U.S. Environmental Protection Agency Industrial Environmental Research EPA-600/7-77-074
Office of Research and Development Laboratory
Research Triangle Park. North Carolina 27711 Jllly 1977
LABORATORY STUDY OF
LIMESTONE REGENERATION IN
DUAL ALKALI SYSTEMS
Interagency
Energy-Environment
Research and Development
Program Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S.
Environmental Protection Agency, have been grouped into seven series.
These seven broad categories were established to facilitate further
development and application of environmental technology. Elimination
of traditional grouping was consciously planned to foster technology
transfer and a maximum interface in related fields. The seven series
are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from
the effort funded under the 17-agency Federal Energy/Environment
Research and Development Program. These studies relate to EPA's
mission to protect the public health and welfare from adverse effects
of pollutants associated with energy systems. The goal of the Program
is to assure the rapid development of domestic energy supplies in an
environmentally—compatible manner by providing the necessary
environmental data and control technology. Investigations include
analyses of the transport of energy-related pollutants and their health
and ecological effects; assessments of, and development of, control
technologies for energy systems; and integrated assessments of a wide
range of energy-related environmental issues.
REVIEW NOTICE
This report has been reviewed by the participating Federal
Agencies, and approved for publication. Approval does riot
signify that the contents necessarily reflect the views and
policies of the Government, nor does mention of trade names
or commercial products constitute endorsement or recommen-
dation for use.
This document is available to the public through the National Technical
Information Service, Springfield, Virginia 22161.
-------
EPA-600/7-77-074
July 1977
LABORATORY STUDY OF
LIMESTONE REGENERATION IN
DUAL ALKALI SYSTEMS
by
J.E. Oberholtzer. L.N. Davidson,
R.R. Lunt, and S.P. Spellenberg
Arthur D. Little, Inc.
20 Acorn Park
Cambridge, Massachusetts 02140
Contract No. 68-02-1332
Task No. 26
Program Element No. EHE624
EPA Project Officer: Norman Kaplan
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, N.C. 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D.C. 20460
-------
TABLE OF CONTENTS
Page
LIST OF TABLES iv
LIST OF FIGURES iv
I. SUMMARY 1
II. INTRODUCTION, PURPOSE AND SCOPE 2
III. EXPERIMENTAL APPARATUS, APPROACH, AND CALCULATIONS 4
A. DESIGN OF THE APPARATUS 4
B. EXPERIMENTAL APPROACH 7
C. EXPERIMENTAL CALCULATIONS 9
IV. RESULTS AND DISCUSSION 11
A. FACTORS AFFECTING LIMESTONE UTILIZATION 11
B. DEWATERING PROPERTIES OF PRODUCT SOLIDS 15
C. SULFATE PRECIPITATION 20
D. FATE OF MAGNESIUM IN LIMESTONE REGENERATION
EXPERIMENTS 23
V. CONCLUSIONS 31
REFERENCES 32
iii
-------
LIST OF FIGURES
FIGURE NO. Page
1 LIMESTONE DUAL ALKALI APPARATUS 5
2 SETTLING CURVES FOR OPEN-LOOP RUNS 16
3 SETTLING CURVES FOR EXTENDED RUNS 108-110 18
4 SETTLING CURVES FOR RUNS 111-114 19
5 RUN 111 SETTLING CURVES - SAMPLED AT DIFFERENT TIMES 21
6 VARIATION IN SETTLED DENSITY WITH ACIDITY/SULFATE
RATIO IN SCRUBBER BLEED 22
7 SULFATE PRECIPITATION IN THIS LABORATORY WORK COMPARED
WITH THAT OBSERVED IN EARLIER STUDIES (Ref. 1) 24
8 CHANGES IN SOLUBLE MAGNESIUM OBSERVED DURING EXTENDED
RUNS 29
LIST OF TABLES
TABLE NO.
I SUMMARY OF LIMESTONE BASED DUAL ALKALI REGENERATION
EXPERIMENTAL RESULTS 12
II COMPOSITION OF SOLIDS PRODUCED IN LIMESTONE DUAL ALKALI
REGENERATION STUDIES 13
III BEHAVIOR OF SOLUBLE MAGNESIUM IN LIMESTONE DUAL ALKALI
REGENERATION STUDIES 25
iv
-------
I. SUMMARY
A series of open- and closed-loop laboratory bench scale experi-
ments have been carried out to study a number of parameters which affect
the reaction of limestone with dual alkali flue gas desulfurization (FGD)
systems process liquors. During the program, which was carried out
from July through November, 1976, acceptable operation in terms of
limestone utilization and dewatering properties of the product solids
was demonstrated for several sets of operating conditions.
By employing a system with four reactors in series, with a total
residence time of two hours, and by using Fredonia limestone (a very
finely ground material, low in magnesium) good operation was achieved
over a reasonable range of liquor compositions. In closed-loop runs at
50°C, utilization of limestone ranged from 78 to 92% with total con-
centrations of sulfite plus bisulfite (TOS) in the scrubber bleed
ranging from 0.56 M to about 1.5 M and the concentrations of sodium
sulfate ranging from about 0.7 M to 1.2 M. No special steps were
required to control the soluble magnesium level in the liquor or
reduce its effects on the regeneration reaction.
Precipitation of calcium sulfate in solid solution with the
calcium sulfite took place according to the same relationship as
determined in prior work for regeneration using lime.
The fact that soluble magnesium levels did not rise in solutions
which were unsaturated with respect to magnesium sulfite, the least
soluble magnesium solid phase, suggested either that magnesium may have
been coprecipitating from those solutions in a manner similar to the
way in which calcium sulfate coprecipitates with calcium sulfite from
solutions unsaturated in gypsum; or that not all the magnesium
present in the limestone dissolved.
-------
II. INTRODUCTION, PURPOSE AND SCOPE
Limestone is used in many of the flue gas desulfurization (FGD)
systems which employ direct slurry scrubbing. In dual alkali FGD
systems, the substitution of limestone for lime, which is generally
used, would be economically attractive, since limestone is considerably
less expensive than lime. However, because lime is more reactive, it
is easier to achieve dependable system operation with lime.
The simple neutralization reaction for limestone with a scrubber
liquor containing sodium sulfite/bisulfite is:
CaC03(s) + 2NaHS03 CaS03(s) + Na2S03 + H20 + C02(g)
The sulfite ion is not sufficiently acidic to react with limestone,
thus bisulfite is the primary reactive species. Because, in comparison
to lime, limestone is very insoluble and only moderately basic toward
bisulfite, regeneration of dual alkali liquors proceeds slowly and the
solids produced can be difficult to dewater. In earlier work (1)
several important parameters which influence the reaction with lime-
stone were identified. The more obvious include: the concentration
of total oxidizable sulfur (TOS) in solution; ionic strength of the
liquor, related, in part, to sodium sulfate concentration; reaction
temperature; and reactor system configuration. It was also observed
that the presence of dissolved magnesium in the process liquor (mag-
nesium is a common impurity in limestone) can lower reaction rates
and produce product solids with poor dewatering properties. Traces
of iron had a similar effect.
Although the earlier laboratory and pilot plant work on limestone
regeneration uncovered important factors influencing the regeneration
reaction, it was not possible to develop a set of process parameters
and reactor conditions which were consistent with both good limestone
utilization and the generation of product solids with good dewatering
characteristics.
The objective for the current program, then, was to develop one
or more sets of acceptable operating conditions by studying the
-------
limestone regeneration reaction in a bench scale laboratory apparatus
which could be controlled carefully. Studies were to be focused on
operation with "concentrated" sodium sulfite/bisulfite/sulfate liquors.
Concentrated dual alkali systems produce liquors saturated with
respect to calcium sulfite upon regeneration and generally contain
active Na concentrations above 0.15 M. Unsuccessful attempts to
regenerate "dilute" (saturated with respect to calcium sulfate upon
regeneration) solutions with limestone have been reported previously
(2).
The plan for achieving the program goal included the following
steps:
• Design and construct a bench scale apparatus for open-
and closed-loop studies;
• Carry out open loop experiments to test the apparatus and
establish baseline operating conditions;
• Close the loop and conduct experiments to study the effects
of temperature, TOS, sodium sulfate, and soluble magnesium.
In order to maximize the chances of finding one or more sets of
acceptable operating conditions, the experimental matrix was designed
to proceed from conditions where success was most likely — high tem-
peratures, low magnesium — to those where acceptable operation was
more doubtful — higher magnesium and sodium sulfate and lower temper-
atures.
-------
III. EXPERIMENTAL APPARATUS. APPROACH, AND CALCULATIONS
A. DESIGN OF THE APPARATUS
A schematic flow sheet of the apparatus designed for closed-loop
operation with limestone is presented in Figure 1. The operation
consisted of three basic steps:
• contacting acidic sodium sulfite/bisulfite/sulfate scrubber
bleed with limestone in a continuous flow reactor system
consisting of four well-mixed tanks in series;
• separating the product solids from effluent in a settler;
• recycling the separated liquor through an absorber in
order to add sulfur dioxide; and back again to the absorbent
regeneration reactor system.
The four reactors in Figure 1 were constructed from glass beakers
with volumes of 0.5, 0.5, 1.0 and 1.5 liters, respectively. Each had
a side-arm blown on the wall and was equipped with baffles, a mixer,
and an external heater. The reactor series utilized gravity flow
with the overflow height of each reactor adjusted as desired. The pH
of the fourth reactor was continuously monitored and recorded if neces-
sary. Each reactor was equipped with a polyethylene coated propeller
with three pitched blades having a diameter of 4.5 cm. Each reactor
had four baffles; each baffle extended inward from the reactor wall a
distance 1/10 the diameter of the reactor. The stirring speeds for the
impellers were calculated to impart approximately the same power per
unit volume for all reactors. Shaft speeds were 350, 350, 500, and 720
RPM from first to fourth reactor, respectively.
Additional components of the system included a settler (with feed
well) which was designed to accumulate and retain product solids for
about 16 hours of continuous operation. A surge tank received the
liquid overflow from the settler. As solids deposited in the settler,
liquid was displaced into the surge tank. Continuous underflow filter-
ing was not employed; solids were removed from the settler periodically.
The amount of liquid removed with the thickened slurry was estimated
and, based on analysis of an aliquot of the liquid, an equal volume
-------
Heater
_ Liquid Line
____ _Gas Line
TC = Temp. Controller
LS = Level Sensor
FC = Flow Controller
pH = pH Sensor
PC = pH Controller
I CW
(Vent
..I
Caustic
Trap
FIGURE 1 LIMESTONE DUAL ALKALI APPARATUS
Surge
-------
having the same composition was synthesized and put back into the
settler.
Regenerated liquor was pumped from the surge tank to the S02
absorber by a Masterflex tubing pump. The flow to the absorber was
controlled on the liquid level in the absorber. In the absorber, the
liquor was acidified by bubbling a mixture of S02 in nitrogen through
it. Since the absorber was operated as a relatively inefficient single-
stage device (simply to acidify the liquor to a desired pH), a consi-
derable amount of SC^ was not removed from the gas. The exit gas was
therefore scrubbed in a caustic trap prior to venting. The scrubber
bleed was pumped to the regeneration reaction system at a constant,
predetermined flow rate using a Masterflex tubing pump. The pH of the
scrubber bleed stream was monitored, and the on/off output of the pH
meter/controller was used to trim the amount of 862 which was mixed
with nitrogen and fed to the absorber.
All reactors were equipped with heaters, and a preheater was also
installed immediately before the first reactor. The heaters were
equipped with on/off temperature controllers. The preheater was parti-
cularly necessary for operation at high temperatures. Silicone rubber
tubing was used wherever possible because of its stability at elevated
temperatures.
A major problem was encountered in obtaining a limestone feeder
to meet the required flow rates. To completely neutralize 1 Jl/h of
simulated scrubber bleed containing 1.45 M bisulfite, a dry limestone
feed of about 1.2 g/m was required. It was not desirable to add the
limestone as a slurry because of the tight water balance (little
evaporation in the absorber); dry feeding the Fredonia limestone with
standard industrial or laboratory equipment proved difficult at the
desired flow rate. Packing of the limestone prevented reproducible
feeding at rates below about 5 g/m, using either a screw feeder or a
rotary plate chemical feeder. Since a suitable off-the-shelf feeder
could not be obtained, a feeder was designed and built at Arthur D.
Little, Inc. It was a novel design employing the properties of a
syringe and plunger to dispense a limestone paste (75% dry solids).
The barrel was a plexiglass tube 5 cm in diameter and 50 cm long. The
-------
plunger was actually a piston, pushed down the barrel by a threaded
rod driven by a gear and motor arrangement. Limestone was extruded
from a constricted opening at the end of the syringe and was fed
directly into the reactor system. Constant feed rates of 0.5-3.0
g/m could be obtained with this syringe-type feeder. The total
capacity of the syringe was apprxomately 1.2 kg. Thus, depending on
the rate of feed, the syringe had the capability to operate unattended
for between 7 and 40 hours.
B. EXPERIMENTAL APPROACH
After assembly of the apparatus was completed, four open-loop
runs were conducted to test the reactor system and to study the effects
of temperature and magnesium concentration on the extent of reaction
and the properties of the solids produced. The first three open-loop
experiments were carried out in a 3-reactor system with reactor time
constants of 15, 15, and 30 minutes for successive reactors. A fourth
reactor with a time constant of 1 hour was added to the end of the
3-reactor train for the remainder of the open- and closed-loop experi-
ments.
In conjunction with the open-loop runs, measurements of magnesium
sulfite solubility were performed to provide estimates, as a function
of temperature, of the levels to which magnesium might build in a
closed-loop system before the precipitation of magnesium sulfite would
occur limiting further buildup.
Initial attempts at unattended closed-loop operation failed due
to both mechanical and process problems. One significant problem was
the blockage of the reactor overflow tubing by large, thin pieces of
scale that appeared to have been formed on the walls of the first
reactor and then chipped off. After two successive problems with
plugging, the system was attended constantly and seven successful
closed-loop runs, ranging in length from eleven to twenty-nine hours,
were carried out. The principal variables studied in the closed-loop
runs were liquor temperature (50 and 90°C), TOS concentration (0.55-
1.6 M), soluble magnesium concentration (1000 and 2000 ppm), and
sulfate concentration (0.65-1.2 M).
-------
In all of the experiments, Fredonia limestone obtained from the
EPA/TVA Shawnee test facility was used as the reagent. The Fredonia
limestone was chosen because it had been characterized, and use in
prior work indicated that it was more reactive than some other forms
of limestone tested (1). Analyses indicated that the limestone con-
tained by weight 96.6% Ca (as CaC03) and 1.1% Mg (as MgC03). Total
available alkalinity (expressed as CaCO-) was 97.2%.
The limestone feeder was initially set so that limestone was fed
at 80-90% of the stoichiometric amount required to completely neutralize
the bisulfite present in the acidic scrubber bleed where feed stoichio-
metry is defined as follows:
„ ... . rmoles CaC03 fed ., n n_„
% stoichiometry = [—r •; _ ] x 100%
0.5 moles HSO, fed
Adjustments were made to the limestone feed rate, if necessary, to hold
the pH of the reactor effluent slurry within the range of 6.2-6.5.
Those adjustments generally reduced the limestone feed shoichiometry
to 70-80%.
In all of the closed-loop experiments, the scrubber bleed feed
rate to the first reactor was kept constant at 15 ml/m. At that feed
rate, the entire system, including absorber, reactors, settler, and
surge tank, had a residence time of about 13 hours.
Since earlier work suggested that poor dewatering of solids was
associated with slow regeneration reaction rates, most of the experi-
ments were carried out at relatively high concentrations of TOS in an
attempt to increase reaction rates. The acidity of the scrubber bleed
fed to the reactor system, H+ (which is predominantly bisulfite, HSO-j")
was controlled so that the ratio HS03~/S03= was generally in the range
of 4-9. Sulfate concentration was varied from about 0.7 M to 1.16 M,
producing TOS/S04= ratios ranging from 0.73 to 2.47.
Because increasing temperature should increase reaction rates, the
effects of temperatures greater than 50°C were explored in several
runs. However, because of the difficulties and costs associated with
the operation of an actual process at elevated temperatures, most of
the runs were carried out at 50°C, which is typical of scrubber bleed
streams.
-------
C. EXPERIMENTAL CALCULATIONS
In addition to limestone feed stoichiometry, discussed above, the
values of two other parameters — limestone utilization and sulfate
precipitation — were important in characterizing the behavior of the
regeneration reaction. Limestone utilization was calculated from the
compositions of the product solids as follows:
„ ,, ji.. • r moles TOS 4- moles sulfate - 0.5 moles Na n -,nnv
Percent Utilization = [-.—; —,• w.—JT m—T3~^ 7—i—«—. - N ] x 100%
(moles Ca + Mg)(Avail. Alkalinity/mole Mg + Ca)J
It should be pointed out that the subtraction of one half of the measured
amount of sodium from the numerator to correct for the soluble sodium/
sulfur salts entrained in the solids is not rigorously correct for solids
produced in reactions with limestone, since some of the TOS in the efflu-
ent exists as HS03~. However, this approximation is adequate since in
most cases utilizations were high and sodium levels in the cake were low.
Available alkalinity per mole of alkaline earth (Ca + Mg) was
determined by dissolving a sample of the limestone in excess acid,
boiling, and back-titrating with base to a pH of about 4. The available
alkalinity was expressed as CaC03 and divided by the total moles of Ca
and Mg in the same sample.
Magnesium is included in the denominator of the utilization equation
because it is assumed that the magnesium in the limestone is a potentially
useful regeneration reagent. As will be discussed later, it is not clear
whether the magnesium in the Fredonia limestone actually took part in
many of the closed-loop runs reported here. The extent to which magnes-
ium participates in the reaction under a given set of experimental con-
ditions is clearly a factor that needs to be determined, particularly
for limestones which contain larger amounts of magnesium than the
Fredonia limestone.
Further, since the amount of total sulfur in the solids which
appears in the numerator is corrected only for sodium/sulfur salts,
the precipitation of soluble magnesium present in the scrubber bleed
as magnesium sulfite during the course of the neutralization reaction
would produce an apparent increase in utilization. This phenomenon
would produce erroneous calculated utilizations only in the case of
systems which are not at steady-state, e.g., open-loop runs where a
-------
significant portion of the soluble magnesium present in the scrubber
bleed is precipitated out during the single pass through the system.
Sulfate precipitation, a measure of the amount of system oxida-
tion that can be accommodated, was calculated as the ratio of CaSO^/
CaS02 in the solids taken from the reactor effluent according to the
relation:
c so _ ,Total moles sulfate in solids - moles soluble sulfate.. ,„„„
4 ~ lTotal moles sulfite in solids - moles soluble TOS J X
where:
-i T 1^1 i r rmoles Na in solids ^ , „. = , ,.
moles soluble sulfate = [ .. 4. . =-. ] x cone, of SO- in liquor
cone. Na+ in liquor 4
i -i LI m«o rmoles Na in solids , ,- „,_., . ,.
moles soluble TOS = [ ....+ . — ] x cone, of TOS in liquor
cone, of Na^ in liquor"
The total moles of sulfate in the washed and dried solids was corrected
for occluded soluble sodium sulfate. The latter was determined by multi-
plying the concentration of sulfate in the process liquor by the ratio
of the amount of sodium in the solids to the concentration of sodium in
the liquor. Similarly, the solid sulfite was corrected for occluded
sodium salts. It should be recognized, however, that no correction was
made for magnesium salts present in the solids. Under steady state
operating conditions, that treatment is completely appropriate, since
sulfur (IV) and sulfur (VI) can be removed both as magnesium salts and
as calcium salts.
10
-------
IV. RESULTS AND DISCUSSION
An overview of the conditions under which each of the four open-
loop and seven closed-loop runs were carried out along with a summary
of the salient results obtained from each are presented in Table I.
Included in the table are the operating temperatures, limestone feed
stoichiometries, and the compositions of the scrubber bleed stream fed
to the regeneration reactor system for each run. For the closed-loop
runs, the scrubber bleed compositions are those measured after steady-
state had been achieved; in the open-loop runs the simulated scrubber
bleed compositions were invariant throughout each run.
The performance of the reactor system in each experiment was
characterized by measurements of pH in the reactor effluent and lime-
stone utilization based on the composition of the solids discharged
from the final reactor. The dewatering properties of the solids
produced were characterized by performing settling tests in a 100-ml
graduated cylinder, determining the meniscus position after 30 minutes
of settling, and calculating the density of the settled solids. In a
number of cases, the percent solids in the wet cake produced by vacuum
filtration on a laboratory Buchner funnel was determined.
The results of analyses of the product solids which were used to
calculate utilization and sulfate precipitation are shown in Table II.
Some of the factors responsible for the observed variations in limestone
utilization are discussed in more detail in the sections which follow.
A. FACTORS AFFECTING LIMESTONE UTILIZATION
Earlier studies of the dual alkali regeneration reaction have
shown that the utilization of limestone is dependent upon a number of
variables including soluble magnesium concentration, reaction tempera-
ture, reactor holdup time, limestone feed stoichiometry, sodium sul-
fate concentration, and bisulfite acidity. It is not possible to
clearly discern the independent effect of each of these variables on
the utilizations obtained in this program because in many cases more
than one independent variable was changed from one experiment to the
next. Furthermore, since the primary aim of this work was to determine
11
-------
TABLE I
SUMMARY OF LIMESTONE BASED DUAL
Scrubber
Steady State
Run
100
101
102
^103
108
109
110
111
112
' 113
114
Type1 :
3R-OL
3R-OL
3R-OL
4R-OL
4R-CL
4R-CL
4R-CL
4R-CL
4R-CL
4R-CL
4R-CL
Cernp ("C) TOS (M)
45
50
90
70
90
50
50
50
50
50
50
1.58
1.61
1.60
1.62
1.43
1.44
1.49
0.56
1.02
1.03
1.33
. H+(M)
1.42
1.45
1.42
1.47
1.26
1.37
1.21
0.49
0.83
0.84
1.20
Ms^ppm)
300
4450
4450
7830
1270
1030
1940
2020
2010
2020
1900
ALKALI REGENERATION EXPERIMENTAL RESULTS
Bleed
Composition
SOU=(M)
0.64
0.69
0.71
0.73
0.80
0.70
0.67
0.77
0.77
0.99
1.16
H+/SOU°
2.2
2.1
2.0
2.0
1.6
2.0
1.8
0.64
1.1
0.85
1.0
Stoich.(%)2
90
86
84
78
65-75
70-75
72
68-75
81
78
78
Reactor
Effluent
6.6
6.0
6.1
6.0
6.2
6.5
6.5
6.4
6.5
6.5
6.5
Limestone
Utiliza-
tion (%)3
88
41
81
79
102
92
90
84
82
81
78
Settling Properties
Filter
Meniscus Cake
(ml) ^ Density5 % Solids
15
23
20
66
26
15
18
32
21
24
30
ND
38
40
12
36
ND(>30)
34
8
23
18
26
ND
ND
ND
ND
42
ND
75
35
56
59
57
Sulfate6 Duration of
Precipitation Run(hrs)
2.7
8.1
6.6
—
5.7
3.6
4.7
12.5
6.6
7.6
6.2
4.1
4.0
4.3
8.8
28.6
28.2
28.8
14.0
11.0
11.0
14.4
^Number of reactors — open/closed loop.
^Percent of amount required to neutralize acidity (see text).
3Calculated from solids composition (see text).
''Meniscus position in 100-ml graduated cylinder after settling for 30 minutes.
Height of dry solids (g) per 100 ml of settled slurry.
6[Solid Sulfate]/[Sol id Sulfite] x 100% (see text).
-------
TABLE II
COMPOSITION OF SOLIDS PRODUCED IN LIMESTONE DUAL ALKALI REGENERATION STUDIES
Run
100
101
102
103
108 l
109
110
111
112
113
114
Ca++
(mmoles/g)
7.37
8.13
6.87
6.09
7.34
7.37
7.50
7.52
7.55
7.44
7.44
TOS
(mmoles/g)
6.48
3.39
5.99
4.81
7.38
6.75
6.65
5.74
5.99
5.70
5.66
S04
(mmoles/g)
0.264
0.338
0.574
0.786
0.462
0.326
0.380
0.830
0.472
0.514
0.508
Mg"""
(mmoles/g)
0.053
0.60
0.83
0.772
0.230
0.104
0.143
0.095
0.124
0.103
0.117
Na+
(mmoles/g)
0.398
0.271
0.680
ND2
0.180
0.35
0.31
0.29
0.26
0.26
0.51
1 Sample taken from settler after last reactor.
2 Not determined.
13
-------
a range of conditions under which acceptable operation could be
realized, operating conditions were changed in relatively small incre-
ments from one experiment to the next.
1. Effect of Soluble Magnesium
The dramatic impact that dissolved magnesium can have on limestone
utilization can be seen by comparing Runs 100 and 101, both open-loop
runs. With all other conditions kept relatively constant, raising the
magnesium level from 300 to 4,450 ppm caused the utilization to
decrease from 88 percent to 41 percent.
2. Effects of Temperature and Reactor Holdup Time
The improved utilization which can be obtained by operating at
elevated temperatures can be seen by comparing the 79 percent and
81 percent utilizations obtained in open-loop Runs 102 (70°C) and 103
(90°C) with the poor utilization of 41 percent in Run 101 (508C).
The fact that the utilization in Run 103 was about the same as
that in 102, even though the temperature was somewhat lower and the
magnesium concentration somewhat higher, was due to the fact that a
fourth reactor with a residence time of 1 hour was added to the series
of three reactors which had been used in Runs 100-102. Although the
solids in the effluent from the third reactor in Run 103 were not
analyzed, and consequently a 3-reactor utilization based on solids
composition could not be calculated, the TOS and acidity in the effluent
from the third reactor during Run 103 indicated that a utilization of
about 65% would have been obtained if only three reactors had been
used.
In the closed-loop experiments, the effect of operation at 90°C
on utilization can be seen clearly by comparing Runs 108 and 109 made
at approximately 1000 ppm magnesium. In Run 108 at 90°C, the utiliza-
tion was 102%, while in Run 109 at 50°C utilization was 92%.
3. Effect of Limestone Feed Stoichiometry
Limestone feed Stoichiometry is another factor affecting, utiliza-
tion. As the limestone feed Stoichiometry is reduced, utilization tends
to increase. The increased utilization, however, is realized at the
expense of a decrease in the extent to which the liquor is regenerated.
In the open-loop runs, reducing the limestone feed Stoichiometry
14
-------
from 90% in Run 100 to 84% and 78%, respectively, in Runs 102 and 103,
was a factor which caused the utilization to drop only from 88 percent
to about 80 percent when magnesium concentration was increased
dramatically.
The effect of feed stoichiometry on utilization in the closed-
loop experiments is not clearcut, because both feed stoichiometry and
scrubber bleed composition were being changed at the same time in many
cases. However, comparison of the results of Runs 109-111 with Runs
112-114 show that generally higher utilizations were obtained in the
former group (84-92% vs. 78-82%) in which the feed stoichiometry was
somewhat lower (68-75% vs. 78-81%).
4. Effects of Sodium Sulfate Concentration and Bisulfite Acidity
The concentrations of sulfate and bisulfite also importantly affect
the limestone regeneration reaction rate and therefore utilization.
Increasing the sulfate concentration raises the ionic strength which
retards the reaction rate; on the other hand, the increased acidities
which accompany elevated bisulfite concentrations tend to drive the
reaction faster. Increasing TOS from 0.56 M to 1.49 M (Runs 111
and 110, respectively) resulted in an increase in utilization from
k
84% to 98% at about the same feed stoichiometry. The antagonistic,
or counterbalancing, effects of TOS acidity and sulfate concentration
are clearly indicated in Runs 113 arid 114 in which there was little
change in utilization (81% vs. 78%) when both TOS (and acidity) and
sulfate were comparably increased. This suggests that the acidity-to-
sulfate ratio in the reactor feed may be an important parameter in
predicting utilization.
B. DEWATERING PROPERTIES OF PRODUCT SOLIDS
With the exception of Run 103 (7830 ppm magnesium) in which very
poor settling behavior was observed, the solids in the product slurries
from all of the other runs settled reasonably well. After 30 minutes
of settling in a 100-ml graduated cylinder, the slurry had generally
settled to 15-30% of its original height.
Settling curves for samples of the product slurry taken at the
end of each of the four open-loop runs are shown in Figure 2. The
solids from Run 100, the run with the lowest magnesium, and Run 102,
15
-------
100
80
60
C
o
o
Q.
C
-------
the run at 90°C, exhibited the best settling behavior. Not only were
the final settled volumes the lowest, but the initial settling rates
were also the most rapid — the solids had settled to essentially their
final settled volumes in fifteen minutes or less. The solids from Runs
101 and 103, the high magnesium runs at 50°C and 70°C, settled more
poorly both in terms of initial settling rate and settled volume after
30 minutes. The settling behavior of the solids from the three-reactor
system used in Run 101 was reasonably similar to that of the effluent
from the third reactor of the four-reactor system used in Run 103.
However, in Run 103, passage of the reaction slurry through the fourth
reactor which had been added to the system resulted in a significant
deterioration in the settling properties of the solids.
The settling curves for the third reactor effluent from Runs 101
and 103, shown in Figure 2, exhibit breaks due to disruption of the
settling process by rising bubbles of CC>2 which were produced as a
result of a continuing reaction in the settled mass. Such bubbling was
observed in settling tests in previous work (1) where low utilizations
were achieved. In Run 101 utilization was only 41%, and in the third
reactor in Run 103 utilization is estimated to have been approximately
65%. Off-gassing was not observed in the final effluent from Run 103
which had reacted for an additional hour in the fourth reactor.
Settling curves for the three extended closed-loop runs (108-110)
and the four shorter closed-loop runs (111-114) are shown in Figures
3 and 4, respectively. All indicate good settling behavior. No signi-
ficant effervescence due to continued reaction was observed in samples
taken from any of the closed-loop experiments (four reactors were used
for all). In a few instances, a number of small bubbles were detected
within the mass of settled solids; however, they did not grow large
enough to rise and disrupt the settling process.
In earlier work (1) the settling properties of the solids were
sometimes observed to deteriorate as a run progressed. Initial
settling rates frequently decreased with time and the form of the
settling curves changed from being concave to becoming almost straight
lines. Care was taken in this program to monitor the settling properties
periodically so that unstable settling behavior would be detected. For
17
-------
100
c
o
(J
—
'E
80
•£• 60
40
20
108
110
I
I
10
15 20
Time (min)
25
30
35
FIGURE 3 SETTLING CURVES FOR EXTENDED RUNS 108-110
-------
100
Time (min)
FIGURE 4 SETTLING CURVES FOR RUNS 111-114
-------
most of the closed-loop runs, the settling behavior was quite stable
throughout the run. The greatest change in settling was observed
throughout the course of Run 111. Settling curves for four samples
taken during and at the end of the run are shown in Figure 5. Although
settling seemed to deteriorate during the first ten hours of the experi-
ment, the last two samples indicate that settling behavior had stabilized
before the experiment was terminated.
In the complete set of experiments which were carried out in this
program, a range of TOS concentrations was studied at more or less con-
stant limestone feed stoichiometry. Thus, the concentrations of solids
in the final product slurries varied. Consequently, the final settled
volumes alone are not the best indicator of the dewaterability of the
product slurries. The settled densities included in Table I take into
account the variation in slurry percent solids. They ranged from eight
grams of dry solids per 100 ml of settled slurry to as high as 40 g/
100 ml. The settled densities observed in all of the open- and closed-
loop experiments are plotted as a function of the ratio HS03~/SC>4= in
the feed liquor in Figure 6. The general increase in settled density
with the ratio HS03~/S04~ supports the thesis that better settling
behavior is exhibited by solids produced in faster reactions.
It is not clear why the settled density of the solids produced in
Run 103 was so low, even though the HSOg'/SO,3 ratio was high and good
limestone utilization was achieved. However, solution analyses indicated
that in Run 103 only about 54% of the TOS was precipitated in the first
two reactors. In all of the other runs, 70-75% of the TOS precipitation
took place in the first two reactors. The slower reaction rates in
the first reactors could have caused the solids to settle poorly. An
explanation of this sort is consistent with the observation made during
the discussion of settling curves in Figure 2 that in Run 103 the
effluent slurry from the fourth reactor settled more poorly than that
from the third reactor.
C. SULFATE PRECIPITATION
The amount of sultate that is precipiated as a calcium salt (as
opposed to being occluded as a soluble salt) is of interest since it
provides a measure of the amount of oxidation that can be accommodated
20
-------
100
80
- 60
o
3
U
c
o>
40
20
10 & 14 hours
7 hours
4 hours
10
15
20
25
30
Time (min)
FIGURES
RUN 111 SETTLING CURVES - SAMPLED AT DIFFERENT TIMES
-------
50
40
o
o
;o
"o
30
NJ
r-o
in
•o
20
c
0)
Q
10
0.5
1.0
1.5
2.0
2.5
[H+]/[S04=
FIGURE 6 VARIATION IN SETTLED DENSITY WITH ACIDITY/SULFATE
RATIO IN SCRUBBER BLEED
-------
without purging soluble salts. In earlier laboratory and pilot plant
studies of the use of lime and limestone to regenerate dual alkali
liquors (1) it was found that a good correlation existed between the
ratio CaSO^/CaSO-j in the product solids and the ratio SQf/SQf in the
regenerated process liquor. The values for the CaSO^/CaSC^ ratio
obtained in these bench scale limestone experiments were calculated
and are included in Table I. The values are also plotted as a function
of S0,~/S0_ in Figure 7. Included on that plot are the values
determined previously in laboratory and pilot plant studies using lime
and limestone. The agreement between the limestone values determined
in this work and the earlier values is surprisingly good.
D. FATE OF MAGNESIUM IN LIMESTONE REGENERATION EXPERIMENTS
Because of the problem which elevated levels of soluble magnesium
can cause when limestone is used for regeneration, an important concern
is the final steady state magnesium concentration in the process liquor.
At steady state the input of magnesium from dissolution of limestone
(and possibly from flyash, if it is collected in the scrubbing system)
is in equlibrium with its removal with the product solids. In the pH
range of 5-7 where most limestone dual alkali systems would operate,
magnesium sulfite would be the least soluble magnesium salt; so its
precipitation would be expected to limit the soluble magnesium level.
Magnesium hydroxide is less soluble than magnesium sulfite, but it
would become the limiting solid phase only at higher pH values.
The changes in soluble magnesium concentration observed across
the regeneration reactor system in each of the experimental runs are
shown in Table III. The initial and final magnesium concentrations
in the feed are the results of analyses of samples of scrubber bleed
entering the reactor system at the beginning and end of each run. Since
the open-loop experiments were fed from a single tank of simulated
scrubber bleed with no recycle, the initial and final values are the
same. The final effluent concentration is that measured in a sample of
reactor effluent taken at the end of a run.
In Run 100, the relatively low, 300 ppm concentration in the feed
rose to 390 ppm upon passage thorugh the reactor system. However,
based on the amount of limestone which was fed and its magnesium content
23
-------
0.25
0.20
o
G
-------
TABLE III
ho
Ul
BEHAVIOR OF SOLUBLE MAGNESIUM IN
LIMESTONE
DUAL ALKALI REGENERATION STUDIES
Scrubber Bleed
Steady State Composition
Magnesium Concentration
(ppm)
Initial Final
Run
100
101
102
103
108
109
110
111
112
113
114
Type1
3R-OL
3R-OL
3R-OL
4R-OL
4R-CL
4R-CL
4R-CL
4R-CL
4R-CL
4R-CL
4R-CL
Temp(°C)
45
50
90
70
90
50
50
50
50
50
50
TOS (M)
1.58
1.61
1.60
1.62
1.43
1.44
1.49
0.56
1.02
1.03
1.33
H+(M)
1.42
1.45
1.42
1.47
1.26
1.37
1.21
0.49
0.83
0.84
1.20
SOU (M)
0.64
0.69
0.71
0.73
0.80
0.70
0.67
0.77
0.77
0.99
1.16
Reactor Reactor
Feed Feed
300
4450
4450
7830
2150
1035
2200
2050
2040
2050
1975
300
4450
4450
7830
1270
1030
1940
2020
2010
2020
1900
Final
Reactor
Effluent
390
4180
1900
6040
960
1030
1890
2040
1990
2010
1900
In Final Reactor Effluent
Mg++(M)
0.016
0.172
0.078
0.248
0.039
0.042
0.078
0.084
0.082
0.083
0.078
SO., (M)
0.64
0.56
0.42
0.42
0.63
0.55
0.64
0.19
0.45
0.45
0.59
MS-H- x so,=
0.010
0.10
0.033
0.10
0.025
0.023
(0.050)
0.016
0.037
0.037
0.046
1 Number of reactors — open/closed loop
-------
(0.13 mmoles/g), the observed increase of 90 ppm could be accounted
for by dissolution of only about 30% of the magnesium which entered the
reactor with the limestone. Analyses of the solids indicated that about
50% of the magnesium dissolved. The fact that a substantial amount of
magnesium remained in the Run 100 product solids (Table II) indicated
that either all of the magnesium entering the system did not dissolve,
or if it did, a substantial amount reprecipitated.
Run 101 was carried out with 4,450 ppm of magnesium in the feed
liquor. That magnesium level was chosen on the basis of a test in
which product solution from Run 100 was saturated with respect to mag-
nesium sulfite at room temperature. Although the limestone utilizaton
achieved in Run 101 was quite poor, a slight decrease in soluble mag-
nesium across the reactor system was observed.
When the preceding experiment was repeated at 90°C as Run 102,
not only was good utilization achieved, but a decrease in soluble
magnesium concentration of greater than 50% was also observed. The
relatively large amounts of magnesium found in the Run 102 product
solids (0.83 mmoles/g) is consistent with the decrease. (A magnesium
content of about 1.0 mmole/g in the solids would correspond to the
observed change in soluble magnesium concentration). Experiment 102
was encouraging not only because good utilization and settling were
achieved, but also because it demonstrated that under those conditions
the soluble magnesium level would not rise above a tolerable level.
A comparison of the magnesium solubilities observed in Runs 101
and 102 (values of the product [Kg**] x [SC>3=], the apparent solubility
product, K ', are included in Table III) suggest an inverse relation-
ship between the solubility of magnesium sulfite and temperature. That
observation is consistent with data reported in Gmelin (3) which indi-
cate that the solubility of magnesium sulfite goes through a maximum
value at about 40°C and then decreases to about 50% of the maximum
value at 90°C. A change in the hydration state of the stable solid
phase from the hexahydrate to the trihydrate in the vicinity of 40°C
is the reason for the maximum in the solubility/temperature relation-
ship.
26
-------
The equilibrium solubilities as a function of temperature for
magnesium sulfite in solutions with ionic strengths representative of
those which would be encoutnered in limestone dual alkali systems were
determined by adding an excess of MgSC^-Bl^O (which had been synthesized
in the laboratory) to synthetic solutions and analyzing the final
concentrations of soluble magnesium and sulfite. With sulfite concen-
trations in the range of 0.6-0.9 M and sulfate at 0.6 M, the following
values for the apparent solubility product, K ', were determined at
sp
the indicated temperatures:
Temperature (°C) [Mg4^] x [S0i=]
25 0.098
50 0.32
70 0.24
90 0.034
The result of the equilibrium solubility measurement at 90°C is in
excellent agreement with the solubility observed in Run 102. However,
the equilibrium K ' measured at 50°C was about three times the value
sp
observed in Run 101; and the calculated K ' for Run 103 at 70°C was
sp
the same as that for Run 101 (0.10) and considerably lower than the
0.24 measured in the equilibrium solubility experiment at 70°C. The
seemingly low values found in Runs 101 and 103 could have been produced
if the samples had cooled while the solids were being separated by
vacuum filtration*. The values which were observed were very close to
the 0.098 value measured at 25°C.
The first closed-loop run, Run 108, was carried out at 90°C with
a magnesium level of 2,150 ppm. That magnesium concentration, slightly
above the solubility limit for the reactor effluent based on the results
of Run 102, was chosen so that the system would have a reasonable
chance of reaching equilibrium by the end of the run. Interestingly,
after nearly 29 hours of operation at 90°C, the magnesium level in the
system had fallen to 1,270 ppm; and with that concentration being fed
to the reactor, the corresponding concentration in the effluent was
960 ppm.
In Run 109 an attempt was made to determine whether or not opera-
tion at 50°C in the presence of about 1000 ppm of soluble magnesium
would be feasible, producing solids with acceptable properties. The
27
-------
regeneration reaction proceeded well, and surprisingly, the soluble
magnesium level remained virtually constant in the system during 28
hours of operation — even though the magnesium level was well below
that corresponding to saturation with respect to magnesium sulfite at
50°C.
Experiment 110 was a repeat of 109 with the soluble magnesium
concentration approximately doubled. Even though the product of [Kg"*"1"]
x [803"] in the reactor effluent at the end of the experiment was
about 0.05, well below the equilibrium value of 0.32 at 50°C, a
significant decrease in soluble magnesium appeared to have occurred
during the 29 hours of operation. A 50-ppm reduction was observed
across the reactors at the end of the experiment.
The remaining closed-loop experiments were all carried out with
about 2000 ppm of magnesium in the feed initially, but in all cases,
the TOS level was lower than that in Run 110. Even though in each of
those experiments the solutions should have been unsaturated with
respect to magnesium sulfite, no measurable rise in soluble magnesium
could be detected.
Soluble magnesium concentrations measured in the feed to, and
effluent from, the regeneration reactor system during the three
extended, closed-loop runs are plotted in Figure 8. Magnesium concen-
trations in the Run 108 reactor effluent were essentially constant at
about 1000 ppm. A gradual decrease in soluble magnesium was observed
in the scrubber bleed as the regenerated liquor lower in magnesium
replaced the prime liquor (which initially contained about 2150 ppm)
in the settler, surge tank and absorber. A discontinuity in the Run
108 scrubber bleed magnesium level was observed at about 20 hours.
The elevated magnesium level was due to the addition of makeup liquor
to the settler which had an erroneously high magnesium concentration.
Levels in both the reactor feed and effluent throughout Run 109
were quite constant at about 1000 ppm. Run 110 was intermediate in
behavior between Runs 108 and 109. In Run 110, with only one exception,
the magnesium concentration in the reactor effluent was always about
70-100 ppm less than that in the corresponding feed sample. That small
decrease in magnesium concentration across the reactor system produced
28
-------
2,500
2,000 -
o.
Q.
vo
+" 1,500 -
1,000 -
Legend:
D Run 108, Scrubber Bleed
• Run 108, Reactor Effluent
A Run 109, Scrubber Bleed
A Run 109, Reactor Effluent
O Run 110, Scrubber Bleed
• Run 110, Reactor Effluent
500
10
15
20
25
30
Elapsed Time (hr)
FIGURE 8 CHANGES IN SOLUBLE MAGNESIUM OBSERVED DURING EXTENDED RUNS
-------
the constant slow decrease in the magnesium level in the reactor
effluent (and for the most part, in the scrubber bleed as well)
throughout the duration of the experiments.
There did not seem to be a clear reason for the slow drop in
magnesium which was observed in Run 110. The apparent solubility
product for magnesium sulfite in the effluent from the reactor was well
below the previously measured equilibrium value. The slow downward
drift was also not due to a corresponding rise in sulfite concentration.
The sulfite concentration varied more or less randomly in the samples
taken (by about + 7%). The water which entered the system as part of
the limestone paste could have been partly responsible for the down-
ward drift but its maximum effect would have been to reduce the magnesium
concentration by only 2% instead of the 4-5% drop observed across the
reactors at most times. The apparent unsaturation of the reactor efflu-
ent with respect to magnesium sulfite in Run 110, and other runs as
well, could, however, have been due to coprecipitation of magnesium
salts with the calcium salts. That same type of coprecipitation
phenomenon involving calcium sulfate produces reactor effluent solu-
tions unsaturated with respect to gypsum.
30
-------
V. CONCLUSIONS
The results of the experiments carried out during the course of
this program demonstrate quite clearly that within certain constraints,
it should be possible to use limestone for the regeneration of dual
alkali liquors without inordinate increases in the complexity of the
system as compared to systems using lime. Using a four-reactor system
with a total residence time of two hours, and finely ground, low-
magnesium (1.1% Mg as MgCO_) Fredonia limestone, utilizations of 78-
92% were achieved over a reasonable range of solution concentrations
without heating the solution above 50°C (typical of actual scrubber
bleed temperatures) and without actively controlling soluble magnesium.
Under those conditions, the solids which were produced settled well and
could be vacuum filtered easily. In most cases they settled to 20-40%
w/w in static batch settling tests; they could be filtered on a
laboratory Buchner funnel to produce cakes having 40-75% solids.
31
-------
REFERENCES
1. Final Report: Dual Alkali Test and Evaluation Program,
Volume II, Arthur D. Little, Inc., 1977. Chapter VI.
2. Ibid., Chatper VII.
3. Gmelin, Handbuch der Anorganischen Chemie, Vol. 27B, No. 1-4,
p. 207.
32
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/7-77-074
2.
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
Laboratory Study of Limestone Regeneration in
Dual Alkali Systems
B. REPORT DATE
July 1977
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO.
J.E. Oberholtzer, L.N. Davidson, R.R. Lunt, and
S.P. SoellenberE
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Arthur D. Little, Inc.
20 Acorn Park
Cambridge, Massachusetts 02140
10. PROGRAM ELEMENT NO.
EHE624
11. CONTRACT/GRANT NO.
68-02-1332, Task 26
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Task Final: 6/76-6/77
14. SPONSORING AGENCY CODE
EPA/600/13
is. SUPPLEMENTARY NOTES T£RL-RTP task officer for this report is Norman Kaplan, Mail
Drop 61, 919/541-2915.
16. ABSTRACT
The report describes a series of open- and closed-loop laboratory bench
scale experiments which were carried out to study parameters which affect the
reaction of limestone with dual alkali flue gas desulfurization system process liquors.
It gives details of several sets of operating conditions which permitted good limestone
utilization to be achieved and product solids with good dewatering properties to be
produced. It discusses the effects of temperature and soluble magnesium on the
behavior of the regeneration reaction as well as the effects of total sulfur (IV) concen-
tration, ionic strength, acidity of the solutions, and reactor system configuration. It
presents data which suggest that regenerated liquors may have been unsaturated with
respect to magnesium sulfite as a result of its coprecipitation with the calcium
sulfite product solids.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
COSATl Field/Group
Pollution
Limestone
Regeneration (Engi-
neering)
Flue Gases
Desulfurization
Alkalies
Magnesium
Pollution Control
Stationary Sources
Dual Alkali Process
Magnesium Sulfite
Calcium Sulfite
13B
08G
14B
21B
07A,07D
13. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
36
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
33
------- |