United States
Environmental Protection
Agency
Office Of Air Quality
Planning And Standards
Research Triangle Park, NC 27711
EPA-454/R-00-0386 V/
September 2000
Air
cvEPA
Source Characterization For
Sewage Sludge Incinerators
Data Quality Assessment Report
Metropolitan Sewer District (MSD)
Mill Creek Wastewater Treatment Plant
Cincinnati. OhiO U-S. Environmental Protection Agerw*
Region 5, Library (PL-12J)
77 West Jackson Boulevard, 12to f toor
Chicago. II 60604-3590^
-------
EPA-454/R-00-038e
SOURCE CHARACTERIZATION FOR SEWAGE SLUDGE INCINERATORS
DATA QUALITY ASSESSMENT REPORT
METROPOLITAN SEWER DISTRICT (MSD)
MILL CREEK WASTERWATER TREATMENT PLANT
CINCINNATI, OHIO
3 Prepared for:
Clyde E.Riley(MD-19)
Emissions, Monitoring and Analysis Division
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711"
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Radiation
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
September 2000
-------
EXECUTIVE SUMMARY
The Clean Air Act Amendments of 1990 required the U.S. Environmental Protection
Agency's (EPA) Office of Air Quality Planning and Standards to establish standards of
performance for sewage sludge incineration. While some emission data exist for sewage sludge
incinerators, data on toxic polychlorinated biphenyls from sewage sludge incinerators are very
limited. The document Final Emissions Report: Source Characterization for Sewage Sludge
Incinerators summarized testing of a multiple hearth incinerator at the Metropolitan Sewer
District Mill Creek Wastewater Treatment Plant in Cincinnati, Ohio in July 1999.
This document presents the results of a Data Quality Assessment of the incinerator
emissions data (toxic polychlorinated biphenyls (PCB), dioxins/furans (D/F), polyaromatic
hydrocarbons (PAHs), as well as CO, CO2,02, and total hydrocarbons (THC) continuous
emissions monitoring data). Specifically considered was the question "Were data generated from
the fourth test run a statistical outlier?" Here, we concluded that Run 4 was a statistical outlier if
the Run 4 concentration of the measured analytes was significantly different from the average of
the Run 2 and 3 concentrations. Of the four test runs conducted, the first was aborted and the
fourth resulted in generally lower emission concentrations than the second and third runs.
Analysis of variance statistical models were utilized to test the hypothesis that the fourth
run was a statistical outlier. The results of the DQA were as follows:
For all tested parameters except CO2 and O2, the fourth run was a statistical outlier
(p < 0.05), and
For CO2 and O2, the fourth run was not a statistical outlier (p 2:0.05).
-------
TABLE OF CONTENTS
Page
EXECUTIVE SUMMARY i
1.0 INTRODUCTION 1
1.1 Summary of Test Program 1
1.2 Summary of the Data Quality Assessment (DQA) 1
1.3 Description of Report Sections . 2
2.0 PRELIMINARY DATA REVIEW 2
3.0 STATISTICAL METHODOLOGY 12
4.0 ANOVA MODEL RESULTS 18
APPENDIX - SAMPLE SIZE DETERMINATION FOR FUTURE DATA COLLECTION EFFORTS
LIST OF TABLES
Table 2-1. Percentage of Compounds for Which Run 4 was 1 to 4 Standard
Deviations Below the Run 2 and 3 Mean 4
Table 2-2. PCS Measurements Only: Percentage of Compounds for Which Run 4
was 1 to 4 Standard Deviations Below the Run 2 and 3 Distributions 6
Table 2-3. Spikes in CEM Measurements During Run 3 . . . ." 7
Table 4-1. ANOVA Model Results 19
LIST OF FIGURES
Figure 2-1. Boxplots of Standardized PCB, D/F, and PAH Measurements by Run 3
Figure 2-2. Boxplots of Standardized PCB Measurements (Uncorrected for
Pre-sampling Surrogate Recovery 5
Figure 2-3. Boxplots of Standardized PCB Measurements (Corrected for
Pre-sampling Surrogate Recovery 5
Figure 2-4. Continuous Emissions Monitoring: Carbon Monoxide 8
Figure 2-5. Continuous Emissions Monitoring: Carbon Dioxide . . 9
-------
LIST OF FIGURES
(Continued)
Page
Figure 2-6. Continuous Emissions Monitoring: Oxygen 10
Figure 2-7. Continuous Emissions Monitoring: Total Hydrocarbons 11
Figure 3-1. Boxplots of ANOVA Model Residuals from PCB, D/F, and PAH Analysis . . 14
Figure 3-2. Boxplots of ANOVA Model Residuals from Uncorrected PCB Analysis ... 15
Figure 3-3. Boxplots of ANOVA Model Residuals from Corrected PCB Analysis 15
Figure 3-4. Boxplots of ANOVA Model Residuals from Carbon Monoxide Analysis ... 16
Figure 3-5. Boxplots of ANOVA Model Residuals from Carbon Dioxide Analysis .... 16
Figure 3-6. Boxplots of ANOVA Model Residuals from Oxygen Analysis 17
Figure 3-7. Boxplots of ANOVA Model Residuals from THC Analysis 17
in
-------
1.0 INTRODUCTION
1.1 Summary of Test Program
The Clean Air Act Amendments of 1990 required the U.S. Environmental Protection
Agency's (EPA) Office of Air Quality Planning and Standards to establish standards of
performance for sewage sludge incineration. These standards are necessary to protect public
health and the environment from any adverse effects of pollutant emissions from sewage sludge
incineration. The regulations will contain general regulatory requirements, pollutant
characterization, and emission limits. To assess control technologies as well as associated
strategies for cost-effective standards, the EPA requires data on emissions from sewage sludge
incinerators. While some emission data exist for sewage sludge incinerators, data on toxic
polychlorinated biphenyls from sewage sludge incinerators are very limited.
The document Final Emissions Report: Source Characterization for Sewage Sludge
Incinerators summarized testing of a multiple hearth incinerator at the Metropolitan Sewer
District Mill Creek Wastewater Treatment Plant in Cincinnati, Ohio in July 1999.
1.2 Summary of the Data Quality Assessment (DQA)
The purpose of the test program was the measurement of mean concentrations of
pollutants in incinerator emissions, sewage sludge, and scrubber water. In the following, the
three test runs are referred to as Runs 2, 3, and 4 (the first run was aborted). As described in the
Emissions Test Report, the air emissions measurements during the final test run tended to be
lower than those for the previous two runs. This DQA report seeks to statistically answer the
question "Was Run 4 a statistical outlier?" by formally testing if the Run 4 air emissions
measurements were statistically significantly different from the Run 2 and 3 air emissions
measurements for the following compounds: toxic polychlorinated biphenyls (PCBs),
dioxins/rurans (D/F), and polyaromatic hydrocarbons (PAHs). A total of 165 compounds were
tested. Additionally, differences between runs for continuous emissions monitoring (CEM) data
(CO, CO2,02, and total hydrocarbons (THC)) were statistically tested.
-------
Note that there is no one definition of an outlier. Various methods may be devised to test
this question under diverse definitions. Here, we will conclude that Run 4 was a statistical
outlier if the Run 4 concentration of the measured analytes was significantly different from the
average of the Run 2 and 3 concentrations.
1.3 Description of Report Sections
Section 2.0 of this report provides a preliminary review of the air emissions
measurements from the three runs, including plots of the continuous emissions monitoring
measurements for carbon monoxide, carbon dioxide, oxygen, and total hydrocarbons. Section
3.0 describes the statistical methodology used to formally answer the question "Was Run 4 a
statistical outlier?" and checks the assumptions required for the statistical tests to be valid.
Section 4.0 presents the results of the data analysis. An Appendix provides supplementary
information regarding sample size requirements for future data collection efforts which use the
same data collection methods as the present test program.
2.0 PRELIMINARY DATA REVIEW
The data used to answer the question of whether or not Run 4 was a statistical outlier
consisted of stack air emission pollutants including PCBs, D/Fs and PAHs, as well as CEM data.
Raw data listings of the stack air emissions pollutant measurements were provided in the
Emissions Test Report. Measurements' for different compounds varied by orders of magnitude;
thus, summarization of the measurements across compounds required some form of
standardization. Within each compound, the mean and standard deviation of the measurements
were computed across Runs 2, 3 and 4. For each run, a standardized value was calculated by
subtracting the compound mean from the measurement and then dividing the result by the
compound standard deviation. If all the runs were similar, then the standardized values would
have been centered around zero. Figure 2-1 illustrates the distributions of the standardized run
measurements across 165 PCB, D/F, and PAH compounds.
-------
O 1
E
T3
N
'-o
l.
O
c
D — 1
V)
3
Run
Figure 2-1. Boxplots of Standardized PCB, D/F and PAH Measurements
by Run (see page 3 for description of boxes)
The boxplots indicate the following for each run:
• The locations of the 25th and 75th percentiles (lower and upper lines of the box).
• The median (middle line in the box).
• The most extreme values within 1.5 * the inter-quartile range1 of the box (the ends
of the solid lines extending from the box).
• Individual values that are most extreme (represented by asterisk stars).
• The horizontal dashed line at zero indicates where the median would lie if the
measurements from the three runs were all coming from the same underlying
distribution, that is, if there were no systematic differences between runs.
Another method used to descriptively summarize the difference between Run 4 and the
previous two runs was as follows. Within each compound, a mean and standard deviation were
computed from Runs 2 and 3 only, and the number of standard deviations that the Run 4
1 The inter-quartile range is defined as the difference between the 75th and 25th percentiles.
3
-------
measurements fell below this mean was calculated. Table 2-1 presents the frequency of
compounds for which the Run 4 measurement was 1, 2,3 or 4 standard deviations below the
mean of Runs 2 and 3. For almost every compound, the Run 4 measurement was at least one
standard deviation below the mean of the Run 2 and 3 measurements. For almost a third of the
compounds, the Run 4 measurement was at least four standard deviations below the Run 2 and 3
mean.
Table 2-1. Percentage of Compounds for Which Run 4 was 1 to 4
Standard Deviations Below the Run 2 and 3 Mean
1
2
3
4
98.18%
85.45%
49.09%
32.73%
For PCB measurements in particular, lower Run 4 emission concentrations may be
attributable to low pre-sampling surrogate recovery in Run 4. After correcting for this, the Run 4
PCB measurements were larger. Figures 2-2 and 2-3 illustrate the standardized PCB
measurements for each run, for the uncorrected and corrected measurements, respectively. The
uncorrected measurements (Figure 2-2) consistently decreased over the three runs. After
correction of the data (Figure 2-3), the run measurements are closer together, but still appear to
have a decreasing trend. Table 2-2 provides the percentage of Run 4 PCB measurements that are
from 1 to 4 standard deviations from the Run 2 and 3 distributions, for the uncorrected and the
corrected data. Again, while correcting for low pre-sampling surrogate recovery reduces the
differences between runs, they still seem to be different. This will be formally tested in Section
4.0 of this report.
-------
2f
1
£
o
o
•o
_N
TD
O
TO
o
-1
-2
Run
Figure 2-2. Boxplots of Standardized PCS Measurements (Uncorrected
for Pre-sampling Surrogate Recovery) (see page 3 for
description of boxes)
£
VI
o
-------
Table 2-2. PCB Measurements Only: Percentage of Compounds for
Which Run 4 Was 1 to 4 Standard Deviations Below the
Run 2 and 3 Distributions
100.00%
100.00%
76.92%
30.77%
84.62%
61.54%
30.77%
15.38%
The PCB, D/F, and PAH data were composited over each day. Continuous
measurements were also taken which provide additional insight into the differences among the
three runs. Figures 2-4 through 2-7 graph the concentrations of CO, CO2, O2 and THC,
respectively, during Runs 2, 3, and 4. As mentioned in the Emissions Test Report in Section 4.4
of the Final Emission Report, "at approximately 11:45 a.m. on Wednesday, July 21, 1999 (Run
3), the upper hearths (2, 3, 4) of incinerator 6 overheated. This caused premature burning and
generation of excess THC emissions. The instantaneous THC concentration hit apeak of 228
ppm. The operator reduced the rate of recycling of shaft cooling air to the lower hearths to
remove some heat from the incinerator. This procedure reestablished control within a few
minutes. This was the only process disruption of any sort during any of the test runs." Similar
spikes in CO, CO2, and O2 concentrations may be seen in Figures 2-4 through 2-6.
Table 2-3 provides some information regarding the size of the spikes around noon during
Run 3. For CO and CO2, the observed concentrations just after noon during Run 3 were 21% and
23% larger, respectively, than the Run 3 maximum morning measurement, and 63% and 19%
larger, respectively, than the Run 3 maximum afternoon measurement. For O2, the observed
concentration just after noon during Run 3, was 16% smaller than each of the Run 3 morning and
afternoon minimum measurements.
-------
Table 2-3. Spikes in CEM Measurements During Run 3
CO
949 to1714ppmdv
2074 ppmdv
898 to 1 275 ppmdv
C02
4. 18 to 6. 74% v
8.26% v
4.91 to 6.94% v
o,
11.28 to 15.03% v
9.49% v
11.25 to 13.6% v
For THC, the concentration levels following the overheating incident dropped off
dramatically (see Figure 2-7) from their Run 2 and pre-Noon Run 3 levels and remained lower
throughout Run 4. This result is consistent with the observed PCB, D/F, and PAH air
measurements, in which Run 2 tended to have the highest levels, Run 4 tended to have the lowest
levels, and Run 3 was in-between.
-------
Run = 2
21 OO-
19OO-
1700-
15OO-
9OO-
High 1607
Low 1248
Mean 1379
-i 1 1 1 1 1 —T r
8:OO 1O:OO 12:OO 14:OO 16:OO 18:OO 2O:OO 22:OO
Time (EDT)
Run = 3
High 2137
Low 918
Mean 1170
8:OO 1O:OO 12:OO 14:OO 1S:OO 18:OO 2O:OO 22:OO
Time (EDT)
Run = 4
21OO-
19OO-
17-OO
15OO
13OO^
900
High 1372
Low 1023
Mean 1126
8:OO 1O:OO 12:OO 14:OO 16:OO 18:OO 2O:OO 22:OO
Time (EDT)
Figure 2-4. Continuous Emissions Monitoring: Carbon Monoxide
8
-------
Run
CVJ
^L
se
10
9
8
7
6
5
A
High 5.70
Low 4.54
Mean 5.16
10
9
Q
7
6
5
A
1O
9
8
6-
5
s:oo 10:00 12:00 i4:oo 16:00 is:oo 20:00 22:00
Time (EDT)
Run — 3
High 8.26
Low . 4.18
Mean 5.50
s:oo 10:00 12:00 1-4:00 ie:oo is:oo 20:00 22:00
Time (EDT)
Run «- 4
High 5.56
Low 4.34
Mean 5.07
8:00 10:00 12:00 14:00 ie:oo ia:oo 20:00 22:00
Time (EDT)
Figure 2-5. Continuous Emissions Monitoring: Carbon Dioxide
9
-------
Run = 2
16-
15-
14
13
11
1O
High 14.6
Lew 12.9
Mean 13.7
s
16-
15-
14-
13-
12-
11-
10
9
8:00 10:00 12:00 14:00 16:00 18:00 20:00 22-.oo
Time (EDT)
Run = 3
High 15.0
Low 9.49
Mean 13.0
s:oo 10:00 12:00 14:00 16:00 18:00 20:00 22:00
Time (EDT)
Run = 4
High 14.6
Lew 12.1
Mean 13.4
-i 1 1 1 1 1 "—' T r
s:oo 10:00 12:00 14:00 ie:oo is:oo 20:00 22:00
Time (EDT)
Figure 2-6. Continuous Emissions Monitoring: Oxygen
10
-------
Run = 2
CJ>
so-
7O-
60
SO-
4O-
3O-
High 7B2 +.
Low 63.2 A A.
Mean 70.8 A.
^ A
A.
8:OO 1O:OO 12:OO 14:OO 16:OO 18:OO 2O:OO 22:OO
Time (EDT)
Run =i= 3
CJ3
8O-
7O-
6O-
50-
4O-
30-
High 73.3
Low 31.3
Mean 54.2
8:OO 1O:OO 12:OO 14:OO 16:OO 18:OO 2O:OO 22:OO
Time (EDT)
Run = 4
80-
ro-
6O-
5O-
4O-
30-
High 57.3
Low 30.6
Mean 37.5
8:OO 1O:OO 12:OO 14:OO 16:OO 18:OO 2O:OO 22:OO
Time (EDT)
Rgure 2-7. Continuous Emissions Monitoring: Total Hydrocarbons
11
-------
3.0 STATISTICAL METHODOLOGY
The determination of whether or not Run 4 was a statistical outlier was formally tested
using an analysis of variance (ANOVA) of the standardized stack air emission PCB, D/F and
PAH data, as well as of the CEM data. Additionally, analyses were performed using each of the
uncorrected and corrected PCB data as described in Section 2.0.
The ANOVA model used for the PCB, D/F and PAH analyses was
(1)
where z,j is the standardized measurement for the i01 run and the j* compound, a; is the mean for
the i'h run, and e^ is the error term. The error terms are assumed to be independent across
compounds and normally distributed within each run.
As mentioned in Section 1 .2, it was concluded that Run 4 was a statistical outlier
provided the mean concentration of the measured analytes was significantly different than the
average of the Run 2 and Run 3 mean concentrations. That is, the null hypothesis
H0: (a2 + aJ/2 = a4 was tested versus the alternative hypothesis H,: (a2 + aJ/2 * a4. According
to our definition of a statistical outlier, these hypotheses were equivalent to H0: Run 4 was not a
statistical outlier and H}: Run 4 was a statistical outlier.
Residuals from the model may be used to visually inspect the normality assumption for
the errors. Figures 3-1 to 3-3 present boxplots of the residuals from the fitted ANOVA models
for the PCB, D/F and PAH analysis and for the uncorrected and corrected PCB analyses. The
residuals appeared to be approximately normally distributed, despite the presence of a few
outliers. Formal tests of the normality hypothesis were conducted using the SAS UNIVARIATE
procedure. These tests supported the normality assumption for all three runs. For the corrected
PCB analysis, however, the formal test of normality did not support the normality assumption for
the Run 3 residuals. However, as shown in Figure 3-3, the departure from normality was not
drastic. Thus, the results presented in Section 4.0 for the corrected PCB analysis may still be
considered valid for the sample size evaluated. These ANOVA tests were performed using the
SAS GLM procedure.
12
-------
The model fit to the CEM data was identical to model (1) other than the fact that the error
terms were not assumed to be independent. Due to the high levels of correlation typically found
between time series observations close in time, the CEM data required a modified ANOVA
approach which accounted for the non-independence of the errors. First, for each of the CO, CO2
and O2 parameters, observations within the same run were assumed to be correlated.
Observations j minutes apart within the same run were assumed to have a correlation of p*. The
correlation structure was allowed to vary between runs with observations from different runs
assumed to be independent. For THC, this model was too complex for the few hourly
observations available during the 6-hour sampling period. A more simplified model was used in
which observations one hour apart were assumed to have a common correlation, again varying by
run. These ANOVA tests were carried out using the SAS MIXED procedure.
Formal tests for normality of the CEM analyses residuals (as shown in Figures 3-4 to
3-7) all support the hypothesis of normality, with the possible exception of the THC residuals
(With only six or seven THC observations per run, the test for normality is inconclusive.)
13
-------
o
o
<
I.OOOOOOOCH
0.00000000
-1.00000000
-2.00000000
3
Run
Figure 3-1. Boxplots of ANOVA Model Residuals from PCB, D/F and
PAH Analysis (see p.3 for description of boxes)
14
-------
0.15f
0.05
0
o
2
-0.5
-0.15
-0.25
3
Run
Figure 3-2. Boxplots of ANOVA Model Residuals from Uncorrected
PCB Analysis (see page 3 for description of boxes)
o
2
~v>
-------
1000
900
800
700
600
500
§ 400
TJ
"S 300
200
100
-100
-200
-300
i
*
*
i
1
Run
Figure 3-4. Boxplots of ANOVA Model Residuals from Carbon Monoxide
Analysis (see page 3 for description of boxes)
D
3
TJ
"w
x>
OC.
-\
-2
3
Run
Figure 3-5. Boxplots of ANOVA Model Residuals from Carbon Dioxide
Analysis (see page 3 for description of boxes)
16
-------
o
3
QC
-2
-3
-4
Run
Figure 3-6. Boxplots of ANOVA Model Residuals from Oxygen
Analysis (see page 3 for description of boxes)
30
20
10
o
3
TJ
'in
ID
OH
-10
-20
3
Run
Figure 3-7. Boxplots of ANOVA Model Residuals from THC Analysis
(see page 3 for description of boxes)
17
-------
4.0 ANOVA MODEL RESULTS
Table 4-1 presents estimated run means and tests of significance for the difference
between the average of the Run 2 and 3 means and the Run 4 mean. Results are given for the
analysis of all 165 PCB, D/F and PAH compounds, for the uncorrected and corrected PCB
compounds only, and for the CO, CO2,02 and THC measurements. For each analysis, if the p-
value was less than or equal to 0.05, then we may reject the null hypothesis H0: Run 4 was not a
statistical outlier in favor of the alternative hypothesis H0: Run 4 was a statistical outlier, at the
5% significance level. The interpretation is that there is only a 5% chance of incorrectly
concluding Run 4 was an outlier, when in fact it was not an outlier. In other words, we will
claim that Run 4 was an outlier only when strong evidence exists in favor of that conclusion.
The results of the DQA were as follows:
• For all but CO2 and O2, we conclude that Run 4 was a statistical outlier.
• For CO2 and O2, we conclude that Run 4 was not a statistical outlier.
18
-------
Table 4-1. ANOVA Model Results
PCB, D/F and PAH
Uncorrected PCB
Corrected PCB3
Carbon Monoxide
(ppmdv)
Carbon Dioxide
Oxygen
Total Hydrocarbons
(ppmdv)
<*2
«3
a*
a2
«3
a*
a2
a3
a4
<*2
«3 •
a.
a2
a3
«4
a2
«3
CC4
a2
«3
a.
0.8312(0.0325)
0.2138 (0.0325)
-1.0450 (0.0325)
0.8785 (0.0216)
0.2033 (0.0216)
-1.0818 (0.0216)
0.9445 (0.1314)
-0.3309 (0.1314)
-0.6136 (0.1314)
1382.46 (12.7632)
1166.56 (43.4856)
1132.57 (14.0177)
5.1852 (0.1574)
5.3495 (0.3063)
5.0319 (0.2170).
13.6666(0.2667)
13.1966 (0.4281)
13.2653 (0.6287)
71.1887 (2.5890)
49.4412 (9.5997)
34.0097 (1.2665)
< 0.0001
< 0.0001
< 0.0001
< 0.0001
0.3956
0.8062
< 0.0001
1 This Is the mean of the 165 standardized compound measurements for run i. The standardized
values are unitless.
2 A test of (a2 + a3)/2 - a4.
3 PCB analytical results were adjusted to account for spiked pre-field surrogate standard
recoveries for each of the 13 congeners analyzed (see Final Emissions Report - Table 6-7a).
19
-------
APPENDIX
SAMPLE SIZE DETERMINATION FOR FUTURE DATA COLLECTION EFFORTS
For each compound, computation of the relative standard error (RSE) of the mean,
defined as the standard error of the compound mean divided by the compound mean, was carried
out along with a study examining the reduction in RSE with an increase in the number of days of
sampling.
The conclusion of the DQA was that Run 4 was a statistical outlier under the definition of
a statistical outlier (See Section 1.2). While the Run 4 mean concentration was significantly
lower than the Run 2 and 3 mean concentration for all but CO2 and O2, this difference could be
interpreted as normal daily variability. Thus, calculation of the RSE, and its reduction with an
increase in the number of days of sampling, were carried out under two possible scenarios:
Scenario 1: Run 4 data represent actual temporal variability and must be incorporated
in order to unbiasedly estimate the true variability in pollutant
concentration means.
Scenario 2: Run 4 data are outlying and do not represent actual pollutant data. As
such, the variability of the pollutant concentration means should be
estimated based only on the data from Runs 2 and 3.
Due to the multiple CEM measurements taken within each run, the CEM observations
may not be assumed to be independent. The correlation between observations close in time is
accounted for through the use of the error structure described in Section 3.0. It is assumed that
future sampling days will have 360 minutely CO, CO2 or O2 measurements, or 6 hourly THC
measurements. The average of the run standard errors is the estimate of a for the RSE
calculation. In this setting, CT represents the standard deviation of the mean of a single run. The
estimate of u is the mean of the measurements.
Tables A-1 to A-10 provide the estimated RSE values for up to 10 runs. The shaded
column indicates the observed RSE for the present data. These values may be used as a guide to
future sample size needs assuming that the future study has the same objectives as the present
20
-------
study and that data are to be collected in the same manner. These RSE calculations are not
applicable to studies which collect data on the same compounds using different methods.
Table A-1. RSE of Toxic PCB Results Using Data from Runs 2, 3 and 4
3,3',4,4'-tetrachlorobiphenyI
(PCB-77)
2,3>3',4,4t-pentachlorobiphenyl
(PCB-105)
2,3,4,4',5-pentachlorobiphenyl
(PCB-114)
2,3',4,4',5-pentachlorobipnenyl
(PCB-118)
2',3,4,4',5-pentachlorobiphenyl
(PCB-123)
3,3',4,4',5-pentachlorobiphenyl
(PCB-126)
2,3,3',4,4',5-hexachlorobiphenyl
(PCB-156)
2,3,3',4,4l,5'-hexachlorobiphenyl
(PCB-157)
2,3',4,4',5,5'-hexachlorobiphenyl
(PCB-167)
19.8% 18.3%
3,3',4,4',5,5'-hexachlorobiphenyl
(PCB-169)
2,2',3,3',4,4',5-heptachlorobiphenyl
(PCB-170)
16.3% 15.3%
14.4% 13.7%
2,2',3,4,4',5,5'-heptachlorobiphenyl
(PCB-180)
2)3)3',4,4',5,5'-heptachlorobiphenyl
fPCB-1891
21
-------
Table A-2. RSE of Toxic PCB Results Using Data from Runs 2 and 3 Only (excluding
Run 4)
S.S'.M'-tetrachlorobiphenyl
(PCB-77)
2,3,3',4,4l-pentachlorobiphenyi
(PCB-105)
2,3,4,4' ,5-pentachlorobiphenyl
(PCB-114)
2,3',4,4',5-pentachIorobiphenyl
(PCB-118)
2',3,4,4',5-pentachlorobiphenyl
(PCB-123)
3,3',4,4',5-pentachlorobiphenyl
(PCB-126)
2,3,3',4,4',5-hexachlorobiphenyl
(PCB-156)
2,3,3',4,4'>5'-hexachIorobiphenyl
(PCB-157)
2,3',4,4',5,5'-hexachlorob]phenyl
(PCB-167)
S.S'A^.S.S'-hexachbrobiphenyl
(PCB-169)
2,2',3,3',4,4',5-heptachlorobipnenyl
(PCB-170)
2,2')3,4)4',5,5'-heptachlorobiphenyl
(PCB-180)
2,3,3',4,4',5,5'-heptachlorobiphenyl
(PCB-189)
22
-------
Table A-3. RSE of Dioxin Results Using Data from Runs 2, 3, and 4
23
-------
Table A-4. RSE of Dioxin Results Using Data from Runs 2 and 3 Only (excluding Run 4)
24
-------
Table A-5. RSE of Furan Results Using Data from Runs 2, 3, and 4
25
-------
Table A-6. RSE of Furan Results Using Data from Runs 2 and 3 Only (excluding Run 4)
26
-------
Table A-7. RSE of PAH Results Using Data from Runs 2, 3, and 4
27
-------
Table A-8. RSE of PAH Results Using Data from Runs 2 and 3 Only (excluding Run 4)
28
-------
Table A-9. RSE for CEMs Using Data from Runs 2, 3, and 4
Table A-10. RSE for CEMs Using Data from Runs 2 and 3 Only (excluding Run 4)
29
-------
TECHNICAL REPORT DATA
(Please read Instructions on reverse before completing)
1. REPORT NO.
EPA^54/R-00-038e
3. RECIPIENTS ACCESSION NO.
4. TITLE AND SUBTITLE
Source Characterization For Sewage Sludge Incinerators
Data Quality Assessment Report
Metropolitan Sewer District (MSD) Mill Creek Wasterwater Treatment Plant
Cincinnati, Ohio
5. REPORT DATE
September 2000
6. PERFORMING ORGANIZATION CODE
7 AUTHOR®
Clyde E. Riley, USEPA
Jeffery A. Ferg, Battelle
Anthony S. Wisbith, Battelle
Dennis A. Falgout, PES
8. PERFORMING ORGANIZATION R2PORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Battelle
505 King Avenue
Columbus, Ohio 43201-2693
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO. 68-D-99-009
12. SPONSORING AGENCY NAME AND ADDRESS
Emissions, Monitoring and Analysis Division
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final; January 99 to September 2000
14. SPONSORING AGENCY CODE
EPA/200/04
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The Clean Air Act Amendments of 1990 require the U.S. Environmental Protection Agency's
(EPA) Office of Air Quality Planning and Standards (OAQPS) to establish standards of performance for sewage sludge
incineration. These standards are necessary to protect public health and the environment from any adverse effects of pollutant
emissions from sewage sludge incineration. The regulations will contain general regulatory requirements, pollutant
characterization, and emission limits. To assess control technologies as well as associated strategies for cost-effective standards,
EPA requires data on PCB, D/F, and PAH emissions from sewage sludge incinerators. While some emission data exist for sewage
sludge incinerators, data on coplanar polychlorinated biphenyls (PCBs) from sewage sludge incinerators are very limited.
The test report summarizes testing of a multiple hearth incinerator at the Metropolitan Sewer
District (MSD) Mill Creek Wastewater Treatment Plant in Cincinnati, Ohio in July, 1999. The emission data collected in this test
program will be used by EPA/OAQPS and EPA's Office Of Water (OW) to support a decision about further data gathering efforts
in support of MACT standards for sewage sludge incinerators. During the testing, a second EPA contractor monitored and
recorded the process and emission control system operating parameters, and prepared Section 4.0, Process Description And
Operation of the report. The report consist of five documents: Executive Summary Report; Volume I-Main Report; Volume II-
Appendices A-J; Volume III-Appendices K-P; and a Data Quality Assessment Report.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS
Field/Group
c. COSATI
PCBs
PAHs
Dioxins/furans
Air Pollution control
18. DISTRIBUTION STATEMENT
Release Unlimited
19. SECURITY CLASS (Report)
Unclassified
PAGES
21. NO. OF
110
20. SECURITY CLASS (Page)
Unclassified
22. PRICE
EPA Form 2220-1 (Rey. 4-77) PREVIOUS EDITION IS OBSOLETE
------- |