUIMnsP,0 f»2 66101
P L N t\ A n
KY. UTILI TIFS-PINEVILLE P I NF V ILLE -F CUR
CINCINNATI GAS f ELECTRICCIMN OHIO 15201
KENTUCKY UTILITIES C3. GHENT
E.KY. RURAL ELECTRIC-DALE " TAtF STAT.'O
C.K.U. STATION NO. 1 15M EAST F
C.!-'.U. E L M. E R S M. I T H 4 3 n 1 M A R D I N
PIG RIVERS E L E C . - C C L E H A M r 0 L T K 1 1: - H i U
HENO. MIJNIC. POW « LIGHT SA!-T
LOUISVILLE GAS ?. E LECTRI CPAHOY S RUN
LCUISVILLE GAS > E LFC TR I CC »Nf RUN
LCUISVILLE GAS >. E LEC TR I C '•' I L L CREEK ?T*T
KY POVIFR-BIG SANDY 6 MI N LOUISA C
i v A - s H A '»' N E r PLANT *
E KY POWER COOP MAYSVILLE
KY.UT ILITIES-RROWN STA. P . 0 . c 0 X 255 BUR
KHJTUCKY UTILITIES C OM? AMCF ,N T P.A L CITY FA
TVA -PARADISE PLANT !i !•' I ME CPA '
E.KY. POWER COOP. -COOPER FU-NSir>E
PIG RIVERS ELECTRIC CORP. REID STATION
COUNTY
1 -7 /, n
.<: / *i U
'710
31 ?0
3110
32.10
3480
3180
3600
3 6 ft 0
4020
1060
COUNTY
440
86 0
19PO
21CO
3380
3R40
3840
3P40
3840
COUNTY
200
280
5PO
720
920
9?fl
1580
1 760
1920
• 1920
1920
2140
2460
2641
2740
2960
2960
3460
4020
STNAME
T Pi LI ^
1 U w -l
i on A
IOWA
IOU<
IOWA
IOWA
I out
ICWA
I n y A
IOWA
. STNAME
KANSAS
KANSAS
KANSAS
KANSAS
K A M S A S
K i M S A S
K 1 'i S A S
KANSAS
KANSAS
KANSAS
STtvA'T.
KENTUCKY
KENTUCKY
KENTUCKY
KENTUC/Y
KENTUCKY
KENTUCKY
KENTUCKY
KENTUCKY
KENTUCKY
KENTUCKY
KENTUCKY
KENTUCKY
KENTUCKY
KENTUCKY
KENTUCKY
KENTUCKY
KENTUCKY
KENTUCKY
KENTUCKY
PLANT
P n o ft
L U t L
0 0 5 !5
03PO
0115
oovc
0015
OOf'O
0005
0025
0190
0015
PLANT
0002
0011
C001
UOC5
0001
0030
0008 • ' '
com
004"
0050
P L A N T
0001
0029
0010
00 0.3
DOT: 6
00?7
0003
0012
0125
0126
0127
0303
0006
0009
0001
0001
0006
OOOr
000 1
POINTS
u
3
-,
i
A
v,
?
I
1
POINTS
1
. 1
I
a
2
1
2
^
1
F1 0 I M T S
1
1
>\
t!
1
2
1
2
5
6
4
T
10
2
J
1
3
2
5
-------
177
PLNMAD
KY. 'UTILITIES - TYRONE
STATE=18
BOX 18 VERS
COUNTY
1110
STNAfE
KENTUCKY
PLANT
0001
P 0 I M T S
1
oes
178
179
C •"• S
1 HO
1C!
1B2
183
IP."
IP.1?
1 ?<••
187
1 i! >\
189
190
191
1 92
193
191
1 9^
19f
197
153
1 5C
200
201
2C2
203
201
PLN1AD
CAJUN ELECTRIC POWER BIG CAJUN NO 2
CENTRAL LA FLEC PTF 1 LENA 71H7 (POYCE)
PLN1AP
Pfi + E liiGNFR PT «".ALLUO nf) 5 OF HWKN5! FT
PKPCO HOP.f.ANTOUM Nf.lMUSG ?Cf-f>1
PETO DICKER.C,ON STATION KICKEFSOn 2P7?3
DFFCO CHALK POINT PLANT AfMJA^.CO
FCTOMAC EOIT.ON U I LL H*"3C0° T ?17n?
PLN"AD
HOLY UMFR PCUTR.MT TOM PLANT . HOLYOKF
VrMERN ".ASS ELECT WST SPR1NCFIF.LD
PLNMAP
JOHN Iv'iRCEN STATION' 19916
D E KA=U! PLANT 255S IJ U'EADOCK HVY 1P732
J C WEADCCK PLANT P^S1^ !1 WEADOCK HWY
COM.SIU'EPS POU.ELM ST.PLT POX 279 1CP17
r.;n=.TKERN fllCHIGAN ELEC COOP 1050 E C. I V I S
LAIjSIMG BO OF UATEP & LTC-HT S72S SOUTH C
JOHN H WARDEN STATION 6]ft SK'LOrn A.WCNUE
HARPOR PEACH 7=^ NORTH HU"ON 1C411
MICHIGAN' 3U3AP. COMPANY TEFE'-JAING MI
LANSING F-OAF.D W'T il LGT ECKE-.T PLT 189C3
ECKfRT F. MOORES PARK STATION ISLAND AVEN
PRES^'UE ISLE STATIC*. PR'SQUE ISLE 0^
f-'Cf.'ROE cOklEP PLANT 3500 E FPOIV'T ST 16161
J R WHITING PLANT 15,75 ERIE C'CAC ^P157
F C COB? PLANT 151 NORTH CAUSEWAY 19110
J H CAI'PRELL PL'.'JT 16900 POLK STREET
HOLLAND BD OF PU3LIC yO"KS CITY HALL
COUNTY
2260
2360
COUNTY
nn
iin
11 ftp
1300
16F.O
COUNTY
369
179P.
1798
COUNTY
.MO
IOC
ICO
760
R60
1310
22 "0
• 2310
23"0
2360
2360
3280
3600
36 CO
3760
41f 0
1160
STNAME
LOUIS! AHA
LOUIS UNA
VAF.YL.AKD
MAr YLAI-.'C"
MARYLAf.'O
MARYL A«,0
KJSYL Af.'D
STNAME
"ASStCMUSETTS
KASSACI'USETTS
KASr.ACHIJSnTS
Klfh Id'-t!
MICHIGAN
IM C h I C A H
M I C H I C A N
MICHIGAN
MICHIGAN
MICHIGAN
KICHK-1K
MICHIGAN
MICHIGAN
WICHIGAM
". I C H I G A V
MICHIGAN
MICHIGAN
'.'. ! C H. I G A N
f, I C H I G A H
MICHIGAN
F L A N T
COOS
0010
PL f- N'T
&001
OOCi'i
rone
0007
oaio
PL ANT
0113
0010
0117
PLANT
00 01
OC11
001 ;?
0015
0007
0007
C 0 C 5
OOC1
OC02
COCE
0016
oai i
on,?3
002<-
0021
CG04
C016
POINTS
2
I
POINT '
I
2
5
•>
2
POINTS
1
1
3
POINTS
1
2
?
1
3
1
1
1
1
'-
t.
i
1
<,
f;
?.
:>
-------
c?s
205
2G£
207
?08
209
210
211
?12
213
211
0CS
215
216
217
218
=> ?19
n ?20
'21
222
223
P O A
225
22?.
?27
223
225
230
231
T 1 r<
?i~b
?3
\
C
2
T.
3
6
2
?
£.
2
1
1
2
1
1
£
b
T
1
3
5
3
1
1
t
POINTS
5
1
2
-------
245
- 216 -
247
249
250
2b2
253
25"
256
257
253
259
260
263
FLNMAO
COLUMBIA WATER & LIGHT/P.O. PCX N/COLUM6
UNIVERSITY OF-fO. POWER PLANT/MO C-E-NE-RA
ST. JOSEPH PSL-LAKE ROAD PLANT/520 FRANC
MO PUF SERV CO PLEASANT HILL BLP. 1
UNION ELEC.BX 119.LA3ADIE PLT.ST LOUIS
SPRINGFIELD UTILIT IES/J" MFS R I VER POWER
SPPINGFIELO UTILITIES/SOUTHWEST PLANT/SF
K. C. POWER £ LIGHT. MGNTROSE PLANT
KC^L. 115 GRAND AVE. K.C.
KCPL. 8700 HAWTHORN K.C.
fO.PUB.SERV.CO. SI3L£Y 10700 EAST 5P HWY
H-JOEP. P&L/21500 E. T*UMAN RD/INDEPENPENC
EMPIRE DIST ELECT CC. ASPUPY PLNT JOFLIN
UNION ELECTRIC, RUSH ISLAND
ASSOCIATED ELECTRIC POWER COOP..NEW KADR
CENTRAL ELECT PWR COOP, CHAMOIS
KANSAS CITY POWER P. LIGHT CO./ST. JCSEP.H
ASSOC ELE. COOP/THOMAS HILL PLNT/EX 158/
UMON ELECTRIC,SIOUX PLT, HWY 94,W.ALTON
UNION FLEC 8200 ^INE RD. MEPAXEC CLAMT
STATE=Z
PLNMAD
265
2&7
MCNTAMA-HAKOTA UTIL, HWY ?3, SIDNEY
KONTANA POWER CO COLSTRIP- UNIT ?1
MONT POWER(COPETTE IB 2536,FILLINGS ?9103
STATE=28
2f 8
?f 9
270
271
ELECTROSTATIC PR.EC IP I T A TOP f-1-76
NEPR PU"LIC FWR DIST SHEL03M STA
LINCOLN ELEC SYS LINCOLN STA
NE9R PUBLIC POWER JELLEVUE NE
?72 fiOHAVE REN STA LAUGHLIN NV P9046
273 NEVSOA POWER CO GARDNER STA
274 SIERRA PACIFIC FO BOX 10100 RENO 89510
COUNTY
380
3-C-O -
520
P.40
l&fiO
IPfcO
i R6.o
2P20
2210
2210
2210
2210
2260
2280
3300
3180
3740
3920
1160
1300
COUNTY
1320
136.0
1720
COUNTY
7PO
1520
1520
21P.O
COUNTY
80
no
2EO
STNAMF.
MISSOURI
M-ISSOU* i
MISSOURI
MISSOURI
M1SSOUR I
MISSOURI
MI S SO US I
MISSOURI
MISSOURI
MISSOURI
MISSOU* I
I-USSCUPI
MISSOURI
MISSOURI
PISSOUK I
MISS OUR I
MISSOURI
MISSOURI
MISSOURI
MISSOURI
S T N A ME
MONTANA
MONTANA
MONTANA
STNAML .
'•JEFRASKA
NEBRASKA
NEBRASKA
ME°RASKA
STNAME
NEVADA
ME V A DA
NEVADA
PLANT
0002
0 004
0004
0003
0003
0005
0039
0001
0021
0022
0031
00K0
OC01
OUlf:
0004
CC02
OC07
0001
OGO 1
OC1C
PLANT
0003
P001
0015
FL«NT
0002
OOC5
0007
0002
PLANT
0001
0006-
0007
POINTS
2
4
3
1
1
5
1
J
1
2
2
. 3
1
2
1
2
1
2
?
1
POINTS
1
1
°OIKTS
2
2
i,
CC'INTS
o
2
-------
CrS PLNM4D
275 PUB SEPV CO OF NH-M.E PR I MACK PL.POW 03501
COUNTY
STMAME PL AM
NEW HAMPSHIRE 0 0 2 (•
POINTS
o?s
276
?77
278
279
280
2H1
res
282
?£.'.
4s, 2 Si
CO
OPS
2e?
2f\f
?e7
2EP.
2HC
2°C
2 '31
2-5?
?CJ
?qi,
295
2?LUMM •
PUB SERV ELEC-HUCSON STA, JERSEY CY, 0^306
JFPSEY CENTRAL POWER *. LIGHT C-ILPEP.T ST
PUB SER ELEC LAMBERT RO TRENTON 01=608
P L N ; M A 0
PATON PUR SERV CO, RATON
ARIZ FU" SERV CO. FOUR CORNERS
PUBLIC. S^RV CO OF UK BOX 2"67 OLR S71C3
FL^KAD
GOUPEY ?TA NY ROUTE 201 JOHNSON
NIAGARA MOHAWK DUNKIRK STEA* S DUNKIRK
JAKTSTCWN FOV.TR ISfi STETLE ST JAME.STOW
JEf-iMSCN .r>TA RT B7 BtlMF.RID
C R HUMTLfY STE RIVER ROAD TOKtV'AND
'C-'E e3« = EEPED 2? AyE'lSWOOP 311 01
CON SOL ED CO CF NY INC AS TOM A 111P5
C3NCOL £0 CO 0^ \Y INC ARTHUR KILL 10311
N Y n i"CS Af.'D E RTE E COP NT N
MLLIKEN STA P.O si LUPLOWVI
r.pEEMOGE STA GREFNIDGE STa C-fT^CEN
COUNTY
en
7SO
1050
2210
22ftO
29 RO
COUNTY
260
1000
lore
COUNTY
610
1000
1 000
1CPO
2000
-------
CPS
-2.98 -
29°
300
301
302
3U3
301
305
3 Of:
307
309
309
310
r«s
311
M 2
TIT
Ml
315
3 16
317
z 1 P
319
320
721
3 r 2
323
3?1
3?f
326
327
3 2 fi
329
330
331
'< 3 2
3 3 3
:. T 1
335
336
337
STATE=J1
FLNMAD , COUNTY
CARCLINA ROWER t. LIGHT co SKYLANO
DUKE POWER CO MARSHALL PLT TERRELL
CF&L CAPF FEAR PLT BX 161 CONCURE
DUKE PCWER COMPANY ALLEN STA BELKONT
HUKE POWER CO RIVERPENO STA hT HOLLY
CAROLINA POWER & LIGHT CO L V SUTTO'J PLT
CFS.L ROXPORO PLT SR 1377 ROXRORO
CPRL UEATHERSPOON FLT .
OOC4 1
P L t M T POINTS
5001 1
50C1 1
5002 1
5C01 4
5001 1
5002 H
5 P 0 9 2
5COP 1
5001 6
5003 1
5001 <•
?OCf 6
5017 1
5 0 C 1 5
5002 2
5052 7
5008 1
5003 3
5002 2
-------
STiTE=36
OPS
PLNMAD
COUNTY
STNAHt'.
P L i-1! T
POINT!
33£ OHIO POWER COMPANY - TIDD PLANT 13913
^39 FAIMESVILLE MUNICIPAL ELECTRIC PL A 41077-
310 CLEVELAND ELEC ILLUtf CO FAST L'KE 11091
311 CLEVELAND FLEC ILLIJM CC AVON LAKF 11012
312 OHIO EDISON - EDGWATER 11052
343 TOLEPO EDISON CO, ACME STATION 13fcOb
.'11 TOLEDO EDISON CO, RAY SHORE STATIO 43616
245 CHIC t'OISON CO - N AVE PLT 11502.
3«f: CELIflA MUNCIPAL UTILITIES 15822
347 DAYTH PWI) .' LT CC F M TAIT FLE GEN 15139
34£ flAYTN POWER R LT CO HUTCHNGS ELC G 45312
319 COLS f. S OHIO ELfC CO PIC1JAY G^M S 13137
350 SUCfEYE SUGARS INC 1587r:
351 SHELPY MUNICIPAL LIGHT PLANT 11875
352 OHIO ECISON CO GORGE PLANT 41210
253 OHIC ECISON CO - MLES PLT 11116
354 COVER KUMICIPAL cCVf.n FLANT 11622
355 fusKi'iouf RIVER. PLANT 15715
256 UNION CARBIDE COPP. - METALS DIV. 15750
357 CRRVILLE MUNICIPAL UTILITIES 11667
o - "
C=!S PLNMAO
25" OKLA Gas ? ELEC PX 1119 PONCft C!fv 71601
259 OKLA. GASRELEC. POX 1270 KUrKOGEE7 41 0 1
'": r S F I M-- A 0
3'0 CUCUESNE LIGHT CO-2P19 WEST CAPSOf.'
2;-l <:t.'OUrr.NE LIC-HT CO-COLFAX C,T A -DUPUE SN E
3e-2 UfST PFNM POWER CC-nUTLFR ST FVTFNSION
.'.62 CUr.UfSNE LIGHT CO-JORDAN ST-CF E^CEMT 7v.ip
->,(.<< ft Pnyrp. C 0 - M E U CASTLE ZT", ^LT-WfST
^65 nufi(JFS'lr LIGHT CO CHE3WICK 15219
36fi UEST PFNN POWER P.O. P 0 " 4 flq 162C1
367 FEfjrJSYL VANI A ELEC1001 5ROAD STREET 15907
VF P-TMHSYLVa',1! s POWER CO'V.O. POX 12f 15(177
7'S KrTROPCLlTtfc EDISCN COP.O. c?v 54P l°f-03
270 c ff. f,;j YLV'N'I A ELFCTFIC Cn'']'JANY 166°3
'""I FHILADtLPHIA ELETU'P L I NT.iCR 0:'-E Y RD 194f-0
372 PE'.'NSYLVAMA ELECTRI1001 FROAD ST. 159C7
.'75 MET PO POL I TAN EDISON CO. DIKE ST. 17P57
374 PHILADELPHIA E L E » 1 IKTUSTRItL H«Y. 19P13
:'7r- F-:r.\SYLV-NlA ELECEAST FRONT T.TFEET 16507
7 K- «rsT PEM;I POVES CO''POOO c»eifi HILL i5f,ci
.'77 FEMrjSYLVAf-.'ia "LECTFIC COP.O. PPX K 15-911
3160
32. °0
3280
2610
3610
37?0
3720
3 P. 20
1210
1500
1500
5100
5660
5710
6500
6700
6720
7100
7100
7160
COUNTY
150C
2000
COUNTY
100
100
100
100
ion
100
26C
260
560
720
821
1660
1820
2340
2360
3080
3720
4213
OHIO
OH 10
OHIO
OHIO
OHIO
01-10
OHIO
OHIO
01' 10
OHIO
OHIO
OHIO
OHIO
OHIO
OHIO
OHIC
OHIO
OHIO
OHIO
OHIO
STMAME
0 K 1. i H 0 ." a
OKLAHOMA
STNtl'C
F'JNNSYLVAI'.1! A
PENNSYLVANIA
PE».'-'?YLV ANI A
PENNSYLVANIA
PENMSYL Vtf-M A
PENNSYLVANIA
PE'iNSYLVANI A
PENNSYLVANIA
PF NfiSYLVANIA
PENMSYLV4MI A
"ENNSYLVAMI i
DENNSYLVANI A
PENNSYLVANIA
PENNSYLVANIA
PENNSYLVANIA.
PENMSYLVANI a
^EMMSYLV ANI f
"ENNSYLVANIA
5302 3
5007
5012
5001 1
5 0 C 3 2
5017 6
5045 4
5011 1
5004 1
5027 ?
5041 A
5001
5002 1
5010 2
50? 7 1
5007 2
5001 1
5001 ' 5
50114 4
5C.06 ! 1
i •
P L i IJ T P C I U T S
OC03 1
0003 1
FLAMT PC HITS
0 C 3 i> -
n o 3 f- - 1
1C37 5
202 f. 6
0 C 2 c 5
d n 1 p 4
C001 2
OC04 2
0021 1
0 C 3 f: 2
C 0 0 3 1
P 0 1 f- 1
C n 1 P 3
0014 2
OC15 2
C 0 C '-i 4
0 C 0 1 2
0 C 0 t 2
-------
O'-'S
P L N M A C
COUNTY
STNAMF.
P L 4 N T
POINTS
- 778 PE NN-S-YL VAN I A- ELE C-T-R I-C- C P. 0-.- -BOX- 2-9-1-571.8
77<3 PrfifoSYLVANI A ELE'ST WHEATFIELO TVJf 15951
38C PENNSYLVANIA "OUER S LIGHT CO. 17532
381 UG I CCPPORATION ROUTE 11 186.?!
?-B2 FIRESTONE TIRE R RUBBER CO. PCX 699 19161
?PJ PENNSYLVANIA POWER
* 01 NTS
-
POINTS
^
3
2
5
3
1
7
2
2
p
2
2
POINTS
1
? "' * .
1
-------
CSS
408
. ___••. _._•._ ~ «_•_•«•«._ __~ CTATF — A£
FLNKAD .
UTAH POWER ?, LIGHT CO PO 899 SLC UTAH
L;T;H POWER • LIGHT PC fi°9 SLC
C;L PACIFIC UTILITY PO 5?c CECAR cnv
UTl^' POVIES R LIC-HT CO PC Rat? SLC 84110
l.n.'-H POV'FP. LIGHT CO PO B^'- JL C
PP.OVO CITY POME" 251 W 800 'lOfTH PROVC
UTAH PCWER f. LIGHT Cn PO (Ci1: SLC H1110
PLNU.AD
F£PCO»1400 N POYAL ST ALEXANDRIA
VE?CC OUTCH GAP RT h i
PLANT PC-IMTS
HA01 1
C C 0 5 1
0002 ?
f- 0 01 ' 1
H 0 0 1 ,1
P 0 0 1 1
E C1 0 1 •' 1
F- 0 0 1 ' ;'
P L t. K T P C I M T S
000? 2
oooi r
0002 1
COt 7 ?
CCf-E 2
001 P I
C 0 f P 1
F L. A N T POINT?
0003 5
C C 0 2 7.
6001 2
0 C 0 2 2
G&04 1
C P 0 c. :>.
OC03 ?
-------
0 B S P L N H A 0
138 . PACIFIC POW.EP.
fi?31
COUNTY
1100
WASH I NCI ON
F- L ' N 7 P 0 I N T S
K-Q 1 0 2
cn
GO
oes
459
410
441
44,?
443
444
44T-
<-46
"47
4< P.
6 <4 Q
"50
451
OFS
452
453
"54
455
456
457
458
459
460
"61
462
"63
464
465
466
46 7
"6 B
469
470
"71
47?
:. 1 „ i^r^i
P L N M A D
VA ELECTRIC R POWER MOUNT STOR1 26739
HARKISOM POWER STATION HAYWOOO
APPALACIAN POWER-KANAWHA R I V ER -GL A SGOW
CABIN CREEK POWER ST.N. OF HGY. 61
RIVESVILLE POWER ST. P.O. "OX 1342 RIVESVI
CMC PCH'ER - MITCHELL PLANT CRESAP
OHIO POUER - K*MKER VVA RT 2 IN C"ESAP
CENTRAL OPERATING PHILIP SPORN OFF US 3?
FORT MARTIN POWER. STATION ?HILES PT.f.AFI
l.'ILLCW ISLAND oGwrp ST. WILLOW IS.
ALPFIC-HT POWER STATION ALBRIGHT
APP PWR JOHN A*»OS FCPOX1000 ST HLBANS
r. I. DUPON'T WASH. WORKS WASHINGTON
t c
FLNMAD
LK SUPERIOR OIST POWER-BAY FRMT PLTf.ieOS
uis PUP SER COP.P JP PULLIAI" PLT
OAIRYLANP PWR CO-0? ALI'A WISCONSIN 54610
WIS FWR 1 LIGHT COLUMBIA 5EM STA FORTAP'
. MADISON GAS FLEC BI.OUNT ?T STATION 5??oi
WIS PWR LIGHT CASSVILLE
OAIRYLANO POWER CO-OP CASSVILLE 53P06
KAMITOUOC PU? UTIL 1203 SOTH ST 54220
WIS FUR SE" CORP ROTHSCH1LO 5"171
HEP-EAST WELLS STA IOB E UCLL? ST ???03
WISCONSIN ELFC PWR ELM RD OAK CRT.EK5?151
WISCONSIN ELF.C "WP. 1035 W CANAL 53"I?
I'ENiSHA ELEC 1 WATER UTIL RIVER ST 51 = 52
WIS ELEC PUR 11^ S WISCONSIN 53071
PKHLAND CN'TR MUNIC FLECT G AGE?-M A I N5358 1
WISCONSIN PWR LIGHT P.OCK RIVEP
WISCONSIN FWR • LIGHT 85? PLEASANT 53511
WISCONSIN PWR LIGHT HOX 356 SHEBOYGAN
DAIRYLAND DWP COOP GENOA WISCONSIN 54632
MENASHA ELECS.HATFK 183 MAIM ST 54952
MARSHFIELO E AND W 2000 S ROSSIS 54449
u
COUNTY
• 560
' 660
760
760
980
1020
10?0
. 1060
1140
1460
1520
1560
2220
COUNTY
120
360
400
600
6SO
1110
1110
1900
1920
2220
2220
?220
2580
260U
29PO
3C60
3060
328C
3620
1020
1060
STNAME.
WEST VIRGINIA
WEST VIRGINIA
WFST VIRGINIA
WEST VIRGINIA
WEST VIRGINIA
WEST VIRGINIA
WEST VIRGINIA
WEST VIRGINIA
WEST VIRGINIA
WrST VIRGINIA
WEST VIRGINIA
WEST VIRGINIA
WEST VIRGINIA
STNAKE
WISCONSIN
WISCONSIN
WI5JCONSI N
W I S C 0 N S I M
WISCONSIN
WISCONSIN'
WI SCOWS I N
WISCONSIN
WISCONSIN
WISCONSIN
WISCONSIN
WISCONSIN
WISCONSIN
WISCONSIN
WISCONSIN
WISCONSIN
WISCONSIN
Wl SCONS II N
WISCONSIN'
WISCONSIN
WISCONSIN
f L :. r T
0003
GC15
CO Of.
CCP7
0 00C
cons
0006
OOOl
0001
1 0 04
COO 1
0006
0001
PLA'IT
000.?
C 0 0 6
OOC1
f-OC1
CSC 2
0001
COO?
3001
00 OP
0006
C01 7
00.??
COOP
CU04
0001
ODD?
oooe
or 01
0 C C 1
OC02
CO 01
POINT?
J
7
2
t
2
2
3
5
2
2
3
3
7
FCIIiTS
3
f,
5
C:
i
2
2
.5
2
4
3
4
7
5
2
2
2
"
I
it
•y
-------
CBS
PLNHAO
COUNTY
STNAKE
PLANT
173 ?LACK HILLS PRL GARNER LAKE RT GILLETTE 80 WYOMING GOC?
171 FFf.C WYOFAK GARNER LAKE RT G ILLET T E8? 7 1 6 80 WYOMING 0001
175 PP&L DAVE JOHNSTON BOX S98 GLENROCK826 37 180 WYOMING 0001
176 UTAH POWER «NO LIGHT K^MIfRT^ 11D WYOMING G001
177 6ASIN ELECT MISSOURI BASIN ° WR . - R 0 JE C T 510 WYOMING COOi
17S JIP ORIDGER POWER PLANT POINT OF POCKS 700 WYOMING OOC7
(Ji
-------
JOJ7Z-I01
REPORT DOCUMENTATION
PAGE
1. REPORT NO.
560/5-83-005
3. Recipient's Accession No.
4. Title and Subtitle _ . .. . . .
Comprehensive Assessment of the Specific Compounds Present in
Combustion Processes, Volume 2. Design for a National Survey
of Emission of Specific Compounds from Coal Fired Utility Boiled
5. Report O*te
August 1985
Plants
7. Author(s)
Robert M. Lucas and Denise K. Melroy
8. Performing Organiiition Rept. No.
9. Performing Organization Name and Address
Midwest Research Institute with subcontract to
Research Triangle Institute
P.O. Box 12194
Research Triangle Park, NC 27709
10. Project/Task/Work Unit No.
1A
11. Contract(C) or Grant(G) No.
68-02-3938
(G)
12. Sponsoring Organization Name and Address
Field Studies Branch
USEPA
401 M Street, S.W.
Washington, DC 20460
13. Type of Report & Period Covered
Final
14.
IS. Supplementary Notes
J. J. Breen, Project Officer
D. T. Heggem;, Work Assignment Manager
16. Abstract (Limit: 200. words)
The emission of several toxic compounds in the polycyclic organic group has been re-
ported from stationary combustion processes. It has been demonstrated that a theoretical
potential exists for the formation of these compounds as the results of combustion of
coal-refuse, wood, municipal refuse, waste oil, and coal. To investigate this topic fur-
ther, a pilot study was designed to obtain data on which to base a national survey. The
overall objectives of the pilot study were to ascertain the number of combustion sites and
the number of days of sampling required at each site to adequately estimate the level and
prevalence of these toxic substances in the emissions from combustion processes and to do
so at a minimum cost.
For each facility a complex, multimedia sampling design was developed for the collec-
tion of solid, liquid, and gaseous influents and effluents. In addition, measurements of
process parameters were also taken. This design allowed for the estimation of the inputs
into the process, the efficiency of the combustion process, and the emissions from the
process. Using the estimates of the variability of the resulting data and cost estimates
based on the experience gained in the pilot, a national survey design was developed. Sam-
pling is planned for seven coal and nine refuse combustion facilities for 5 days each.
Estimates of the levels of toxic substances are anticipated to have a precision of ± 5 to
± 60%.
17. Document Analysts ,a. Descriptors
Combustion,
Survey design,
PAH, PCB, PCDD, PCDF
b. tdfpitifieri/Open-Cnded Terms
c. CO3ATI
. Availobil'ty Et.'.trrri-nt
Release to Public
19. r.ocu.ity Cl.ns (This Report)
Unclassified
j TO. Security Clasi (Tlii; f-jy.ft
j Unclassified
21. No. of Pages
61
:». ANM-L'iS.lG)
OP'JIO ,'.\L FCKM T.7i ;s-77)
(Fomivrly N1IS-3:,)
Pcpartiiicii! c', Cernmcice
------- |