Acurex Project 7418
A REVIEW OF FGD. SYSTEM OPERATION AND PROCESS CONTROLS
.• -
t
FINAL REPORT
Acurex Corporation/Energy & Environmental Division
South/East Regional Operations
Route 1, Box 423
Morrisville, North Carolina 27560
December 1979
Prepared for
Lead Engineer Larry Jones
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Emission Standards and Engineering Division
Research Triangle Park, North Carolina 27711
68-02-3064
-------
Acurex Project 7488
A REVIEW OF FGD SYSTEM OPERATION AND PROCESS CONTROLS
FINAL REPORT
John Chang and Anthony Marimpietri
Acurex Corporation/Energy & Environmental Division
South/East Regional Operations
Route 1, Box 423
Morrisville, North Carolina 27560
December 1979
Prepared for
Lead Engineer Larry Jones
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Emission Standards and Engineering Division
Research Triangle Park, North Carolina 27711
Contract 68-02-3064
-------
TABLE OF CONTENTS
Page
1.0 Introduction 1
2.0 Summary 1
3.0 Review of EPA Test Data 2
4.0 Process Controls for FGD Systems 11
5.0 Further Reduction of Efficiency Variability 15
6.0 Conclusions 1?
7.0 References 18
8.0 Attachments 19
-------
7488/B
12-18-79
1.0 INTRODUCTION
On June 11, 1979, the U.S. Environmental Protection Agency (EPA)
promulgated revised new source performance standards (NSPS) for sulfur
dioxide (S02) emissions from utility boilers. The standards were based in
part on a series of flue gas desulfurization (FGD) system test data. The
data are available in air pollution emission test reports included in Utility
Boiler Docket OAQPS-78-1. In a petition to EPA for reconsideration of the
S02 percent reduction standards, the Utility Air Regulatory Group (UARG)
presented a statistical analysis of these data (OAQPS-78-1, VI-A-5). EPA
has also performed a statistical analysis of these data (OAQPS-78-1, VI-B-13).
The EPA analysis differs from the UARG analysis in that different conclusions
are reached based on distinctions in the type of FGD system, process control,
and mode of operation of the FGD systems tested.
The purpose of this report is to discuss the conditions under which
these data were obtained and to explain the distinctions in the FGD systems
tested. Existing S02 removal efficiency test data are examined to determine
representative efficiency variabilities. This report also focuses on the
control of FGD efficiency variation by process control systems. Typical
process control modules used in FGD systems in the United States and the
more advanced, automatic control modules used in Japan are also discussed.
Other factors that can affect FGD performance variation are considered.
2.0 SUMMARY
In considering the relevance of the test data to a projection of
efficiency variability in new FGD systems, only lime/limestone systems were
included because they are the primary (but not only) basis for the NSPS.
-------
7488/B
12-18-79
In Addition, they are the only type of FGD system control equipment venders
can supply in sufficient quantity to meet the total projected market demand
_ _ - - _.... _ f
(OAQPS-78-1, II-A-33). In addition, lime/limestone systems, in comparison
to regenerable FGD systems, are considered to be less responsive to process
control systems because of the slower reaction rate of absorbent. Thus,
the process control variation from a lime/limestone system is the "worst
case" example with respect to applying process control. In evaluating the
90 percent reduction requirement, FGD application on high-sulfur coal is
the most relevant example because low-sulfur coal applications would be
affected by the 0.6 Ib/million Btu emission floor or the 70 percent minimum
control level.
An examination of the test conditions shows that, with the exception
of three tests, the facilities tested experienced equipment malfunction or
operating problems or were otherwise not characteristic of well-designed
and well-operated lime/limestone systems on high-sulfur coal applications.
The three facilities that experienced no malfunctions or abnormal operating
conditions were Shawnee, Pittsburgh 2, and Louisville. Of these, Louisville
experienced the largest efficiency variability and Shawnee the smallest.
The degree of supervision over the process was greater at Shawnee than at
any other system tested. Close process supervision and other factors dis-
cussed below have resulted in good process stability (low efficiency
variability). New FGD system designs can be expected to utilize new control
systems (as has been done in Japan) and benefit from the experience of the
best U.S. and Japanese facilities.
3.0 REVIEW OF EPA TEST DATA
During the development of new source performance standards (NSPS) for
S02 emissions from electric utility generation units, the U.S. Environmental
-------
7488/B
Protection Agency (EPA) conducted various field studies of FGD systems to
better define the performance and operational characteristics of lime/lime-
stone slurry FGD processes. Since the promulgation of the Utility Boiler
NSPS, the UARG has further analyzed the EPA data to evaluate the capability
of FGD systems to achieve 90 percent S02 removal (OAQPS-78-1, VI-A-5).
Some of the data used in their analysis were obtained from FGD systems without
adequate process control systems and from FGD systems not operating under
normal conditions.
In fact, the EPA test data were obtained under varying conditions.
Table 1 lists the locations of the FGD systems tested in EPA field studies
in the order in which they are discussed below along with the type of scrubber
modules for each facility. Only tests conducted on lime/limestone FGD systems
and high-sulfur coal are considered in the final analysis because the most
stringent S02 percent reduction requirement, 90 percent S02 removal, is
applied only to high-sulfur coal. Review of the operating conditions and
types of FGD systems tested shows that only the data from Pittsburgh 2.
Shawnee, and Louisville (the last three in Table 1) were obtained from lime/
limestone slurry FGD systems under normal operating conditions, treating
flue gas from high-sulfur coal./Two other tests (Eddystone and Mitchell)
were conducted on non-lime/limestone systems and thus are not discussed.
Four other tests (Conesville A and B, Lawrence and Pittsburgh 1) were con-
ducted under conditions that did not represent well-operated lime/limestone
slurry FGD systems for high-sulfur coal. The specific reasons these four
FGD systems do not represent good operation are discussed below.
At the Conesville Power Station of Columbus and Southern Ohio Electric
Company, two modules (A and B) of the turbulent contact absorber (TCA) system
-------
7488/B
12-18-79
Table 1. EPA FGD SYSTEM DATA BASE
Site
Scrubber module
Comments
0 Eddystone (Philadelphia, Pennsylvania) Magnesium oxide
O Mitchell (Chicago, Illinois)
^Conesville A (Columbus, Ohio)
Conesville B (Columbus, Ohio)
Lawrence (La Cygne, Kansas)
Wei 1 man-Lord
TCA/lime
TCA/lime
Rod and spray
towers/1imestone
O Pittsburgh 1 (Bruce Mansfield, Pennsylvania) Venturi/lime
O Pittsburgh 2 (Bruce Mansfield, Pennsylvania) Venturi/lime
t1 Shawnee (TVA, Paducah, Kentucky) TCA/lime
O Cane Run (Louisville, Kentucky) TCA/lime
Not a lime or limestone
system
Not a lime or limestone
system
Intermittent operation
Low data availability
Intermittent operation
Low data availability
Low absolute S02
emission rate
Poor pH control
Step change of L/G
-------
7488/B
12-18-79
were tested. Lime was used as the scrubbing reagent (OAQPS-78-1, IV-A-8).
The many equipment failures encountered during the tests reduced the
reliability of the modules to a low level. The problems included:
(1) pressure surges in the piping,
(2) maldistribution of gas flow,
(3) excessive mist entrainment,
(4) acid rainout,
(5) failure of the plastic coating in the stack flues,
(6) failure of the plastic piping,
(7) plugging in the thickener lines,
(8) corrosion in the presaturator, and
(9) detachment of the rubber lining in the scrubber and recycle tank.
These problems resulted in excessive downtime for repair and consequently
led to short run duration and low data availability* (OAQPS 78-1, IV-D-611).
The longest continuous run of a module was 10 operating days. A more typical
run duration lasted 4 to 6 days. Over a 6-month period from June 15, 1978,
to December 15, 1978, module A was operated intermittently for 60 days;
however, 24-hour average S02 removal efficiency data were available for
only 24 days. Module B was operated 63 days, and the data were available
for only 21 days. These represent 40 percent and 33 percent data availa-
bility, based on scrubber operating time for Modules A and B, respectively
(OAQPS-78-1, IV-A-19). These figures are low compared with the 85.7 percent
data availability over a 49-day period at the Shawnee test site (OAQPS-78-1,
JU
Data availability is defined as the number of days for which 24-hour
average S02 removal efficiency data are available divided by the possible
number of operating days. An operating day is defined as 18 or more hours
of operation in a 24-hour period beginning at midnight.
-------
7488/B
12-18-79
IV-J-20). In the latter stage of data acquisition at the Conesville Power
Station, EPA staff determined that the FGD system could not be operated
under maximum removal conditions (see Attachment 1) because of original
design problems, and that improvement was not likely without major equipment
changes. The tests at the Conesville facility were subsequently terminated
by EPA. Because of the continual system plugging and resulting intermittent
operation during the Conesville tests, the efficiency variability of the
»» m__m_u_i_ii_Mir-r'. i " r~i B.-I.- ' ..i_i i ii ii - i - - _j.j-__.ii-.- __.-i_i.- —i- - ii
data did not represent that of a well-controlled, normal FGD system operation.
In tests at Kansas Power and Light Company's Lawrence Unit No. 4, limestone
slurry was used in both rod scrubbers and spray towers to absorb S02 from
low-sulfur coal (0.5 percent sulfur) flue gas (OAQPS-78-1, II-I-5). This
system was originally intended to achieve S02 removal of 73 percent, but
because of the low sulfur content of the coal used, the system was able to
achieve high removal. The median S02 removal efficiency during the test
was 96.6 percent. The combination of low-sulfur coal and high S02 removal
by the FGD resulted in an extremely low median outlet S02 emission of
0.03 lb/106 Btu (OAQPS-78-1, IV-J-22). Therefore, small variations in outlet
S02 concentration appeared as a significant efficiency variability, especially
when log-normal distribution [In (1-eff.)] is used to describe the efficiency
variability (OAQPS-78-1, VI-A-5).
As stated above, the Lawrence test results were obtained under significantly
different conditions (i.e., low-sulfur coal and high efficiency S02 removal)
than the other tests discussed in this report. If the test results were
combined, the results would be misleading, especially because the statistical
methods employed exaggerate variability. The results from the Lawrence
facility, therefore, should not be used to evaluate the process stability
(i.e., efficiency variability) of high-sulfur coal FGD systems.
-------
7488/B
12-18-79
The FGD system used at the Bruce Mansfield Station of Pennsylvania
Power Company consists of venturi absorber modules with lime slurry used as
the reagent (OAQPS-78-1, IV-D-611). Pittsburgh 1 and 2 (Table 1) represent
the data collected in two different periods of operation at the same plant.
The pH meter of the FGD process control system did not work during the first
period of testing. Initial problems occurred with pH meter electronics;
later, the suspended solids in the slurry tended to abrade and break the
flow-type pH probes. Operators had to record hourly scrubber and absorber
slurry pH readings with a portable pH meter. Thus, the entire FGD system
was not under proper control during the first period, and the data obtained
can not be used to interpret system performance adequately.
During the second test period at Bruce Mansfield, the FGD system operated
under good pH control conditions (OAQPS-78-1, II-D-432). The electronic
problem with the pH meter had been resolved, and other improvements made
before the second test period began included the relocation of the pH meter
and modification of the sampling procedure. Scrubbing pH was better controlled,
and S02 removal efficiency was consistently improved. In addition, the
modules operated without any substantial development of hard scale (gypsum)
or plugging (OAQPS-78-1, II-B-86). For these reasons, the data obtained
during the second test porjnri a+ Pit.tshurgh represent the performance of a 5
well-operated FGD system equipped with manga] proce.^ mnt.rnls.
Testing at the Shawnee Power Station, which is operated by the Tennessee
Valley Authority (TVA), was performed with a pilot-scale (10 MW) TCA system
using lime slurry (OAQPS-78-1, III-B-4). During these tests, data availability
was 85.7 percent, and neither pH meter malfunctions nor serious scaling
occurred. The process was also closely monitored by skilled technicians
-------
7488/B
12-18-79
(OAQPS-78-1, II-D-440). Therefore, data from TVA's Shawnee plant represent
normal FGD system performance of an existing system using manual pH controls
coupled with close operator attention. With respect to operator attention,
the Shawnee test was superior to any other.
Tests were also conducted at the Louisville Gas and Electric Company's
Cane Run No. 4 plant. This FGD system is a TCA using lime as the reagent.
Many problems occurred during the FGD system's initial operating phase,
poor gas distribution and excess pressure drop, which resulted in low S02
removal efficiency (OAQPS-78-1, IV-A-8). To improve the S02 removal effi-
ciency, additional spray headers were installed in the TCA and additional
slurry recirculation pumps were retrofitted. Following these modifications,
no pH meter malfunctions were reported, and plugging and scaling have not
been problems (OAQPS-78-1, II-I-7).
The use of three fixed-speed pumps to control the slurry recirculation
rate, however, resulted in FGD system operation at two levels. Two pumps
were used when boiler load was lower than 150 MW and three pumps when boiler
load was higher than 150 MW (see Attachment 2). This arrangement introduced
a step change in the liquid-to-gas ratio (L/G) when boiler load varied across
150 MW (see Figure 1). Because this facility was used as a peaking unit,
the load varied during normal operation from 90 MW in the morningtoapeak
value of 175 MW in the afternoon (OAQPS 78-1, II-I-301). As shown in Figure 1,
the unusual pumping arrangement caused the L/G to vary fmm 35 tn fin ja
(a relatively wide range) whenever load changed. Examination of the data
shows that this abrupt change in recirculation rate occured in a 1-day operating
period (OAQPS 78-1-76/56, II-B-75). Such a variation of L/G can strongly
affect the performance of a TCA system as measured by S02 removal efficiency.
-------
70 -i
SO -,
50 -
L/G
(gal/macf)
30
i
80
I » r
100
120
HO 160
180
BOILER LOAD (MW)
Figure 1. Liquid-to-gas (L/G) ratios vs. boiler load changes
-------
7488/B
As the Shawnee data show, for a TCA/lime system, a change in the ratio of
scrubber slurry to flue gas from 60 to 30 gal/macf could cause the S02 removal
efficiency to decrease drastically (from 85 to 55 percent) when other operating
parameters were fixed (OAQPS-78-1, II-A-75). Adjustments in other process
parameters, such as pH, can partially compensate for this variability in
efficiency; however, the relatively slow response time of pH controls limits
their ability to follow rapid L/G changes.
The test data show Louisville to have the largest efficiency variability
TT|H •!•>•,! ^
of the three tested facilities which experienced no malfunctions and were
in constant operation during testing. The degree of supervision over the
process is not known and may have been a contributing factor to the process
variation in addition to the L/G ratio fluctuations. The degree of process
supervision at Shawnee was relatively high (see Sections 4.0 and 5.0), and
such process supervision is a factor in achieving good FGD process stability
(i.e., low efficiency variability).
In summary, the preceding analysis shows that data from Pittsburgh 2,
Shawnee TCA, and Louisville were obtained under normal operating conditions
and without malfunction of pH control meters. Table 2 shows the logarithmic
means and standard deviations of those three sets of data (OAQPS-78-1, VI-A-5).
Among them, Louisville has the highest S02 removal efficiency variability
(standard deviation). This may have resulted from the unusual arrangement
of the slurry pumps, which led to wide L/G variations. Also, the degree of
process supervision is not known. In contrast, the Shawnee TCA data give
the lowest performance variability because of careful operation by skilled
personnel.
10
-------
7488/B
12-18-79
Table 2. RESULTS OF STATISTICAL ANALYSIS OF PITTSBURGH 2 AND SHAWNEE DATA
(OAQPS-78-1, VI-B-13)
Site Process variation and mean
Geometric standard
Mean deviation
Cane Run (Louisville, Kentucky) 83.3-84.4 0.295-0.343
Pittsburgh 2 (Bruce Mansfield, Pennsylvania) 85.4 0.212
Shawnee TCA (TVA, Paducah, Kentucky) 88.5 0.182
4.0 PROCESS CONTROLS FOR FGD SYSTEMS
Existing FGD systems in the United States, including the Louisville,
Shawnee, and Pittsburgh facilities, use manual pH controls. A major reason
for pH control is the prevention of scaling1 (OAQPS-78-1, II-I-8). Continuous
maintenance of high S02 removal efficiency in new FGD systems requires additional
controls and operating procedures designed to stabilize the process. This
section describes the use and limitations of pH controls and other controls
expected to be installed on new FGD units that will improve process stability.
In a typical pH control scheme, a stream of scrubber slurry from the
recirculation loop is tapped and passed through a vessel containing the pH
sensor, as shown in Figure 2. If the inlet S02 concentration increases,
the increased S02 absorption reduces the pH of the scrubber slurry. When
the pH drops below a set point, the flow of reagent increases, as does the
rate of reaction with absorbed S02 in the scrubber hold tank. The pH sensor,
in turn, monitors the resulting increase in pH, and the controller responds
11
-------
RECIRCULATION PUMP
Figure 2. Lime/limestone scrubber system pH control mode.
2
CM
o
C\J
o
i
GAS OUTLET
RECIRCULATION SLURRY
GAS INLET
REAGENT
FEED
12
-------
7488/B
12-18-79
by reducing the reagent slurry flow rate. This is a continuous process
that oscillates within a band across the set pH point.2
A number of factors determine the width of the pH control band. The
H is constrained by two factors: (1) the reagent feed rate required
to achieve a given degree of S02 removal and (2) the pH level below which
hard scale (gypsum) formation will tend to occur. These factors depend on
both system design and operating conditions such as inlet S02 concentration
and liquid-to-gas ratio.
The maximum pH of the slurry is generally limited by the reagent
*
utilization factor, which may include calcium carbonate (CaC03) scaling in
lime systems. An adequate margin must be provided within the upper and
lower pH limits so that a little "overshoot" will not cause scale formation
2
problems.
The most common problems associated with pH control systems are plugging
and pH sensor erosion. For example, in the Pittsburgh FGD system, a flow-type
sensor is used to measure the slurry pH (OAQPS-78-1, II-I-6). To do so,
slurry sample must be forced through the sensor. The sample lines to and
from the sensor elements can plug, and the sensor elements are constantly
subject to erosion. During the Pittsburgh 1 test period, plugging and pH
sensor erosion led to control system failure and caused unstable operating
conditions (OAQPS-78-1, IV-D-611). After experimentation, the pH measuring
equipment was moved to a more desirable location in the scrubber cycle, and
the sampling techniques were improved. With these modifications, the pH
sensor worked satisfactorily (OAQPS-78-1, II-1-6). The FGD system was under
The amount of reagent reacted with S02 divided by the amount of reagent
added to the system.
13
-------
7488/B
12-18-79
controlled, stable operating conditions during the Pittsburgh 2 test period
at Bruce Mansfield.
The plugging and erosion problems encountered in pH control systems
may also be solved through the use of dip-type pH sensors. The dip-type
sensor is inserted into the slurry tank and can be manually removed for
periodic maintenance and calibration. Since the slurry does not flow through
the cell when dip-type sensors are used, the measuring electrode, which is
made of glass, is not subject to breakage as it is when flow type sensors
are used. In the Louisville FGD system, six sets of dip-type pH electrodes
were installed in the recirculation tank. Each set was checked daily against
the others, and recalibration and repairs were performed as needed. This
o
pH control system has worked well, and no maintenance problems were reported.
The majority of lime or limestone scrubber installations in the
United States have used manual pH control. EPA has demonstrated the perfor-
mance capabilities of the pH control method in a pilot-scale FGD system
(Shawnee TCA) and in full-scale FGD systems (Pittsburgh 2 and Cane Run).
The efficiency variability of these systems is shown in Table 2.
Control system performance can be improved by (1) conversion of the
feed-back type of process controls to feed-forward controls and (2) the use
of automatic controls. In pH control systems, which are feed-back controls,
the pH instruments do not detect system variations before the slurry hold
tank chemistry has been upset. Thus, the reagent (lime or limestone) is
not added until a depletion has already occurred. In contrast, a feed-forward
process control uses instruments at the FGD absorber inlet to maintain the
balance of the FGD system as inlet conditions change. Signals from the
inlet S02 continuous monitoring equipment enable the operator to anticipate
14
-------
7488/B
12-18-79
the variations in the FGD system. Thus, slurry tank reagent feed rate may
be adjusted before the absorbent in the hold tank is significantly depleted.
Automatic controls improve on those described above by anticipating
the effects of changes in S02 inlet concentration and gas flow rate caused
by boiler load changes. Such systems, which are currently used in Japan,
have the potential to reduce variations in S02 removal efficiency (OAQPS-78-1,
II-I-359). These automatic process controls measure flue gas volume and
S02 concentration at the scrubber inlet to determine automatically the slurry
makeup volume requirements. Fine tuning of the makeup feed rate is maintained
by pH control. In addition to the automatic controls, a feed-forward control
scheme maintains the stoichiometric ratio between the absorbed S02 and the
added reagent. Scale formation is avoided by pH adjustment. Thus, through
the use of automatic controls and feed-forward process controls, the balance
of the FGD process chemistry is maintained, and the S02 removal efficiency
is stabilized (OAQPS-78-1, IV-B-4).
5.0 FURTHER REDUCTION OF EFFICIENCY VARIABILITY
In addition to process controls, the following factors may influence
FGD system stability and should receive close attention.
Qualified Full-Time Personnel
Qualified full-time personnel are needed to operate and maintain an
FGD system. An FGD system involves a chemical process that must be monitored
by well-trained operators to maintain the chemical balance of the system.
For example, the Shawnee TCA results were achieved using skilled personnel
to monitor and control the process (OAQPS-78-1, II-D-440). The results
demonstrate that better performance can result when an adequately trained
staff is dedicated to the operation of an FGD system. In Japan, assigning
15
-------
7488/B
12-18-79
adequately trained operators to FGD process control functions is considered
a major factor in achieving good process stability (OAQPS-78-1, II-I-359;
OAQPS-78-1, II-B-96).
Proper FGD System Design
FGD control systems should be designed so that process stability is
manageable. Proper operating margin and flexibility should be incorporated
into FGD systems to provide greater tolerance. For example, an adequate
margin must be provided within the upper and lower pH limits for the pH
control band to diminish the effect of overshoot on scale formation. The
appropriate slurry flow rate must be determined so that the variation of
L/G with boiler load change will not upset FGD system performance.
Stabilization of Inlet S02 Concentration by Coal Blending
An inherent property of coal is its variability in sulfur content.
Fuel sulfur is unevenly distributed in a coal seam, and even coal shipments
from the same mine will have a range of sulfur content (OAQPS-78-1, III-B-4).
Thus, all naturally occurring coals can be expected to produce variations
in the S02 concentrations of boiler flue gas. Coal blending is one method
of reducing the variation in fuel sulfur concentration and, consequently,
of reducing the control system's burden of maintaining FGD process stability.
However, the influence of the variability of coal sulfur content on the
efficiency of the scrubbing operation has not been found to be a major factor
(OAQPS-78-1, VI-B-13).
Quality Control of Lime and Limestone
The reactivity of reagents, especially limestone, is affected by both
their particle size distribution and their inert content. The quality of
the reagents should be controlled to maintain an accurate stoichiometric
16
-------
7488/B
12-18-79
ratio between inlet S02 and reagent feed rate and thus avoid system chemistry
upset (OAQPS-78-1, II-I-359).
6.0 CONCLUSIONS
The following observations are summarized from the above discussion:
1. The performance test results from Cane Run at Louisville, Kentucky,
Pittsburgh 2 at Bruce Mansfield, Pennsylvania, and TVA Shawnee TCA at Paducah,
Kentucky, were obtained under normal operating conditions and without pH
control malfunction. However, the S02 removal efficiency variability of
the Louisville test was higher than that of the other two tests. It may
have resulted from step change coupled with a wide range in L/G ratio which
can have adverse effects on FGD system stability. In addition, operator
attention at Shawnee was superior to any other system tested.
2. Good results can be achieved with pH controls through careful
attention to probe location and operating procedures, as demonstrated by
the Pittsburgh and Shawnee tests.
3. Existing FGD systems can be improved by using automatic control
systems designed to keep proper balance in the stoichiometric ratio between
reagent fed and S02 absorbed and by maintaining slurry pH stability. Such
controls are currently used in Japan. Future FGD systems equipped with
those advanced control modes are expected to achieve performance variability
equal to or better than the Shawnee and Pittsburgh tests.
4. Several other factors, including adequately trained full-time
operators, contribute to the achievement of process stability. These factors
should be considered in conjunction with well-designed process controls in
new FGD installations.
17
-------
7488/B
12-18-79
7.0 REFERENCES
1. Lime/Limestone Scrubber Operation and Control Study. Electric Power
Research Institute. Palo Alto, California. Publication No. EPRI FP-627.
October 1978.
2. Jones, D. G. , 0. W. Hargrove, and T. M. Morasky. Lime/Limestone Scrubber
Operation and Control. Control Technology News, Journal of the Air
Pollution Control Association. 29:1099-1105. October 1979.
3. Lime FGD Systems Data Book. Electric Power Research Institute. Palo,
Alto, California. Publication No. EPRI FP-1030. May 1979.
18
-------
7488/B
12-18-79
8.0 ATTACHMENTS
1. Kelly, W. M. Trip Report—CSDEC Conesville Unit 5 S02 Monitoring
Project, June 27, 1978.
2. Chang, J. C. S. Telephone Conversation Report, Control of L/G at
Louisville Plant FGD (Cane Run), November 15, 1979.
19
-------
ATTACHMENT 1
UNITED STATES ENVIRONMENTAL
JUN 271978 Offlce of Air Quality PIanni
DATE: Research Triangle Park, North"Carolina 27711
Trip Report—CSOEC Conesville Unit 5 S02 Monitoring Project
FROM: Winton E. Kelly, Project Officer I
Field Testing Section, Emission Measurement Branch,.ESED (MQ-l'3)
T0: J. E. McCarley, Jr.s Chief, Field Testing Section
Emission Measurement Branch, ESED (MD-13)
I. PURPOSE
The purpose of this visit was to review the S0? monitoring instru-
mentation, the contractor's progress in instrument specification testing,
and to arrange for acquisition of boiler process data.
II. PLACE AND DATES
The visit was conducted at Columbus and Southern Ohio Electric
Company, Conesville Plant, Conesville, Ohio, on June 15-16, 1978.
III. PRINCIPAL ATTENDEES
CSOEC
Tom Reed
Harl Todd
UOP. Air Correction Division
Dave Lovetere
PEDCo
Tony Wisbith
Dave Howie
Jon Allen
Environmental Protection Agency
Winton E. Kelly
IV. DISCUSSION
The instrument systems were found to be appropriately installed
and operating properly. PEDCo had completed final specification testing
on June 14. Final results were not available at the time of the visit,
but PEDCo personnel felt confident that all performance criteria would
be achieved. The only instrument operating difficulty has been
PA FORM 1320-6 (REV. 3-76)
-------
deterioration of the automatic blowback valves on the probe systems.
This problem was corrected by installing an additional solenoid valve in
the high pressure air lines. A minor difficulty was observed in the
timer cams for the Dupont track and hold circuits, but correction is a
matter of easy adjustment. Based on observations of PEDCo's procedures
and preliminary review of reference method measurements, the instrument
systems are providing reliable, accurate results.
Since all installations and performance tests had been completed,
PEDCo was instructed to begin the formal test period as of June 15, 1978.
Process data will be collected beginning on the same date. The PEDCo field
personnel are compiling very complete logs of instrument operation and
specific instructions were given to ensure that all periods of missing or
unusable data are properly accounted. PEDCo was instructed to use the time
allowances currently under consideration for routine calibration and
maintenance; 1 hour per day for calibration and 8 hours per month for
routine maintenance. Additional testing was directed to determine a
relative accuracy for the 02 analyzer. Procedures to be included in
revised Specification 3 were described and PEDCo was instructed on their
use.
Based on their work in ambient monitoring studies, PEDCo is aware
of the need for quality assurance procedures and was familiar with the
approaches currently proposed by QAB/EMSL for ambient monitors. In order
to provide monthly data for instrument relative accuracy, PEDCo was
instructed to perform Method 6 at both test locations and Method 3 analysis"
on samples extracted at the oxygerf analyzer at 30-day interval's. A multi-
point calibration will also be performed at these intervals. Daily
calibration data (zero and calibration drift) will be available from
instrument log records. As an additional procedure, EMB could.perform
regular audits with NBS gases.
The scrubber was surveyed to determine the easiest way to monitor the
amount of flue gas bypassed. There are presently no access ports available
in the bypass flue. Flow exiting each module is monitored using calibrated
annubars. Total flow from the boiler is not monitored. It was decided
that the initial approach should be to calculate the total furnace exhaust
flow based on F-factor, coal feed, oxygen, temperature, and moisture results.
The sum of gases exiting modules will be subtracted to determine bypass.
It should be noted that this may prove to be an inadequate approach since
small errors in the larger flow rate determinations could result in large
percentage errors in the absolute value of the bypass. As soon as a base
of data is available, calculation will be performed to estimate the accuracy
of this procedure. In order to confirm the annibar calibrations, PEDCo
will perform velocity traverses now and at 30-day intervals at each scrubber
outlet.
-------
In discussions with CSOEC personnel, it was determined that the
easiest way to obtain the necessary boiler parameters would be to
have the scrubber operator record the necessary information on a revised
scrubber log sheet. Since the operator calls the boiler room at hourly
intervals, information in addition to boiler load can be obtained. In
order to simplify this procedure, requested boiler parameters were limited
to: (1) boiler load, MW, (2) total coal feed in past hour, (3) scrubber
inlet temperature, and (4) boiler 02. Copies of the forms to be used by
CSOEC were obtained. Weekly test intervals will run from midnight Thursday
to midnight Thursday. Strip charts will be collected at corresponding
intervals for the analyzer system.
In further discussion with CSOEC personnel, it was determined that
the operating results (percent S02 removal) reported earlier (record of
communication—Winton Kelly dated May 30, 1978) were obtained when the
system was operated under maximum removal conditions for the unit compliance
and acceptance tests. These conditions could not be maintained for more than
one to two weeks. The required slurry usage was excessive and the system
thickener was too small to handle the resultant fine material. The fines
buildup quickly leads to blinding and complete system pluggage. This was
experienced during the test and as a result, only one module was operational
during the visit. The other module was being cleaned for restart. Efficiency
levels observed during the earlier tests would not be applicable to longer
term averages.
At the time of this visit, the operating module was treating about
50 - 60 percent of the flue gases, at full boiler load. The resulting ~
emission levels ranged from 550 to 650 ppm at 5 percent 02 on June 16.
On June 15, the S02 concentrations ranged from 450 to 550 ppm at 7.5
percent Q?. The inlet S02 concentrations were 2900 to 3000 ppm at 5
percent 0~. S0? efficiencies based on these averages are 75 to 79 percent.
No attempt was made to evaluate the variability of the data since most of
the recent operating time was with only one operational module and
current strip charts would have to be removed from the recorders.
CSOEC practices require that when one module is down, the remaining
module treats as much gas as possible. This condition probably overloads
the system, resulting in less than design S02 removals. An availability
test on the system should be performed during our monitoring study. During
this test, at least one module must be available continuously for 30 days
and still be in operational condition at the test conclusion.
b
In additional discussions, it was determined that the scrubber
system was designed for automatic control, and the inlet and outlet S02
levels were control parameters. However, CSOEC has not been able to run
on automatic for many reasons, one of them being the lack of successful
analyzer operation. They also feel that the cost of manpower necessary
for instrument maintenance would be excessive.
-------
4
V. CONCLUSIONS AND RECOMMENDATIONS
The instrument system design appears to have incorporated all
improvements developed in earlier tests. The system should yield high
levels of data availability. PEDCo is performing in an excellent manner
and conscientiously attending to information gathering concerning instru-
ment operation. A formal program of quality assurance has been formulated
and will be instituted. Additional relative accuracy tests will be
performed on the oxygen analyzers. Scrubber bypass will be estimated
using a different technique unless the calculations prove inadequate.
The results obtained during compliance testing are not expected to be
representative of long-term operation. Current operation with one
module is about 75 to 80 percent S02 removal on about half the gas flow.
No estimate can be made from test information concerning operation with
both modules on line.
cc: Charles Sedman
Ken Durkee
Dick Gerstle, PEDCo
Larry Jones
Lou Paley, DSSE
Roger Shigehara
Gene Smith
FILE: 78-SPP-28
ESED:EMB:W.KELLY:mew:Rm730MU;x5243:6/27/78
-------
ACUREX
Aerotherm
Person placing call:
Subject:
ATTACHMENT 2
TELECOMMUNICATION
Date:
Contact: My.
JLoa/sri/te Want
Followup required: D Yes D No Responsible person/date
Name:
Phone Number:
Title:
Synopsis of discussion:
$
^^ <<
SL.
Copies to be routed to:
m EED 04H
------- |