FLUID BED STUDIES OF THE LIMESTONE BASED
FLUE GAS DESULFURIZATION PROCESS
BY
A. SKOPP
J.T. SEARS
R.R. BERTRAND
FINAL REPORT
MAY 16. 1967 - AUGUST 27, 1369
Prepared under Contract No. PH 86-67-130 for
National Air Pollution Control Administration
Division of Process Control Engineering
ESSO RESEARCH AND ENGINEERING COMPANY
Government Research Division
Linden, New Jersey
GR-9-FGS-69
-------
- 1 -
FOREWORD
This report describes studies carried out by the Esso Research
and Engineering Company for the National Air Pollution Control Administra-
tion under Contract No. PH 86-67-130. The studies examined the feasibility
of desulfurizing flue gas in a fluidized bed with limestone materials,
both when discarding the sorbent and when regenerating the sulfated stones
and reusing the material in a cyclic process. The work was performed
over the period May 15, 1967 to August 27, 1969 at the Esso Research
Center in Linden, New Jersey. Mr. Allen Potter was the contract monitor
for the National Air Pollution Control Administration.
The authors wish to express their appreciation to Thomas
Coughlin and Henry Silakowski for performing the laboratory experiments.
We would like to thank Dr. A. C. Frost for his work during the initial
phases of the program.
-------
- ii -
TABLE OF CONTENTS
Page
FOREWORD i
1. SUMMARY 1
2. INTRODUCTION 5
3. DESULFURIZATION WITH LIMESTONE AND DOLOMITES 6
3.1. Apparatus 6
3.1.1. Gas Manifold System 6
3.1.2. Fluid Bed and Fines Recovery System 6
3.1.3. S02 Detection System 6
3.2. Experimental Procedures 9
3.3. Data Analysis 10
3.4. Measured Fluid Bed Reactivity of 12
Coarse Sorbents
4. IMPROVING SORBENT UTILIZATION 19
4.1. Effect of Bed Attrition on Sorbent Utilization 19
4.2. Effect of Physical and Chemical Modifications 23
on Sorbent Reactivity
4.2.1. Effect of Calcining Conditions on 23
Oxide Utilization
4.2.2. Reactivity of the Hydroxide 27
4.2.3. Studies with Uncalcined Limestones 28
and Half Calcined Dolomites
4.2.4. Reactivity of Metal-Oxide Modified Stones 31
4.2.4.1. Metal Oxide Modification of 31
N-1359 Lime
4.2.4.2. Copper Oxide Doped, Half- 32
Calcined Dolomite
4.3. Reactivity and Use of Finely Sized Sorbent 33
4.3.1. Small Particle Reactivity 33
4.3.2. Residence Time Required for Small Particles 34
in Fluid Bed Result in Uneconomical Design
4.4. Miscellaneous Studies of Coarse Sorbent Fluid 36
Bed Desulfurization
4.4.1. Sorbent Attrition Rate 36
4.4.2. Calcination in a Continuously Operated Bed 37
4.4.3. Effect of Fly-Ash 38
-------
5.
- iii -
TABLE OF CONTENTS
(Cont'd)
5.1.
REGENERATION STUDIES
5.2.
5.3.
5.4.
5.5.
6.
Thermodynamic Considerations in Sorbent
Regeneration
Experimental Apparatus
Experimental Procedures
Regeneration-Definition Studies
5.4.1.
5.4.2.
5.4.3.
5.4.4.
Reduction of CaS04
Sorbent Reactivity After First Reduction
Agglomeration
5.4.3.1. Fly Ash Effects
5.4.3.2. Self-Agglomeration Without
Fly-Ash
Possible Sulfite Formation and Thermal
Regeneration
Cyclic Performance
5.5.1.
5.5.2.
5.5.3.
Reactivity of Cycled Sorbents
Reduction Efficiency of Cycled Sorbents
Attrition of Cycled Sorbents
6.1.
PROCESS DESIGN CONSIDERATIONS
Conventional Power Plant
6.2.
6.3.
6.4.
6.5.
7.
Desu1furizer:
6.1.1.
6.1.2.
6.1.3.
6.1.4.
Regenerator
6.2.1.
Sizing Reactor
Sizing: Limestone Particles
Fluidization Requirements
Sorbent Selection
By-Product Processing
Schematic Design:
Conventional Power Plant
Future Process Variation:
Fluidized Bed Combustion
Possible Process Innovations:
Sulfur Production
7.1.
ECONOMICS OF FLUID BED LIMESTONE DESULFURIZATION
7.2.
7.3.
Basis for Cost Estimates
Process Installation in Existing Power Plant
Future Conventional Power-Plant Process Installation
7.3.1.
7.3.2.
Single Pass Desulfurization - 200 MW Plant
The Regenerative Process - 200 MW Plant
Page
39
39
42
45
45
45
49
50
50
50
51
51
51
53
55
58
58
58
59
60
61
62
62
62
64
65
66
66
67
68
68
70
-------
7.4.
7.5.
7.6.
- iv -
TABLE OF CONTENTS (Cont'd)
7.3.3.
The Regenerative Process - 1000 MW Plant
Direct Injection - 200 MW Plant
Regenerative Desulfurization in a Fluidized-
Bed-Combustion, 1000 MW Plant
Economics Summary
8.1.
STATUS AND RECOMMENDATIONS
Present Status
8.
8.2.
8.3.
8.4.
8.5.
NOMENCLATURE
REFERENCES
APPENDICES
I.
Technical Problem Areas
Other Process Considerations
Commercialization Lead Time
Re commenda tions
DESULFURIZATION KINETICS, BASIC DATA
II.
REDUCTIVE REGENERATION BASIC DATA
III.
ERROR ANALYSIS - DESULFURIZATION TO CaS04 FORMATION
EQUILIBRIUM PREDICTIONS OF POSSIBLE REACTIONS
IN DESULFURIZATION
IV.
V.
GIBBS FREE ENERGY OF FORMATION OF RELEVANT SPECIES
VI.
RESIDENCE TIME AVERAGING FOR THE EFFECTIVE RATE
CONSTANT IN A FLUIDIZED BED
VII.
ECONOMIC EVALUATION FOR A ONCE-THROUGH FLUID
BED LIMESTONE DESULFURIZATION PROCESS USING SMALL
PARTICLES
VIII.
FLUIDIZED BED REGENERATIVE DESULFURIZATION PROCESS
DETAILS
Page
73
77
78
79
81
81
81
82
83
83
84
85
87
88
91
94
96
104
105
109
111
-------
- 1 -
1.
SUMMARY
A study has been completed to determine the technical and economic
potential of removing sulfur dioxide from the flue gas of fossil
fuel-burning power plants by using inexpensive limestones or dolomites
as S02 sorbents in a fluid bed. The fluid bed desulfurization process
involves the reaction of S02 with CaO and oxygen at about l600°F.
CaO
+
S02
+
1/2 02
. CaSO 4
Initial experiments were made to define the kinetics of the
above reaction with precalcined sorbent. Five different limestones
and dolomites were studied in an externally heated, 3" diameter, fluid bed
reactor using a simulated flue gas. The data obtained showed that the
fluid bed reactivity depended on the sorbent used and on the degree of
sorbent utilization. For example, the rate constant determined for the
above reaction at l600°F for 2.8 mID diameter particles of a high CaO
limestone was found to drop from about 3000 rnin-l to 700 min-l as the
CaO utilization changed from 5 to 20%. Over this same range of CaO
utilization, the reactivity of a calcite dolomite that was investigated
varied from 7000 rnin-l to 3000 min-l. These rapid declines in reactivity
are due to the formation of Ca804 product shells which reduce the
diffusivity of 802 and 02 into the interior of the particles.
The rate of the desulfurization reaction is important in deter-
mining efficiency of desulfurization and sorbent utilization.
Thus, for the limestone cited above, with a nominal gas-solids contact
time of 0.1 sec., 100% 802 removal will occur with fresh limestone while a
maximum removal of only 48% will be achieved when the stone is 25% reacted.
In an attempt to improve the fluid bed reaction rates, a number
of sorbent and process modifications were investigated. These consisted of:
--Trying to remove the surface sulfate shell by operating the
fluid bed under particle-attriting conditions.
--Modifying the physical and chemical form of the sorbents
--Using finely sized sorbents
--Regenerating the sorbent by reducing the CaS04 to CaO.
Both an increase in bed height and in fluidizing velocity were
found to increase the amount of particle attrition in the fluid bed.
The dolomite tested attrited more readily than the limestone which was
investigated but in neither case was the attrition selective to the
surface sulfates, nor was it able to give increased sorbent utilization.
-------
- 2 -
Improving the calcination, using hydrated lime, using half-
calcined dolomites, and doping the sorbents with catalytic metal oxides
were all investigated as ways of increasing the reactivity. Pressure
induced calcinations, carried out at temperatures where the equilibrium
dissociation pressure of the limestone was greater than atmospheric
pressure, gave some improvement in sorbent reactivity over that obtained
with diffusion controlled calcination where the equilibrium dissociation
pressure was less than atmospheric. However, the small improvement was
not judged to merit any special calcining procedure. Hydration of lime
gave very small particles which were no more reactive than similarly sized
particles of calcined limestone. Metal oxide-doped stones and half-
calcined dolomites also gave no improvement in reactivity.
An increase in sorbent reactivity was found with decreasing
particle size but reasonably long fluid bed residence times were still
required to give good sorbent utilization. Thus, for example, it was
estimated that 60~ particles of one of the limestones which was studied
would have to spend an average of about 40 minutes in the reaction zone
to provide 90% flue gas desulfurization with stoichiometric addition
rates of the stone. The high-recycle, transfer-line reactor system needed
for such an operation would result in desulfurization costs estimated
to be above $2.20/ton coal. The concept of operating a "hetero-reactor
system" in which recycle requirements are reduced by using a bed of coarse
particles to prevent rapid entrainment of the much finer reactive particles,
was found to offer little or no advantage to the use of coarse particles
alone. .
Reductive regeneration of coarse sulfated sorbents was felt to
offer the potential of both decreasing sorbent requirements and producing
a concentrated S02 stream with by-product value. As a first step in
studying the regenerative fluid bed desulfurization process, the equilibrium
relationships involved in the reduction of CaS04 to CaO and S02 were
analyzed to define conditions for the reduction. Temperatures above
1900°F were found to be needed to reduce CaS04 and produce an effluent
gas containing SOZ. With CO or HZ as reductant, the reduction reaction is:
CaS04
+
CO(HZ)
6
~ CaO
+
COZ(HZO)
+
SOz
Under highly reducing conditions, considerable CaS is formed instead of
CaO. Experimentally, a Z:l ratio of COZ/CO was chosen to keep sulfide formation
in the particles at l-Z%.
Using an externally-heated, 2-1/2" ID ceramic fluidized-bed
reactor, a gas containing 10% CO or HZ was found to reduce CaS04 at
ZOOO°F to yield an effluent containing nearly 9% SOZ, the reaction apparently
proceeding to equilibrium. The oxide produced was capable of reacting
with SOZ. Cycling of stones between desulfurization and reductive
regeneration demonstrated continued reactivity for S02 by
regenerated particles. Continued exposure to high temperature caused a
decrease in reactivity, presumably due to porosity loss. However, desulfuriza-
tion activity was maintained at or above initial levels for 10 cycles
lasting about 110 hours with a calcined limestone. A nine percent
-------
- 3 -
802 effluent from the regenerator WaS produced in each of these regeneration
cycles. The limestone had a very low breakdown-elutriation rate, less than
O.025%/hr at 2 ft/sec. fluidizing velocity.
Although fly ash (from the combustion of coal) was found to have
no effect on the adsorption, data obtained at the higher temperature
corresponding to regeneration showed that the hot fly ash could cause
agglomeration of sorbent particles. Self-agglomeration of the sulfated
lime particles was also found to be possible at regeneration temperatures.
The data obtained in this program provided the basis for the
conceptual design of once-through and regenerative fluid bed desulfurization
processes for use in conventional power plants. Both the once-through and
regenerative processes would use a coarse limestone fluid bed reactor
operating at about l600°F. A temperature near l600°F is needed to obtain
sufficient reaction rates and this means locating the desulfurization
reactor somewhere in the superheater region of a modified boiler. Current
fluid bed technology and process design limits desulfurization reactor
diameter to about 50 feet. For a 200 MW plant, one such reactor would be
required while larger power plants would use multiple reactors of this size.
Using 1100 to 5000~ limestone particles, the once-through design
calls for a 24" deep settled bed operating at a superficial velocity of 15
ft/sec. Assuming ideal gas-solids contacting, 87% gas desulfurization is
calculated with three times stoichiometric sorbent addition rates. The
make-up sorbent could be uncalcined limestone since experiments have shown
it to calcine rapidly in the l600°F fluid bed.
The regenerative design assumes the same sized sorbent and
superficial velocity with a settled bed depth of 15". Ninety percent flue
gas desulfurization is calculated at a sorbent makeup rate equivalent to
one-third that required for stoichiometric reaction with the 802 in the
flue gas. The regenerator would be relatively small with a cross-sectional
area about 1/30 that of the desulfurizer. Partial combustion of fuel
would provide the high temperature gas needed to reduce the sulfated sorbent
and maintain the regenerator temperature at about 2000°F. The expected
effluent stream of 9% 802 could be prepared for use in a contact sulfuric
acid plant or could be reduced further to yield sulfur.
Both the once-through and the regenerative designs are based
on extrapolation of the activity data obtained on a low activity limestone
(i.e., the lowest activity of the five sorbents evaluated in this program).
Therefore, these designs are felt to be conservative with respect to most
other potential sorbents.
Economic evaluations, based on the above design consideration,
predicts the costs shown on the next page. A positive feature of these
processes is that the costs are relatively insensitive to the price of the
sorbent assumed in this study, i.e., $2.05/ton.
-------
- 4 -
Estimated Costs of Limestone Fluidized
Bed Desulfurization Processes
Plant Size Plant Operating Costs, $/Ton Coal Burned @
Process MW Investment 90% Plant Load Factor 60% Plant Load Factor
Once-Through 200 $3,030,000 2.35 3.18
Limestone Process
Regenerative 200 $4,585,000 1.02 2.01
Limestone Process 1000 $16,950,000 .44 1.14
The predicted costs for the regenerative process compare favor-
ably with those of other proposed desulfurization techniques when applied
to new pulverized coal boilers. The cost and time involved in modifying
existing boilers to accomodate a fluid bed desulfurizer, however, rules
out its use in such equipment.
Technical problems, such as gas solids contacting, passage of
fly-ash through the fluid bed grid, and agglomeration in the regenerator
remain to be investigated. Poor contacting in the adsorbing bed would
lead to lower than predicted gas desulfurization. In the designs used
for the economic evaluations, a grid containing many small gas inlet holes
was assumed in order to limit gas bubble size and provide good contacting.
With such a design, it would have to be demonstrated that fly ash would
not compact and clog the grid. In the regenerator, poor contacting could
result from agglomeration of fly-ash and sorbent. This could give incomplete
regeneration, lower than predicted S02 concentrations, and less efficient
use of reductant. If these problems cannot be avoided by changes in
regenerator design or operating conditions, then it would be necessary to
use a gas or liquid fuel to supply the reducing gas. This would not have
much effect on process economics since the fuel requirements for
sorbent regeneration are only about 3 or 4% those of the power plant
boiler.
The major effort in this study was devoted to the application
of fluidized bed desulfurization in conventional power plant boilers. However,
limited consideration was also given to its application in fluidized bed
combustion (FBC) boilers. In FBC, the fuel would be burned in a fluidized
bed of solids which contained a part or all of the steam raising surface.
These solids could then be limestone for an in situ desulfurization system.
As coal is burned above the grid in FBC boilers, the problem of passing
fly ash through a grid encountered with the conventional boiler is eliminated.
Achieving good gas-solid contacting would be much easier as gas velocities
are expected to be much lower in an FBC plant.
Fluid bed combustion itself is still only in the research stage.
Nevertheless, desulfurization in an FBC system employing sorbent regeneration
appears potentially attractive from both a technical and economic standpoint.
It is therefore recommended that future development of limestone--based
fluid bed desulfurization be directed at adapting the process to fluidized
bed combustion schemes.
-------
"
- 5 -
2.
INTRODUCTION
Among the many proposed processes being investigated to remove
502 from power plant flue gas, the direct injection of pulverized limestone
or dolomite into the boiler has the advantage of low initial capital cost
and moderate operating cost because of the use of inexpensive reactants.
The use of powdered limestone and dolomite as reactants to reduce the
sulfur dioxide content in flue gases is based on the reaction of 502 with
the CaO formed on calcination of the raw stone. Unless coupled with a wet
scrubbing system, direct injection of limestone does not achieve high
desulfurization efficiency or high sorbent utilization.
As part of the National Air Pollution Control Administration's
program to investigate the use of limestone and dolomites for flue gas
desulfurization, Esso Research and Engineering Company was granted a con-
tract to study the feasibility of using limestone and dolomite sorbents in
a fluid bed reactor system. The commercial application of such a fluid
bed desulfurization process would involve delivering the hot flue gases
(1600°F) to a separate reactor and subsequently returning them to the boiler.
The advantages of this process appear to reside in the potential for good
sorbent utilization, high desulfurization efficiencies, easy continuous material
handling, and the possibility for cyclic use of the sorbent.
-------
i
- 6 -
3.
DESULFURIZATION WITH LIMESTONES AND DOLOMITES
3.1.
Apparatus
The high temperature fluidized bed unit built for the limestone
desulfurization studies described in this report is shown in Figures 1 & Z.
unit consists of three major components, a gas manifold system for blending
a simulated flue gas, a high temperature fluid bed and fines recovery system,
and an infrared analyzer for monitoring the SOZ content of the effluent
gas.
The
3.1.1.
Gas Manifold System
The simulated flue gas used in these studies was formed in three
blending steps. NZ' COZ, and air were first mixed at room temperature in
a common carrier line. This mixture was then heated to over ZOO°F and
injected with steam in a 300-350°F furnace. Transport lines heated to above
150°F prevented water vapor from condensing out as the Nz-COZ-air-HZO
mixture proceeded either directly to a Z5 foot long preheat coil below the
fluid bed, or to the by-pass stream used for calibration of the SOZ detector.
S03 formation was minimized by the addition of SOZ to the mixture just before
it left the preheat coils and entered the fluid bed.
3.1. Z.
Fluid Bed and Fines Recovery System
The inlet gas leaving the Z5-foot preheat coil entered the 3"
I.D. 5-foot long Incoloy 800 fluid bed reactor through a 100~ opening
screen placed over a plate perforated with 3/3Z" diameter holes spaced
1/4" apart. A 3/8" LD. sample withdrawal port and a thermocouple well
extended down from the grid through the interior of the preheat coil windings.
The fluid bed reactor and the preheat coil were heated by a 6-foot long,
18 KW furnace manufactured by Electro Applications Incorporated, Houston,
Pa.
The top of the reactor was coupled to an insulated, 7" I .D. ,
18" high expanded section which returned high climbing slugs of particles
to the fluid bed. The gases leaving the bed passed through a heated line
at 400°F into a glass cyclone separator, then to a porous stainless steel
filter to remove the entrained fines. The cyclone filter-system was located in
a small oven kept at 400°F to prevent possible sulfuric acid condensation.
3.1. 3.
SOZ Detection System
The gases leaving the cyclone-filter unit passed through an
air-cooled condensing coil and a water-cooled condenser. The condensed
water was then separated from the gas stream in a cyclone separator.
Most of the gases then proceeded directly to the vent. However, .
up to Z CFM was side-tracked to a refrigerating unit which lowered the gas
temperature to 35°F and condensed out more water. This minimized inter-
ference from HZO absorption in the S02 IR analyzer.
-------
ELECTRIC
FURNACE
INCOLOY
REACTION
TUBE
FL U ID
BED
l~::)~
..":.:
~:~:i:;~
-,'
.":."
,:.'.:.
CYCLONE
AND
FIL TER
OVEN
..',
,'t'
.~:.::..:
:",:
,,:
f..,
l~~
;.:;~~
BY-PASS
LINE
.;:.
..~ :.
.: ;~:.:
:::/~
0"'-:
~~r-~l~
;" ~::
;f.::f
il{l.
!(i;i
;~.~'.:.
,-.;..
. ;~~ "
ANALYZER VENT
VENT
AIR
CONDENSER
H20
WATER
CONDENSER
H20
(S TEAM
CONDENSA TE)
THERMOCOUPLE
WELL
~t)\
. :.:;~'; SO
2
STEAM
S TEAM ADDITION
FURNACE
Figure 1
35°F
COOLER
H20
.
-...J
I. R.
ANALYZER
AND
RECORDER
Flow Diagram for the High Temperature Fluid Bed Desulfurization Unit
--SAMPLE WITHDRAWAL
PORT
;d
-------
- 8 -
Figure 2
High Temperature Fluid Bed Desulfurization Unit
,r-"
~,
"'
~~:~-~'~-.::.. ,...;?>~J~'-- ::.'
A Beckman I,R. analyzer model #IR3lS,used for determining the 502
content of the effluent gas, was calibrated before and after each run by passing
the simulated flue gas through the by-pass line going directly from the
stearn addition box to the condensing coils. This arrangement gave a cali-
bration curve which was not only quick and easy to prepare, but representative
of the actual gases being used in the experimental work. It took into account
the solubility and consequential removal of 502 and C02 in any water
condensed out of the gas, as well as the interference of H20 and C02 in the
I.R. analyzer.
-------
3.2.
- 9 -
Experimental Procedures
The experiments were made by charging the reactor with a known
amount of adsorbent, fluidizing the adsorbent with nitrogen while heating
it to the desired temperature, and then introducing other gases to produce
a simulated flue gas. The simulated flue gas used for most of the
experimental program had a composition corresponding to about that
obtained from combusting a heavy fuel oil containing 4% sulfur with 5%
excess air.
Table I
Simulated Flue Gas Composition
Gas Constituent Mole %
N2 74.73
C02 14.00
H20 10.00
02 1.00
S02 0.27
Six different stone types were investigated in this program,
a dolomite (N-1337), a calcite dolomite (N-135l), a limestone (N-1359),
a dolomitic limestone (N-1360), an algal limestone (N-1363), and a ferrous
limestone (N-1343), with primary emphasis given to the N-1337 and N-1359 stones"
Designation
N-1337
N-1343
N-135 1
N-1359
N-1360
N-1363
Table 2
Descriptions of Limestones and
Dolomites Investigated
Weight Percent, Calcined (2)
Quarry Source Type Stone (l) CaO ~ 8i02 ~ Al?01
Chas Pfizer Co. Dolomite 54 44 0.92 0.33 0.15
(Gibsonburg, 0.)
Hopper Bros. Quarries Fossili, ferrous 94 .85 2.98 0.66 0.73
(Ashland, Neb.) limestone
Jeffery Limestone Co. Calcite dolomite 54 28.5 8.2 7.0 1.55
(Parma, Mich.)
Grove Lime Co. Limestone 97 1.16 1.07 0.22 0.29
(Stephen City,
Va.)
Monmouth Stone Co. Dolomi te, 81 13.0 3.65 1.25 0.27
(Monmouth, Ill.) limestone
Casey Stone Co. Alga 1 66 1.42 21.0 2.50 6.60
(Casey, Ill.)
-------
- 10 -
3.3.
Data Analysis
Figure 3 shows the type of S02 adsorption data that were obtained
in the program. Conversion of CaO to CaS04 (i.e., sorbent utilization)
was calculated by integrating the area between the S02 breakthrough curve
and the feed concentration line to obtain the amount of S02 adsorbed.
The difference in weight of the batch of particles before and after a
run provided an independent check on the total utilization of limestone.
On the average, the chart records gave a calculated CaS04 formation
3.2% (:t1.7% at 95% confidence limits) greater than the actual weight
gain for the sorbent (see Appendix III). This is well within possible errors
from rotameter calibrations or decreased weight gain from possible
disproportionation of adsorbed CaO-S02 into CaS and CaS04.
Fig;ure 3
Typical S02 Adsorption-Breakthroug;h Curve
3000
- - - - - _u - - - - -~
,
I
!
;
i
i
!
I
t
- ---
----------
2500
Inlet 802 Concentration
rn
cu
o
300
2000
+J
~
OJ
;:j
,...,
44
44
I'
-------
- 11 -
The sorbent utilization at 20% S02 breakthrough was used in
this program as. a relative measure of sorption rates. Actual adsorption
rates were also determined. The equation used to obtain these rates
was derived assuming complete backmixing of the sorbent and plug flow
of the gas. Based on these assumptions, a kinetic reaction model was
developed to provide a relationship between a reaction rate coefficient,
R(X), and the following experimentally measurable quantities: the inlet
S02 gas phase concentration, Co; the outlet S02 concentration, CH; the
settled bed height, Ho; the superficial velocity, V; and the fraction CaO
utilized in the sorbent, X. The reaction rate coefficient, R(X), is the rate
of S02 removal from a volume of gas per unit time, per unit volume of
particulate bulk phase. The reaction rate coefficient is written as a
function of the sorbent CaO utilization, X, to show that it is dependent on
the degree of sorbent utilization. Thermodynamic analysis indicates that
MgO in calcined dolomites and limestones cannot react near 1600°F to effect
desulfurization, for both MgS04 and MgS03 are unstable at this temperature
(see Appendix V for details). As shown in Section 4.2.3. of the report,
even at lower temperatures where it can theoretically react, MgO still
does not form MgS04 to any appreciable degree. Thus, only CaO should react
to form the sulfate and permanently remove S02 from flue gas.
Considering the reaction of 802 with the sorbent on passin$
through the lamella shown in the following figure we can write the
material balance over a differential height, dz:
Figure 4
Material Balance Diagram
CH H
C + dz
z A
i L
C
Co
°
V
302 removed from gas = S02 absorbed by sorbent
-V'A'(C -C ) = A.dz'(l-E;).R(X).C = -V.A.dC
z z+dz z z
. IE. the void fraction in the
where A is the reactor cross-sect1ona area, 1S .
bed and C is the S02 concentration at the point Z. R(X) is the react10n
rat~ coeff~cient per unit volume of particulate phase. .On rearrang~me~t .
and integration over the boundary conditions of the flu1d bed one 0 ta1ns.
(1)
-------
- 12 -
JCH
dCz
U-
Co Cz
~:
R(X) (1- t. ) dz
( 2)
CH
In - =
Co
R(X) . (l-E.. )
U
. H
(3)
where H is the expanded bed height.
Assuming H = Ho/(l-£), equation (3) was derived to yield the final rate
equation (4) from which R(X), the reaction rate coefficient could
be determined from experimentally measured parameters.
lr~ = -R(X) .~
o
(4)
3.4.
Measured Fluid Bed
Reactivity of Coarse Sorbents
The initial fluid bed experiments were made using coarse,
16 to 18 mesh, pre-calcined particles of the dolomite N-1337 and the
limestone N-1359. Temperature was maintained constant at l600°F,
superficial gas velocity was varied from 2 to 6 ftlsec, and settled bed
height from 1 to 8 inches. The calculated reaction rate coefficients
are shown in Figures 5 and 6 as a function of sorbent CaO utilization (X).
Analyses of the data contained in these figures showed that the reactor model
~. 4) fit the experimental data over a wide range of superficial velocities
a t different fixed bed heights. However, as the settled bed height was increased,
the calculated reaction rate decreased. This is attributed to gas
distribution and contacting variations in the laboratory unit. Slugging
and channeling in a fluid bed are a function of the bed height. Slugs
of gas bubbles form at high gas velocities and extend across the entire
reaction area, causing inefficiences in desu1furization. However, they
are not fully formed upon entrance to the fluid bed and only form as the
gas travels through the bed. At low bed heights, insufficient time is
available before the gas leaves the bed to form these slugs. Preferred
routes of gas passage through the bed may also exist and such channeling
also decreases gas contacting. Different bed heights and reactor grid
configurations will result in different modes of channeling through the
particles. These different variations are difficult to predict and
analyze, and are generally empirically correlated, as was done in this
study.
-------
Figure 5 Figure 6
Reaction Rate Coefficient, Dolomite 1337 Reaction Rate Coefficient, Limestone 1359
10 10
U(ft/sec) Ho (in) U(ft/sec) .!!o.-(in)
l::. 2.0 2 . 0 1.5 2
\ . 2.5 2 V l::. 2.0 2
0 3.0 2 \ . 3.0 2
B ... 4.0 2 8 .. 4.0 2
. 5.0 2 . 5.0 2
V 0 6.0 2 V 6.0 2
\ . 4.0 4 . 4.0 1
V 4.0 1 A 0 4.0 4
V 16/18 Mesh 0 4.0 8
...., 6 \ 16000F ...., 16/18 Mesh f-'
I I 6 w
~ ~ 1600°F
'g 'g
0 V C")
C") \0. \
a a
...., ....,
........ \\ ........
:x :x
'-' 4 '-' 4
~ ~
~A
0"
2 ~~ 2
8~
0 0
0.2 0.4 0.6 0.8 1.0 0.1 0.2 0.3 0.4 0.5
CaO Utilization, X CaOUti1ization, X
-------
- 14 -
The reaction rate obtained for 6/8 mesh sorbent is shown in
Figure 7 for N-1359. For this particular sorbent and particle size
distribution. no effect of bed height was observed; however. the limited
data available prevent drawing any valid conclusions as to the presence
or magnitude of a bed height effect.
Figure 7
Reaction Rate Coefficient. Lime (1359)
5
4
Ho (inches)
.
.
2
1
U = 4 ft/sec
6/8 Mesh
l600°F
~ 3 .
I
Q
~
M
a
~ 2
~
~
'-'
~
1
o
0.1
0.2
0.3
0.4
0.5
CaG Utilization, X
The reaction rate coefficient. R(X). can be related to k(X).
the reaction rate constant, in the following manner:
R(X) = k(X).Y(k,Ho)
(5)
where the term Y. the contacting factor, is obtained in the actual
laboratory reactor used to study the reaction. The contacting factor. Y,
is essentially the efficiency of gas-solid con.tacting. It is a function
of the reaction rate constant, k, and the settled bed height. Ho. The
magnitude of the contacting factor depends on tontacting efficiency in
the actual laboratory reactor used to study the reaction and is determined
in part by the gas distribution provided by the grid system used. the
ratio of bed diameter to bed height. and the particle diameter employed.
Various models have been proposed for calculating contacting
efficiency in fluidized bed systems (3) and under conditiops of slugging
bed operation (4¥5). These models ar~based on the use of relatively
small diameter particles.
In the coarse particle systems used in this work we have
determined k(X) for two different sorbents by using equation (5)
in a correlation of the reaction, R(X). with Ho. This correlation
is based on a simplified linear model which assumes that the contacting
efficiency, Y. can be expressed as follows:
-------
- 15 -
Y(H ) = l-m(k) H (0 (Y <1)
o 0
(6)
where m = experimentally determined constant,
H
o
= settled bed height.
This model implies a contacting factor of 1.0 at zero (0)
bed height only. While this boundary is probably correct, it is possible that
the contacting efficiency remains at 1.0 for small finite bed heights.
If this is true, it would slightly effect the value of k finally obtained.
The simple linear model presented should be replaced in more refined
calculations by a more general exponential model as Y ----+ o.
By plotting values of R from figures 4 and 5 vs Ho at values
of X, and extrapolating to zero bed height, k(X) was obtained for both the
N-1337 and N-1359 stones at various values of X. The ratio, R/k=Y, was
found to be nearly the same at a given value of k, which showed that
the efficiency factor was indeed primarily dependent on Ho and k with
negligible variation due to stone composition and density differences.
Figure 8 presents the contacting factors so obtained and illustrates
the marked dependence of contacting efficiency on Ho and k in the laboratory
reactor. It is important to realize that similar contacting efficiencies
would not be expected for a commercial reactor having a different grid
design and a vastly different ratio of bed diameter to bed height.
><
n
H
0
oW 0.6
()
co
~
co
i=:
0.-1
.w 0.4 4 "
()
co
.w
i=:
0
u
1.0
0.8
0.2
o
Figure 8
Contacting
Factor - Rate Constant Dependance
in Laboratory Reactor
H =
o
1 inch
2
"
2
4
Reaction Rate Constant,
-------
- 16 -
The reaction rate constants for the 16/18 mesh dolomite (N-1337)
and lime (N-1359) have been calculated and are shown in Figure 9. Thus,
all the experimental data have been reduced to universal, simple reaction
rate results from which ultimate sorbent performance in good contacting
fluid beds can be predicted. In Figure 10, the desulfurization rate constants
for the three other stones, N-l35l, N-1363. N-1360, are presented along
with N-1359 desulfurization data for 6/8 mesh particles. The rate constants
for the two dolomitic stones (N-l337, N-l351) are seen to be substantially
higher than those of the high CaO limestones. Lime particles (N-1359)
appear to have the lowest reaction rate of any stone tested.
Fig;ure 9
Reaction Rate Constants for
Lime and Dolomite (1359, 1337)
16 H
o
1359, 1337 8
10 -16/+18 mesh 1600°F
H 8
I
P
OM
Ei
C'1 6 c
0
H
,.-... 4
:x:
'-' c
~
2
00 10
CaO Utilization, X
-------
6
M
I
i::
"M
S
C") 4
o
M
,-.,
~
'-'
~
2
- 17 -
Figure 10
Reaction Rate Constants for Various
Stones (See Table 2 for Compositions)
8
Ho=2 inches
l600°F
6/8 Mesh
o
o
0.1
0.5
0.2
0.3
0.4
CaO Utilization, X
-------
- 18 -
An indication of the change of the rate constant for
N-1359 lime particles as a function of temperature has been obtained.
The ratio of the rate constant, k, at temperature T to the rate constant
at l600°F was calculated by assuming contacting efficiency characteristics
similar to those in Figure 8 for particles other than 16/18 mesh size.
In Figure 11, kT/k1600°F is shown as a function of temperature.
This represents an outline of the decrease in the desu1furization reaction
rate as the temperature is lowered to l200°F from 1600°F. The present data
are consistent with the 28 kcal/mole activation energy found by Battelle
(6) in the range l500-2150°F for simultaneous calcination - desulfurization.
It might be noted that weight gain due to 002 pick-up was found at the lowest
temperatures; however, recarbonation was only of the order of magnitude
of the S02 reaction rate.
Figure 11
Rate Constant Temperature Dependence for N-1359 Lime
1.0
........
i:J:.<
°
a
a
~ 0.8 .
~
---
H
~
'-"
(J) 0.6
.w
~
(1j
.w
(J)
~
0 0.4
u
Q)
.w
(1j
~
Q) 0.2
:>
'M
.w
(1j
.--i
~
0
1200 1300 1400 1500 1600 1700
Temperature, of
-------
- 19 -
4.
IMPROVING SORBENT UTILIZATION
The rapid decrease in reactivity, shown in Figures 9 and 10 for
the different partially reacted sorbents, indicates that the attainment of
high sorbent utilization (e.g., >50%) in a commercial flue gas
desulfurization process would require long sorbent residence times if
coarse particles were used. Consequently, fluid bed reactors would have
to be designed for large solids hold-up. In order to improve sorbent
utilization and so decrease this hold-up, four different process variations
were investigated in our experimental program. These process variations
consisted of the following:
inducing surface attrition of the sorbent in the fluid
bed in order to continuously remove sulfated material.
making inexpensive physical or chemical modifications of
the sorbent to increase its reactivity.
using finely sized sorbent.
using coarse sorbent with continuous reductive regeneration
to the oxide and recovery of the S02 for conversion to a
salable product.
4.1.
Effect of Bed Attrition on Sorbent Utilization
Natural surface attrition, resulting from particle
contact within the fluid bed, was investigated as a means of
removing the Ca804 formed at the surface of the sorbent.
to particle
continuously
Attrition in a bed of given sized particles can be increased
in two ways. Either an increased superficial velocity or an increased
bed height should increase the degree of grinding of the particles with
one another. An increased bed height has the added advantage of decreasing
the competing shell formation rate, since more particles receive an
equal amount of 802 for a given period of time.
Table 3 and Figure 12 show the effect of these variables on
the overall attrition rate for N-1359 and on oxide utilization at the 20%
breakthrough point. A comparison of two runs made at a bed height of ap-
proximately six inches, one at a superficial fluidizing velocity of 3.5 ft/sec,
the other at a superficial velocity of 7.0 ft/sec, shows that a doubling
of the superficial velocity more than quadruples the overall attrition
rate. However, the increased attrition rate (0.33 gms/min) is only
about one-quarter of the formation rate of fresh oxide surface required
for stoichiometric reaction with the now doubled rate of S02 passing
through the bed, and the oxide utilization drops slightly from 6.7 wt. %
to 5.0 WL %.
-------
Table ~
EFFECT OF ATTRITION ON OXIDE UTILIZATIONS
FOR -12 +14 HESH PARTICLES OF CALCINED N-1359
Bed Charge - Grams Superficial Amount of Attrition Overall Formation Rate Total Oxide (CaO + MgO)
(Approximate Height - Gas Velocity before 20% Break- Attrition Rate of fresh oxide Utilization at 20%
inches) ft/sec through - Grams gm/min req ui red-gmhnin Breakthrough Point
580 grams 3.5 5 .07 0.6 6.7%
(.... 6 inches)
580 grams 7.0 il .33 1.2 5.0%
(V1 6 inches)
1175 grams 7.0 71 1.2 1.2 2.6% N
a
(VI 12 inches)
2345 grams 5*-7.0 703** 2.3 1.2 14.8%
('-""24 inches)
* Operating difficulties necessitated higher bed pressures with correspondingly lower
superficial velocities.
** Operating difficulties gave a poor material balance; amount of attrition was a value
calculated from the differences between the before and after bed weights and the S02 flowrate.
-------
s::
.~
13
--
13
CO
I
I1J
.j.J
rn
p::
s::
o
.~
.j.J
.~
H
.j.J
.j.J
<
- 21 -
Fi~UX"e 12
Effect of Bed Height on
Attrition Rate of N-1359
3
2
Formation Rate of Fresh Surface
Required for Stoichiometric
Reaction with Inlet S02 - gm/min
at V = 7.0 ft/sec.
1
-----------
./
./
.........
/'
-- - - -- -- -----
o
o
6
12
18
Bed Height - Inches
24
30
-------
- 22 -
In the expectation that an increase in the bed height would
increase the degree of internal grinding of the particles and promote
attrition of the sulfate shell, additional runs were made at bed heights
of twelve and twenty-four inches at a superficial fluidizing velocity of
7.0 ft/sec (Figure 12).
Increasing the bed height to twelve inches greatly increased
the overall attrition rate to 1.2 gm/min, equal to the 1.2 gm/min formation
rate of fresh oxide surface required for stoichiometric reaction with
the S02. However, the oxide utilization dropped to 2.6%. Such loss
of oxide utilization was also observed in some of the later studies.
One possible explanation for this behavior is that since the sulfated
surface is slightly "sticky", and hence possibly malleable at these
temperatures (as indicated by the difficulty in dumping a fixed or
settled bed), a moderate degree of attrition may actually have deformed
and sealed surface pores of the sulfated shell.
With the expectation that an even higher degree of attrition
could overcome this possible sealing effect of attrition, the bed height
was again doubled to 24". The resulting 2.3 gm/min attrition rate was
nearly twice the formation rate of fresh oxide surface required for
stoichiometric reaction with S02, but only 14.8 mole % oxide
utilization was obtained. This low utilization suggested that the
attrition was not selectively removing sulfated surface to expose
reactive oxide but was breaking down the particles in their entirety.
This was confirmed in studies with the dolomite, N-1337. Using 6" to 24"
deep beds of calcined 12 to 14 mesh N-1337, with superficial gas
velocities up to 7.0 ft/sec, it was found that unreacted dolomite broke
up more readily than the limestone to produce fines in the 50 to 100
mesh range. Under similar conditions, the same sorbent but having 30%
of its CaO sulfated, attrited as readily as the unreacted dolomite and
produced fines of about the same composition as the remaining sorbent.
If the surface were being selectively attrited, the fines would have
had a higher sulfate content than that of the bulk adsorbent.
-------
- 23 -
4.2.
Effect of Physical and Chemical
Modifications on Sorbent Reactivity
A number of physical and chemical modifications of the limestone
(N-1359) and the dolomite (N-1337) were tested to determine if these
could provide more reactive sorbents whose decreased requirements would
more than offset the costs of modifying the sorbents. Improving the
calcination, using the hydroxide form of the sorbent, using half calcined
dolomites, and doping the sorbents with catalytic metal oxides were all
investigated as ways of improving reactivity.
4.2.1.
Effect of Calcining Conditions
on Oxide Utilization
The following two types of calcination were studied in detail for
the limestone N-1359: one which removed the C02 from the carbonate sites
by diffusion (i.e., the equilibrium dissociation pressure remained below
1 atm.), and one where the C02 left the carbonate sites at a dissociation
pressure above atmospheric pressure. Table 4 and Figures 13 and 14
summarize the results obtained from this study. The diffusion controlled
calcination gave oxide utilization in the 1.5 to 3.5% range and the
pressure induced calcination gave utilizations in the range of 6 to 8%.
The pressure-induced calcinations were carried out in a
l' x l' x 2' stainless steel box whose walls and top were maintained at
1800-1850°F in a large Lindberg furnace (see Figure 15). Small
quantities of particles were dropped through a pipe into the box, where
they were rapidly heated by radiation to 1665°F. They then proceeded
to evolve C02 at atmospheric pressure. The high surface heat flux
probably permitted small entrapped pockets of carbonate to overheat,
overpressure, and finally rupture to form needed voids in a popcorn
manner near the particle surface. Unca1cined stone situated further
away from the outside surface was insulated by the layer of calcined
stone, and so the absence of a high heat flux probably minimized pore
formation towards the center of the particle. Higher box wall temperatures
to induce higher particle surface temperatures were avoided to minimize
the dead-burning of completely calcined particles. Total calcination
time was governed by the time it took a previously added batch of particles
to calcine and reach 1800-1850°F before they were covered with a new batch
of carbonate. 1100 grams of N-1359 could be calcined in one hour.
Some additional conclusions regarding the selection of ca1cinating
conditions can be drawn from the data in Table 4. Glasson (7) found
that dead burning with the consequential loss of specific surface area
increased with temperature and the C02 concentration above the calcined
oxide. This same effect is apparent in our data. Thus, Runs 6, 7, and
12 using material calcined at temperatures between 1450 and 1625°F gave
higher utilizations than Runs 9, 10, and 11 which used material calcined
at 1750°F, where both the temperature and C02 concentrations (itself a
function of temperature) were relatively high. This trend is shown by
the dashed line in Figure 13. The series of runs made with material
-------
% CaO
Utilized
Run II (at 20% B.T.)
6 3.49%
7 3.52%
9 2.81%
10 2.04%
11 2.30%
13 2.56%
16 1. 50%
12
3.25%
19
6.66%
4
5.43%
Table 4
Effect of Calcining Conditions on
Oxide Utilizations for -12 +14 Mesh BCR-1359
Calcining Conditions
802 Reacting Conditions
Fluidized Bed Calcinations (Diffusion Controlled) - Fluid Bed Run
3.5 ft/sec
Gently fluidized bed, 2 1/2 hrs up to l450°F,
held at l450°F for 1/2 hr
Fluidized bed, up to l600°F in 1 hr, dumped
Fluidized bed, up to l750°F in 1/2 hr, dumped
Fluidized bed, up to l750°F in 1/2 hr, held at
l750°F for 1 hr
Fluidized bed, up to l750°F in 1 hr held at
l750°F for 1/2 hr
Fluidized bed, up to l400°F for 8 hr, some attrition
Violently fluidized bed, up to l300°F for 12 hours -
high at tri tion
580 gm bed, l600°F, V ~
29i13 fines*
580 gm bed, l600°F, V =
1 gm fines
580 gm bed, l600°F, V =
0.5 gm fines
580 gm bed, l600°F, V = 3.5 ft/sec
0.3 gm fines
3.5 ft/sec
3.5 ft/sec
580 gm bed, l600°F, V = 3.5 ft/sec
0.8 gm fines
580 gm bed, l600°F, V = 3.5 ft/sec
8.3 gros fines
580 gm bed, l600°F, V - 3.5 ft/sec
no fines
N
~
Fixed Bed Calcination (Diffusion Controlled) - Fluid Bed Run
Fixed bed calcination, V = 1.2 ft/sec, top of bed 580 gm bed, l600°F, V = 1.2 ft/see
at l625°F in 7 hrs - difficult to dump fixed bed - very difficult
to dump. No fines
Particles exposed to l800°F walls-heated for 1 - 2
hours from radiation - no purge gas
Fixed bed with purge-walls to l700°F, partially
pressure-induced, partially diffusion controlled
calcination
Note:
All runs had breakthrough curves of the type shown in Figure 4.
* Amount of material collected in the cyclone separator.
580 gm bed, l600°F, V =
5 gm of fines
580 gm bed, l600°F, V =
1 gm fines
3.5 ft/sec
3.5 ft/sec
-------
iN:
o.r::
N M
::J
~ ~ 8 -
.r::
t:: ...,
O~
OM (1)
~ ~ 6
N~
OM
.-I N
"ri0
~ tI.1 4
o
(1)
u
~ 2
o
1200
.r::
M
::J
o
~ ~ 3
...,
t::~
o (1)
OM QJ
..., H
(1)~
N
OM N
.-10
OM tI.1
~ ~ 2
o
ON
(1)
U
~
10
- 25 -
Figure 13
Effect of Calcining Conditions on
Oxide Utilization for N-1359
Particle Size: - 12 + 14 Mesh
Reactor Conditions: 580 gm bed, l600°F, V = 3.5 ft/sec
Time at Calcination Temperature: Variable
~ - Diffusion Controlled
. - Part Pressure Induced, Part Diffusio
Controlled.
A - Pressure Induced
A
.
x
x
-
--
....--- X
----.
x
1400
1600
1800
2000
Calcination Temperature - Deg Fahr.
Figure 14
Effect of Time at Calcination Temperature
on Oxide Utilization for N-1359
Calcination Temperature - l7500p
Diffusion Controlled
o
1/2
1
1-1/2
Time at Calcination Temperature - Hours
-------
1 LOADING PIPE
/ TO OUTSIDE TOP
THERMOCOUPLE I / OF FURNACE
~ ~~I~@
FURNACE@I ~I. ~ CS5 ;~/('
HEATERS ---- ~. r::::j I - ~ >-;:::::;.
~~ r~ V I ~ ~/~ ~~
I . C I tS? ~ :?=: ~R
~ ~ ' ~ t ~;~ @ ~5
,~ ;- ,- I ~ Ji? p (5 ;g
r<::2 ~ ;i ig~~1 $~ pg
;:g>~ 'bW~~'~~~
sg ;22 ~tg ~-~~~_~~_f@
re=:>9 )S ~g I KS ::2S 1~<8
«;? '.-c- '. - (<:::::> rg ><::::::> ~
~ . (0 o:~~o~ ~ ::~
'<=::> '" I 0,,000Q. OC) ('. 0;:::::: c (~ t~ /S:
(g "," o~~o~o-():O' 00/':0- ,~~ (-~/-~
(~ / OOQ~~"~/?~
~ ,/ IOO°:.:::JO ~ ~ ~
sg ,/,/ c'oo@oo~~ p
~<=>",/ ~ ~ tg
'~ I", p/2
;;../// I ~. @g
(gL. Vcg
N
0'>
CALCINING PARTICLES
STAINLESS STEEL
BOX
Figure 15
Radiant Heat Calciner for Pressure-Induced Calcinations
-------
- 27 -
calcined at l750°F, but maintained at calcination temperature for varying
lengths of time, also showed the decrease in oxide utilization resulting
from an increase in dead burning as exposure time was lengthened. This
is shown in Figure 14. Undoubtedly the pressure-induced calcinations
carried out in a pure C02 atmosphere at higher temperatures had an even
higher degree of dead-burning. In these cases, however, the resultant
pore formations evidently offset this deleterious effect.
All runs referred to in Table 4 had sharp breakthrough curves
implying that further diffusion-controlled reaction would be slow.
Although some increase in sorbent reactivity can thus be obtained by
operating under pressure induced conditions, large improvements in
sorbent utilization are judged to be unlikely.
4.2.2.
Reactivity of the Hydroxide
To determine if the hydroxide would be the preferred form of
the sorbent for fluid bed desulfurization, experiments were carried out
with the hydroxide made from N-1359 limestone. The hydroxide was made
by calcining the carbonate, slowly adding the stoichiometric amount of
ice water needed to form the hydroxide, and then drying the hydroxide
for five hours at 100°F.
A 325 cm3 bed of 25 to 50 mesh hydroxide made in this way
was tested with simulated flue gas at a superficial velocity of 1.2
ft/sec and a 600°F temperature. This temperature was selected because
it is below the thermodynamic dissociation temperature of Ca(OH)2 in
the flue gas atmosphere (see Appendix IV, Section IV-2).
Total S02 removal occurred until 3.5 wt% of the Ca(OH)2 had
been consumed and then there was a very sharp S02 breakthrough. The
discharged bed material was found to contain 30% C02' Evidently, the
carbonation proceeded at a sufficiently high rate to hinder sulfation,
probably through pore blockage.
Recarbonation is thermodynamically unfavorable at temperatures
above about l450°F (see Appendix IV, Section IV-2) , but at such
temperatures, the hydroxide reconverts to the oxide. To determine if
the oxide made from such an intermediate hydroxide was more reactive
than that made directly from the carbonate, the following experiment was
made. 250 cm3 of the 25 to 50 mesh hydroxide was run in the fluid bed
at a velocity of 1.2 ft/sec and a temperature of l600°F. Under these
conditions, 41.9 wt%of the CaO reacted at the 20% S02 breakthrough point.
A comparable run with just the calcined N-1359 sorbent (i.e., no
intermediate hydroxide formation) yielded only 22.2% CaO utilization
at the 20% breakthrough point.
The increased oxide utilization of the hydroxide-treated
material might be explained in two ways. The hydroxide formation-
calcination steps could conceivably "activate" a hitherto unreactive
material through crystal rearrangements. A more likely explanation,
however, is an increased porosity, which might be caused from the
agglomeration of very small particles to form the -25 + 50 U.S. mesh
bed particles. During the hydration step, severe decrepitation was
-------
- 28 -
noted with the formation of a high percentage of very small particles.
In a commercial process, if these particles were not retained in the
bed for a sufficient time to insure good particle utilization, there
would be no advantage to using intermediate hydroxide. On the other
hand, a fluid bed designed to retain such particles would also retain
small particles of normally calcined N-1359 which later experiments
showed to be about as active as the more difficult to prepare hydroxide.
Therefore, no advantage is seen for using the hydroxide.
Studies With Uncalcined Limestones
and Half Calcined Dolomites
4.2.3.
At temperatures above approximately l500°F, both MgS03 and
MgS04 are thermodynamically unstable in a typical flue gas environ-
ment (see Appendix IV). At temperatures below about l450°F.
on the other hand, CaC03 is thermodynamically stable at the 14% C02
concentration of flue gas (Figure~l, Appendix IV). Consequently, the
only way that both the calcium and magnesium portions of a dolomite can
be utilized for S02 adsorption is for the process to be operated at a
temperature where the CaC03 remains uncalcined but where S02/02 is
able to displace C02 from the carbonate
In an attempt to evaluate this displacement reaction, 12 to 14
mesh of the uncalcined limestone N-1359 was used in a one foot bed at
600° and l200°F. At the lower temperature, there was no noticeable
reaction while at l200°F, the calculated rate coefficient was less than
2% of the coefficient for calcined material at l600°F.
In contrast to the limestone, the half calcined dolomites
were found able to adsorb S02 at temperatures down to 1000°F.
Three dolomites, N-1337, 1351 and 1360, were calcined at 1200 to l300°F
in a S02-free flue gas atmosphere. Under these conditions, the MgC03
was converted to the oxide while the CaC03 remained unchanged.
CaC03 . Mgc03
.
CaC03 . MgO + C02
(7)
Weight losses sustained during calcination corresponded to those expected
from the above reaction as Table 5 shows.
The reactivity of half-calcined materials at low temperatures
was much lower than the fully calcined stones at l600°F and appeared
to be less dependent on utilization (see Figure 16). Comparative results
for the three stones are summarized in Table 5 for reaction at l300°F.
-------
- 29 -
Table 5
Sulfur Dioxide Capture by Half-
Calcined Dolomites at l300°F
Half Calc'n. % CaO Utilization
Mol. % of CaO in Wt. Loss at constant k
Sorbent CaO + MgO Theor. Actual (k ~ 100 min-l)
N-1337 46.7 25.4 26.9 87
N-1351 57.5 18.0 18.1 52
N-1360 81.9 8.0 9.6 11.5
Theoretically, the half-calcined dolomites can remove S02 by
reaction with CaC03, MgO or both. The failure of the MgO to react to any
meaningful degree was confirmed in these studies by analysis made of
sulfated stones, Table 6. The sorbent weight increase, and the percent
sulfur and C02, determined by analyzing the final product, agreed most
closely with those values calculated on the basis of no MgO reaction.
Table 6
Analysis of Sulfated Half-Calcined Dolomites
, .~--:--:--_.-
Calculated Assuming % of
MgO Sulfated Is:
N-1351
Observed 0% 10% 20%
10.9 11.0 10.8 10.5
10.0 11.0 12.6 14.2
11.5 14.5 16.8 18.9
Wt. % Sulfur
Wt. % C02
Weight Gain, %
N-1337
Wt. % Sulfur
Wt. % C02
Weight Gain, %
15.9
4.8
17.2
~5.5
3.5
17.8
14.9
6.0
24.2
14.5
8.5
27.4
-------
M
I
0 4
.--i
~
.--i
I
!::
"g
"'"' 3
~
'-/
p:;
6
5
2
1
o
- 30 -
Figure 16
Reaction Rate Coefficient for Calcined Dolomite (1337)
7
T = 1250°F
16/18 Mesh
T = 1600°F
(Fully Calcined)
o
0.4
0.8
0.2
0.6
CaG Utilization, X
1.0
-------
- 31 -
As Figure 16 shows, however, its activity was much lower than that of
the fully calcined dolomite. Consequently, the use of half calcined
dolomites is not judged to be attractive, even though they can be used at
lower temperatures than fully calcined dolomites or limestones.
4.2.4.
Reactivity of Metal-Oxide
Modified Stones
The modification of sorbent by doping with metal oxides was
also investigated as a means of increasing reactivity. Doping was carried
out by contacting the uncalcined sorbent with a solution of metal nitrate,
washing the doped sorbent, followed by drying and calcination. Treatment
of the sorbent with nitrate solutions of copper (II), aluminum and iron (III)
resulted in considerable evolution of C02 suggesting reaction with CaC03.
The treatment with the copper solution produced a hard copper oxide shell
around the sorbent. Treatment with the aluminum and ferric solutions
resulted in the formation of gelantinous aluminum and ferric hydroxide.
4.2.4.1.
Metal Oxide Modification of N-1359 Lime
As shown in the following table no improvement in CaO utilization
was observed when the calcined, doped N-1359 lime was used for desulfurization.
In most cases, a definite deterioration of sorbent performance was observed.
All the breakthrough curves with exception of the copper-treated sample
were similar to that obtained with untreated sorbent, but were displaced
toward lower utilizations. The sorbent treated with the copper solution
had a gentler breakthrough curve, but did not give an increase in
oxide utilization.
Table 7
Modification of 1359 Lime - Doping With Metal Oxides
Treatment Calcium Oxide Utilization at % SO~ BT
1 20 50
None .18 .20 .21
None .18 .20 .21
Nickel .17 .20 .21
Copper (II) .14 .19 .22
Copper (II) .09 .16 .20
Aluminum .07 .10 .12
Cobalt .12 .14 .16
Cobalt .07 .09 .12
Ferric .11 .14 .16
-------
- 32 -
The effect of modifying the sorbent with copper oxide was
further investigated by measuring the rate of 502/02 reaction using
treated and untreated sorbent calcined under identical conditions.
The following table shows that no major change in reactivity occurred on
modifying the sorbent.
Table 8
Reaction Rate of 802/02 With
Modified N-1359 Lime at l600°F
Fraction CaO
Utilized
Reaction Rate Coefficient (Min-l)
16/18 Mesh 12/14 Mesh
Copper Oxide Control Copper Oxide Control
0.05
7430
6400
0.10
2770
3500
3180
3950
0.20
530
100
A copper modified sample of uncalcined N-l359 limestone was
used for desulfurization at l300°F. Effective desulfurization under
these conditions was not successful. A measurement of the rate of 502/02
reaction resulted in a rate coefficient of only 156 min-l at 5% CaD
utilization - a factor of at least 50 less than with calcined sorbent
at 1600°F.
4.2.4.2.
Copper Oxide Doped,
Half-Calcined Dolomite
Dolomite (N-1337) was treated with an aqueous cupric salt
followed by calcination at l200°F to deposit a film of copper oxide on
the sorbent. A comparison of the rate of 502/02 reaction at l200°F
using the modified dolomite and a normal half-calcined dolomite showed
a slight decrease in sorbent reactivity on modification with copper.
This effect is illustrated in Table 9.
Table 9
Rate of 802/02 Reaction With Modified
Half-Calcined Dolomites at l200°F
Dolomite (1337), 14/16 Mesh Particles
Fraction CaO
Utilized
Normal Half-
Calcined Dolomite
Copper Modifie~ Half-
Calcined Dolomite
0.1
0.2
0.3
-1
1000 min
-1
790 min
-1
860 min
-1
470 min
-1
640 min
-1
205 min
-------
- 33 -
4.3.
Reactivity and Use of
Finely Sized Sorbent
The use of finely sized sorbent, alone and in combination
with coarser material, was investigated as a way of achieving both good
S02 removal efficiency and high sorbent utilization in a fluid bed flue
gas desulfurization process.
4.3.1.
Small Particle Reactivity
The rate constant data presented in Figures 9 and 10 and the
less detailed results on intermediates and small particle sizes available
in Appendix I were analyzed for a dependence of rate constants on particle
size. Using assumptions of similar contacting efficiency characteristics
and CaD utilization-rate constant characteristics for different particle
sizes, rough dependence of rate constants on particle size were obtained.
For 1337 dolomite, 60% utilization, the following dependency
was found in the diameter range 100-3000~,
-0 5
k1337 d- d .
(8)
For 1359 lime, 20% utilization, a different dependency was found,
-1 3
k1359'~ d .
(9)
Although the power coefficiencies may easily be in error by 1/3,
appear from these first approximations that 1359 lime has a much
marked dependence on particle mesh size than 1337 dolomite.
it does
more
If only gaseous diffusion into the porous particle were the
limiting mechanism, one would expect a d-2 proportionality on the rate
constant, assuming a constant diffusion coefficient in a continuously
reacting outer shell (~). Thus, our experiments showed that a simplp
diffusion model does not adequately represent the data.
The uneven powers determined for the effects of particle diameter
could be explained by many different reaction mechanisms. Among the possibil-
ities are the following: (a) a mixture of chemical reaction and diffusion
control, (b) a variation in the diffusion constant with particle size
if Knudsen diffusion is important, (i.e., change in pore size distribution
from calcination), and (c) the necessity of molecular reorientation of
the solid with chemical reaction. With such a number of unresolved
possible reaction mechanisms, it would seem advisable to base engineering
predictions only on experimental, empirical correlations and not on
mechanistic schemes.
-------
4.3.2.
- 34 -
Residence Time Required for
Small Particles in Fluid Bed
Result in Uneconomical Design
The results of Section 4.3.1. relating the reaction rate
constant to particle size have been combined with the detailed kinetic
data on 16 to 18 mesh N-1337 presented in Section 3.4. to predict
utilization versus residence time for individual 60~ and 1100~ particles
at 1600°F. The results obtained are shown in Figure 17. The analysis
is based on 9070 desulfurization of a 2700 ppm S02 containing flue gas.
Individual particle residence time required to achieve 50% utilization
with the small particles is calculated to be about 8 minutes compared to
one hour for the large particles. At 90% sorbentO utilization, the small
particles require about 40 minutes, down from the 5 hours required for
the large particles.
1.0
60]1
0.9
0.8
0.7 -
'"0
(])
N
''-;
r-i 0.6
''-;
+J
~
0 0.5 L
~
u
~
0
''-;
+J 0.4
()
~
,..
~
0.3
0.2
0.1
0 I I I I ~ I 1
0 20 40 60 80 100 120 140 160 180
Particle Residence Time - Minutes
Figure 17
CaO Utilization for Single Particle N-1337 80rbent
2700 PPM 802 in Flue Gas
90% Desu1furization
-------
- 35 -
These residence time requirements for small particles would not
result in a conventional fluid bed design for desulfurization but rather
in the design of a high dust-loaded transfer line reactor using many
stages of very large cyclones for solids recycle. An analysis of such a
design, detailed in Appendix VII, indicates the process to be economically
unattractive and of questionable technical feasibility.
In an attempt to reduce costs by decreasing cyclone requirements
and designing a more conventional fluid bed, the "hetero-reactor concept"
was investigated. The hetero-reactor concept involves the use of coarse
particles in conjunction with fine particles to prevent the immediate
entrainment of the latter, while still obtaining the benefit of its high
reactivity. An analysis was made of the expected performance of a hetero-
fluid bed reactor assuming that a settled bed height of 2 ft. would be
required to obtain the necessary gas-solids contacting for removing 90%
of the S02 from the flue gas. With this bed height~ stoichiometric
addition ofllOO~ fresh sorbent would provide an average residence time of
about 8 hours. Calculations based on Figure 17 and the model developed
in Appendix VI indicate that the particle utilization would be 90% without
any fine particles present so there would be no advantage in using them.
The situation changes somewhat if suitable gas-solids contacting can be
obtained with a bed height of only 1 ft. In this case, the average
particle residence time is reduced to about 4 hours. At stoichiometric
addition, this is unable to provide 90% desulfurization with only the
coarse particles present. If reactor operation at a one-foot settled bed
height can be achieved 1 then it is necessary to consider the cost savings
from use of a lower bed height as compared to the cost of using both coarse
and fine particles in a hetero-reactor operation. For a one-foot settled
bed height, the operating cost savings for a fluid-bed desulfurizer sufficient
to treat the flue gas from a 800 MW power plant are estimated to be
about $195,000/yr or 89/ton coal burned (Appendix VII). Based on an
annual N-1337 dolomite requirement of 548 tons/year for a 1.0 stoichiometric
feed rate, this operating cost savings can support an expenditure of about
$0.36 per ton of sorbent for operation of the hetero-reactor system. This
cost savings assumes no treatment of the coarse sorbent is necessary to
prevent immediate entrainment of the fines. In experiments we have made with
a hetero reactor, this has not been found to be the case. Observed fine
particle residence times have been about the same as gas residence times
indicating almost immediate entrainment. In a one foot bed, such low
residence times would give an even lower utilization to the fine particles
than to the coarse ones. It is likely therefore that some treatment of
the coarse sorbent would be required to provide a surface capable of holding
the fine particles. Such treatment would undoubtedly cost more than $.36/ton
limestone. Coupled with the improbability of a one foot bed being able to
provide sufficiently good gas-solids contacting~ the hetero-reactor concept
is judged to offer no advantage over a coarse particle system.
-------
- 36 -
4.4.
Miscellaneous Studies of Coarse
Sorbent Fluid Bed Desulfurization
Three additional factors were studied that could effect the
use of coarse Limestone or dolomitic sorbent in a fluid bed flue gas
desulfurization process. These factors are sorbent attrition, calcination
and interaction with hot fly-ash.
4.4. L
Sorbent Attrition Rate
Attrition, followed by elutriation, is important in a coarse-
particle bed, since it causes sorbent to be transported out of the reactor
before full utilization has occurred. By measuring the amount of material
entrained overhead in a run, a measure of stone stability can be obtained.
Particles entrained out of the bed will be those fragments whose terminal
velocities are less than the gas velocity through the disengaging section
of the laboratory reactor.
A systematic study of the attrition of calcined dolomite N-1337
was made to ascertain the effect of gas velocity on particle
break-up. Figure 18 shows the increase in attrition plotted against
(U-Uo), the actual minus the minimum fluidizing velocity. U-Uo is used
since no attrition would be expected below the minimum fluidization
velocity, the transition point from a static bed to the fluidized
attriting stage.
Although the number of stones tested was limited, a comparison
of calcined particles' attrition rates showed the dolomite to have less
physical integrity than high CaD lime particles.
Table 10
Relative Attrition of Limestones
Measured from overhead loss of calcined
materials in initial desulfurization
Stone No., N-
Relative Attrition Rates
1337 (dolomite)
1
1343 (ferrous limestone)
0.125
1359 (limestone)
0.185
The data from Figure 18 and Table 10 were used for an
approximation of the expected attrition in a high velocity, large fluidized
bed in Section 6. where the design of a commercial-size reactor is
described.
-------
- 37 -
Figure 18
Effect of Velocity on Particle Attrition
4
+J
Q
Q)
s.
~
on
C\! 3
!-<
+J
~
~
>,
..c .
!-< 2
::r::
-
rJ)
rJ)
0
H 1
+J
...c: .
bO
on .
Q)
~ 0
~ 0 1 2 3
4 5
Excess Superficial Velocity, (U-Uo)
4.4.2.
Calcination in a
Continuously Operated Bed
The use of unca1cined limestone as sorbent make-up to the
fluid bed is naturally preferred to the use of the more expensive pre-
calcined stone. To confirm that unca1cined sorbent could be charged
to a fluid bed desu1furization process, parallel runs were made in which
the stoichiometric quantities of calcined and unca1cined 70 to 100
mesh N-1337 particles needed to react with 85% of the incoming 802, were
added at 5 minute intervals to a gently fluidizing (V = 1.6 ft/sec) 200 cm3
bed of 70 to 100 mesh glass particles. The measured outlet S02 con-
centration plots were almost exactly identical, indicating that the
reaction characteristics of prior-calcined and bed-calcined particles
were the same.
In view of these runs it is
previously ground to proper size, can
large, continuously operating unit.
apparent that raw carbonate stone,
be used as a feed material to a
-------
- 38 -
4.4.3.
Effect of Fly-Ash
The effect of fly ash was evaluated by adding about 3-5% fly
ash to the bed of sorbent prior to an experimental run. The bed could
then be brought to temperature with either a trickle flow of gas or gas
flowing at full velocity.
The following two points were demonstrated in the fly-ash
studies: (1) coal fly-ash does not stick or agglomerate with lime
particles in the desulfurization reactor at l600°F, and (2) fly-ash
elutriates out of the bed at fluidizing velocities equal to or greater
than 2 ft/sec. Thus, fly-ash should have no detrimental effect on
the adsorption of S02 in the proposed fluidized bed reactor.
-------
- 39 -
5.
REGENERATION STUDIES
The regeneration of sulfated sorbent was investigated as a way of
reducing fresh stone requirements, and as a way of producing a saleable by-
product to reduce processing costs.
5.1.
Thermodynamic Considerations
in Sorbent Regeneration
The limestone regeneration technique investigated in this program
involves the reaction of partially sulfated sorbent with reducing gas
at high temperatures. Under appropriate reducing conditions, reactions
(10) and (11) shown below can be made to predominate. They yield
regenerated CaO and a concentrated S02 stream. Competing with these two
reactions are reactions (12), (13) and (14) which can also take place under
reducing conditions.
(10) CaS04 + CO - CaO + S02 + C02
(11) CaS04 + H2 -+ CaO + S02 + H20
(12) CaS04 + 4CO ~ CaS + 4C02
(13) CaS04 + 4H2 ~ CaS + 4H20
(14) 3CaS04 + CaS ~ 4CaO + 4S02
Reactions (12) and (13)areundesirab1e since they convert sulfate to sulfide
without liberation of S02, and they consume large amounts of reductant.
To study conditions under which the desired reaction would be
predominant, analyses were made of the equilibrium relationships of the
system represented by the above equation. Equilibrium data calculated by
Wheelock & Boylan (2) were used initially to provide a qualitative picture
of the CaS04 reduction system. These data showed that reactions (10), (11)
and (14)were favored by high temperatures while the other reactions
were favored at lower temperatures. Gas composition had an effect through
the CO/C02 and H2/H20 ratios. High ratios promoted a high S02 concentration
via reaction (10) and (11) but also promoted reactions (12) and (13) to an
even greater extent (i.e., the equilibrium for these reactions depends
on the fourth power of the CO/C02 and H2/H20 ratios). A ratio of 1/2 produced
mostly CaO and S02 in preference to CaS at most temperatures.
In order to quantify these effects of temperature and gas composition,
equilibrium compositions for both the solid and gas phases were calculated
for a series of hypothetical regeneration systems. These calculations were
made by a high speed digital IBM 7090 Computer using a program developed by
the Rand Corporation'(1Q,ll). The Rand program calculates the equilibrium
composition of a system - subject to material balance restraints - by
minimizing the free energy of the system. A partial summary of the results
-------
-40 -
obtained using this program are given in Table 11. It can be seen that
temperatures above l3000K (1880°F) are required for regeneration by CO or H2
of the sulfate to the oxide. At lower temperatures, the reaction proceeds
to the sulfide, while temperatures above l4000K (2060°F) offer no advantage
in terms of equilibrium conversion. It should be noted that complete conversion
of Ca804to CaS, without S02 formation, is predicted for reduction with
pure methane at: 2240°F.
Tab1,e 11
Equilibrium Reduction of CaS04
Inlet Gas Composi- Outlet Gas Comp os i ti on % of Reacted caS04
tion, Mole % Mole % Converted to
Temp, of C0 !!2 C02 H20 ~ ~ .92L ~ B2- CaO CaS
1880 5 0 10 0 85 0 15 0 85 0 100
1880 0 5 0 10 85 0 0 15 85 0 100
1880 10 0 20 0 70 0 30 0 70 0 100
1880 0 10 0 20 70 0 0 30 70 0 100
1880 15 0 30 0 55 0 45 0 55 0 100
1880 0 15 0 30 55 0 0 45 50 0 100
2060 5 0 10 0 85 4.0 14.4 0 81.6 95 5
2060 0 5 0 10 85 4.0 0 14.4 81.6 95 5
2060 10 0 20 0 70 7.7 27.7 0 64.6 95 5
2060 0 10 0 20 70 7.7 0 27.7 64.6 95 5
2060 15 0 30 0 55 11.1 40 0 48.8 95 5
2240 5 0 10 0 85 3.9 14.4 0 81. 7 94 6
2240 0 5 0 10 85 3.9 0 14.4 81. 7 94 6
2240 0 10 0 20 70 7.5 0 27.7 64.8 94 6
2240 15 0 30 0 55 10.7 40.2 0 49.1 94 6
2240 10 15 0 30 55 10.8 0 40.2 49.0 94 6
2240 ---- 100% CH4 ------ 0 33.3 66.7 0 0 100
Table 11 shows 802 concentrations ranging up to about 11 mole %.
Higher 802 concentrations would have been predicted had reducing gases been
specified with CO plus H2 concentrations in excess of 15%. The problem,
however, with using a more strongly reducing gas is one of maintaining the
regenerator in thermal balance. This problem is more easily explained
with reference to Figure 19 which shows a material and thermal balance for the
regenerator. Entering the regenerator is partially-sulfated sorbent at the
temperature existing in the desulfurizing bed; i.e., 1600°F. The sorbent's
temperature is raised to about 2000°F, and after regeneration is returned to
the absorbing bed. The fluidizing gas is a high temperature reducing gas
produced by the partial combustion of a fuel. At the outlet of the reactor,
the gas temperature has fallen to approximately that of the bed, while its
composition has changed as shown in Table 11.
The predominant reactions (10) and (11) occurring in the
regenerator at 2000°F are endothermic and remove a considerable amount of
Heat of Reaction, 6.H
Reaction Cal/g-mole @ 2060°F
(10) Ca804 + CO --- CaD + 8°2 + C02 43,400
(11) Ca804 + H 2 ----+ CaO + 802 + H20 50,800
-------
Q)
..-I
o
u e
..-/ I
.u 00
CtI-......
.0..-1
CtI CtI
..-/ U
"0"-'::
-,CtI
P-I-I
CtI Q)
..-I t::
CtI Q)
~oo
.u Q)
s:: ~
/I.1.
1-1
U) 0
U) .u
Q) U)
U :J
:<.0
/I.1 e
o
u
Sulfated Sorbent
from Absorber
1600°F
150
100
50
o
- 41 -
Figure 19
Thermal Balance Diagram-Regeneration
802 Rich Gas Stream
2000°F
Partially Regenerated
Sorbent to Absorber
2000°F
Regenerator
ZOOO°F
CaS04 + CO - CaO + SOZ + COZ
CaS04 + HZ - CaO + SOZ + HZO
Heat Loss to Surroundings
Combustor
Fuel
Air
Figure 20
Thermal Balance of the Regenerator
Mole % (CaO utilization) into
Regenerator less Mole % out
of Regenerator
. 10% Difference
x ZO% Difference
-----
---
-50
o
Thermally Deficient System
5
10
15
Regenerator Effluent, Mo1e% So
2
-------
- 42 -
heat from the system. In addition, sensible heat is removed from the
regenerator by the circulating sorbent which exits at a higher temperature
than it enters. The amount of heat so removed depends on the solids circulation
rate. Finally, even with a well insulated regenerator, some heat will leak
to the surroundings. To maintain the regenerator temperature, these thermal
requirements must be satisfied by a change in the enthalpy of the gas
flowing through the system. Assuming equilibrium conditfuns, the inlet gas
composition determines the outlet composition and therefore its enthalpy
at the specific temperature (i.e., ZOOO°F). The composition and enthalpy
of the inlet gas depend only upon the fuel used, the air to fuel ratio,
and the air and fuel temperatures. In effect then, the change in enthalpy
that will occur in the gas passing through the system will depend only on
the inlet conditions at the combustor.
These considerations are what limit the maximum SOZ concentration
in a commercial system to about 13%. Figure ZO shows the thermal balance
of a system, using coke as fuel, as a function of the S02 concentration
in the regenerator effluent (and thus on the CO + HZ concentration in the
inlet gas~ Curves are presented for variations in per-pass conversion of
the circulating solids (mole % CaS04 in entering solids - mole % CaS04
in existing solids). Above about 13% SOZ in the outlet gas, not enough
heat from the fuel is available for endothermic reduction and the system is
thermally deficient. In a real system where there is some heat loss from the
regenerator to the surroundings, the tolerable SOZ limit will actually be
somewhat below this value.
5.2.
Experimental Apparatus
The regeneration studies of sulfated particles were carried
out in an alumina-ceramic fluidized bed unit. The equipment is pictured
in Figure 21 and a schematic of the unit is shown in Figure Z2. Reducing
gas was made by blending HZ, CO, HZO, C02 and NZ in the desired proportions.
Methane could also be used if desired and SOZ was available for calibrating
the SOZ analyzer.
Additional provision was made for the introduction of
oxygen to the reactor effluent to combust any sulfur, H2S or COS formed
during the reductiun. The reactor itself was a Z inch i.d. alumina tube.
An earlier stainless steel reactor was found to corrode badly at the
2000°F operating temperature. An inlet section of packed alumina
cylinders, 1/8" x 1/8", served as a dispersing grid. The gasketed
reactor tube ends were placed outside of a small Lindberg furnace.
This kept the rubber gaskets at low temperatures so that they could
perform through numerous duty cycles. The outlet from the fluid bed
fed into an oxygen combustion chamber where any H2S, COS, or S that might
be formed could be combusted.and then into a cyclone filter and water
condenser.
As in the absorption experiments, sulfur dioxide measurement
was performed by an infra-red Beckman analyzer (IR 3l5A). The range of
the instrument was selected for O-15M% SOZ in the effluent. It was cali-
brated as outlined in the desulfurization kinetic studies (Section 3.1.3).
-------
- 43 -
Figure 21
Fluidized Bed
Regeneration Unit
-------
Thermo-
couple
~
~ ~ _I
eu ;j '"0
s:: ,....f OJ
'"' ""' I'i1
~ . M I
~ 'g~ J'
rr.:I ;j tJ
,....feu
-
.j>-
I.R. Analyzer
and
Recorder
Flow Diagram for the Fluidized Bed Regeneration Unit
-------
- 45 -
5.3.
Experimental Procedures
In the regeneration studies ~ different reducing conditions
were first employed to determine operating conditions for achieving high
SOZ effluent concentrations. Reductants were CO~ HZ and CH4~ as obtained
from commercial gas cylinders. Limestone particles were first sulfated
by reaction with simulated flue gas. Direct reduction on commercial
calcium and magnesium sulfates was also carried out in these preliminary
studies.
To determine the cyclic capacities of regenerated sorbents, a batch
of material would be subjected to repeated cycles of adsorption and regenera-.
tion. In these cyclic experiments~ both the adsorption and the regeneration
were run in the regenerator unit so as not to expose the particles to un-
necessary temperature variations or air. Adsorption was performed first,
following the desulfurization procedures outlined in Section 3.Z. Upon
termination of the absorption phase~ nitrogen alone was passed through the
bed to keep it fluidized while the temperature was raised for regeneration.
The reductant mixture was then intorduced and maintained until regeneration
was complete. The SOZ effluent concentration was continuously monitored
during a cycle. Regeneration yielded a reactive material for reuse as a de-
sulfurizer. Weighings of the discharge were performed after each complete
cycle as a check on loss of material through attrition and of actual sulfa-
tion changes as obtained from recorder charts. Differences of only l-Z% were
recorded on average~ well within experimental error~ which indicates that no
systematic error was occurring.
5.4.
Regeneration-Definition Studies
5.4.1.
Reduction of CaS04
Initial experiments with CaS04 (i.e.~ anhydrite) demonstrated
that SOZ was generated by a reducing gas containing CO or HZ' The SOZ
effluent concentrations obtained for 10 and 15 mole % reducing gas at
various temperatures are compared to equilibrium predictions in Figure Z3.
Hydrogen appeared to produce higher SOZ concentrations than CO~ but not
enough data were obtained to reach a definite conclusion on this. In
the partial combustion of coal or hydrocarbons to produce the reducing
gas ~ normally a Z/l ratio of (CO/C02) / (H2/HZO) is dictated by equilibrium
for (CH2)n fuel. Thus~ mostly CO would be present in a commercial
reducing gas and any small difference in reactivity between CO and HZ
can be neglected.
-------
~
C1J
.....
~
..
N
o
U)
.j..J
~
C1J
:::I
.....
4-1
4-1
~
Figure 23
Sulfur Dioxide in Regeneration Effluent
12
Reducing gas: 10% CO-H2/20 C02-H20/ba1 N2
Reducing gas: 15% CO-H/30 COZ-H20/bal N2
10
8
IS!
~
8
6
/~ ---es/
/ E;)
~ I. .
I Predicted
. Equi1.
I
I
I
I
I
I
I
I
I
I
,
U(ft/sec)
0.8 1.2
o IS)
&
e~
.
4
Reductant
2
H2
CO
.
IS!
o
/"
/
8/
~
/
I
I ,
I
I
I
I
I
I
I
.
1.6
8
.
Calc. predict zero equi1 at 1860°F 25/50
Mesh CaS04 particles Ho = 6 inches
ISIJ
1---
~
e
.
Predicted
Equi1.
.j::--
0\
o
1800
2000
2100
2200 1800
1900
1900
2000
Reducing Temperature, of
2100
2200
-------
- 47 -
Reducing conditions were kept mild by the use of a 1:2 ratio of
reductant to its oxidized species in order to inhibit calcium sulfide
formation during reduction.
Reduction of partially sulfated limestone and dolomites also
was accomplished readily with CO/COZ mixtures. A typical chart record obtained
of a batch reduction of a fluid bed of sulfated limestone is presented
in Figure 24. With good gas-solids contacting, the 802 level in the
effluent was found to depend on reductant concentration, as in the case
of anhydrite reduction. The maximum level was quite independent
of the amount of 8C2 absorbed by the different stones-
Figure 24
Typical Record of Regeneration Effluent
9
8
7
iN! 6 "0
Q)
.j.J
.j.J ~
s:: 5 ell
~ .j.J "0
U) Q)
.-I .j.J
4-1 4 ~ ~
4-1 C\I
~ .-I .j.J
1'1:, U)
s:: 3
-M N ~
a
C"~ U .-l
a 2 I'r-<
U)
N a
:z; U
1 ~j
0
I -L-
0 20 40 60
Time, Minutes
Reducing Gas
20% CO/10% CO
T .. 2000°F
25/50 Mesh Par~s
80
100
-------
- 48 -
It is worth mentioning that the time lag, th, for holdup in
the system from feed to recorder was about 1.5 minut,esat U=2 ft/sec.
However, on given occasions, it was noticed that if the reductant flow
were stopped3 802 was observed in the effluent at the plateau level up
to7.th times before falling to zero concentration, which indicates a
mixing or desorption control mechanism.
An increase in the ratio of C02/CO or H20/H7 from 2:1 did not
increase the 802 effluent concentrations. However 3 it should be pointed
out that approximately 90% efficiency of the reducing species to produce
802 was already occurring. Lowering the C02/CO or HZofH2 ratio~ on the
other hand3 did decrease the S02 level. With pure H2 plus inert NZ3
only 25-30% efficiency of reductant to produce S02 was achieved. CaS
formed~ as well as H2S, The ratio of CaS/ (CaD + 502) in the effluent
was determined to be about 5:2.
With particles larger than 25/50 mesh~ difficulties encountered
with fluidization agglomeration (discussed in Section 5.4.3.) produced
low 802 concentrations and caused uncertainties in the analysis of
regeneration of these sized particles. However, experiments were made
in a 5/8" Ld. packed bed and these indicated that regeneration rates
with particles up to 2.5 rom in size are comparable to rates with the
small diameter stones.
Reduction with methane did not proceed readily. Metha~, in
the presence of C02 and H20 did react with sulfates material to evolve
some 802, but the 802 concentration was significantly lower than that
obtained with CO or H2 (Table 12). The results with a CH4/N2 feed
produced about one 802 molecule per methane molecule, in contradiction
to thermodynamic analysis which predicts no 802 at equilibrium, but
this is still unsatisfactory for sulfated sorbent regeneration.
Table 12
Sulfur Dioxide Concentration From
Methane Reduction of Calcium sulfate
Gas Flow = 2 ft/sec
Mole Ratio = 2:1, Balance N2
% 802 in
1950°F
1900°F
Effluent
2000°F
2050°F
% Me thane
5
10
1.5
2.5
2.3
4.5
2.5
4.3-5~0
6.0
-------
- 49 -
With all reductants, the 802 concentration in the reactor
effluent was found to have about a first order dependence on the inlet
reductant concentration in the range 5-15%. With CO or H2' reduction
appears to be equilibrium limited, as may be deduced from the equilibrium
comparisons of Figure 23 and the fact that a 802 concentration plateau
is maintained until exhaustion of the sulfate in the stone (Figure 24).
The low 802 effluent concentrations with CH4 reduction are
too low for practical use as feed to a sulfuric acid plant. However,
the 8 to 10% sulfur dioxide effluent levels that are obtained with CO/HZ
reduction ~ high enough to serve as a feedstock for by-product processing
in a H2S04 plant.
5.4.2.
Sorbent Reactivity After First Reduction
As shown in Section 4.2.3., magnesium oxide does not react
with sulfur dioxide in a flue gas to any appreciable extent. This is
true even of MgO formed by reducing MgS04. Thus, the fact that magnesium
sulfate can be reduced at temperatures as low as l400°F cannot be profitably
utilized in a regenerative desulfurization system.
However, calcium oxide does exhibit reactivity for sulfur dioxide
after its formation by the reduction of calcium sulfate. In fact, flue
gas desulfurization by once-regenerated particles proceeded at greater
rates than that obtained with freshly calcined limestone. A comparison
is presented in Table 13.
Table 13
Reaction Rate Coefficient of Once Regenerated Materials
Desulfurization, l600°F; U = 2 ft/sec; 25/50 Mesh Particles
Regeneration Temp., 2000°F; Desulfurization in ceramic unit
X = Fraction of CaO utilized
Material
Reaction Rate Coefficient (sec-l)
Unregenerated Regenerated
CaS04 Anhydrite
X = 0.50
9-22
Dolomite (1337)
X = 0.52
X = 0.56
X = 0.72
8
1-2
/"V 0
45-50
41.5
16
Limestone (1359)
X = 0.25
X = 0.47
4.5
-./0
30
3.5
-------
- 50 -
5.4.3.
AKKlomeration
5.4.3.1.
Fly Ash Effects
Experiments were made to determine the effect of fly ash and
sorbent agglomeration during high temperature regeneration. Experimentally,
the procedure was to add about 3-5% fly ash to a bed patch prior to an
experimental run. The bed was then brought to temperature, about 2000°F,
with either a trickle flow of gas, or gas flowing at full velocity.
Agglomerated lumps of sorbent and fly ash (and fly ash alone)
were found and these caused uneven fluidization and thus subsequent
uneven, lower S02 concentrations in the regenerator effluent. The lumps
clogged the reactor, as found by examination at the conclusion of a run.
The lumps were fragile to physical shock at room temperature, but their
strength characteristics at high temperature are unknown.
The discovery of high temperature particle-fly ash agglomeration
might create problems in an actual commercial regenerator. If the fly-
ash is transported out of the reactor before temperature (and agglomeration)
can be reached and is not returned by the cyclones, no serious consequences
would develop. If partial combustion of coal is used to produce the
reducing gas, the fly ash will already be at temperature and agglomeration
difficulties could occur.
5.4.3.2.
Self-Agglomeration
Without Fly-Ash
It has been definitely established that self-agglomeration of
sorbent particles can occur at temperatures near 2000-2100°F. This
phenomenon appears possible without the necessity of iron or silica
agents as fluxing materials (i.e., fly-ash), although the presence of
ferrous materials enhances agglomeration. During one experiment
investigating reduction characteristics, calcium sulfate particles
(pure material) formed a semi-solid cylinder from the great majority
of particles in the batch. Other experiments produced smaller agglomerated
lumps. These lumps are very fragile at room temperature and fall apart
. at touch, similar to the fly-ash agglomerated lumps. It appears that
this agglomeration problem becomes more acute as the temperature is raised.
Some problems could occur in fluidization in the regenerator,
based on this evidence, which would result in gas by-passing, low reductant
utilization, and low S02 concentrations in the effluent gas. However,
providing sufficient fluid bed agitation, whether by mechanical stirring
or preferably by better fluidization, should alleviate the type of
agglomeration as a potential problem.
-------
- 51 -
5.4.4.
Possible Sulfite Formation and
Thermal Regeneration
Thermodynamic analysis indicated that thermal decomposition
of calcium sulfate begins near 2450°F while the sulfite begins to decompose
near l200°F. Apparently, however, either a molecular form of CaS03
or a strongly sorbed S02 species can exist to much higher temperatures in
lime. This was shown by passing a gas of S02 (1%)/N2/C02 (but without
oxygen) through a fluidized bed of lime at l600°F. The gas was desulfurized
efficiently until about 40% of the stone was utilized. Undoubtedly some
disproportionation of CaS03 to CaS and CaS04 also occurred. At 2000°F,
a feed stream of pure N2 produced an effluent stream of about 14% 502.
About two thirds of the sorbed sulfur compounds were stripped by this
procedure, indicating that a sulfur dioxide/sulfite specie was present
in large amounts.
Although formation of CaS03 is preferred from the standpoint
of regeneration, this phenomenon would be difficult to utilize. Efficient
combustion demands that excess oxygen be present since sub-stoichiometric
(and near-stoichiometric) combustion produces H2S, cas and unburned carbon.
Under excess oxygen conditions, the formation of CaS04 would predominate.
5.5.
Cyclic Performance
5.5.1.
Reactivity of Cycled Sorbents
Particles of limestone N-1359, dolomite N-1337, and calcium
sulfate anhydrite were used in a cyclic sequence series of experi-
ments on flue gas desulfurization and regeneration. All three materials were
found to retain sorption reactivity upon cycling. The data for lime
N-1359, for absorption are presented in Figures 25 and 26. Small correction
factors have been applied to the data in Figure 26 for variations in
bed height and gas velocity so that all the data refer to the same initial
conditions. Note that the reactivity first increases with cycling,
then decreases with further use, but even after 8 cycles (approximately
115 hours in the hot fluid bed) the capacity of the stone was still equal
to the initial capacity. In contrast, the desulfurization reactivity of
anhydrite dropped from the first cycle towards 30-50% of the initial
activity after 4-5 cycles.
The good maintenance of reactivity upon cycling of the limestone
suggests the use of such material in a continuous recycle system. As
expected, continued cycling did eventually cause a decline in capacity
as "dead-burning" of the stones presumably occurred from exposure to
the regenerator temperature of 2000°F. It is known that dead-burning
begins to occur at about 1900°F (12). Dead-burning decreases reactivity
by decreasing the porosity of the particle as it rearranges into a more
packed material. This limits diffusion paths into the particle and
provides a limit to the reaction rate.
-------
"0
Q)
+J
U
Q)
I-t J»
I-t +J
o .,..,
u u
....f',3
tIS Q)
:>:>
o
8 UJ
3HB
N"O
o t::
tI] II]
~+J
A.!:
N bO
.,..,
~ Q)
::r::
ft
t::"O
o Q)
.,.., P=1
+J
tIS +J
N t::
.,.., «I
,....!+J
.,.., fIJ
+J s::
:::> 0
U
o
c'3~
50
40
30
,....!
II]
6
8
Q)
~
N
o
en
20
10
~
- 52 -
Figure 45
Breakthrough Curves on Cycling
Cycle
1 87
524
3
o
0.6
6
25/50 Mesh Limestone
(1359); Sorption 1600°F
Regeneration 2000°F
U=2 ft/sec,
Ho=8 in for Cycle 1
0.1
0.2
0.3
0.4
0.5
Fraction CaO Utilization
Figure 26
Desu1furization Reactivity Upon Cycling
0.5
0.4
0.3
0.2
0.1
Cycle Number
-------
- 53 -
The initial increase in reactivity upon cycling is speculated
to be due to a change in the physical structure of the particle. Upon
calcination, a porous macroscopic structure is formed, while molecular
vacancies are formed by evolution of C02 from the carbonate. Time is
required for reorientation-expansion of the molecular spacing, presumed
necessary in order to accommodate a sulfate group. However, upon
regeneration and evolution of 802' the molecular spacing remains as
reoriented to provide a more reactive, accommodating particle. The
opposite phenomenon of dead-burning SOon overshadows molecular reorientation,
however, and the capacity begins to decrease. Calcined dolomite also
exhibited this phenomenon of initial activity increase (Table 13). On
the other hand, calcium oxide generated from anhydrite had an initial
large reactivity which decreased rapidly with cycling as dead-burning
occurred. In this case geometric accommodation already was present
and no molecular reorientation was possible or necessary.
From this speculation, one would predict that the important
parameter limiting cyclic capacity is exposure time to the high (2000°F)
regeneration temperatures, rather than cycle number per see In the
experiments with the lime (1359), the sorbent received exposures of
15-20 hours to dead-burning temperatures, including experimental adjustments,
etc., but only 4-6 hours of actual reduction time were used. In an
efficient closed-circulation system, it would thus seem that the number
of cycles equivalent to 15 hours of high temperature exposure should be
possible before there is a loss in sorbent reactivity below the initial
rate.
Anhydrite, as well as limestone, might be suitable for use in
a recycle, regenerative system. The desu1furization rate for the anhydrite
is high enough initially that the faster activity decrease can be tolerated,
and still the average sorbent utilization would be as high or higher than
lime (1359). This would need to be verified more fully, however, if
anhydrite or gypsum were to be used as a sorbent.
5.5.2.
Reduction Efficiency of Cycled 80rbents
The time dependence of the effluent 802 concentration from
batch reduction had a typical history, depicted earlier in Figure 24,
for all the runs of a series of cycle experiments. The maximum 802
plateau level was maintained near 9% for each cycle as Figure 27
indicates.
-------
- 54 -
ri9;ure 27
Effluent From Regenerator-Cyclic Performance
I
T
I
I
10
r-I
Q)
:>
Q)
~ 8
;:j
CIS
Q)
~
CIS 6
r-I
p..,
I
N
O 4
tI)
~
2
o
I
%
I
T
~
1
C02/CO
10% CO
2000°F
2 : ~ B a1. N 2
5
6
7
8
9
10
Cycle Number
Regeneration was complete each cycle. Graphical integration of
time records of S02 in effluent streams found fair-to-good calculated material
balance between absorption and regeneration (1-10% difference in total S02
between the two integrations). Weight balances at the conclusion of a cycle
were always good (0-1% variation) and provided a check on completeness of
regeneration. Calcium sulfide did not build up in the particles and
remained low at the 1-2% level (Table 14). This might be expected as 02
will oxidize any sulfide to S02 and CaS04 during the desulfurization cycle.
Table.l4
Sulfide Content in Limestone From Cyclic Regeneration
Reg. Temp = 2000°F, C02/CO = 2:1, CO% = 10%, Bal. N2
% Sulfide
Cycle Concluded
o
0.75
1.7
1.45
Start
1
5
10
-------
- 55 -
5.5.3. Attrition of Cycled Sorbents
Attrition of lime 1359 was found to
Figure 28 illustrates. Attrition of dolomite
large on the first cycle (50%), and was still
cycle as seen in Table 15.
be small upon cycling as
1337, however, was quite
quite large on the succeeding
Figure 2R
Physical Integrity of Lime (1359)
1-1 0.4
..c:
-
"0
co
Q) 0.3 U = 2 ft/sec
..c:
1-1
Q)
:> 0.2
0
~
0.1 .
.
o
~
0
"ri Q)
+Jr-f 4
co C)
"ri»
I-It.)
+J
;:j 1-1 3
r-f Q)
~p..,
I
C"O
0 co 2
"ri Q) .
+J..c:
"ri 1-1
1-1 Q)
+J :> 1
<0
0
~+J 0
1 2 3 4 5 6 7 8 9 10
Cycle
-------
Stone
Dolomite 1337
Cycle 1
Cycle 2
Dolomi te 1337
Cycle 1
Cyc Ie 2
Lime, 1343
Cycle 1
Lime, 1359
Cy-cle 1
Cycle 2
Cycle 3
56 -
Table 15
Attrition During Cyclic Use of Sorbents
25/50 Mesh Particles, Sorption at
1600oF, Regeneration at 2000°F, in-
situ regeneration unless otherwise
noted.
Cycle Time
(hr)
% CaO Utiliz.
@ SO 2 oul;1Ji:F=o, 2
Gas Vel. Loss Initial Bed
(ft/sec) Overhea,d,(%) Hei~ht (cm)
7.5
5.5
2
2
52.5
72.0
53
2.9
10
4.6
Preattrited 6 hrs in glass column, V = 2.45 ft/sec, 26% loss overhead
[S02]in= 6000 ppm
2
2
13.7
18.1
2.9
2.9
8.5
22.8
2
- Exploratory experiments, [S02]' - 1%
1.n
5
Inappropriate
Flue Gas
27
27
3.3
2.5
4
3.3
2.9
21.7
3
7.5
10
1.7
o
19.3
15.2
15
17
15
14
-------
- 57 -
The data on the 1359 lime show its attrition to be small.
Initial rounding of some sharp particles appears to have resulted in
elutriation and loss of smaller particles, but this stone is basically
quite stable. '
The direct use of the attrition data generated in this
laboratory system for the design of large systems is not recommended.
It is known that attrition from particle collisions with walls is important
and probably dominant in determining particle size. In laboratory equipment,
such as the present apparatus, slugging occurs, causing particles to be
carried into the top wall and to brush against the side walls. In large
units slugging is impossible; instead bubbles break in the open disengaging
sections. Attrition is then mainly due to particle-particle collisions.
Wall interaction is less important as the surface wall/volume ratio is,
much smaller. Thus, attrition may be appreciably less in large systems,
operating at the same velocity, than obtained in our experiments although
qualitative rating of stones from the present data should be valid. Thus,
the N-l337 dolomite would still be expected to attrit much more readily
in a large commercial unit than would the N-1359 limestone. The amount
of attrition obtained with the dolomite in the laboratory unit is
unacceptably high. However, at this point it is not possible to completely
rule out the commercial use .of this sorbent, or other sorbents on physical
grounds without additional testing in larger equipment.
-------
- 58 -
6.
PROCESS DESIGN CONSIDERATIONS
Of the var~ fluid bed processing alternatives that were researched
in this program, only the use of coarse, untreated, sorbent particles,
preferably with regeneration, was considered to offer potential applica-
tion to commercial flue gas desulfurization. This section discusses the
design of a coarse particle fluid bed desulfurization system.
6.1.
Desulfurizer:
Conventional Power Plant
6.1.1.
Sizing:
Reactor
In order to insure efficient desulfurization at low power loads,
the design calls for a l600°F bed at 90% boiler load. At 60% load, one
could expect a temperature drop of about 100°F. Stone would need to
be fed more rapidly to compensate for reactivity loss despite longer
gas residence time.
Figure 29 shows the relationship between the reactor bed
diameter and gas velocity at l600°F for a system treating the flue gas
from a 200 MW coal-burning power plant. The maximum diameter of a fluid
bed at today's state-of-the-art is 55'-60'. Using present technology,
a bed of 50 ft. diameter could be utilized in a desulfurization process.
With such a bed, gas would flow through the reactor at a superficial
velocity of 15 ft./sec at full load.
Figure 29
Reactor Size-Gas Velocity
125
75
Basis: 200 MW power plant
(90 x 106 ACFH)
~ 100
4-<
n
...
OJ
.I.J
OJ
S
('(I
"0-4
~
'0
OJ
I'Q
Present maximum fluid
bed size
/ / / / / / / I / /\ / /
50
r-!
('(I
s::
,..
OJ
.I.J
'S::
I-f
25
o
o
5
10
15
20
25
Fluidization Velocity, ft/sec
-------
- 59 -
As the bed would operate near l600°F, insulation would be required
to limit heat loss. By using internal insulation, carbon steel would
have enough structural strength to be used for the shell body.
Ducts to connect the required external desulfurizer and the boiler would
be insulated to reduce heat loss. Erosion by fly-ash, accelerated by the gas,
would be kept low by limiting the gas velocities to near 100 ft/sec. This
means perhaps a 17' x 17' duct for a 200 MW boiler.
6.1.2.
Sizing:
Limestone Particles
For small particles the entrainment velocity is about 70 times
the incipient fluidization velocity (3); however, as the particles become
larger, different hydrodynamic effects occur and entrainment takes place
at only 10 times the incipient fluidization velocity (13). A more severe
restriction on velocity limitations and particle size distributions is
encountered, then, with large particles.
Figure 30 presents a graph of the diameter of calcined limestone
particles versus gas entrainment velocity. To avoid entrainment at the
15 ft/sec superficial velocity, a minimum particle diameter of 110~ was
specified. A maximum diameter of 5000~ was selected to obtain a
range o~ sizes for good ~luidization which could be obtained from a
ri~ure 30
Particle Entrainment Velocity
50
Conditions: Calcined porous particles
fluidized by flue gas @ l600°F
40
u
w
00
"
~ 30
~
~
~
~
~
u
0 20
M
W
~
00
~
~ 10
,
o
o
500
1000
1500
2000
2500
Particle Diameter, ~
-------
- 60 -
limestone supplier. Calculations for sulfated particles using the root-
~an-squared average diameter of 2450~ give an incipient fluidization
velocity of 4.3 ft/sec - using Leva's correlation (14). The range of
particle sizes selected, 1100-5000~ , is a more narrow relative range of
sizes than employed in fluidization processes such as catalytic cracking.
A large range is normally necessary to provide good uniform fluidization;
however, the above range should provide good characteristics for these larger
particles. The reactor design calls for an enlarged-diameter, disengaging
section to reduce the gas velocity by one-half, to ensure limitations on
the carry-over of small particles.
6.1.3.
Fluidization Requirements
Modern boilers use thin, flat-walled sides. With these walls,
the pressure drop through the boiler is limited by the pressure differential
that can be tolerated between the atmosphere and the inside of the boiler
without the occurrence of an explosion or expensive reinforcement of the
buck stays. The use of forced and induced draft fans can keep the average
pressure inside the boiler near atmospheric; air is forced into the boiler at
pressures above atmospheric and flue gas is drawn out below the outside
pressure to reduce the maximum differential across the boiler walls. A
maximum pressure drop of about 20" H20 through the fluid bed might be
tolerated in an integrated boi1er-desu1furization system in such a balanced
draft case. Depending on the sulfation level, which determines the density
of the particles, a bed height of 15-24" might be tolerab~e.
To insure uniform gas distribution without channeling, a grid
pressure drop of 30% of the bed pressure drop is needed (15). In addition,
gas bubble sizes must be minimized to limit gas by-passing of the particles.
Contacting efficiency would otherwise limit the desulfurization that could
be obtained. Large grid openings lead to large bubbles according to
the relation (2):
v ~ G6/5
B
(15)
Here VB is the bubble volume and G is the volumetric gas flow rate through
the opening above the minimum fluidization velocity. Now G is proportional
to a grid design with numerous small holes as gas inlets.
A flat plate grid with I" high imbedded buttons, containing small gas
inlet holes, was assumed in this design. These buttons are similar to
those being investigated by the Pope Evans and Robbins Co. for use in a
fluid bed packaged boiler (16). Although the use of such a grid would
insure the formation of onlY-small bubbles initially, the bubbles could
grow rapidly. The maximum bubble size is a function of particle size (3) and
for the present large particles it is approximately the bed diameter. In
even the shallow beds visualized, bubble growth is so rapid that bubbles
could be several feet across at the top of the bed. To limit the bubble size,
vertical pipes might be employed in a similar manner to that used in other
large particle processes (12). The increase in vertical surface area would
-------
- 61 -
serve to reduce the equivalent effective bed diameter - and thus reduce
bubble size, as bubbles only can grow to the containing vessel
diameter.
6.1. 4.
Sorbent Selection
Obviously, a sorbent must be chosen for a fluid bed process
which does not attrit too easily in a fluid bed. As shown in Section 4.4.1.,
the N-1359 limestone has gaad stability. Consequently, it was selected
for the process designs even though its initial reactivity was lower than
any of the ather sarbents evaluated in this pragram (see Figure 10).
It was felt that if the stane with the lowest rate constant measured from
a variety of&ones could desulfurize flue gas efficiently in the system
as designed, then most other stones would also adequately desulfurize flue
gas in this system. Naturally, the use of a more reactive stone, with
comparable attrition characteristics wauld result in a lower process cost
than that develaped far the use of N-1359 sarbent.
The rate constant far 2380-3360 ~ and 1000-119~ particles
of stone 1359 are given in Figures 9 and 10. The actual rate constant that
should be used in calculating desulfurizatian is an average aver the
residence time distribution and particle size. An expanential residence
time distributi.on has been used to calculate the effective rate constant
for a given bed hold-up time (see Appendix VI for details).
A 802 removal .of appraximately 90% was chasen as the
design criterian.. Far single-pass limestane use, a maximum .of 87% .of the
S02 could b.e remaved at rated laad with 29% utilized particles. A 3/1
molar stoichiometric ratio .of CaD to S02 wauld need ta be added to a 2 ft.
bed - assuming a rate constant average .of 900 min-l at the designated _gas
velocity .of 15 ft/sec. The upper limit for desulfurizatian .of 87% assumes
ideality - gas plug flow with no gas by-passing in the fluidized bed.
Real svstems wauld have bubbles and some gas by-passing~ desulfur~~~~io~
efficiency would thus be less than calculated.
In a regenerative pracess, appraximately 92% desulfurizatian
wauld be obtained in a 15 inch bed (settled depth) under ideal gas
cantacting and 17% average calcium .oxide utili~atian. The average
rate canstant is 1800 min-l for this case. The pressure drop
thraugh the 15 inch bed wauld be much lawer than in the single-pass, 24
bed. The regenerative pracess wauld be more canservative, and fairly
safe, with respect ta structural bailer strength.
inch
One might expect an incremental laad on the boiler's cyclane-
precipitator dust callectian system due ta attrited limestane. A nine ta
twenty-seven percent increase is estimated far the single-pass system using
the data of Sectian 5.5.3. With the regenerative pracess, the dust
laading increase in the bailer wauld depend an the limestane additive
rate, which is determined by the length .of time the sarbent effectively
-------
- 62 -
maintains its reactivity. Based on about a 1:3 CaO to S02 mole ratio
make-up stone addition, a dust loading increase of 3% is estimated. It
would be expected that the cyclones and electrostatic precipitators normally
installed in a boiler should handle this limestone dust without modification.
6.2.
Regenerator
The fluidized bed regenerator would be much smaller than the
desu1furizer. A reducing gas with a (C02 + H20) :(CO + H2) ratio of 2:1
would be used, as discussed in Section 5. This gas (at 2000°F) could be
obtained by partially combusting coal or methane with air - pure, or
diluted with flue gas. The cheaper coal should be used providing the
resulting fly-ash proves no problem. An assumed gas velocity of 11 ft/sec
at rated load sets the internal diameter of the regenerator at 10.8 ft.
A high velocity is required to maintain agitation and decrease agglomeration
problems.
Cyclones would be provided to capture and return any entrained
or attrited particles. They would also serve to protect subsequent
equipment from particle fouling. Pneumatic tubes would probably be used
to transport the material to and from the desulfurizer downcomers to the
regenerator since mechanical feeders would tend to attrite the particles.
The grid is assumed to be a firebrick arch as used in some present-day
limestone kilns. The inlet openings could be straight through holes, as good
contacting is not as necessary to obtain conversion in the regenerator as
it is in the desulfurizer. This is because the reduction reaction is fast
and at equilibrium. A bed depth of 3 feet is estimated to be sufficient
for regeneration. The average composition of particles being returned
to the desulfurizer would be about 5 mole % sulfate.
6.2.1.
By-Product Processing
The nine percent sulfur dioxide concentration from the regenerator
is high enough for use in a contact plant to produce sulfuric acid.
Clean-up of the gas stream for use in a catalytic contacting plant would
follow the outline presented by Wheelock and Boylan (18) for a process
converting anhydrite to sulfuric acid. A standard sulfuric acid plant,
except for the sulfur burning section, would be employed to obtain 98%
sulfuric acid for by-product sale.
Schematic Design:
Power Plant
Conventional
6.3.
The design considerations presented in Section 6.1. and 6.2. are
incorporated into the schematic presented in Figure 31. This diagram presents
the salient features in the integrated system but it is not drawn to scale.
-------
Figure 31
Limestone Fluidized Bed Desulfurization Process-Conventional Power Plant
REGENERATION UNIT WITH BY-PRODUCT PROCESSING
Dry Gas (S02' 02'
C02, N2 to catalytic
conversion and absorp-
tion sections of acid
plant
Dry in g
Tower
Mist
Elec-
tro-
static
Precip
Compressor
Air
Force
Draft
Discard
Limestone
Makeup
~
....
Coal
INTEGRATED DESULFURIZER
Boiler
/
-
----
'"
....
Desulfurizer
dp = 2.4 rnrn
+
~ ~ Supe
=
=
Pre-
eater
Air -
Forced
Draft
,
Electro-
static
~recip-
f?
fl
()
'"
w
Induced
Draft
-------
- 64 -
6.4.
Future Process Variation:
Fluidized Bed Combustion
The use of fluidized bed combustion is currently being researched
in the United States and in England as a means of raising steam in
packaged and utility sized boilers. In fluidized bed combustion, advantage
is taken of the excellent heat transfer coefficients obtainable in a fluid
bed (i.e., 50 to 100 BTU/hr/ft2/oF) to reduce high temperature boiler
tube requirements (and costs) and to maintain bed temperature at about
1500 to l800oF. Most of the fluidized bed consists of coarse inert
material, with the fuel (usually coal) occupying only a small portion
of the total bed.
Research has been directed at injecting finely pulverized
limestones and dolomites into fluid bed combustor to effect desulfurization
of the gases. With this type of system, sorbent utilizations have been
low (i.e., 25-30%) and several times stoichiometric sorbent addition rates
have been found to be required to provide high levels of S02 removal.
An interesting variation of the fluid bed combustion process is to combust
coal in a coarse lime particle bed. Fly ash would be entrained out of
the bed by the gas, leaving the sulfated lime, which can be sent to a
regenerator with less fear of agglomeration. Small diameter particles should
be used because smaller particles and a wide size distribution offer
better fluidization conditions, and thus less tendency to agglomerate (see
Section 5.43). Present analysis indicates that gas velocities near 3 ft/sec
would be used in the fluidized bed combustor for economic savings (19);
then particles with davg = 1000 or smaller could possibly be used without
entrainment difficulties. The Consolidated Coal Company has shown that
such a coarse limestone fluid bed combustion system can work, that S02
capture is good, and that the sulfated sorbent can be reductively regenerated
(20).
Design problems would be simplified in such a desulfurization scheme,
compared to a conventional plant fluid bed desulfurizer, since the fluidized
bed is also the combustion unit. Duct work and grid problems would be
eliminated or substantially reduced. Stable stone particles would, of course,
still be needed. As described for a conventional plant, high efficiency
desulfurization should occur in a cyclic, regenerative scheme with the 2-3 foot
bed height presently envisioned for fluidized bed combustion. The regenerator
would be basically the same as that described for a conventional plant scheme.
Areas of potential problems, such as fly-ash separation and
desulfurization efficienc~would need to be investigated. Combustion
efficiency would be important. Low combustion might cause worse problems
than with coarse particle combustion as the unburned carbon recycle load
could increase; however, high combustion efficiency with fine pulverized
coal might eliminate much of the costly cyclone system for carbon recycle
presently expected to be needed in fluidized bed combustion.
-------
- 65 -
6.5.
Possible Process Innovations:
Sulfur Production
Sulfur would be a more desirable by-product than sulfuric acid
because of easier transportation, storage, and sales possibilities.
It is therefore of interest to determine how the regeneration procedure
might be modified to produce sulfur rather than sulfuric acid.
One method of producing sulfur might be to use a coke reduction
process after concentration of the SOZ from the regeneration by liquefaction.
The COS formed over coke could be catalytically decomposed to sulfur.
This mode of operation, however, appears too expensive and involved for a
utility to operate. Sulfur could be produced by operating a modified
Claus Process on the regenerator off-gas. In such a process, HZS would
be added to the regenerator off-gas in sufficient quantities to react with
S.o2 according to the equation: .
SOz
+
ZHZS -.-. 3S
+
2HZO
( 16)
The reaction could be operated at about 600°F over an alumina catalyst.
The sulfur formed would be condensed downstream of the catalyst. Most of
the sulfur would be collected as product wi th the remainder used to produce
the HZS added to the regenerator off-gas. This HZS could be made by steam
reforming natural gas together with S as proposed by Princeton Chemical
Research (21).
An alternate route to sulfur might be to add CO or HZ to the
regenerator off-gas in sufficient quantities to reduce the SOZ to
sulfur:
CO and HZ
much were
SOz + 2CO ---+ 2C02 +
addition would have to be very
to be added, COS and HZS would
S
(17)
carefully controlled; if too
be produced:
S
+
co --.. COS
(18)
Ryason and Harkins (Z2) have researched a direct reduction technique for
controlling SOZ (and NOx) emissions in flue gas. They found a copper
oxide on AlZ03 catalyst to be very effective at temperatures near 10000F
for promoting the desired reactions.
Still another method would involve the lower temperature reduction
of CaS04 to CaS instead of CaO and SOZ' The CaS might then be treated with
steam and COZ as Squires (Z3) has suggested to yield HZS:
CaS + COZ + HZO ~HZS + CaC03
The HZS could then be partially combusted to yield a Z:l
which would be converted to sulfur in a Claus Process.
(19)
mixture of HZS:SOZ
-------
- 66 -
7.
ECONOMICS OF FLUID BED LIMESTONE DESULFURIZATION
The cost to desulfurize flue gas from fossil-fueled power
plants with a limestone-basedt fluidized-bed process has been estimated
for a 200 MW plant using both the regenerative and the single-pass
limestone processes and for a 1000 JvM plant using the regenerative system.
Comparable economics are also presented for direct injection techniques for
desulfurizationtand rough estimates of fluid bed combustion/desulfurization
costs are given.
7.1.
Basis for Cost Estimates
Details as to equipment size, material flow rates and costing
are presented in Appendix VIII. Equipment was sized (Section 6) to handle
full-rated capacity power loads. Summarized investment costs and opera-
ting charges are presented in the text. The base case is assumed to be
a coal burning plantt using 3.5% S coal at the rate of 3000 ton coal/yr/MW
capacity at 90% load or 8000 hrs/yr. Flue gas desulfurization in the
fluid bed of 90% was designated at a 90% load factor. Eighty-five percent
conversion of the sulfur in the coal to by-product H2S04 was assumed in
regenerative systems (1-2% loss in the sulfuric acid plant, 2-3% loss
with limestone discardt and 90% actual desulfurization efficiency).
Costing procedures were standardized to insure valid economic
comparisons and analyses for different processes. If they were
available, equipment cost for heat exchangerst etc., were taken from Chemical
Engineering Progress cost files and updated. Fans and electrostatic precipitator
costs were obtained by manufacturer estimates. Installed costs of reactors
were estimated from experience for construction of large scale equipment
with the different materials specified. Limestone storage and feeder
costs were taken from the experience of TVA (24). The sulfuric acid
plant costs were taken from literature quotes, such as those by Connor (~).
No cost was assumed for installation of the process into a new power plant
designed to accommodate the desulfurization system. Costs for land for
the regenerative-by-product sub-system were not included.
Total construction costs were obtained from these costs by
the use of factors for electrical wiring, piping, buildings. etc.
Lang's analysis (~) was followed closely for these factors. Purchased equipment
costs served as a base factor. Installation costs for foundations and
labor were 43% of purchase costs. Extra costs for foundations of the
absorber and its flue gas duct work were assessed at 15% of installation
costs. Piping costs ranged from 25%-60% of installed equipment costs
for fluid-solid and fluid-fluid processes, respectively. Costs for
instrumentation, utilities and electrical wiring, and building and land
improvements ranged from 23% of piping-installation costs for the absorber
to 50% for other parts of the process. A final direct cost was obtained
for the process from the sum of these costs. An engineering, contingency,
and construction fee of 35% of the direct cost was added to obtain the
total construction cost. Working capital was estimated for 60 days
operating supplies and labor.
-------
67 -
In determining operating costs the following criteria were used:
capital charges (taxes, interests, investment return) were 14% of
investment (construction costs plus working capital), yearly maintenance
is estimated at 5% of investment costs, power for operating induction
fans, pumps, etc., should be available at 4 mils/kw. hr at the power station,
limestone costs $2.05/ton delivered; coal costs $4.20/ton. labor is
available at $3.00 hr/man and $l2,000/yr supervision, and ' plant and payroll
overhead are 50% x (maintenance + labor) and 40% of labor, respectively.
There would be a thermal effect on the boiler due to the need
to heat-up limestone. At low stone addition rates, excess heat would be
released from the exothermic reaction of sulfate formation and a small
savings in fuel would result. At high addition rates, the endothermic
heat of calcination would have to be made up by burning additional fuel in
the furnace. In the cost estimates, these effects have been assessed at a
cost or credit equivalent to $6.05/ton coal needed to make up power output
variations.
The costs for a 200 MW plant, operating at a 60% and a 90% load
factor, being desulfurized in turn by a direct injection process, a single-pass
fluidization system, and a regenerative system producing sulfuric acid,
are estimated. A scale-up to a 1000 MW plant regenerative desulfuriza-
tion process was also performed. The costs at a 60% load factor may be
the most significant, as the average power load on a grid, and thus on
a single station averaged over its life, is 58-60%.
7.2.
Process Installation in
Existing Power Plants
The use of inexpensive limestone to eliminate sulfur dioxide
pollutants from coal flue gas is very appealing. It would be desirable
to install a process utilizing limestone in presently operating power
plants. However, the cost of desulfurization must include modification
costs to the boiler in addition to direct costs associated with the
desulfurization process equipment. These costs for a fluidized bed
desulfurization process operating at 1600°F would be quite prohibitive.
Present day boilers range in size, but a convenient typical
size is a 200 MW unit. As will be shown later, a cost of $l-2/ton coal
($600,000-$l,200,000/yr) can be assessed to operate the process equipment
for this size unit. To install this fluidized bed process, the boiler
must be cut in the superheater region. Installation of duct work from the absorber
to the boiler would necessitate moving half the boiler. As boilers and
tubes are hung from the top, an almost complete reconstruction of half
the boiler would be required. This might cost from $1-3 million. In
addition the boiler would not be operating for a period of 6-18 months. A
charge of $2.75 to $8.2 million would be assessed as the incremental price to
keep customers supplied with purchased power while the plant is being
modified. This figure is obtained by estimating a cost of $15,OOO/day to
buy power at a premium rate over the internal cost of producing power.
At. a 14% yearly fixed charge rate on this total of ~3.75-$11.2, an
additional cost of $0.88-$2.60/ton coal would be levied on a fluidized
bed process. Even the extremely optimistic case of $0.88/ton coal is
excessive (the more realistic estimate is $1.65 ton/coal). These additional
costs remove a fluidized bed desulfurization process from consideration for
application to existing boilers.
-------
f""
- 68 -
7.3.
Future Conventional Power-Plant Process Installation
7.3.1.
Single Pass Desulfurization - 200 MW Plant
A single-pass process using stable limestone such as N-1359 to
desulfurize the gas in a 200 MW plant would require a $3,030,000 investment.
The general outline of the process has been described and sized in Section 6
for a 200 MW unit. The construction cost breakdown is given in detail in
Appendix VITI, including unit size specifications and flows. Capital
costs are summarized in Table 16.
Table 16
200 MW Power Plants - Single Pass Fluidized Bed
Desulfurization/Stable Limestone
Investment Costs
Item
Costs
Direct Costs;
Desulfurizer
Duc ts
Limestone Handling
Induction Fans and Motors, and
Desulfurizer-?ower Plant
Total
$
882,000
314,000
566,000
Engineering, Construction, Contingency Fee
Construction Costs
Working Capital
Total Investment
430,000
$2,192,000
758,000
2,950,000
80,000
$3,030,000
($15.l5/kw)
Basis:
87% Desulfurization, 3:1 Limestone/S02 addition ratio, 3.5% S coal.
The yearly operating costs for 60% and 90% load factors for
the single pass process are given in Tables 17 and 18. Power needs
are fairly large because induction fans are needed to overcome a A P of about 26"
H20 through the fluidization absorber. Labor requirements are 2 operators/
shift, plus a supervisor, plus one extra man for limestone handling.
Tabl,e17
200 MW Power Plant - Single Pass Fluidized Bed
Desulfurization/Stable Limestone
Yearly Operating Costs; 90% Load Factor
Item
Capital Charges (14% Investment)
Maintenance (5% Investment)
Power (4 mils/kw hr.)
Limestone ($2.05/ton)
Labor
Payroll Overhead (40%)
Plant Overhead
Thermal Effect
Costs
Total
Cost/Ton Coal
$425,000
151,000
140,000
420,000
70,000
28,000
110,500
72,000
$1,406,500
$2.35
Basis: 87% Desulfurization, 3:1 Limestone/S02 ratio, 3.5% S
coal, 90% load factor, ideal gas contacting.
-------
- 69 -
Table 18
200 MW Power Plant - Single Pass Fluidized Bed
Desulfurization/Stable Limestone
Item
Yearly Operating Costs; 60% Load Factor
Cost
Capital Charges (14% Investment)
Maintenance (5% Investment)
Power (4 mils/kw hr.)
Limestone ($2.05/ton)
Labor
Payroll Overhead (40%)
Plant Overhead
Thermal Effect
Total
Cost/Ton Coal
$425,000
151,000
93,500
322,000
70,000
28,000
110,500
711000
$1,271,000
$3.18
Basis. 87% Desulfurization, 3:45Limestone/S02 ratio, 3.5% S coal,
60% load factor, ideal gas contacting.
Lower operating costs would result if a more reactive material
than N-1359 was available. If a physically s table material of reactivity
similar to dolomite N-1337, could be found, the costs of operating a single-
pass desulfurization process would be quite similar to that with 1359 lime-
stone. However, only a 1. 35; 1 CaO (in dolomite) /S02 ratio would be required
kineticly to obtain 89% flue gas desulfurization. The limestone handling
costs would therefore be less, as less material is handled. However
the desulfurizer costs and size would be about the same. About $160,000 would be
saved in limestone handling investment costs (i.e., total investment =
$2,870,000). The operating costs are summarized in Table 19. A net thermal
credit is obtained at 90% load operation as less stone is calcined and heated
up than with the N-l359 limestone, while the same heat release would be
obtained from sulfate formation.
Table 19
200 MW Power Plant - Single Pass Fluidized Bed
Desulfurization/Stable Dolomite
Yearly Operating Costs; 90% Load Factor
Item
Cost
Capital Charges (14% Investment) $ 402,000
Maintenance (5% Investment) 143,000
Power (4 mi]s /kw hr) 140,000
Dolomite ($2.05/ton) 350,000
Labor 70,000
Payroll Overhead (40%) 28,000
Plant Overhead 106,500
Thermal Effect Credit 6,000
Total $1,228,000
Cost/ton coal $2.05
Assumptions: 89% Desulfurization, 1.35:1 CaO-Dolomite/S02 ratio,
coal, 90% load factor, ideal gas contacting.
3.5% S
-------
- 70 -
Table 20
200 MW Power Plant - Single Pass Fluidized Bed
Desulfurization/Stable Dolomite
Yearly Operating Costs; 60% Load Factor
Item
Cost
Cost/ton Coal
$ 402,000
143,000
93,500
295,000
70,000
28,000
106,500
64,000
$1,202,000
$3.00
Capital Charges (14% Investment)
Maintenance (5% Investment)
Power (4 mi1s/kw hr)
Dolomite ($2.05/ton)
Labor
Payroll Overhead
Plant Overhead
Thermal Effect
Total
Bas is :
89% Desu1furization, 1.7:1 Limestone/S02 ratio, 3.5% S coal,
60% load factor, ideal gas contacting.
7.3.2.
The Regenerative Process - 200 MW Plant
The desulfurizer costs in this case are the same as in the single-
pass process. Limestone requirements would be less, so smaller facilities
are needed to handle the stone. A 15 inch settled bed depth would be employed,
as opposed to the 24 inch bed in the single-pass system. A shallower bed
could be used since regeneration permits operation with a low sorbent
utilization and high reactivity stone in the desulfurizer without increasing
limestone requirements. The bed pressure drop would therefore be less than
with a single pass system, and less powerful induction fan facilities are
required. Capital costs are summarized in Table 21, including the costs for
the regeneration sub-syste~ The acid plant includes a contacting-absorption
section, but no sulfur burning equipment is needed. Off-site costs of 20%
were included to obtain total construction costs for a working unit.
-------
- 71 -
Table 21
200 MW Power Plant - Regenerative Fluidized
Bed Desulfurization/Stable Limestone
Investment Costs
Item
Direct Costs:
Desulfurizer
Ducts
Limestone Receiving-Disposal
Limestone Recirculation .
Fans & Motors, Desulfurizer-Power
Plant
Regenerator
Regenerator Effluent Preparation
for Acid Plant
Total
Eng., Constr., Contigency Fees
Sulfuric Acid Plant, Constructed
Construction Costs
Working Capital
Total Investment
Cos ts
$
882,000
314,000
150,000
205,000
285,000
290,000
300,000
$2,426,00
848,000
1,185,000
$4,459,000
126,000
$4,585,000
($22.92/kw)
Basis:
90% Desulfurization; 1:3 Limestone/S02 addition rates, 3.5% S coal,
205 H2S04 ton/day plant capacity.
Yearly operating costs are estimated for a 200 MW plant.in
Tables 22 and 23. The price of $14.00/ton H2S04 is estimated return,
exclusive of costs for transportation and marketing, on the product. The
actual price .will be set by outside competition, which will in turn be
controlled by the sulfur prices prevalent. The above price is the sulfuric
acid return price expected when sulfur prices are $30/ton. The sulfur
price in the 70's is expected to be in the range, $25-35/ton (Q) . The
personnel required to operat€ the desulfurization process should be
sufficient to also operate the automated acid plant.
-------
- 72 -
Table 22
200 MW Power Plant - Regenerative Fluidized Bed
Desulfurization/Stable Limestone
Yearly Operating Costs; 90% Load Factor
Item
Capital Charges (14% Investment)
Maintenance (5% Investment)
Power (4 mils/kw hr)
Limestone ($2.05/ton)
Labor
Plant Overhead
Payroll Overhead (40%)
Operating Water (50~/M gal)
Reducing Coal ($4.20/ton)
Thermal Effect
Operating Cost
Sulfuric Acid Return ($14.00/ton)
Net Desulfurization Cost
Cost/ton coal
Cost
$ 643,000
230,000
132,000
50,000
70,000
150,000
28,000
10,000
118,000
Credit 57,000
$1,374,000
761,000
$ 613,000
$1. 02
Assumptions:
3.5% S coal, 8000 hr operation yearly, 90% load factor,
85% conversion of sulfur to sulfuric acid, 90% desu1furization
in actual operation.
-------
- 73 -
Table 23
200 MW Power Plant - Regenerative Fluidized Bed
Desulfurization/Stable Limestone
Yearly Operating Costs; 60% Load Factor
Item
Capital Charges (14% Investment)
Maintenance (5% Investment)
Power
Limestone ($2.05!ton)
Labor
Plant Overhead
Payroll Overhead
Operating Water (50~IM gal)
Reducing Coal ($4.20/ton)
Thermal Effect
Operating Cost
Sulfuric Acid Return ($14.00/ton)
Net Desuliurization Cost
Cost/ton coal
Assumptions :
Cos t
$ 643,000
230,000
88,000
42,000
70,000
150,000
28,000
7,000
80,000
Credit 29,000
$1,309,000
503,000
$ 806,000
$2.01
3..5% S coal, 8000 br.. operation yearly, 60% load
factor, 85% conversion S ~ H2S04, 90% desulfurization
in actual operation, 2.4:1 S02/limestone make-up ratio.
7.3.3.
The Regenerative Process - 1000 MW Plant
Five desulfurization reactors of the size previously described would
be used in this design, as larger diameter reactors are felt to be impractical.
The flue gas must, therefore, be divided into five streams to be
transported to the individual absorber. Thus, more duct work is required
for this 1000 MW plant desulfurization unit than might be expected on
scale-up from a 200 MW unit.
Costs for the 1000 MW system were calculated separately for
each piece of equipment, where possible. If direct estimation could not be
used, the 0.6 factor (~) was used for economic scale-up from a 200 MW
unit. Other costing procedures followed the outline used for a 200 MW
regenerative system.
.
-------
- 74 -
Table 24
1000 MW - Regenerative Fluidized Bed
Desulfurization/Stable Limestone
Investment Costs
Item
Direct Costs
Desulfurizers
Ducts
Limestone Receiving-Disposal
Limestone Recirculation
Fans and Motors, Desulfurizers-Power
Plant
Regenerator
Regenerator Effluent Preparation for
Acid Plant
Total
Eng., Constr., Contigency Fees
Sulfuric Acid Plant, Constructed
Construction Costs
Working Capital
Total Investment
Cos ts
$ 4,410,000
2,070,000
350,000
955,000
740,000
780,000
770 ,000
10,075,000
3,525,000
3,040,000
16,640,000
310,000
$16,950,000
($16.95/kw)
Bas is :
92% peak desulfurization, 1:3 Limestone/SOZ addition rates,
3.5% S coal, 1025 H2S04 ton/day plant capac1ty.
The operating costs are given in Table 25 and 26 for two load
factors. The same criteria were used for sulfuric acid credit, etc., as in
the 200 MW unit case. Labor needs were estimated at 4 men/shift + 1
supervisor + 1 day time limestone disposal man.
Since the future value of sulfuric acid is uncertain, the effect
of varying acid prices is graphed in Figure 32. The variations in sulfur
content also affect desulfurization costs (Figure 33). In deriving the
costs, it was assumed that the whole system was sized to be able to handle
sulfur coals as high as 3.5-4%.
-------
- 75 -
Table 25
1000 MW Regenerative Fluidized Bed
Desulfurization/Stable Limestone
Yearly Operating Costs; 90% Load Factor
Item
Capital Charges (14% Investment)
Maintenance (5% Investment)
Power (4 mils/kw hr)
Limestone ($2.05/ton)
Lab or
Plant Overhead
Payroll Overhead (40%)
Operating Water (50~/M gal)
Coal ($4.20/ton)
Thermal Effect
Operating Cost
Sulfuric Acid Return ($14.00/ton)
Desulfurization Cost
Cost/ton coal
Assumptions:
Cost
Credit
$2,370,000
848,000
660,000
250,000
118,000
483,000
48,000
50,000
590,000
284,000
5,133,000
3,805,000
$1,328,000
44~
3.5% S coal, 8000 hr operation yearly, 90% load factor, 1:3
limestone/S02 ratio, 85% conversion of sulfur to sulfuric acid
90% deSu1furizatiOri in actual operation '
Table 26
1000 MW Regenerative Fluidized Bed
Desulfurization/Stable Limestone
Yearly Operating Costs; 60% Load Factor
Item
Capital Charges (14% Investment)
Maintenance (5% Investment)
Power (4 mils/kw hr)
Limestone ($2.05/ton)
Labor
Plant Overhead
Payroll Overhead
Operating Water (50~/M gal)
Coal ($4.20/ton)
Thermal Effect
Operating Cost
Sulfuric Acid Return ($14.00/ton)
Desu1furization Cost
Cost/ton coal
Assumptions:
Cost
Credit
$2,370,000
848,000
440,000
210,000
118,000
483,000
48,000
33,000
393,000
144,000
$4,809,000
2,535,000
$2,274,000
$1.14
3.5% S coal, 8000 hr operation yearly, 60% load factor,
1:2.4 limestone/S02 make-up ratio, 90% desulfurization
in actual operation.
-------
- 76 -
Figure 32
Net Desu1furization Costs as a Function
of By-Product Credit
.-I
~
0 2.00
tJ
~ Basis: 3.5% S coal
0 85% conversion of
~
.......... coal sulfur to H2S0
-------
7.4.
Direct Injection - 200 MW Plant
- 77 -
The capital investment required for a direct injection process using
a 2:1 limestone/S02 mole ratio was taken directly from a TVA report (~).
One could break down their direct costs into two major items: (1) $569,000
is required to receive limestone, transport it to the boiler and cart it away;
(2) $581,000 is required to crush the limestone to fine particl~and then
collect the particles in cyclones and electrostatic precipitators so as not
to contaminate the air. The engineering fees and working capital bring the
investment to $1.578 million.
It should be noted that the design specifications for this process
call for 66% desulfurization, although only 40-50% S02 removal has been
generally obtained with a 2:1 limestone to S02 ratio.
Yearly operating costs are given in the following table:
Table 27
200 MW Power Plant - Direct Injection
Desulfurization Costs
Item
Capital Charges (14% Investment)
Maintenance (5% Investment)
Limestone ($2.05/ton)
Power (4 mils/kw hr)
Labor
Payroll Overhead (40%)
Plant Overhead (50% Main. and Labor)
Thermal Effect
Total Cost
Cost/ton coal
90% Load Factor
Yearly Cost
60% Load Factor
Yearly Cost
$221,000
29,000
285,000
21,000
39,500
16,000
59,000
21,600
$792,000
$1. 24
$221,000
29,000
190,000
14,000
39,500
16,000
59,000
14,400
$683,000
$1. 71
Assumptions:
3.5% S coal, 8000 hr yearly operating time,
a 2:1 1imestone/S02 addition ratio.
-------
- 78 -
7.5.
Regenerative Desulfurization in a Fluidized-
Bed-Combustion, 1000 MW Plant
Order of magnitude estimates of costs for desulfurization with
limestone in the fluidized-bed-combustion system (FBC) described in Section 6.4.
have been made, assuming favorable operating characteristics, which must
still be demonstrated. The costs cited in this sub-section are therefore
less firm than previous process estimates. Only the incremental costs for
desulfurization above a non-desulfurizing FBC plant have been estimated.
The desulfurization reactor would be the fluidized-bed-combustion
chamber. Since no modifications for the desulfurization scheme are required,
no reactor costs have been assessed. No additional charge has been .
assessed for draft fans as they too are already required for the FBC scheme.
The costs for the regenerative process are then as follows:
Table 28
Desulfurization in a 1000 MW - Regenerative
Fluid Bed Combustion Plant With Stable Limestone
Incremental Investment Costs
Item
Costs
Direct Costs
Limestone Receiving-Disposal
Limestone Recirculation
Regenerator
Regenerator Effluent Preparation
for Acid Plant
Total
Eng., Constr. Contigency Fees
Sulfuric Acid Plant, Constructed
$
350,000
540,000
780,000
770,000
Construction Costs
Working Capital
Total Incremental Investment
$2,540,000
890,000
3,040,000
$6,470,000
200,000
$6,670,000
($6.67/kw)
Basis:
90% Desulfurization, 1:3 Limestone/S02 molar addition ratio,
3.5% S coal, l600°F bed operating temperature.
-------
- 79 -
Table 29
Desulfurization in a 1000 MW - Regenerative
Fluid Bed Combustion Plant with Stable Limestone
Item
Capital Charges (14% Investment)
Maintenance (5% Investment)
Power (4 mils/kw hr)
Limestone @ 2.05/ton
Labor + Overhead
Plant Overhead
Water @ 10~/M gal
Reducing Coal ($4.20/ton)
Thermal Effect
Total Cos t
H2S04 Return ($14/ton acid)
Net Profit
Profit/ton coal
Operating Costs
Costs
60% Load Factor
90% Load Factor
935,000
334,000
40,000
167,000
98,000
202,000
33,000
393,000
Credit 190,000
$2,012,000
2,535,000
$ 523,000
l7~
935,000
334,000
60,000
250,000
98,000
202,000
50,000
590,000
285,000
$2,234,000
3,805,000
$1,571,000
5Z~
Basis:
90% Desulfurization, 1:3 Limestone/SOZ molar addition ratio,
3.5% S coal, l600°F Bed opereating temperature
The incremental investment costs would be quite low, at $6.67/kw,
in comparison with other desulfurization processes. Although a profit is
estimated, operating problems could require additional investment
to alleviate the difficulties, and costs 'could correspondingly increase.
7.6.
Economics Summary
Table 30 summarizes the capital and operating costs developed
for the different fluid bed limestone desulfurization cases. The high cost
of modification to a power plant presently in operation would prohibit the
process being used for this application. Instead, systems easily adapted
to present plants, such as direct limestone injection, perhaps coupled with
a wet scrubber, would be employed for desulfurization in these cases.
The ability to re-use limestone after regeneration cuts limestone
costs significantly. The return from sale of a by-product (H2S04) also reduces
desulfurization costs. These factors give a significant advantagem a
regenerative system as compared to a single-pass use of limestone.
f1
~),
~,
l
-------
- 80 -
Net operating costs of the external fluidized bed scheme appear
competitive with other proposed processes. Most estimates for otnp.r
processes are also in the range $0.25-$1.00/ton coal, for high load factors
and 800-1000 MW plants. As expected, new (or large) plants operating
at a 90% load factor are much more economical than those operating at 60% load,
which is more nearly the average grid load factor. The quick estimate
for fluidized-bed-combustion desulfurization shows greater economic
long-term potential for this combined power-desulfurization scheme
than the conventional power plant modified with a desulfurization unit.
Table 30
Cost Summary for the Limestone Fluid Bed
Desulfurization Processes
Basis:
About 90% Gas desulfurization, 3.5%
Limestone or dolomite at $2.05/Ton,
H2S04 return of $14/ton.
S fuel
Plant Size Plant Investment Operating Costs - $/Ton Coal
Process MW $MM $/KW 90% Load Factor 60% Load Factor
Once-Through Limestone Process 200 3.03 15.15 2.35 3.18
Once-Through Dolomite Process 200 2.87 14.35 2.05 3.00
Regene rati ve Limestone Process 200 4.585 22.92 1.02 2.01
1000 16.950 16.95 .44 1.14
Direct Injection Limestone 200 1.579 7.90 1.24 1.71
Process
Regenerative Limestone Fluid 1000 6.67(1) 6 .67 (1) .52(2) .17( 2)
Bed Combustion Process
(1) Incremental investment over that required for a
fluid bed combustion system without desulfurization.
(2) Net profit over the costs of operating a fluid
bed combustion system without desulfurization
-------
- 81 -
8.
STATUS AND RECOMMENDATIONS
8.1.
Present Status
The regenerative system provides considerable economic advantages
over a process with single pass use of limestone and would be the preferred
desulfurization system. The economics, necessarily preliminary at this
stage, are competitive with other integrated power plant-desulfurization
processes. Although this process would not be installed in presently
operating power plants because of high boiler modification costs, it is a
possibility to provide a relatively low cost S02 process for the large con-
ventional plants of the future. The process, however, appears to have a
much greater potential if fluidized bed combustion is developed commercially.
The process, as presented, yields the by-product, sulfuric acid.
Modifications of the regenerative scheme might enable the process to produce
the more versatile product, sulfur. The important points, however, are
that a saleable by-product could be produced. and the regenerative sys-tem
can work.
Another favorable aspect of this regenerative process is the
relative insensitivity of the total cost on the use and price of limestone.
Only 3% of the total costs are limestone purchase costs. Even if the
price and life of limestone particles were in considerable error, the
economics would npt be significantly effected and the economic
conclusions would not be changed.
8.2.
Technical Problem Areas
Technical feasibility has been shown for a good part of the
process. However, there are several potential problem areas in this process
which would need to be more thoroughly investigated if the process were to
be developed.
Bubble formation and growth in a large particle, shallow fluidized
bed is difficult to predict from bench scale experiments. Fluid-solid
contacting depends on bubble formation, as mentioned in the text. Poor
contacting leads to low desulfurization efficiencies. The contacting factors,
defined in Section 3.4., are- difficult to predict and might range from
0.1 to 0.9. A contacting factor of 0.50 would mean a 70% S02 removal
instead of 90%, and a reduction in the production of H2S04 and subsequent
sales credit. Costs would then increase by 25~/ton coal. It is
anticipated that engineering development on grid and bed design would
be required to obtain high contacting factors by reducing bubble
growth.
A probable grid-bed design has been mentioned in the text (Section 6.1.3).
Grid materials of construction must retain sufficient tensile strength to
support the weight of the bed at 1600°F. A supported stainless steel (310)
plate was used in the design study although further study may indicate that a
ceramic material is required. The gas inlet ports are small holes
and are necessary to keep the initial bubble size small. Numerous vertical
rods scattered throughout the bed hopefully would deter bubble growth.
-------
- 82 -
In the assumed grid design, numerous small holes, drilled in grid buttons
at right angles to the direction of gas flow through the bed, have been assumed
to provide good fluidization, but the use of small gas-inlet holes presents
a problem. It is known that fly ash (or any particle) will tend to follow
straight paths when entrained by gas flow; they do not make sharp
turns readily. Instead they tend to impinge against elbows, obstacles,
or surfaces which force a turn. The possibility of compaction
and clogging of the gas inlet orifices by fly ash thus exists. Fly ash
clogging or impingement wear of the button must be demonstrated to be small
before complete technical feasibility of our fluid bed process has been
shown.
Fly-ash impingement in the grid would not be present in a
fluidized bed combustion process, since the coal would be fed above the
grid. The grid itself would be cooled by inlet air, and materials for
construction would be easily obtainable.
Means to prevent agglomeration of the lime particles in the
regenertor still need to be definitely established to insure a viable
cyclic process. Possibly, bed agitation and correct particle size
distribution will suffice, as described in the text. If not, reducing
gas would have to be made by the partial combustion of oil or gas. Other
reducing gases (e.g., reformer gas) would not be able to satisfy the
thermal requirements of the system.
8.3.
Other Process Considerations
The process equipment is quite large. The desu1furizers are 50
feet high with large duct work into the 50 foot diameter beds. As a
comparison, a conventional 1000 MW power plant burns coal in perhaps a
50' x 70' chamber and extends upward 180' (plus 80' for support girders)
before returning through the superheater tubes, etc.
The operating temperature of 1600°F requires the placement of
the desulfurizers in the secondary superheater region. Even in a
specially designed system, the economics of power operation would be affected
by this placement. At very low power loads, the temperature in the
superheater region might decrease down to 1400°F. This would adversely
affect reaction rate although balanced to some extent by the lower gas
velocities and consequently longer gas residence times. Lower
reaction rates would necessitate higher relative addition rates of limestone
for the same desu1furization efficiency. Thus, the large power units, which
are favored economically in this process, should be operated at high loads
for economy with the smaller units in the power grid used to accomodate
the daily changes in power consumption.
The regenerator itself is not large. However, as in all desu1furiza-
tion processes which produce a by-product, considerable land is needed
to accomodate the by-product plant. The cost of land must be included as
added investment for all such processes and this may rule out their use
in some cases.
-------
- 83 -
The 90-odd new fossil fueled plants over 500 MW capacity projected
for construction by 1990 by the Federal Power Commission (~) will be built
mostly in the central part of the U.S. Easy access to limestone deposits
is thus foreseen. The highly congested Northeast with its higher land costs
in the big cities but great need of ambient sulfur dioxide control will not
have many new fossil-fueled plants, and thus little need for the present
regenerative desulfurization process.
8.4.
Commercialization Lead Time
To carry the development of the present conventional process
through to commercial status, several steps must be taken. The fly-ash
and the contacting problems must be investigated. Preliminary information
could be gained on the compaction problem in present laboratory size
equipment. Further answers must await the construction of a 1 to 2
foot diameter semi-works fluid bed. To minimize scale-up problems, a
proto-type unit of the order of 20 feet in diameter would be needed.
As the power industry demands high reliability, a complete commercial size
200 MW integrated power plant-process would need to be financed and
demonstrated for reliability before the industry would accept it for use.
The time scale for this development under a crash program might be 4 years
until commercial plant construction, followed by several years of demonstra-
tion of the reliability of the process in a power plant.
8.5.
Recommendations
Adaption of the present process to fluidized bed combustion
power cycle, presently being developed, is potentially quite attractive.
Fly ash compaction problems should not occur; gas-solid contacting problems
would be reduced; and the economics predict a possible profit, assuming
favorable operating conditions on this first analysis. The future of this
process is therefore judged to lie with development of fluidized bed
combustion. Such development is estimated to be at the commercial plant
demonstration state in the mid 1970's. If a concurrent program on
regenerative limestone desulfurization with fluid bed combustion is carried
on, an integrated fluid bed combustion desulfurization system could be
developed almost as rapidly as other fluid bed combustion systems. Con-
sequently, we recommend that future development on limestone-based fluid
bed desulfurization be directed at adapting the process to fluidized bed
combustion schemes.
-------
- 84 -
NOMENCLATURE
CH = Concentration of 502 exiting from a fluidized bed.
Co = Initial concentration of 502 into a fluidized bed.
d = Limestone particles average "diameter" as measured by passage through
given mesh sizes.
D = Reactor diameter.
k = Psuedo first order rate constant on 502 concentration for sulfation
reaction
G = Volumetric gas flow rate.
H = Initial settled bed height
o
R = Psuedo rate coefficient for sulfation reaction at specified conditions
in a fluidized bed.
T = Temperature
U = Superficial gas velocity in a fluidized bed.
U = Superficial gas velocity at incipient fluidization.
o
V = Volume of gas bubbles in a fluidized bed.
B
x = Fraction of CaO in limestone particles that has apparently reacted to
form CaS04.
Y = Contacting factor, essentially the efficiency of contact between
solids and gas in a fluidized bed.
~ G; = Gibbs free energy of formation of a compound
~P = Pressure drop through a given system, usually the fluidized bed and
grid combined.
-------
(10)
(11)
(12)
(13)
- 85 -
REFERENCES
(1)
Harvey, R. D., "The Mineralogy and Petrography of Carbonate Rocks
Related to Sulfur Dioxide Reactivity," NAPCA sponsored 3rd Limestone
Symposium, St. Petersburg, Florida, December 1, 1967.
(2)
"Status Report of Alkaline Additive In-House Research," Prepared for
the National Center for Air Pollution Control, Cincinnati, Ohio,
Process Control Engineering Section by Bituminous Coal Research.
(3)
Davidson, J. F. and Harrison, D., "Fluidized Particles," Cambridge
University Press, 1963.
(4)
Matsen, J. M. and Tarmy, B. L., "Scale-up of Laboratory Fluid Bed
Data: The Significance of Slug Flow," AIChE Meeting, Tampa, Florida,
May 1968.
(5)
Hovmand, S. and Davidson, J. F., "Chemical Conversion in a Slugging
Fluidized Bed," Trans. Inst. Chern. Eng., 46(6) :T190-T203, 1968.
(6)
Coutant, R. W., et a1., "Investigation of the Reactivity of Limestone
and Dolomite for Capturing S02 from Flue Gas," Battelle Memorial
Institute, Summary Report, 1968.
(7)
Glasson, D. R., "Reactivity of Lime and Related Oxides. VII. Crystal
Size Variations in Calcium Oxide Produced from Limestone,"
J. Appl. Chern., 11:201-6, June 1961.
(8)
Levenspiel, 0., "Chemical Reaction Engineering," J. Wiley and Sons,
Inc., 1962, pp. 346-367.
(9)
Wheelock, T. D. and Boylan, D. R., "Production of Sulfur Dioxide and
Lime from Calcium Sulphate," The Ind. Chern., 36(430):590-4, 1960.
White, S. B., Johnson, S. M. and Dantzig, G. B., "Chemical Equilibrium
in Complex Mixtures," Rand Corp. Bulletin, P-l059.
Clasen, R. J., "The Linear Logarithmic Programming Problem," Rand
Corp. Bulletin, RM-3707.
Boynton, R. S., "Technology of Lime and Limestone," Interscience
Publishers, New York, 1966.
Bourgeois, P. and Grenier, P., "The Ratio of Terminal Velocity to
Minimum Fluidizing Velocity for Spherical Particles," Candian J. Chern.
Eng., 46:325-8, Oct. 1968.
-------
(14)
(15)
(16)
(17)
(18)
(19)
(20)
(21)
(22)
(23)
(24)
(25)
(26)
(27)
(28)
- 86 -
Leva, M., "Fluidization," McGraw-Hill, New York, 1959.
Zenz, F. A. and Othmer, D. F., "Fluidization and Fluid-Solids System,"
Rheinhold Publishing Corp., New York, 1960.
Visits to and djscussion with the Pope Evans and Robbins Co.,
Alexandria, Virginia.
Vo1k, W., Johnson, C. A. and Statler, H. R., "Effect of Reactor
Interna1s on Quality of Fluidization," Chern. Eng. Prog., 58(3):44-7,
1962.
Wheelock, T. D. and Boylan, D. R., "Sulfuric Acid from Calcium Sul-
fate," Chern. Eng. Prog., 64(11):87-92, 1968.
Williams, D. F., "Power Station Boilers: Preliminary Costing and
Considerations," First International Conference on Fluidized Bed
bution, sponsored by NAPCA. Hueston Wood, Ohio, 1968.
Design
Com-
Zie1ki, C. W., Lebowitz, H. E., Struck, R. T. and Gorin, E., "Sulfur
Removal During Combustion of Solid Fuels in a Fluidized Bed of Dolo-
mite," Div. of Fuel Chemistry of ACS, 13(4):13-29, 1969.
"S02 Removal: Cheaper Process Piloted," C&EN, 46(34):22, 1968.
Ryasan, P. P. and Harkins, J., "Studies on a New Method of Simultaneously
Removing S02 and NOx from Combustion Gases," J. Air Pollution Control
Assoc., 17(12):796-99, 1967.
Squires, A. M., "Cyclic Use of Calcined Dolomite to Disu1furize Fuels
Undergoing Gasification," Advances in Chemistry, Series No. 69,
p. 205-229, 1967.
"Sulfur Oxide Removal from Power Plant Stack Gas: Conceptual Design and
Cost Study; Sorption by Limestone or Lime; Dry Process," Prepared for
NAPCA by TVA. CFSTI:PB-168-972, 1968.
Connor, J. M., "Economics of Sulfuric Acid Manufacture," Chern. Eng.
Prog., 64(11):59-65, 1968.
Lang, H. J., "Cost Engineering in the Process Industries," C. H. Chilton,
Ed., McGraw-Hill, New York, 1960.
Manderson, M. C., "The Sulfur
1968.
Outlook," Chern. Eng. Prog., 64(11);47-53,
"Electric Power: Environment Counts," C&EN, 47(2):12, 1969.
-------
- 87 -
APPENDICES
-------
APPENDIX I
DESULFURIZATION KINETICS, BASIC DATA
Standard Conditions:
Bed Temperature, 1600°F; S02 inlet conc.,
2700 ppm. Calcination of stones: 6-7 hrs @ 1700°F
Studies Performed in Unit Described in Section 3,
Figure 1
U.S. Fraction S02 in Effluent at Utilization, X, of
Run N- Mesh U (ft/sec) Ho (in.) 0.05 0.10 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.6 0.7 0.8 0.9
- - - - - - - -
A - Standard Conditions
R-17 1337 16/18 3.2 2 .01 .04 .12 .21 .33 .46 .62
R-18 1337 16/18 2.1 2 .04 .12 .22 .36 .53 .77
R-23 1337 16/18 3.2 2 .02 .09 .19 .29 .40 .54 .72 .95
R-30 1337 12/14 3.2 2 .17 .34 .46 .565 .66 .76
R-31 1337 12/14 1.1 2 .03 .15 .31 .50 .71
R-34 1337 16/18 1.9 2 .01 .08 .19 .31 .42 .52 .61 .70
R-35 1337 16/18 1.3 2 .03 .15 .32 .46 .58 .70 .83 00
00
R-36 1337 16/18 6.2 2 .30 .54 .73 .90
R-38 1337 16/18 2.6 2 .01 .05 .12 .22 .35 .51 .70
R-39 1337 16/18 4.2 2 .03 .10 .19 .29 .39 .50 .62 .73
R-40 1337 16/18 5.1 2 .05 .14 .24 .30 .44 .54 .64 .73
R-41 1337 16/18 6.2 2 .13 .31 .48 .60 .72 .83 .93
R-46 1359 16/18 1.6 2 .005 .11 .57 .81
R-47 1359 16/18 2.1 2 .01 .02 .09 .29 .54 .73
R-48 1359 16/18 3.2 2 .03 .10 .25 .58 .80
R-49 1359 16/18 4.2 2 .01 .06 .13 .25 .46 .74
R-50 1359 16/18 5.3 2 .01 .07 .17 .34 .59 .89
R-51 1359 16/18 6.2 2 .01 .06 .14 .26 .40 .58 .77 .91
R-52 1359 16/18 3.1 2 .01 .07 .18 .33 .65 .86
R-54 1359 16/18 4.2 4 .04 .25 .69 .92
R-61 1359 16/18 4.2 4 .08 .32 .85
R-62 1359 16/18 4.2 8 .40 .92
R-63 1359 16/18 4.2 2 .04 .15 .57 .81
R-64 1359 16/18 4.2 1 .01 .06 .13 .23 .35 .49 .62 .72 .78
R-65 1359 16/18 4.2 2 .02 .09 .22 .48 .74
R-66 1359 16/18 4.2 1 .04 .10 .17 .28 .42 .56 .68 .78 .86
-------
AP~ENDLX 1 (Conttd)
, II.
U.S. FractionS02 in Effluent at Utilization, X, of
Run N- Mesh U (ft/sec) Ho (in.) 0.05 0.10 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.6 0.7 0.8 0.9
- - - - - - -
R-67 1337 16/18 4.2 4 .05 .14 .23 .39 .46 .51
R-68 1337 16/18 4.2 1 .01 .04 .07 .11 .15 .20 .25 .31 .37
R-72 1359 6/8 4.2 2 .05 .17 .39 .62 .79
R-73 1359 16/18 4.2 2 .04 .17 .61 .90
R-74 1359 6/8 4.2 .69 .85
R-76 1359 6/8 4.2 1 .44 .48 .66 .83
R-77 1351 6/8 4.2 1 .18 .27 .36 .45 .54 .60 .66 .71 .76
R-78 1360 6/8 4.2 1 .26 .41 .51 .58 .65 .71 .76 .79
R-79 1363 6/8 4.2 1 .28 .46 .61 .71 .79 .87 .94
R-80 1351 6/8 4.2 1 .15 .29 .41 .51 .60 .66 .72 .77 .82
Modifications
B - T=1200op (Half-Calcined Stone)
R-60 1337 14/16 1.7 2 .20 .25 .30 .34 .38 .42 .46 .51 .56
CXJ
'"
(Copper Modified)
R-44 1337 14/16 1.7 2 .14 .29 .42 .53 .64 .75 .87
C - T=1600op (Metal-Doped) (Nickel)
R-20 1359 18/25 1.1 2 .40 .95
(Aluminum)
R-26 1359 16/25 1.1 2 .20 .70
(Cobalt)
R-22 1359 18/25 1.1 2 .36 .82
R-27 1359 16/25 1.1 2 .32 .76
(Iron)
R-25 1359 16/25 1.1 2 .40 .82
(Copper)
R-21 1359 18/25 1.1 2 .04 .26 .68
R-28 1359 16/25 1.1 2 .01 .13 .53
-------
U.S.
Run N- Mesh U (ft/sec) H() (in.)
- - - D - Temperature Effects T=1500°F
R-45 1359 4/10 4.0 2
J-129 1359 25/50 2.0 3.2
T=1413°F
J-122 1359 25/50 2.0 3.2
T=1400°F
R-81 1359 6/8 3.8 1
T=1250°F
R-82 1359 6/8 3.5 1
E - Particle Size
R-14 1337 16/25 1.1 2
R-29 1337 20/25 1.1 2
A-37 1337 40/50 0.5 4
A-38 1337 70/100 0.5 4
A-39 1337 140/170 0.5 3
A-47 1359 50/100 0.5 3
APPENDIX I (Cont'd)
Fraction S02 in Effluent at Utilization, X, of
0.05 0.10 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.6 0.7
------------
.32 .77 .93
.01 .34 .74
.03 .12 .32
.52 .61 .70 .80
.66 .75 .87
0.8
0.9
--
\D
.15 .51 .79 0
.08 .24 .43 .67 I
.02
.02
.00 .10
.15
-------
- 91 -
APPENDIX II
REDUCTIVE REGENERATION BASIC DATA
Basic representative data on reduction of sulfated particles,
followed by absorption of S02 from flue gas, are presented in this section.
A.
Desu1furization in unit described
Regeneration in unit described in
Material: MgS04' 50/100 Mesh, Ho
in Section 3(D)
Section 5 (R)
= 6 in.
Temperature [SO,,) (' ) % X (20% Out
Run CF) Gas Composition L J.'lC:1.X S02 In )
A-15R 1600 10% H2/90% N2 7.6
A-16R 1600 10% CO/90% N2 8.9
A-17R 1400 10% CO/90% N2 11.3
D 1350 Simulated Flue Gas 0
A-18R 1300 10% CO/90% N2 8.3
D 1350 Simulated Flue Gas 0
A-21R-l 1500 10% H2/20% H20/70% N2 9.1
R-2 1500 20% CO/20% C02/60% N2 17.3
A-23R-1 1700 5% H2/95% N2 6.8
R-2 1700 10% H2/90% N2 10.0
R-3 1700 10% H2/5% CO/15% N2 13.9
D 1350 Simulated Flue Gas 0.005
B. Desu1furization in unit described in Section 3(D)
Regeneration in unit described in Section 5 (R)
25/50 Mesh CaS04 (anhydrite) particles
Fluidizing Gas 2/1 CO?/CO or H20/H2; bal. N2; Ho = 6" in Reg.,
3" in Desulf. . (initially) .
Temperature [S02J (Max) % X (20% Out
Run Cycle (oF) Gas Composition S02 In )
A-37R-1 2060 10% CO 7.3
R-2 2060 20% CO 9.3
A-38R-l 2060 10% H2 8.7
R-2 2060 15% H2 10.3
D 1600 Flue Gas 0.55
A-39R-1 2160 10% CO 6.6-7.3
R-2 2160 15% CO 8.6-9.5
D 1600 Flue Gas 0.31
A-4uR-1 1960 10% CO 6.3-7.3
R-2 1960 15 % CO 8.1-9.6
D 1600 Flue Gas 0.44
A-41R-1 1960 10% H2 8.4-8.7
R-2 1960 15% H2 9.9-10.2
D 1600 Flue Gas 0.47
-------
- 92 -
APPENDIX II ~ont'd)
Temperature [S02] (Max) % X(20% SO? Out)
1\.Ul1 Cycle (oF) Gas Composition In
A-43R-1 1860 10% H2 5.2-6.0
R-2 1860 15% H 6.6-6.9
D 1600 Flue tas 0.42
A-44R-1 1960 10% CO 5.7-6.7
R-2 1960 15% CO 8.0-9.6
A-45R-1 1 2060 10% H2 8.6
R-2 2060 15% CO 9.3
D-1 1600 Flue Gas 0.48
R-2 2 2060 10% CO 8.0
D-2 1600 Flue Gas 0.20
R-3 3 2060 10% CO 8.3
D-3 1600 Flue Gas 0.17
R-4 4 2060 10% CO 8.4
D-4 1600 Flue Gas 0.127
R-5 5 2000 10% CO 8.6
D-5 1600 Flue Gas 0.114
R-6 6 2000 10% CO 8.2
C.
Desu1furization in unit described in Section 3 (D)
Regeneration in unit described in Section 5 (R)
U = 2 ft/sec, Ho = 8 in; 2/1 C02/CO; Material 25/50
Mesh N-1337
Temperature [S02] (Max) % xC Out)
Run Cycle CF) Gas Composition 20% SO? In
J-58D 1 1600 Flue Gas 0.45
R 2000 8.1% CO 8.2
J-59D 2 1600 Flue Gas 0.212
R 2000 8.1% CO 7.6
J-62D 3 1600 Flue Gas 0.154
R 2000 8.1% CO 8.0
J-68D 4 1600 Flue Gas 0.175
R 2000 10% CO 9.1
D. Desu1furization, Regeneration in one unit;
U = 3 ft/sec; Material 25/50 N-1337,
Preattrited at U = 6 it/see for 6 hr.
Temperature [S02] % X ( Out)
Run Cycle (oF) Gas Composition (Max) 20% SO? In
J-60D 1 1600 Flue Gas 0.35
R 2000 8.1% CO 7.8
J-61D 2 1600 Flue Gas 0.151
R 2000 8.1% CO 8.1
U = 2 ft/sec H = 6 in, Calcined Dolomite, N-1337
0
J-74D 1 1600 Flue Gas 0.525
R 2000 10% CO 8.7
J-75D 2 1600 Flue Gas 0.720
R 2000 10% CO 8.8
-------
- 93 -
APPENDIX II (Cont'd)
E.
Desulfurization, Regeneration in unit described in Section 5.
U = Z ft/sec, Ho = 8.1 in, Z/l COZ/CO,
Material Z5/50 lime (N-1359)
F.
Regeneration of particles with pure HZ/NZ feed mixtures,
followed by NZ/OZ mixtures to oxidize CaS. Z5/50 Mesh, N-1359 Lime.
Temperature Feed Gas Composition [SoZJ(Max)% Effluewt
Run (OF) (excluding N2) (Minus HZO),
Fluidized Bed
J-131R-l ZOOO Z5% HZ 7.Z
(with °z overhead to 10.1
react with HZS
R-Z 2000 15.5% 0z 8.Z
Packed Bed
J-135R-l ZOOO 10% HZ 3.5 (includes oxidized HZS)
R...,Z ZOOO 16% °z 4.8
J-136R-l ZOOO 9.Z% HZ Z.8
3.6 (includes oxidized HZS)
R-Z ZOOO 9.Z% HZ, 2.3% S02 0
R-3 ZOOO 16% 0z 7.0
-------
- 94 -
APPENDIX III
ERROR ANALYSIS - DESULFURIZATION TO CaS04 FORMATION
The amount of S02 sorbed by the lime particles from the flue gas
was calculated from infrared detector monitoring and from weight gain of
the particles. The following runs were chosen to determine any
difference between the methods.
SUMMARY, WEIGHT DIFFERENCE DEPENDENCE
ON ANALYSIS PROCEDURE
Run
Z, Chart - Weight
(gillS S03/155 gm bed)
+4.6
+2.2
-0.6
-2.7
+4.2
+7.0
-10.5
+9.2
+13.3
-8.2
+5.3
+11.6
+5.4
+12.7
+9.5
+3.0
+5.5
+9.4
+5.7
+4.5
+6.8
+10.6
+9.8
+10.8
+15.0
-8.5
+1.5
+11.4
-5.8
+8.0
Z2
R-18
-23
-30
-31
-34
-35
-36
-38
-39
-41
-46
-47
-48
-49
-50
-51
-52
-54
-61
-62
-63
-65
-66
-67
-68
-72
-73
-77
-79
-80
21.2
4.4
0.4
7.3
17.7
49.0
111.0
84.7
177 .5
67.2
28.2
135.0
29.2
162.0
90.5
9.0
30.3
88.5
32.6
20.3
46.3
112.5
96.0
117.0
225.0
72.2
2.2
130.6
33.7
64.0
n=30
+ 150 . 7
2,165.0
-------
- 95 -
Then a statistical analysis gives:
Z = +5.02,
%2 = 3.2,
Standard Deviation,
S= J ~ Z2 - (z) 2 = 6.85
n
%S=4.5
This difference. is significant, as at 95% confidence limits (using
Studentized statistics):
Deviation
S
t . r;;;
= Z~ .05V n-l =
3.2 + 1. 7%
-------
- 96 -
APPENDIX IV
EQUILIBRIUM PREDICTIONS OF POSSIBLE REACTIONS
IN DESULFURIZATION
A typical flue gas contains about 15 vol. % C02' 1-10% H20, (for
coal, oil), 3% 02 and 0.15 to 0.30% S02. Possible reactions that might
occur can be obtained from equilibrium calculations. Here we shall only
consider whether an individual reaction should proceed or not.
IV-I. Hydroxide Formation
Displacement of carbonate by water vapor to form hydroxides
should not occur as the amount of C02 in the flue gas is large enough
to prevent displacement by H20 at all temperatures up to the carbonate
dissociation temperatures, i.e., reaction 1\ is always displaced to the
left.
cac03(s) + H20(g)
,
\.
Ca(OH)2(s) + C02(g)
(1)
IV-2. Carbonate Dissociation
The carbonates dissociate as follows:
~
CaC03(s) ~ CaO (s) + C02(g)
(2)
The dissociation pressure, PC02' will increase with temperature
as shown in Figure A-1. Whenever the prevailing flue gas partial pressure
of C02 is greater than the equilibrium PCOZ' CaO formation will be inhibited.
At approximately 1450°F for CaC03 and 680°F for MgC03, the PC02 will equal
the partial pressure of C02 in the flue gas. Further heating will liberate
C02 with the attendant oxide formation. It was noted by Battelle (~)
that the dissociation temperature of the magnesium carbonate constituent
of the dolomite was slightly higher than that of pure magnesium carbonate
(730°F vs. 680°F). This temperature difference is caused by the weak
chemical bonding of the carbonates in the dolomite as illustrated by
Equation 3:
CaC03 'MgC03(s) ~ CaC03 'MgO(s) + C02(g)
(3)
Such bonding will also affect the magnesium carbonate reactions
described in later sections. Since these predictions are based on a pure
magnesium carbonate reactant, it should be recognized that the results
will be slightly modified for bonded magnesium carbonates.
-------
+2.0
+1.1
N
o
u
p..
()() .
o
H
- 2.0
-3.0
-4.0
- 97 -
Figure A-I
Equilibrium C02 Pressures for MgCO) and CaCO) Dissociation
0.0
15% C02
------------------
------------
Flue Gas Level
(1 atm total pressure)
-1.0
Q
~
)(
1600
1400
1000
600
800
1200
700
Tempera ture, of
-------
- 98 -
Once the oxides are formed from the carbonates, they will not
react with the water vapor in the flue gas according to Equation (4).
The hydroxide dissociation temperature is 400°F for Mg(OH)Z and 7Z0°F
for Ca(OH)z. Below these temperatures, the results of Section IV-l are
applicable, and Equation {I) is driven to the left.
------\
caO(s) + HZO(g) ~
~a(OH) Z(s)
(4)
From these observations, it is apparent that the COZ and HZO
levels present in flue gases permit dolomite materials to exist only as
carbonates or as oxides. Magnesium and calcium carbonates dissociate
at approximately 680°F and l450°F respectively, under a "usual" flue gas
COZ partial pressure equal to 0.15 atmospheres.
IV-~ SOZ Reactions in the Carbonate Phases
Magnesium and calcium carbonates are each capable of forming
sulphites and sulphates with the SOZ in the flue gas below their dissociation
temperatures.
IV-3.1.
Sulphite Formation
SOz can directly displace the COZ in the carbonates to form
the sulphites.
CaC03(s) + S02(g) ~' CaS03(s) + COZ(g)
(5)
Figure A-Z indicates that these displacement reactions
proceed at temperatures less than 840°F for the CaC03 and 460°F
MgC03 when the SOZ cQncentration is at a 0.Z5% level. However,
90% of the SOZ removed from the flue gas, the maximum allowable
for these reactions become 570°F and 3Z0°F, respectively.
will
for the
with
temperatures
Equation 5 moves to the right even though the COZ concentration
in the flue gas is assumed to be 600 times greater than the SOZ concentration
(15 vol. % C02' .OZ5% SOZ). This occurs because SOZ is far more acidic
than C02' Since the displacement of a weak acid with a strong acid can
be expected to be exothermic, the reaction should proceed at low SOZ
concentrations at lower temperatures. Figure A-2 shows that this expectation
is correct.
Once the sulphites are formed they have a strong tendency to
become oxidized to the sulphates by the low (0.1 - 5%) oxygen concentration
in the flue gas as shown in Figure A-3. However, at relatively low
temperatures their reaction rates could be relatively slow.
IV-3.Z.
Sulphate Formation
The SOZ and 0z concentrations in the flue gas can jointly
displace the COZ in the carbonates to form directly the sulphates
according to the equation:
CaC03(s) + SOZ(g) + l/Z 0z ~ CaS04(s) + COZ(g)
(6)
-------
N
o
CJ)
~ -4.
bO
o
H
-1.
-2.
-3.
- 5.
-6.
Figure A-2
CONVERSION OF CALCIUM AND MAGNESIUM CARBONATES TO THEIR
SULPHITES AT A CO PARTIAL PRESSURE EQUAL TO 0.15 ATMOSPHERES*
2
--- - ~- - - - -- - - - - --
Original Flue Gas Level - 0.25% 802
(1 atm total pressure)
-------------~-----------------
------------
Final Flue Gas Level - 0.025% 802
------------------------
(1 atm total pressure)
~
~
')I. !
Co
<
400
300
200
Temperature, of
*Calculations given by Erdos, E., Collection Czechoslov. Chern. Commun, 12, 2152 (1962).
-------
.0.0
.2.0
.4.0
.6.0
.8.0
-10.0
N
-...
r-4
C'( .12.0
o
Po.
bO -14.0
o
...:I
.16.0
-18.0
- 100 -
Figure A-3
Equilibrium 02 Pressures Required
for the Oxidation of Su1fites to Sulfates
p 1/2
°2
(3 x 10-2 atm) Flue Gas Level
~---------------------------------
- 26.0
- 28.0
- 30 .0
2200
1500
1000
Temperature, of
I
500
-------
0.0
-2.0
- 101 -
Figure A-4
THE EQUILIBRIUM PARTIAL PRESSURES OF S02 AND 02 FOR SULFATE
FORMATION FROM MAGNESIUM AND CALCIUM CARBONATES
Flue Gas Level -
PS02 = 0.25 x 10-2 atm
-----------------------------------
-4.0
-6.0
-8.0
1/2
N
.........
M
N
p..,0 -12.0
N
o
~ - 14 . 0
IJO
.3 -16.0
-18.0
- 20 .0
- 22 .0
- 24 . 0
- 26.0
- 28 .0
-30.0
P02 = 3 x 10-2 atm
PCO = 0.1 atm
2
.......
Carbonate Dissociation Temperatures
~
<5'0
<; x
C'~J
rO
'.1
<;(:
-Q..J
~
~
C'o
",J
1600
I
1000
\
I
700
Tempera ture, of
X
.1,/
<
OJ
<
C'O< ,
rO
'.1
<;
tlQ)
~
~C'o
J "
8~i'
"
.1/<
~J'
I
400
-------
Figure A-5
THE EQUILIBRIUM PARTIAL PRESSURES OF S02 AND 02
FOR MAGNESIUH AND CALCIUH SULFATE DECOMPOSITION REACTION~
-1.0
-2.0
1/2
Log PS02'P02
-3.0
Original Flue Gas Level
i-'
a
N
~& PS02 = 0.25 x 10-2 atm
o
-4.0 ~ P02 = 3 x 10-2 atm
--------~-----______I~~!~~~~~~~----~~X-----------
0. PS02 = 0.25 x 10-3 atm J'Q if
~
-------
- 103 -
1/2
As indicated in Figure A-4, the partial pressure product PS02,P02
is sufficiently high in the furnace to permit these conversions to occur
at temperatures up to the dissociation temperatures of the carbonates,
IV-4,
S02 Reaction in the Oxide Phase
S02 and 02 can similarly react with the oxides to form the
sulphates according to the equation:
CaO(s) + S02(g) + 1/2 02(g)~ CaS04(s)
(7)
Figure A-5 shows that when 90% of the S02 has been removed the
d P P 1/2.
partial pressure pro uct S02' 02 1n the flue gas is still sufficiently
high to cause sulphate formation at temperatures up to 2200°F for CaO
and l500°F for MgO. At temperatures lower than the carbonate dissociation
temperatures (680°F for MgC03 and 1450°F for CaC03) some of the oxides
would form intermediate carbonates and sulfites which would then be converted
to sulphates as described in Section IV-3.
IV-5.
Summary of Reaction Regions
Table A-l lists possible reactions which could take place in
desulfurization over the various temperature ranges from 200°F to 2000°F.
TABLE A-I
POSSIBLE REACTIONS FOR THE OXIDES AND CARBONATES OF
MAGNESIUM AND CALCIUM IN FLUE GASES FROM 200°F TO 2000°F
Possible Reaction Temperature Range
CaC03 + S02 ---+ CaS03 + C02 200°F - 570°F
CaS03 + 1/2 02 --. CaS04 20fF - 2000°F+
CaC03 + S02 + 1/2 02 --. CaS04 + C02 200°F - 1450°F
Cao + S02 + 1/2 02 -. CaS04 1450°F - 2000°F
XgC03 + 502 --. MgS03 + C02 200°F - 320°F
XgS03 + 1/2 02 ----to MgS04 200°F - 2000°F+
XgC03 + S02 + 1/2 02 ----to Mgs04 + C02 200°F - 680°F
~(gO + S02 + 1/2 02 --.. MgSo4 680°F - 1500°F
-------
- --- --
Al'Pt:NUIX v
GIBBS FREE ENERGY OF FOB}~TION OF RELEVANT SPECIES
The free energies used in calculating equilibria constants and compositions in
Section 4.1 and Appendix V are listed below. The basic reference for this data
is "JANAF Thermochemical Data," Dow Chemical Co., Midland, Michigan, supplemented
by Wheelock and Boylan (see Reference ~).
FREE ENERGY OF FORMATION VARIATION WITH TEMPERATURE
°
11 Gf (kca1/mo1e)
500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600
440 620 800 980 1160 1340 1520 1700 1880 2060 2240 2420
OK
@
Compound of
CO
C02
CS
CS2
COS
CH4
H20
H2S
S2
S02
S03
S20
MgO
MgC03
Mg(OH)2
MgS
MgS03
MgS04
CaO
CaC03
Ca(OH)2
CaS
CaS03
CaS04
-39.31
-94.46
30.79
5.04
-45.59
-5.49
-51.16
-10.13
8.58
-71. 79
-81. 9 2
-26.59
-218.19
-235.68
-177.30
-79.68
-201. 07
-255.506 -246.09
-138.25 -135.80
-249.96
-191. 76
-115.00
-238.79
-291.00
-37.14
-94.40
34.60
8.41
-43.73
-7.85
-52.36
-9.60
11.81
-71. 92
-84.31
-24.90
-130.76
-232.28
-184.53
--80.48
-41. 47
-94.51
27.07
1.84
-47.36
-3.05
-49.92
-10.54
5.53
-71. 56
-79.44
-28.10
-125.62
-219.14
-170.12
-78.78
-194.33
-236.65
-133.35
-243.96
-184.77
-112.54
-231. 51
-280.10
-43.61 -45.74 -47.96 -49.96 -52.05 -54.13 -56.19 -58.24 -60.28
-94.56 -94.60 -94.63 -94.66 -94.68 -94.70 -94.72 -94.73 -94.74
22.12 19.95 17.79 15.64 13.50 11.37 9.25 7.14 5.04 I
-3.85 -4.01 -4.16 -4.33 -4.48 -4.64 -4.80 -4.96 -5.11 ~
-50.37 -50.62 -50.86 -51.10 -51.35 -51.59 -51.83 -52.07 -52.30 ~
-0.53 2.03 4.63 7.25 9.89 12.54 15.20 17.86 20.52 I
-48.65 -47.35 ~46.04 -44.71 -43.37 -42.02 -40.66 -39.38 -37.93
-12.16 -11.01 -9.84 -8.67 -7.49 -6.31 -5.13 -3.95 -2.77
--------------------------------------- 0 -------------------------------------
-72.57 -70.82 -69.07 -67.33 -65.58 -63.84 -62.10 -60.37 -58.64
-78.21 -74.23 -70.26 -66.31 -62.36 -54.43 -54.50 -50.59 -46.69
-32.09 -30.55 -29.02 -27.50 -25.97 -24.45 -22.93 -21.41 -19.89
~123.06 -120.51 -117.78 -115.00 -112.22 -109.44 -106.18 -101.20 -96.25
-212.66 -206.25 -199.72
-163.02 -155.99 -148.86
-79.13 -76.70 -74.08 -71.41 -68.73 -66.05 -62.89 -58.01 -53.15
-228.52 -217.69 -206.76 -195.85 -185.01 -174.24 -163.06 -150.24 -137.52
-130.90 -128.50 -126.05 -123.60 -121.05 -118.45 -115.05 -113.30 -110.70
-238.01 -232.16 -226.18 -220.36 -214.36 -208.45 -201. 77 -196.73 -190.84
-177.85
-110.01 -107.60 -105.30 -102.00 -100.50 -98.00 -94.00 -93.00 -89.60
-225.47 -216.72 - 20 7 .97 -199.63 -191.18 -182.64 -173.35
-269.10 -258.50 -248.00 -236.60 -227.05 -216.75 -205.05 -199.90 -186.30
-------
- 105 -
APPENDIX VI
RESIDENCE TIME AVERAGING FOR THE EFFECTIVE
RATE CONSTANT IN A FLUIDIZED BED
The analysis of the reaction of S02/02 with limestone in the fluid
bed reactor has thus far been limited to consideration of a very specific
idealized flow pattern, namely plug flow. Real reactors never fully satisfy
the flow requirements of such idealized models. Deviations from ideality can
be cavsed by channeling of the fluid through the vessel, by recycling of
fluid within the vessel, or by the existence of stagnant regions of fluids in
the vessel. The experimental data obtained in our laboratory data showed
evidence of a large deviation from ideality, which was treated by assigning
a contacting factor which related the actual performance of the reactor to
that which would have been expected under conditions of ideal flow of the
fluid.
The prediction of the performance of the commercial desulfurization
reactor will also be very dependent on how ideal the flow distribution is at
the velocity and bed height used. Lacking sufficient information to estimate
the ideality of flow in the commercial unit, the following prediction of the
commercial unit performance will be limited to considering the maximum
performance that can be expected. This assumes ideal flow conditions and the
applicability of the plug flow model used in analysis of the laboratory reactor
data. This approach is equivalent to setting the contacting factor used in
analyzing the laboratory reactor data at unit value. This assumption is
clearly incorrect, but does serve to establish the limiting case for perfor-
mance of the commercial unit.
Since it is not possible to attain complete knowledge about the
commercial reactor system, let us be less ambitious and ask how we can at
least describe the system adequately enough to yield information useful in
the design of the unit. The commercial unit, unlike the laboratory reactor,
will be in a dynamic state as far as solids entering and leaving the reactor~
Fl
w
F2
Where Fl
F2
W
feed rate of uncalcined stone to the reactor - Ib moles/hr
removal rate of sulfated product - Ib moles/hr
bed hold up - Ib/moles
-------
- 106 -
The feed and removal rates, Fl and FZ' are adjusted to maintain a
uniform quantity of material, W, in the fluidized bed. The backmixing of
solids which occurs in the fluid bed results in an age distribution for the
solids, i.e., the solid particles being removed from the bed in the overflow
stream FZ have been in the bed for various lengths of time. This variation
in the residence time of the solids is of course reflected in their degree
of CaO utilization.
The kinetic data obtained in Section 3.3. presented the reaction rate
constant as a function of particle utilization,X. In order to use this rate
information it is necessary to know how the utilization of the particle
changes with the time the particles are in the bed, (the particle age or par-
ticle residence time 8).
Assume an increment of fresh particles of weight, W, added to this
fluid bed at time e = O. The rate of S02 uptake by these particles is
+ dNSO
2
r.W.e
S02
(8)
de
Where
NSO
Z
amount of SOZ (mass)
e
time particle is in bed
r
rate of S02 uptake per unit mass of particles
average S02 concentration seen by particles.
eSO
2
At steady state,eSOZ is constant during the residence time of the particles
at some average between SOZ in and S02 out.
Expressing equation
(8)
in terms of the rate constant k
+
dNSO
2
k(X)
de
fs
W . eSO
2
(9)
--
Here ~ bulk density of solid and k has been written k(X)
that it is a part of the particle utilization eX).
to indicate
The particle utilization X
to the S02 capacity of the charge.
is the fraction of 802 uptake compared
X
=
wt 802 uptake
capacity for S02
=
NS02
W .0(
(10)
where 0(
=
mass capacity of S02 per unit mass of stone.
-------
- 107 -
Differentiating equation (10)
dXB
=
dNS02
W .0<
(11)
Substituting eq.
(11) into eq.
(9) and separating variables,
=
CS02
Iso;
dX
. d8
(12)
k(X)
On integration between x=o and X at time 8, we obtain after
rearrangement:
s:
dX
k(X)
CS02
fs'~
8
(13)
The integral in eq. (13) is done numerically to various values
of X and the particle residence time to obtain that X given by:
8 = (area under the curve from
X=O to X)
'~'cA
(14)
CSO
2
A logarithmic mean concentration for CSO is the appropriate averaged con-
centration. 2
In order to predict the limiting performance of the commercial unit
it is necessary to know the age distribution of the reacting solids. It has
been found that the following particle age distribution function, 1(8), is
quite accurate for perfectly mixed solids*.
1(8)
=
1
e
exp (-8/8)
(1'; )
Here 8
is the average particle residence time defined as
8
W
F
(16)
and 8 is the particle residence time.
*E. B. Nauman and C. N. Collinge, Chern. Eng. Sci. 23, 1317 (1968)
-------
- 108 -
The analysis of the reaction rate constant presented earlier showed
that the rate constant, k(X), depended on the degree of CaO utilization,
X. The degree of CaO utilization, X, for a particle injected into the fluid
bed reactor was shown above to depend on the length of time the particle was
in the reactor, S. These considerations suggest that the reaction constant,
k(X) for a given sorbent particle will depend on the time that this particle
has been in the reactor. The expression for this dependence of the rate con-
stant on the particle residence time, k(S), can readily be obtained from the
expressions previously determined for the dependence of the rate constant on
the degree of CaO utilization, k(X), and the dependence of the degree of
CaO utilization on the particle residence time XeS).
The desu1furization performance for the commercial reactor will
be made up of the contribution of all the particles in the fluid bed, each
with its own particular residence time. The average rate constant for the
commercial reactor, k, is obtained by integrating the product of the expres-
sions for the variation of rate constant with particle residence time, k(8),
and the age distribution for the particles, I(~) over all values of particle
residence time, S. cP
k = So k(8) .1(8) dS
(17)
This expression is solved numerically.
The desu1furization performance can now
rate expression previously described for the plug
specifying the operating parameters
CH
C' =
Co
be calculated, using the
flow reactor model, by
H
exp (-k . Y . ~ )
U
(18)
where the contacting factor, Y, is the factor representing the departure
from ideal flow conditions in the commercial reactor.
The overflow stream from the fluid bed, F2, contains, as mentioned
earlier, a wide distribution of particle ages, each with a degree of CaO
utilization dependent on its history in the bed. An analysis similar to that
used above leads to an expression for the average CaO utilization, X, in
the commercial reactor.
x - l""X(8)oI(8)d8
(19)
-------
- 109 -
APPENDIX VII
ECONOMIC EVALUATION FOR A ONCE THROUGH FLUID
BED LIMESTONE DESULFURIZATION PROCESS USING SMALL PARTICLES
A coal-fired, 800 MW power plant burning a 3% sulfur fuel was
selected as the basis for evaluating a fine particle fluid bed desulfuri-
zation process. Process designs were based on experimental data obtained
with the N-1337 dolomite. The designs call for operating the bed at
about l600°F and fine-grinding the dolomite before charging it to the
fluid bed. Under these conditions, our experiments show that calcination
will be rapid and active sorbent will be produced. The use of fines
necessitates a once through system, as fly ash could not be separated
from the sulfated dolomite, which would render a regenerative system in-
operable.
Economic evaluations have been made for several cases designed
to remove 80% of the S02 from the flue gas while utilizing about 85% of
the CaO portion of the dolomite (the CaO represents about 28 wt% of the
uncalcined dolomite for N-1337).
The best design from a combined engineering
of a 73 foot diameter bed operating at a 24
using 60 to 100~ particles. Four stages of
drop through the system is estimated at 48"
and cost standpoint consisted
ft/sec superficial velocity and
cyclones are used; pressure
H20.
A high superficial gas velocity was chosen to keep bed size down. Never-
theless, a large fluid bed must be used to handle the huge volume of flue
gas generated by the 800 MW plant (85 x 106 SCFH). At these high veloci-
ties, there would be virtually total solids entrainment by the gas passing
through the reactor. A minimum of three, preferably four, stages of
cyclones would be required to collect these solids and return them to
the bed. These cyc~ones are necessarily quite large to handle the huge
volume of flue gas. The designs call for eight primary cyclones
(each of which is 22 ft in diameter) discharging into 24-secondary cyclones
(each of which is 12 ft in diameter) which in turn discharge into 72 tertiary
cyclones (each of which is 8 ft in diameter). Taken together, the investment
for these cyclones constitutffiabout 45% of the total investment.
Four large induced draft fans are needed to overCOme the added
pressure drop of the desulfurization system. With these fans discharging
into the bottom of the stack at a positive pressure, the pressure in the
fluid bed itself would be lower than atmospheric.
-------
- llO -
TABLE A-2
ECONOMIC EVALUATION OF FINE PARTICLE
FLUIDIZED BED DESULFURIZER FOR 800 MW POWER PLANT
Investment
Desu1furizer (includes Foundation)
Ducts
Induced Draft Fans, Buck Stags
Feed and Disposal System
Limestone Crusher
Tota 1
Contigency Fees
Working Capita 1
Tota 1
Yearly Operating Costs
Investment)
Capital Charges (14%
Plant Maintenance
Labor and Supervision
Payroll Overhead
Plant Overhead
Power
Fuel
Limestone
Therma 1 Effect
Total
$/Ton coal
60%
Load Factor
$2,050,000
732,000
80,000
32,000
407,000
603,000
7,000
657,000
o
$4,624,000
$2.90
Cost
$ 7,722,000
400,000
1,270,000
770,000
300,000
$10,462,000
3,650,000
529,000
$14,641,000
90%
Load Factor
$2,050,000
732,000
83,100
33,000
'407,000
905,000
10,000
980,000
-10,000
$5,273,000
$2.20
For the hetero-reactor system, assuming a one-foot settled bed
height can achieve 90% desulfurization with stoichiometric sorbent
addition, the cost savings shown below would be obtained for a 800 MW
plant.
Capital Charges (fans and drives)
Power Cost (to overcome 6P)
Maintenance Charge
Total Annual Cost Savings, $
With a 90% Load Factor
41,000
140,000
~OOO
$195,000 (89/Ton Coal Burned)
-------
- 111 -
APPENDIX VIII
FLUIDIZED BED DESULFURIZATION PROCESS DETAILS
VIII-I. Cos ting Procedures
Criteria used to estimate purchase prices or on-site construction
costs of special process equipment are listed below. Using Lang's factors
as outlined in the text section 6, the final direct cost of construction
was obtained from these prices.
TABLE
A-3
PURCHASE PRICE ESTIMATION PROCEDURES
Item
Method
Process Equipment-Installed Costs
1. Steel Shells
2.
Insulation
a.
$1.OO/lb carbon steel
$1.80/lb stainless steel
$20/ft2 for 9" thick insulation -
6" firebrick, 3" ceramic
$15/ft3 firebrick
25~/stamped, snap-in gas inlet
button
$27/ft2 grid surface area ~ for
vertical rods to deter bubble
growth
a.
b.
b.
3.
Grid
a.
b.
Process Equipment-Purchase Price
1. Electrostati c Precipitator
2. Heat Exchangers
3. Compressors, Pumps, Fans
a. Small Units
b. Large Fans
$1.50/CFM gas throughput
Chern. Eng. News Cost Files, Jan '58
Chern Eng News Cost Files, Jan '61
Direct Manufacturer Estimates
VIII-2. Unusual Equipment - Details, 200 MW
Plant, Regenerative Process
VIII-2.1. Desulfurizer
Specifications: 3/8" thick steel shell, 50' overall height;
insulation protection of shell and grid underside against fly-ash impingement -
6" firebrick, 3" ceramic undercoating; 48.5 ft I.D. fluid bed x 15' shell
height; 70' ft I.D. disengaging section x 12' shell height; tapered body
to duct work; 1" thick stainless steel grid, with 750 gas inlet buttons
(appr. 1" dia) /ft2 grid surface area, supported by girders for strength.
Maximum Flow Conditions: Limestone particles, 1000-5000jl fed
at 3 tons/hr; 90 x 106 ACFH flue gas @ l600°F; U superficial = 15 ft/sec
in bed, U superficial = 7.2 ft/sec in disengagement section, U = 150 ft/sec in
por ts; Ho 1'/15 inches; ~ P I!/ 20 inches H20.
inle t
! i
-------
- 11Z -
VIII-Z. Z. Ducts
Specifications: 3/8" thick steel shell with insulation protection,
17' x 17' square ducts connecting boiler with the absorber; length 150'.
Maximum Flow Conditions:
Ugas =100 ft/sec @ l600°F.
VIII-].3. Regenerator
Specifications: 3/8" thick steel shell, 35' overall height,
10.5' I.D.; insulation protection of shell as in absorber; 3' firebrick
arch grid perforated with straight bore holes; 6.2 ft diameter cyclones
with return tubes (design, ZO-86 in Perry's Chern. Eng. Handbook). Multiple
particle downcomers, 1 ft2 flow area
Maximum
U ~ 11 ft/sec; 3.8
CO/HZ, ZO% COZ/H20
CaS04/CaO into bed
coal boiler feed).
Flow Conditions: 6.6 x 105 SCFH @ ZOOO°F, reducing gas,
tons coal/hr partially combusted with air to yield 10%
gas; Ho = 3 feet; 6 P = 4Z inches HZO; recirculate 0.17
at 42 tons/hr, send out 0.05 CaS04-CaO (for 3.5% sulfur
VIII-Z.4. Induction Fans
Specifications: 2 fans to over.come pressure drop through
desulfurization reactor, handling 34 x 10° ACFH gas. At 65% machine efficiency,
Z600 horsepower to the motors is required.
VII 1-3. Unusual Equipment - Details, 1000 MW
Plant, Regenerative Process
VIII-3.l.
Desulfurizers
Five units, each of specifications and flows described in VIII-Z.l.
VIII-3.2.
Ducts
thick
17' x
total
Specifications: One conduit, 38' x 38', insulated, 3/8"
shell, to connect to boiler at each end of absorbers; five ducts,
17', split from this conduit to carry flue gas to the absorbers;
length of small ducts, 900 ft.
Maximum flow conditions:
Ogas /V 100 ft/sec @ l600°F.
VIII-3.3.
Regenerator
Specifications: 3/8" thick steel shell, Z4' I.D., 40' overall
length; 4' firebrick arch grid, perforated with gas inlet holes, 8' diameter
cyclones.
Maximum flow conditions: 33 x 105 SCFH @ ZOOO°F reducing gas,
U ~ 11 ft/sec, 19 tons coal/hr partially combusted with air to yield 10%
CO/HZ, ZO% COZ/HZO gas; Ho = 3 feet, 6 P = 4Z inches HZO; recirculate 210
tons/hr of CaS04-CaO.
-------
- 113 -
VIII-3.4.
Induction Fans
6 Specifications:
170 x 10 AFCH gas.
2 fans to overcome pressure drop, handling
-------
Unclassified
- 114 -
~SS~
SecuritY Classiflcation-'T'hi s Pa2:e
DOCUMENT CONTROL DATA. R & D
(Saeu,lIy cla.,"'eallon 01 Ima, body 01 abal,ael and 'nda,,'n, _nol.IIOII ...ual ba .nt.red ",h.n the 0118'." re_tt I. clu.W.d,
1. ORIGINATING ACTIVITV (CorpoNf. .utltor) 28. IIEPOIIT Sa:CUIIITY CLA"II'ICATION
Esso Research & Engineering Company Unclassified
P.O. Box 8 2b. GROUP
Linden, New Jersey 07036
3. REPORT TITLE
Fluidized Bed Studies of the Limestone-Based Flue Gas
Desulfurization Process
.. DESCRIPTIVE NOTES (7'yp. 0/,_, _d Inclu./v. d.t..)
Summary Report - May 15, 1967 - August 27, 1969
s. AU THDR(SI (",.t nam., mlddl. /nm.I, '..t n.....)
Alvin Skopp. John T. Sears, Rene R. 13ertn,.nrJ
a. REPORT DATE 7.. TOTAL NO. 01' PAGES rb. NO. ~~ IIEI'S
August, 1969 114
Ia. CONTRACT OR GRANT NO. "..OllIGINATOR'S REPORT NUMBERCS,
PH 86-67-130
b. PROJECT NO. GR-9-FGS-69
e. lib. ::'J.H:':"~~PORT NOca, (An,. 0111., n_ban ...t...,. b. ."',,.ad
d.
10. DISTRI8UTION STATEMENT
Distribution of this report requires prior approval of the National Air
Pollution Control Administration
II. SUPPLEMENTARY NOTES 12. Sponsoring Act~v~ty
Process Control Engineering Program
National Air Pollution Control
Administration
13. A8STRACT
A conceptual design of a fluidized bed flue gas desulfurization
process has been developed based on the ability of coarse lime particles
to react with and remove S02 from combustion flue gases in a fluidized bed.
Reduction of the sulfated lime particles formed in such a process has
been demonstrated to provide reactivated particles. These particles can
again react with the sulfur dioxide in a flue gas to form sulfates. The
effluent from reduction has a high S02 concentration and is suitable for
the production of sulfuric acid. Thus, a cyclic, regenerative flue
gas desulfurization process can be envisioned that produces a saleable
by-product to offset processing costs.
Experiments were performed to determine and improve the reactivity
of different limestone and dolomitic sorbents in a fluidized bed, and to
define conditions for regeneration of the sorbents. A conceptual design
of a system was formulated from these data. A coarse particle, high-gas-
velocity fluidized bed operating at l600°F was the basis of the best design.
Regeneration was carried out with a producer-type.gas at about 2000°F.
Preliminary economics indicated that the sys tern was not applicable in
presently operating boilers but a grass-roots power plant might operate
competitively. However, other general considerations, such as the large
size and location of the equipment in the boiler train, make a commercial
process appear unlikely in its present form. Adaption of this process
to fluidized bed combustion schemes is quite promising, and future process
develoDment should Droceed in this direction.
ESSO
1473
Unclassified
Security Cla..Ulcation - This Page
------- |