xvEPA
United States
Environmental Protection
Agency
Industrial Environmental Research
Laboratory
Research Triangle Park NC 27711
Research and Development
EPA-600/S7-82-032 Oct. 1982
Project Summary
Studies of Flue Gas
Desulfurization at Louisville
Gas and Electric's Paddy's
Run Station
0. W. Hargrove, Jr., G. P. Behrens, and W. E. Corbett
Between the Spring of 1973 and the
Fall of 1976, Louisville Gas and
Electric's Paddy's Run lime flue gas
desulfurization (FGD) system logged
more than 4000 hours of operation
without any major process or me-
chanical problems. Due to this oper-
ating success when other similar
systems were encountering numerous
problems, EPA funded a 6-month
evaluation study at Paddy's Run. A
program was implemented to charac-
terize the system in its normal mode of
operation and to conduct tests which
would simulate conditions typical of
other lime-based systems.
The Paddy's Run FGD system
normally uses carbide lime, a by-
product in the production of acetylene,
as the alkaline additive. Contaminants
in the carbide lime, notably thiosulf ate,
are responsible for minimizing the
sulfite oxidation rate which helps to
maintain scale-free operation of the
system. Substitution of commercial
lime for carbide lime resulted in
gypsum scaling in the scrubber.
Magnesium addition (—3000 ppm) with
commercial lime reestablished scale-
free operation and improved SO2
removal dramatically. Addition of
either carbide lime or commercial lime
with magnesium resulted in sulfite
oxidation levels well below those
required to operate subsaturated with
respect to gypsum (less than about 16
percent). Under these conditions, all
of the sulfate formed precipitated in a
solid solution with calcium sulfite
hemihydrate.
Additional tests examined the effect
of chloride addition, lime addition,
point location, and the reaction tank
volume.
This Project Summary was devel-
oped by EPA's Industrial Environmen-
tal Research Laboratory, Research
Triangle Park, NC, to announce key
findings of the research project that is
fully documented in a separate report
of the same title (see Project Report
ordering information at back).
Introduction
Louisville Gas and Electric (LG&E)
Company's Paddy's Run No. 6 flue gas
desulfurization (FGD) system was one of
the first commercial-scale FGD units to
be operated successfully in the U.S.
Between its. start-up in the Spring of
1973 and the Fall of 1976, the unit
logged close to 4000 hours of operation'
without any major process or mechan-
ical problems. Because of the demon-
strated success of the Paddy's Run
system, the EPA sponsored a test
program to determine the factors which
account for the successful operation of
this FGD system.
At the time that this test program was
conceived, a wide range of operating
problems were being encountered in
many FGD units which were operating.
One of the more serious process
-------
problems was chemical scaling in the
scrubber. A method being investigated
by many experimenters to avoid this
problem was to operate the FGD system
in such a manner that the slurry liquor
remains subsaturated with respect to
calcium sulfate dihydrate (gypsum).
Recognizing this fact, the EPA conceived
a program whose overall objectives
were to characterize the performance of
the LG&E system in its "normal"
operating mode, and to determine the
features which make it possible for this
system to operate in a subsaturated
gypsum mode. Additional tests were
planned to simulate conditions more
typical of those encountered in other
lime-based systems and to determine
the effects of changes in important
process operating parameters.
The intent of the program was to
provide useful information which could
be applied to other FGD systems.
Ultimately, it was hoped that data from
the successful operation at Paddy's Run
could be extrapolated to other lime/
limestone systems.
System Description
The flue gas from the Paddy's Run No.
6 boiler (65 M W) first passes through an
ESP and then is treated in one of two
parallel marble-bed scrubber modules.
The inlet flue gas typically contains
1500-3000 ppm of S02. During the
EPA test program, the boiler was
operated at half load, requiring the use
of only one scrubber module.
The major features of the FGD unit
are shown in Figure 1. Each scrubber
module contains two marble beds in
series, followed by two banks of chevron
mist eliminators. The scrubbing liquor is
introduced to the scrubber through
spray nozzles below each bed. The
scrubber effluent liquid streams consist
of the overflow liquors from the two
marble beds as well as the scrubber
bottoms liquor. Approximately 7 cm (3
in.) of marbles are contained on each
bed. The downcomer weirs are of such a
height that the active gas/liquid contact
zone on each bed is about 25 cm (10 in.).
A unique feature of this system is the
mixing well into which the three
scrubber effluent streams, the clarifier
overflow liquor, and the lime additive
slurry flow before they enter the main
reaction tank. The effective residence
time of the mixing well is about 30
seconds, whereas the residence times
of the main hold tank and surge tank are
about 35 and 5 minutes, respectively (all
of these residence times are for half-
load conditions). The residence time of
the liquor on the beds averages 30
seconds, while the normal scrubber
L/G is about 7.5 l/Nm3 (56 gal./103
scf).
The alkaline additive normally used at
Paddy's Run is carbide lime, purchased
from a local acetylene manufacturer.
The two basic steps in producing
acetylene are: (1) commercial lime and
coke are heated to form calcium
carbide, and (2) the calcium carbide is
then reacted with water to form acetylene
and calcium hydroxide, commonly
called carbide lime. This regenerated
carbide lime retains some impurities
from the processing which are carried
into the lime FGD system at Paddy's
Run. The carbide lime additive is stored
as a 25 percent slurry in an additive tank
Gas
out
Natural f
gas ' |
22,500 l/min.
, L/G =7.5 l/Nm3
10% solids
Flue gas -»
180,000
Nm3/hr
Reaction tank
770,000 I
-+40% solids
to pond
Scrubber
bottoms
tank
102,000 I
Figure 1. Simplified flow diagram - Paddy's Run No. 6 FGD system.
2
and is fed to the draft tube on pH
demand. The pH of the reaction tank
effluent is generally 8.
The surge tank effluent is the scrubber
feed stream. In addition, a portion of this
stream is routed to the clarifier to
maintain a circulating slurry solids level
of 10 weight percent. The clarifier
underflow (about 20 percent solids) is
sent to a set of two vacuum filters which
produce a filter cake containing 35-40
weight solids.
System makeup water comes from:
(1) mist eliminator wash water, (2)
additive slurry, (3) turning vane spray
wash water, and (4) seal water for
pumps and mixers. Mist eliminator and
turning vane spray washes are intermit-
tent; seal water and the lime slurry are
continuous makeup water sources.
Program Objectives
The overall objective of the test
program was to determine quantitatively
the reasons for the operating success of
the Paddy's Run System. This unit had
not experienced problems such as
chemical scaling or mist eliminator
pluggage which have been associated
with many other lime and limestone
SOa scrubbing systems. One important
aspect of this trouble-free operation is
the system's ability to operate subsat-
urated with respect to gypsum. This
permits the system to operate without
gypsum scaling problems.
It has been established that subsatur-
ated conditions are related to sulfate
coprecipitation with calcium sulfite.
Laboratory studies have verified the
formation of calcium sulfite-sulfate
hemihydrate solid solution1. These
studies indicate that sulfate can be
incorporated into the calcium sulfite
lattice to a maximum sulfate to total
sulfur ratio of about 0.16. In addition,
the sulfate/total sulfur ratio in the
solids is directly related to the gypsum
relative saturation in the liquor. As long
as the sulfate production rate does not
exceed the rate at which sulfate can be
incorporated into the sulfite lattice, the
liquor from which the solids precipitate
will remain subsaturated with respect
to gypsum. Gypsum scaling is not a
problem in a liquor which is subsaturated
with respect to gypsum.
There were several areas of interest
regarding operation in the subsaturated
mode. The primary area is the sulfite
oxidation rate. System parameters
which may impact the oxidation rate or
otherwise influence the subsaturated
mode of operation include additive type
-------
(carbide lime vs. other additives),
reaction tank configuration, and soluble
ion concentrations. It was anticipated
that studies involving these parameters
would yield insight into the question of
why subsaturated gypsum operation is
possible at Paddy's Run.
To fulfill the overall program objectives,
four test phases were originally out-
lined:
Phase I—Carbide Lime Testing.
Phase II—Commercial Lime Testing.
Phase III—Hold Tank Modification
Testing.
Phase IV—Chloride/MgO Testing.
The objectives of each test phase are
discussed below.
Phase I—Carbide Lime
Testing
The initial phase of testing involved
characterizing the LG & E Paddy's Run
No. 6 FGD system in normal operation;
i.e., using carbide sludge as the lime
additive. The results of this test phase
served as the "base case" for compar-
ison with subsequent test results.
Phase II—Commercial Lime
Jesting
The second phase of testing was per-
formed using commercial lime as the
alkaline additive rather than carbide
lime. The major goal of this phase was to
identify differences in operation which
might be caused by the change in the
lime additive. Particular attention was
given to monitoring oxidation rate
effects because oxidation was known to
be a key variable in operating sub-
saturated with respect to gypsum.
All subsequent tests were conducted
using commercial lime as the additive.
Phase III—Hold Tank
Modification Testing
Lime systems have the advantage of
a very rapid additive dissolution rate.
Taking only this factor into account,
high lime utilization efficiencies should
be attainable even for very short reac-
tion tank residence times. In one portion
of this test phase, the reaction tank resi-
de nee time was reduced to study the im-
pact of this change on the performance
of the system.
A mechanical feature of LG&E's
Paddy's Run FGD unit, which is unique
to that particular system, is the mixing
well. It has been suggested that a key to
the success of the LG&E system is as-
sociated with the operation of this mix-
ing well. This hypothesis was tested in a
second reaction tank modification test
which eliminated the effect of the mixing
well. This was accomplished by changing
the lime addition point from the mixing
well to the reaction tank.
Phase IV—Chloride/MgO
Testing
The final phase involved testing the
effects of chloride and magnesium
levels on subsaturated gypsum opera-
tion. It had been reported that increased
chloride ion concentrations may
decrease the amount of sulfate which
can be purged from the system as a
solid solution with calcium sulfite.
Indications that increasing magnesium
concentrations enhance the coprecip-
itation of calcium sulfate with calcium
sulfite had also been reported. Sup-
porting evidence for these claims was
not found in recent laboratory studies
by Jones'3; however, measurements
of these effects in an actual operating
system were desired during this test
phase.
All major tests outlined for each test
phase were conducted during the test
program. However, due to differences
encountered between commercial and
carbide lime operation, the order of the
tests had to be changed. As discussed in
the Results section, operation with
commercial lime led to scaling in the
scrubber. Therefore, magnesium addi-
tion was employed to eliminate the scal-
ing conditions prior to the hold tank
modifications test phase. Consequently,
the test phases were renumbered to:
Phase I—Carbide Lime Testing.
Phase II—Commercial Lime Testing.
Phase III—MgO Addition Testing.
Phase IV—Hold Tank Modification
Testing.
Phase V—Chloride Addition Testing.
Magnesium addition was necessary
throughout the last three test phases to
maintain system operability.
Results
Phase I—Carbide Lime
Testing
Phase I was conducted over a 6-week
period beginning mid-November 1981.
The FGD system operated according to
LG&E's standard operating procedures
during this period. (The only change was
that the relatively high magnesium-
content liquor normally used to slurry
the carbide lime was not used, so that
carbide and commercial lime could be
compared with low dissolved magnesium
concentrations in the scrubbing liquor.)
The scrubber feed pH set point was 8.0
which resulted in scrubber effluent
liquor pHs of 4.5-5.0. The L/G during
this operating period (and through most
of the program) was about 7.5 l/Nm3 (56
gal./103 scf).
The S02 removal efficiency ranged
from 75 to 83 percent according to the
DuPont on-line analyzer. The DuPont
analyzer measured inlet SC>2 concentra-
tions between 1800 and 2000 ppm. Wet
test methods showed inlet SC>2 concen-
trations of 2000 to 2500 ppm. The SO2
removal efficiency calculated from wet
test results was about 5 percent lower
than that measured from DuPont
analyzer results because of higher
manually determined outlet concentra-
tions.
The concentrations of important
dissolved ions which were measured in
the system during Phase I are plotted in
Figure 2. Note that the calcium level
increased by about 50 to 70 percent
across the scrubber due to the dissolution
of solids in the low pH scrubber liquor.
Also, note that the dissolved sulfate
level also increased across the scrubber,
but only by about 20 to 40 percent.
The calcium solids dissolution which
occurred in the Paddy's Run system at
base case conditions is a significant
operating characteristic. Roughly 30 to
50 percent of the total alkalinity
required for SO2 removal was being
derived from calcium sulfite dissolution
in the scrubber. A significant portion of
the sulfate ion increase was also due to
solids dissolution since some sulfate
exists in a solid solution with the
calcium sulfite. The large amount of
solid phase alkalinity required was due
to the relatively low L/G employed
which limited the total available liquid
phase alkalinity. The relatively long
slurry residence time on the marble
beds (20-30 seconds) provided the time
required for the relatively slow solid
dissolution reactions to occur. Thus, the
mechanical design of the Paddy's Run
marble bed scrubbers resulted in the
large percentage increase in calcium
ion concentration which was observed
across the scrubber.
If the concentrations of both Ca" and
SOJ increase across the scrubber, the
gypsum relative saturation will also
increase. This effect is shown in Figure
3. Since the system oxidation level
remained below 15 percent during
Phase I, the scrubber feed liquor
remained subsaturated with respect to
gypsum. The lower bed downcomer
-------
liquor was also subsaturated during
most of this test phase, except for a
short period when the system oxidation
level approached 15 percent and the
inlet relative saturation approached 1.0.
However, the increased gypsum relative
saturation across the scrubber, which
was caused primarily by calcium solids
dissolution, was not sufficient to create
scaling problems during Phase I.
30
.o
c
S c
•y. g s
Q
30
25
2O
75
10
C
O
!0
t3
I
"o
-------
between 0.9 and 1.0. Since the increases
in Ca++ and SOI across the lower bed
were approximately equal during the
two phases, the increase in gypsum
relative saturation was about 30 percent
in both Phase I and Phase II. The
resulting lower bed effluent gypsum
relative saturation was between 1.1 and
1.3 during Phase II.
Gypsum relative saturations less than
1.3 are below what is generally consid-
ered to be the critical scaling limit.
However, the lower bed effluent stream
is a composite of the liquor leaving the
entire lower bed. Regions on the bed
which were burdened with higher than
average S02 loadings would have
experienced higher than average gypsum
relative saturations. In this situation,
localized scaling might be expected.
This is exactly what was observed
throughout Phase II, even when the
scrubber liquid rate was raised to its
maximum value.
The results from Phases I and II
support the data gathered during
coprecipitation laboratory studies which
were performed by Radian for EPA2. In
both phases, the gypsum relative
saturation of the scrubber feed liquor
•B fs
1.0
I
0.15
o
•S 0.10
a
£
0.05
Scrubber
effluent
*
\/777///m
Phase I
Results
Phase II
Results
approached equilibrium as the solid
sulfate/total sulfur ratio approached
0.16. This is about the maximum
amount of sulfate which can be incor-
porated into the calcium sulfite hemi-
hydrate crystal lattice. Infrared scans
run on solid samples from both phases
indicate that all of the sulfate in the solid
phase present was calcium sulfite-
sulfate solid solution. If any of the
gypsum solids formed in the scrubber
during Phase II were carried to the
reaction tank, they were redissolved
because the reaction tank liquor was
subsaturated with respect to gypsum.
Additionally, further work sponsored
by the EPA has identified several
reduced sulfur compounds as well as
polynuclear aromatics in the carbide
lime2. Thiosulfate is one of these com-
pounds which is a known inhibitor of the
oxidation reaction of sulfite to sulfate. It
is very likely that the successful opera-
tion of the Paddy's Run system with car-
bide lime is a result of these contami-
nants minimizing the sulfite oxidation
rate. When commercial lime was used
without inhibitors, the sulfite oxidation
increased to the point where the
increase in gypsum relative saturation
across the scrubber resulted in chemical
scaling.
Phase HI—MgO
Addition Testing
Once it was recognized that scaling
under base case conditions was un-
avoidable, a decision to begin magnes-
ium addition was made. Addition of
magnesium should increase liquid
phase alkalinity and decrease the re-
quirement for solids dissolution across
the scrubber. It was hoped that lower
solids dissolution rates would alleviate
the scaling.
The most dramatic effect of magne-
sium addition was its impact on the liquid
phase most alkalinity and the resulting
SO2 removal efficiency. Table 1 sum-
marizes this effect. Magnesium concen-
trations of 170 mmole/liter (—4000 ppm)
resulted in removal efficiencies above
Table 1. Effect of Magnesium Concentration On SOz Removal Efficiency
Dissolved Magnesium S02 Removal
Content (ppm) Efficiency (%)
99 percent. At magnesium concen-
trations of about 90-95 mmole/liter
(—2000 ppm), removals between 90and
95 percent were obtained which were
still significantly higher than the 75 per-
cent achieved with either commercial or
carbide lime alone.
In addition to the increased SO2
removal efficiency, the higher magnes-
ium levels had a major impact on system
operability. This is shown in Figure 6.
When the magnesium concentration
was maintained at 170 mmole/liter
(4000 ppm), essentially no calcium
dissolution occurred in the scrubber.
Under these conditions, the sulfate/total
sulfur ratio in the solid phase was about
0.05 and the gypsum relative saturation
was below 0.2 throughout the scrubbing
loop. However, when the magnesium
concentration was dropped to 90
mmole/liter (—2000 ppm), the scrubber
effluent pHs dropped, and calcium
dissolution was noted in the scrubber.
During the same period the solid phase
sulfate/total sulfur ratio rose to 0.13,
and gypsum scaling was observed on
scrubber internals. The gypsum scale
thus formed was dissolved as the
magnesium level was again increased.
It was noted, however, that operation
with magnesium addition is not without
its drawbacks. High magnesium levels
appear to lead to solids settling and
dewatering problems. During Phase III,
operation with magnesium concen-
trations above 180 mmole/liter (—4400
ppm) led to clarifier operating problems.
The increased magnesium levels in-
creased the soluble sulfite and sulfate
levels and created conditions which
were conducive to excessive nucleation.
This created small solid crystals which
would not settle very efficiently. Clarifier
settling problems were observed for
almost a week following 2 days of
operation at 4400 ppm magnesium. A
reduction in the magnesium level
reestablished the formation of rosette
crystal structures which settled much
more efficiently. Figure 7 is an electron
micrograph of scrubber solids when the
dewatering system was performing
satisfactorily.
Figure 5. Comparison of oxidation
rate and gypsum relative
saturation values - Phase I
vs. Phase II.
Note: Inlet flue gas contained
—2000 ppm SO2 and the dissolved
chloride concentration in the
scrubber liquor was <500 ppm.
200 (baseline}
2000
4000
75
90-95
99+
-------
Dissoved Mg
Concentration 750
(mmole/liter)
100
Dissolved Ca++ 20
Concentration JQ
(mmole/liter)
rt
CO 0 0 ° °0
0 o °° o '
00 ° °
Lower bed
**? Exit liquor
/ I Feed liquor
AA— -f< X_> f^-Affr-*-*! A^ •» « « - -,
Gypsum
relative
saturation
/.o
7.0
0.5
n
.Tj'Z.oive/'Aet/
/ 1 Ew'r //fl't/o/'
//V""
~^ L_
liquor
-J--C P.- C3- . -
SOf
Total S
in solids
0.1
0.05
°0
o
O Ooo oO.O^_
75 20 25 30 ' "5 ' ' 70 75 ' 20
Jt//?e 7S77 ^u/y 7377
'2
Figure 6. Impact of magnesium level on oxidation and gypsum scaling potential
(Phase III).
Phase IV—Hold Tank
Modification Jesting
Two reaction tank modifications were
tested during Phase IV operation. First,
the mixing well effect was eliminated by
relocating the lime additive point from
the mixing well to the main reaction
tank. In a second test, the residence
time of the reaction tank system was
reduced from about 35 minutes to about
5 minutes by bypassing the mam
reaction tank. Both tests were conducted
with Mg(OH)2 addition.
Operation in both configurations
proved to be successful. As expected,
the effects of magnesium on SOa
removal for each configuration were
quite similar to those shown in Phase III
However, there were some differences
in the gypsum relative saturations and
the sulfite oxidation fraction for the two
configurations. These differences are
shown in Figure 8. Oxidation was
somewhat higher during operation in
the reduced residence time configuration.
Mixing Well Elimination Test
Basically, the effects of magnesium
on total system operation in this
configuration were similar to those
experienced in Phase III. With the lime
additive point change, the system
remained operable at magnesium levels
above 100 mmole/liter (2400 ppm).
Reduced reaction
Dissolved
(mmole/liter)
Dissolved
Ca ^Concentration
(mmole/liter)
Gypsum
relative
saturation
^SO"
Total S
2OO
1
.5
.u
tn G
0. 15
O.05
^ Mixing we/1 elimination test •* ^-Volume
0000000000 oo
W°°OO wu°
Lower bed
. ""Exit liauor "*» A A
A ^sFeed liquor^ ^ A
A A « - •"•
A
JLower bed j>
*• Exit liquor
a*-Feed liquor** °
. n A A
A * n O O
o °
o 0
°0 - _«0°
0 0 U ° 0 0
test-*
oo
A
T
,
i
A .
•u-^
25
10
Figure 7. Typical rosette formed in
lime system (July 5,1977).
Figure 8.
15 20
July 1977 August 1977
Impact of magnesium on system operation during Phase IV.
24
-------
Slurry streams subsaturated with
respect to gypsum were present through-
out the system, and the solid phase
sulfate/total sulfur ratio remained
about 0.05 to 0.07. When the magnesium
level was reduced and maintained
below 100 mmole/liter, however,
significant calcium dissolution occurred
in the scrubber, and the oxidation
fraction and gypsum relative saturations
approached inoperable levels.
The main effect expected from the
lime additive point change was an
alteration in the product crystal structure.
When the lime additive enters the
mixing well, lime particles dissolve
directly into the scrubber effluent liquor,
which is relatively high in soluble sulfite
and has a relatively low pH. Locally high
relative saturations thus occur in the
region of the dissolving solid Ca(OH)a
particles which promotes nucleation. In
the configuration where lime enters the
mam reaction tank, the dissolving lime
particles encounter reaction tank liquor
which has a higher pH and lower
soluble sulfite concentration. Since the
localized relative saturations generated
in this situation would not be expected
to be as high, more orderly crystal
growth should be promoted by the "no
mixing well" configuration.
The results from the draft tube
elimination test support this theory.
Shortly after the lime additive point was
moved from the mix well, the crystal
morphology began changing from the
rosettes which were produced during
Phase III to structures which were more
platelet-like. During this transition,
solids settling problems were encount-
ered. However, continued operation in
the "no mixing well" configuration and a
reduction m magnesium concentration
resulted in calcium sulfite platelets
similar to those produced in many
limestone FGD systems. The large
platelets formed during the latter
portion of this test settled well and
clarifier and filter efficiencies were
restored. Figure 9 is an electron
micrograph of the platelet solids.
Reduced Reaction Tank
Volume
The objective of the reduced reaction
volume test was to assess system
operability with the main reaction tank
eliminated from the scrubbing loop. The
system did remain operable with the
major difference being increased oxida-
tion in the reduced volume configuration
(Figure 8). This increased oxidation was
thought to have been caused by liquid
level control problems in the mixing
well. At times, the upper agitator blade
in the mixing well was exposed to the
air causing a frothing effect which could
have increased 02 transfer into the
slurry relative to normal operation.
Other than this effect, operation with a
small reaction tank volume was very
similar to that with the normal reaction
tank configuration.
Phase V—Chloride Addition
Testing
The final test phase was devoted to
determining the effect of increased
chloride levels on system chemistry.
CaCb was added to the system to
increase the chloride level from 10
mmole/liter (350 ppm) to 80 mmole/liter
(2800 ppm). Since operating time was
limited during this phase, the magnesium
level was increased simultaneously to
maintain system operability. A ratio of 1
mole of magnesium added for each 2
moles of chloride was found to maintain
the scrubbing solution's liquid phase
alkalinity. The results of Phase V testing
are shown m Figure 10.
Since the magnesium concentration
was increased in conjunction with the
increase in chloride, extensive data
concerning the independent effect of
chloride are not available. However,
data gathered early in this phase, before
the system was at steady state, indicate
the chloride effect. An initial spike in
chloride concentration of 270 mmole/liter
(9600 ppm) occurred. The results of
Figure 9. Calcium sulfite hemihydrate
platelets formed during "no
mixing well" test.
operating at this high chloride concen-
tration were as expected. The calcium
concentrations were higher than nor-
mal; an initial spike was noted in the oxi-
dation rate; significant dissolution of
solid phase calcium in the scrubber was
evident; and scaling conditions were
observed on the lower bed. As the
chloride concentration in scrubbing
slurry and the clarifier liquor equalized
at about SOmmole/liter (2800 ppm), the
solid oxidation fraction dropped to
below 0.1 and subsaturated conditions
were again measured in lower bed ef-
fluent liquor. The need for calcium
solids dissolution was eliminated as the
magnesium/chloride ratio increased m
the scrubber feed liquor.
As mentioned previously, the magne-
sium concentration was increased to
offset the addition of chloride. Since a
magnesium level of 125 mmole/liter
(3000 ppm) had maintained system
operability at the conclusion of Phase IV
and since 1 mole of magnesium electri-
cally balances 2 moles of chloride, the
magnesium concentration was increased
to about 160 mmole/liter to offset the
70 mmole/liter increase in chloride.
Since the magnesium sulfite and
carbonate salts are soluble, the liquid
phase alkalinity was essentially the
same in Phase V (after the initial spike in
chloride) as during Phase IV. The low
calcium dissolution rates in the scrubber
measured between August 25 and 30
reflect this.
Solid Solution Chemistry
A major objective of this program was
to'determine the basis for the operation
of the Paddy's Run system in a mode
which was subsaturated with respectto
gypsum. During the course of this
program, the solid phase sulfate/total
sulfur ratio was found to have a major
impact on the gypsum saturation level
in the system's reaction tank liquor. This
same result was also found in a
laboratory study previously conducted
by Radian1. In Figure 11, the field data
from the LG&E test program are
compared with the correlation of solid
phase sulfate/total sulfur versus gypsum
relative saturation derived from the
laboratory experiments. While there is
scatter due to (1) unsteady-state opera-
tion, and (2) uncertainties in the gypsum
relative saturations and analytical
results in Phase I, the results from the
field testing show that the reaction tank
liquor remained subsaturated when the
solid phase sulfate/total sulfur level
remained below about 0.16.
-------
250
Dissolved
Mg^andCL
ion concentration
(mrnole/liter)
Qissolved Ca++
ion concentration
(mmole/'liter)
Gypsum
relative
saturation
Sutfate
Total S
In solids
1.5
1.0
0.5
O
0.15
0.10
0.5
r Lower bed
Exit liquor
-\
O
O O
O
24 25 26 27 28 29 30 31
August 1977
Figure 10. Phase V system chemistry parameters.
Conclusions
• Carbide lime contains oxidation
inhibitors which allow the Paddy's
Run system to operate in a scale-
free mode. Without these oxidation
inhibitors, testing with commercial
lime resulted in gypsum scaling in
the scrubber.
• Magnesium addition markedly
improved SC>2 removal efficiency by
increasing the liquid phase alkalinity
(98 percent removal at —3000 ppm
Mg++ with low chloride levels).
• Addition of magnesium (—3000
ppm) enabled the Paddy's Run
commercial lime system to operate
without scaling by (1) reducing the
sulfite oxidation fraction, and (2]
reducing the amount of calcium
sulfite dissolution which occurred m
the scrubber.
• Extremely high sulfate magnesium
concentrations (—4000 ppm Mg1"*
with low chloride levels) resulted in
the formation of solid crystals with
poor settling/dewatering character-
istics.
• An increase in chloride tended to
increase sulfite oxidation and the
tendency to scale. Magnesium
addition can offset this effect.
• Lime systems can function with low
reaction tank residence times due
to the rapid dissolution of lime. The
size of the reaction tank and the
lime addition point location can
affect the crystal structure.
References
1. Jones, Benjamin F., Philip S. Lowell,
and Frank B. Meserole. Experimental
and Theoretical Studies of Solid
Solution Formation in Lime and
Limestone SO2 Scrubbers, Volumes
I and II, EPA-600/2-76-273a,b
(NTIS PB264953 and 264954)
October 1976.
2. Holcombe, L. J. and K. W Luke.
Characterization of Carbide Lime to
Identify Sulfite Oxidation Inhibitors,
EPA-600/7-78-176 (NTIS PB286646).
September 1978.
-------
CO
Q.
-------
u
I
o
5'
I
O
NJ
05
00
n,
rn
s
o
i > "o m •
fliii
en
3
s
------- |