United States Office of Air Quality EPA-450/3-86-012
Environmental Protection Planning and Standards October 1986
Agency Research Triangle Park NC 27711
Air
Primary Aluminum:
Statistical Analysis
of Potline
Fluoride Emissions
and Alternate
Sampling
Frequency
-------
•'r-
•- i
r
EPA-450/3-86-012
Primary Aluminum:
Statistical Analysis of Potline
Fluoride Emissions and Alternate
Sampling Frequency
Emission Standards and Engineering Division
U.S. Environmental Protection Agency
Region 5, Library (PL-12J)
77 West Jackson Boulevard, 12th Floor
Chicago, IL 60604-3590
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Radiation
Office of Air Quality Planning and Standards
Research Triangle Park, NC 27711
October 1986
-------
DISCLAIMER
This report has been reviewed by the Office of Air Quality Planning and Standards, U.S. Environmental Protection
Agency, and approved for publication as received from the Radian Corporation. Approval does not signify that the
contents necessarily reflect the views and policies of the U.S. Environmental Protection Agency, nor does mention of
trade names or commercial products constitute endorsement or recommendation for use. Copies of this report are
available from the National Technical Information Services, 5285 Port Royal Road, Springfield, Virginia 22161.
-------
TABLE OF CONTENTS
Section paqe
1.0 INTRODUCTION .......................... j.j
1.1 PURPOSE ............... ]_!
1.2 OVERVIEW OF TECHNICAL APPROACH ......... '. 1-2
1.3 REPORT OVERVIEW .................... .' ^3
2.0 SUMMARY ............................ 2-1
2.1 BACKGROUND INFORMATION .......... 2-1
2.2 OVERVIEW OF THE DATA ANALYSIS ......... '.'.'.'.'.'. 2-4
2.2.1 Approach ...................... 2-4
2.2.2 Data Analysis Results and Conclusions ....... 2-6
2.3 EMISSIONS MODEL .......... 2 8
2.4 PROBABILITY-OF-AN-EXCEEDANCE GRAPH AND ITS USE !!*'*' 2-9
2.5 CONTROL CHART MONITORING ...... o 12
2.6 REGULATORY PROCEDURES OVERVIEW ..... ! ........ 2-16
2.7 ADDITIONAL MONITORING TECHNIQUES ..... ....... 2-17
2.8 EMISSION MODELS NOT FITTING ALUMAX MODEL ..'.'.'.'.'.'.'. 2-18
3.0 DATA ANALYSIS ......................... 3_i
3.1 ROOF MONITOR DATA ........... 3.1
3.2 DRY SCRUBBER DATA ........... ......... 3 15
3.3 DATA DISTRIBUTION .......... ......... 3 J6
3.4 TIME SERIES ANALYSIS ....... ......... 3 25
3.5 COVARIANCE OF THE ROOF MONITOR AND DRY SCRUBBER.' .'.'*'* 3-25
3.6 CHOICE OF DATA BASE .................. ] [ 3_28
4.0 PROPOSED EMISSIONS MODEL .................... 4_j
4.1 EMISSIONS MODEL AND PROBABILITY-OF-AN-EXCEEDANCE RELATION 4-1
4.2 PROBABILITY-OF-AN-EXCEEDANCE GRAPH AND REQUIREMENTS FOR
ITS USE ........................ 4.3
5.0 CONTROL CHART MONITORING
5.1 CONTROL CHART THEORY ........... 5.1
5.2 CONTROL CHART FOR REGULATORY USE ..... .' ....... 5.5
5.3 ALUMAX CONTROL CHARTS ............ ..... 5.7
5.4 DERIVATION OF ALUMAX CONTROL CHARTS ..... '.'.'.'.''' 5-13
5.5 REGULATORY USE OF ALUMAX CONTROL CHARTS ........ '. '. 5-17
6.0 REGULATORY PROCEDURES ..................... 6_j
7.0 REFERENCES ........................... 7_}
i
-------
TABLE OF CONTENTS (continued)
Section
APPENDICES
A DATA TABLES A_!
B CONTROL CHART PARAMETERS B_j
C HYPOTHESIS TESTING c_j
D SUMMARY OF PRIMARY ALUMINUM PLANT PERFORMANCE D-l
ii
-------
LIST OF TABLES
2-1 Example Criteria For Reduced Performance Test Schedule ..... 2-11
3-1 Summary Statistics For Roof Monitors ............ 3.7
3-2 Summary Statistics For Each Roof Monitor ......... , . . 3.9
3-3 Roof Monitor Paired Comparisons ........... ... 3-14
3-4 Summary Statistics For Dry Scrubbers ........ '.'.'.'.'' 3-16
3-5 Summary Statistics For Each Dry Scrubber ........ . . . . 3-17
3-6 Kolmogorov D Test For Goodness of Fit ...... ....... 3-24
3-7 Variances and Covariances of Roof Monitor and Dry Scrubber
Emissions ........................... 3.27
4-1 Probability of An Exceedance, ALUMAX Potroom Group ....... 4-6
5-1 Emissions Data for ALUMAX Potroom Group 101G/161W, LB F/T Al 5-8
5-2 Emissions Data for ALUMAX Potroom Group 101H/161E, LB F/T Al '. ' 5-9
5-3 Emissions Data for ALUMAX Potroom Group 103G/162W, LB F/T Al* 5-10
5-4 Emissions Data for ALUMAX Potroom Group 103H/162E, LB F/T Al 5-11
5-5 Potroom Group Warning Limits, ALUMAX Data ........... 5-14
6-1 Example Criteria For Reduced Performance Test Schedule ..... 6-2
iii
-------
LIST OF FIGURES
Figure _
Page
2-1 Schematic Plant Layout 2 2
2-2 Probability of an Exceedance. ...'.'.] 2"in
2-3 X Control Chart For ALUMAX Potroom Group 1016/isiw 2~n
2-4 Sx Control Chart For ALUMAX Potroom Group 101G/161W '.'.'.'.'.'. 2-14
3-1 ALUMAX Monthly Average Emissions, Roof Monitor 101G 3.3
3-2 ALUMAX Monthly Average Emissions, Roof Monitor 101H ' 34
3-3 ALUMAX Monthly Average Emissions, Roof Monitor 103G ' 35
3-4 ALUMAX Monthly Average Emissions, Roof Monitor 103H . 3-5
3-5 Roof Monitor Annual Average, Potline 101. 3 in
3-6 Roof Monitor Annual Average of Monthly Standard Deviations!
Pot! me 101 3 ,,
3-7 Roof Monitor Annual Average, Potline 103'. . . 310
3-8 Roof Monitor Annual Average of Monthly Standard Deviation
Potline 103 ' 3_13
3-9a Normal Probability Plot of Roof Monitor Monthly Means '. 3.19
Detrended Normal Probability Plot of Roof Monitor Monthly Means 3-19
Normal Probability Plot of Roof Monitor Monthly Means 3 20
3-10b Detrended Normal Probability Plot of Roof Monitor Monthly Means 3-20
3-11 Monthly Mean Values of the Four Roof Monitors, Ib F/T Al 3 21
3-12 Logarithm of the Monthly Means of the Four Roof Monitors'.'.'.' 3-22
4-1 Probability of an Exceedance 4.4
5-1 Control Charts for X and s 5 2
5-2 Monthly Mean Emissions for ALUMAX Potroom Group ioiG/ieiW ' 5-4
5-3 X Control Chart for ALUMAX Potroom Group 101G/161W With Respect
Standards X and o 5 15
5-4 S Control Chart f8r ALUMAX Potroom Group ioiG/161W,'with ' ' '
Rispect to Standards X and o ... = 1fi
o o o-io
IV
-------
1.0 INTRODUCTION
1.1 PURPOSE
Radian Corporation provided engineering support services to the
U. S. Environmental Protection Agency (EPA) Emission Standards and
Engineering Division (ESED) in the evaluation of total fluoride emissions
data from the Mount Holly Plant of ALUMAX of South Carolina, a primary
aluminum reduction plant. Work was performed under EPA Contract
Number 68-02-3816, ESED Project No. 85/11. Emissions of fluorides from the
potroom groups at the ALUMAX plant are at levels well below the standard.
Performance data were evaluated to establish a basis for reducing the
frequency of required performance tests. The ALUMAX plant analysis may be
applicable to other primary aluminum plants in the industry performing well
below the standard. (See Summary of Primary Aluminum Plant Performance in
Appendix D). The ALUMAX study therefore was documented in this report in a -
fashion that facilitates similar studies potentially to be conducted by
State and local regulatory groups on other plants.
The evaluation of test data was limited to emissions from potroom
groups at prebake plants. The standard for primary aluminum reduction
plants (40 CFR Part 60, Subpart S) limits emissions of total fluorides (F)
from potroom groups at prebake plants to 1.9 Ib/ton of aluminum produced
(Ib F/T Al). The potroom group limit applies to the combined emission rates
from the primary control device, a dry scrubber installed on the exhaust gas
stream from the pots, and from the miscellaneous fugitive losses vented from
the potroom enclosure through roof monitors. At ALUMAX, monthly performance
tests are required on the roof monitors. The primary control devices were
exempted from the monthly testing requirements consistent with the
provisions of the standard [40 CFR 60.195(b)], and instead are subject to an
annual performance test schedule.
If a less frequent monitoring schedule for the roof monitor emissions
is to be established for a plant with performance well below the standard,
some assurance must exist that the probability of exceeding the standard '
under the less frequent testing schedule remains at an acceptable level. In
addition, quantitative criteria must be established as the basis for
decisions to approve less frequent test schedules. This report addresses
these concerns.
1-1
-------
1.2 OVERVIEW OF TECHNICAL APPROACH
The following tasks were undertaken to address the concerns that a less
frequent monitoring schedule pose:
• Task 1 - Analyze the ALUMAX emission data from dry scrubbers and
roof monitors to characterize emissions data so that a model could
be developed to predict emissions from the potroom groups and the
entire plant as a function of time.
• Task 2 - Prepare control charts based on ALUMAX data 1) to
establish objective criteria for measuring "performance," 2) to
detect emission performance changes, and 3) to monitor operating
performance in between performance tests.
• Task 3 - Using the emission-versus-time model developed in Task 1,
prepare a graph of the probability of an exceedance, P(Ex), as a
function of the overall mean emission rate, X, and standard
deviation of the mean, S-. The probability of an exceedance graph
may be used by a regulatory authority as the quantitative basis
for approving a less frequent performance test schedule, assuming
that the emission performance of the potroom group, or the plant
does not change.
• Task 4 - Prepare a report 1) summarizing the results of the study
of the ALUMAX data and 2) outlining how a regulatory agency might
qualify a primary aluminum potroom group for a reduced performance
test schedule, 3) how changes in overall emission levels could be
detected on a reduced performance test schedule, and 4) how the
regulatory agency might be assured that the emission levels do not
change between performance tests.
Tasks 1 and 3 provide the quantitative criteria for making the decision
to qualify a potroom group for a reduced performance test schedule. Task 2
offers the means to detect changes in overall emission levels with the use
of control charts using performance test results. The most serious problem
a regulatory agency faces in granting a reduced performance test schedule is
the uncertainty that exists that plant operating procedures affecting
1-2
-------
emissions may become lax in between required performance tests. Task 4
addresses this risk by suggesting the use of a control chart which monitors
plant performance continuously, whether or not the plant or potroom group is
on a reduced performance test schedule.
1.3 REPORT OVERVIEW
Section 2 summarizes the overall approach to characterizing the ALUMAX
emissions data and gives the results, recommendations, and conclusions of
the data analysis (Task 1) and the model development (Task 3). Section 2
also summarizes the control chart parameter results and summarizes the
recommendations on their use (Task 2).
Section 3 gives the details of the analysis of the ALUMAX data
(Task 1).
Section 4 contains the emission model development results, and gives
the procedure for constructing the Probability of Exceedance versus X and S-
chart (Task 3). X
Section 5 outlines the theoretical basis of control chart theory and
develops its application to monitoring the ALUMAX potroom groups (Task 4).
Section 6 overviews the procedure of how a primary aluminum potroom
group may be qualified for reduced frequency scheduling. It then outlines
how a regulatory agency may detect any overall change in emission levels,
and how assurance can be given that the potroom group emissions are not
changing in between performance tests.
Section 7 lists the references cited in this study.
1-3
-------
2.0 SUMMARY
This section summarizes the data analysis approach and results of the
study of fluorides emissions from the ALUMAX primary aluminum reduction
plant. To aid understanding of the analysis, the production process, points
of emission, and the data set are discussed briefly in the background
information, Section 2.1. Section 2.2 presents an overview of the data
analysis approach as well as some results and conclusions. Section 2.3
summarizes the results of the development of an emissions X versus time model.
Section 2.4 discusses the Probability-of-an-Exceedance graph derived with the
model and its use to qualify a potroom group for reduced frequency performance
tests. Section 2.5 summarizes the concepts and results of control chart
monitoring applied to a primary aluminum plant and Section 2.6 outlines
regulatory procedures as they apply to reduced performance test schedules in
primary aluminum plants. Section 2.7 discusses additional monitoring
techniques that give the regulatory authority additional assurance that the
potroom group emissions are not changing in between performance tests on a
reduced test schedule. Section 2.8 briefly summarizes the development of an
emissions model and regulatory procedures for a potroom group that does not
fit the ALUMAX model.
2.1 BACKGROUND INFORMATION
The ALUMAX plant contains four potroom groups, as illustrated
schematically in Figure 2-1. A potline is a long line of pots in which
aluminum is reduced from bauxite. Aluminum reduction in an individual pot is
a batch process. However, individual pots are charged on a regular,
alternating schedule and the potline as a whole is presumed to be a
continuous process from an emissions standpoint. The reasonableness of this
presumption is more apparent when the monthly period of the sampling
schedule is considered. Any model of emissions versus time is based on the
presumption that the emissions from the process can be considered continuous
and that periodic testing (e.g., monthly) provides a reasonable
approximation of emissions during the interval between tests.
2-1
-------
Hoof Monitor
Hoof Monitor 101H<
I
I
^ Pol
Dry Sorubbor
161W
Pot
\ /
I I
lino
I I
I I
I I
I I
I I
I I
lino 1
oa
1
I
D
•X
1
1
1
ry Serubbor
i
1
1
^
Potroom Group
Hoof Monitor 103O
Potroom Group
Moot Monitor 103M>
"»N Hoof Monitor 103H,
:?*-*—N. .< -=J-S- ..
i- ~*r
^ Potllno
J.
Dry tombbor
itaw
1
1
~f^
oa v-^
1
f
Potllno. 104
Or
I
I
Dry Sorubbor
ieai
I
Potroom Group
Potroom Group
Figure 2-1. Schematic Plant Layout.
2-2
-------
During the reduction process, the pots are enclosed and exhausted through
a manifold system to a dry scrubber primary control system. Exhaust gas leaks
and other fugitive emissions escape through a ventilation vent in the roof of
each building referred to as a roof monitor. Emissions testing is required on
the primary control device and on the roof monitors. A potroom group includes
those pots vented to a common primary control device. As illustrated in
Figure 2-1, the pots in a potline are segregated into two potroom groups. The
concentration of pollutants exiting the roof monitor for a potroom group is
determined at a single location in the roof vent. Flow or velocity is
determined at several locations and averaged to calculate the total loss from
the roof monitor.
Emissions data are reported by the plant for the roof monitor and primary
control device for each potroom. The summation of these emission rates is
compared to the 1.9 Ib F/T Al standard to determine compliance. For the ALUMAX
plant, data were reported for the following emission points and intervals
(emission unit numbers refer to Figure 2-1):
Potroom Group Roof Monitor 101G
Dry Scrubber 161W
Potroom Group Roof Monitor 101H
Dry Scrubber 161E
Potroom Group Roof Monitor 103G
Dry Scrubber 162W
Potroom Group Roof Monitor 103H
Dry Scrubber 162E
Consecutive Monthly
Observations
56
17
56
17
54
15
54
15
Annual
Observations
2
2
2
2
Both monthly and annual observations were submitted for the dry scrubbers
because about a third of the way through the 4-year period for which data
were submitted, the sampling frequency changed from one test per month to
one test per year.
2-3
-------
2.2 OVERVIEW OF THE DATA ANALYSIS
2.2.1 Approach
In order to develop an emissions model from the ALUMAX data, it was
necessary to characterize the data and to answer certain key questions, which
were:
• Should the roof monitor monthly emissions be averaged on a plantwide
basis, or should the roof monitor monthly mean emissions be
considered separately for each roof monitor?
• Should the dry scrubber monthly emissions be averaged on a plantwide
basis, or should the dry scrubber emissions be considered separately
for each dry scrubber?
• Should the potroom group monthly emissions be averaged on a
plantwide basis, or should the potroom group emissions be considered
separately for each potroom group?
Additional questions that remained were:
t What was the frequency distribution of the data—normal,
lognormal, or another distribution?
• Was there any relationship (i.e., correlation) between roof monitor
emissions and dry scrubber emissions in the same potroom group?
• Were the roof monitor monthly mean emissions data and dry scrubber
emissions data autocorrelated?
The purpose of the data analysis described below was to answer all the
above questions, or to provide the basis for making reasonable assumptions
that would be consistent with the ALUMAX emissions data base as a whole. It
was necessary to assume, for example, that the dry scrubber emissions data
were not autocorrelated because there was an insufficient number of
consecutive months' data to test each dry scrubber separately for
autocorrelation of the monthly mean emissions. This assumption was shown to
be consistent with the roof monitor monthly mean emissions data, which were
not autocorrelated.
2-4
-------
The data analysis may be outlined as follows (where appropriate, specific
references to the detailed discussion sections are provided):
• Monthly means were computed and verified for each roof monitor and
dry scrubber separately. (Section 3)
• Statistics were calculated for monthly means and the monthly
standard deviations (of the daily performance test readings for each
month) of the roof monitor emissions for all four roof monitor
combined--maximum, minimum, mean, standard deviation, range,
skewness and kurtosis. (Table 3-1)
t The same statistics were calculated for each roof monitor
separately. (Table 3-2)
• Statistics were calculated for monthly means and the monthly
standard deviations (of the daily performance test readings for each
month) of the dry scrubber emissions for all four dry scrubbers
combined—maximum, minimum, standard deviation, range, skewness and
kurtosis. (Table 3-4)
t The same statistics were calculated for each dry scrubber
separately. (Table 3-5)
• The means of the roof monitor and dry scrubbers monthly mean
emissions were tested for equality between roof monitors and between
dry scrubbers. (Tables A-4, A-5, and 3-3)
• A plot was made of cumulative normal probability distribution of the
monthly mean roof monitor emissions and their deviation from the
expected value to show graphically how well the combined roof
monitor emissions data fit a normal distribution. (Figures 3-9a and
3-9b)
• A plot was made of cumulative probability of the logarithm of the
monthly mean roof monitor emissions and their deviations from the
expected value to show graphically how the combined dry scrubber
emissions fit a lognormal distribution. (Figures 3-10a and 3-10b)
2-5
-------
The Kolmogorov D statistic was calculated for the combined roof
monitor monthly mean emissions data and for each roof monitor
separately so as to test the goodness-of-fit for the data compared
to the normal and to the lognormal probability distributions.
(Table 3-6)
The autocorrelation function and the partial autocorrelation
function were calculated for each roof monitor monthly mean
emissions data set. These were used to characterize the
autocorrelation structure of the roof monitor monthly mean
emissions. (Tables A-9 through A-12)
Covariance matrices were calculated for roof monitor and dry
scrubber monthly mean emissions in the same potroom group.
Correlation coefficients were calculated from each covariance matrix
to determine if a correlation exists between the emissions from the
roof monitor and the dry scrubber in the same potroom group.
Correlation coefficients were also calculated by regressing monthly
mean roof monitor data on monthly mean dry scrubber data. Plots of
roof monitor monthly mean emissions versus dry scrubber monthly mean
emissions gave visual interpretations of correlation. (Section 3.5)
2-2-2 Data Analysis Results and
Overall mean roof monitor emissions, X", were 0.809 Ib fluoride per ton of
aluminum (Ib F/T Al) with a mean standard deviation of monthly means, S-,
equal to 0.0174 Ib F/T Al . The dry scrubber overall mean emissions, X *'were
equal to 0.07443 Ib F/T Al with a mean standard deviation, S-, equal to
0.050 Ib F/T Al. There were sufficient differences between the overall means
for each roof monitor and dry scrubber to justify treating each dry scrubber
and roof monitor separately rather than using pooled values for the emissions
model. The monthly mean emissions for roof monitors ranged from 0.750 to
0.870 Ib F/T Al. The monthly mean emissions for dry scrubbers ranged from
0.046 to 0.101 Ib F/T Al .
2-6
-------
The graphical and statistical tests (Kolmogorov D statistics) showed that
the monthly mean roof monitor emissions fit a normal distribution better than
they fit a lognormal distribution. A normal distribution was therefore
selected as appropriate to characterize the probability distribution of the
ALUMAX monthly mean roof monitor data. The dry scrubber monthly mean emission
data showed no evidence in their skewness and kurtosis values to indicate that
their distribution was other than normal. It was assumed that the dry
scrubber mean monthly emissions data were normally distributed.
The autocorrelation functions and partial autocorrelation functions of
the roof monitor monthly mean emissions data indicated that there was
insignificant autocorrelation structure. This lead to the conclusion that the
roof monitor monthly mean emissions were independent (of time), random
variables.
The correlation coefficients showed that there was no significant
correlation between the roof monitor monthly mean emissions and the dry
scrubber monthly mean emissions in the same potroom group. This result
supports the assumption that the emissions variables within a potroom group
were independent of one another.
Because the monthly mean emissions data for each roof monitor were found
to follow a normal distribution, and because no autocorrelation behavior was
found, the same model could be used for each potroom group. The facts that
the monthly mean emissions from each roof monitor and each dry scrubber were
significantly different from one another suggest that the monthly mean
emissions should be calculated separately for each potroom group. Because
insignificant correlation was found between the roof monitor and dry scrubber
emissions within a potroom group, the roof monitor monthly mean emissions
(Xj), and the dry scrubber monthly mean emissions (X2) could be considered to
be independent of one another, which affects the form of the model for the
potroom group monthly mean emissions.
For the potroom groups, Section 4 postulates the total emission rate to
be the sum of the roof monitor emissions and the dry scrubber emissions. The
monthly variances for the potroom group are the sums of the variances of the
monthly roof monitor emissions and the dry scrubber emissions. For the ALUMAX
potroom groups, overall monthly means ranged from 0.8070 to 0.9618 Ib F/T AT.
Standard deviations ranged from 0.1340 to 0.2114 Ib F/T Al.
2-7
-------
The conclusions reached in analyzing the data were:
• The roof monitor monthly mean emissions were normally distributed,
random variables;
• There were significant differences between the monthly mean
emissions for the individual roof monitors;
t There were significant differences between the monthly mean
emissions for the individual dry scrubbers; and
• The monthly mean roof monitor emissions and the monthly mean dry
scrubber emissions in the same potroom group showed insignificant
correlation.
2.3 EMISSIONS MODEL
An emissions model was developed so that it could be possible to predict
for a potroom group the probability of an exceedance during normal operating
conditions. The probability of an exceedance is related to measurable
operating parameters, the overall monthly mean emission rate and the standard
deviation of the test measurements used to calculate the monthly means. Such
a Probability-of-an-Exceedance graph provides a regulatory authority with
quantitative criteria for selecting which potroom group qualifies for a
reduced frequent performance test schedule, and for choosing a performance
test frequency.
The lack of autocorrelation of the ALUMAX roof monitor emissions data,
and the determination that the data were normally distributed, simplified the
preparation of the Probability-of-an-Exceedance graph. The initial design
proposed for this study included the possibility that the monthly mean
emissions data might have been autocorrelated. If that had been the case, the
Probability-of-an-Exceedance graph would have had to be developed through
computer simulations. Because the emissions were not autocorrelated, and
2-8
-------
because the distribution followed a normal Gaussian curve, the probability of
an exceedance was calculated analytically from readily available tables for
the normal probability distribution.
The emissions model (developed in Section 4.0) for the ALUMAX data was
that of a normally distributed random variable with a known mean and variance
for each potroom group. In a potroom group, the total monthly mean emissions
are the sum of the dry scrubber emissions and the roof monitor emissions; the
variance of the monthly mean emissions is equal to the sum of the variances of
the dry scrubber emissions and the roof monitor emissions.
The emissions model developed for the ALUMAX potroom groups is given in
Equations 2-1 and 2-2.
(2-1)
Variables X and S- are the monthly means and standard deviations,
respectively, of emissions from the potroom groups. Variables subscripted 1,
refer to the roof monitor emissions while those subscripted 2 refer to the dry
scrubber emission rates.
2.4 PROBABILITY-OF-AN-EXCEEDANCE GRAPH AND ITS USE
The Probability-of-an-Exceedance graph is given in Figure 2-2. It was
developed based on the emissions model and statistics (see Section 4) to
provide a regulatory authority with a basis for qualifying a potroom group for
a reduced frequency sampling schedule. It applies to plants whose emissions
are normally distributed random variables (i.e., not autocorrelated) .
The plot shows the probability of exceedance, P (Ex), on the ordinate and
the monthly mean emission rate, X, on the absissca. Curves are given for
several values of the standard deviation of monthly mean emission rates, S-.
Horizontal lines designating one expected exceedance per year and one in ten
years also are shown on the figure. Once a potroom group has sufficient data
(48 consecutive months of monitoring data) to determine that their monthly
mean emission rates are not autocorrelated, and that their monthly mean
2-9
-------
.001.1
41
U
01
U
§
o
5K
•a
•8
M
ft.
.000001
.0000001
.0001
.00001
0.7
0.9 1.1 1.3 1.5 1.7
X, Monthly Mean Emissions, Ib F/T Al
Figure 2-2. Probability of an Exceedance.
2-10
-------
emissions are normally distributed, it can be located on the graph by its
overall monthly mean emission rate, X, and standard deviation, S-. Its
location on the Figure 2-2 determines the Probability-of-an-Exceedance for the
potroom group. An example shown in the figure is that of ALUMAX potroom group
103H/162E. Its probability of exceedance is about 0.000002 or about one
exceedance in over 10 thousand years. The other potroom groups at ALUMAX have
a lower probability of exceedance. This information, along with a table such
as that in Table 2-1 may be used by a regulatory agency to qualify a potroom
group for a reduced performance test schedule.
The regulatory authority must make the selection of which regions of the
graph qualify for various performance test intervals. Table 2-1 is given as
an example of the criteria that may be used. ALUMAX, potroom group 103H/162E
for instance would qualify for a 12-month monitoring schedule using these
criteria. It must be stressed, however, that the decision about monitoring
schedules are the purview of the regulating authority and must be made based
on their judgment. The Probability-of-an-Exceedance graph serves only as a
tool for distinguishing aluminum plant performance levels quantitatively .
TABLE 2-1. EXAMPLE CRITERIA FOR REDUCED PERFORMANCE TEST SCHEDULE
Example
Probability of Performance
an Exceedance, P(Ex) Test Schedule3
P(Ex) > 0.001 one/month
0.0001 < P(Ex) < 0.001b one/quarter
0.00001 < P(Ex) < 0.0001C one/6 months
P(Ex) < 0.00001 one/12 months
Selection of the performance test schedule is the purview of the regulatory
authority.
P = 0.001, signifies that the probability of an exceedance is 1 in 1,000
measurements on a monthly schedule, or once every 83.33 years because of
random variation.
CP = 0.0001, signifies that the probability of an exceedance is 1 in 10,000
measurements on a monthly schedule, or once every 833.33 years because of
random variation.
2-11
-------
2.5 CONTROL CHART MONITORING
Once a potroom group has qualified for reduced performance test
scheduling, the regulatory agency must be able to identify emission test
results higher than the overall average caused by random variation, from those
caused by changes in the overall emissions level. The control chart methods
proposed in this study are well suited to make that distinction.
Control charts monitoring performance test results are based on past
performance and ongoing results. Each performance test averages three daily
readings of Ib F/T Al for a monthly mean emission value, X; a standard
deviation of the three readings is also determined. Separate control charts
plot X and $x values for each performance test. For the ALUMAX potroom
groups, X and $x were found to be random variables. For each successive
performance test, random variation is exhibited, but knowing the overall
means, X and $x, it is possible to set limits on the expected variations,
assuming no changes occur in the overall potroom group average emissions or
their variability. Control charts can identify either warning limits or
control limits (or both). Warning limits are usually set at the mean value +
2-sigma, or two standard deviations above and below the mean. Control Limits
are generally set at the mean value + 3-sigma, or the three standard
deviations above and below the overall limits. Warning limits cover about 95%
of the monthly means and standard deviations; control limits cover about 99%
of the monthly means and standard deviations during normal random behavior of
the process.
Results of periodic performance tests, whether monthly, quarterly,
semi-annually, or annually, are located on their respective control chart
immediately following the test. If the points fall outside the control limits
of the control charts, the process, or operation is considered "out of
control", and management, operating personnel, and the regulatory agency are
signaled that something other than random variation is causing the emissions
to be larger (or smaller) than normally expected.
Figures 2-3 and 2-4 give an X-control chart and an $x control chart
respectively for ALUMAX potroom group containing Roof Monitor 101G and Dry
Scrubber 161W. Shown on the charts are horizontal lines for the overall
means, the central lines, and the upper and lower warning limits, UWL and LWL,
2-12
-------
ti-z
'MI91/9IOI dn(U9 uiocuiod XVWnitf -ioj 1^43
'£-3
-------
Figure 2-4. S Control Chart For ALUMAX Potroom Group 101G/161W.
2-14
-------
respectively. The latter limits are set at 2-sigmas (i.e., standard
deviations) above and below the central lines. The means + 2-sigmas for a
normally distributed variables should include, on the average, about
95 percent of the individual X or S readings. Only one datum in 20, on the
average, should fall outside UWL or LWL by random variation only. If there is
a change in operating procedure, or equipment failure, that affects emissions,
whether increased or decreased, it will be detected on the control charts by
one or more consecutive points falling outside the warning limits.
Figure 2-3 shows several monthly means above or below the warning
limits—January 1982, September 1982, February 1983, April 1983, June 1983,
August 1983, October 1984, and December 1984. Figure 2-2 shows five monthly
standard deviations above the upper warning limit, and none below the lower
warning limit. These are above the average expected number (which would be
about 3 for each graph) of "violations" of UWL and LWL, but the graphs show
both the X and S chart to recover quickly, i.e., there are not two
A
consecutive months for which the monthly means or standard deviations are
above or below UWL and LWL. The control charts demonstrate random behavior
and no overall increasing or decreasing trends. The X chart (Figure 2-3)
shows data from December 1981 through August 1982, nine consecutive months,
below the central line. Whenever more than seven consecutive points are
(1) increasing, (2) decreasing, (3) above the central line, or (4) below the
central line, a "run" has occurred and the process is said to be "out of
control." The system is seen to exhibit random behavior afterwards, so that
the early out-of-control condition may relate to plant startup causes. Aside
from the early excursion into an out-of-control condition in the X chart, both
Figures 2-3 and 2-4 show a process that is "in control."
Control charts for the other potroom groups at ALUMAX could be plotted
and would resemble those in Figures 2-3 and 2-4. The key parameters for
establishing these charts are the mean, standard deviation, the central line,
UWL, and LWL for emissions from each group. From these the warning and
control limits are constructed. Table 5-5 lists these key parameters for
possible use at ALUMAX.
An important feature of the use of control charts is that an
out-of-control situation generally does not mean that the NSPS emission level
(1.9 Ib F/T Al) has been exceeded. It indicates that the emissions have
2-15
-------
exceeded the range expected by random variation based on the past history of
the plant operating procedures and emission levels. An out-of-control
situation, or a situation in which the warning limits have been exceeded may
require recertification by the regulatory agency for a reduced performance
test schedule. Possible regulatory provisions such as this are summarized in
Section 2.6 below and are fully developed in Sections 5 and 6.
2.6 REGULATORY PROCEDURES OVERVIEW
The ALUMAX model may be used for any potroom group whose monthly mean
emissions are normally distributed, random variables. The latter implies that
the monthly mean emissions are not autocorrelated. Initial qualification of a
potroom group for a reduced performance test schedule requires a minimum of 48
consecutive months of performance testing to verify 1) that the monthly mean
emissions are not autocorrelated, and 2) they fit a normal Gaussian frequency
distribution. If both criteria are met, the regulatory authority may decide
on a performance test schedule based on their past performance. An example
set of quantitative criteria that might be used to qualify a potroom group for
reduced monitoring frequency is offered in Table 2-1. The probability of an
exceedance value for a potroom group may be determined in Figure 2-2 as a
function of X, the overall monthly mean emissions, and S-, the standard
deviation of the monthly mean emissions, determined from the 48 months of
testing.
During the period of the reduced performance test schedule, the potroom
group should maintain two control charts, X and S , for performance test
results. Initially, the control chart parameters are based on the data
gathered during the 48 months of operation needed to qualify the potroom group
for reduced performance test schedule. Central lines and upper and lower
warning limits are identified for each chart according to the methods
described in Section 5 and Appendix B.
Whenever the performance test gives results outside the warning limits on
any of the control charts, appropriate action must be taken by the plant
operating personnel. The action recommended is to look for any cause in the
plant that may have contributed to the result's lying outside the warning
limits, and to perform an additional performance test the next month. The
2-16
-------
purpose of these actions is for the regulatory agency (and the plant operating
personnel) to be able to make a judgement as to whether the violation of the
warning limit was caused by random variation or whether the overall plant
operating emission level has changed. Procedures are outlined in Sections 5
and 6 for making such a judgement with a known probability (usually 95 percent
probability) of making the correct judgement.
If the conclusion is reached based on a series of monthly performance
tests that the overall emission level of the potroom group has either
increased or decreased from the initial operating levels, the potroom group
must be recertified for the appropriate performance test schedule, using the
Probability-of-an-Exceedance graph Figure 2-2 and Table 2-1, and new control
chart parameters derived.
2.7 ADDITIONAL MONITORING TECHNIQUES
The inherent weakness in monitoring fluoride emissions with the
techniques outlined above is that the regulatory authority must assume that
there is no change in operating or maintenance procedures that affect emission
rates in between performance tests. If the potroom group is on an annual
performance test schedule, there is a risk that the overall emission level may
have changed, or even that the standard may have been exceeded for several
months without detection.
To help prevent such occurrences, it is suggested that maintenance
practices that reduce emissions be monitored on a daily or weekly basis.
Control chart theory as described for use with emission test data, may be used
with maintenance practices as well. At least one additional plant operating
parameter should be monitored with control charts on a continuing basis before
and after the potroom groups are given a reduced frequency performance test
schedule. The additional parameter can be one or more of the normal routine
operating practices that are carried out. For example, ALUMAX inspects each
potline weekly to determine how many panels on each pot are damaged or
improperly fitted and need replacing, or tightened in place. In addition, the
vent hoods over each pot are inspected daily to verify that they are correctly
in place, and they have not been left retracted too long after an anode has
?o
been replaced. Records are kept each week of the number of panels and the
2-17
-------
number of hoods that need adjustment or maintenance. The results of these
panel and hood inspection reports give indications of the operating efficiency
and performance of the plant operating personnel. The number of loose and
ill-fitting panels and hoods also relates to fluoride emissions levels.
Past history of hood and panel inspections at ALUMAX or another plant
could be used to develop X and $x charts for each potroom with their
appropriate warning or control limits. Violations of the limits on the hood
and panel inspection control charts could indicate a change in normal
operating and maintenance practices at the plant which could affect emissions.
For regulatory purposes, such additional control charts could be required to
be used weekly to monitor ongoing operating and maintenance performances. Any
violation of the control limits on these surrogate monitoring parameters could
trigger additional performance tests in between the regularly scheduled
performance tests to verify whether or not the potroom group is still in
control from the standpoint of their emissions level and its variability.
Such a regulatory requirement would give added assurance to the
regulatory authority that the potroom group emissions are not changing in
between regularly scheduled performance tests. The additional cost to the
plant would be much smaller than the savings gained on performance test cost
on a reduced frequency schedule.
2.8 EMISSION MODELS NOT FITTING ALUMAX MODEL
Recapitulating the essential features of the ALUMAX model are: 1) the
potroom group mean monthly emissions are normally distributed, 2) the potroom
group monthly mean emissions are random variables, and 3) the roof monitor and
dry scrubber emissions are independent of one another. Should a potroom
group's data not fit all of these three requirements, some adjustments to the
ALUMAX model should be made, before a new potroom group should be qualified
for reduced performance testing by the methods recommended in earlier sections
of this report.
The simplest adjustment to make would be if the roof monitor monthly mean
emissions and the dry scrubber emissions displayed covariance. This would
require adding a covariance term to Equation 2-2. The equation for variance
(standard deviation squared) for the potroom group would then read
2-18
-------
$x - $xl + $x2 + 2r Sxl Sx2' (2-3)
where r - the correlation coefficient of Xj and X2, and S-, is the standard
deviation of Xj and S-2 is the standard deviation of JL.
The last term in Equation 2-3 accounts for the covariance of the roof
monitor and dry scrubber monthly means.
One S- has been calculated for a new potroom group from Equation 2-3, it
may be used with X, the overall mean, Figure 2-1, and Table 2-1, by the
regulatory authority to qualify the potroom group for a reduced performance
test schedule as indicated above. No adjustment of the control chart
parameters is required from the control chart procedures detailed in Sections
5 and 6, and Appendix B because of the covariance of the roof monitor and dry
scrubber emissions.
If the potroom group emissions are not normally distributed, then a new
Probability-of-Exceedance graph is needed to replace Figure 2-2 (and 4-1). If
the emissions are lognormally distributed, tabulations are available for that
distribution. Other non-normal distributions may require a computer
simulation in order to calculate the probabilities for X and S-, so that a new
Probability-of-Exceedance graph can be constructed for qualifying the potroom
group.
Finally, if there is an autocorrelation structure to the data, i.e., the
potroom group monthly mean emissions are not random variables, then a new
Probability-of-an-Exceedance graph (Figure 2-2) will be required.
Autocorrelation complicates the model to the point that the best way to
calculate the probability of an exceedance for a given X" and S-, is through
computer simulation. No change would be required in the control chart
procedures outlined above or in Section 5 because of the autocorrelation
structure of the data.
As a result of the analysis of the ALUMAX data, it is not thought likely,
however, that monthly mean emissions from a potroom group in a primary
aluminum plant would not meet the requirements needed to use the ALUMAX model
developed in this study.
2-19
-------
3.0 DATA ANALYSIS
The analysis of the ALUMAX emission data is discussed in this section.
Section 3.1 examines the roof monitor data to define the overall means,
standard deviations and other statistics for each roof monitor separately
and for all four monitors combined. Annual monthly averages for each roof
monitor are plotted to discern trends and differences between monitors.
Section 3.2 examines the dry scrubber data in a similar manner, except that
after April 1982 only annual performance tests were reported for 1983 and
1984. The purpose of the analysis of the dry scrubber data was to determine
if there were significant differences in the operating characteristics
(means, standard deviations, etc.) between dry scrubbers.
Section 3.3 addresses the question of the choice of the data
distribution to be used in the model. It is necessary to determine whether
the emission data are normally distributed, lognormally distributed, or
follow some other frequency distribution. Because the roof monitor data
account for about 90 percent of the total emissions, the frequency
distribution of the roof monitor emissions will be used to develop the
model. There are insufficient dry scrubber data to test adequately various
frequency distributions for goodness-of-fit, therefore the dry scrubber
emissions will be assumed to be normally distributed.
Section 3.4 explores the time series character of the data, looking
particularly for autocorrelation behavior. Because th NSPS emission limit
for potroom groups in primary aluminum plants is based on the sum of the
emissions from roof monitors and dry scrubbers, the covariance of the
monthly means for these emission points was investigated. The covariance of
these two parameters is analyzed and reported in Section 3.5.
Finally in Section 3.6 the character of the emissions from the ALUMAX
plant is summarized and a choice is made of how the data are to be treated
in the proposed model.
3.1 ROOF MONITOR DATA
The data base used in this study was obtained from the Mount Holly
10 J
plant of ALUMAX of South Carolina. ' These data are provided in Tables A-l
3-1
-------
and A-2 in the Appendix. The roof monitor data are shown in Figures 3-1
through 3-4 for Roof Monitors 1016, 101H, 103G, and 103H, respectively. The
data appear to be similar in their average emission levels, variability, and
sequential behavior.
Table 3-1 gives summary statistics for all four of the ALUMAX roof
monitor data grouped together. The variable labeled "XBAR" is the monthly
mean value of three daily determinations (occasionally only two daily
readings were available for a specific month). Variable "SX" is the
standard deviation of the two or three daily emissions for a given month.
Both these emission variables carry units of pound(s) of fluoride emissions
per ton of aluminum produced.
Statistics shown in Table 3-1 are the number of cases, the minimum, the
maximum, the range, the mean, the standard deviation, the skewness, and the
kurtosis. Positive skewness measures a frequency distribution which
exhibits a "tail" to the right of the most frequent value which is longer
than the tail to the left of the most frequent value. A lognonnal
distribution for example would have a positive skewness, usually greater
than 2.0. An approximately normal distribution would have a skewness near
zero. Kurtosis, or the coefficient of kurtosis, measures the peakedness of
a distribution. For a normal distribution the kurtosis has a value of 3.0.
If the kurtosis exceeds 3, the distribution has longer tails than a normal
distribution with the same variance; if the kurtosis is less than three, the
distribution is more peaked than the normal distribution with the same
variance. For a lognormal distribution9 the kurtosis is a function of the
standard deviation^ , the kurtosis ranges from 0.0 to 6.235 X 1027 as a
ranges from 0 to 4.0. A lognormal distribution having a standard deviation,
equal 0.150 would have a kurtosis of 0.372; if a equals 1.0 the kurtosis
would equal 110.94.
While the skewness and kurtosis statistics are, in themselves,
insufficient to test the data for normality, or lognormality, they can give
such information as:
1. Does the set approximate normal (or lognormal) behavior?
2. Do two or more data samples appear to be approximately from the
same population?
3. Are the data skewed to the right or to the left?
3-2
-------
EMI55K+IS - LBS FLUORIDE/TON ALUMINUM
U3
C
-1
(D
to
00
I
u»
X
o
IP
-s
o»
ID
0>
o
3
XI
o
o
o
*—»
o
r,
D
t-
O
i_i i_i
:-i bi
_i i_
O
-—a
-a.
tT5a
C3--_
"-Q-
H
D
s m
D T]
m
m
1.1 '
-------
ALUMAX MONTHLY AVERAGE EMIS-: .jf--
1 J5 -
2
5 1 2 -
*?
3 i.--
H-
Ci
a
X O.3 -
U
u.
cn o a -
i
" O.T -
DT
cl
oa -
MQwrrot-? 1 01 H
S!
f
•
! , i
^Aj\^fUjuf |!"'
*" li 1 ' ^ j '
i ! "t,. f 1 L
! '^ i 4 a
i -
n
1 1 1 • — i , ,
Figure 3-2. ALUMAX Monthly Average Emissions, Roof Monitor 101H
3-4
-------
ALL'1 MAX MONTHLY AVERAGE EMI
1 0,3 a
0 1.1-
~v
— j -| _
•?
2
LU
Cl
3 °'S -
'._'
_t _
-o. '_' ,"P -<
..?
3
f'i 4=' .-
;
LJJ
•r o a -
L.T
"r O 4 •
01
n.3 -
[
i
i
I1
t
!
1
ca
t1
1
A.?
1 i J ! 17
\ ;> i S
I ' )j!-
i1 !
: : C
! i f
•fl 1 )'
i '• i
i
i
i
|
|
j
| [
f
i
it'
°
[
}
i 1
! 1
; i
, !
i, !'
i l>
ci'1
c
i
p
1
j
ill
III
ij
J
I
c
t
1 '
1
i
!
II
j
f
_ . Ij3^1 || j ci
ft ,i'i.i ' | i
i > IA . 1
.11 - i
I!' '• '
••» ; 4 . .-* '
: 1 7 ', . f 'i i
: ! ' ^ ^ i 1
1 : .' ', ,• ' ! i/1
13
r
j
if)
.2ij
4O
Figure 3-3. ALUMAX Monthly Average Emissions, Roof Monitor 103G
3-5
-------
M
nn
ALL'MA/ MONTHL.Y AVERAGE EMISSION"?
1 .4
ct O 3 -
0 3
°
f
f|
I Ou5H
'•lU
!
Figure 3-4. ALUMAX Monthly Average Emissions, Roof Monitor 103H
3-6
-------
TABLE 3-1. SUMMARY STATISTICS FOR ROOF MONITORS
XBAR
SX
N of Cases
Minimum
Maximum
Range
Mean
Standard Dev
Skewness
Kurtosis
220
0.33330
1.31330
0.98000
0.80938
0.17414
0.01942
0.21706
220
0.01530
0.52920
0.51390
0.14361
0.08298
1.03680
1.82004
3-7
-------
Table 3-2 shows statistics, computed for each roof monitor separately.
Roof monitor overall means range from 0.750 to 0.870 of pound fluoride per
ton of aluminum produced (Ib F/T AT) while the standard deviation mean values
range from 0.121 to 0.165 Ib F/T Al.
Further detail is given in Table A-3 in the Appendix, Section 6.0, in
which the ALUMAX roof monitor data are further subdivided by roof monitor by
year. These results are summarized in Figures 3-5 through 3-8. Figure 3-5
shows roof monitor annual averages for Roof Monitors 101G and 101H, both of
which are in the same building at the Mount Holly plant. The overall means
(from Table 3-2) are also shown on the figures. Note that the annual means
for RM101H are significantly larger than those for RM101G. The overall mean
for RM101H is about 14 percent higher than that of RM101G. The annual means
of the monthly standard deviations for RM101G and 101H are shown in
Figure 3-6. No apparent trend is clear, although the overall mean standard
deviation for RM101H is about 24 percent higher than the overall mean for
RM101G.
Figure 3-7 shows roof monitor annual averages for RM103G and RM103H.
Figure 3-8 shows the annual averages of the monthly standard deviations. The
overall means for RM103H are larger than the overall means for RM103G (again
about 14 percent for the overall monthly means, and 28 percent for the
standard deviations). The annual averages for RM103H are again all greater
than the annual averages for RM103G for both the emissions and the standard
deviations.
To determine if there was a significant difference between the means of
the roof monitors, a one way analysis of variance was performed with
monitors as treatments. The results are given in Table A-4 of the Appendix
and show that there is a significant difference between the mean values of
the monitors, at less than the 0.001 level of significance (i.e., >99.9%
probability). The overall means of the roof monitor emissions were tested
by the least significant difference method, or Isd method, to determine
which means differed from each other.3 The calculations are given in Table
A-5 in the Appendix and are summarized in Table 3-3. The table also shows a
schematic plant layout for the Mount Holly Plant of ALUMAX, showing the
relative locations of the roof monitors and dry scrubbers. The paired
difference table shows differences of the overall monitor means and
3-8
-------
TABLE 3-2. SUMMARY STATISTICS FOR EACH ROOF MONITOR
THE FOLLOWING RESULTS ARE FOR: RM$a = 101G THE FOLLOWING RESULTS ARE FOR: RMS = 101H
TOTAL OBSERVATIONS: 56 TOTAL OBSERVATIONS: 56
N of Cases
Minimum
Maximum
Range
Mean
Standard Dev
Skewness
Kurtosis
XBAR
56
0.43940
1.15330
0.71400
0.76001
0.14596
0.26243
0.01560
SX
56
0.01970
0.31190
0.29220
0.13281
0.07367
0.95984
0.20925
N of Cases
Minimum
Maximum
Range
Mean
Standard Dev
Skewness
Kurtosis
XBAR
56
0.56600
1.28500
0.71900
0.86961
0.86961
0.51151
0.60766'
SX
56
0.01990
0.32600
0.30610
0.16486
0.16486
0.40070
-0.71778
THE FOLLOWING RESULTS ARE FOR: RMS = 103G
TOTAL OBSERVATIONS: 54
THE FOLLOWING RESULTS ARE FOR: RMS = 10.?K
TOTAL OBSERVATIONS: 54
N of Cases
Mini mum
Maximum
Range
Mean
Standard Dev
Skewness
Kurtosis
XBAR
54
0.33330
1.13000
0.79670
0.74987
0.020989
0.02584
-0.60669
SX
54
0.01530
0.44030
0.42500
0.12129
0.08647
1.24572
2.02579
N of Cases
Mini mum
Maximum
Range
Mean
Standard Dev
Skewness
Kurtosis
XBAR
54
0.37000
1.31330
0.94330
0.85761
0.19465
-0.14520
0.47816
SX
54
0.03540
0.52920
0.49380
0.15511
0.08325
1.75156
6.06089
*RM$ = 101G is the designation for ALUMAX Roof Monitor 101G.
3-9
-------
Figure 3-5. Roof Monitor Annual Average, Potline 101
3-10
-------
Figure 3-6. Roof Monitor Annual Average of Monthly Standard Deviations,
Potline 101.
3-11
-------
MOMv rat
9.741?
Figure 3-7. Roof Monitor Annual Averages, Potline 103
3-12
-------
Figure 3-8.
Roof Monitor Annual Averages of Monthly Standard Deviations
Potline 103.
3-13
-------
TABLE 3-3. ROOF MONITOR PAIRED COMPARISONS
• oot Hooltor 101
Moot Monitor lOIMv
1
1
N^X' PA<
Dry >oruk»or
Pot
|
lint
lino
-_,
L_ ^_
oa
I
1
0
\
I
1
I
ry Serukhor
I
1
1
1
/
Potrooin Orono Potroom droup
Hoof Monitor 10»ON *«•« Monitor 103H
1
1
^ Pot
Dry ••rifekor
i«aw
Pot
I I
Ino 1
1
1
1
1
1
1
11*4
^F —
oa ^
I ,
Or
r
104
1I2I
Polrooa Or»«»
Polroom Or»«»
ROOF MONITOR PAIRED COMPARISONS
Paired Difference Table
101 G
101 H
103 G
103H
0.09760**
0.01200
0.10774**
103G
0.01014
0.11974**
101H
0.10960**
The Table is read as follows:
0.10774 is the difference between the mean emissions from Roof
Monitors 103H and 103G. The double asterisk indicates those
differences that are highly significant, i.e., the level of
significance is less than 0.01 (probability >0.99).
3-14
-------
indicates which differences are significantly different from zero and which
are not. The difference comparisons can be verified qualitatively by
reviewing the emission levels indicated in Figures 3-5 and 3-7.
The results of these analyses suggest that the roof monitors should be
treated separately in the modeling phase of this study rather than attempting
to pool the data of all four monitors.
3.2 DRY SCRUBBER DATA
Table 3-4 give summary statistics for all four dry scrubbers combined.
Note that the overall mean XBAR for the dry scrubber is less than one-tenth
that of the overall mean XBAR for the roof monitors. The mean standard
deviation of the dry scrubber emissions data is about one-fifth that of the
roof monitor data. The overall dry scrubber mean is 0.07443 Ib F/T Al
and the mean standard deviation is 0.02856 Ib F/T Al. The data do not appear
to exhibit a consistent or significant skewness, and the kurtosis is near that
of a normal distribution (= 3.0 for the normal distribution).
Table 3-5 gives statistics for each dry scrubber. The means range from
0.046 to 0.101 Ib F/T Al, and the standard deviations calculated from the
daily emissions used for the monthly means range from 0.015 to 0.034 Ib F/T
Al.
The paired difference table for dry scrubbers is shown below:
Paired Difference Table
162E
0.01319
0.05421**
0.03531*
162W
0.02212
0.01840
161E
0.04102*
161W
161E
162W
The Table is read as follows:
0.05421 is the difference in the overall mean emissions for dry
scrubbers 161E and 162E. The double asterisk indicates the difference
to be highly significant at the 0.01 level (probability >99%).
The single asterisk denotes a significance level at least 0.05
(probability >95% but <99%). No asterisk accompanying a difference
indicates the difference is not significantly different from zero.
3-15
-------
The analysis of variance which is reported in Table A-6 in the appendix
indicates that there is a highly significant difference (P-0.004) in the mean
emissions between dry scrubbers. Table A-7 gives calculations to show which
dry scrubbers differ.
3.3 DATA DISTRIBUTION
A key consideration in developing a model for predicting emissions from
potroom groups in primary aluminum plants is to determine the frequency
distribution of the data. ALUMAX has asserted that they believe the
emissions data to follow a lognormal distribution.
The analysis given below for the roof monitors indicates that the roof
monitor data follow a normal distribution rather than a lognormal
distribution. Much continuous emission data are lognormally distributed,
TABLE 3-4. SUMMARY STATISTICS FOR DRY SCRUBBERS
XBAR
N of Cases
Minimum
Maximum
Range
Mean
Standard Dev
Skewness
Kurtosis
72
0.01500
0.26270
0.24770
0.07443
0.05017
1.50491
2.36320
72
0.00150
0.18870
0.18720
0.02856
0.03559
2.24836
5.57309
3-16
-------
TABLE 3-5. SUMMARY STATISTICS FOR EACH DRY SCRUBBER
THE FOLLOWING RESULTS ARE FOR: DS$a = 161E
TOTAL OBSERVATIONS: 19
N of Cases
Minimum
Maximum
Range
Mean
Standard Dev
Skewness
Kurtosls
XBAR
19
0.01500
0.08200
0.06700
0.04634
0.02208
0.19063
-1.22424
SX
19
0.01500
0.06580
0.06430
0.01488
0.01843
2.19785
3.46346
THE FOLLOWING RESULTS ARE FOR: DS$= 161W
TOTAL OBSERVATIONS: 19
N of Cases
Minimum
Maximum
Range
Mean
Standard Oev
Skewness
Kurtosls
XBAR
19
0.01870
0.26270
0.24400
0.08736
0.06443
1.33261
1.17746
SX
19
0.00170
0.18870
0.18700
0.04161
0.04815
1.66006
2.51859
THE FOLLOWING RESULTS ARE FOR: DS$ = 162E
TOTAL OBSERVATIONS: 17
THE FOLLOWING RESULTS ARE FOR: DS$ - 1621V
TOTAL OBSERVATIONS: 17
N of Cases
Minimum
Maximum
Range
Mean
Standard Oev
Skewness
Kurtosls
XBAR
17
0.03800
0.21970
0.18170
0.10055
0.05059
0.72971
-0.15357
SX
17
0.00300
0.14240
0.13940
0.03396
0.03747
1.75293
2.40400
N of Cases
Minimum
Maximum
Range
Mean
Standard Oev
Skewness
Kurtosls
XBAR
17
0.02070
0.16100
0.14030
0.06524
0.03797
1.18821
0.59005
SX
17
0.00450
0.10580
0.10130
0.02385
0.02731
2.05867
3.25029
aDS$ * 161E 1s the designation for ALUMAX Dry Scrubber 161E.
3-17
-------
however the ALUMAX data are monthly averages of two or three consecutive days'
testing. The analyses given below indicate the roof monitor data to be
approximately normally distributed.
Figure 3-9a is a probability plot of all the roof monitor data combined.
The ordinate is the expected normal value of that observation for the standard
normal distribution, with a mean of 0 and variance of 1, and the abscissa is
the monthly mean, XBAR. If the data are from a normal distribution, the
plotted values will lie on a straight line. In Figure 3-9a, at an expected
value of zero the XBAR value is seen to be about 0.8. On a cumulative scale,
the zero on the ordinate of the diagram represents 50 percent of the distribu-
tion, which is the median. The upper plot therefore shows that approximately
50 percent of the XBAR values are less than 0.8 Ib F/T Al. Table 3-1 gave the
overall mean X" - 0.80938. For a normal distribution the mean and the median
are coincident and occur at the most frequent value, or for a cumulative
distribution they occur at the 50 percent probability level. Asterisks on the
plot indicate that one datum occurs at the point plotted while the numbers 2
through 9 indicate the number of data occurring at the indicated point.
An additional plot of the data which graphically tests the hypothesis of
normality is given in Figure 3-9b. The detrended normal probability plot
shows the differences between the expected normal values and the standardized
value of the observation. Note that the vast majority of the monthly mean
emissions data hover around zero (±0.2). If the data follow a normal proba-
bility distribution exactly, all the points would be plotted at "DEVIATION
FROM EXPECTED VALUE" equal to zero.
Figure 3-10a plots the roof monitor mean monthly emission data on the
same scale after having transformed XBAR to the logarithm (base e), LXBAR. If
the data followed a lognormal distribution, the curve would be linear. A
concave upward trend is visually apparent in Figure 3-10a. The deviation from
the expected value for LXBAR (Figure 3-10b) and shows much more curvature.
These graphical tools are evidence that the mean monthly roof monitoring data
are normally distributed.
Figures 3-11 and 3-12 show stem-and-leaf plots5 of XBAR and LXBAR,
respectively. The plots are essentially frequency histograms rotated 90°
clockwise from the usual frequency diagrams. The value of a stem-and-leaf
display is that actual numerical values (to two significant digits past the
3-18
-------
I- -1
Q.
X
•2
-3
3**
22*3*
543*
S5344
399
*976
699
679
2674
#28 2
**43
*2 **
0.2 0.4 0.6 0.8 1.0 1.2 1.4
XBAR
Figure 3-9a. Normal Probability Plot of Roof Monitor Monthly Means
Q.
X
0.4
0.2
0.0
-0.2
S-0.4
Q-O.6 •*•
-0.8
0.2
*
# * *
* *2*23*6
953*4*2
* 53994 353*3* **
* 3 *
#675 ***
2 *2 3 57959974
0.4
0.6
0.8
XBAR
1.0
1.2
1.4
Figure 3-9b. Detrended Normal Probability Plot of Roof Monitor Monthly Means
3-19
-------
3 +•
a.
x
-1
-
3 .+
— *•-
-1.5
*
*
* *
23
234
553
757
99
99
399*
>99
#279
»372
**33*
*2 **
**
-1.0
-0.5
LXBAR
0.0
0.5
Figure 3-10a. Normal Probability Plot of Roof Monitor Monthly Means
1.5 -t-
LU
1.0
a.
2 0.5
i
0.0
°-0.5
*2
**
* 3* *
2*** 34524
22223 959* *2 *
*43 3**9993
27999993
•1.5
Figure 3-10b.
-l.o
-0.5
LXBAR
0.0
0.
Detrended Normal Probability Plot of Room Monitor
Monthly Means
3-20
-------
STEM-AND-LEAF PLOT OF VARIABLE: XBAR
SMALLEST VALUE AT TOP OF PLOT IS: 0.333
3 37
***OUTSIDE VALUES***
4 12233
4 89
5 013334
5 556688999
6 000003334
6 566666778888899
7 H 0000000011112223333334444
7 5555667777777888999
8 M 000000001112222222223444444
8 5555666667777778888999999
9 H 000000011222334
9 55557777777889
10 000122334
10 558
11 00011223
11 58
***OUTSIDE VALUES***
12 89
13 1
Figure 3-11. Monthly Mean Values of the Four Roof Monitors, Ib F/T AT
3-21
-------
STEM-AND-LEAF PLOT OF VARIABLE: LXBAR
SMALLEST VALUE AT TOP OF PLOT IS: -1.099
-10 9
-9 9
-8 96632
***OUTSIDE VALUES***
-7 2
-6 985
-6 3221
-5 9976
-5 442111000
-4 9555
-4 32110000
-3 H 9887777655555
-3 4444333221111110
-2 M 99988887766666555
-2 44332222221111100
-1 9999999997777766666555
-1 H 4443333332222111110000
-0 9999988877665
-0 4443332222221
0 000011234
0 5589
1 00111224
1 6
2
2 557
Figure 3-12. Logarithm of the Monthly Means of the Four Roof Monitors
3-22
-------
decimal) are given in the display. The "stems" are the numbers in the
vertical column (3 through 13) representing XBAR from (0.3 to 1.3). The
leaves are the numbers on the right of the empty space next to the stems and
represent the actual XBAR values. For example in the first row at the top of
the plot "3 37" means that the smallest XBAR - 0.33 and the next larger
value of XBAR = 0.37.
The locations of the median (M) and the hinges (H) are indicated by the
M or H's in the area between the stem and leaves. The median splits the
sorted or ordered data in half, and the hinges split each half once more. M
is printed on the line which includes the actual value of the median, i.e.,
the median lies between 0.80 and 0.84. H is printed on the stems containing
the hinges. The "H-Spread" is the difference between the values of the two
hinges. The "inner fences" are defined as follows:
lower fence - lower hinge - (1.5 X H-Spread)
upper fence = upper hinge + (1.5 X H-Spread)
Any values outside the inner fences are printed on separate lines and
separated from the inner values by a line of text: ***OUTSIDE VALUES***.
Figure 3-11 shows the XBAR values to be approximately symmetrically
distributed about the median (or mean) value M. Figure 3-12 shows the
logarithm of XBAR (LXBAR) to be skewed (a longer tail...) to the top or the
small values of LXBAR. A lognormal distribution is skewed toward the larger
values of LXBAR which is opposite to the skew of Figure 3-12. These figures
provide additional evidence that the normal distribution represents the roof
monitor average monthly emission data better than the lognormal distribution.
Finally, XBAR and LXBAR were tested for goodness of fit using the
Kolmogorov D statistic. The results of the test are given in Table 3-6 for
the combined data and by roof monitor and indicate that for Roof Monitors
101G, 101H, and 103G, there is no significant difference between the normal
and lognormal distributions. For Roof Monitor 103H and the combined data,
however, the test indicates there is a significant difference between the
sample and hypothesized distribution for the lognormal distribution. The
computer printouts of the Kolmogorov D statistic tests are given in the
Appendix, Table A-8. Table 3-6 also shows that for the combined roof monitor
data the Kolmogorov D test suggests that the overall roof monitor emissions
3-23
-------
TABLE 3-6. KOLMOGOROV Da TEST FOR GOODNESS OF FITb
Roof Monitor XBARC
Combined Data n.s.
101G n.s.
101H n.s.
103G n.s.
103H n.s.
LXBARd
s.
n.s.
n.s.
n.s.
s.
D is calculated as the maximum absolute difference between the hypothesized
cumulative distribution and the sample cumulative distribution over all data
points.
n.s. - indicates the Kolmogorov D statistic is not significantly large to
warrant rejection of the null hypothesis, H at a 0.05 significance
level. °
s. - indicates D statistic to be significantly large; i.e., can reject
HQ at a 0.05 significance level.
CHQ: XBAR data fits a normal distribution.
HQ: LXBAR data fits a normal distribution.
3-24
-------
data fit a normal distribution but do not fit a lognormal distribution. The
conclusion drawn from the analyses given here is that the roof monitor data
fit a normal frequency distribution as well as or better than a lognormal
distribution.
3.4 TIME SERIES ANALYSIS
To investigate the autocorrelation structure of the roof monitor data,
the autocorrelation functions (ACF) and the partial autocorrelation (PACF)
4 7
functions ' were calculated and given in Tables A-9 through A-12 in the
Appendix. Tables A-10 and A-ll show that the emissions data exhibit no
significant autocorrelation structure, whereas Tables A-9 and A-12 indicate a
very weak autocorrelation structure--the first order autocorrelation
coefficients, p,, equal 0.333 and 0.282, for roof monitors 101G and 103H,
respectively. These values for the four roof monitors are so low as to
indicate that a random model would be appropriate, i.e., no autocorrelation
structure should be incorporated into the model. The autocorrelation
functions and partial autocorrelation functions were also calculated for the
monthly standard deviations. The plots of ACF and PACF for the monthly
standard deviations which are given in Tables A-13 through A-16 in the
Appendix show no autocorrelation whatsoever and suggest that the standard
deviations are independent of time.
As indicated earlier, there were an insufficient number of consecutive
readings of the dry scrubber emission data to do a meaningful time series
analysis. A minimum number of about 50 is recommended. It will therefore
be assumed that the dry scrubber emissions data are independent of time, and
normally distributed. The fact that the roof monitor data were found not be
autocorrelated adds credence to this assumption.
3.5 COVARIANCE OF THE ROOF MONITOR AND DRY SCRUBBER DATA
The analysis given above all point toward a random, time independent
model which will be discussed in Section 4 below. Because the primary
aluminum NSPS specifies 1.9 Ib F/T Al as the maximum emission rate of
fluoride from a potroom group, it is necessary to add the roof monitor
emissions to the dry scrubber emissions in order to obtain the emissions from
the potroom group.
3-25
-------
The sum of the monthly means of the roof monitor emissions and the
monthly means of the corresponding dry scrubbers will be defined as:
X - Xj + X2, (3.!)
where:
X = monthly mean emissions from a potroom group.
Xj = monthly mean emissions from the roof monitor of the potroom
group, and
X2 * monthly mean emissions from the dry scrubber of the potroom
group.
The estimate of the variance of the sum of the variables is:
S'x ' $2xl + S2x2 + 2 r Sxl Sx2 (3-2)
where,
2
S - = the estimate of the variance of x,
2
S -j - the estimate of the variance of x,,
2
S x2 * the estimate of tne variance of x-,
r = the correlation coefficient of Xj and JL, and
r S-j S-2 = the covariance between x, and x2.
The purpose of this section is to determine if there exists a
significant correlation between the monthly mean roof monitor emissions and
the monthly mean dry scrubber emissions. If there is insignificant
correlation, the covariance term in the calculation of the variance of the
total monthly emissions will be assumed equal to zero.
The calculations of the covariance matrices for roof monitor and dry
scrubber emissions in the same potroom group are given in Table A-21 of the
Appendix. The roof monitor variances, the dry scrubber variances, their
covariances, and the coefficients of correlation are listed in Table 3-7.
The data show that the covariance of the Roof Monitor 103H-Dry Scrubber 162E
3-26
-------
TABLE 3-7. VARIANCES AND COVARIANCES OF ROOM MONITOR AND DRY SCRUBBER EMISSIONS
Roof Monitor
Dry Scrubber
Pair
RM101G-DS161W
RM101H-DS161E
w RM103G-DS162W
i
™ RM103H-DS162E
ar = Sxl x2 /
Variance
of X2,
2
Sx2
0.004151
0.000487
0.001442
0.002560
Covariance
of X1 X2,
Sxlx2
0.000675
0.000143
0.000278
0.004384
Coefficient3
of
Correlation,
r
0.06777
0.03909
0.04677
0.47375b
Number
of
Observations
19
19
17
17
the value of r = 0.47375 is not significantly different from zero based on 15
degrees of freedom
-------
pair is apparently much larger than the covariance of the other roof
monitor-dry scrubber pairs. However, a statistical test of significance
indicates that the coefficient of correlation for the Roof Monitor 103H-Dry
Scrubber 162E pair is not significantly different from zero.
The determination that none of the correlation coefficients between the
roof monitor-dry scrubber pairs within a potroom group is significantly
different from zero supports the assumption, but does not guarantee, that
within a potroom group the roof monitor - dry scrubber emissions are
independent of one another.
3.6 CHOICE OF DATA BASE
To summarize the results of the studies given in Sections 3.1 through
3.5, the following conclusions may be drawn:
• There is a significant difference between means of monthly roof
monitor data.
• There is a significant difference between the means of monthly
standard deviations of the roof monitor data.
• There is a significant difference between the means of the dry
scrubber emissions.
t There is a significant difference between the monthly means and
standard deviations of potroom groups.
• The roof monitor monthly means are normally distributed.
• The roof monitor monthly means and monthly standard deviations are
independent of time.
• There is no significant correlation between the means of the
roof monitors and the dry scrubbers in the same potroom group.
In view of the results reported above it is recommended that the proposed
model simulate the potroom group emissions for each pair of roof monitors and
dry scrubbers separately.
3-28
-------
4.0 PROPOSED EMISSIONS MODEL
The data analysis results from Task 1 (see Section 3) indicate that the
probability of exceedance (of a 1.9 standard) for the Alumax plant can be
determined analytically as a function of X and S- for each potroom group.
^
An analytical determination of the probability of exceedance is appropriate
for the following reasons:
1. The monthly average emissions from the potroom groups exhibit
insignificant autocorrelation; and
2. The monthly average emissions from the potroom groups are normally
distributed.
4.1 EMISSIONS MODEL AND PROBABILITY-OF-AN-EXCEEDANCE RELATION
The probability graph is independent of the potroom group and depends
only on the data's not being autocorrelated and their having a normal
distribution. The emissions model for the ALUMAX data is developed below,
and the derivation and construction of the Probability-of-an-Exceedance
graph is explained. The use of the Probability-of-an-Exceedance graph in
regulatory procedures is explained in Sections 5 and 6.
The proposed statistical methodology for determining the probability of
exceedance vs. X and S- for each potroom group is based on the emission
characteristics derived for the ALUMAX data, as follows:
1. Assume monthly average emissions from roof monitors, denoted by
Xj, are normally distributed.
2. Assume monthly average emissions from the dry scrubbers, denoted
X2 are normally distributed.
3. Assume Xj and X^ are independently distributed. Therefore,
monthly mean emissions from a potroom group, X - X, + 5L, will be
normally distributed with mean,
X - Xj + X2 (4-1)
4-1
-------
and variance,
c2 -2 . C2
* x ~ 5 xl + b x2 (4-2)
where:
X = monthly mean emissions from a potroom group,
Xj = monthly mean emissions from the roof monitor(s) of the
potroom group,
X2 = monthly mean emissions from the dry scrubber(s) of the
potroom group,
X = overall mean of the monthly mean emissions from a potroom
group,
2
S - = estimate of the variance of the monthly mean emissions from
a potroom group,
2
S -j = estimate of the variance of the monthly roof monitor mean
emission over a period of time, and
S -2 = estimate of the variance of the monthly means emission of
the dry scrubber emissions over the same period of time as
SV
4. The probability (Pr) that a monthly average total emission value,
X, will exceed the standard of 1.9 Ib/ton aluminum can be
calculated as follows:
Pr (X > 1.9) » Pr
ax
Prf(Z) > 1=1^
where; ^
X - monthly mean emission value from a potroom group
u- » the true mean of the monthly mean emissions from a
potroom group
4-2
-------
a- = the true standard deviation of the monthly mean emissions
J\
from a potroom group
Z » the standard normal random variable
X = overall mean of monthly mean emissions from a potroom group;
an estimate of
S- = the standard deviation of the monthly mean emissions from a
potroom group; an estimate of a-.
A
The probability of Z greater than any value, can be determined from a
table for the cumulative normal distribution function.8 Probabilities may
be calculated for any value of X and S- with the method described above.
^
For example, suppose the overall mean, X of a hypothetical potroom
group Y is 1.3, with S- = 0.2. Then graphically the normal distribution can
be depicted as follows:
The probability of any monthly mean emission exceeding the 1.9 standard
would be calculated as follows:
Pr (X > 1.9) = Pr (Z > ((1.9-1.3J/0.2))
Pr (Z > 3) - 0.0013.
The probability value, 0.0013, is found in a table which provides the
probability of Z being greater than any value between 0 and 4.99. (Some
tables may provide probabilities of Z being less than any value between
-4.99 and 0. The probabilities are equivalent as a result of the
symmetrical properties of the normal distribution.)
4.2 PROBABILITY-OF-AN-EXCEEDANCE GRAPH AND REQUIREMENTS FOR ITS USE
After the probabilities have been calculated over an appropriate range
of X and S- values, they can be plotted as in Figure 4-1. The abscissa
represents the various X values, while the ordinate expresses the
probability of a random exceedance of the standard for any month. Each
curve on the graph represents the various S- values as they relate to the X
A
values and the probabilities of exceedance. In other words for a certain
pair of values, X and S-, 1) find X on the abscissa, 2) follow a straight
line up from the abscissa to where it intersects the appropriate curve for
the S- value and 3) follow a horizontal line from that point of intersection
to the ordinate to find the probability of an exceedance.
4-3
-------
0
1
•O
41
01
A
s
.0001.=
«w s=
•a
•8
Vi
p-
.00001
.000001
.0000001
0.7
0.9 1.1 1.3 1.5 1.7
X, Monthly Mean Emissions, Ib F/T Al
Figure 4-1. Probability of an Exceedance
4-4
-------
Two horizontal lines are drawn on Figure 4-1 and labeled "ONE
EXCEEDANCE IN 10 YEARS," and "ONE EXCEEDANCE IN ONE YEAR." The location of
these lines is calculated as follows. In 10 years there would be 120
chances for an exceedance based on a monthly sampling schedule. The
probability"of one exceedance in 10 years is 1 divided by 120, or 0.008333
which is the level for the line labeled "ONE EXCEEDANCE IN TEN YEARS." For
the line labeled "ONE EXCEEDANCE IN ONE YEAR," the probability is 1 divided
by 12, or 0.0833.
The X and S- values for the potroom groups of the AlUMAX plant are
A
presented in Table 4-1 along with the normal Z variate and probability of an
exceedance P(Ex).
The location of the potroom groups for the ALUMAX plant is well below
the probability of the ONE EXCEEDANCE IN 10 YEARS line. In fact, only
potroom group 103H/162E has a probability that can be found in the range of
most cumulative normal probability tables. Its probability of exceedance of
the standard for any month is 1.8 x 10" . Potroom Gr.oup 103H/162E is
plotted on Figure 4-1. The remaining three potroom groups have a
probability of exceedance less that 1X10 . This means that by random
variation alone, at the ALUMAX plant one might expect that there would be no
more than one exceedance per 100,000 years. It should be emphasized that
Figure 4-1 accounts for random variation only. Figure 4-1 is not valid if:
• There is equipment failure at the plant;
• The plant operating conditions and procedures change; or
t There is a change in the regular maintenance and cleanup routines.
The following requirements must be fulfilled before using Figure 4-1 to
provide probabilities of exceedance for a given potroom group's data:
1. A sufficient number of monthly observations must be available, no
less than three years to identify the distributional properties
and no less than four years to identify the existence of any auto
correlation structure. The analysis of the ALUMAX data has lead
to the belief that perhaps random monthly mean observations would
not exhibit significant autocorrelation structure for the same
type primary aluminum potroom group as the ALUMAX facility.
4-5
-------
TABLE 4-1. PROBABILITY OF AN EXCEEDANCE, ALUMAX POTROOM GROUP.
POTROOM
GROUP
101G/161W
101H/161E
103G/162W
103H/162E
X
0.8473
0.9162
0.8070
0.9618
— •
Sx
0.1520
0.1340
0.2114
0.2024
Z
6.926
7.342
5.170
4.635
P(Ex)
-------
5.0 CONTROL CHART MONITORING
The control chart is the method proposed to monitor plant operating
performance, especially during periods of reduced performance test
frequency, to determine changes in plant operations which may affect
emission levels. Control chart theory and its application to the ALUMAX
aluminum plant data are the subject of this section. Limitations of the
regulatory use of the performance test control charts will also be
discussed, and a possible method to overcome those limitations is proposed.
5.1 CONTROL CHART THEORY
Knowledge of the behavior of chance variations is the foundation on
which control-chart analysis rests. If a group of data is studied and its
variation conforms to a statistical pattern that might reasonably be
produced by chance causes, then it is assumed that no special assignable
causes are present. The conditions which produce this variation are said to
be "in control," or "under control." They are under control in the sense
that, if chance causes alone are at work, then the amount and character of
the variation may be predicted, and it is not possible to trace the
variation of a specific instance to a particular cause. On the other hand,
if the variations in the data do not conform to a pattern that might
reasonably be produced by chance causes, then is is concluded that one or
more assignable causes are at work. In this case, the conditions producing
the variation are said to be "out of control."
An example of control charts is given in Figure 5-1. Set numbers given
in the figure refer to sets of five machined product specimens each which
were milled in two groups, Group I and Group II. Two control charts are
shown, one for the average value, X, of a critical dimension on the
specimens for each set, and the other for their standard deviation, S.
Central lines for each chart are shown as well as the upper control limits,
UCL, and lower control limits, LCL. The upper and lower
5-1
-------
1
—Y-
"X
trA4&-
'_. j__i__a - 4~£ . f- i
/2 -
; ^gf Mi
_ r_i_i. - -t
Figure 5-1. Control Charts for % and s
5-2
-------
control limits are set at the 3-sigma limits above and below the central
line, and are based on the variability of the specimens in each set number
and on the number of specimens per set. If an average, X, or standard
deviation, S, falls outside the control limits, the milling process is out
of control. Sets 3 and 5 for Group I, and Sets 6, 8, and 9 are seen to be
outside the control limits, so the manufacturing process for Groups I and II
are both out of control. In addition to the fact that the points fall
outside the control limits on the X chart, the pattern of the averages for
each group of sets does not exhibit random variation. For each group the
averages begin high and then successively decrease to an out-of-control
condition and then rise again. This pattern is a second indication of an
out-of-control process. Examination of the S control chart for the same
data groups, gives no indication that the standard deviation (the
variability) is out-of-control. While the X control chart in this example
shows an out-of-control condition, it does not indicate the cause of the
condition. P.ersons familiar with the manufacturing process -- a milling
process in this case -- would have to examine the entire procedure to make a
judgement on an assignable cause for the condition.
An example of a control chart for a primary aluminum potroom group is
given in Figure 5-2 for the ALUMAX Potroom Group containing Roof Monitor
101G and Dry Scrubber 161W. Figure 5-2 is the X chart for monthly mean
emissions for the potroom group, plotted as a function of time. Each month,
three measurements are taken and the average of those measurements is
plotted on the figure. An S control chart, plotting the standard deviations
of the three measurements each month versus time may also be prepared, but
is not given here. The overall mean emission is plotted as the central line
at about 0.849 Ib F/T Al. Upper and lower control limits, UCL and LCL,
respectively are plotted at 1.140 and 0.554 Ib F/T Al. Figure 5-2 shows
out-of-control points to have occurred in January 1981 and August 1982. The
system recovered in each case by having the succeeding month's X fall within
the control limits. Comparing Figures 5-1 and 5-2 emphasizes the random
nature of the monthly mean emission which is clearly seen in the increasing
and decreasing values of X in succeeding months as seen in Figure 5-2.
There is one additional out-of-control criterion to be seen in Figure 5-2.
5-3
-------
Figure 5-2. Monthly Mean Emissions for ALUMAX Potroom Group 101G/161W.
5-4
-------
Nine consecutive points from December 1981 through August 1982 lie below the
central line. Whenever seven or more points lie above or below the central
line, an out-of-control condition exist for any control chart.
As indicated in the previous example, an out-of-control condition
generally indicates that there is an assignable cause, or an identifiable
reason, for the condition. No direct indication is given by the control
chart what the reason is for the process' being out-of-control. The plant
management and operating personnel will need to examine every step of the
day-to-day operation to discover the assignable cause. Notice that the
out-of-control condition in August 1983 indicated potroom group emissions
about 1.24 Ib F/T Al, still well below the emission limit of 1.9 Ib F/T AT.
Thus an out-of-control condition does not necessarily mean the potroom group
is exceeding the emission limit.
The type of control chart described above is useful for monitoring the
performance of a process to determine when the process is in control. For
purposes of regulatory monitoring a process, especially during periods of
reduced performance a second type of control chart, described below in
Section 5.2, is ideally suited.
5.2 CONTROL CHART FOR REGULATORY USE
As described in more detail in Section 6, when a potroom group is
petitioning for a reduced performance test schedule, it presents a history
of emission performance for a period of at least 48 consecutive months of
sampling. If the data analysis shows no autocorrelation, and the data are
normally distributed then an overall mean, X, and the standard deviation of
the monthly means, S-, and the standard deviation of the monthly means, So,
are calculated to locate the potroom groups on Figure 4-1. Depending on the
zone the potroom group is located in, the regulatory authority may qualify
the potroom group for a less-frequent-than-monthly performance test schedule.
On the less frequent performance test schedule, e.g. annual frequency,
the regulatory agency needs to be sure that if an overall change occurs in
the emission level of the potroom group, that change is detected as soon as
possible.
5-5
-------
The control charts best suited for regulatory use are one in which
there is control with respect to given standards, X and
-------
5.3 ALUMAX CONTROL CHARTS
Emissions data for potroom groups of the Mount Holly plant of ALUMAX of
South Carolina, provide the basis for control charts. These data are listed
in Tables 5-1 through 5-4. Potroom Group 101G/161W was designated as the
potroom group which contained Roof Monitor 101G and Dry Scrubber 161W. XBAR
values in the table record the monthly mean emissions. For roof monitors
the data were provided for each month from January 1981 (or March 1981)
through August 1985.
For the dry scrubbers, monthly readings were given through about
May 1985, and thereafter, only annually. For example, Table 5-1 shows
values of XBAR for Dry Scrubber 161W derived from measured values from
January 1981 through May 1982, then constant values of 0.0862 Ib F/T Al are
shown from June 1982 through August 1983. The value for September 1983 is
derived from measured emission values. From October 1983 through
September 1984, another constant value of 0.0889 is recorded. For
October 1984, the value 0.0603 is the measured value for 1984, followed by
the constant values of 0.0874 from November 1984 through August 1985.
Consistent with regulatory requirements, measured monthly mean dry scrubber
emission values were terminated after May 1982. Thereafter, only annual
measurements were taken -- September for 1983 and October for 1984. For
purposes of constructing the control charts for the potroom group emissions
for the months for which dry scrubber readings were not taken, the overall
dry scrubber average values up to that time were used.
Thus for the months from June 1982 through August 1983, 0.0862 was
assumed to be the emission from Dry Scrubber 161W and was the calculated
average emission for the data from January 1981 through May 1982. For the
period from October 1983 through September 1984, 0.0889 was assumed to be
the emission from Dry Scrubber 161W, and was the mean of dry scrubber
emission data from January 1981 through May 1982 plus September 1983.
Similarly, the value 0.0874 for the period from November 1984 through
August 1985 was the mean of the measured Dry Scrubber 161W values from
January 1981 through May 1982, plus September 1983 and October 1984.
5-7
-------
Roof Dry Potroom
Monitor Scrubber Group
101-G 161-W 101G/161W
Date
01/81
02/81
03/81
04/81
05/81
06/81
07/81
08/81
09/81
10/81
11/81
12/81
01/82
02/82
03/82
04/82
05/82
06/82
07/82
08/82
09/82
10/82
11/82
12/82
01/83
02/83
03/83
04/83
05/83
06/83
07/83
08/83
09/83
10/83
11/83
12/83
01/84
02/84
03/84
04/84
05/84
06/84
07/84
08/84
09/84
10/84
11/84
12/84
01/85
02/85
03/85
04/85
05/85
06/85
07/85
08/85
Average
XBAR
0.6053
0.7167
0.7370
0.8010
0.8473
1.0200
1.0100
0.8267
0.7267
0.8867
0.9060
0.7320
0.4393
0.5170
0.7700
0.7033
0.6867
0.7467
0.6333
0.7167
0.9700
0.7667
0.6467
0.8500
0.6833
0.4833
0.7000
0.5500
0.8200
1.0300
0.8467
1.1533
0.7967
0.9533
0.8433
0.6600
0.8000
0.7700
0.8267
0.8600
0.7000
0.6967
0.6700
0.9033
0.7333
0.9900
0.7767
0.5367
0.5967
0.7733
0.6100
0.5967
0.6367
0.7167
0.6650
0.9203
XBAR
0.0580
0.0200
0.0853
0.0187
0.1523
0.2627
0.0727
0.0737
0.0873
0.0553
0.0500
0.0410
0.0753
0.2033
0.0210
0.0517
0.1363
0.0862
0.0862
0.0862
0.0862
0.0862
0.0862
0.0862
0.0862
0.0862
0.0862
0.0862
0.0862
0.0862
0.0862
0.0862
0.1350
0.0889
0.0889
0.0889
0.0889
0.0889
0.0889
0.0889
0.0889
0.0889
0.0889
0.0889
0.0889
0.0603
0.0874
0.0874
0.0874
0.0874
0.0874
0.0874
0.0874
0.0874
0.0874
0.0874
XBAR
0.6633
0.7367
0.8223
0.8197
0.9996
1.2827
1.0827
0.9004
0.8140
0.9420
0.9560
0.7730
0.5146
0.7203
0.7910
0.7550
0.8230
0.8329
0.7195
0.8029
1.0562-
0.8529
0.7329
0.9362
0.7695
0.5695
0.7862
0.6362
0.9062
1.1162
0.9329
1.2395
0.9317
1.0422
0.9322
0.7489
0.8889
0.8589
0.9156
0.9489
0.7889
0.7856
0.7589
0.9922
0.8222
1.0503
0.8641
0.6241
0.6841
0.3607
0.6974
0.6841
0.7241
0.8041
0.7524
1.0077
0.8473
Roof Dry Potroom Potroom
Monitor Scrubber Group Group
101-G 161-W 101G/161W 101G/161W
STO DEV STD DEV VARIANCE
0.1924
0.0646
0.0545
0.2296
0.1562
0.1453
0.0872
0.0814
0.0902
0.0907
0.1975
0.2533
0.1513
0.0197
0.2425
0.0945
0.0651
0.1553
0.1692
0.1861
0.1114
0.1155
0.1001
0.1442
0.1193
0.0929
0.1054
0.1345
0.0700
0.2706
0.0451
0.1234
0.1662
0.3027
0.1234
0.3119
0.0964
0.1082
0.1069
0.1114'
0.1127
0.0751
0.1480
0.3073
0.1266
0.0624
0.0351
0.0551
0.1102
0.2991
0.0608
0.1041
0.2219
0.1595
0.0358
0.0676
0.0485 0.03937001
0.0017 0.00417605
0.1037 0.01372394
0.0032 0.05272640
0.1887 0.06000613
0.0931 0.02977970
0.0337 0.00873953
0.0064 0.00666692
0.0186 0.00848200
0.0274 0.00897725
0.0107 0.03912074
0.0053 0.06418898
0.0192 0.02326033
0.0604 0.00403625
0.0026 0.05881301
0.0110 0.00905125
0.0667 0.00868690
0.0634 0.02813765
0.0634 0.03264820
0.0634 0.03865277
0.0634 0.01642952
0.0634 0.01735981
0.0634 0.01403957
0.0634 0.02481320
0.0634 0.01825205
0.0634 0.01264997
0.0634 0.01512872
0.0634 0.02210981
0.0634 0.00891956
0.0634 0.07724392
0.0634 0.00605357
0.0634 0.01924712
0.0780 0.03370644
0.0643 0.09576178
0.0643 0.01936205
0.0643 0.10141610
0.0643 0.01342745
0.0643 0.01584173
0.0643 0.01556210
0.0643 0.01654445
0.0643 0.01683578
0.0643 0.00977450
0.0643 0.02603849
0.0643 0.09856778
0.0643 0.02016205
0.0117 0.00403065
0.0627 0.00516330
0.0627 0.00696730
0.0627 0.01607533
0.0627 0.09339210
0.0627 0.00762793
0.0627 0.01476810
0.0627 0.05317090
0.0627 0.02937154
0.0627 0.00521293
0.0627 0.00850105
STD DEV
0.19841
0.06462
0.11713
0.22962
0.24495
0.17256
0.09349
0.08165
0.09209
0.09475
0.19778
0.25334
0.15251
0.06353
0.24251
0.09514
0.09320
0.16774
0.18068
0.19659
0.12816
0.13174
0.11849
0.15752
0.13510
0.11246
0.12299
0.14868
0.09443
0.27792
0.07780
0.13873
0.18359
0.30943
0.13915
0.31844
0.11586
0.12585
0.12474
0.12863
0.12975
0.09886
0.16136
0.31395
0.14198
0.06348
0.07185
0.08346
0.12678
0.30559
0.08734
0.12151
0.23059
0.17139
0.07220
0.09219
0.14990
TABLE 5-1. EMISSIONS DATA FOR ALUMAX POTROOM GROUP 101G/161W, LB F/T Al
5-8
-------
Date
01/81
02/81
03/81
04/81
05/81
06/81
07/81
08/81
09/81
10/81
11/81
12/81
01/82
02/82
03/82
04/82
05/82
06/82
07/82
08/82
09/82
10/82
11/82
12/82
01/83
02/83
03/83
04/83
05/83
06/83
07/83
08/83
09/83
10/83
11/83
12/83
01/84
02/84
03/84
04/84
05/84
06/84
07/84
08/84
09/84
10/84
11/84
12/84
01/85
02/85
03/85
04/85
05/85
06/85
07/85
08/85
Average
Roof
Monitor
101-H
X8AR
0.6740
1.2850
0.8500
0.9740
0.7500
1.1333
0.8933
0.6833
0.8700
0.9133
0.8007
0.8070
0.5660
1.0337
0.9200
0.8500
0.7867
0.7833
0.8067
0.8200
1.1033
0.9367
0.7700
0.8833
0.7833
0.8533
0.7200
0.6867
0.9500
0.9067
0.9800
1.0433
0.8233
0.8167
0.8900
0.8367
0.8933
0.8200
0.8033
0.8767
0.7633
0.8400
0.7467
1.0067
0.7900
0.9700
1.0033
0.9733
1.0567
0.7033
0.3600
0.7000
1.0467
0.9277
1.0253
0.7073
Dry F
Scrubber C
161-E ]
XBAR
0.0200
0.0350
0.0173
0.0333
0.9820
0.0737
0.0623
0.0810
0.0703
0.0417
0.0367
0.0697
0.0187
0.0397
0.0150
0.0303
0.0430
0.0453
0.0453
0.0453
0.0453
0.0453
0.0453
0.0453
0.0453
0.0453
0.0453
0.0453
0.0453
0.0453
0.0453
0.0453
0.0537
0.0457
0.0457
0.0457
0.0457
0.0457
0.0457
0.0457
0.0457
0.0457
0.0457
0.0457
0.0457
0.0570
0.0463
0.0463
0.0463
0.0463
0.0463
0.0463
0.0463
0.0463
0.0874
0.0463
'otroom
!roup
L01H/161E
XBAR
0.6940
1.3200
0.8673
1.0073
0.8320
1.2070
0.9556
0.7643
0.9403
0.9550
0.8374
0.8767
0.5847
1.0734
0.9350
0.8803
0.8297
0.8286
0.8520
0.8653
1.1486
0.9820
0.8153
0.9286
0.8286
0.8986
0.7653
0.7320
0.9953
0.9520
1.0253
1.0886
0.8770
0.8624
0.9357
0.8824
0.9390
0.8657
0.8490
0.9224
0.8090
0.8857
0.7924
1.0524
0.8357
1.0270
1.0496
1.0196
1 . 1030
0.7496
0.9063
0.7463
1.0930
0.9740
1.1127
0.7536
0.9162
Roof
Monitor
101-H
STD OEV
0.1216
0.3234
0.1036
0.1785
0.3219
0.1601
0.2223
0.1069
0.1808
0.1115
0.1398
0.1236
0.0666
0.0199
0.2778
0.0436
0.1222
0.1582
0.2098
0.0265
0.1168
0.1380
0.1323
0.1357
0.1350
0.1401
0.2931
0.0950
0.1908
0.1115
0.0520
0.1850
0.1102
0.1537
0.0624
0.1041
0.2003
0.3260
0.1429
0.2970
0.1701
0.2338
0.0569
0.1060
0.2718
0.2600
0.1518
0.2775
0.3252
0.2710
0.2651
0.1311
0.1595
0.1300
0.2400
0.0405
Dry Potroom
Scrubber Group
161-E 101H/161E
STD DEV VARIANCE
0.0035 0.01479881
0.0122 0.10473640
0.0031 0.01074257
0.0153 0.03209634
0.0641 0.10772842
0.0178 0.02594885
0.0102 0.04952133
0.0225 0.01193386
0.0078 0.03274948
0.0047 0.01245434
0.0095 0.01963429
0.0101 0.01537897
0.0015 0.00443781
0.0074 0.00045077
0.0020 0.07717684
0.0076 0.00195872
0.0095 0.01502309
0.0187 0.02537693
0.0187 0.04436573
0.0187 0.00105194
0.0187 0.01399193
0.0187 0.01939369
0.0187 0.01785298
0.0187 0.01876418
0.0187 0.01857469
0.0187 0.01997770
0.0187 0.08625730
0.0187 0.00937469
0.0187 0.03675433
0.0187 0.01278194
0.0187 0.00305369
0.0187 0.03457469
0.0081 0.01220965
0.0182 0.02395493
0.0182 0.00422500
0.0182 0.01116805
0.0182 0.04045133
0.0182 0.10660724
0.0182 0.02075165
0.0182 0.08854024
0.0182 0.02926525
0.0182 0.05499368
0.0182 0.00356885
0.0182 0.01156724
0.0182 0.07420648
0.0658 0.07192964
0.0233 0.02358613
0.0233 0.07754914
0.0233 0.10629793
0.0233 0.07398389
0.0233 0.07082090
0.0233 0.01773010
0.0233 0.02598314
0.0233 0.01744289
0.0233 0.05814289
0.0233 0.00218314
Potroom
Group
101H/161E
STD DEV
0.12165
0.32361
0.10364
0.17915
0.32821
0.16108
0.22252
0.10924
0.18095
0.11160
0.14011
0.12400
0.06661
0.02123
0.27781
0.04425
0.12256
0.15929
0.21064
0.03243
0.11828
0.13924
0.13361
0.13696
0.13628
0.14133
0.29369
0.09682
0.19170
0.11305
0.05526
0.18594
0.11049
0.15477
0.06500
0.10568
0.20113
0.32650
0.14404
0.29757
0.17108
0.23452
0.05974
0.10754
0.27242
0.26817
0.15356
0.27848
0.32601
0.27201
0.26612
0.13314
0.16118
0.13207
0.24113
0.04672
0.16628
TABLE 5-2. EMISSIONS DATA FOR ALUMAX POTROOM GROUP 101H/161E, LB F/T AT.
5-9
-------
Roof Dry Potroom
Monitor Scrubber Group
103-G 162-W 103G/162W
Date
03/81
04/81
05/81
06/81
07/81
08/81
09/81
10/81
11/81
12/81
01/82
02/82
03/82
04/82
05/82
06/82
07/82
08/82
09/82
10/82
11/82
12/82
01/83
02/83
03/83
04/83
05/83
06/83
07/83
08/83
09/83
10/83
11/83
12/83
01/84
02/84
03/84
04/84
05/84
06/84
07/84
08/84
09/84
10/84
11/84
12/84
01/85
02/85
03/85
04/85
05/85
06/85
07/85
08/85
Average
XBAR
0.5430
0.9405
0.5933
0.5800
0.5800
0.7067
0.8620
0.7590
0.7953
0.0280
0.6323
0.7483
0.3333
0.7500
0.9200
0.9000
0.6567
0.8100
0.5033
0.9067
0.7333
0.4967
0.5367
1.0033
0.5633
0.7067
0.4333
1.1200
0.8433
1.1167
0.6000
0.8000
0.9300
0.8867
0.7300
0.5500
0.7033
0.9267
0.8167
0.9533
0.9533
0.7300
0.6667
0.6033
0.6633
0.9800
0.4100
0.7167
0.7300
0.6367
1.1300
0.9067
0.9410
0.9237
XBAR
0.0207
0.0560
0.0910
0.1610
0.0657
0.0590
0.0797
0.1263
0.0493
0.0410
0.1137
0.0450
0.0290
0.0430
0.0493
0.0678
0.0678
0.0678
0.0678
0.0678
0.0678
0.0678
0.0678
0.0678
0.0678
0.0678
0.0678
0.0678
0.0678
0.0678
0.0678
0.0473
0.0665
0.0665
0.0665
0.0665
0.0665
0.0665
0.0665
0.0665
0.0665
0.0665
0.0665
0.0450
0.0652
0.0652
0.0652
0.0652
0.0652
0.0652
0.06S2
0.0652
0.0652
0.0652
XBAR
0.5637
0.9965
0.6843
0.7410
0.6457
0.7657
0.9417
0.8853
0.8446
0.0690
0.7460
0.7933
0.3623
0.7930
0.9693
0.9678 .
0.7245
0.8778
0.5711
0.9745
0.8011
0.5645
0.6045
1.0711
0.6311
0.7745
0.5011
1,1878
0.9111
1.1845
0.6678
0.8473
0.9965
0.9532
0.7965
0.6165
0.7698
0.9932
0.8832
1.0198
1.0198
0.7965
0.7332
0.6483
0.7285
1.0452
0.4752
0.7819
0.7952
0.7019
1.1952
0.9719
1.0062
0.9889
0.8070
Roof Dry Potroom Potroom
Monitor Scrubber Group Group
103-G 162-W 103G/162W 103G/162W
STO DEV STD OEV VARIANCE
0.0459
0.0771
0.2219
0.0721
0.0954
0.2434
0.1231
0.1919
0.0979
0.0115
0.1429
0.2008
0.0551
0.0900
0.0500
0.4403
0.0252
0.1442
0.0252
0.3179
0.0306
0.1537
0.2434
0.2318
0.0551
0.0950
0.0153
0.0819
0.1701
0.2914
0.1229
0.0700
0.0173
0.1815
0.1179
0.0200
0.0723
0.1804
0.1833
0.0231
0.1419
0.0173
0.0666
0.1412
0.1531
0.0436
0.1493
0.1050
0.1113
0.2194
0.0854
0.0493
0.0358
0.0510
0.0045 0.00212706
0.0390 0.00746541
0.0148 0.04945865
0.0746 0.01076357
0.0163 0.00936685
0.0098 0.05933960
0.0204 0.01556977
0.1058 0.04801925
0.0059 0.00961922
0.0053 0.00016034
0.0359 0.02170922
0.0066 0.04036420
0.0060 0.00307201
0.0166 0.00837556
0.0093 0.00258649
0.0376 0.19527785
0.0376 0.00204880
0.0376 0.02220740
0.0376 0.00204880
0.0376 0.10247417
0.0376 0.00235012
0.0376 0.02503745
0.0376 0.06065732
0.0376 0.05514500
0.0376 0.00444.977
0.0376 0.01043876
0.0376 0.00164785
0.0376 0.00812137
0.0376 0.03034777
0.0376 0.08632772
0.0376 0.01651817
0.0123 0.00505129
0.0365 0.00163154
0.0365 0.03427450
0.0365 0.01523266
0.0365 0.00173225
0.0365 0.00655954
0.0365 0.03387641
0.0365 0.03493114
0.0365 0.00186586
0.0365 0.02146786
0.0365 0.00163154
0.0365 0.00576781
0.0161 0.02019665
0.0356 0.02470697
0.0356 0.00316832
0.0356 0.02355785
0.0356 0.01229236
0.0356 0.01365505
0.0356 0.04940372
0.0356 0.00856052
0.0356 0.00369785
0.0356 0.00254900
0.0356 0.00386836
STD DEV
0.04612
0.08639
0.22239
0.10374
0.09678
0.24358
0.12478
0.21912
0.09807
0.01266
0.14734
0.20090
0.05542
0.09151
0.05085
0.44188
0.04526
0.14903
0.04526
0.32010
0.04848
0.15823
0.24630
0.23482
0.06671
0.10216
0.04059
0.09012
0.17420
0.29381
0.12851
0.07106
0.04039
0.18512
0.12341
0.04161
0.08099
0.18404
0.18689
0.04319
0.14651
0.04039
0.07594
0.14211
0.15717
0.05628
0.15347
0.11086
0.11685
0.22226
0.09252
0.06080
0.05048
0.06219
0.12647
TABLE 5-3. EMISSIONS DATA FOR ALUMAX POTROOM GROUP 103-G/162W, LB F/T AT
5-10
-------
Roof Dry Potroom
Monitor Scrubber Group
103-H 162-E 103H/162E
Date
03/81
04/81
05/81
06/81
07/81
08/81
09/81
10/81
11/81
12/81
01/82
02/82
03/82
04/82
05/82
06/82
07/82
08/82
09/82
10/82
11/82
12/82
01/83
02/83
03/83
04/83
05/83
06/83
07/83
08/83
09/83
10/83
11/83
12/83
01/84
02/84
03/84
04/84
05/84
06/84
07/84
08/84
09/84
10/84
11/84
12/84
01/85
02/85
03/85
04/85
05/85
06/85
07/85
08/85
Average
XBAR •
1.1287
0.8043
1.2900
1.0867
1.0567
0.8400
1.1057
0.6680
0.9750
0.8937
0.8230
0.8710
0.6033
0.7733
0.8967
0.8700
1.1833
1.1167
1.0200
0.7100
0.9033
0.4200
0.3700
0.8700
1.1067
0.8200
0.8933
0.8600
0.9733
0.7500
0.9067
0.6867
0.6667
0.7700
0.6900
0.5967
0.8633
0.9700
0.9167
0.8233
0.8867
1.3133
0.8700
0.7467
0.8233
0.7067
0.4200
0.6367
0.9300
0.7933
0.9033
0.9520
0.9345
0.9217
XBAR
0.0797
0.0380
0.1653
0.1623
0.1000
0.2197
0.1420
0.1240
0.1150
0.0480
0.0600
0.1017
0.0497
0.0687
0.1143
0.1059
0.1059
0.1059
0.1059
0.1059
0.1059
0.1059
0.1059
0.1059
0.1059
0.1059
0.1059
0.1059
0.1059
0.1059
0.1059
0.0443
0.1020
0.1020
0.1020
0.1020
0.1020
0.1020
0.1020
0.1020
0.1020
0.1020
0.0767
0.1006
0.1006
0.1006
0.1006
0.1006
0.1006
0.1006
0.1006
0.1006
0.1006
0.1006
XBAR
1.2084
0.8423
1.4553
1.2490
1.1567
1.0597
1.2477
0.7920
1.0900
0.9417
0.8830
0.9727
0.6530
0.8420
1.0110
0.9759
1.2892
1.2226
1.1259
0.8159
1.0092
0.5259
0.4759
0.9759
1.2126
0.9259
0.9992
0.9659
1.0792
0.8559
1.0126
0.7310
0.7687
0.8720
0.7920
0.6987
0.9653
1.0720
1.0187
0.9253
0.9887
1.4153
0.9467
0.8473
0.9239
0.8073
0.5206
0.7373
1.0306
0.8939
1.0039
1.0526
1.0351
1.0223
0.9618
Roof Dry
Monitor Scrubber
103-H 162-E
STD
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
DEV STD
1382
1816
1493
0404
0551
0400
2626
0354
1202
2619
2066
2195
1250
0862
1935
1758
2183
2150
1562
1345
2021
1709
1249
1931
5292
0866
1457
0800
1595
0964
0473
1266
0513
1952
1136
1531
2458
3516
1626
1250
0643
1401
1803
2060
1739
1210
0794
0902
1819
0833
0.1767
0.
0.
1911
2072
0.1045
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
DEV
0649
0040
0352
0470
0078
1424
0985
0201
0229
0030
0092
0341
0092
0127
0327
0524
0524
0524
0524
0524
0524
0524
0524
0524
0524
0524
0524
0524
0524
0524
0524
0047
0508
0508
0508
0508
0508
0508
0508
0508
0508
0508
0290
0498
0498
0498
0498
0498
0498
0.0498
0.0498
0.
0.
0.
0498
0498
0498
Potroom Potroom
Group Group
103H/162E 103H/162E
VARIANCE
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
02331125
03299456
02352953
00384116
00309685
02187776
07866101
00165717
01497245
06860061
04276820
04934306
01570964
00759173
03851154
03365140
05040065
04897076
02714420
02083601
04359017
03195257
01834577
04003337
28279840
01024532
02397425
00914576
02818601
01203872
00498305
01604965
00521233
04068368
01548560
02602025
06299828
12620320
02901940
01820564
00671513
02220865
03334909
04491604
03272125
01712104
00878440
0.01061608
0.03556765
0.00941893
0.03370293
0.03899925
0.04541188
0.01340029
STD DEV
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.15268
.18164
.15338
.06197
.05564
.14791
.28046
.04070
.12235
.26192
.20680
.22212
.12534
.08712
.19624
.18343
.22449
.22128
.16475
.14435
.20879
.17874
.13544
.20008
.53179
.10121
.15483
.09563
.16788
.10972
.07058
.12669
.07219
.20169
.12444
.16129
.25100
.35526
.17035
.13492
.08195
. 14903
.18261
.21192
. 18088
.13085
.09372
.10303
.18858
.09705
.18358
.19748
.21309
.11576
.16512
TABLE 5-4. EMISSIONS DATA FOR ALUMAX POTROOM GROUP 103H/162E, LB F/T Al
5-11
-------
As indicated in the model proposed in Section 4, the monthly mean
potroom group emissions, X, equal the sum of the roof monitor emissions X,
and the dry scrubber emissions. That is:
Similar procedures were employed to estimate the monthly standard
deviations of the dry scrubbers for those months for which performance tests
were not conducted. In this case, however, variances were averaged, from
which standard deviations were calculated. For example, the standard
deviation, 0.0634, used for the period from June 1982 through August 1983,
was calculated from the mean variance calculated for the months from
January 1981 through May 1982. That is:
S6/82 = f^l/Sl + $22/81 + $23/81 + "' + s24/82 + s25/82>/17l1/2' ^
S6/82-8/83 = °'0634 * [(0-04352 + 0.00172 + 0.10372 + 0.00322 + ... +
0.01102 + 0.00672)/17]1/2. (5.3)
For the period from October 1983 through September 1984:
S- = ^ + $2
10/83-9/84 = l/Sl + $2/81 +
S10/83-9/84 = °'0643 lb F/T A1- (5-4)
From November 1984 through August 1985:
Sll/84-9/84 ' [
-------
o
Where S . = the potroom group monthly variance estimate for month i,
2
S ,. = the roof monitor monthly variance estimate for month i,
11 and
2
S 2.j = the dry scrubber monthly variance estimate for month i.
For example for January 1981, it can be seen from the first row of numerals
in Table 5-1,
S^. = (0.1924)2, (5-7)
S22- = (0.0485)2, and (5-8)
S21/8] = 0.03937001, or (5-9)
Sl/81 = °-1984 1b F/T Al. (5-10)
•
The monthly mean value of standard deviations, S , is seen from
A
Table 5-1 to be 0.1499 Ib F/T Al, while the overall monthly means from the
potroom group, X, is 0.8473 Ib F/T Al. X and S" for the other ALUMAX
J\
potroom groups are given in Tables 5-4 through 5-6.
Table 5-5 shows the results of the calculation of the upper and lower
warning limits. Details of the calculations are given in Appendix B, and
are based on the methods used by the ASTM . Figures 5-3 and 5-4 show the X
and S control charts for ALUMAX Potroom Group 101G/161W, including Roof
^
Monitor 101G and Dry Scrubber 161W. Control charts for the other ALUMAX
potroom group are not included, but will be similar to Figure 5-4 and 5-5,
using the data of Table 5-5 for their construction.
5.4 DERIVATION OF ALUMAX CONTROL CHARTS
A control chart would be required for X and S for each potroom group.
^
Thus for the ALUMAX plant, eight control charts would be needed, two for
each potroom. Each month for which sampling is done, the monthly mean X.,
and the standard deviation of the three daily readings, S . are calculated.
^ 1
5-13
-------
TABLE 5-5. POTROOM GROUP WARNING LIMITS, ALUMAX DATA
Pot room Group
101G/161W
101H/161E
103G/162W
103H/162E
aL1m1ts are based
X1 and sx1.
Central
Line, X
o
0.8473
0.9162
0.8070
0.9618
on a monthly sample
X Charta
UWL
1.043
1.133
0.972
1.177
size of 3, I.e.*
LWL
•
0.652
0.700
0.642
0.747
three dally
Central
Line, 0QC
0.1499
0.1663
0.1265
0.1651
readings per month
Sx Chart3
UWL
0.272
0.302
0.229
0.299
to calculate
LWL
0
0
0
0
Standard central line, X = X, overall mean of monthly means for ALUMAX data.
cStandard Deviation, ao = s for ALUMAX data.
-------
C3
i—t
o
VI
a.
O
O
O .0
J- fc>
O T3
a. c
m
•— «
O i
-o
o -o
oo
^= o
o ••->
o o
i. Q>
4-> O.
c w
O d)
<_> a:
IX
CO
I
m
0)
s_
3
cn
-------
91-S
0 0
• j> pus )
'MI91/9IOI
-------
Their values could either be plotted on the X and $x chart or numerically
compared with the UWL and LWL values in Table 5-7. If LWL < X. < UWL for
the X chart, and if LWL < Sx1 < UWL for the S chart, then the process is in
control. Graphically, if X. and Sx1 fall between UWL and LWL, the
indications are the process is in control. Should any of the inequalities
cited above not be true, i.e., should X. or S j fall outside the warning
limits, then action is appropriate. An example of regulatory actions that
might be taken is given in the next section.
5.5 REGULATORY USE OF ALUMAX CONTROL CHARTS
The regulatory action taken if the monthly Xi and S . fall outside the
warning limits should relate to the risk involved. For example, the risk of
an exceedance, or the risk of any process change (or emission change) going
undetected, is inversely proportional to the frequency of performance
testing. A yearly schedule of performance testing results in a higher risk
of an undetected emission level change than a quarterly schedule of testing.
The example that follows assumes an annual performance test schedule.
It uses corrective action having stringent requirements and guidelines.
This was intentionally done to err on the side of being overly cautious so
as not to permit any potential change in overall emission level to be
unchallenged and unverified. As the regulatory procedure gains more
experience, then the periodic review of the procedures that are recommended,
might relax some of the requirements of quarterly performance test schedule
compared to the requirements of an annual schedule. Such a relaxation would
be consistent with the fact that the risk of an undetermined change on a
quarterly performance test schedule is less than the risk for an annual
schedule.
The most important control chart criterion to monitor, from the
regulatory standpoint, is the Upper Warning Limit, UWL, on the X control
chart. If the overall emission level of the potroom group is increasing,
the most likely indication would be the violation of the UWL on the X
control chart. Showed X. for any month be greater than UWL on the X chart,
5-17
-------
additional action must be taken immediately to determine if the violation is
caused by random variation, or by an overall increase in emission levels
greater than XQ. The method for making this judgement is given in the
example below.
On the otherhand, if X. is less than LWL on the X chart, or if a
monthly standard deviation, Sxi lies outside the warning limits of the
Sx chart, then that occurrence too should be treated as a potential change
in maintenance or operational practice which may indicate a change in
emissions. Unexpectedly low levels of X^ e.g., below LWL, may indicate a
sampling problem which is giving a falsely low measurement.
Values of Sx1 greater than UWL on the $x chart may indicate a
larger-than-expected daily fluctuation in the fluoride emissions, or it may
indicate a careless analytical technique, or even an error in the
calculation of the results. Similarly, lower-than-usual values of S . may
indicate invalid sampling or analysis techniques. In short, any violations
of the warning limits on the X chart or the $x chart should be investigated
as a potential change in operating or maintenance procedures in the potroom
group.
In the discussion that follows, the logic of the regulatory actions
will be outlined as briefly as possible. The mathematical formulas on which
the logic and the decisions are based are given in detail in Appendix C,
HYPOTHESIS TESTING.
To give an example of the regulatory action that may be taken, the
following set of conditions are assumed:
1. Potroom Group 1016/161W has requested an annual performance test
schedule based on their previous performance schedule of 3 samples
per month and 56 months of testing.
2. Potroom Group 101G/161W is permitted to go on an annual sampling
and performance test schedule based on the location of the potroom
group in Figure 4-1 and % - 0.8473, and S- - 0.1520 Ib F/T Al.
3. Warning Limits and Central Lines in units of Ib F/T Al have been
established as follows:
5-18
-------
Warning Limits Central Lines
UWL LWL ~J
X chart 1.043 0.652 0.847
Sx chart 0.302 0.000 --- 0.1499
So long as the potroom group can demonstrate control, annual sampling and
performance tests can continue. To demonstrate control, both the annual
means (of three consecutive daily readings) X. and the standard deviations
of the three readings, $xi must fall within the established warning limits.
That is, an exceedance or violation of the control limits occurs when any of
the following inequalities is untrue.
0.652 < X. < 1.043, and
0.0 < Sx. < 0.302.
If any one of the four limits is violated, certain actions are required.
From the regulatory standpoint, the most critical limit is the upper warning
limit on X., 1.043. If X. > 1.043, the overall level of emissions for the
potroom group may have increased, which may invalidate the basis for
granting a reduced sampling schedule.
The discussion that follows focuses on j^. Discussions on S follow
the discussion dealing with Xi exceedances of the control limits.
1. Let X be the first annual mean potroom group emission rate,
Ib F/T Al that gives X. > 1.043.
2. Whenever X. > 1.043, the potroom group will be required to sample
the following month, calling the mean for the following month JL.
3. After X2 has been determined, three possibilities exist.
a. X2 > 1.043
b. X2 < 1.043, but [(Xj + X2)/2] < 1.043
c. X2 < 1.043, but [(Xj + X2)/2] > 1.043
4. If 3(a) occurs, then the potroom group is required to do
performance tests monthly for the next six months, determining Xv
w w *» .1 . .—, -J
4* 5'
5-19
-------
5. If 3(b) occurs, the potroom group may return to annual sampling
and performance tests.
6. If 3(c) occurs, sampling will be required a third month
consecutively giving X3.
7. Three possibilities exist, after sampling the third consecutive
month,
a. X3 > 1.043
b. X3 < 1.043 and [(Xj + X2 + fy/3] < 1.043, or
c. X3 < 1.043 and [(Xj + X2 + X3)/3J > 1.043
8. If 7(a) occurs, then sampling will be required for the next
5 months, giving X4, XB, Xg, X?, and Xg.
9. If 7(b) occurs, then the potroom group may return to an Annual
sampling schedule.
10. If 7(c) occurs, then sampling will be required for the next
5 months, giving X4, Xg, Xg, X7, and Xg.
11. If 3(a), 7(a), or 7(c) occurs, a total of 8 months of consecutive
testing will have accurred. Based on these eight months of
testing, a revaluation of the overall operating level, i.e., the
8-month mean, will be done to decide whether the 8-month mean is
significantly larger than the initial mean, XQ, upon which the
reduced performance test is based. The decision will be based on
the outcome of an hypothesis test on the means (XQ and the 8-month
mean, X,) described in Appendix C.
12. If the hypothesis test indicates that no change has occurred in
the emissions level, then the potroom group may return to
annual sampling.
13. If the hypothesis test causes the decision to be made that an
increased emissions level has occurred, then a revaluation of
whether the potroom group still qualifies for reduced sampling
schedule by relocating Xj and S-j on Figure 4-1 and see'if the
potroom group still qualifies for reduced sampling.
14. If the potroom group no longer qualifies for annual sampling, then
it must return to the appropriate sampling and analysis schedule
as determined by the regulatory agency, e.g., see Table 2-1.
5-20
-------
15. If the potroom group still qualifies for annual sampling according
to the guidelines set by the Regulatory Authority in Figure 4-1,
then new values of XQ,
-------
5. If five consecutive annual readings fall below the central line,
Xo, (and at the same time greater than LWL) the action items of'
Step 1 should be done:
• check all analytical instruments for leaks,
t check all calculations for error, and
• Recalibrate the instruments before the next performance test.
6. If six consecutive annual readings fall below the central line
(and at the same time greater than LWL), Xo return to monthly
sampling for the next 5 months, obtaining X4, Xg, X7, and Xg (also
the corresponding standard deviations).
7. Proceed as in Steps 3 and 4.
The probability of a run of five to six values of X. less than the central
value on the X chart is very small, but the possibility cannot be discounted
completely or ignored.
For the Sx control chart, there is no possibility of having less than
Sx1=0.0, since Sx- is defined as the positive root of ( ? (X. - X)2/(n-l))1/2
If SX1 > UWL (=0.302 for the example used here), there is an'obvious
instability in the daily emissions measurements for the given month. The
instability may be occurring because of (a) random variation; (b) change in
operating procedures causing highly variable emissions measurements;
(c) change in sampling and analysis precision, perhaps faulty equipment; or
(d) an error in the calculation of results. If the problem results from (a)
or (d), the only response needed is to correct any known error. Should (b)
or (c) be the cause of Sx1 > 0.302, corrective measures should focus on
answering the question, "What change in sampling analysis, or operating,
procedures have occurred?", and taking the appropriate steps to correct'the
problem.
A reasonable regulatory course of action for the standard deviation
readings, similar to that used for X., is outlined below:
1. Let Sxl be the first annual standard deviation of the three daily
measurements in a given month that exceeds UWL.
5-22
-------
2. If Sxl > 0.302, then the potroom group will be required to do a
performance test the next month.
3. After $x2 has been determined, three possibilities exist:
(a) Sx2 > 0.302,
(b) Sx2 < 0.302, but [(Sxl + Sx2)/2] > 0.302, and
(c) Sx2 < 0.302, but [(Sxl + Sx2)/2] < 0.302.
4. If 3(a) or 3(b) occurs, the potroom group should continue monthly
performance tests for the next six months, determining S ,, S .,
Sx5' Sx6' Sx7' an(* ^x8* Proceed
5. If 3(c) occurs, the potroom group may return to annual sampling.
6. Redetermine X for the latest eight readings. Test the hypothesis
that the true mean of the last 8 is greater than the true mean of
the original 56 data points (Appendix C) .
7. If the test indicates that X for the last eight data has
increased, then recalculate X and S- and recheck the eligibility
A
of the potroom group for annual sampling from Figure 4-1.
8. If the potroom group using the new X and S- qualifies for annual
sampling, recalculate XQ, a0, UWL, and LWL for the potroom group
and return to annual sampling using the revised control chart
parameters.
9. If the test given in Step 6 indicates that no change has occurred
in the overall monthly mean emissions, then return to annual
sampling using the original control chart parameters.
10. If the potroom group using the new X and S- does not qualify (in
Step 8) for annual sampling, then it shall return to the
appropriate sampling and per performance test schedule based on
Figure 4-1 and Table 2-1.
Should X^ and S^ exceed their respective limits simultaneously, the
procedures outlined for the X parameters should take precedence.
5.6 LIMITATIONS OF REGULATORY USE OF CONTROL CHART
The basic assumption of control chart theory is that the sampling
interval is frequent enough to characterize the process, and that little
5-23
-------
process change occurs between sampling periods. Currently, the NSPS
for primary aluminum plants requires monthly performance testing. A reduced
performance test schedule could be set for quarterly sampling, semi-annual,
annual, or other sampling intervals. The longer interval between
performance tests, the slower would be the detectable response to any
systematic or overall change in emission levels and the lower the level of
assurance that changes in emissions performance are detected by the reduced
monitoring frequency. Furthermore, during periods between
performance tests the regulatory authority must take on faith that plant
operations and emission rates remain constant. The presumption that plant
operations remain constant between tests is the basic limitation of the X
and Sx control chart procedure for sanctioning reduced monitoring frequency.
Control chart theory, however, can be applied again in a different
manner to offset this limitation and help ensure compliance with the NSPS.
The premise is to track and chart parameters other than the performance test
data during the periods between tests as a surrogate monitoring method.
Fortunately, various routine maintenance procedures and operational
practices done by primary aluminum plant personnel are ongoing and occur on
each shift, daily, or on some other regular periodic interval. Listed below
are some routine procedures and daily occurrences that can be related to the
plant operating performance and to the fluoride emissions.
• The number of loose vent covers on the aluminum cell housings each
day;
• The number of bent vent covers;
• The number of gaskets changed per day (or per week);
• The number of cells showing buildup of electrolytic cell minerals
on the top edge of cell;
• The number of locations per day requiring clean up of electrolyte
cell materials'from the potroom floor;
t The pounds of electrolyte spilled and swept up from the floor per
day; and
t The time it takes to change each anode.
t The number of days required between baghouse filter replacement.
5-24
-------
Some of these are measurable parameters that could be "monitored"
between performance tests.
There are many other daily routine tasks not listed above that need to
be done to keep the plant operating efficiently and "in control."
Documentation and quantification of a defined set of these tasks during the
period of monthly sampling would permit the construction of a control chart
that would relate to operating and maintenance performance of the plant.
When the potroom group qualifies for reduced performance test
scheduling, the operating/maintenance control chart would continue to be
maintained on the same periodic schedule it operated on before the potroom
group, qualified for reduced performance testing. Any change in the
operating performance would be detected in this control chart, and could
signal a possible out-of-control condition which affects the emissions
level. The operating/maintenance control chart could serve as a surrogate
performance monitoring tool, and with experience and the appropriate data
could be related to fluoride emission rates (Ib F/T Al). Out-of-control
conditions with the operating/maintenance control charts would trigger an
increased performance test frequency in much the same way as did the X and
S- control charts described in the previous section.
The specific operating/maintenance parameters to be monitored would be
determined by evaluating plant operations on a case-by-case basis. The
concept, however, is described here for those regulatory authorities that
would require assurance that plant operations have not changed between test
periods.
5-25
-------
6.0 REGULATORY PROCEDURES
This section outlines the steps that are recommended and the data that
are required to qualify a potroom group in a primary aluminum plant for a
reduced frequency performance test schedule. Quantitative criteria are
offered for making the decision as to which potroom groups are qualified for
a reduced frequency performance test schedule, and what the frequency of
testing should be. Suggestions are given for regulatory procedures while
the potroom group is on a reduced frequency performance test schedule.
Step 1. Assemble at least four years of consecutive monthly results
of emission rates, Ib F/T Al, from roof monitors in potroom
groups. Four years of dry scrubber emission rates are also
needed, on whatever periodic schedules have been specified by
the appropriate regulatory agency. It is assumed that each
monthly performance test measurement on the roof monitor or
dry scrubber includes three separate daily measurements from
which the monthly means and standard deviations may be
determined.
Step 2. Determine if there is a significant covariance of the
roof monitor and dry scrubber emissions. If the covariance
is insignificant, Equation 4-2 may be used to determine the
monthly standard deviation for the potroom group, otherwise a
covariance term must be added as in Equation 3-2. Equation
4-1 may be used to determine the monthly mean emissions for
the potroom group.
Step 3. Determine if the potroom group monthly mean emissions are
autocorrelated. If they are not significantly autocorrelated,
Figure 4-1 and a table such as Table 6-1, may be used to
qualify the potroom group for a reduced schedule performance
test. If there is significant autocorrelation a new Prob-
ability-of-an-Exceedance graph must be derived.
6-1
-------
TABLE 6-1. EXAMPLE CRITERIA FOR REDUCED PERFORMANCE TEST SCHEDULE
Example
Probability of an Performance
Exceedance' P Test Schedule3
P > °-001 one/month
0.0001 < P < O.OOl" one/quarter
0.00001 P< 0.0001C one/6 months
P <0-00001 one/12 months
a
authority1.°f ^ performance test schedule is the preview of the regulatory
P - 0.001 means that one exceedance would be expected with every 1,000
measurements or about 83.3 years, on the original monthly performance test
schedule assuming only random variation. performance test
P = 0.0001 means that one exceedance would be expected with every 10 000
833-33
6-2
-------
Step 4. Determine the overall monthly mean and monthly standard
deviation for the potroom group for the entire time period
available in Step 1.
Step 5. Determine the frequency distribution of the roof monitor
data. If the frequency distribution is normally distributed,
and there is no significant autocorrelation to the data, the
Probability-of-an-Exceedance graph (Figure 4-1) developed for
this study may be used; otherwise a new Probability-of-an-
Exceedance graph must be developed.
Step 6. Determine probability of an exceedance from Figure 4-1. This
assumes that they are normally distributed random variables.
Step 7. Regulatory authority selects appropriate performance test
schedule from Probability of Exceedance for potroom group
determined in Step 6. Example criteria for a reduced
performance test schedule are given in Table 6-1.
Step 8. Construct performance test X and SY control charts and follow
A
a monitoring protocol determined by the regulatory authority.
A protocol for monitoring control charts on a reduced
performance test schedule is suggested in Section 6.
Step 9. Construct surrogate parameter X and Sx control chart to
monitor operating performance while on reduced performance
test schedule. The surrogate parameter is a routine
measurement or housekeeping procedure that is done routinely
by operating personnel whether or not the potroom groups is
on reduced performance test schedule. See Section 2 for the
details.
6-3
-------
7.0 REFERENCES
1. "Four Year Statistical Analysis," ALUMAX of South Carolina (1985).
2. Daniels, M., ALUMAX of South Carolina, letter to Robert McCalister [sic],
dated September 19, 1985.
3. Steel, Robert G.D. and James H. Torrie, "Principles and Procedures of
Statistics," 2nd. ed. McGraw-Hill Book Company. New York (1979).
4. Pankratz, A., "Forecasting With Univariate Box - Jenkins Models.
Concepts and Cases," John Wiley & Sons, New York (1983).
5. Tukey, John W., "Exploratory Data Analysis," Addison-Wesley Publishing
Company, Reading, MA (1977).
6. Charavarti, I.M., R.G. Laha, and J. Roy, "Handbook of Methods of Applied
Statistics, Volume I. Techniques of Computation, Descriptive Methods,
and Statistical Inference," John Wiley & Sons, Inc., New York (1967).
7. Anderson, T.W., "The Statistical Analysis of Time Series," John Wiley &
Sons, New York (1971).
8. Hald, A., "Statistical Tables and Formulas," John Wiley & Sons, Inc.,
London (1952).
9. Aitchison, J., and J.A.C. Brown, "The Lognormal Distribution,"
Cambridge University Press, Cambridge (1957).
10. "ASTM Manual on Presentation of Data and Control Chart Analysis,"
ASTM Special Technical Publication 15D., American Society for Testing
and Materials, Philadelphia, PA 19103 (1976).
7-1
-------
11. Duncan, Acheson J., "Quality Control and Industrial Statistics,"
3rd Edition, Richard D. Irwin, Inc., Homewood, Illinois (1965).
12. Hastings, N.A.J., and J.B. Peacock, "Statistical Distributions,"
John Wiley & Sons, New York (1975).
13. Hald, A., "Statistical Theory with Engineering Applications,"
John Wiley & Sons, Inc., New York (1952).
14. Snedecor, George W., and William G. Cochran, "Statistical Methods,"
7th ed., The Iowa State University Press, Ames. (1980).
15. SAS® Institute, Inc., "SAS/ETS® User's Guide," Version 5 Edition,
Gary, NC: SAS® Institute, Inc. (1984).
16. Letter from R. C. Dickie, ALUMAX, to Jack R. Farmer, U. S. EPA,
August 27, 1985, Attachment entitled "Emission Test Results,
Potline 1 & 2, October 1984.
17. Letter from Hurtis L. Givens, Alcan Aluminum to Eric Noble, U. S.
EPA, undated (received 6/28/85). Attachment entitled "Potline
Roof Emissions, Total Fluorides."
18. Letter from Hurtis L. Givens, Alcan Aluminum to Eric Noble, U. S.
EPA, undated (received 11/20/85). Attachment entitled "NSPS Total
F Emissions Data."
19. Letter from Robert E. Hurt, Noranda Aluminum to Eric Noble, U. S.
EPA, August 14, 1985. Attachments entitled "Table I. Summary of
Results Fluoride Emissions for 39B System" and "Table I. Summary
of Results Potroom Group III Fluroide Emissions."
20. Dickie, R. C. in a memorandum to Robert A. McAllister, dated January 29,
1986 "Hood Inspection Summary,"
7-2
-------
APPENDIX A
DATA TABLES
A-l
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
101-G ROOF MONITOR RESULTS 1981
DATE
LOG
TEST*
P F LB G F LB T F LB
ACFM
SCFM
'/. ISO
81/01/12
81/01/13
81/01/14
31/02/02
81/02/03
81/02/04
81/03/09}
81/03/101
81/03/li)
81/03/30
81/03/31
81/04/OJ^
"81/05/lTt
81/05/1 2\
81/0.5/13J
81/06/08
81/06/09
81/06/10
81/07/07\
81/07/081
81/07/09-1
31/08/10
81/08/11
81/08/12
81/09/28
81/09/29
81/09/30
31/10/05
81/10/06
81/10/07
31/1 1/09
31/1 1/10
81/1 1/1 1
81/1 1/30
81/12/01
81/12/03
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101 -G
006
007
008
012
013
014
203
205
207
232
233
234
244
246
248
262
264
267
272
274
277
294
296
298
312
313
314
315
313
320
333
338
340
341
343
347
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.548
.526
.232
.497
.247
.406
.425
.428
.395
.639
.364
.330
.510
.510
.520
.478
.642
.510
.480
.520
.510
.330
.430
.370
.470
.320
.280
.540
.260
.480
.328
.369
.616
.540
.335
,226
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.198
.134
.156
.168
.445
.384
.329
.353
.281
.426
.326
.268
.340
.506
.230
.372
.386
.650
.570
.550
.400
.410
.490
.400
.350
.400
.360
.450
.560
.420
.422
.471
.512
.429
.427
,239
0
0
0
0
0
0
0
0
0
1
0
0
0
1
0
0
1
1
1
1
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
.75
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
I
CO
DATE
101-H ROOF MONITOR RESULTS 1981
LOG
TEST**
P F LB
G F LB T F LB
ACFM
SCFM
X ISO
81/01/19
81/01/20
81/02/17
31/02/18
81/02/19
81/03/09
81/03/10
81/03/11
81/04/06
81/04/07
81/04/08
81/05/1 1
81/05/12
81/a5/13
81/06/08
81/06/09
81/06/10
81/07/07
81/07/08
81/07/09
81/08/10
81/08/1 1
81/08/12
81/08/31
81/09/01
81/09/02
81/10/05
81/10/06
81/10/07
81/1 1/09
81/1 1/10
81/1 1/1 1
81/1 1/30
8! -' ', 2/0!
31/12/02
101-H
101-H
101-H
101-H
101-H
101 -H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
i 0 1 -H
101-H
016
017
019
020
021
204
206
208
235
236
237
245
247
249
263
265
266
273
275
276
295
297
299
301
303
304
316
317
319
334
336
339
342
344
346
0.374
0.404
1 .087
0.654
0.369
0.416
0.464
0.609
0.772
0.459
0.538
0.600
0.240
0.480
0.730
0.410
0.640
0.660
0.420
0.335
0.340
0.350
0.250
0.550
0.400
0.630
0.400
0.340
0.460
0.352
0.279
0.305
0.352
0 . 564
0.394
0.214
0.350
0.569
0.450
0.760
0.342
0.371
0.359
0.406
0.419
0.326
0.500
0.210
0.230
0.410
0.560
0.650
0.490
0.340
0.420
0.410
0.390
0.310
0.340
0.280
0.410
0.430
0.530
0.580
0.579
0.374
0.513
0.323
0.356
0.432
0.538
0.760
1 .656
1 .063
1 .136
0.764
0.835
0.963
1 .130
0.878
0.364
1.090
0.450
0.710,
1 .140
0.970
1 .290
1 . 150
0.760
0.770
0.750
0.740
0.560
0.890
0 .680
1 .040
0.330
0.370
1 .040
,: .?3i
0.653
0.313
0.675
0.920
0.826
1 ,454
1 ,839
2,078
2,198
2,242
1 ,354
1 ,455
1 ,464
1 ,577
1 ,361
1 ,789
1 ,791
1 ,753
1 ,733
2,249
2,106
2,179
1 ,748
1 ,322
1 ,563
1 ,311
1 ,289
1 ,385
2,172
2,040
1 ,776
1 ,514
1 ,425
1 ,626
1 ,599
1 ,548
1 ,608
2,260
2,269
2,160
1 ,472
1 ,850
2,081
2,206
2,187
1 ,379
1 ,462
1 ,463
1 ,551
1 ,371
1 ,786
1 ,708
1 ,694
1 ,676
2,140
1 ,931
1 ,995
1 ,656
1,727
1 ,462
1 ,260
1 ,216
1 ,285
2,042
1 ,944
1 ,697
1 ,450
1 ,343
1 ,545
1 ,505
1 ,508
1 , 493
2,253
2,202
2,092
85.9
111.3
103.9
103.8
99.7
100.0
95.9
93.3
99.0
99.4
101 .3
104.2
103.3
94.3
102.7
105.0
102.0
103. 9
99.8
102.7
113.7
113.7
1 13.6
104.0
107.2
104.3
112.5
107.3
105.6
108.9
111.9
103.6
103.0
109.0
107. 1
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
103-6 ROOF MONITOR RESULTS 1981
DATE
LOG
TEST**
P F LB 6 F LB T F LB ACFM
SCFM
y. iso
81/03/02
81/03/03
81/03/04
81/04/13
31/04/14
81/05/26
81/05/27
81/05/33 -
81/06/01
81/06/02
81/06/03
81/07/27
81/07/28
81/08/03
81/08/04
81/08/05
31/09/08
81/09/09
81/09/10
81/10/12
81/10/13
81/10/14
81/11/02
81/1 1/03
31/1 1/04
81/12/07
31/12/08
fll /I ?/fi9
103-6
103-G
103-6
103-6
103-6
103-6
103-6
103-6
103-6
103-6
103-6
103-6
103-6
103-6
103-G
103-G
103-6
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
200
201
202
238
239
251
252
254
256
258
260
282
235
287
288
290
292
306
303
310
321
323
325
327
329
331
343
350
352
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
£
0
0
0
0
0
0
0
0
0
0
0
c
.327
.333
.348
.525
.599
.430
.350
.200
.270
.310
.290
.350
.260
.350
.630
.280
.406
.458
.341
.276
.405
= 299
.332
.294
.257
.248
.225
.305
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
p_.
U •
CK
0.
0.
0.
0.
0,
0.
0.
0.
0.
0.
0.
163
230
224
361
406
410
180
220
380
190
310
240
220
310
340
380
390""
545
445
343
573
380
541
535
428
212
299
305
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
o
1
0
0
0
0
0
0
0
0
0
0
.490V
.5681
.571'
.386
.995
.840
.530
.410
.640
.500
.600
.590
.480
.670
.970
.660
.48tl
T79A
.004 \
.786
.620
.978
.679
.872
.829
.635
.460
.524
.609
1 ,945
1 ,700
1,878
1,592
1 ,890
2,082
2,157
1 ,818
1 ,450
1 ,977
1 ,820
1 ,585
1 ,489
1 ,454
1 ,640
1 ,692
1 ,553
1 ,560
1 ,470
1 ,460
1 ,787
1 ,758
1 ,765
1 ,912
1 ,919
5,250
i
1 ,745
1 ,779
1,744
1 ,918
1 ,697
1 ,360
1 ,568
1 ,853
1 ,938
2,048
1 ,728
,391
,371
,736
,505
,409
,382
1 ,533
1 ,606
1 .453
1 ,465
I .380
1 ,392
1 ,718
1 ,705
1 ,715
1 ,355
1 .845
5,035
i ,736
1 ,71 7
1 ,740
100.2
98.7
98.3
102.2
105.7
117.3
109.3
105.3
108.6
104.1
105.2
93.3
104.3
1 12.7
1 14.6
106.3
112.3
108. 1
109.5
105.6
103.4
101 .1
111.0
109.5
104.6
1 14.4
102.3
102. 1
109. 1
£>.<<+ O
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
103-H ROOF MONITOR RESULTS 1981
DATE
LOG
TEST*
P F LB
G F LB T F LB
ACFM
SCFM
V. ISO
3=
I
cn
81/02/23^
81/02/24
81/02/25,
81/04/20
81/04/21
31/04/22
103-H
103-H
103-H
103-H
103-H
103-H
81/05/261 103-H
81/05/27 103-H
31/05/28J 103-H
81/06/01 103-H
81/06/02 103-H
81/06/03 103-H
31/07/27 103-H
81/07/28 103-H
81/07/29 103-H
31/03/03 103-H
31/03/04 103-H
31/03/05 103-H
81/09/08 103-H
31/09/09 103-H
81/09/10 i 103-H
81/1 O/O^
81/10/13
81/10/1*
31/11/0^
31/1 1/04>
103-H
103-H
103-H
103-H
103-H
31/12/07] 103-H
31/12/03 103-H
81/12/097 103-H
022
023
024
241
242
243
250
253
255
257
259
261
283
284
236
289
291
293
307
309
31 1
322
324
326
330
332
349
351
353
0
0
0
0
.
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.913
.383
.646
.673
- • ->
.330
.950
.720
.660
.500
.520
.620
.460
.630
.500
.370
.340
.330
.447
.638
.431
.409
.281
.372
.372
.512.
.496
.539
.494
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
u
0
0
0
.342
.267
.335
.335
.322
.359
.460
.410
.690
.620
.520
.470
.650
.420
.500
.430
.530
.450
.667
.726
.407
.299
.357
.237
.518
.547
.244
.606
.251
1
1
0
1
0
0
1
1
1
1
1
1
1
1
1
0
0
0
1
1
0
0
0
0
0
1
0
1
0
.260
.150
.980
.014
.703
.696
.400
.120
.350
.130
.050
.080
.110
.060
.000
.300
.830
.340
.1 14
.364
.839
.707
.638
.659
.888
.060
.739
.196
.746
2,067
2,134
1 ,930
1 ,651
1 ,635
1 ,237
2,521
2,429
2,378
1 ,324
2,016
1 ,905
1 ,573
1 ,664
1 ,563
1 ,586
1 ,547
1 ,518
1 ,416
1 ,464
1 ,377
l-,667
1 ,694
1 ,566
1 ,799
2,034
2,387
2,284
2,309
2,039
2,152
1 ,924
1 ,574
1 ,673
1 ,266
2,412
2,303
2.224
1 ,719
,913
,792
,469
,574
,466
,506
1 ,440
1 ,452
1 ,323
1 ,369
1,315
1 .603
1 ,648
1 ,519
1 ,735
1 ,951
2,336
2,262
2,325
99.3
105.0
105.6
101 .9
99.5
102.3
98.6
103.4
100.8
103.3
115.5
104.9
107.7
98.8
107.6
114.6
109.6
105.4
105.9
100.2
92.8
101 .5
101 .4
96.2
115.9
1 18.7
89.4
104.7
98.6
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
101-G ROOF MONITOR RESULTS 1982
DATE
i_OC
P F LB
G F LB T F LB
ACFM
SCFM
'/. ISO
82/01/11
82/01/12
82/01/13
32/02/01
82/02/02
32/0 2/0 3
82/03/08
32/03/09
82/03/10
82/04/05
82/04/04
82/04/07
82/04/26
32/04/27
82/04/28
82/06/01
82/06/02
82/0 A/0 3
32/07/19
82/07/20
82/07/21
32/08/09
82/08/10
32/08/1 1
32/09/14
32/09/15
82/09/16
32/09/27
32/09/28
82/09/29
82/11/08
32/11/09
32/11/10
32/12/06
32/12/07
.ri -o / fl 1 X 7s O
101-G
101-G
101-G
101 -G
101-G
101-G
101 -G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101 -G
10 1 -G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101 -6
.1 ri i _ r.
360
362
364
366
368
370
384
386
388
401
403
405
419
421
423
425
427
429
437
439
441
461
463
465
479
431
483
491
493
495
509
511
513
515
517
«=; i o
0.192
0.373
0.180
0.271
0.297
0.220
0.170
0.560
0.590
0.370
0.370
0.400
0.330
0 . 240
0.450
0.270
0.290
0.350
0.240
0.210
0.320
0.130
0.210
0.350
0.330
0.380
0.470
0.380
0.380
0.450
0.180
0.240
0.320
0.310
0.450
":• 3 AH
0.144
0.240
0.189
0.223
0.217
0.319
0.320
0.350
0.330
0.440
0.260
0.270
0.370
0.380
0.300
0.360
0.420
0 .530
0.350
0.2-?:
0.490
0.350
0.530
0.530
0 .560
0.480
0 . 620
0.320
0.320
0.520
0.430
0.330
0.450
0.420
0.570
n . 4<=>Q
0 . 336"
0.613
0.362-
0 . 499\
0.514
o.ssa
0 . 490n
0.910
0.910-
0.810
0.630
0.670,
0.69(1
0.620
0.750.
0 . 620
0.700
0.920
0.59ff
o . 4>'C
0.820-
0.520
0.740
0.890J
0 . 950~
0.370
1 .090-
0.70CT
0.700
0.97a
0.610"
0.570
0.76Qj
0.73(3
I nQ!0
0 -810
1,446
1 ,918
1 ,889
,986
,726
,935
2,668
2,672
2,575
\ 2,236
2,277
J 2,251
2,531
2,656
2,560
2,145
2,146
2,080
1 ,358
1 ,363
1 ,382
1 ,390
1,916
2,352
1 2,374
2,305
/ 2,325
2,277
2,283
2,310
2,622
2,441
2,493
2,268
2,236
2.094
1 ,491
1 ,910
1 ,902
1 ,974
1 ,741
1 ,926
2,661
2,726
2,569
2,177
2,202
2,247
2,533
2,559
2.455
2,054
1 ,991
2,023
1 ,723
1 ,799
1 ,732
1 ,762
1 ,787
2,163
2,213
2, 131
2,166
2,175
2,193
2,194
2.640
2,413
2,477
2,194
2,185
2,061
102.8
108.1
101 .6
101 .4
102.2
105.2
94.8
101 .7
105.9
105.3
109.7
101 .7
105. 1
103.2
108. 1
10S. 1
110.6
106.5
1 10.6
109.9
110.4
109. 1
98.5
109. 1
104.8
101 ,7
100.7
101 . 1
100.5
102.4
96.6
95.9
95.0
99.o
100 .2
103.0
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
DATE
LOG
101-H ROOF MONITOR RESULTS 1982
TEST M P F LB Q F LB T F LB ACFM
SCFM
'/. ISO
82/01/11
82/01/12
82/01/13
82/02/01
82/02/02
82/02/03
82/03/08
82/03/09
82/03/10 .
82/04/05
82/04/06
82/04/07
82/04/26
82/04/27
82/04/28
82/06/01
82/06/02
82/06/03
82/07/19
82/07/20
82/07/21
82/08/09
82/08/10
82708/1 1
82/09/14
82/09/15
82/09/16
82/09/27
82/09/28
82/09/29
82/1 1/08
82/1 1/09
82/1 1/10
82/12/06
82/12/07
32/«. 2/Q8
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101 -H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101 -H
101-H
101-H
101-H
101-H
101-H
101-H
,1 0 1 -M
361
363
365
367
36?
371
385
387
389
402
404
406
420
422
424
426
428
430
438
440
442
462
464
466
480
482
484
492
494
496
510
512
514
516
518
520
0.422
0.327
0.263
0.558
0.625
0.756
0.270
0.650
0.780
0.400
0.420
0.530
0.270
0.380
0.450
0.230
0.370
0.330
0.300
0.370
0.520
0.380
0.360
0.400
0.460
0.410
0.370
0.330
0.460
0.520
0.420
0.350
0.310
0.320
0.450
0.380
0.192
0.163
0.331
0.469
0.431
0.262
0.320
0.400
0.320
0.380
0.460
0.340
0.420
0.380
0.460
0.390
0.450
0.330
0 .260
0.600
0.360
0.430
0.440
0.450
0.780
0.590
0.710
0.400
0.530
0 . 5'30
0.500
0.320
0.410
0.490
0.590
0.420
0.614
0.490
0.594
1.027
1 .056
1 .013
0.600
1 .060
1 .100
0.800
0.880
0.870
0.680
0.760
0.920
0.610
0.820
0.750
0.570
0.970
0.880
0.810
0.800
0.850
1 .230
1 .000
1 .080
0.780
0.990
1 .040
0.920
0.670
0.720
0.310
1 .040
0.800
2,098
2,416
2,342
2,342
2,103
2,050
2,591
2,562
2,568
2,242
2,211
2,270
2,272
2,277
2,288
2,220
2,250
2 , 254
2,296
2,322
2,348
1,848
1 ,836
1 ,955
2,350
2,327
2,359
2,348
1 ,952
2,140
2,225
2,025
1 ,993
1 .323
2,101
2,085
2,138
2,478
2,357
2,305
2,097
2,033
2,610
2,591
2,581
2,188
2,201
2,260
2,200
2,136
2,258
2,163
2,096
2,088
2,186
2,1 15
2,183
1 ,743
1 ,743
1 ,836
2,221
2,200
2,181
2,241
1 ,840
2,022
2,196
1 ,984
1 ,955
1 ,751
2,058
2,074
101 .7
98.3
100.1
115.5
114.4
103.8
97 k 2
97.5
101 .9
107.6
106.3
106.4
108.4
106.2
108.1
111.2
1 10.5
110.3
100.0
103.5
96.4
95.5
95.8
95. 1
104.7
101 .6
105.3
108.5
109.2
108.3
103.1
103.3
105.8
106.9
104.8
105.7
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
103-G ROOF MONITOR RESULTS 1982
DATE
LOG
TEST *
P F LB
G F LB T F LB
ACFM
SCFM
'/. ISO
I
CO
32/01/04
82/01/05
J2/01/06
J2/02/08
82/02/09
32/02/10
J2/03/01
82/03/02
82/03/03
J2/04/12
32/04/13
82/04/14
J2/04/20
)2/04/21
82/04/22
92/06/14
12/06/15
32/06/16
82/07/26
J2/07/27
12/07/28
82/08/02
12/08/03
12/08/04
82/09/20
82/09/21
12/09/22
J2/10/04
82/10/05
J2/10/06
(2/11/01
82/11/02
92/11/03
J2/ 12/20
d2/ 12/21
• 3 1 , 1 O -"> O ,
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
i riQ—r;
354
356
358
372
374
376
378
380
382
407
409
411
413
415
417
431
433
435
443
445
447
449
451
453
485
487
489
497
499
501
503
505
507
521
523
=••:><=;
0.293
0.352
0.379
0.303
0.324
0.598
0.150
0.130
0.160
0.390
0.320
0.290
0.350
0.560
0.370
0.280
0.480
0.230
0.280
0.340
0.310
0.230
0.300
0.310
0.310
0.240
0.220
0.350
0.430
0.560
0.380
0.380
0.340
0.270
0 . \ 60
n ^ j£.ft
0.238
0.218
0.417
0.321
0.317
0.382
0.170
0.150
0.230
0.360
0.520
0.370
0.520
0.410
0.560
0.450
0.920
0.290
0.350
0.320
0.370
0.420
0.550
0.620
0.180
0 . 290
0.260
0.420
0.250
0.700
0.380
0.320
0.400
0.290
0. 160
n '^iin
0.532
0.569
0.796
0.624
0.641
0.980
0.330
0.280
0.390
0.750
0.840
0.660
0.870
0.970
0.920
0.730
1 .400
0.570
0.630
0.660
0.680
0.650
0.850
0.930
0.500
0.530
0.480
0.770
0.680
1 .270
0.760
0.690
0.740
0.570
0 . 320
n . xnn
1 ,974
1 ,950
1 ,994
2,113
2,121
2,165
1 ,341
1 ,303
1 ,317
2,387
2,254
2,253
2,361
2,503
2,525
1 ,902
2,058
2,337
2,045
2,154
2,158
2,150
2,091
2,136
2,047
2,087
2,112
2,004
2,005
1 ,953
1 ,916
1 ,896
1 ,924
1 ,358
1 ,900
1 . R33 <
I ,912.0
1 ,936.0
1 ,929.0
2,117.0
2,043.0
2,131 .0
1 ,349.0
1 ,307.0
1 ,301 .0
2,303.0
2,275.0
2,141 .0
2,368.0
2,445.0
2,553.0
1 ,789.0
1 ,912.0
2,176.0
1 ,926.0
2,006.0
2,031 .0
2,006.0
1 ,961 .0
1 ,997.0
1 ,908.0
2,011 .0
2,035.0
1 ,894.0
1 ,893.0
1 ,332.0
1 ,827.0
1 ,302.0
I ,827.0
1 ,889.0
1 ,869.0
\ .337 . 0.
107.9
104.2
107.9
106.0
107.4
109.5
109.2
109.0
1 10.9
101 . 1
106.3
106.4
103.8
101 .4
100.6
1 10.3
109.1
103.4
110.7
106.4
104.7
109.5
110.0
109.4
103.5
104.6
98.6
103.9
100.1
106.6
106.4
104.6
103.1
106.9
104.7
i n ? . P.
-------
3'SJ I
E'ZOT
I *6T I
P'901
T*80T
T*80T
6*SOT
0*80T
8* TOT
S'SOT
0'30T
S'pOI
9*801
Z*60I
Z*60T
Z'OTT
8*30T
3*£OT
6*£OT
8*pH
T*90T
6* L6
3*.30T
0*90T
T*£OT
S'60I
9*9TT
OS I X
136' I
888' T
P98' T
£68* T
9S6* T
9£6* T
3p6* T
036* T
0^3*3
908*3
STS*3
093*3
8£3'3
p96* T
~S.fr 1*2
8Fl'3
383*3
T00'3
/160'3
*
2££* I
TZT'3
690*2
ZET'2
OSO'3
WdOS
OE6 ' I
Z T 6 ' 1
OE8' T
6TO'3
3SO*3
3£0*3
EfrO'3
EEO*3
93fr*3
699 'Z
996 ' 3
131*3
j _
t* w C- '-'
938 '3
303*3
691*3
ISO* 3
090 ' 3
/19fr* T
88S* T
3£I*3
690*3
E9T'3
S60'3
OP 1*2
WdD*
Or3* 0
08S*0
0/19*0
030* T
030* T
09£'0
038*0
OSZ'O
003* T
036 '0
OP6'0
006*0
03 T' T
066 '0
03p' T
000* T
0^9*0
Op6*0
08Z*0
03TM
089*0
0/16*0
0£8*0
oez*o
009*0
08p" 0
606*0
690* T
£89*0
38/1*0
ZrO* T
81 d i
OS I ' C
083 '0
033*0
OEE'O
0^9 '0
0^ '0
OOE'O
OSS'O
098*0
08S*0
OOS'O
OS9*0
038*0
OSP'O
06S*0
n T & • n
028 0
03S'0
03E*0
OOS'O
09S*0
09E*0
06E*0
Op£*0
O^p'O
OOE'O
3£E*0
3p£*0
163*0
22t7*0
9TE*0
T££*0
81 d 9
,~ ~f. .V a ^
Uc- » >-
OG'£ ' '
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
101-Q ROOF MONITOR RESULTS 1983
DATE
LOG
TEST *
P F LB G F LB T F LB
ACFM
SCFM
7. ISO
83/01/10
83/01/1 1
83/01/12
83/02/07
83/02/08
83/02/09
83/02/28
83/03/01
83/03/02
83/03/28
83/03/29
83/03/30
83/05/02
83/05/03
83/05/04
83/06/06
83/06/07
83/06/08
83/07/11
83/07/12
83/07/13
o1?.-' 03/0 8
W *• <' W W^ » W
83/08/09
33/08/10
83/09/12
83/09/13
83/09/14
83/10/10
83/10/1 1
83/10/12
83/11/07
83/11/08
83/i 1/09
33/1 1/28
Q Q X 1 'i s?9
101-G
101-G
101-6
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
. 501 -G
001
003
005
025
027
029
035
037
039
051
053
056
073
075
077
085
087
039
107
109
111
117
119
121
144
146
143
167
169
171
193
195
197
205
207
0.23
0.43
0.44
0.38
0.28
0.19
0.40
0.32
0.26
0.19
0.16
0.29
0.32
0.35
0.38
0.31
0.43
0.51
0.34
0.30
0.43
0.42
0.60
0,23
0.31
0.42
0.37
0.35
0.33
0.31
0.33
0.27
0.43
3.231
0.31
0.34
0.29
0.21
0.16
0.23
0.41
0.37
0.34
0.32
0.29
0.41
0.58
0.44
0.39
0.69
0.34
0.30
0.46
0.55
0.45
0.63
0.70
0.39
0.51
0.53
0.63
0.87
0.66
0.43
,0.66
0.54
0.59
Q.26
0.55\
0.78
0.72J
0.59^
0.44
0.4Z
o.aij
0.691
0.6QJ
0.511
0.44
0.70.
0.90
0.79
0.77J
1 .01
0.77
1.31J
G.80\
0.35
0.89
1.12"
1 .05
1 .29.
0.621
0.82
0.95;
0 .63
1 .23
1 .00.
0.74
0.98j
0.8L
i .02
0,419
1 ,792
2,456
2,454
1 ,515
1 ,604
1,599
1 ,758
1 ,777
2,069
2,134
1 ,671
1 ,921
2,057
2,090
2,457
1 ,978
1 ,993
1,972
1 ,951
1 ,805
1 ,742
2,122
2,284
2,031
1 ,921
1 ,948
__ 1,985
1 ,939
1 ,828
2,290
2,348
' 1,671
1 ,704
\ 1,937
1,956
1 ,749
2,403
2,428
1,526
1,628
1 ,622
1,745
1 ,733
2,030
2,080
1 ,653
1 ,905
1 ,943
1 ,964
2,376
1 ,348
1 ,861
1 ,357
1 ,815
1 ,674
1 .607
1 ,992
2,085
1 ,351
1 ,774
1 ,803
1 , 367
1 ,841
1 ,708
2,120
2,314
1 ,605
1 ,653
1 ,850
I ,935
101 .5
98.9
98.1
102.0
103.5
102.6
107.7
106.0
101 .8
90 .6
99.3
105.3
105.7
103.4
95.3
99.4
101.2
102.7
97.9
100. 1'
107.2
110.0
99.5
123.3
96.4
105. 1
93. 1
109.9
1 17.0
102.7
93.0
1 10.4
111.2
109.0
ioc .5
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
101-H ROOF MONITOR RESULTS 1983
DATE
LOG
TEST
P F LB G F LB T F LB
ACFM
SCFM
Y. ISO
83/01/10
83/01/11
83/01/12
83/02/07
83/02/08
83/02/09
83/02/28
83/03/01
83/03/02
83/03/28
83/03/29
83/03/30
83/05/02
83/05/03
83/05/04
83/06/06
83/06/07
33/06/08
83/07/11
83/07/12
83/07/13
83/08/08
83/08/09
33/08/10
83/09/12
83/09/13
83/09/14
83/ 10/10
83/10/11
83/10/12
83/11/07
83/1 1/08
83/1 1/09
83/1 1/2B
83/0 ' /29
W *^/ ft d s *~ f
01 / 1 1 /QO
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
. 101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
i c :, -'-
i n 1 -w
002
004
006
026
028
030
036
038
040
052
054
055
074
076
078
086
083
090
108
110
112
113
120
122
145
147
149
163
170
172
194
196
198
206
208
? \ n
0.30
0.34
0.41
0.25
0.37
0.38
0.21
0.35
0.30
0.31
0.32
0.25
0.34
0.33
0.46
0.26
0.32
0.27
0.34
0.25
0.25
0.31
0.35
0.40
0.24
0.25
0.33
0.22
0.29
0.36
0.32
0.24
0.13
0.28
Q .45
Q .39
0.35
0.58
0.37
0.56
0.37
0.63
Q.28
0.70
0.32
0.46
0.36
0.34
0.71
0.40
0.61
0.69
0.66
0.51
0.70
0.69
0.69
0.51
0.81
0.75
0.53
0.50
0.62
0.42
0.60
0.56
0.50
0.67
0.76
0.43
0.41
0 .53
0.65
0.92
0.78
0.81
0.74
1 .01
0.49
1 .05
0.62
0.78
0.69
0.59
1 .05
0.73
1 .07
0.95
0.99
0.73
1 .04
0.95
0.95
0.83
1 .16
1 .14
0.77
0.75
0.95
0.64
0.89
0.92
0.82
0.91
0.94
0.72
0.87
0.92
1 ,874
1 ,915
1 ,930
1,781
2,213
2,151
1,978
1,977
1 ,996
1 ,892
1 ,900
1 ,814
1 ,990
2,146
2,426
1 ,664
1 ,915
1 ,922
2,007
1 ,732
1 ,732
2,048
2,176
2,056
1 ,846
1 ,974
1 ,934
1 ,312
1 ,357
2,055
1 ,744
1 ,782
1 ,803
1 ,652
1 ,782
1 ,859
1 ,848
1 ,881
1 ,921
1 ,736
2,207
2,125
1 ,971
1 ,924
1 ,920
1 ,324
1 ,898
1,794
1 ,894
2,017
2,320
1 ,531
1 ,787
1 ,793
1 ,851
1,588
1 ,588
1 ,374
2,011
1 ,931
1 ,681
1 ,834
1 ,804
1 ,729
1 ,282
1 ,902
1 ,717
1 ,674
1 ,724
1 ,544
1 ,724
1 ,313
104.6
103.2
104.9
103.2
97. Q
105.1
103.1
107.0
99.4
98.3
97.2
100.1
108.3
107.8
100.3
107.2
103.0
100.3
1 10.3
112.9
112.9
107.4
100.7
103.3
107.0
103.4
99.4
101 .1
118.5
102.5
104.8
102. 1
95.9
98.9
105.0
101.5
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
103-G ROOF MONITOR RESULTS 1983
i
i—>
INJ
DATE
LOC
TEST #
P F LB
G F LB T F LB
ACFM
SCFM
'/. ISO
83/01/17
83/01/18
83/01/19
83/01/31
83/02/01
83/02/02
83/03/08
83/03/09
83/03/10
83/04/04
83/04/05
83/04/06
83/04/25
83/04/26
83/04/27
83/06/13
83/06/14
83/06/15
83/07/05
83/07/06
83/07/07
83/08/15
83/08/16
33/03/17
83/09/06
83/09/07
83/09/08
83/10/17
83/10/18
83/10/19
83/10/31
83/11/01
83/11/02
33/12/05
83/12/06
:3": ••• \ -?/n7 .
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
"< H3-R
007
009
Oil
019
021
023
041
043
045
057
059
061
067
069
071
091
093
095
103
105
101
123
125
127
138
140
142
173
175
177
187
139
191
21 1
213
2 i 5
0.16
0.29
0.42 •
0.50
0.59
0.41
0.28
0.27
0.33
0.35
0.33
0.24
0.20
0.21
0.23
0.49
0.57
0.58
0.48
0.33
0.53
0.30
0.56
0.88
0.23
0.21
0.30
0.37
0.28
0.31
0.40
0.39
0.36
0.41
0 ,49
0 .35
0.16
0.21
0.37
0.47
0.66
0.38
0.32
0.32
0.17
0.36
0.47
0.37
0.23
0.2.
0.21
0.65
0.62
0.45
0.37
0.34
0.43
0.49
0.65
0.47
0.33
0.30
0.44
0.50
0.45
0.49
0.51
0.57
0.56
0.55
0.53
0 .32
0.32
0.49
0.80
0.97
1 .25
0.79
0.60
0.59
0.50
0.71
0.80
0.61
0.43
0.42
0.45
1 .14
1 .19
1 .03
0.85
0.67
1 .01
0.79
1 .21
1 .35
0.55
0.51
0.74
0.37
>. (tfTs^p.
0.30
0.92
0.95
0.92
0.96
1 .02
0,63
1 ,147
1 ,834
1 ,754
1 ,968
2,073
2,074
1 ,841
2,032
2,048
2,030
2,398
2,542
1 ,557
1 ,656
1 ,597
1 ,863
1 ,946
1 ,991
1 ,823
1 ,795
1 ,333
1 ,823
1 ,820
1 ,736
1,796
1,317
1 ,783
1 ,985
731 ,989
1 ,991
2,008
2,209
2,133
2,207
2,243
2.. 0 1 4
1 ,144
1 ,854
1 ,789
1 ,926
2,016
1,957
1 ,763
1 ,934
2,047
2,024
2,355
2,484
1 ,508
1 ,612
1 ,543
1 ,769
1 ,361
1 ,379
1 ,686
1 , 687
1 ,694
1 ,725
1 .758
1 ,647
1 ,667
1 ,679
1 .658
1 ,916
1 ,897
1 ,895
1 ,990
2, 143
2,074
2,129
2,123 .
. 2,003
103.3
100.6
95.0
104.6
104.2
102.4
108.1
98.6
100.7
101 .5
107.1
104.7
102.3
101 .4
104.7
103.3
96.4
103.2
106.5
104.2
103.2
95.0
90.6
107.7
102.5
102,2
96.6
104.5
94.3
103.3
102.1
98.0
100.3
101.9
105.5
96.4
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
103-H ROOF MONITOR RESULTS 1983
DATE
LOG
TEST
P F LB
G F LB
T F LB
ACFM
SCFM
'/. ISO
CO
83/01/17
33/01/13
83/01/19
83/01/31
83/02/01
83/02/02
83/03/08
83/03/09
83/03/10
83/04/04
83/04/05
33/04/06
83/04/25
33/04/26
£3/04/27
33/06/13
33/06/14
83/06/15
83/07/05
33/07/06
33/07/07
33/08/15
33/08/16
33/08/17
83/09/06
S 3/0 9/0 7
33/09/03
33/10/17
83/10/13
33/10/19
83/10/31
33/1 1/01
33/11/02
33/12/05
83/12/06
83/1 2/^7
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
1Q3-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
008
010
012
020
022
024
042
044
046
053
060
062
068
070
072
092
094
096
104
106
102
124
126
123
139
141
143
174
176
173
133
190
192
212
214
216
0.15
0.34
0.17
0.27
0.47
0.36
1 .13
0.56
0.31
0.31
0.28
0.32
0.33
0.41
0.46
0.30
0.47
0.35
0.51
0.46
0.40
0.26
0.31
0.25
0.30
0.30
0.35
0.17
0.35
0.33
0.31
0.33
0.24
0.27
0.33
0.30
0.13
0.17
0.10
0.56
0.60
0.34
0.54
0.47
0.32
0.32
0.49
0.45
0.40
0.53
0.55
0.48
0.47
0.51
1 0.57
0.59
0.40
0.42
0.55
0.46
0.59
0 .57
C .61
0.33
0.46
0.33
0.40
0.35
0.37
0.43
0.64
0 .29
0.33
0.51
0.27
0.33
1 .03
0.70
1 .67
1 .03
0.62
0.62
0.77
0.77
0.73
0.94
1 .01
0.78
0 .94
0 . 36
1 .03
1 .05
0.79
0.63
0.36
0.71
0.39
0 .37
C .96
0.55
0.30
0.71
0 .71
0.68
0.61
0.76
0 .97
0.53
1 ,152
1 ,834
1 .617
1 ,890
2,043
2,034
2,424
2,406
2,421
2,421
2,710
2,679
2,434
2,191
2,061
2,261
2,211
"•> 1 i c
«. , i i -•
2,046
2,090
1 , 937
2,032
2,215
2,265
2,010
2,074
2,036
1 ,915
2, 161
1 ,330
2,048
1 ,919
1 ,936
1 ,63i
1 ,580
1 ,736
1 ,148
1 ,364
1 ,676
1 ,332
1 ,954
1 ,971
2,312
2,256
2,361
2,361
2,610
2,527
2,342
2,085
1 , 970
2,149
2,063
1 ,967 .
1 ,391
1 , 922
1 ,816
1,912
2,077
2,111
1 . 33?
1 ,335
1 .930
1 ,333
2,063
1 ,795
1 ,994
1 ,866
1 ,336
1 ,607
1 , 530
1,71 9
135.6
94.0
92.3
103.9
104.4
100.2
108.8
99.9
98.6
98.6
100.3
97.2
104.1
108.3
103.4
99.5
105.7
1 10.0
105.3
109.3
120.0
109.9
91 .7
98.2
106.8
100.6
102.6
107.5
95.9
106.2
101 .3
103.5
107.7
104.4
107.3
94.0
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
101-G ROOF MONITOR RESULTS 1984
I
I—«
^
DATE
LOG
TEST**
P F LB
G F LB T F LB
ACFM
SCFM
ISO
34/01/09
34/01/10
84/01/11
84/02/06
84/02/07
84/02/08
84/02/27
84/02/28
84/02/29
84/04/09
84/04/10
34/04/11
84/04/30
34/05/01
84/05/02
34/06/11
84/06/12
34/06/13
84/07/02
34/07/03
84/07/04
34/08/13
84/08/14
34/08/15
84/09/04
34/09/05
84/09/06
34/10/15
84/10/17
84/10/29
84/10/30
34/10/31
34/12/10
84/12/1 1,
3 4/1 2. '-12
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
1 0 1 -G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
! 0 1 -G
\ o :, -3
218
220
222
243
247
249
251
253
255
269
271
273
275
277
279
237
239
291
306
308
310
324
326
328
330
332
334
353
355
357
371
373
375
399
401
403
0.28
0.21
0.39
0.46
0 .36
0.43
0.50
0.45
0.35
0.55
0.39
0 .46
0.25
0.30
0.40
0.40
0.43
0.24
0.22
0.23
0 .20
0.62
0.33
0.39
0.35
0 .37
0 .3?
0 .53
0 .45
0 .40
0 .39
0 .36
0 .33
0 .18
0 .33
0 -27
0.59
0.48
0.45
0.41
0.29
0,37
0.43
0.41
0.36
0.43
0.36
0.38
0.39
• 0.33
0.43
0.34
0.31
0.37
0 .62
0.34
0.40
0.62
0.29
0 .47
0.27
0.49
0.32
0.50
0.56
0.53
0.41
0.42
0.42
0 .30
0 .26
e - 23
0.8?
0.69
0.34
0.86
0.65
0 .80,
0.92
0.35
0.71.
0.98
0.76
0.34
2,000
1 ,933
2,290
2,137
1 ,945
1 ,777
1 ,997
2,136
1 ,616
1 ,786
2,244
1 ,689
0.64\ 1,896
0.631
0.83
0.74
0.74
0.61
0.84
0.57
0 .60,
1 .23
0.62
— Q . 36,
0.6l
0.87
0.71
1 .04
1 .01
0.92
2,009
1 ,741
1 .980
1 ,953
1 ,359
2,077
2,092
2,130
2,037
2,073
2,004
2,047
2,153
2,177
2.273
2.040
2', 136
0.31"] 2,23C
0.73 2,075
0.74J 2,078
0 . 43n
0.59
G . 5*
1 ,415
1 ,461
1 ,389
*
1 ,955
1 ,894
2,267
2,130
1 , 966
1 ,790
1,930
2,111
1 ,622
1 ,723
2,210
1 ,644
1 .787
1 ,926
1 ,645
1 .852
1 ,312
1 ,723
1 ,940
2,004
1 ,972
1 ,876
1 .398
1 ,345
1.911
2,038
2,069
2,157
1 .924
2,004
2,077
1 ,974
1 , 968
1 ,386
1 , 429
1 ,861
97.5
109.0
36.2
93.2
93.2
101 .3
109.1
101 .2
103.5
106.8
97.3
104.5
103. 1
100.0
105.3
101 .9
102.7
112.5
108.7
109.5
103.8
107.3
107.1
107.6
1 14.9
99.8
100.7
102.4
1 10 .4
106.9
107.4
105.3
108.2
104.7
105.3
99. 1
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
101-H ROOF MONITOR RESULTS 1984
DATE
LOG
TEST*
P F LB
G F LB T F LB
ACFM
SCFM
7. ISO
84/01/09
34/01/10
34/01/11'
84/02/06
84/02/07
84/02/08
84/02/27
84/02/28
84/02/29
84/04/09
84/04/10
34/04/11
84/04/30
34/05/01
34/05/02
34/06/1 1
34/06/12
34/06/13
84/07/02
84/07/03
84/07/04
34/03/13
34/03/14
34/08/15
34/09/04
84/09/05
34/09/06
34/10/15
84/10/17
84/10/29
84/10/30
34/10/31
34/12/10
&&/?.?./'. 1
S <•-.-' '" ' 2
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
1 3 1 -ri
101-H
101-H
101-H
101-H
101-H
101-H
101-H
* 0 A -H
; 7 . -k'
219
221
223
246
248
250
252
254
256
270
272
274
276
273
280
288
290
292
307
309
311
325
327
329
331
333
335
354
356
358
372
374
376
400
402
4Qd
0.29
0.25
0.47
0.73
0.38
0.25
0.22
0.43
0.43
0.34
0.40
0.26
0.28
0.34
0.22
0.27
0 .66
0.29
0.30
0.21
0.28
0.42
0.41
0.50
0.37
0.46
0.25
0.37
0.48
0.32
0.30
0.39
0.42
0.30
0 .41
0,60
0.46
0.57
0.65
0.43
0.40
0.26
0.47
0.53
0.34
0.44
0.31
0.38
0.43
0.59
0.37
0.42
0.45
0.42
0.39
0.53
0.53
0.49
0.58
0.62
0.48
0.56
0.24
0.74
0.65
0.34
0.73
0.45
0.72
0.42
0.52
0.67
0.74
0.32
1.12
1 .16
0.79
0.51
0.68
0.96
0.77
0.78
1 .21
0.64
0.77
0.93
0.59
0.70
1.11
0.71
0.70\
0.73
0.81-1
0.9H
0.99
1.12)
0.82)
1 .02
0 . 49J
1 .11)
1 .13
0 . 67j
1 .03"
0.34
1 .14,
0.72
0 .93
' O"7
A . £* r '
1 ,647
1 ,421
1 ,730
1 ,347
1 ,745
1 ,629
1 ,628
1 ,574
2,030
1 ,303
2,063
t ,752
1 ,797
1 ,361
1 ,319
1 ,734
1 .949
2,025
1 ,739
1 , 769
1 ,757
2,133
1 ,931
2.234
2,077
2,147
2, 195
1 ,998
1 ,781
1 ,39^
1 , 995
2,023
2 . 349
1 ,477
i , 688
1 ,948
1 .616
1 ,371
1 ,672
1 ,319
1 .763
1 ,650
1 ,571
1 ,554
2,027
1 ,746
2,017
1 ,725
1 ,732
1 ,749
1 ,764
1 ,642
1 ,377
1 ,908
1 ,656
1 ,637
1 ,632
• 1 ,947
1 .794
2,034
1 ,917
2,103
2,036
1 ,397
1 ,668
1 ,776
"\ 1 , 852
! 1,391
J 2,210
1 1 , 425
1 1 , 635
j 1,901
84. 1
113.5
93.5
117.4
104.5
95.3
103.4
105.7
95.2
113.3
100.4
108.0
109.3
93.1
103.4
102.3
106.4
99.0
102.6
102.9
107.1
99.9
106.3
101 .6
102.6
111.4
105.7
109.1
110.2
109.4
105.3
99.9
100.3
102.3
1 12.3
104.3
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
103-6 ROOF MONITOR RESULTS 1984
DATE
LOG
TEST**
P F LB G F LB
T F LB
ACFM
SCFM
'/. ISO
84/01/16
84/01/17
84/01/18
84/01/30
84/01/31
84/02/01
84/03/05
84/03/06
84/03/07
84/04/02
84/04/03
84/04/04
84/05/14
84/05/15
84/05/16
34/06/05
84/06/06
34/06/07
84/07/16
84/07/17
84/07/18
84/08/07
84/08/08
84/03/09
84/09/10
34/09/1 1
34/09/12
34/10/01
34/10/02
84/10/03
84/11/05
84/1 1/06
84/11/07
34/12/03
34/12/04
34/12/05
103-Q
103-Q
103-0
103-0
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
224
226
228
233
235
237
257
259
261
263
265
267
281
233
235
293
295
297
312
314
316
313
320
322
336
333
340
347
349
351
377
379
331
393
39=
397
0.25
0.34
0.24
0.28
0.25
0.25
0.37
0.34
0.33
0.50
0.47
0.34
0.23
0.43
0.45
0.42
0.42
0.50
0.41
0.52
0.40
0.37
0.34
0.29
0.24
0.33
0.29
0.45
0.33
0.26
0.15
0.32
0.47
0.50
J .44
0 .411
0.33
0.52
0.46
0.29
0.23
0.30
0.39
0.23
0.43
0.44
0.63
0.40
0.37
0.46
0.51
0 .51
0.52
0.48
0.39
0.46
0.63
0.34
0.40
0.45
G.47
0.37
0.30
0.31
0.23
0.23
0.34
0.43
0.26
0.52
.
0 .59
0.63
0.36
0.70
0.57
0.53
0.55
0.75
0.62
0.74
0.94
1 .10
0.74
0.61
0.38
0.96
0.94
0.94
0.98
0.30
0.98
1 .08
0.71
0.74
0.74
0.71
0.70
0 .59
0.76
0.56
0.49
0.49
0.73
0.72
1 .01
0 . 93
1 .00
1 ,603
1 ,456
2,024
1 ,683
1,733
1,573
1 ,460
1 ,504
1 ,574
2,195
2,026
1 ,711
1 ,723
2,039
1 .634
1 ,915
2,029
1 ,974
2,189
2,032
2,073
1 , 942
1 ,696
1 ,776
2,175
2,402
2,344
2,727
2,228
2.344
2,292
2,076
2,240
2,303
2,367
1 ,997
1 ,581
1 ,445
1 ,977
1 ,666
1 ,757
1 ,598
1 ,383
1 ,414
1 ,539
2,162
1 ,956
1 ,620
1 ,607
1 ,91 1
1 ,634
1 ,762
1 ,873
1 ,861 ...
2,025
1 ,966
1 ,923
1 ,792
1 ,571
1 ,622
2,028
2,234
2,170
2,624
2,171
2.231
2,209
2,029
2,228
2,241
2.339
1 , ?*3
105.1
102.8
108.0
104.2
93.9
95.5
107.8
113.9
101 .1
95.5
105.2
103.4
105.4
96.7
103.7
103.1
91 .5
100.3
99.9
101 .9
101 .0
99.1
107.5
102.1
103.3
107.9
102.3
105.2
108.6
102.3
98.1
109.1
103.7
105.3
102.1
106. 1
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
DATE
LOG
103-H ROOF MONITOR RESULTS 1984
TEST** P F LB G F LB T F LB ACFM
3CFM
'/. ISO
34/01/16
84/01/17
84/0 1/rB
84/01/30
84/01/31
84/02/01
84/03/05
84/03/06
84/03/07
84/04/02
84/04/03
84/04/04
34/05/14^
84/05/15
84/05/16J
34/06/05]
84/06/06
34/06/OTj
84/07/16
34/07/17
34/07/18
84/08/07
34/03/08
34/08/09
34/09/10
34/09/1 1
84/09/12
34/10/01
84/10/02
34/10/03
84/1 1/05
34/1 1/06
84/1 1/07
34/12/03
84/12/04
84/12/05
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
225
227
229
234
236
238
258
260
262
264
266
263
282
234
286
294
296
298
313
315
317
319
321
323
337
339
341
348
350
352
373
330
382
394
396
398
0.33
0.27
0.43
0.31
0.24
0.21
0.26
0.33
0.50
0.31
0.23
0.54
C.52
0.47
.0.39
0.37
0.38
0.21
0.36
0.36
0.38
0.49
0.33
0.41
0 .30
0.43
0.35
0.46
0 .28
0.27
0.49
0.41
0.42
0 .23
0 .37
0 .37
0.30
0.34
0.39
0.46
0.30
0.27
0.37
0.51
0.62
0.40
0.55
0.32
0.53
0.39
0.39
0.53
0.50
0.46
0.50
0.60
0.46
0.96
0.94
0.76
0.37
0.59
0.57
0.52
0.38
0.32
0 .53
0.28
0.34
0.57
0.38
0 .20
0.64
0.61
0.82
0.77
0.54
0.48
0.63
0.84
1 .12
0.71
0.33
1 .37
1.10
0.36
0.79
0 .91
0.38
0 .63
0.36
0.96
0.84
1 .45
1 .32
1 .17
0.67
1 .02
0.92
0.98
0.67
0.59
1 .02
0 .69
0.76
0.30
0.75
0 .57
1 ,491
1 ,543
1 ,465
1 ,502
1 ,470
1 ,231
1 ,964
1 ,913
1 ,388
1 ,893
1 ,787
1 ,356
2,046
1 ,396
1 ,733
2,006
2,161
1 ,634
1 , 975
2,111
1 ,872
2,160
1 ,358
1 ,393
2,238
2 , 349
2,253
2_415
2,322
2,307
2,031
2, 123
2,243
2,374
2,480
1 , 633
1 ,478
1 ,513
1 ,424
1 ,459
1 ,453
1 ,242
1 ,882
1 ,313
1 ,873
1 ,355
1 ,749
1 ,749
1 ,930
1 ,751
1 ,761
1 ,398
1 ,996
1 ,5i4
1 ,826
1,919
1 ,729
1 , 972
1 ,632
1 ,708
2,035
2,205
2,051
2,301
2,250
2,214
1 ,923
2,063
2,231
2,305
2,454
1 , 532
99.8
1 10 .0
105.3
106. 1
1 13.1
112. 1
100.6
95.9
105.5
93.0
102.6
100 .0
102.3
93.0
103.3
107
92.2
109. 6
113.1
111.6
111.9
98.9
101 . 1
95.3
110.0
113.0
105.6
107.7
111.6
105.5
100 .3
103.2
99.0
104.3
107.9
106.9
-------
DATE
Table A-l. ALUMAX ROOF MONITOR DATA.
LOG
101-G ROOF MONITOR RESULTS 1985
TEST** P F LB G F LB T F LB ACFM
SCFM
ISC
85/01/07
35/01/08
35/01/09
35/02/11
85/0 2/ 12
35/02/13
35/03/04
85/03/05
85/03/06
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
405
407
409
426
428
430
435
437
439
0.420
0.450
0.250
0.370
0.410
0.140
0.203
0.209
0.378
0.2500
0.2000
0.2200
0.6700
0.4300
0.3100
0.3580
0.3720
0.3000
0.670
0.650
0.470
1 .040
0.830
0.450
0.566
0.530
0.679
1 ,567.0
1 ,543.0
1 ,548.0
1 ,320.0
1 ,474.0
1 ,440.0
2,251 .0
1 ,556.0
1 ,607.8
1 ,545.0
1 ,525.0
1 ,536.0
1 ,763.0
1 ,450.0
1 ,420.0
2,165.0
1 ,485.0
1 ,581 .7
103.4
106.0
107.0
106.6
1 12.7
105.9
104.2
112.7
109.0
I
I—>
CO
DATE
LOG
101-H ROOF MONITOR RESULTS 1985
TEST**
P F LB
G F LS
T F LB
ACFM
SCFM
Y. ISO
85/01/07
35/01/08
85/01/09
85/02/11
85/02/12
85/02/13
35/03/04
85/03/05
85/03/06
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
406
408
410
427
429
431
436
433
440
0.370
0.540
0.700
0.370
0.160
0.090
0.582
0.303
0.377
0.400
0.460
0.710
0.360
0.300
0.330
0.510
0.612
0.193
0.770
0.990
1 .410
0.730
0.960
0.420
1 .093
0.920
0.570
1 ,484.0
1 , 857 . 0
1 ,817.0
976.0
1 ,564.0
1 ,387.0
1 ,655.0
1 ,511 .6
1 ,946.1
1 ,436.0
1 ,311 .0
1 ,788.0
955.0
1 ,529.0
1 ,363.0
1 ,603.0
1 ,433.2
1 ,921 .5
100.2
104.3
94.8
102.4
106.3
105.5
109.2
103.2
107.6
-------
DATE
.oc
Table A-l. ALUMAX ROOF MONITOR DATA.
103-G ROOF MONITOR RESULTS 1935
TEST** P F LB G F LB T F LB ACFM
SCFM
ISO
85/01/14
35/01/15
85/01/16
85/02/04
85/02/05
85/02/06
103-G
103-G
103-G
103-G
103-G
103-G
411
413
415
420
422
424
0.16
0.17
0.27
0.41
0.20
0.23
0.14
0. 13
0.32
0.42
0.41
0.49
0 .30
0.35
0.58
0.82
0.61
0.72
1 ,630
1 ,532
1 ,361
2,304
2,049
2,077
1 ,602
1 ,563
1 ,354
2,277
2,030
1 ,983
103.2
103.1
107.5
102.3
103.6
108.3
DATE
LOC
103-H ROOF MONITOR RESULTS 1985
TESTS P F LB G F LB T F LB ACFM
SCFM
ISO
35/01/14
85/01/15
85/01/16
85/02/04
85/02/05
35/02/06
103-H
103-H
103-H
103-H
103-H
103-H
412
414
416
421
423
425
0.25
0.23
0.15
0.41
0.26
0.22
0.19
0.26
0.13
0.33
0 .37
0.33
0.45
0.43
0.33
0.73
0.63
0.55
1 ,079
1 ,265
1 ,202
2,430
1 ,526
1 ,463
1 ,065
1 ,254
1 ,209
2,451
1 ,531
1 ,435
103.4
106.3
94.0
100.1
112.3
103.4
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
101-G ROOF MONITOR RESULTS 1985
DATE
LOG
TEST**
P F LB
G F LB T F LB
ACFM
SCFM
V. ISO
ro
o
85/01/07
85/01/08
85/01/09
85/02/11
85/02/12
85/02/13
85/03/04
85/03/05
85/03/06
85/04/16
85/04/17
85/04/18
85/05/06
85/05/07
85/05/08
85/06/10
85/06/11
85/06/12
85/07/08
85/07/09
85/07/10
85/08/12
85/08/13
85/08/14
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
101-G
405
407
409
426
428
430
435
437
439
455
457
459
461
463
465
485
487
489
497
499
501
515
517
519
0.420
0.450
0.250
0.370
0.410
0.140
0.210
0.210
0 . 38 C
0.270
0.290
0.240
0.320
0.430
0.190
0.200
0.540
0.300
0.284
0.311
0.343
0.636
0.385
0.426
0.250
0.200
0.220
0.670
0.430
0.310
0.360
0.370
0.300
0.210
0.390
0.390
0.380
0.390
0.200
0.410
0.360
0.340
0.419
0.321
0.316
0.314
0.459
0.542
0.670
0.650
0.470
1 .040
0.830
0.450
0.570
0 .580
0 .680.
0.480
0.680
0.630
0.700
0.820
0.390
0 .610
0.900
0.640
0.703
0.632
0.660
0 . 950
0.843
0.968
1 ,567
1 ,543
1 ,548
1 ,820
1 ,474
1 ,440
2,251
1 ,556
1 ,607
1 ,537
1 ,644
1 ,404
1 ,754
2,088
1 ,902
,735
,515
,543
,629
,420
1 ,521
1 ,795
2,283
2,231
1 ,545
1 ,525
1 ,536
1 ,763
1 ,450
1 ,420
2,165
1 ,485
1 ,581
1 ,472
1 ,572
1 , 354
1 ,680
1 ,972
1 ,809
1 ,591
1 ,405
1 ,438
1 ,513
1 ,307
1 , -533
1 ,666
2, 138
2,093
108.4
106.0
107.0
106.6
112.7
105.9
104.2
112.7
109.0
104.7
101 .2
103. 1
103.2
101 .7
106.2
1 15.4
111.6
105.8
107.0
99.1
102.3
96.9
96.6
96.0
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
101-H ROOF MONITOR RESULTS 1985
I
ro
DATE
LOG
TEST**
P F L8
G F LB T F LB
ACFM
SCFM
y. iso
85/01/07
85/01/08
85/01/09
85/02/1 1
85/02/12
85/02/13
85/03/04
85/03/05
85/03/06
85/04/16
85/04/17
85/04/18
85/05/06
85/05/07
85/05/08
85/06/24
85/06/25
85/06/26
85/07/08
85/07/09
85/07/10
85/08/12
85/08/13
85/08/14
101-H
101 -H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
101-H
406
408
410
427
429
431
436
438
440
456
458
460
462
464
466
491
493
495
498
500
502
516
518
520
0.370
0.540
0.700
0.370
0.160
0.090
0.580
0.310
0.380
0.400
0.260
0.240
0.340
0.550
0.460
0.428
0.338
0.436
0.631
0.483
0.31 1
0.331
0.275
0.363
0.400
0 .460
0.710
0.360
0.800
0.330
0.510
0.610
0.190
0.450
0.430
0.340
0.530
0.630
0.640
0.584
0.440
0.556
0.643
0.523
0.484
0.420
0.425
0.307
0.770
0 . 990
1 .410
0.730
0.960
0.420
1 .090
0.920
0.570
0.840
0 .680
0.580
0.870
1 .180
1 .090
1 .012
0.778
0.993
1 .274
1 .007
0.795
0.751
0.700
0.671
1 ,484
1 , 357
1 ,817
976
1 ,564
1 ,387
1 ,655
1 ,511
1 ,946
1 ,664
1 ,488
1 ,628
1 ,695
2,225
2,096
2,280
2,019
1 ,847
2,290
2,193
1 ,948
1 ,944
1 ,858
1 ,936
1 ,436
1 ,81 1
1 ,738
955
1 ,529
1 ,363
1 ,603
1 ,438
1 ,921
1 ,572
1 ,419
1 ,573
1 ,604
2,072
1 ,995
2,116
1 ,893
1 ,696
2,104
2,000
1 ,760
1 ,784
1 ,713
1 ,788
100.2
104.3
94 . 8
102.4
106.8
105.5
109.2
108.2
107.6
111.2
111.0
115.1
110.8
111.9
112.7
99.8
109.9
106.5
107.4
110.6
114.8
102.0
102.8
106.0
-------
Table A-l. ALUMAX ROOF MONITOR DATA.
ro
ro
DATE
103-G ROOF MONITOR RESULTS 1985
LOG
TEST*
P F LB
G F LB T F LB
ACFM
SCFM
'/. ISO
85/01/14
85/01/15
85/01/16
85/02/04
85/02/05
85/02/06
85/03/11
85/03/12
85/03/13
85/04/01
85/04/02
85/04/03
85/05/13
85/05/14
85/05/15
85/06/03
85/06/04
85/06/05
85/07/15
85/07/16
85/08/05
85/08/06
85/08/07
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
103-G
411
413
415
420
422
424
441
443
445
449
451
453
467
469
471
479
481
483
503
505
509
511
513
0. 160
0.170
0.270
0.410
0.200
0.230
0.460
0.270
0.290
0.360
0.400
0.260
0.690
0.600
0.470
0.370
0.460
0.440
0.408
0.298
0.436
0.629
0.350
0.140
0.180
0.320
0.420
0.410
0.490
0.380
0.480
0.320
0.150
0.490
0.260
0.520
0.440
0.670
0.550
0.480
0.410
0.446
0.730
0.437
0.346
0.573
0.300
0.350
0.580
0.820
0.610
0.720
0.830
0.750
0.610
0.510
0.890
0.510
1 .210
1 .040
1 .140
0.930
0.940
0.850
0.854
1 .028
0.873
0.975
0.923
1 ,630
1 ,532
1 ,361
2,304
2,049
2,077
1 ,975
1 ,707
1 ,625
1,582
1 ,508
1 ,633
1,737
1,956
2,424
2,023
1 ,963
2,176
1,559
1 ,624
1 ,656
1 ,837
2,054
1 ,602
1 ,563
1 ,354
2,277
2,030
1 ,983
1 ,940
1 ,635
1 ,580
1 ,538
1 ,481
1 ,603
1 ,625
1 ,819
2,257
1 ,846
1 ,791
2,007
1 ,440
1 ,491
1 ,565
1 ,726
i ,918
103.2
108.1
107.5
102.3
103.6
108.8
94.9
109.3
98.4
108.8
89.9
105.3
106.0
97.3
98.5
107.7
110.1
106.3
112.8
105.2
101 .3
102.6
iOG.3
-------
Table A-Il. ALUMAX ROOF MONITOR DATA.
CO
DATE
103-H ROOF MONITOR RESULTS 1985
LOG
TEST*
P F LB G F LB T F LB
ACFM
SCFM
'/. ISO
85/01/14
85/01/15
85/01/16
85/02/04
85/02/05
85/02/04
85/03/11
85/03/12
85/03/13
85/04/01
85/04/02
85/04/03
85/05/13
85/05/14
85/05/15
85/06/24
85/06/25
85/06/26
85/07/15
85/07/16
85/08/05
85/08/06
85/08/07
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
103-H
I03-H
103-H
103-H
412
414
416
421
423
425
442
444
446
450
452
454
468
470
472
492
494
496
504
506
510
512
514
0.250
0.230
0.150
0.410
0.260
0.220
0.510
-0 .490
0.510
0.390
0.330
0.440
0.140
0.430
0.440
0.501
0.323
0.355
0.220
0.520
0.454
0 .432
0 .340
0.1900
0.2600
0.1800
0.3300
0.3700
0.3300
0.3100
0.6500
0.3300
0.4400
0.3700
0.4200
0.5600
0.5600
0.5800
0.6150
0.4190
0.6440
0.5680
0.5611
0.4290
0.5600
0.5020
0.450
0.480
0 .330
0.730
0 . 530
0.550
0.820
1 .140
0.830
0.820
0.700
0.860
0.700
0.990
1 .020
1.116
0.742
0.998
0.788
1 .081
0.883
1 .040
0.842
1 ,079
1 ,265
1 ,202
2,430
1 ,526
1 ,468
1 ,803
1 ,646
1 ,576
1 ,695
2,523
1 ,513
2,252
2,222
2,208
2,440
1 ,924
1 ,790
2,144
2,234
2,348
2,210
2,242
1 ,065
1 ,254
1 ,209
2,451
1 ,531
1 ,435
1 ,761
1 ,552
1 ,520
1 ,677
2,550
1 ,480
2,104
2,054
2.046
2,257
1 ,762
1 ,622
1 ,964
2,042
2, 198
2,063
2,070
103.4
106.3
94.0
100 . 1
112:8
103.4
95.3
104. 1
96 . 4
104.0
98.0
106.2
104.5
102.8
111.2
96.7
102.0
97.4
96.2
108.4
102.4
98.5
104.0
-------
TABLE
EMISSIONS SUMMARY 161 EAST.
DATE
81/01/19
81/01/20
81/02/17
81/03/10
81/03/11
81/04/06
81/05/12
81/05/13
81/06/08
81/07/07
81/07/08
81/08/11
81/08/12
81/09/01
81/09/02
81/10/05
81/10/06
81/11/10
81/12/01
81/12/02
82/01/12
82/01/13
82/02/03
LOCATION
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
TEST**
015
016
017
020
021
022
007
008
009
013
014
015
040
041
042
065
066
067
092
093
094
119
120
121
135
137
138
154
156
157
180
178
179
191
192
193
219
220
221
225
226
227
AL/P/D
2,932.7
2,932.7
2,935.0
2,913.4
2.913.4
2,913.4
2,920.0
2,924.3
2,924.3
2,952.8
2,952.8
2,952.8
2,876.4
2,840.9
2,840.9
2,957.1
2,957.1
2,957.1
2,980.8
2,980.8
2,980.2
2,863.6
2,863.6
2,360 .0
2.860.2
2.828.5
2,822.5
2,869.0
2,870.4
2,870.4
2,882.9
2,882.9
2,882.9
3,046.6
3,045.3
3,045.3
3,018.0
3,018.1
3,015.8
3,036.1
3,036.1
3,038.0
7. ISO
106.4
101.3
107.1
98.8
105.3
97.1
94.7
94.1
99.4
96.7
104.1
104.4
105.6
91 .5
103.3
112.4
98.2
103.2
98.6
108.7
94.0
102.3
92.4
93.8
114.6
97.9
115.9
82.7
105.7
87.0
96.1
100.7
102.2
100.9
98.3
101.2
98.5
96.4
105.4
T F LB/T
0.0181
0.024/
0 . 0 1 8J
0.029]
0.027
0.04£
0.014]
0.020|
0.018J
0.030"
0.057
0.020
0.035"
0.155
0.056J
0.093"
0.058
0.07Q,
0.074A
0.058]
O.OS5J
0. 107^
0.067J
0 .06?]
0.079n
0.063
0.064^j
0.047]
0.0381
0.040J
0.026]
0.044
0.040J
0.07ll
0.079
0.059J
0.017
0.020
O.Oli
0.0481
0.034
0.037..
-------
TABLE A-2. DRY SCRUBBER
EMISSIONS SUMMARY 161 EAST.
DATE
82/03/09
82/03/10
82/04/06
82/04/07
82/04/27
82/04/28
83/09/22
83/09/23
84/10/08
84/10/09
84/10/12
LOCATI
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
161-E
ON TEST**
243
244
246
257
258
259
275
276
277
158
159
160
365
366
367
AVERAGE :
COUNT: 57
AL/P/D
2,968.8
2,968.8
2,962.7
3,029.0
3,030.4
3,030.4
2,997.9
2,997.9
3,000.5
3,050.4
3,050.4
3,055.8
3,097.2
3,091 .5
3,094.1
2,958.2
X ISO
101 .7
96.8
100.8
100.1
108.3
101 .1
117.4
108.4
105.7
106.8
103.4
102.6
108.4
110.1
116.1
101 .9
T F LB/T
0.013^
0.015
0.01?J
0.032"
0 .022
0 .037,
0.04«Tl
0.049
0.032J
0.06H
0 .045
0.055-
0.020"^
0.018
0.1 33J
0.046
A-25
-------
TABLE A-2.
DRY 3c£u#
EMISSIONS SUMMARY 161 WEST.
DATE
81/01/12\
)
81/01/1§X
8 1/0 2/0 2^
81/02/03J
81/03/09
81/03/31
81/04/OJ
81/05/11
81/06/09
81/07/08
81/07/09
81/08/10
81/08/31
81/10/07
81/10/08
81/11/09
81/11/30
82/01/11
82/02/01
LOCATION
161-W
161-W
161-W
161-W
161-W
161-W
161-W
161-W
161-W
161-W
161-W
161-W
161-W
161-W
161-W
161-W
161-W
161-W
161-W
TEST*
009
010
Oil
023
024
025
005
006
004
010
Oil
012
037
038
039
068
069
070
095
096
097
116
117
118
132
133
134
158
159
160
177
175
176
188
189
190
216
217
218
222
223
224
AL/P/D
2,929.4
2,994.4
2,911 .8
2,985.8
2,982.9
2,982.9
2,917.8
2,917.8
2,917.3
2,956.5
2,956.5
2,956.5
2,864.8
2,864.8
2 , 864 . 8
2,980.0
2,980.0
2,980.0
2,980.8
2,975.5
2,975.5
2,874.8
2,874.8
2,874.8
2,828.3
2,828.3
2,828.3
2,870.2
2,870.2
2,877.9
2,887.3
2,887.3
2,887.3
3,048.9
3,048.9
3,048.9
3,017.4
3,017.4
3,017.4
3,032.1
3,032.1
3,032.1
A-26
y. iso
90.4
107.8
94.8
100.4
105.1
101 .0
108.8
102.3
96.4
93.9
102.6
100.7
101 .7
100.7
94.1
114.5
109.4
110.0
105.2
112.7
102.6
107.3
89.8
89.4
112.5
99.5
99.0
86.6
98.1
87.9
104.9
102.7
96.1
91 .8
91 .5
100.9
91 .0
103.7
91 .3
T F LB/T
0.028s)
0.032
0.1147
0.021\
0.018
0.021J
0.2051
0.023
0.028,
0.021]
0.020 1
0.015/
0.052~1
0.370
0.035^
0 . 364"
0.181
0.243-
0.035")
0.100
0.083>
0.070^
0.081
0.070-
0.070
0.107
0.085_
0.087]
0.039
0.040
0.057]
0.039
0.06pJ
0.047"1
0.039
0.037
0.058
0.096
0.072
0.273
0.172
0.165
-------
Table A-2. DRY SCRUBBER
EMISSIONS SUMMARY 161 WEST.
DATE LOCATION TEST* AL/P/D X I SO T F LB/T
82/03/08 161-W 240 2,931.7 88.8 0.020
241 2,931.7 98.2 0.024
242 2,931.7 114.4 0.019
82/04/05 161-W 254 3,026.6 104.1 0.041
255 3,026.6 105.2 0.051
256 3,026.6 113.3 0.063
82/04/26 161-W 272 2,996.1 112.2 0.086
273 2,996.1 104.0 0.212
274 2,996.1 109.5 0.111
83/09/20 161-W 155 3,055.7 106.0 0.225
156 3,055.7 119.2 0.093
33/09/21 161-W 157 3,054.9 110.9 0.087
84/10/22 161-W 368 3,096.1 103.2 0.050
369 3,096.1 107.0 0.073
370 3,096.1 112.1 0.058
AVERAGE: 2,962.7 101.9 0.087
COUNT: 57
A-27
-------
TABLE A-2. D/ZY
EMISSIONS SUMMARY 142 EAST.
DATE
81/02/23
81/04/20
81/05/26
81/04/02
81/06/03
81/07/28
81/07/29
81/08/04
81/08/11
81/09/09
81/09/10
81/10/14
81/10/15
81/11/04
81/11/05
81/12/08
81/12/09
82/01/04
82/02/09
82/03/02
82/03/03
82/04/13
82/04/14
LOCATION
162-E
162-E
162-E
142-E
162-E
142-E
142-E
142-E
142-E
142-E
142-E
142-E
142-E
142-E
142-E
142-E
162-E
142-E
142-E
142-E
142-E
142-E
162-E
TEST*
AL/P/D
X ISO
T F LB/T
025
023
024
019
020
021
053
054
055
042
043
044
107
108
109
110
111
112
142
143
144
144
145
144
172
173
174
197
198
199
213
214
215
231
232
233
237
238
239
243
264
265
2,990.8
2,990.8
2,990.8
2,920.9
2,920.9
2,920.9
2 , 957 . 9
2,957.9
2,957.9
2,971 .2
2,971 .2
2,971 .2
3,001 .7
3,001 .7
3,001 .6
2,997.1
2,997.1
2,997.1
2,838.5
2.844.2
2,844.2
3,012.0
3,012.2
3,012.2
2,946.6
2,946.6
2,945.7
3,026.8
3,026.8
3,023.2
2,936.4
2 , 996 . 4
2,996.4
2,987.4
2,987.4
2,987.4
2,956.7
2,956.7
2,960.8
3,017.9
3,022.2
3,022.2
A-28
102.0
103.4
102.9
104.7
97.9
101 .0
100.8
100.3
101 .0
110.1
106.9
109.3
108.1
103.9
107.5
114.3
108.3
96.6
108.3
100.9
98.1
96.8
95.5
99.4
90.0
100.9
91 .3
100.1
102.4
101 .8
97.1
99.3
96.5
101 .2
101 .8
100.2
109.6
97.2
102.6
108.5
101 .6
112.9
0.024
0.151
0.064
0.042
0.038
0.034
0.181
0.125
0.190
0.214"
0.151
0.122
0.109T
0.096
0.095J
0.3^8"
0.207
0.084"
0.210
0.209
0.187
0.1451
0.122
0.105.
0.140
0.095
0.110
o.osT
0.048
0.045.
0.052
0.058
0.070
0.030
0.084
0.141
0.0391
0.055
0.055_>
0.083
0.064
0.05?
-------
TABLE A-2.
EMISSIONS
DRY SCRUBBER
SUMMARY 162 EAST.
DATE
82/04/21
83/10/23
83/10/26
84/09/17
84/09/18
84/09/19
LOCATION
162-E
162-E
162-E
162-E
162-E
TEST*
269
270
271
182
183
134
359
360
361
AL/P/D
X ISO
2,998.9
2,998.9
2,997.5
3,070.3
3,070.3
3,075.8
3,077.5
3,080.4
3,085.1
104.7
113.4
104.5
108.6
110.8
109.4
110.3
107.1
113.4
T F LB/T
0.145
0.080
0.118
0.0481
0.046
0.032J
0.106
0.076
0.048
AVERAGE:
COUNT:
51
2,985.8
103.4
0.104
A-29
-------
TABLE A-2.
EMISSIONS SUMMARY 162 WEST.
DATE LOCATION
TEST**
AL/P/D
X ISO
T F LB/T
81/03/02
81/04/13
81/05/27
81/03/28
81/06/01
81/07/27
81/08/03
81/09/08
81/10/12
81/11/02
81/11/03
81/12/07
82/01/04
82/01/03
32/02/08
82/03/01
82/04/12
162-W
162-U
162-W
162-W
162-W
162-W
162-W
162-W
162-W
162-W
162-W
162-W
162-W
162-U
162-W
162-W
162-W
002
003
001
016
017
018
056
057
058
101
102
103
104
105
106
114
115
113
139
140
141
161
162
163
168
170
169
194
195
196
210
211
212
228
229
230
234
235
236
260
261
262
2,955.5
2,955.5
2,955.5
2,904.7
2,904.7
2,899.2
3,003.9
3,016.9
3,016.9
2,981 .7
2,981 .7
2,981 .7
3,004.4
3,004.4
3,004.4
2,995.0
2,991 .5
2,995.0
2,841 .7
2,841 .7
2,841.7
3,013.7
3,013.7
3,013.7
2,949.6
2,948.1
2,949.6
3,027.2
3,027.2
3,027.2
2,934.8
2,994.3
2,994.8
2,987.6
2,987.6
2,987.6
3,958.2
2,958.2
2,958.2
3,014.3
3,014.3
3,014.3
A-30
101 .4
102.1
101 .3
112.0
107.2
1 10.0
105.6
91 .8
100.9
106.9
110.0
104.7
107.4
98.2
109.7
104.8
93.8
103.9
106.6
90.6
100.8
90.8
97.6
90.4
96.5
90.0
96.7
102.3
101 .9
93.0
102.4
90.9
101 .4
86.3
98.4
99.7
106.0
101 .3
104.9
111.0
100.8
108.3'
0.021
0.016
0.025
0.033
0.034
0.101
0.084^
0.081
o . 108J
0.113
0.247
0.123
0.053
0.084
0.060
0.056
0.031
0.070
0.103
0.071
0.065
0.248
0.075
0.056
0.056
0.045
0.047
0.017
0.040
0.027
0.120
0.146
0.075
0.039
0.052
0.044
0.029
0.035
0.023
0.036
0.062
0.031
-------
TABLE A-2. DRY SCRUBBER
EMISSIONS SUMMARY 162 WEST.
DATE
82/04/20
83/10/24
84/09/24
84/09/25
84/09/27
LOCATION
162-U
162-W
162-W
162-W
162-W
TEST*
266
267
268
179
130
181
362
363
364
AL/P/D
7. ISO
2,997.5
2,997.5
2,997.5
3,071 .0
3,041 .0
3,071 .0
3,098.3
3,103.2
3,101 .5
104.8
95.3
112.6
1 10.0
104.3
112.0
109.9
112.2
114.0
T F LB/T
0.043
0.060
0.045
0.044
0.061
0.037
0 .063
0.040
0 .032
AVERAGE:
COUNT:
51
3,006.4
102.2
0.065
A-31
-------
TABLE A-3a. ROOF MONITOR 1016 STATISTICS BY YEAR.
THE FOLLOWING RESULTS ARE FOR:
RM* > 101G
YEAR
TOTAL OBSERVATIONS:
81.000
12
THE FOLLOWING RESULTS ARE FOR:
RM* « 101G
YEAR - 84.000
TOTAL OBSERVATIONS: 12
XBAR
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
3KEWNESS
KURTOSIS
12
0.60530
1.02000
0.41470
0.81795
0.12401
0.18829
-0.72386
SX
12
0.05450
0.25330
0.19880
0.13691
0.06824
0.40106
-1.26730
XBAR
SX
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
12
0.53670
0.99000
0.45330
0.77195
0. 11849
-0.08179
-0.05152
12
0.03510
0.30730
0.27220
0. 11210
0.0693S
1.89794
3.46135
THE FOLLOWING RESULTS ARE FOR:
RM* • 101G
YEAR
TOTAL OBSERVATIONS:
82.000
12
THE FOLLOWING RESULTS ARE FOR:
RM* * 101G
YEAR * 35.000
TOTAL OBSERVATIONS: 3
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
XBAR
12
0.43930
0.97000
0.53070
0.70387
14012
10351
0
-0
0.00133
SX
12
0.01970
0.24250
0.22280
0.12957
0.05856
0.01948
-0.12057
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
XBAR
3
0.59670
0.77330
0.17660
0.65943
0.09878
0.69615
-1.50000
SX
0.06160
0.29910
0.23750
0. 15697
0.12547
0.58963
-1.50000
THE FOLLOWING RESULTS ARE FOR:
RM* - 101G
YEAR > 83.000
TOTAL OBSERVATIONS:
12
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
XBAR
12
0.48330
1.15330
0.67OOO
0.79333
0.19334
0.20463
-0.59513
SX
12
0.04510
0.31190
0.26680
0. 15545
0.09007
0.79149
-0.78111
A-32
-------
TABLE A-3b ROOF MONITOR 101H STATISTICS BY YEAR.
THE FOLLOWING RESULTS ARE FOR:
RM* - 101H
YEAR
TOTAL OBSERVATIONS:
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSI3
81.000
12
XBAR
SX
12
0.67400
1.28500
0.61100
0.88616
0.17799
0.97064
0.28560
12
0. 10360
0.32340
0.21980
0.17450
0.07781
1.08287
-0. 16844
THE FOLLOWING RESULTS ARE FOR:
RM* « 101H
YEAR ' 84.000
TOTAL OBSERVATIONS:
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
12
XBAR
12
0.74670
1.00670
0.26000
0.87388
0.09457
0.20677
-1.42490
SX
12
0.05690
0.32600
0.26910
0.20784
0.08325
-0.30967
-1.01287
THE FOLLOWING RESULTS ARE FOR:
RM* ' 101H
YEAR
TOTAL OBSERVATIONS:
82.000
12
THE FOLLOWING RESULTS ARE FOR:
RM* ' 101H
YEAR • 85.000
TOTAL OBSERVATIONS: 3
XBAR
SX
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
12
0.56600
1.10330
0.53730
0.85497
0. 13798
-0.13451
0.27422
12
0.01990
0.27780
0.25790
0.12062
0.07544
0.48347
-0.24193
N OP CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
XBAR
0.70330
1.05670
0.35340
0.87367
0. 17704
0.13077
-1.50000
SX
3
0.26640
0.32520
0.05880
0.28753
0.03270
0.69140
-1.50000
THE FOLLOWING RESULTS ARE FOR:
RM* • 101H
YEAR
TOTAL OBSERVATIONS:
83.000
12
XBAR
SX
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
12
0.68670
1.04330
0.35660
0.85750
0.10376
0.09260
-0.65552
12
0.. 05200
0.29310
0.24110
0. 13608
0.06526
1.03494
0.86419
A-33
-------
TABLE A-3c. ROOF MONITOR 103G STATISTICS BY YEAR.
THE FOLLOWING RESULTS ARE FOR:
RM* = 103G
YEAR » 81.000
TOTAL OBSERVATIONS: 10
THE FOLLOWING RESULTS ARE FOR:
RM* > 103G
YEAR > 83.000
TOTAL OBSERVATIONS: 2
XBAR
SX
XBAR
SX
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
10
0.53100
0.94050
0.40950
0.68908
0. 14487
0.46481
•1.18160
10
0.04590
0.24340
0.19750
0. 12434
0.06942
0.71031
-1.03369
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
2
0.41000
0.71670
0.30670
0.56335
0.21687
0.00000
-2.00000
2
0. 10510
0. 14930
0.04420
0. 12720
0.03125
0.00000
-2.00000
THE FOLLOWING RESULTS ARE FOR:
RM* - 103G
YEAR * 82.000
TOTAL OBSERVATIONS: 12
THE FOLLOWING RESULTS ARE FOR:
RM* = 103G
YEAR
TOTAL OBSERVATIONS:
84.000
12
XBAR
SX
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
12
0.33330
0.92000
0.58670
0.69922
0. 18343
-0.55016
-0.62370
12
0.02S20
0.44030
0.41510
0.13966
0.1 2858
1.22035
0.55509
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
XBAR
12
0.55000
0.98000
0.43000
0.77305
0. 14888
0.15177
-1.36177
SX
12
0.01730
0. 18330
0.16600
0.09673
0.06331
0.02243
-1 .53934
THE FOLLOWING RESULTS ARE FOR:
RM* ' 103G
YEAR - 83.000
TOTAL OBSERVATIONS: 12
XBAR
SX
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
12
0.43330
1 . 1 2000
0.68670
0.79500
0.22934
-0.04009
-1.20337
12
0.01530
0.29140
0.27610
0.13131
0.09154
0.34984
-1.11994
A-34
-------
TABLE A-3d ROOF MONITOR 103H STATISTICS BY YEAR.
THE FOLLOWING RESULTS ARE FOR:
RM« « 103H
If EAR . g (. 000
TOTAL OBSERVATIONS: 10
THE FOLLOWING RESULTS ARE FOR-
RM« » IQ3H
YEAR
TOTAL OBSERVATIONS: 12
84.000
XBAR
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
sx
10
0.66800
1.29000
0.62200
0.98488
0.18449
-0. 12938
-0.75163
10
0.03540
0.26260
0.22720
0. 12847
0.08746
0.41515
-I. 15534
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
XBAR
1.32144
SX
12
0.59670
1 .31330
0.71660
0.85056
0.18036
1.23816
12
0.06430
0.35160
0.28730
0. 16978
0.07395
1. 15325
1.21949
THE FOLLOWING RESULTS ARE FOR:
RM» . 103H
YEAR * 82.000
TOTAL OBSERVATIONS:
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
12
XBAR
12
0.42000
1.18330
0.76330
0.84922
0.21083
-0.35875
-0.16665
SX
12
0.08620
0.21950
0. 13330
0. 17530
0.04256
-0.77000
-0.47223
THE FOLLOWING RESULTS ARE FOR-
RM« » 103H
YEAR
TOTAL OBSERVATIONS: 2
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
XBAR
85.000
SX
0.42000
0.63670
0.21670
0.52835
0.15323
0.00000
-2.00000
2
0.07940
0.09020
0.01080
0.08480
0.00764
0.00000
-2.00000
THE FOLLOWING RESULTS ARE FOR:
RM« ' 103H
YEAR » 83.000
TOTAL OBSERVATIONS: 12
XBAR
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
12
0.37000
I.10670
0.73670
0.79778
0.18536
-0.66855
0.72182
SX
12
0.04730
0.52920
0.48190
0.15298
0.12826
2.26157
4.45037
A-35
-------
TABLE A-4. ANALYSIS OF VARIANCE WITH MONITORS AS TREATMENTS.
nNnLv'CJlS OF- WimANCF UTTH MUM I I MRS Air. IRRITMENIS.
ou. v,.,!,: xnr.r-. H: .V" i-un.ifpiF f: - 'i° r-ruiAWP MUi.uriE f^:: . 1
AtlAl VtilB Ur VAR
'JOLJRCfi HUM Hf hit,!(I.U\'L,'i I»f- Nf ,'ifl - fU.'! l^Fvt. I -KATlf) Fr>
0.7/.?9/ 3 o.'.M^iL 8.157-: . OOO
-------
TABLE A-5. COMPARISON OF ROOF MONITOR ANNUAL MEANS.
***
4.2
4 '-3 Q,Q_I1OQ
/*.0*+
/QL&. -._. £>i&?76O.x*-.
/0_3_G -
#*
jg f AP
^
//
A-37
-------
TABLE A-6. ANALYSIS OF VARIANCE WITH DRY SCRUBBERS AS TREATMENT.
CO
oo
ANALYSTS (IF VARIANCE WITH DRV SCRUHBhR AS I RkAl MKI'll S.
DEI1 VAR:
XBAR
N: .'V Mill riPI.E ft: . 4LB SQUARED MUl.TTPl.E R
SOURCE
DBCRUB
SUN-OF -SQUARES
') „')?. i:? 12
M., i -i ,"-,10
AMALYS IS OF VAR IAIMCE
DF ME AM-SQUARE- F" -RATIO
•'. o. Ol O•
-------
TABLE A-7. LEAST SIGNIFICANT DIFFERENCE - DRY SCRUBBER MEANS.
J.
3 »
/hi
-fr
» 68
4 -! a
a -z. o
4-3 o.
3 @ •
J • - o
»7-'
*•*•
if-
17/7
(o aj) =^/-^fe /g.ooi-x,? fa)
r s?
a.
O
a .
**
3-01
A-39
-------
TABLE A-8a. ROOF MONITOR DATA KOLMOGOROV D STATISTIC.
UNIVARIATE
VARIABLE^XBAR
MOMENTS
1
o
N
MEAN
STD DEV
SKEMNESS
USS
CV
T:MEAN-O
SON RANK
NUM "- 0
D: NORMAL
220
0.809367
0.174152
0.0195071
150.759
21 .517
68.9334
12155
220
0.0317748
SUM MCTS
SUM
VARIANCE
KURTCIS1S
CSS
STD MEAN
PROB> 1 f 1
PROB>ISI
PROB>D
220
178.061
0.0303288
0,249447
6.642
0.0117413
0.0001
0.0001
>0.15
QUANT ILES
100% MAX
75% Q3
50% MLH
25% C4i
0% KIN
RANGE
Q3-Q1
MODE
1 .3133
O.Vli
0,81335
0,700B1?5
0.3333
0.98
0,219175
0.7167
99%
95%
90%
10%
5%
1%
1 .28895
1.1161?
1 .02477
0.1393A4
0.5177
0.3784
V«KintiLt-LXBAR
MOMENTS
N
MEAN
STD DEV
3KEWNES5
USS
CV
T:MEAN^-O
SGN RANK
NUM -- 0
DI NORMAL
220
-0.236448
0.230251
-0.781481
23.9101
-97.3789
-15.2316
-10853
220
0.0644181
SUM WG1S
SUM
VARIANCE
KUKTOSIS
CSS
STD MEAN
PROB> 1 1 1
PROB>ISI
PROB>D
220
-52.0187
0.05301'J4
i .18344
11.6104
0,0155235
0.0001
0.0001
0.024
100% MAX 0,272543
75% «3 -0,0833836
50% MFD -0.206602
25% UJ -0,3515499
0% MIN -1,09871
RANGE
03-01
MODE
1.37126
0,272118
-0,333098
99%
95%
90%
10%
5%
1%
0,253827
0,109928
0,024467
- 0,521484
-0,658376
-0,972695
-------
TABLE A-8b. ROOF MONITOR 101G. KOLMOGOROV D STATISTIC.
HON-101G
UNIVARIATE
VARIABLE^LXBAR
MOMENTS
N
MEAN
STD DEV
SKEWNESS
USS
CV
TIMEAN--0
SON RANK
NUM "= 0
D: NORMAL
MON^IOIG
56
-0.292971
0.19607
-0.320428
6.92101
-66.9248
-11.1817
-779
56
0.0599727
SUM WGTS
SUM
VARIANCE
KURTOS1S
CSS
STD MEAN
PROB>MI
PROU>ISI
PROB>U
56
-16.4064
0.0384436
0,270236
2.1144
0,026201
0.0001
0.0001
>0.15
UNIVARIATE
VAR1ABLE-XBAR
MOMENTS
N
MEAN
STD DEV
SKEWNESS
USS
CV
TJMEAN^O
SGN RANK
NUM "=-- 0
D*. NORMAL
56
0.759979
0.145988
0.26947
33.516
19.2095
38.9564
798
56
0.06253'J1
SUM WG1S
SUM
VARIANCE
KURTOSIS
CSS
STD MEAN
PROB>I 1 1
PRUBXSI
PROB>D
56
42.5588
0.0213125
0, 130776
1.17219
0.0195085
0.0001
0.0001
>0.15
100% MAX
7'JX. 03
50% MED
25% 01
0% MIN
RANGE
G3-G1
MODE
GUANI1LLS(DEF=
0, 142627
0, 165878
-0,29863
0,406096
-0,822573
0.9652
0,240218
•0.333098
100% MAX
75% 03
50% MED
25% 01
0% MIN
RANGE
03-01
MODE
j,1533
0,84715
0,74185
0,66625
0,4393
0.714
0.1809
0.7167
99%
95%
90%
10%
5%
1%
0,112627
0.021266
-0.02-13368
-0,54079
-0.669823
-0,822573
M>
99%
95%
90%
10%
5%
1%
1.1533
1.0215
0.976
0.58269
0,511945
0.4393
-------
TABLE A-8c. ROOF MONITOR 101H. KOLMOGOROV D STATISTIC.
MOK'^IOIH
UNIVARIATE
VARIABLE-XBAR
MOMENTS
N
MEAN
STD DEV
SKEMNESS
USS
CV
TtMEAN^O
SGN RANK
NUM "~ 0
D: NORMAL
MON^IOIH
UNIVARIA
56
0.869611
0.132073
0.525695
43.3079
15.1876
49.2726
798
56
0.0716823
TE
SUM WGTS
SUM
VARIANCE
KURTOSIS
CSS
STD MEAN
PROBXTI
PROBXSI
PROB>D
56
48.6982
0,0174432
0,780922
0.959377
0,017649
0.0001
0,0001
>0.15
VARIABLE^LXBAR
MOMENTS
N
MEAN
STD DEV
SKEMNESS
USS
CV
T J MEAN~0
SGN RANK
NUM "- 0
Dt NORMAL
56
-0.150875
0.isJ0734
0.00353465
2.52437
-99.9065
-7.49032
-667
56
0.0535399
SUM WGTS
SUM
VARIANCE
KURTOSIS
CSS
STD MEAN
PROB> 1 1 1
PROB> 1 S 1
PROB>U
56
-8,44898
0.022/206
0,431585
1 .24963
0,0201426
0.0001
0.0001
>0.15
QUANTILESUiEh": 4)
100% MAX
75% 03
50% MKH
25% «1
0% MIN
RANGE
G3-Q1
MODE
1 . 285
0.9615
0,85165
0,784i5
0.566
0.719
0.18085
0.7833
99%
95%
90%
10%
5%
1%
1.285
1.1078
1 .04432
0.70231
0,681905
0.566
100%
75%
50%
25%
0%
MAX 0,2'J0759
UK -0,0356677
MFLi -0, 1601J81
Oi -0, 243157
MIN -0,569161
RANGE
03-Q1
MODE
0.81992
0,207489
-0,24424
4)
99%
95%
90%
10%
5%
1%
0,250759
0, 10233
0,0433648
-0,353383
-0,382877
-0,569161
-------
TABLE A-8d. ROOF MONITOR 103G ' KOLMOGOROV D STATISTIC,
UNIVARIATE
VARIABLE-XBAR
MOMENTS
N
MEAN
STD DEV
SKEMNESS
U5S
CV
T:MEAN^O
SGN RANK
NUM "" o
D:NURMAL
54
0.749867
0.186611
0.0265797
32.2099
24.8859
29.5286
742.5
54
0.0857499
SUM MtilS
SUM
VARIANCE
KURTOSIS
CSS
STD MEAN
PROB> 1 T 1
F'ROUXSI
PRO If > II
54
40.492B
0.0348237
-0.546949
1.B4566
0.0253945
0.0001
0.0001
>0.15
QUANTILES(DEK>4>
100% MAX
75% 03
50% MED
25% Dl
0% MIN
RANGE
03-U1
MODE
1 .13
0,910025
0,73165
0.598325
0,3333
0.7967
0.3117
0.73
99%
95%
90%
10%
5%
1%
1.13
1 .11752
0.96665
0.51715
0,427475
0.3333
MONM03G
UNIVARIATE
VARIABLE^LXBAR
N
MEAN
STD DEV
SKEUNESS
USS
CV
T:MEAN=-O
SGN RANK
NUM "« 0
D:NORMAL
MOMENTS
54
-0.320701
0.265456
-0.599186
9.28858
-82.7736
-8.87779
-695.5
54
0.0897283
SUM HGTS
SUM
VARIANCE
KURTUSIS
CSS
STD MEfcN
PROB>ITI
PRDBXSI
PRU6>D
54
-17,3178
0.0704667
0,223004
3.73473
0.0361239
0.0001
0.0001
: o. 15
GUANflLESUiEF 4)
00% MAX
75% Q3 -
50% MED
25% Oi
0% MIN
RANGE
03-Q1
MODE
0, 127218
0,0943031
-0,312456
-0, 513633
- i ,09871
1 .22093
0.41933
-0.314711
99%
95%
90%
10%
5%
1%
0, 122218
0,111116
-0.0340146
-0,6'J9/81
-0.850143
-1,09871
-------
T/b'.E /,-6e. P.OOF MONITOR 103H KOLMOGOROV D STATISTIC.
UNIUARIATE
MOMENTS
MEAN
STD DEV
SKEUNESS
USS
ru
L* V
T:MEAN-O
SON RANK
NUM "~ 0
DJ NORMAL
54
0.857611
0. 194649
-0.149388
41 .7249
22.6967
32.3768
742.5
54
0.0880425
SUM HUTS
SUM
VARIANCE
KUR10SIS
CSS
STD MEAN
PROBXTI
PROBXSI
PROB>D
54
46.311
0.0378884
0.64548
2.00808
0.0264884
0.0001
0.0001
> 0 . 1 5
OUANTILESUiEF-4)
100% MAX
752 03
50% MEU
25% Qi
0% KIN
RANGE
Q3-Q1
MOPE
1,3133
0.9565
0.87
0, '/4 0025
0.37
0.9433
0,216475
0.87
99%
95%
90%
10%
5%
1%
1,3133
1.20997
1,1117
0.62
0.42
0.37
UNIVARIATE
VARIABLE=--LXBAR
MOMENTS
MEAN
STD DEV
SKEHNESS
nee
U 9 w
cv
TtMEAN^O
SON RANK
NUM "- 0
Dt NORMAL
54
-0.182323
0.252576
-1.14209
5.17616
-138.532
-5.30453
-554.5
54
0.12845
SUM WGTS
SUM
VARIANCE
KURTOSIS
CSS
STD MEAN
PROBXTI
PROBXSI
PRUB>V
54
-9.84544
0.0637945
2,27079
3,38111
0,0343712
0.0001
0.0001
0.024
QUANTILES(DEF-4>
100% MAX 0,272543
75% B3 -0,0445075
50% MED -0.139262
25% Ri -0,30ii9!j
0% MIN -0.994252
RANGE
03-01
MODE
1.2668
0,256687
-0.139262
NOTE? SAS INSTITUTE. SAS CIRCLE. BOX
8000, CARY, N.C, 27511-8000
99%
95%
90%
10%
5%
1%
0,272543
0.18989
0.10588
-0,478399
-0.86/501
-0,994252
-------
TABLE A-S AUTOCORRELATION AND PARTIAL AUTOCORRELATION FUNCTIONS FOR ROOF MONITOR.
' 101G
ALUMAX DATA FOR ROOF MONITOR 101-G.
PLOT OF XBAR
NUMBER OF CASES = 56
MEAN OF SERIES = 0.760
STANDARD DEVIATION OF SERIES
PLOT OF AUTOCORRELATIONS
0.145
LAG
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
CORR
.333
.132
.077
-.107
-.080
-. 144
-.231
-. 183
-.075
.079
-.014
.101
.023
.096
-.014
SE -1.0 - . 8
. 134
. 148
.150
.150
.152
.153
.155
.161
.165
.165
. 166
.166
.167
.167
.168
-.6 -.4 -.2
<
<
( ***
< ***
< ****
<******
< *****
< **
<
< *
(
\
< *
0 .2 .4 .6 .81.0
*****)***
**** )
** )
)
)
)
)
)
** )
)
*** )
* )
*** )
)
PLOT OF XBAR
NUMBER OF CASES = 56
MEAN OF SERIES = 0.760
STANDARD DEVIATION OF SERIES = 0.145
PLOT OF PARTIAL AUTOCORRELATIONS
LAG CORR SE -1.0 -. 8 -.6 -.4 -.2
.0
.2
.4
.6
.8 1 .0
i
2
3
4
5
6
7
8
9
10
11
12
13
14
15
.333
.024
.029
-. 161
-.004
-.115
-.150
-.078
.034
.124
-.125
.101
-. 102
. 122
-.186
. 134
.134
. 134
.134
.134
. 134
. 134
.134
. 134
.134
.134
.134
.134
. 134
.134
< ;*****>***
< !*
< !*
(*****!
( *!
< *** ;
( ****!
< **:
< :*
( !****
( ****!
< :***
< ***:
( :****
<*****:
>
>
>
>
>
>
>
>
>
>
>
>
>
>
A-45
-------
TABLE A-10. AUTOCORRELATION AND PARTIAL AUTOCORRELATION FUNCTIONS FOR ROOF
'MONITOR IOIH.
ALUMAX DATA FOR ROOF MONITOR 101-H.
XBAR COPIED FROM SYSTAT FILE INTO ACTIVE WORK AREA
PLOT OF XBAR
NUMBER OF CASES = 56
MEAN OF SERIES = 0.870
STANDARD DEVIATION OF SERIES
0. 131
PLOT OF AUTOCORRELATIONS
LAG CORR
1 -
2
3 -
4
5 -
6 -
7 -
8
9 -
10
11 -
12
13 -
14 -
15
>PACF
.161
.063
.232
.068
.005
.136
.001
.021
.053
.006
.051
.064
.113
.018
.179
XBAR
SE -1.0 - . 8
134
137
138
144
145
145
147
147 .
147
148
148
148
148
150
150
-.6 -.4 -.2 .0 .2 .4 .6 .8 1.
(*****; )
< ;** )
(*****! )
< : ** )
( *! )
< ****'. >
< *; )
< :* )
< **: )
< :* )
< **! )
< ! ** )
< ***: )
< *: )
( ! ***** )
PLOT OF XBAR
NUMBER OF CASES * 56
MEAN OF SERIES =
STANDARD DEVIATION OF
0.870
SERIES
0.131
PLOT OF PARTIAL AUTOCORRELATIONS
LAG CORR
SE
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
-.161
.038
-.222
-.002
.021
-.201
-.037
.030
-.144
-.020
-.039
-.035
-.122
-.082
.173
.134
.134
.134
.134
.134
.134
.134
.134
.134
.134
.134
.134
.134
. 134
.134
-1.0 -.8 -.6 -.4 -.2
+ _-.._4.. --- + ---- + +----+
<*****:
( !
(*****!
< *:
< :
<*****:
< *!
(i
•
< ****!
< *:
< *!
< *!
( ****!
< ***!
0 .2
)
* )
)
)
* )
)
)
* )
)
)
)
)
)
)
.4 .6 .8
+_„--+----+-
( !*****)
A-46
-------
TABLE A-11. AUTOCORRELATION FUNCTION AND PARTIAL AUTOCORRELATION FUNCTION
FOR ROOF MONITOR 103G.
ALUMAX DATA FOR ROOF MONITOR 103-G.
XBAR COPIED FROM SYSTAT FILE INTO ACTIVE WORK AREA
PLOT OF XBAR
NUMBER OF CASES = 54
MEAN OF SERIES = 0.739
STANDARD DEVIATION OF SERIES
PLOT OF AUTOCORRELATIONS
LAG CORR
0.208
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
-. 119
.018
-.065
-.076
.029
.032
-.172
.082
-.002
.035
-.128
.097
.129
.128
-.156
SE -1.0 -.8 -.6
136
138
138
139
139
139
140
143
144
144
145
147
148
150
152
-.4 -.2
( ***
(
< **
( **
<
(
(*****
(
( *
(
( ****
<
(
<
( ****
0 .2 .4 .6 .8 1.0
* )
\
\
* )
* )
)
*** )
>
* )
)
*** )
**** )
**** )
)
PLOT OF XBAR
NUMBER OF CASES = 54
MEAN OF SERIES = 0.739
STANDARD DEVIATION OF SERIES =
PLOT OF PARTIAL AUTOCORRELATIONS
LAG CORR SE -1.0 -.8 -.6
0.208
-.4 -.2
.0
.2
.4
.6
.8 1 .0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
-.119
.003
-.063
-.093
.010
.034
-.181
.038
.024
.014
-.154
.095
.180
.118
-.151
.136
.136
.136
.136
.136
.136
.136
.136
.136
.136
.136
.136
.136
.136
.136
***:
i
i
**:
***!
:*****!
****:
: *
!*
i
i
*
!*
:*
! *** >
:*****)
i ***
****:
A-47
-------
TABLE A-12. AUTOCORRELATION AND PARTIAL AUTOCORRELATION FUNCTION FOR
ROOF MONITOR 103H.
ALUMAX DATA FOR ROOF MONITOR 103-H.
XBAR COPIED FROM SYSTAT FILE INTO ACTIVE WORK AREA
PLOT OF XBAR
NUMBER OF CASES = 54
MEAN OF SERIES = 0.858
STANDARD DEVIATION OF SERIES
0. 193
PLOT OF AUTOCORRELATIONS
LAG
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
CORR
.282
.129
.038
-.049
-.387
-.198
-.081
-.024
-.021
. 125
.130
.163
.182
.251
.019
SE -1.0 -.8 -.6 -.4 -.2
.136 <
.146 <
.149 <
.149 ( **
. 149 ***(******
. 167 < *****
.171 • < ***
.172 < *
.172 < *
. 172 <
.173 <
.175 <
.178 <
.181 <
. 188 <
0 .2
*****)**
**** )
* )
)
\
)
)
)
**** )
**** )
***** )
***** )
*******
*
.4 .6 .8 1 .<
)
)
PLOT OF XBAR
NUMBER OF CASES = 54
MEAN OF SERIES = 0.858
STANDARD DEVIATION OF SERIES =
PLOT OF PARTIAL AUTOCORRELATIONS
LAG CORR SE -1.0 -.8 -.6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
.282
.054
-.013
-.069
-.391
.003
.054
.033
-.028
-.017
.039
.136
.149
.186
-.103
.136
.136
.136
.136
.136
.136
.136
.136
.136
.136
.136
.136
.136
.136
.136
0. 193
-.4 -.2 .0 .2 .4
._+ + -+ + +--
< J*****)**
.6
•+-•
**
< **:
****<*****:
:*
! **
!*
*:
*:
!*
! ****
; ****
:*****)
***; )
>
)
)
)
>
)
>
)
>
)
>
)
A-48
-------
TABLE A-13 AUTOCORRELATION AND PARTIAL AUTOCORRELATION FUNCTION
FOR THE STANDARD DEVIATION OF MONTHLY MEANS FOR
ROOF MONITOR 101G.
ALUMAX DATA FOR ROOF MONITOR 101-G.
THE VARIABLE IS THE STANDARD DEVIATION OF THE MONTHLY READINGS.
SX COPIED FROM SYSTAT FILE INTO ACTIVE WORK AREA
PLOT OF SX
NUMBER OF CASES = 56
MEAN OF SERIES = 0.133
STANDARD DEVIATION OF SERIES
PLOT OF AUTOCORRELATIONS
LAG CORR SE -1.0 -.8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
PLOT OF SX
NUMBER OF CASES = 56
MEAN OF SERIES = 0.133
STANDARD DEVIATION OF SERIES =
PLOT OF PARTIAL AUTOCORRELATIONS
LAG CORR SE -1.0 -. 8 -. 6
0.073
-.6 -.4 -.2
.061
.171
.115
.003
.189
.101
.044
. 101
.061
.057
.079
.182
.210
.207
.034
. 134
.134
.138
. 140
. 140
.144
. 145
.146
.147
.147
.148
.149
.152
.158
.162
.4 -.2 .0 .2 .4
< **! )
-------
TABLE A-14. AUTOCORRELATION AND PARTIAL AUTOCORRELATION FUNCTION FOR THE STANDARD
DEVIATION OF MONTHLY MEANS FOR ROOF MONITOR 101H.
ALUMAX DATA FOR FOCF MONITOR l-.-l-H.
PLOT OF 'iX
NUMBER OF CASES = 56
MEAN OF SERIES =
STANDARD DEVIATION OF
O
165
SERIES
0. 082
PLOT OF AUTOCORRELATION'S
i.AG CORR
3E
-.3 -.6
.6
1
/•^
,J|
4
5
6
7
9
9
10
11
12
13
14
15
. 0 1 0
. 183
.118
. 142
-.046
-.043
- . 0^3
. Ill
- . O03
. 129
. 057
-. 003
. 150
-. 146
-.070
. 134
.134
. 133
. 1 40
. 142
. 143
. 143
. 144
. 145
. 145
. 148
. 143
. 143
. 151
. 153
•**
*••*•
#**
* )
###*•*)
*** )
**•*
****
#»•
#•***
PLOT OF SX
NUMBER OF CASES -
MEAN OF SERIES =
STANDARD DEVIATION
OF SERIES ~
-.032
PLOT OF PARTIAL. AUTOCORRELATIONS
L AG
CORR
1
*-,
-;
4
5
fi
-~T
a
9
10
1 1
12
13
14
15
. 0 1 0
.133
. 119
. 114
- . 090
-.111
-.114
. 145
. 089
. 151
. 038
— . 1 35
. O75
•-. 133
."»""* "7
~" * ' / /
. 1 34
. 134
. 134
. 134
. 134
. 134
. 134
. 134
. 134
. 134
. 134
. 134
. 134
. 134
. 134
- .3 - . a - . 4 - .
(
•
<>
^
<
i^
,,
.
(
.;
(
2 . '.' ..2 - 4 .6
i * '
*.* + +* i
\ *• »-f- >
! !-•** )
** •• ! )
*»» ' )
«••*•»•! >
:•**-•*• >
! »*•*
; **•** '•
: *
A-50
*••**•*-!
„ -j
-h —
-------
TABLE A-15. AUTOCORRELATION AID PARTIAL AUTOCORRELATION FUNCTION FOR THE STANDARD
DEVIATION OF MONTHLY MEANS FOR ROOF MONITOR 103G.
DATA FOR I" OOF MONITOR 103-6
SX IS THE STANDARD DEVIATION OF THE DAILV READINGS EACH
SX COPIED FROM SYSTAT FILE INTO ACTIVE WORI- AREA
PLOT OF bX
NUMBER OF CASES = 54
MEAN OF SERIES = 0.121
STANDARD DEVIATION OF SERIES = 0.036
PLOT OF AUTOCOPRELAF[ONS
LAG CORR SE -1.0 -.3
MONTH.
.4
1
2
-•;
4
5
6
-7
3
q>
10
11
12
13
I j.
15
-.221
-.076
— ~'29
.267
-. 163
. 110
__ /"> ~7 ^T
. 19.->
- ,. O4O
. 115
-. 125
- . 003
._ _ t'lJQ
. 173
.015
. 136
. 143
. 143
. 150
. 15S
. L \'3 «1
. 1*3
. 164
. lea
. Io8
. 1 69
. 171
. 17!
. 1 72
. 175
(#**•***
,, *.*.».**
**
*•*•##•**)
*•*•*
PLOT OF SX
NUMBER OF CA'BES - 54
MEAN CF -lEPiu? = _ -.'.121
,3rANDARO GE'.-'CATION OF "ERIE3 =
f'LOT OF PARTIAL AUTOCORRELATIONS
LAG
1
2
-
4
er(
6
-7
a
9
10
1 1
12
13
14
15
CORR
- . 221
-. 132
-.296
. 140
-.151
. 049
. 0 L 6
. 123
. 136
. Io3
. 059
-.'D36
-.054
. 065
. 073
iE -1.0 -.3 -.6 -.4 -.2
^ + < H -i -t
.136 (*****!
.136 •: *^**
.136 **<***##
. 176 <
.1-0 < ***•*
. 1 76 ''
.136 <
.136 <
. 1 36
. 136 '-
.136 <
. 136 '- *
. 13o '; **
:;': A-5i ;
•I- .2 .4 .6
1- H -1
'
,'
i
* •»• )- * ;
,'
••* ;
* >
*«••*•** )
*-*** }
»•»-***;
** ,'
>
't
** )
i. o
-------
TABLE A-16 AUTOCORRELATION AND PARTIAL AUTOCORRELATION OF FUNCTION FOR THE
STANDARD DEVIATION OF MONTHLY MEANS FOR ROOF MONITOR 103H .
MA* L'ATA FOR ROOF" MONITOR K'3-H.
3X IS THE -STANDARD DEVIATION OF THE DAILY READINGS EACH
SX COPIED TROM SvSTAT FILE INTO ACTIVE WORK. AREA
MONTH.
PLOT OF 3X
NUMBER OF CASfIS - 54
MFAN OF SERIE3 -- '•'• '-^
STANDARD DEVIATION ''F SCRIED --
PLOT OF AUTOCORRELATIONS
LAG COPR ':!-- -• 1 - '1 • <*
0.032
1
2
-;
4
5
6
—
n
9
!. 0
1 1
i 2
1 -
14
i *"'
. 035
-.023
-.117
. OO6
•-. 063
-. 142
- . • J I 2
--.. 037
. 0 1 4
. 035
-. 151
— f~i~i"i
. 20'J
. -1/92
. 1.24
. 136
. 137
. 177
. 13?
. i ':."•
. I 40
. 143
. 1 4 :••
1 42
i 4^
,. 1 4.'
., J. 4 <3
„ i 4 /
. Ll'2
. 153
a -.4 -.2
i
( *•
i, *•**
(
( **
0 -2
»•*# )
)
)
)
I
.4 .6
«••*-*•
, *#*•» )
PLOT OF 3X
NUMBER OF Ctt'3C.3 = ^'f-
MErtN OF C'-;R:E" - -.155
STANDARD C-CV [AT [Q") Or JIRIE3 =
PLOT Of r'Ai
LAG C'JRK'
.8
- . a
.4
1
-v
-
4
5
-,
7
3
9
.10
1 I
12
1 3
14
. OC5
-.036
-.112
(.-f .— i cr
-.074
- . 1 4fc
.013
- . 064
- . 0 1 3
-. 092
-. 130
-.030
. 190
' . 005
. 136
. 136
. 136
1 ""** **-%
•( "T »
. I3o
. 136
. L36
. 13-D
, I. 3v3
. 136
. i 36
. 136
. I3o
•**•*«•
*
***
+*••*•
A-52
!*•****»
#*# >
-------
TABLE A-17 STATISTICS FOR DAILY ROOF MONITOR READINGS.
TOTAL OBSERVATIONS: 655
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
655
0.24000
1.67000
1.43000
0.30860
0.22068
0.34915
0.34027
A-53
-------
TABLE A-18. STATISTICS FOR DAILY READINGS BY ROOM MONITORS.
THE FOLLOWING RESULTS ARE FOR:
RM* = [QIC
TOTAL OBSERVATIONS: 168
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
168
0.33600
1.31000
0.97400
0.76042
0.19158
0.30920
-0.03543
THE FOLLOWING RESULTS ARE FOR:
RM* - 101H
TOTAL OBSERVATIONS: 168
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
168
0.42000
1.65600
1.23600
0.87027
0.19946
0.41516
0.60178
THE FOLLOWING RESULTS ARE FOR:
RM* > 103G
TOTAL OBSERVATIONS: 160
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
160
0.28000
I.40000
1.12000
0.74748
0.22181
0.33714
-0. 11291
THE FOLLOWING RESULTS ARE FOR:
RM* ' 103H
TOTAL OBSERVATIONS: 159
N OF CASES
MINIMUM
MAXIMUM
RANGE
MEAN
STANDARD DEV
SKEWNESS
KURTOSIS
159
0.24000
1.67000
1.43000
0.85586
0.24225
0.32794
0.43047
A-54
-------
TABLE A-19. REGRESSION ON POTLINE 101 ROOF MONITORS AND DRY SCRUBBERS.
THIS IS A DEGRESSION ON RM101S AND D5I61W.
DEP VAR: APAR N: 19 rtULTIPLE ': . •>„3 SL'LAKEC .-";;. T!.'
ADJUSTED R
VARIABLE
CONSTANT
03XBAR
l-'l-R '
CQEFFtCIEMT
' DF, WHFRE M= 19, ADD DF=
'3TD. ERFOR STD. CQEF.
SOURCE ::UM-OF-'=OIJAPC5 DF
PESIDUAL
1
17
113 OF VARIANCE
>'iN- SQUARE F-RATIQ P
'.''.'"019^4 . o/a .73.
THIS IS M KCGRESSION LIiN RMlOiH AMD D3161E.
CEP VAP: XE--.P N: 19 MULTIPLE R: .07^ SJUAPED ^
ADJUSTED P
VARIABLE
CQNSTANT
D3XDAR
1- <1-R .. »-(N-l) /DF, WHERE N= 19, AND DF= 17:
COEFFICCENT 3TD. ERROR 3TD. COEF. TCLEPANCE
0.359275
•:•. 277209
0.092767
1.316276
0.079124
ANALYSIS OF VARIANCE
SOURCE "UJM-QF-SajMRES DF MEAN-SQUARE
1
REGRESS IOr
RE3ICUAL
r
0. C'00754
0.029979
"'26
A-55
-------
TABLE A-20. REGRESSION ON POTLINE 103 ROOF MONITORS AND DRY SCRUBBERS.
THIS IS A FF6KESSIQN ON RMlOTG AND DS162W.
MULTIPLE R: -047 SOUAKSD ^TIFLE - • <>-
OEP VAR: -E|AR
ADJUSTEO R = ^l-.N-n. ^ ^
, --• -
RE M= 17, AND DF= I"*: •'-'"-
COEFFICIENT 3TO. ERROR 3TD. CQEF. TOLERANCE
7rM - - ,-> .'79643 0.000000 . 3'.'l-3
CONSTANT "•"'.', "^ i'oA'l ~'l 0.046736 1 . 0'"":".'O -IJ
DSXBAR '"'• ''•-':-:i4 '
riAl-VSIS OF VARIANCE
SOURCE SUM-OF-50MARE3 OF MEAN-SQUARE
. . - -,&, i i "i. (''O0856
REGRESSION '-•':';';7^ .s ,'-,.,"-,26076
RESIDUAL "• ''I'*-' l" ' '
P
0-7 - 3T,9
THIS IS A REGRESSION ON RM107H AND DS162E.
Dep.,HR, XPAR N= 17 MULTIPLE P, -474 3CU^ED -ULT IPLE R: .2*
ADJUSTED R"- l-.l-9")*'N-l)/DF, WHERE N= 17, AND '" ^ 1^5: • i"
•'ftRIABLE COEFFICIENT 3TD. ERROR 3TD. COEF. "CL.EFANCE F
- -, — -i- i-> i ,-iQIOpi. M . ('l)(l'< ' .
CONSTANT ,j.7.._'.>o8 - -^ «= ; ;^ ..-„-,,-„.!..
DSXBAR 1.712671 O.J—O". ).*./— i.-
ANALYSIS OF VARIANCE
c.OURCE SUM-OF-SOU^rES DF MEAN-SQUARE
pcer-ESSIDN -.120174 1 ^ .^^J
RESIDUAL '.415171 15 O.<->-^.
-------
TABLE A-21. COVARIANCE MATRIES FOR ROOF MONITORS AND DRY SCRUBBERS.
f>
cn
THIS IS THE CORRELATION MATRIX FDR RMHHG-DS161 W. rHI5 IS ,HE cOVARIANCE MATRIX FDR RMK>3H~DS1O
1 1 . OOi )675 0 . 004 151
NUMBER OF OBStRVATIONS:
CCK'ARIANCE MATRIX
XBAR
DSXBAR
XBAR UE
O. '173458
MUMBE:R OF OBSERVATIOMS: i?
THIS IS THE CORRELATION MATRIX FOR R
COVAKIANCE MATRIX
XBAR
XBAR
DSXBAR
NUMBER OF OBSERVAriON^:
XBAR
0 . ' -•'» '0487
THIS IS THE ITOVARIANCE MATRIX FUR RM1O3G-DS161 W.
rOVAftIANCE MA f FI X
XBAR
DSXBAR
XBAR
0.O24500
O.OO027S
DSXBAR
NUMBER OF UFc-iERVAT(ONii: 17
-------
FIGURE A-l. PROBABILITY PLOT OF DAILY ROOF MONITOR READINGS.
THIS IS A PLOT OP THE DAILY EMISSIONS C=QM THE ALUMAX DATA.
NOfcMAL PROBABILITY PLOT
EXPECTED
VALUE
-4
OEVIATTOM
PPQM
EXPECTED
VALUE
636 35*
9999
9909
9999
'7999
796
_'*
DETRENDED rJGRMAL PPOBABILITV PLOT
-1.0
? 59999999999939^^96
^* 4999953
379
*43 *
0.5
1.5
A-58
-------
FIGURE A-2. PROBABILITY PLOT OF THE LOGARITHM OF THE DAILY ROOF MONITOR READINGS.
THIS
IS A PLOT OF THE LOGARITHM OF THE ALUfAX FOf.F MONITOR
THe LOGARITHM OF THE PATLY .vnNITOF .-£AD1,NRO IS l.Cv .
Fl.CT
EXPECTI-:D
VHl 'J:-;
154
7909
- 1 .
:-,£-]- e:{rr,j pen NORMAL PR
f-l.or
59364
9
-93
LJX
A-59
-------
FIGURE A-3a. PLOT OF RM 101G EMISSIONS VS. DS162W EMISSIONS.
W
PLOT OF RM101G EMISSIONS VS. DS162£ EMISSIONS.
—t--
1.0
0.8 +
0.6
0.4
0.00
*
*
# * *
0.05
1—
0. 10
0. 15
DSXBAR
0.20
0.
FIGURE A-3b PLOT OF RM 101H EMISSIONS VS DS161E EMISSIONS
PLOT OF RM101H EMISSIONS VS. DS161E EMISSIONS.
XBAR
1.4 +
1.2
1.0
0.8
O.6 -i-
0.4 +
0.:
0.00
0.02
# *
*
*
O.O4 0.06
DSXBAR
0.08
A-60
-------
FIGURE A-4a. PLOT OF RM 103G EMISSIONS VS DS161W EMISSIONS.
PLOT OF RM103G EMISSIONS VS.DS1A1W.
XBAR
1.0
0.8
0.6
0.4
0.2 «•
**
** *
*
0.00
0.05
0. 10
DSXBAR
0. 15
0.2(.
FIGURE A-4b PLOT OF RM 103H EMISSIONS VS DS162E EMISSIONS
PLOT OF RM103H EMISSIONS VS. DS 162E EMISSIONS.
XBAR
1.4 -i-
1.2
l.O
0.8
0.6
# * * *
* *
*
o.oo
0.05
0.10 0.15
DSXBAR
0.20
0.25
A-61
-------
APPENDIX B
CONTROL CHART PARAMETERS
Factors for computing control limits:10
All of the factors are functions of n only. Integer variable n is the
number of samples that are taken per measurement. The constant 2 or 3 results
from whether the 2-sigma limits or the 3-sigma limits are used. The 2-sigma
limits are used with the upper and lower warning limits; the 3-sigma limits
are used to calculate the upper and lower control limits.
c
n-1
- 3-sigma limits _ 2-sioma limits
A • 3/Vn~ A4 - 2/VrT
A
3
B3 - l-(3/C4)(l-C2)1/2
B4=l-(3/C4)(l-C2)1/2
g
B5 - C4-3(1-C) B
B6 - C4-3(1-C42)1/2
For n = 3
0! = 1.0 , (-1/2)!= (*)1/2
C4 = (1/2)1/7/0! = 0.88623
B-l
-------
TABLE B-l. CONTROL CHART FACTORS FOR SAMPLE SIZE OF THREE
A
A3
B3
B4
85
B6
3-siqma limits
= 3//3 = 1.732
= 3/(C4/3) = 1.954
= 0
= 2.568
= 0
= 2.276
2-siqma limits
A4
A5
B7
B8
B9
B10
= 2//3 = 1.1547
= 2/(C4/3) = 1.3i
= 0
= 2.045
= 0
= 1.813
Note that B3, BS, B7, and Bg calculate <0 by formulas
given above, however B3, BS, B7, and Bg are always >0.
Calculations of Control Limits
Central Lines
For a potroom group the central lines of the X and s chart are
calculated as follows:
(a) For samples of equal size
X = (Xj + X2 - X3 + ... + Xk)/k (1)
sx = (Sj +s2 + s3 + ... + sk)/k, (2)
where ^ = the monthly mean potroom group emission for month i, Ib F/T AT,
s.. = the monthly standard deviation of three daily readings, Ib F/T
AT.
k = the total number of months for which emissions data are
available.
(b) For samples of unequal size the central lines are calculated as
follows:
X" = (nj Xj + n2X2 + n3*3 + ... + nX)/(n+n + ... + n), (3)
l + n2S2 + "• + nksk)/(nl + "2 + "3 +••• + nk), (4)
where ni = the number of samples (daily readings) taken for month i.
B-2
-------
Control Limits
(a) Upper and lower control limits - No Standard Given.
For X Chart
UCL = X + A3 sx
LCL = X - A, sv
-------
Warning Limits Calculation
Central lines
X Chart, XQ = X for potroom group
$x Chart, aQ - §x for potroom group
Table B-2 gives the overall means, X, of the monthly mean emission rates for
the ALUMAX potroom groups in Ib F/T aluminum, and the mean of the monthly
standard deviations, ix. For each potroom group the number of monthly
observations, n, was three.
TABLE B-2. ALUMAX DATA FOR POTROOM GROUPS
Potroom Group
101G/161W
101H/161E
103G/162W
103H/162E
X
0.8473
0.9162
0.8070
0.9618
~sx
0.1499
0.1663
0.1265
0.1651
n
3
3
3
3
Table B-3 shows the results of the calculation of upper and lower warning
limits for the ALUMAX potroom groups. The warning limits were based on the
constants for the 2-sigma control chart factors given in the equations above,
and in Table 1. The upper and lower control limits were not calculated (with
respect given standards, XQ and aQ because they will not be used for
regulatory monitoring. The warning limits are to be used because their use
would generally signal a process or maintenance procedure change before the
upper and lower control limits would.
B-4
-------
TABLE B-3. POTROOM GROUP WARNING LIMITS, ALUMAX DATA
= __.= s;_3; = = =
X Chart3
Chart
Potroom Group
101G/161W
101H/161E
103G/162W
103H/162E
Central .
Line, XQb
0.8473
0.9162
0.8070
0.9618
UWL
1.043
1.133
0.972
1.177
LWL
0.652
0.700
0.642
0.747
Central
Line,
-------
APPENDIX C
HYPOTHESIS TESTING
In hypothesis testing the null hypothesis defines what test is being
conducted. The hypothesis always relates to true values, or population
values. The test statistic is a calculated value used to test the
hypothesis. The test statistic uses sample parameters, as contrasted to
population parameters.
For example, the true mean (of a population) is generally symbolized as
u, whereas the mean of a sample is symbolized as X.
In the hypothesis test needed in the application used for this report,
the hypothesis to be tested is:
H • u, > u ,
0 1 0
where UQ is the true mean value of the monthly mean emissions used to
qualify a potroom group for a less frequent performance test schedule, and
Uj is the true mean of a set of emission measurements (8 in the examples in
this study) taken after the upper warning limits have been violated on the X
control chart for a given potroom group. The judgement that needs to be
made is whether the overall emissions level (the true population mean, ju.)
has increased from the original level JLI .
In making a decision based on a hypothesis test, two types of mistakes
are possible.
a. the decision can be made that the hypothesis is correct, when in
fact it is not; and
b. the decision can be made that the hypothesis is incorrect, when in
fact it is true.
C-l
-------
The test is constructed to minimize the probability of making the second
type error, at what is taken as , the level of significance. For most
cases, is taken as 0.05, indicating a 5 percent probability of making an
error of type b above. That is, there would be only a 5 percent error that
the hypothesis is rejected when in fact it is correct. The hypothesis test
is outlined below:
Hypothesis: HQ : Uj > UQ
Test Statistic:
- (Xj - X0) / [sp
- 1) Sj2 * (n0 - 1) s2Q)
with df = (PJ + nQ - 2)
Critical Values: ^,0.95'
If t > tdf 0>g5, then accept HQ, otherwise reject H . Critical value
*df 0.95 1s student's t statistic, one-tailed at the 0.05 level of
significance with df degrees of freedom.
Example:
Table 5-3 gives the 56 values of Xj (XBAR) on which X is based for
Potroom Group 101G/161W.
XQ = X = 0.8473
S- = Sn = 0.1520
A 0
As given in Appendix D, UWL for the X chart = 1.043. Suppose the following
consecutive monthly mean performance tests were conducted:
Xj = 1.0496
X2 = 0.9224
X3 = 0.8090
C-2
-------
X4 = 0.8857
X5 = 1.0524
Xg = 0.8357
X7 = 1.0270
Xg = 0.9063
For the last 8 monthly mean emissions:
Xj = the mean of X,, X^, X.,, ..., Xg, and
Sj = the standard deviation
r^ - 8
Therefore, Xj = 0.9360
Sj = 0.0960
nj - 8
The question, is "Has the true mean emission level from which the last
8 samples were taken changed from the original level where X = 0.8473,
SQ = 0.1520, and nQ = 56?"
Therefore, test the hypothesis:
Hn : u. > u
0 '1 '0
Test Statistic:
t = (0.9360 - 0.8473) / [s ^1/56 + 1/8],
sp = [((55(0.1520)2 + 7(0.0960)2) / (56 + 8-2)]1/2
s = 0.14675
t = 1.5992
df = 62
^2,0.95 '
C-3
-------
Because t is less than tg2)(h95, the null hypothesis, HQ must be rejected.
That is to say, that there is not a significant difference between 0.9360
and 0.8473 at the 5 percent level of significance. The conclusion that must
be reached is that from the data we have collected, we can not say that the
overall mean has increased for the last 8 runs.
C-4
-------
APPENDIX D
SUMMARY OF PRIMARY ALUMINUM PLANT PERFORMANCE
TABLE D-l. TOTAL FLUORIDE EMISSIONS FROM POTROOMS SUBJECT TO NSPS
Plant
Code
A
B
C
D
E
Plant
Type
CWPBa
CWPB
CWPB
CWPB
vssb
Number
Monthly
Tests
51
48
34
16
22
Measured Emissions
Ib F/T Al
M1n.
0.55
0.42
0.67
0.71
0.88
Max.
1.30
1.31
3.48
1.49
3.11
Avg.
0.89
0.86
1.27
1.04
1.49
NSPS
Limit
Ib F/T Al
1.9
1.9
1.9
1.9
2.0
Number of
Exceedances References
0 16
0 16
3C 17,18
0 19
3d 20
aCWPB - Plant uses center-worked prebake pots.
bVSS - Plant uses vertical stud Soderbrg pots.
cTwo failures occurred In same month, one on retest.
Specific reason for failures not reported. Plant conducts one test run/line each month,
so each reported test is the average for all three lines.
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO. 2.
EPA-450/3-86-012
4. TITLE AND SUBTITLE Primary Aluminum NSPS :
Statistical Analysis of Potline Fluoride Emissions
and Alternate Sampling Frequency
7. AUTHOR(S)
&. PERFORMING ORGANIZATION NAME AND ADDRESS
Office of Air Quality Planning and Standards
Environmental Protection Agency
^ Research Triangle Park, NC 27711
12. SPONSORING AGENCY NAME AND ADDRESS
DAA for Air Quality Planning and Standards
Office of Air and Radiation
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
October 1986
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPOR T '-, ..
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-02-3816
13. TYPE OF REPORT AND PERIOD COVE==. "
Final
14. SPONSORING AGENCY CODE
EPA 200/84
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Statistical analyses were performed on 4 years of fluoride emissions data
from a primary aluminum reduction plant. These analyses were used to develop
formulae and procedures for use by regulatory agencies in determining alternate
sampling frequencies for secondary (roof monitor) emissions testing on a case-
by-case basis. Monitoring procedures for ensuring compliance even with a reduced
test frequency are also addressed.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
Air pollution
Aluminum industry
Fluorides
Standards of performance
18. DISTRIBUTION STATEMENT
Unlimited
b.lDENTIFIERS/OPEN CiMD^D TERMS
Air Pollution control
19. SECURITY ClAS-/77iu Report)
Unclassified
20. SECURITY CLASS (This page)
Unclassified
c. COSATI Field/Group
13B
21. NO. OF PAGES
172
C2. PrilCE
EPA Form 2220-1 (R«v. 4-77) PREVIOUS EDITION is oeso .ETE
-------
INSTRUCTIONS
1. REPORT NUMBER
Insert the EPA report number as- it appears on the cover of the publication.
2. LEAVE BLANK
3. RECIPIENTS ACCESSION NUMBER
Reserved for use by each report recipient.
4. TITLE AND SUBTITLE
Title should indicate clearly and briefly the subject coverage of the report, and be displayed prominently. Sot subtitle, if used, in
type or otherwise subordinate it to main title. When a report is prepared in more than one volume, repi-al (lie primary title. add vubmnr
number and include subtitle for the specific title.
5. REPORT DATE
Each report shall carry a date indicating at least month and year. Indicate the basis on which il «.is selected ft:*.. Jalc i>< man: .
approval, date of preparation, etc./.
6. PERFORMING ORGANIZATION CODE
Leave blank.
7. AUTHOR IS)
Give name(s) in conventional order (Joftn R. Doe, J. Robert Dov. etc.). List author's aliilution it it Jitters Iroin the pcriuunim:
zation.
8. PERFORMING ORGANIZATION REPORT NUMBER
Insert if performing organization wishes to assign this number.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Give name, street, city, state, and ZIP code. List no more than two levels of an orpam/uiiunal IJrcarc^y.
10. PROGRAM ELEMENT NUMBER
Use the program element number under which the report was prepared. Subordinate numbers may be mcliiUeil in parent hoes.
11. CONTRACT/GRANT NUMBER
Insert contract or grant number under which report was prepared.
12. SPONSORING AG3NCY NAME AND ADDRESS
Include ZIP code.
13. TYPE OF REPORT AND PERIOD COVERcD
Indicate interim final, etc., and if applicable, dates covered.
14. SPONSORING AGbNCY CODE
Insert appropriate code.
15. SUPPLEMENTARY NOTES
Enter information not included elsewheie but useiul, such as: Prepared in cooneration with, translation <>l, I'revnleii ai conlin-mu <••«.
To be published in, Supersedes, Supplements, etc.
16. ABSTRACT
Include a brief f2CO \vords or les*) factual :umr.iary of the Most significant information Contained in tin report. II the r-noct <. '««JH»* j
significant bibliography o* literature survey, mention it here.
17. KEY WORDS AND DOCUMENT ANALYSIS
(a) DESCRIPTORS - Select from the Thesaurus of Engineering and Scicntilic Terms the propc; autliori/cd ICMIIS thai identity iht-mj.;^
concept of the research and are sufficiently specific ind p'ecisc to be used as index entries lor cataloging.
(b) IDENTIFIERS AND OPEN-ENDED TERMS - Use identifiers for project names, code names, cquiprn.nl designators, etc. Use
ended terms written in descriptor fcum for those subjects for which no descriptor exists.
(c) COSATI MELD GROUP - Field and g/oup assignments are to be taken from the 1965 TOSA11 Subject <. jtepory Li:.:. Since ihc
jority of documents are multidijciplinary in nature, the Primary Held/Group assignment* s) will be specific discipline, area of humaa
endeavor, or type of physical object. The application(s) will be crosvrclerencc-1 with secondary I icIuVdrnup .ssignments tlut will s
the primary posting! s).
18. DISTRIBUTION STATEMENT
Denote releasability to the public or limitation for re"sonJ other than security for example "Release Unlimited." ( i c any availahiiwy «<>
the public, with address and pnce.
19. & 20. SECURITY CLASSIFICATION
DO NOT submit classified reports to the National Technical Infer ma tic n service.
21. NUMBER OF PAGES
Insert the total number of pages, including this one ind unnumbered pages, but exclude distribution list, ;l any.
22. PRICE
Insert the price set by the National Technical Information Service or the Government Printing Office, if known.
EPA Form 2220-1 (R.v. 4-77) (R.v.r,-)
------- |