EMB Report'No. 77SPP23A
AIR POLLUTION
EMISSION TEST
O
VOLUME I: FIRST INTERIM REPORT:
CONTINUOUS SULFUR DIOXIDE MONITORING
AT STEAM GENERATORS
August, 1978
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Waste Management
Office of Air Quality Planning and Standards
Emission Measurement Branch
Research Triangle Park. North Carolina
-------
EMB REPORT NO. 77SPP23A
August 1978
AIR POLLUTION EMISSION TEST
Volume I: First Interim Report
Continuous Sulfur Dioxide
Monitoring at Steam Generators
by
W. E. Kelly and C. Sedman
Emission Measurement Branch, ESED
U. S. Environmental Protection Agency
Research Triangle Park, N. C. 27711
and
J. R. McKendree and R. P. Quill
Monsanto Research Corporation
Dayton Laboratory
Dayton, Ohio 45407
Contract 68-02-2818
Work Assignment 2
Prepared for
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
Office of Air Quality Planning and Standards
Emission Standards and Engineering Division
Research Triangle Park, North Carolina 27711
-------
TABLE OF CONTENTS
List of Figures v
List of Tables vii
Summary of Results ix
Sections
I. INTRODUCTION AND DESCRIPTION OF TEST SITES 1
A. Introduction 1
B. Description of Test Sites , 2
1. Cane Run Unit No. 4, Louisville Gas & 2
Electric Co., Louisville, Ky.
2. Bruce Mansfield Unit No. 1, Pennsylvania 4
Power Company
3. La Cygne No. 1, Kansas City Power and 6
Light Co.
4. Eddystone Unit No. 1, Philadelphia 8
Electric Co.
II. DATA GATHERING AND REDUCTION 11
A. Data Gathering Systems 11
1. Cane Run No. 4, Louisville Gas and 11
Electric Company
2. Bruce Mansfield Unit 1, Pennsylvania 15
Power Company
3. La Cygne Unit #1, Kansas City Power and 19
Light Company
4. Eddystone No. 1, Philadelphia Electric 22
Company
B. Data Reduction 24
C. Improvements and Research Needed 34
III. DATA ANALYSIS 37
A. Data Collection Interval Selection 37
B. Data Calculation Assumptions 40
C. Unusable Data 42
D. Usable Data and Statistical Analyses 42
IV. RECOMMENDATIONS AND CONCLUSIONS 61
ill
-------
LIST OF FIGURES
Figure Page
1-1 Cane Run Station No. 4 Scrubbing Plant Gas Flow 3
1-2 Bruce Mansfield Station No. 1 Scrubbing System 5
Gas Flow
1-3 La Cygne Station No. 1 Scrubbing Plant Gas Flow 7
1-4 Eddystone Station No. 1 Scrubbing Plant Gas Flow 9
II-l Instrument System Schematic, Cane Run No. 4 12
Louisville Gas and Electric Company
II-2 Instrument System Schematic, Bruce Mansfield No. 1 16
Pennsylvania Power Company
II-3 Instrument System Schematic, La Cygne No. 1 20
Kansas City Power and Light Company
II-4 Instrument System Schematic, Eddystone No. 1 23
Philadelphia Electric Company
II-5 Sample Strip Chart Data 27
II-6 Sample Data Transcription Form 28
II-7 Sample Data Listing 30
II-8 Sample Average Listing and Statistical Summary 31
III-l Inlet SO2 Variation, Cane Run No. 4 38
III-2 Outlet S02 Variation, Cane Run No. 4 39
III-3 Probability Vs. Linear inlet S02 Concentrations 45
from Louisville Site
III-4 Probability Vs. Log Inlet S02 Concentrations 46
III-5 Probability Vs. Linear Outlet S02 Emissions 47
from Louisville Site
III-6 Probability Vs. Log Outlet S02 Emissions 48
III-7 Probability Vs. Percent S02 Removal 52
III-8 Probability Vs. 100 Minus Percent S02 Removal 53
III.-9 Louisville Data for 7 and 14 Day Averaging Periods 59
-------
LIST OF TABLES
Table Page
III-l Categorization of Data 43
III-2 Inlet S02 Emissions (24-Hr. Averaging Period) 49
III-3 Outlet S02 Statistics (24-Hr. Averaging Period) 51
III-4 S02 Removal Statistics (24-Hr. Averaging Period) 54
III-5 Variation in S02 Concentrations in lb/106 BTU 56
(24-hour averages)
III-6 Cumulative Louisville Statistics Vs. Averaging Time 58
vii
-------
SUMMARY OF RESULTS
A program was begun in May, 1977, to acquire S02 monitoring data
in support of possible revisions to the New Source Performance
Standard for fossil fuel fired steam generators, originally pro-
mulgated in December, 1971. Four sites were chosen for monitor-
ing and the monitoring equipment was installed and tested. Data
gathering began during July-September, 1977 at all sites and has
continued through early 1978. Monitoring was discontinued at
one site in October, 1977, with little useful data accumulated.
Monitoring at another site was discontinued in December, 1977,
after yielding only 85 days of data.
Data from the two remaining sites and data from another site ob-
tained through other EPA programs were analyzed along with the
15 days data from one abandoned site. Mean S02 emissions ranged
from 0.23 to 1.22 Ib/Mbtu while mean S02 removal efficiencies
ranged from 81.4 to 96.0 percent. Analyses of the variability
indicate that:
• Dampening of variable S02 emissions by scrubbing does
occur, especially with more reactive absorbents and
good pH control.
• The variability of outlet S02 emissions is substan-
tial, such that caution should be used in developing
emission standards "never to be exceeded".
• Longer averaging times decrease the expected variabil-
ity significantly.
IX
-------
Insufficient data have been collected to examine the effects of
long averaging period,, i.e., 30 days or -longer. The monitoring
systems used to generate these data were not the type systems
which would be recommended by EPA to gather data in such a study,
rather the systems were simply adaptations of existing on-site
S02 monitoring equipment to generate data at minimum delay and
cost. As a result, the reliability of these systems during this
study does not reflect the current capability of SC>2 monitors and
data retrieval systems.
-------
I. INTRODUCTION AND DESCRIPTION OF TEST SITES
A. INTRODUCTION
The objective of this program was to obtain continuous monitoring
data for sulfur dioxide emission rates and removal efficiencies
for flue gas desulfurization (FGD) systems. The primary use of
this data was to perform statistical analyses on variations as a
function of averaging time. These analyses, along with other in-
formation, were used to determine the appropriate averaging per-
iod for the revised standard of performance for new coal-fired
steam generators.
Sulfur dioxide and oxygen concentrations were measured upstream
and downstream of the FGD system using continuous instrumental
monitors. Copies of boiler operation logs and FGD system opera-
tion logs were obtained to document process operating conditions
during the monitoring periods.
All instruments were subjected to the appropriate performance
specification test procedures1 to assure the accuracy of the
sulfur dioxide and oxygen measurements.
The monitoring program included testing at four steam generators.
These were: (a) Cane Run Unit No. 4, Louisville Gas and Elec-
tric Co., Louisville, Kentucky, (b) Bruce Mansfield Unit No. 1,
Pennsylvania Power Co., Shippingport, Pennsylvania, (c) La Cygne
Unit No. 1, Kansas City Power and Light Co., La Cygne, Kansas,
federal Register, Volume 40:194, October 6, 1975.
-------
and (d) Eddystone Unit No. 1, Philadelphia Electric Company,
Philadelphia, Pennsylvania.
Due to the magnitude of the program, four contractors were in-
volved. Monsanto Research Corporation coordinated the overall
program and processed the collected data. Midwest Research In-
stitute, Scott Environmental Technology, Inc., and York Research
Corporation had the responsibility of setting up monitoring
equipment, performing the instrument performance specification
tests, conducting routine calibrations and maintenance, and for-
warding all data to Monsanto for processing.
B. DESCRIPTION OF TEST SITES
The four FGD systems serving steam generators included in this
monitoring program are generally of different design and reactant
types. These systems were designed several years ago (e.g.,
Bruce Mansfield designed in 1971) and, therefore, do not incorpo-
rate the latest design features expected at new installations.
Additionally, the size and type of the steam generators are dif-
ferent. Further description of the individual sites is presented
below.
1. Cane Run Unit No. 4, Louisville Gas and Electric Co.,
Louisville, Kentucky
Unit No. 4 at the Cane Run Station has an electric generating
capacity of 181 megawatts. The boiler flue gases are passed
through two parallel electrostatic precipitators for particulate
removal and then to an FGD system designed by American Air Filter
Corporation. The FGD system is divided into two modules (see
Figure 1-1). In each module, the flue gases first pass through
a quench section for temperature reduction, then through a mobile-
bed absorber equipped with sprays, then through a set of Chevron
-------
CANE RUN STATION NO. 4 SCRUBBING PLANT GAS FLOW
UJ
BALANCED DRAFT FAN
QUENCHER
REHEATERS BEING INSTALLED HERE
BOILER
ELECTROSTATIC
PRECIPITATOR
INLET SAMPLING
i nrATinws
HJvrM 1 wIM J
ELECTROSTATIC
PRECIPITATOR
OUTLET SAMPLING
LOCATIONS
DAMPERS
EMISTER
SCRUBBER
MODULE
SUPPLY TO
REACTOR
RETURNJOLUTION
TO REACTOR
DEMISTER
1
QUENCHER
i
SCRUBBER
MODULE
•SUPPLY TO
REACTOR
BALANCED DRAFT FAN
RETURN SOLUTION
TO REACTOR
Figure 1-1
-------
demisters, also equipped with sprays, and then through a direct,
oil-fired stack gas reheater near the stack breechings.
Carbide lime slurry is used as the sulfur removal reactant.
The FGD system was retrofitted to an existing boiler with an
electrostatic precipitator particulates control system. The
scrubbing modules are equipped so that stack gases may be by-
passed.
The sulfur dioxide and oxygen measurement locations at the FGD
system inlet are downstream of the precipitators while the outlet
measurement locations are in the stack breechings. Monitoring
was started in July, 1977.
2. Bruce Mansfield Unit No. 1, Pennsylvania Power Company
Unit No. 1 at the Bruce Mansfield Station has an electric generat-
ing capacity of 825 megawatts. The boiler flue gases are passed
directly to the FGD system. The system is a tandem venturi-
absorber type designed by Chemico. The FGD system is divided in-
to six venturi-absorber modules, or trains (see Figure 1-2). The
boiler flue gases pass from the boiler into a common manifold for
distribution to the six modules. The flue gas is first contacted
with scrubbing slurry in the venturi section for particulates re-
moval, then additionally contacted in the absorber section for
sulfur dioxide removal. The gases then pass through direct oil-
fired gas reheaters prior to entering the stack.
Thiosorbic lime slurry is used as the sulfur dioxide removal re-
actant.
The FGD system was designed and constructed at the same time as
the boiler, so that the entire system is an integral unit. No
provisions are included for direct bypass of flue gasses around
-------
BRUCE MANSFIELD STATION NO. 1 SCRUBBING SYSTEM GAS FLOW
CO
<
o
o
CQ
INLET
SAMPLING
LOCATION
OUTLET SAMPLING
LOCATIONS
STACK
(4 FLUES)
DAMPERS
Figure 1-2
-------
the FGD system. Any combination of the six modules may be used
under normal circumstances should difficulties be encountered in
a module. However, if a module is taken out of service, the
boiler output must be reduced.
One of the Unit 1 chimney flues was undergoing liner repairs dur-
ing the test period. Therefore, the unit was operated at half-
load with only three modules available.
Partial monitoring began in late August, 1977, with the first
complete data beginning in mid-September.
3. La Cygne No. 1, Kansas City Power and Light Co.
Unit No. 1 at the La Cygne station is a cyclone-fired boiler
which drives a 820 megawatt generator. The boiler flue gases are
passed directly to the FGD system. The system is comprised of
eight tandem venturi-absorber modules (see Figure 1-3). The FGD
system and the boiler were designed and constructed as a complete
package. Each module can be isolated from the system for individ-
ual maintenance. The module outlets lead to a common manifold
and then to six induced draft fans. The outlets of the fans are
combined into two ducts leading to the stack. Stack gas reheat
is by indirect steam heat exchange, supplemented by hot air ad-
dition via a side stream from the combustion air preheaters.
Limestone slurry is used as the sulfur dioxide removal reactant.
Monitoring instrumentation was installed in early August, 1977,
and partial data was collected through November 5, 1977. Because
of instrument operation difficulties an4. atypical FGD system
operation, only a limited amount of data was obtained.
-------
LA CYGNE STATION NO. 1 SCRUBBING PLANT GAS FLOW
o
QQ
AIR
HEATER
INLET SAMPLING
L0(
NATION " A "
AIR
HEATER
I CO MANIFOLD |
Ji
--r^
D
SCRUBBER
-F
C
SCRUBBER
T -f-
H
SCRUBBER
T
G
SCRUBBER
DAMPERS.
/I \\
> *F
B
SCRUBBER
T *-\-
F
SCRUBBER
}_., i
" -f-
A
SCRUBBER
E
SCRUBBER
REHEATER
REHEATER
REHEATER
REHEATER
_*
— H
, ^
* -x:
-ffBN
^/ " B
^
OUTLET SAMPLING
LOCATIONS •»
11 C
" <
(ST;
i
* ii ^^a
# ^^
/TD\
[(^]
(ffw)
*(™}
^CIC
r^
Figure 1-3
-------
4. Eddystone Unit No. 1, Philadelphia Electric Company
Unit No. 1 at the Eddystone station is rated at 325 megawatts
generating capacity. The boiler flue gases are passed through a
mechanical collector/electrostatic precipitator section for par-
ticulates removal, then to the FGD system. The total flue gas
flow is split into three streams, each equipped with a venturi
scrubber where the flue gases are contacted with river water for
additional particulate removal. One of the three venturi scrub-
bers is followed by a sulfur dioxide removal absorber. Magnesium
oxide slurry is used as the sulfur dioxide removal reactant.
Testing in this program has been limited to the total venturi-
absorber train. Stack gas reheat is accomplished by direct oil
firing.
^
The FGD system was retrofitted to the existing Unit No. 1. Flue
gas bypass is provided in order to protect the boiler on Unit 1.
Monitoring instrumentation was operational since mid-August, 1977,
and testing was discontinued in late December, 1977. However,
only a small number of days of data were obtained due to limited
concurrent operating time of the boiler, the sulfur dioxide re-
moval absorber, and the instrument systems.
-------
EDDYSTONE STATION NO. 1 SCRUBBING PLANT GAS FLOW
BOILER
MECHANICAL
COLLECTOR
ELECTROSTATIC
PRECIPITATOR
^
A
X_j
*
B
Y—
A"
C
PARTICULATE
SCRUBBER
PARTICULATE
SCRUBBER
DADTTPIII ATT
ft
J L
1 r
X—
J L
1 r
PCUCATCD
DCLirAlTD
INLET SAMPLING
LOCATION " A "
SCRUBBER
OUTLET SAMPLING
LOCATION " B "
FAN
FAN
LA- S02 SCRUBBER -ft^T- REHEATER
FAN
BY-PASS DAMPERS
FOR EACH INDIVIDUAL TRAIN
-e/-*
TO STACK
Figure 1-4
-------
II. DATA GATHERING AND REDUCTION
A. DATA GATHERING SYSTEMS
The sulfur dioxide monitoring equipment at all test locations
were DuPont Model 460 Photometric analyzers2. The basic instru-
ment configuration varied from site to site and is discussed be-
low. The oxygen monitoring equipment varied from site to site;
however, the basic detection principle was the same. The oxygen
measurement equipment is described in detail in the site descrip-
tions below. Continuous instrumental moisture measurements were
not performed. The moisture contents of the sample streams were
determined by manual procedures, as discussed for each test site.
1. Cane Run No. 4, Louisville Gas and Electric Company
A schematic diagram of the monitoring system used at this facil-
ity is given in Figure II-l. The sulfur dioxide instrument is a
single DuPont 460 four-point, dual range analyzer. The two
scrubber module inlet concentrations are measured on a 0-4000 ppm
range and the two module outlets after reheating are measured on
a 0-500 ppm range. In the event that the module outlet concen-
trations exceed 500 ppm, the measurement range can be manually
switched so that all points are measured on the 0-4000 ppm range.
The instrument operates on approximately a 14-minute cycle. Dur-
ing this cycle, each of the four measurement points is sampled.
An automatic instrument zero and sample probe backflush occurs
2Mention of a specific company or product name does not consti-
tute endorsement by the Environmental Protection Agency.
11
-------
Figure II-l: Instrument System Schematic, Cane Run No. 4
Louisville Gas and Electric Company
-------
immediately prior to each measurement in the cycle. The concen-
tration measurement result is printed on a multipoint chart re-
corder in the scrubber control room. This concentration record
represents the instantaneous value at the time the recorder
printer is activated. Therefore, a single instantaneous result
is available for each of the four locations in each 14-minute
period. A secondary single-pen continuous trace recorder was
used as a backup for the control room primary recorder.
The oxygen measurement equipment consisted of a Scott Model 150
paramagnetic analyzer3. The sample for oxygen measurement was
withdrawn from the Dupont instrument downstream of the sulfur di-
oxide measurement cell. Therefore, the oxygen measurements were
performed on the same sample stream used for sulfur dioxide mea-
surement. Initially, the oxygen analyzer was connected directly
to the Dupont analyzer without any further sample conditioning
prior to oxygen measurement. The system was later modified to
incorporate a condenser for moisture removal prior to oxygen mea-
surement. Oxygen concentrations were determined on a 0-25 per-
cent by volume range. The measurements were recorded on a
single-pen continuous trace recorder. For each 14-minute instru-
ment cycle, approximately 2-1/2 minutes of continuous trace
record was available for each of the four sample points.
The moisture content of the sample stream was determined by man-
ual techniques. Moisture tests were performed on the scrubber
inlet and outlet streams, and on the analyzer sample stream im-
mediately following the knockout trap in the DuPont analyzer.
The temperature of the sample stream(s) was also measured. It
was found that the sample streams are cooled to approximately
10°C above ambient after the trap. Because of this cooling, the
samples were not at stack conditions and were assumed to be
3Mention of a specific company or product name does not consti-
tute endorsement by the Environmental Protection Agency.
13
-------
saturated at the temperatures measured at the knockout traps.
Early in the test program, before the incorporation of a dryer,
the moisture content of the oxygen sample stream was the same as
that for the sulfur dioxide sample stream. After incorporation
of the dryer, the oxygen measurement sample was dry.
A performance specification test was performed on the sulfur di-
oxide and oxygen instrument system by Scott Environmental Tech-
nology, Inc. **
The major problem areas encountered during the monitoring program
at this facility were interfacing for and measurement of oxygen.
It was necessary to modify the sample extraction apparatus and
the analyzer itself several times before a satisfactory combina-
tion was obtained. The oxygen analyzer was out of service the
majority of the period between August to November. While the
equipment configuration operated satisfactorily, daily inspection
and maintenance were necessary to insure operation.
The only other major problem encountered at this facility was the
consistently high oxygen concentrations measured at the scrubber
module outlets. It was originally concluded that there was leak-
age past the blowback air valves installed on the sample probes
at these locations. However, later experiments, after replace-
ment of the air valves failed to correct the problem, showed that
the placement of the probes was the difficulty. A damper is
located upstream of the probe and turning vanes in the duct are
located immediately after the probe. These physical configura-
tions tend to stratify the flow into the lower half of the duct,
especially at low boiler load conditions. The sample probes ex-
tended only four feet into the duct from the top, causing the
4EMB Project No. 77-SPP-20, "First Interim Report - Continuous
S02 Monitoring at Cane Run No. 4, Louisville Gas and Electric
Co.". Scott Environmental Technology, December, 1977.
14
-------
sample to be extracted from an eddy or low flow zone. This prob-
lem was corrected by extending the probes to about ten feet into
the duct, so that samples were always extracted from the main
flow zone in the duct.
Clogging of the sample probes was not a major problem at this fa-
cility. Particulates are removed by electrostatic precipitators
prior to the inlet sampling locations and the stack gas reheaters
prevent wet gas conditions at the outlet sampling locations. The
brief outages caused by sample line plugging could have been
avoided by using a modified automatic cleaning procedure.
2. Bruce Mansfield Unit 1, Pennsylvania Power Company
A schematic diagram of the monitoring system used at this facil-
ity is presented in Figure II-2. The sulfur dioxide instrument
used at the inlet manifold is a DuPont Model 460 two-point ana-
lyzer. The concentration measurement range is 0-5000 ppmv. The
instrument operates on a 10-minute cycle, where each of the two
measurement points is sampled for 5 minutes. Immediately prior
to each measurement, an automatic instrument zero and probe back-
flush occurs. The concentration measurement result is recorded
on a continuous trace chart recorder in the scrubber control
room. For each 10-minute cycle approximately 4-1/2 minutes of
continuous chart record are available for each inlet location.
The FGD system outlet is equipped with two DuPont 460 four-point,
dual range analyzers. Each of these analyzers measures three
module outlets and the common reheater outlet. Concentration
measurements are normally performed on a 0-500 ppm range, but the
range can be manually switched to 0-1000 ppm should 500 ppmv be
exceeded. Each instrument normally operates on a 12-minute cy-
cle., where each point is monitored for 3 minutes, with an auto-
matic zero and probe backflush occurring prior to each measurement.
15
-------
A TRAIN OUTLET
B TRAIN OUTLET
C TRAIN OUTLET
E TRAIN OUTLET
D TRAIN OUTLET
F TRAIN OUTLET
Figure II-2: Instrument System Schematic, Bruce Mansfield No. 1
Pennsylvania Power Company
-------
However, for this program, the three module outlets were not mon-
itored.
The concentration measurement result is printed on a multipoint
chart recorder in the scrubber control room. This concentration
record represents the instantaneous value at the time the re-
corder printer is activated. Therefore, a single instantaneous
value was available for each 3-minute period. A secondary single-
pen continuous trace recorder was used as a backup for the pri-
mary control room recorder.
The oxygen measurement equipment at each location consisted of a
Taylor OA1375 paramagnetic analyzer, with a chiller condenser
system for sample gas drying. The sample for oxygen measurement
was withdrawn from the DuPont instrument downstream of the mea-
surement cell. The oxygen measurements were conducted on the
same sample that the DuPont instrument used for sulfur dioxide
measurement. Oxygen concentrations were determined on a 0-25
percent by volume range. The measurements were recorded on a
single-pen continuous trace recorder at each inlet and outlet in-
strument. At the inlet location, oxygen results were available
for 4-1/2 minutes at each test point for each 10-minute cycle.
At the outlet locations, oxygen data were available for 2-1/2 ,
minutes of each 3-minute cycle. A performance specification test
was performed on the sulfur dioxide and oxygen instrument systems
by York Research Corporation6.
The moisture contents of the sample streams were determined by
manual procedures. Tests were performed at the inlet and outlet
locations, and immediately after the knockout trap in the DuPont
5Mention of a specific company or product name does not consti-
tute endorsement by the Environmental Protection Agency.
6EMB Project No. 77-SPP-19, "First Interim Report - Continuous
S02 Monitoring Program at Bruce Mansfield No. 1, Pennsylvania
Power Company", York Research Corporation, January, 1978.
17
-------
analyzer. It was found that the moisture content of the sample
stream after the knockout trap was significantly less than flue
gas moisture content, therefore, the measured results at the in-
strument were used for calculation purposes. Since dryers were
incorporated in the oxygen sample conditioning systems, all
oxygen measurements are on a dry basis.
The major problems encountered at this facility were the measure-
ment of oxygen and plugging of sample probe filters. The oxygen
interface system (pump, chiller, condenser, valving) required a
very high maintenance effort to correct plugging and leaks. The
chiller system for the gas condenser had numerous breakdowns.
The oxygen analyzers also failed due to electronic malfunction.
After many modifications to the interface systems, the oxygen
analyzer at the inlet location was operated with a much reduced
maintenance requirement. The paramagnetic analyzer system at the
ABC reheater outlet was never brought to an acceptable operating
reliability level. This analyzer was replaced by a Thermox7
Model WDG Zirconium oxide cell instrument. After this modifica-
tion, there were no further difficulties with oxygen measurement
at that location.
A second problem has been severe plugging problems at the outlet
sampling location. During the test program there were problems
with the absorber mist eliminator and the stack gas reheaters
were not operated, resulting in wet gas conditions. The wet
carry-over from the scrubber modules quickly caked on the sample
probe filters and could not be cleaned by an air backflush. At
times, the operating period between required filter cleaning was
only two to three hours. A probe backflush system incorporating
a combination of hot water and air was fabricated and installed.
7Mention of a specific company or product name does not consti-
tute endorsement by the Environmental Protection Agency.
18
-------
This system operated for two weeks with success until water sup-
ply lines began freezing due to cold weather.
The oxygen measurement and probe plugging problems were compounded
by the physical layout of the monitoring systems at this facility.
The control stations and recorders for the sulfur dioxide instru-
ments are located in the control room approximately one hundred
yards from the analyzer cabinets. As a further complication, the
sampling probes are located about one-hundred yards away from and
150 feet above the analyzer cabinets. Because of these physical
separations, two men were required to rigorously calibrate and
service the systems.
No major problems were encountered with the operation of the sul-
fur dioxide monitors other than sample interfacing. A non-
linear response was identified at the inlet analyzer during the
specification tests and was repaired.
3. La Cygne Unit #1, Kansas City Power and Light Company
A schematic diagram of the equipment used at this facility is
given in Figure II-3. The inlet sulfur dioxide instrument is a
single-point, dual-range DuPont Model 460 analyzer. Concentra-
tion measurements are normally made on a 0-5000 ppm range but can
be manually switched to a 0-10,000 ppm range when 5000 ppm values
are exceeded. The instrument operates on a 10-minute cycle, with
a high pressure probe backflush and an automatic zero every cycle.
The measurement result is recorded on a continuous trace chart
recorder. A Beckman8 Model F3 paramagnetic oxygen analyzer was
originally installed on the sulfur dioxide sample stream exhaust.
However, this arrangement did not function reliably due to inter-
face problems. The system was modified to provide a separate
8Mention of a specific company or product name does not consti-
tute endorsement by the Environmental Protection Agency.
19
-------
to
o
FGD SYSTEM INLET MANIFOLD
V02 PROBE.
\S02 PROBE"
Figure II-3
Instrument System Schematic, La Cygne No. 1
Kansas City Power and Light Company
-------
probe and sample line system for the oxygen analyzer. This ar-
rangement operated satisfactorily.
The outlet sulfur dioxide monitor was a DuPont Model 460 two-
point, dual-range analyzer. The concentration measurements were
normally conducted on a 0-1500 ppm range, with manual switching
capability to a 0-3000 ppm range when the low range was exceeded.
The analyzer operated on a 10-minute cycle, with each sample
point being analyzed for 5 minutes each cycle. A probe backflush
and automatic instrument zero occurs each cycle. The concentra-
tion results were recorded on a continuous trace recorder.
A successful oxygen measurement system installation was not com-
pleted at the outlet location. After several failures, it was
decided that a separate probe and sample line system would be
necessary. This system was not installed due to test termination
at this facility.
The moisture contents of the sample streams were determined by
manual procedures. At the inlet, the flue gas moisture content
was measured. At the outlet, the sample stream temperature im-
mediately after the knockout trap was measured and saturated con-
ditions at that temperature were assumed. These moisture contents
were always less than those measured at the outlet flue gas ducts.
A performance specification test was conducted on the sulfur di-
oxide instrument at the outlet location by Environmental Protec-
tion Agency personnel.9 Tests were begun at the inlet analyzer
system, but it was determined that repairs to the DuPont instru-
ment were necessary. The performance of the analyzer was signif-
icantly improved after replacement of several tubes in the
3"First Interim Report, La Cygne, Kansas Power Plant S02 Con-
tinuous Monitoring Study", EMB Project No. 77-SPP-22, Environ-
mental Protection Agency, January, 1978.
21
-------
photometric detector. A retest of the inlet analyzer was not
completed due to test program termination. ,
The only major problem that was encountered other than oxygen
measurement was plugging of the sample probe filters at the out-
let location. This was due to the wet condition of the stack
gas.
4. Eddystone No. 1, Philadelphia Electric Company
A schematic diagram of the monitoring system used at this facil-
ity is given in Figure II-4. The sulfur dioxide measurement in-
strument is a single DuPont Model 460 dual-range, five point
analyzer. The FGD system inlet S02 concentration is measured on
a 0-5000 ppm range, while the system outlet concentration is mea-
sured on a 0-500 ppm range. The range is automatically switched
depending on the sample being analyzed. The instrument normally
operates on a 20-minute cycle with a four-minute analysis at each
of the five points. For this program, the instrument was ad-
justed to measure at only three points.
The concentration measurement result is printed on a multipoint
chart recorder in the scrubber control room. This record repre-
sents the instantaneous concentration when the recorder printer
is activated. Therefore, a single instantaneous result is avail-
able for each location in each 12-minute period.
Continuous oxygen measurement was not conducted at this facility.
Due to the intermittent operating nature of the FGD system and
the problems previously encountered with oxygen measurements,
manual procedures were used. The moisture content of the sample
streams were determined by manual measurements at each sample
site.
22
-------
S02 SCRUBBER OUTLET
to
uJ
1-C SCRUBBER INLET
£ TO OTHER SYSTEM SAMPLE POINTS
Figure II-4: Instrument System Schematic, Eddystone No. 1
Philadelphia Electric Company
-------
An instrument performance specification test was performed by
Scott Environmental Technology, Inc.10
The major problem encountered at this facility was frequent clog-
ging of the sample probes and lines. This was due to the inter-
mittent nature of operation and the absence of high-pressure
backflush cleaning provisions.
An instrument malfunction was repaired at the beginning of the
test program. The auto zero mechanism required rebuilding to ob-
tain proper operation.
B. DATA REDUCTION
At the beginning of the continuous sulfur dioxide monitoring pro-
gram, an evaluation was conducted to determine a data recording
and reduction procedure that could be rapidly implemented and
would yield the highest confidence level for data handling re-
liability. Automatic data logging equipment with later automatic
reduction by computer was considered and rejected due to the un-
availability of equipment in inventory and the relatively long
delivery times required for procurement. Automatic tape record-
ing of data with subsequent reduction by computer was rejected
due to the relatively long time requirements for tape translation
to computer compatible input.
Due to the time restrictions on the initiation of continuous mon-
itoring, the only alternative remaining was manual reduction of
strip chart records and using punched cards for computer data
input. This procedure requires a very large manpower effort and
would not be practical for any program other than a short-term
10EMB Project No. 77-SPP-20, "First Interim Report - Continuous
S02 Monitoring Program at Eddystone No. 1, Philadelphia Electric
Company", Scott Environmental Technology, Inc., December, 1977.
24
-------
project. For routine monitoring, an automatic data logging and
calculation system is the only feasible approach for the amount
of data required to monitor flue gas desulfurization system per-
formance .
Monsanto Research Corporation was responsible for data reduction
and summarization in this project. The strip chart records were
collected from the recorders along with copies of the appropriate
boiler and scrubber control room logs at weekly intervals by the
contractors operating the monitoring systems at the test sites.
These were: Bruce Mansfield No. 1 - York Research Corporation;
La Cygne No. 1 - Midwest Research Institute; Cane Run No. 4; and
Eddystone No. 1 - Scott Environmental Technology, Inc.
The site contractors also provided narrative logs regarding in-
strument system performance and calibrations during the weekly
operating period.
A partial listing of the types of data that were received from
each test site is given below:
Cane Run No. 4
S02 strip chart (2 outlets - 2 inlets on same chart)
02 strip chart (2 outlets - 2 inlets on same chart)
Scrubber logs
Boiler logs
Calibration information
Bruce Mansfield No. 1
S02 inlet chart
02 inlet chart
S02 outlet chart
02 outlet chart
25
-------
Bruce Mansfield No. 1 (Continued)
Scrubber logs
Boiler logs
Additional comments on a daily basis
Eddystone No. 1
S02 strip charts (inlet-outlet on same chart)
Boiler logs
Scrubber logs
Manual test results and daily comments
La Cygne No. 1
S02 inlet chart
02 inlet chart
S02 outlet chart
Daily comments
Samples of the chart records from Cane Run No. 4 are presented in
Figure II-5.
The chart records were first logged in as received and then were
reviewed for possible data gaps. The charts were then reduced to
tabular format as shown in Figure II-6. Where only a single in-
stantaneous point was available in a 15-minute period, that re-
sult was used. When more than one point, or where a longer con-
tinuous record was available in a 15-minute period, the average
value was entered for a 15-minute value. The unit load vales
were obtained from the boiler logs.
After data transcription the boiler logs, scrubber logs, and
daily narratives were reviewed so that appropriate comments could
be included prior to data listing.
26
-------
to
Cht. Soeed: 6Vhr.
: 0-254 •!.!:
Inlet, Outlet, N and S n:| ,,
il.\ •'
. I . •!• I ••! Hi'l,j;'hll!llillli!ii
iffieiMi:
;;:j:;il.;.!:!;:!.iiiliiiil^iiJNlljilllLfcTnlriTiTn
Louisville SO; j
Cht. Soeed: 2"/hr. .,
Range: Inlet 0-4000 ppra 7
3:,|ii.. •:,:,,.
:i! |ij! i ! !
j'''! I'l'i: l!!i Plant Chart C.G4E)
I ! ' i '• I" f:- I ::|':"i:!:T'!!i!"'l:!!:M!u
'i >ff _ 30 i ' 30 I 40 i 50 . . \ . ' 60 ' ' "
!IP Hinn
iUoiilllll
?;•! rllii'iii'-'lilHiliil:!!:)!!^
10 I ;:-70 ;!; •! 60 | • 50 !
. :,i .• , " h.
Figure II-5: Sample Strip Chart Data
-------
LOUISVILLE OXYGEN DATA
(transcribed from strip chart)
co j
CO
LOUISVILLE SO2 DATA
(transcribed from strip chart)
Date
10-11
Time
1000
15
30
45
1100
15
30
45
1200
15
30
45
1300
15
30
45
1400
15
30
45
1500
15
30
45
Boiler Load
BLR
95
100
102
105
102
100
North Scrubber
In Out
1
70
73
74
73
71
72
73
73
78
76
76
78
76
78
77
72
75
78
77
78
78
79
76
75
3
21
25
25
25
23
24
21
25
23
19
32
19
19
23
23
16
20
19
23
20
26
27
27
18
South Scrubber
In Out
5
67
70
70
71
73
71
72
74
76
76
75
75
74
76
77
73
75
77
75
76
76
78
79
75
7
26
38
23
40.
23
27
34
31
43
28
42
23
44
27
42
26
40
27
32
32
46
31
45
28
-O2 Data-
Date
9-22
Time
0430
45
0500
15
30
45
0600
15
30
45
0700
15
30
45
0800
15
30
45
0900
15
30
45
1000
15
30
45
1100
Boiler Load
BLR
North Scrubber
In Out
1
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6.5
6
6
6
6
6
8
8.5
8
8.5
3
12
11
11.5
10.5
10.5
10
12.5
11
11
11
12.
11
12
13
12.5
12
12.5
12.5
12
12
11.5
11.5
10
11
11
11.5
11
South Scrubber
In Out
5 7
6 13
6 13
6 12
5.5 12.5
6 13
5.5 12.5
6 12.5
6 13
6 13
5.5 12.5
6 13
6 13
6 13
6 11
6 13.5
6 13
6 ..13
6 13
5 12.5
6 12
5.5 12
5.5 12
5 12
6 12.5
6 12
5.5 11.5
6 12
Figure II-6: Sample Data Transcription Form
-------
The transcribed data were then audited by spot-checking a 15-
minute data point for each day of transcribed data. After
audited, the data were keypunched and processed by computer to
obtain a data listing and calculated results for each 15-minute
data point. The computer listings were reviewed for keypunch or
other transcription errors. The listings were then sent to EPA
for editing. This edit procedure incorporated the review of
boiler, scrubber, and instrument operating logs so that periods
of scrubber outage, bypass, startup or shutdown could be identi-
fied and coded so that these data would not be included in the
calculation of averages and summaries. A sample of a 15-minute
data listing at Cane Run No. 4 is presented in Figure II-7. For
this system there are two separate inlets and outlets. The data
listing format is the same at the other facilities except that
no results are entered in the appropriate columns where less than
two inlets or two outlets are present.
After data editing was completed, average summaries were prepared.
Averages based on the 15-minute data were prepared for consecu-
tive 1-hour, 3-hour, 8-hour, and 24-hour intervals. In order to
calculate an average result for a single interval, it was speci-
fied that at least 75 percent of the 15-minute data points be
available for that interval. For example, an 8-hour average could
only be calculated when 24 of the possible 32 15-minute data
points were available. When less than 75 percent of the data were
available for an interval, an average was not calculated and
blanks were entered in the summary printouts.
After each 30 days of average interval data, a statistical sum-
mary was prepared. The parameters calculated were the mean
(based on the interval averages), and the statistical measures of
variation. Samples of 24-hour interval averages and the sta-
tistical summary of the averages are given in Figure II-8.
29.
-------
U)
o
*
* LOUISVILLE
*
*
* BLR
* LOAD
* OflTE TIME MW
* 30 107
* 45 107
* 081077 0600 134
* 15 130
* 30 134
* 45 134
* 081077 0700 145
* 15 145
* 30 145
* 45 145
* 081077 0800 160
* 15 160
* 30 160
* 45 160
* 081077 0900 160
* 15 160
* 30 160
* 45 160
* 081077 1000 160
* 15 160
* 30 160
* 45 160
* 081077 1100 ,160
* 15 160
* 30 160
* 45 160
* 081077 1200 160
* 15 160
* 30 160
* 45 160
* 081077 1300 IbO
* 15 160
* iO 160
* 45 160
* 081077 1400 160
* 15 160
* 30 160
* 45 160
081077 1500 160
15 160
30 160
45 160
081077 1600 150
15 150
30 150
•45 IbO
081077 1700 160
15 160
******************]
CD
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
o
0
0
0
0
0
0
0
0
0
0
n***l
INLET
S02
CONC 02
PPM CONC
WET PCT
2572 4.9
2572 4.9
2654 4.9
2654 4.9
2654 4.9
2695 4.9
2940 4.9
2613 4.9
2940 4.9
2041 4.9
2082 4.9
2409 4.9
2041 4.9
3022 4.9
2491 4.9
3062 4.9
2572 4.9
3226 4.9
2491 4.9
2450 4.9
2450 4.9
2450 4.9
2450 4.9
2450 4.9
2409 4.9
2409 4.9
2409 4.9
2409 4.9
2409 4.9
2409 4.9
2409 4.9
2409 4.9
2409 4.9
2409 4.9
2368 4.9
2409 4.9
2409 4.9
2450 4.9
2409 4.9
2409 4.9
2368 4.9
2368 4.9
2532 4.9
2858 4.9
2450 4.9
3062 4.9
2450 4.9
2817 4.9
************
(NORTH)
H20 LB/
PCT MMBTU
6.1 5.944
6.1 5.944
6.1 6.133
6.1 6.133
6.1 6.133
6.1 6.227
6.1 6.794
6.1 6.039
6.1 6.794
6.1 4.718
6.1 4.812
6.1 5.567
6.1 4.718
6.1 6.982
6.1 5.756
6.1 7.077
6.1 5.944
6.1 7.454
6.1 5.756
6.1 5.661
6.1 5.661
6.1 5.661
6.1 5.661
6.1 5.661
6.1 5.567
6.1 5.567
6.1 5.567
6.1 5.567
6.1 5.567
6.1 5.567
6.1 5.567
6.1 5.567
6.1 5.567
6.1 5.567
6.1 5.473
6.1 5.567
6.1 5.567
6.1 5.661
6.1 5.567
6.1 5.567
6.1 5.473
6.1 5.473
6.1 5.850
6.1 6.605
6.1 5.661
6.1 7.077
6.1 5.661
6.1 6.510
********** **
OUTLET
S02
CONC 02
PPM CONC
WET PCT
244 7.1
186 7.1
269 7.1
259 7.1
259 7.1
244 7.1
293 7.1
244 7.1
308 7.1
403 7.1
2g8 7.1
293 7.1
298 7.1
298 7.1
293 7.1
298 7.1
298 7.1
298 7.1
328 7.1
391 7.1
303 7.1
387 7.1
333 7.1
401 7.1
328 7.1
347 7.1
333 7.1
416 7.1
328 7.1
357 7.1
249 7.1
352 7.1
328 7.1
382 7.1
293 7.1
352 7.1
303 7.1
387 7.1
289 7.1
333 7.1
342 7.1
323 7.1
342 7.1
279 7.1
284 7.1
293 7.1
284 7.1
284 7.1
*************
(NORTH)
:
:
H20 LB/ :
PCT MMBTU :
6.9 0.674 :
6.9 0.513 :
6.9 0.742 !
6.9 0.715 :
'6.9 0.715 :
6.9 0.674 !
6.9 0.609 :
6.9 0.674 :
6.9 0.850 :
6.9 0.836 !
6.9 0.823 :
6.9 0.809 :
6.9 0.823 :
6.9 0.823 :
6.9 0.809 :
6.9 0.823 :
6.9 0.823 :
6.9 0.823 :
6.9 0.904 :
6.9 1.079 :
6.9 0.836 :
6.9 1.066 :
6.9 0.917 :
6.9 1.106 !
6.9 0.904 :
6.9 0.958 }
6.9 0.917 !
6.9 1.147 :
6.9 0.904 :
6.9 0.9P5 :
6.9 0.688 :
6.9 0.971 :
6.9 0.904 :
6.9 1.052 I
6.9 0.809 :
6.9 0.971 :
6.9 0.836 ;
6.9 1.066 :
6.9 0.796 :
6.9 0.917 :
6.9 0.944 :
6.9 0.690 :
6.9 0.944 :
6.9 0.769 :
6.9 0.782 :
6.9 0.809 :
6.9 0.782 :
6.9 0.782 :
****»*******]
t*****
EFF
88.7
91.4
87.9
88.3
88.3
89.2
88.1
88.8
87.5
82.3
82.9
85.5
82.6
88.2
85.9
88.4
86.2
89.0
84.3
80.9
85.2
81.2
83.8
80.5
83.8
82.8
83.5
79.4
83.8
82.3
87.6
82.6
83.8
81.1
85.2
82.6
85.0
81.2
85.7
83,5
82.7
83.7
83.9
88.4
86.2
68.6
86.2
88.0
t*****
*
» INLET (SOUTH) OUTLET (SOUTH)
*
* S02 S02 :
* CONC 02 CONC 02 :
* PPM CONC H20 LB/ PPM CONC H20 LB/ :
* WET PCT PCT MMBTU WET PCT PCT MMBTU I EFF
* 2572 4.9 6.1 5.944 284 7.1 6.9 0.782 : 86.8
* 2532 4.9 6.1 5.850 293 7.1 6.9 0.809 ! 86.2
* 2777 4.9 6.1 6.416 333 7.1 6.9 0.917 I 85.7
* 2736 4.9 6.1 6.322 440 7.1 6.9 1.214 I 80.8
* 2654 4.9 6.1 6.133 387 7.1 6.9 1.066 : 82.6
* 2736 4.9 6.1 6.322 440 7.1 6.9 1.214 J 80.8
* 2654 4.9 6.1 6.133 367 7.1 6.9 1.012 : 83.5
* 2777 4.9 6.1 6.416 342 7.1 6.9 0.944 t 65.3
* 2736 4.9 6.1 6.322 318 7.1 6.9 0.877 t 86.1
* 2940 4.9 6.1 6.794 313 7.1 6.9 0.863 8 87.3
* 3267 4.9 6.1 7.548 298 7.1 6.9 0.823 I 89.1
* 3553 4.9 6.1 8.209 293 7.1 6.9 0.809 J 9Q.1
* 2532 4.9 6.1 5.850 293 7.1 6.9 0.809 : 86.2
* 3757 4.9 6.1 8.681 298 7.1 6.9 0.823 : 90.5
* 3675 4.9 6.1 8.492 293 7.1 6.9 0.809 : 90.5
* 3226 4.9 6.1 7.454 293 7.1 6.9 0.809 t 89;1
* 3634 4.9 6.1 8.398 293 7.1 6.9 0.809 : 90.4
* 3471 4.9 6.1 8.020 293 7.1 6.9 0.809 : 89.9
* 2491 4.9 6.1 5.756 460 7.1 6.9 1.268 t 78.0
* 2450 4.9 6.1 5.661 347 7.1 6.9 0.958 : 83.1
* 2450 4.9 6.1 5.661 416 7.1 6.9 1.147 ! 79.7
* 2450 4.9 6.1 5.661 347 7.1 6.9 0.958 : 63.1
* 2450 4.9 6.1 5.661 391 7.1 6.9 1.079 : 60.9
* 2450 4.9 6.1 5.661 352 7.1 6.9 0.971 ! 82.8
* 2450 4.9 6.1 5.661 391 7.1 6.9 1.079 ! 80.9
* 2450 4.9 6.1 5.661 357 7.1 6.9 0.985 : 82.6
* 2450 4.9 6.1 5.661 401 7.1 6.9 1.106 : 80.5
* 2409 4.9 6.1 5.567 391 7.1 6.9 1.079 : 80.6
* 2450 4.9 6.1 5.661 406 7.1 6.9 1.120 ! 80.2
* 2409 4.9 6.1 5.567 391 7.1 6.9 1.079 : 60.6
* 2409 4.9 6.1 5.567 436 7.1 6.9 1.201 : 78.4
* 2409 4.9 6.1 5.567 352 7.1 6.9 0.971 : 82.6
* 2409 4.9 6.1 5.567 328 7.1 6.9 0.904 : 83.8
* 2409 4.9 6.1 5.567 382 7.1 6.9 1.052 J 81.1
» 2409 4.9 6.1 5.567 347 7.1 6.9 0.958 : 82.8
* 2409 4.9 6.1 5.567 367 7.1 6.9 1.012 t 61.8
* 2450 4.9 6.1 5.661 342 7.1 6.9 0.944 : 83.3
* 2409 4.9 6.1 5.567 387 7.1 6.9 1.066 J 80.9
* 2450 4.9 6.1 5.661 362 7.1 6.9 0.998 ! 62.4
* 2409 4.9 6.1 5.567 362 7.1 6.9 0.998 ! 62.1
* 2368 4.9 6.1 5.473 333 7.1 6.9 0.917 : 83.2
* 2368 4.9 6.1 5.473 338 7.1 6.9 0.931 : 63.0
* 2368 4.9 6.1 5.473 308 7.1 6.9 0.850 ! P4.5
* 2532 4.9 6.1 5.850 279 7.1 6.9 0.769 : 86.9
* 2858 4.9 6.1 6.605 293 7.1 6.9 0.609 ! 87.7
* 2940 4.9 6.1 6.794 284 7.1 6.9 0.7P2 ! 88.5
* 2736 4.9 6.1 6.322 284 7.1 6.9 0.7P2 : 87.6
* 2817 4.9 6.1 6.510 : 284 7.1 6.9 0.782 : 68.0
*********************************************************
Figure II-7: Sample Data Listing
-------
*****************»******************************»*******************»*************»**»*»*»**»*********»*»***»»»»»*»„„,**»»»»»»»»
t
* 24 HOUR AVGS
* LOUISVILLE
*
» DATE TINE MRS
* 072177 1200 48
* 072277 2100 96
* 072377 2*100 71
* 072477 2<»00 75
* 072577 2400 96
* 072677 2400 96
* 072777 2400 63
* OB0977 2400 85
* 081077 2400 96
* 081177 2400 94
* 081277 2400 90
* 081677 2400 89
* 081777 2400 96
* 082277 0400 79
* 082377 2<»00 96
* 062477 2400 91
* 082577 2400 66
* 082677 2400 96
* OB2777 2400 90
* 082877 2400 96
* 082977 2400 96
* 083077 2400 96
* 083177 2400 96
» 090177 2400 96
* 090277 2400 94
INI NORTH)
S02
PPM LB/
DRY MMBTU
2389 5.164
2875 6.238
290n 6.300
2843 6.167
2787 6.045
2844 6.169
2018 4.403
2649 5.746
3175 6.887
29o3 6.298
2492 5,405
2667 5.766
2906 6.304
2741 5.951
2709 5.882
2503 5.434
2695 5.851
2633 5.717
2858 6.206
3023 6.564
2508 5.447
2351 5.108
OUT(NORTH)
S02
PPM LB/
DRT MMBTU
133 0.342
27 0.07U
34 0.088
34 0.086
277 0.710
290 0.744
373 0.956
411 1.024
510 1.307
227 0.581
430 0.960
391 1.002
353 0.906
287 0.737
368 0.945
358 0.92U
371 0.953
357 0.918
424 1.089
516 1.326
362 0.93U
290 0.744
* IN(SOUTH) : OUT(SOUTh)
* ;
* SP2 : S02
* PPM LB/ : PPM LB/
EFF » DRY MMBTU : DRY MMBTU
93.4 * 2830 6.140 : 95 0.244
98.9 * 2911 6.314 : 20 0.050
98.6 * 2920 6.335 ! 23 0.060
98.6 * 2934 6.365 : 30 0.078
68.2 * 2612 6.100 : 352 0.854
87.8 » 2876 6.239 : 332 0.851
63.9 * 2969 6.341 : 394 0.936
61.8 * 2677 5.808 : 383 0.981
81.2 * 3196 6.934 : 506 1.267
9l.o * 2878 6.243 : 244 0-626
79.6 * 2224 4.825 : 427 1.058
62.7 « 2635 5.715 : 395 0.997
65.7 * 2865 6.260 : 372 0.954
87.7 » 2734 5.935 : 327 0.839
83.9 * 2758 5.987 : 411 1.056
63.5 * 2599 5,642 : 389 0.967
63.8 * 2726 5.918 t 422 1.071
64.0 * 2665 5.787 : 419 1.076
62.6 * 2861 6.210 : 483 1.240
79.9 * 3052 6.625 : 572 1.469
63.1 * 2586 5.619 : 421 1.081
65.4 * 2535 5.504 : 391 1.003
i
EFF
96.0
99.2
99.1
98.8
65.5
86.2
83.7
82.9
61.4
90.2
75.2
82.4
64.6
66.0
82.4
83.0
80.9
81.5
80.2
78.0
80.8
81.9
**********
t ;
INLET OUTLET :
TOTAL TOTAL :
LB/ LB/ : TOTAL
MMBTU MMBTU i EFFY
*
5.662 0.293 t 94.8
6.276 0.060 : 99.0
6.318 0.074 : 98.8
6.266 0.083 : 98.7
6.073 0.762 : 87.1
6.204 0.797 : 67.1
5.372 0.946 : 82.4
5.777 1.002 : 82.6
t
*
6.911 1.287 I 61.4
6.271 0.603 : 90.4
5.115 1.009 : 80.3
5.750 0.999 : 82.6
6.282 0.930 t 85.2
5.943 0.788 : 86.7
5.935 1.001 : 83.1
5.536 0.944 : 83.0
5.885 1.012 t 82.8
5.752 0.997 : 82.7
6.208 1.165 t 81.2
6.595 1.397 : 78.8
5.533 1.006 : 61.8
5.306 0.874 : 83.5
***************************
********************
24 HOUR AVGS
LOUISVILLE
30 DAY STATS
MEAN
STD DEVIATION
AVG DEVIATION
MAXIMUM
MINIMUM
RANGE
PCT ST DEV
********************
*****************
IN(NORTH)
S02
PPM LB/
DRY MMBTU
2703. 5.868
255.6 0.5508
193.5 0.4179
3175. 6.887
2018. 4.403
1156. 2.464
9.45 9.387
t****************
*****************
OUT(NORTH)
S02
PPM LB/
DRY MMBTU
310. 0.768
141.3 0.3572
106.4 0.2716
516. 1.326
27. 0.070
489. 1.256
45.55 45.314
*****************
********
EFF
66.6
5.94
4.67
98.9
79.6
19.3
6.9
********
******************
* IN(SOUTH) :
* ;
» S02 :
* PPM LB/ :
* DRY MMBTU :
«__v v_<»«w •
_«v v_*«w f
* 2785. 6.038 :
« 203.7 0.4377 :
* 155.1 0.3313 :
* 3196. 6.934 :
* 2224. 4.825 :
* 972. 2.109 :
* 7.32 7.249 :
******************
****************
OUT (SOUTH)
S02
PPM LB/
DRY MMBTU
337. 0.853
156.7 0.3976
116.8 0.2928
572. 1.469
20. 0.050
552. 1.419
46.53 46.622
****************
***********
*
*
*
*
EFF
85.5 *
6.89 *
5.20 *
99.2 *
75.2 *
24.0 *
8.1 *
***********
***************
INLET OUTLET
TOTAL TOTAL
LB/ LB/
MMBTU MMBTU
5.953 0.820
0.436 0.375
0.352 0.280
6.911 1.397
5.115 0.060
1.795 1.337
7.354 45.740
****************
**********
TOTAL
EFFY
86.1
6.3
4.9
99.0
78.8
20.2 *
7.3 *
**********
Figure II-8: Sample Average Listing and Statistical Summary
-------
A second statistical computation was performed for the sulfur di-
oxide removal efficiencies of the FGD systems. This procedure
calculated a frequency distribution for the percent occurrence of
sulfur dioxide removal efficiencies for each averaging interval.
The calculation procedures used to convert the analyzer outputs
for sulfur dioxide and oxygen concentrations to mass emission
factors are given in 40 CFR 60 Subpart D. This procedure is
known as the F-factor approach and is given below:
_
E =
CFK
T=M
20.9
°2
Where E = Emission factor - Ib/million Btu
C = S02 concentration - ppmv, wet basis
F = Stoichiometric conversion factor, 9820 dscf/million Btu
for subbituminous coal
K = Conversion factor, 1.659 x 10~7 Ib/dscf per ppmv
02 = Oxygen concentration, percent by volume as measured
M = Moisture fraction as measured (for dried samples, M=0)
The sulfur dioxide and oxygen concentration results were obtained
by multiplying the strip chart readings as a percent of scale by
the appropriate calibration factor.
The emission factor was calculated for each FGD system inlet and
outlet test point. The sulfur dioxide removal efficiency for a
module or set of modules is calculated by:
Efficiency = Ein " E out x 100%
tin
32
-------
When more than inlet and/or outlet test point was monitored, the
total system emission factors and sulfur dioxide removal effi-
ciencies were calculated by a weighted average procedure. The
equation for this calculation for total system efficiency is
given by:
EFF total = EFFA(FA) + EFFB(FB)
Where,
EFF f 1 = total system efficiency
EFF = efficiency of module or module set A
£\
F = fraction of gas flow through module or model
/\
set A
EFFB = efficiency of module or module set B
Fn = fraction of gas flow through module or module
D
set B
For all data in this study the scrubber systems have operated
such that the flows through two module sets have been essentially
equal, or there has only been one module set operating. Thus, a
simple average has been used for total system performance.
The major problems that were encountered in the data reduction
procedures were related to strip chart records. Cases were en-
countered where time, periods of chart stops, starts, and ad-
vances, and unmarked instrument range changes, were not properly
noted, as well as periods when pens failed to ink. These prob-
lems were corrected by requiring that the site contractors devote
additional attention to assuring that accurate and adequately'in-
formative chart records were obtained. During data editing, the
33
-------
boiler, scrubber, and instrument log sheets were adequate to de-
scribe periods of missing or unuseable data.
C. IMPROVEMENTS AND RESEARCH NEEDED
The majority of the problems encountered in this program were due
to the necessity of using previously installed instrumentation
and adapting other instrumentation into a system that was not de-
signed as a unit and as such, could not perform at optimum levels.
For example, most of the sulfur dioxide instruments were ordered
and installed five to seven years ago when less was known about
the most useful installation locations. Additionally, in some
cases the instrument systems were never properly installed and
operated.
None of the sulfur dioxide systems were designed to allow straight-
forward measurement of oxygen. In order to meet the program time
objectives, available oxygen measurement equipment had to be
adapted and this adaptation was time consuming and required more
maintenance effort than is desirable.
It appears that the modification to the probe backflush system
which incorporated a combination of water and air will solve the
probe plugging problems in those cases where dry air is not an
adequate cleaning procedure.
/
While the present procedure for moisture content of the sample
streams is adequate for intensive monitoring efforts where man-
power is continuously available to record results, this procedure
is not a desirable long-term approach. Research is recommended
to define the moisture measurement requirements of various instru-
ment systems, and the best way to achieve these requirements.
This result could either be by continuous instrumentation for
moisture measurement, or by sample conditioning to a known stan-
dard condition.
34
-------
The improvement that will provide the greatest benefit is an
automatic data logging and calculation system. This has been the
most time consuming part of this program and manual data analysis
is not recommended for routine purposes. Data processing equip-
ment suitable for this purpose is commercially available.
In summary, the monitoring system operating problems that have
been encountered in this program have been the result of inade-
quate initial system design, or the compromises made in the ^in-
terest of rapid test initiation. Based on the experience de-
veloped in this program, a monitoring system that would be
convenient to use, yield high operating reliability, and provide
data automatically converted to the units of the standard could
be designed.
35
-------
III. DATA ANALYSIS
A. DATA COLLECTION INTERVAL SELECTION
At the beginning of the monitoring program it was necessary to
establish a minimum data collection frequency. Since essentially
all of the installed sulfur dioxide systems were multiple sam-
pling point analyzers, no truly continuous measurements would be
available. Review of the systems indicated that a 15-minute
cycle for a 4-point analyzer was the practical minimum. Since
all other analyzers would yield at least one result during a
15-minute cycle, this was selected as a standard collection fre-
quency for all sites to maintain calculation consistency. For
those systems where either a multiple number of instantaneous
values, or a period of continuous values are available in a 15-
minute period, the average of these values is used as the data
point. This procedure is valid so long as the measured parame-
ters do not vary significantly during a 15-minute period. Sam-
ples of the output at an inlet and outlet test point at Cane Run
No. 4 where the analyzer was held in a continuous mode are given
in Figures III-l and III-2. The inlet sulfur dioxide concentra-
tions are essentially constant over a 15-minute period. The
outlet sulfur dioxide concentrations vary about ± 1 to 2 scale
divisions about the mean during a 15-minute period. For readings
greater than 50 percent of full scale, the potential variation
for an instantaneous reading from the 15-minute mean would be at
most ± 4 percent of the concentration result. This estimate is
based on steady operation at constant load.
37
-------
M !.+
TT
-H
U-
Figure III-l: Inlet S02 Variation, Cane Run No. 4
38
-------
Figure III-2: Outlet S02 Variation, Cane Run No. 4
39
-------
The selection of a 15-minute minimum data interval is also con-
sistent with the requirements given in 40 CFR 60 Subpart A Sec-
tion 60.13.e.z.
B. DATA CALCULATION ASSUMPTIONS
In order to calculate the mass emission factors and recovery ef-
ficiencies from sulfur dioxide concentrations, it was necessary
in some cases to use average results. At Cane Run No. 4, the
oxygen monitor experienced frequent outages early in the test
program. For those periods where oxygen data were not available,
an average result for oxygen concentration was used. These
averages were based on test results obtained when the monitor was
in operation at similar boiler megawatt loadings. The periods
where averages were used can be identified in the data by observ-
ing those periods of time where the oxygen listings are constant.
The moisture levels chosen for use in the calculations are based
on a series of actual moisture and temperature measurements. The
measurements yielded moisture contents in both the inlet and out-
let sample streams of between 2 and 7 percent by volume. An
average value of 4 percent by volume was used at the inlet and
outlet for all data since the measurements available did not
justify daily moisture level modifications. The maximum error
that can be encountered in the emission factor calculations due
to this assumed moisture is about 3 percent. For the sulfur di-
oxide removal efficiency calculation, an error of 3 percent in
both the inlet and outlet emission factors will result in a max-
imum error of 0.5 percent of the percent removal efficiency.
This is due to the cancelling effect of the division calculation.
At Bruce Mansfield No. 1, no oxygen estimations were necessary.
The reported data are based on measurements of all parameters.
Moisture measurements were taken at weekly intervals and the re-
sults for that week were used in the calculations.
40
-------
At Eddystone No. 1, oxygen measurements were by manual procedures
during the scrubber operating periods. It was found tha't the
variability was small and, therefore, one average result was used
for inlet and outlet. Moisture levels were based on manual mea-
surements .
At La Cygne No. 1, oxygen data on a limited basis were available
for the inlet location. No data were available for the outlet.
For estimation purposes, the outlet oxygen concentration was as-
sumed to be equal to the inlet. This calculation yields only an
estimate of the sulfur dioxide efficiency since there is probably
some dilution between the inlet and outlet measurement points.
Moisture levels were chosen based on average results obtained by
manual measurements. Because no complete outlet data were
gathered at La Cygne No. 1, statistical analyses as described in
the following sections were not performed.
In addition to the data from EPA monitoring sites, data acquired
from the Wellman-Lord demonstration unit at Northern Indiana
Power Service Company's (NIPSCO) D. H. Mitchell No. 11 generating
unit, Gary, Indiana, were also included in the statistical analy-
ses which follow. A description of the Wellman-Lord process and
its installation at the Mitchell plant may be found in the pro-
ceedings: Symposium on Flue Gas Desulfurization, Hollywood,
Florida, November, 1977, EPA-60017-78-058, Page 650.
For brevity the test sites are identified in the following sec-
tions as follows:
Test Site Identifier
Cane Run No. 4, LGE Louisville
Bruce Mansfield No. 1, PPC Pittsburgh
Eddystone No. 1, PE Philadelphia
Mitchell No. 11, NIPSCO Chicago
41
-------
C. UNUSABLE DATA
Normally the S02 monitoring system is not operated when the
boiler is down, this time being used for repair, maintenance, and
calibration of instruments. During periods when the scrubber is
by-passed, the SC>2 monitors at all EPA test sites remained in
service. These data, along with data obtained while the scrubber
was starting up, not having achieved normal operation, were not
included in the data analyzed for this report. However, data was
not deleted due to process upsets unrelated to start-ups or shut-
downs. All other unused data for the purposes of this report
were due to the monitoring system, resulting from outright fail-
ures, repair, or routine calibration and maintenance. At the
Louisville test site, some data were pre-empted by performance
testing of the monitoring equipment.
Table III-l illustrates the population of unusable data at two
test sites by category. As shown, monitoring system problems and
repairs accounted for about one-half of unusable data at the
Louisville and Pittsburgh test sites. Boiler failures accounted
for roughly 25 percent of non-data periods, while monitoring sys-
tem maintenance accounted for less than 5 percent. Monitoring
failures, calibration, and maintenance accounted for 31 and 20
percent of operating time for the two modules at the Louisville
site, and nearly 44 percent at the Pittsburgh site. The appar-
ently poor availability of the monitoring systems is not indica-
tive of the state-of-the-art of S02 monitoring systems. The
shortcomings of these particular systems were discussed in detail
earlier in this report.
D. USABLE DATA AND STATISTICAL ANALYSES
The remaining valid data was statistically analyzed according to
the ground rules described earlier in this chapter, for the
42
-------
Table III-l. CATEGORIZATION OF DATA
Sites
Louisville
Module "A" Module "B"
Pittsburgh
Usable Data Points
7582
9414
3890
Unusable Data:
Boiler down 1444
Scrubber down 414
Start-up 54
Monitor failure/repair 3851
Calibration/maintenance 165
Initial performance test 876
Unaccounted losses 466
14852
1444
414
54
2019
165
876
466
14852
1245
2646
372
8153
43
-------
following variables as a function of 24-hour averaging time, as-
suming all data were normally distributed:
mean
standard deviation
average deviation
percent standard deviation
SC>2 removal efficiencies
Subsequent analyses showed, in fact, that all data were log-nor-
mally distributed. The inlet data findings differed from previ-
ous coal sampling studies where coal sulfur data were normally
distributed.
Figures III-3 and III-4 show inlet S02 data from the Louisville
site plotted on normal and log-normal probability paper, respec-
tively, for the period July 23 - December 15, 1977. Data after
December 15 resulted from a different coal feed and were deleted
to make the statistics valid. Apparently, the best fit was the
plot of inlet S02 values against log probability of occurrence
(Figure III-4). From this plot the mean value of S02 inlet emis-
sion rate was 5.75 Ib/MBTU and the geometric mean 1.08 lb/MBTU.*
Data from the remaining sites were also plotted in Figure III-4.
Table III-2 summarizes the results. Mean inlet SC>2 values ranged
from 5.12 to 6.30 lb/MBTU with geometric standard deviation from
1.07 to 1.12 lb/MBTU.
Figures III-5 and III-6 show the outlet S02 data from the Louis-
ville site as functions of normal and log-normal probability.
Again, the best fit appears to be log-normal probability. Mean
*The mean is y-intercept at 50 percent value; geometric standard
deviation is the ratio of the 93.3 percent and 6.7 percent less
than values, raised to the 1/3 power.
44
-------
UT
99.99 99.9 99.8
7.0
vO
o
to
I 6.0
o
o
S1
5.0
4.0
PERCENT LESS THAN
99 98 95 90 80 70 60 50 40 30 20 10 5 2 1 0:5 0.2 0.1 0.01
T—T
0.01 Q05 0.10.2 Q5 1 2 5 10 20 30 40 50 60 70 80 90 95 98 99 99.8 99.9 99.99
PERCENT GREATER THAN
Figure III-3. Probability Vs. Linear Inlet SO2
Concentrations from Louisville Site
-------
. JJ9.99 99.9 99.8
1UU| 1—i—
PERCENT LESS THAN
99 98 95 90 80 70 60 50 40 30 20 10 5 21 0.5 0.2 0.1 0.01
CO
oo
o
o
o
o
CM
o
on
10
1.0
LEGEND:
* LOUISVILLE
• PITTSBURGH I
• PITTSBURGH II
o PHILADELPHIA
o CHICAGO
0.010.05 0.10.2 Q5 1 2 5 10 20 30 40 50 60 70 80 90 95 98 99 99.8 99.9 99.99
PERCENT GREATER THAN
Figure III-4. Probability Vs. Log Inlet SO2 Concentrations
-------
PERCENT LESS THAN X
99.99 99.9 99.8 99 98 95 90 80 70 60 50 40 30 20 10 5 21 0.5 0.2 0.1 0.01
2.0
oa
vO
o
1.2
CSJ
o
CO
0.4
0.01 Q05 0.10.2 0.5 1 2
5 10 20 30 40 50 60 70 80 90 95 98 99
PERCENT GREATER THAN X
99.8 99.9 99.99
Figure III-5. Probability Vs. Linear Outlet S02
Emissions from Louisville Site
-------
CO
10
PERCENT LESS THAN
,99.99 99.9 99.8 99 98 95 90 80 70 60 50 40 30 20. 10 5 21 0.5 0.2 0.1 0.01
CO
co
CO
CO
CM
O
CO
1.0
0.1
LEGEND:
* LOUISVILLE
• PITTSBURGH I
• PITTSBURGH II
a PHILADELPHIA
o CHICAGO
I I
0.01 Q05 Q1Q2 Q5 1 2
5 10 20 30 40 50 60 70 80 90 95 98 99 99.8 99.9 99.99
PERCENT GREATER THAN
Figure III-6. Probability Vs. Log Outlet S02 Emissions
-------
Table III-2. INLET S02 EMISSIONS
(24-hr. Averaging Period)
SITE
Louisville
Pittsburgh No. 1
Pittsburgh No. 2
Philadelphia
Chicago
NO. OF POINTS
82
20
11
8
25
MEAN INLET S02
lb/106 BTU
5.75
6.30
5.40
5.10
6.30
GEOMETRIC STANDARD DEVIATION
lb/106 BTU
1.08
1.07
1.11
1.05
1.12
-------
and geometric standard deviations were graphically determined to
be 0.89 and 1.35 Ib/MBTU respectively.
The remaining S02 emission data for four other test sites were
plotted (Figure III-6) on log-normal probability paper. All five
sites exhibited a reasonably good data fit. Results of graphical
analyses are tabulated in Table III-3.
Pittsburgh I and II data represent the two periods of operation -
Period I when pH control was a severe problem and Period II when
pH control was improved.
Table III-3 shows the geometric standard deviation of outlet
emissions to vary from 1.2 to 1.5.
Since both the inlet and outlet S02 levels were judged to be log-
normally distributed, it was assumed that the performance of each
scrubber system, expressed as a percentage S02 reduction
,nn (1-outlet)
_l_ \J \J —« —.. , 1
inlet J
would also be log-normally distributed. Figure III-7 shows a
plot of S02 reduction versus log-normal probability. In the
range (70 to 98 percent removal) shown, the straight line rela-
tionships of data and subsequent graphical analysis are prone to
greater error. Therefore, a better plot is one which transforms
the data to the more favorable area on the logarithmic chart,
i.e., the 1 to 4 range.
Such a transform is 100-(S02 removal, percent) which in reality
is outlet/inlet x 100 or percent SO2 remaining. When plotted as
shown, in Figure III-8, the straight line fits allow much better
graphical interpretation of data. Table III-4 shows the results
50
-------
Table III-3. OUTLET S02 STATISTICS
(24-hr. Averaging Period)
SITE
Louisville
Pittsburgh No. 1
Pittsburgh No. 2
Philadelphia
Chicago
NO. OF POINTS
89
20
11
8
25
MEAN OUTLET S02
lb/106 BTU
0.88
1.21
0.77
0.155
0.67
GEOMETRIC STANDARD DEVIATION
lb/106 BTU
1.38
1.37
1.21
1.48
1.19
-------
1000
,99.99 99.9 99.8 99 98 95
to
o
UJ
Q.
10
PERCENT TIME BELOW
80 70 60 50 40 30 20 10 5 2 1 0.5 0.2 0.1 0.01
LEGEND:
* LOUISVILLE
• PITTSBURGH I
• PITTSBURGH II
a PHILADELPHIA
o CHICAGO
0.010.05 0.2 0.5 1 2 5 10 20 30 40 50 60 70 80 90 95 98 99 99.8 99.9 99.99
PERCENT TIME ABOVE
Figure III-7. Probability Vs. Percent S02 Removal
-------
Ul
CO
100
99.99 99.9 99.8
PERCENT LESS THAN
99 98 95 90 80 70 60 50 40 30 20 10 5 21 0.5 0.20.1 0.01
X
I
I
10
i
o
LU
Q_
10
i 1 T
1 T r
1
0.010.050.10.2 Q5 1 2
LEGEND:
* LOUISVILLE
• PITTSBURGH I
• PITTSBURGH II
a PHILADELPHIA
o CHICAGO
'I' I' 1 1 1 1 1 1 1 1 L.
10 20 30 40 50 60 70 80
PERCENT GREATER THAN
I I
90 95 98 99 99.8 99.9 99.99
Figure III-8. Probability Vs. 1QO Minus Percent S02 Removal
-------
Table III-4. S02 REMOVAL STATISTICS
(24-hr. Averaging Period)
SITE
MEAN SO2 REMOVAL
NO. OF POINTS EFFICIENCY, PERCENT
GEOMETRIC STANDARD PERCENT
DEVIATION
01
Louisville
Pittsburgh No. 1
Pittsburgh No. 2
Philadelphia
Chicago
89
20
11
8
25
83.8
81.4
85.3
96.8
90.0
1.060
1.057
1.029
1.014
1.015
-------
of these graphical interpretations. The means (converted back to
percent removal) ranged from 81.4 to 96.8, while geometric stan-
dard deviations ranged from 1.014 to 1.060.
Some questions exist regarding the "dampening" of inlet sulfur
variations by flue gas scrubbing. Though many parameters enter
into the design of FGD systems, the reactivity of the system
chemistry and the degree of process control play an important role
in dampening of S02 emissions as well as the amount of excess re-
actant present.
In all the systems monitored, process controls for maintaining the
desired reactant concentrations (measured by pH) were usually
operated in the manual rather than the automatic mode.
Using the values for the mean and geometric standard deviation
from Tables III-2 and III-3, the variations in S02 concentration
were compared for several confidence levels as shown in Table
III-5. The dampening effect is measured by comparing inlet var-
iation to outlet variation and is calculated as shown in the
table.
Since each system is operated under a different set of conditions,
e.g., coal fired, boiler loading, type of scrubber module, excess
reactant, few conclusions can be drawn. It is noteworthy, how-
ever, that a negative dampening is observed for the first batch
of Pittsburgh data where pH control was very poor and that a
70 percent dampening resulted when pH control improved. The more
reactive sodium, magnesium oxide, and magnesium oxide enriched
lime systems showed considerably more dampening than the lime
system alone, as would be expected under comparable operating
conditions.
Table III-5 can be used to illustrate the impact of a "never to
be exceeded" regulation where scrubbers are the control of choice.
55
-------
Table III-5.
VARIATION IN S02 CONCENTRATIONS IN LB/106 BTU
(24-hour averages)
68.7%
Confidence Level
95%
Percent Dampening = inlet.-outlet x 100
inlet
99.7%
Louisville:
Inlet
Outlet
Percent Dampening1
Pittsburgh:
#1 Inlet
Outlet
Percent Dampening
#2 Inlet
Outlet
Percent Dampening
Philadelphia:
Inlet
Outlet
Percent Dampening
Chicago:
Inlet
Outlet
Percent Dampening
±0.46
±0.33
28.2
±0.44
±0.45
-0.0
±0.59
±0.16
72.9
±0.26
±0.07
70.8
±0.76
±0.13
82.9
±0.96
±0.80
15.7
±0.91
±1.07
-17.5
±1.25
±0.36
71.2
±0.52
±0.18
64.7
±1.60
±0.28
82.5
±1.49
±1.43
4.0
±1.42
±1.92
-35.2
±1.99
±0.59
70.4
±0.80
±0.35
56.8
±2.55
±0.46
82.0
56
-------
At the 99.7 percent confidence level the variation in S02 emis-
sions is rather large in comparison to the present standard for
new coal-fired boilers above 250 MBTU/hr - 1.2 Ib S02 per MBTU
input. The best system examined has a ± 0.35 Ib/MBTU variation
on a 24-hour basis, implying that a lifetime mean of 0.85 Ib/MBTU
SC>2 emissions would be required in order to comply 99.85 percent
of the time. The lime system with a ± 1.43 Ib S02/MBTU expectan-
cy could not always comply, and would have to average 0.40 Ib
S02/MBTU over its lifetime to be in compliance 97.5 out of every
100 24-hour periods or roughly 356 days per year.
Finally, the effects of longer averaging periods upon performance
were examined. Since insufficient data exists for a rigorous
examination, only the Louisville system was considered at 7 and
14-day periods. Figure III-9 and Table III-6 summarize the find-
ings. The differences in mean values are due to more data points
used in the 24-hour case (89) versus 84 24-hour points in the 7
and 14-day cases.
As expected, the geometric standard deviation decreases with in-
creased averaging time, increasing the confidence of compliance
with a given emission standard or percentage removal requirement.
57
-------
Table III-6. CUMULATIVE LOUISVILLE STATISTICS VS. AVERAGING TIME
Averaging Period
Percent Recovery 24-hour 7 days 14 days
No. of Periods
Geometric Mean
Geometric STD Deviation
89
83.8
1.06
12
83.1
1.05
6
83.1
1.03
S02 Emissions, lb/106 BTU:
No. of Periods
Geometric Mean
Geometric STD Deviation
89
0.88
1.38
12
0.95
1.2(5
6
0.95
1.17
58
-------
IV. RECOMMENDATIONS AND CONCLUSIONS
A. Although the availability of the monitoring systems during
the first four months of EPA's study ranged from 56 to 80
percent, most problems were a result of the system design.
B. The oxygen monitoring instrumentation is the least reliable
piece of equipment in the S02 monitoring system used for this
study. Oxygen data by Orsat analysis or a suitable backup
system may be necessary on occasions.
C. Research was necessary to define the expected sample condi-
tions with respect to moisture content, and to develop the
most acceptable procedure for moisture measurement or esti-
mation.
D. Manual data acquisition and reduction has required a very
high manpower investment. It is recommended that automatic
data systems to acquire the individual instrument outputs and
convert to units of the standard, averaged over the specified
intervals, be used for routine monitoring.
E. SO2 scrubbers do have a dampening effect upon inlet S02 vari-
ations over longer 24 hour or greater averaging periods. The
instantaneous dampening effect is masked by short-term varia-
tions in scrubber performance due to system response lags.
Dampening of 70 to 80 percent, as calculated by outlet stan-
dard deviation divided by inlet standard deviation, has been
shown in these tests on a 24-hour basis.
61
-------
F. Longer averaging periods have the effect of lowering the var-
iance of scrubber performance as measured by S02 removal ef-
ficiency and outlet S02 emissions.
62
------- |