SEPA
United States
Environmental Protection
Agency
Industrial Environmental Research
Laboratory
Research Triangle Park NC 27711
Research and Development
EPA-600/S7-82-026 June 1982
Project Summary
Mechanisms <*£ Dry SO2
Control Processes
Cathy Apple and Mary€.
..,
This report identifies and eyptua&s
the postulated physical^rlanges, .
chemical reactions, antf" reaction^X
mechanisms involved in two dry flue*'
gas desulfurization (FGD) technolo-
gies: line spray drying and dry injec-
tion of sodium compounds.'"^pftjcs
covered are: (1) chemical reactions
and physical changes, (2) proposed
reaction mechanisms and mathemati-
cal models, (3) process parameters
which affect the reactions and rate
controlling steps, and (4) needs for
additional data to verify the proposed
mechanisms and the effects of the
various process parameters. The infor-
mation used in developing this report
was obtained from a review of pub-
lished articles and technical papers.
The reactions involved in lime spray
drying are gas-liquid phase reactions.
SOz removal depends on the moisture
content of the drying lime slurry
droplet. Initially, the moisture content
of the droplet is high and the rate of
reaction is controlled by the diffusion
of SO2 to the surface of the slurry
droplet. It appears that the greatest
amount of SOa is removed at this time,
and Babcock and Wilcox has mathe-
matically modelled SOa removal
during this period. As the moisture
content of the droplet is reduced by
evaporation, the dissolution of
Ca(OH)2 into ions becomes the rate
limiting factor. SO2 removal by the
spray-dried solids is limited by the
dissolution of Ca(OH)a and by the
diffusion of SO2 through the CaSO3 •
1/2H2O product that has precipitated
on the surface of the lime particle.
y
The reactions involved in the re-
moval of SO2 by the dry injection of
finely ground sodium compounds are
gas-solid phase reactions. Two steps
are involved in the removal of SO2: (1)
NaHCO3 is thermally decomposed to
Na2CO3 (producing small pores in the
sorbent particles, which increase the
available surface area and the re-
activity of the sorbent); and (2) the
SO2 reacts with Na2CO3 to produce
Na2SO3 (beginning at the surface of
the particle and moving inward,
leaving behind a layer of reacted ash
which tends to plug the pores that
were formed by thermal decomposi-
tion, reducing the reactivity of the
sorbent). Therefore, it has been
postulated that initially the rate of
reaction is controlled by the thermal
decomposition of NaHCO3 to NaaCOs
and then becomes controlled by the
diffusion of SO2 through the ash layer.
Mathematical models describing SO2
removal for these cases can be derived
from the classical unreacted core
model.
This Project Summary was devel-
oped by EPA's Industrial Environ-
mental Research Laboratory, Research
Triangle Park, NC, to announce key
findings of the research project that is
fully documented in a separate report
of the same title (see Project Report
ordering information at back).
Summary
This report identifies and evaluates
the postulated physical changes, chem-
ical reactions, and reaction mechanisms
involved in dry flue gas desulfurization
-------
(FGD). Basically, this work is a review of
the available published theories describ-
ing the reactions involved in lime spray
drying and dry injection of sodium
compounds. The topics covered in this
report are:
• The physical changes and chemical
reactions that occur in the dry
scrubbing systems
• Proposed reaction mechanisms
and mathematical models.
• Process parameters which affect
the reactions and the rate controll-
ing steps
• Additional data needed to verify or
expand on the proposed mechan-
isms and the effects of the various
process parameters
The information used in developing
this report was obtained from previously
published articles and technical papers
Much of the information was obtained
from papers presented at the October
1980 EPA Symposium on Flue Gas
Desulfurization.
The information on spray drying is
more complete and consistent than the
information on dry injection. Spray
drying has been investigated more
thoroughly and is presently being
applied to commercial-scale units. Dry
injection is still being investigated on
the pilot scale, although recent tests
have been conducted on a system
designed to treat 20 MWe equivalent of
flue gas.1
Note that the term stoichiometry used
in this report represents the moles of
fresh sorbent introduced to the system
divided by the moles of SC>2 entering the
system. For lime, the moles of sorbent is
based on the equivalent males of CaO
introduced, and for sodium compounds
the moles of sorbent is based on the
equivalent moles of NazO entering the
system. This method of describing dry
scrubbing stoichiometry is commonly
used by system vendors, and the results
they report are based on this definition.
The conventional FGD definition for
stoichiometry in wet systems is the
moles of sorbent introduced to the
system divided by the moles of SC>2
removed by the system. Compared with
the conventional FGD definition of
stoichiometry, the definition used by dry
scrubbing vendors makes the system
look more favorable because it does not
account for SC>2 removal efficiency.
However, note that the dry scrubbing
definition of stoichiometry is the same
definition that is conventionally used in
chemistry (i.e., the moles of A needed to
react completely with B without ac-
counting for efficiency).
Lime Spray Drying
Spray drying involves contacting SO2-
laden flue gas with an atomized sorbent
slurry in a spray dryer The SO2 is
absorbed by the slurry droplets, while
the droplets are dried by the hot flue gas.
An ESP or fabric filter is used to collect
the dried solids exiting the spray dryer.
Information on dry FGD with lime
spray drying was evaluated to determi ne
the reactions and reaction mechanisms
involved in the removal of SC>2. The
results of pilot-scale lime spray drying
tests were examined to establish the
reaction mechanisms and rate con-
trolling steps. Proposed qualitative and
quantitative models describing the
spray drying process were also eval-
uated. Additionally, the important
process parameters affecting the spray
drying reactions were qualitatively
evaluated to determine their influence
on the rate controlling steps of the
reaction and on the degree of S02
removal achievable. Finally, data gaps
in published literature were identified.
These data gaps prevent complete
verification and expansion of the
proposed models.
Qualitative and Quantitative
Reaction Models for Lime
Spray Drying
Models developed for lime spray
drying must take into account the
simultaneous physical changes and
chemical reactions that occur in the
spray dryer. The overall chemical
reaction takes place between the
gaseous S02 and the dissoluted lime
sorbent to form calcium sulfate and
sulfite salts:
(1 )
Ca(OH)2(s)
CaSOa • 1/2H2O(S, + 3/2 H2Om
CaSO3 • 1/2H2O(Si + 1/2O2 + 3/2
HaOm^CaSO • = 2H2Ois, (2)
Simultaneously, the reaction surface is
physically changed as the slurry droplet
dries.
The reactions that occur are basically
gas-liquid phase reactions, because
moisture must be present for the
reactions to occur. Thus, the moisture
content and composition of the slurry
droplet dur.ing and after drying are
extremely important in determining the
nature of the rate controlling steps. The
slurry droplet is typically composed of
many small particles of porous solid
lime in an aqueous medium. As the
slurry droplet dries, water evaporates
from its surface, leaving behind a solid
agglomerate composed mainly of sulfite
and sulfate product solids and unreacted
lime. The dried solids also may contain
significant amounts of residual moisture,
depending on the length of drying time
and the approach to saturation at the
dryer outlet.
The most widely accepted qualitative
model suggests that the chemical
reactions that occur are intimately
linked to three stages of droplet drying.2
These drying stages have been char-
acterized as:
• Constant rate period. The greatest
amount of water is evaporated from
the droplet during this period, and
drying is controlled by the rate of
diffusion of water vapor from the
saturated surface of the droplet.
• First falling rate period. During this
period, the lime particles in the
slurry droplet begin to touch one
another due to the loss of moisture,
and drying is controlled by the rate
of diffusion of water vapor from an
unsaturated surface.
• Second falling rate period. Drying
during this period is controlled by
the rate of diffusion of moisture
through and around the tightly
packed lime particles.
During the constant rate stage of
drying, the rate of SO2 removal appears
to be controlled by the diffusion of the
S02 in the flue gas to the surface of the
slurry droplet. The reaction that occurs
between the gaseous SCbandthewater
in the droplet is characterized by the
following equation:
S02(g)
S02
:(aql
(3)
As evaporation proceeds during the two
falling rate periods, less water is
available for the dissolution of the solid
Ca(OH)2 into ions; therefore the dissolu-
tion of the Ca(OH)2 becomes the rate
limiting factor.3
If the spray-dried solids maintain
adequate moisture, further reaction
may continue in the downstream
collection device. This is termed the
pseudo-equilibrium period. Reactions
during this period are limited by the rate
of diffusion of S02 through the CaS03 •
1/2 H20 that has formed on the surface of
the lime particles and through the pore
system of the lime particles.
This qualitative model provides the
basis for development of theoretical and
-------
empirical quantitative (mathematical)
models. Once these models have been
evaluated and verified against test data,
they can be used with confidence to
predict S02 removal under given
conditions
A mathematical model was developed
by Babcock and Wilcox (B&W)4 to
characterize the reactions that occur
during the constant rate drying period.
The qualitative model states that the
rate of reaction is controlled by the
diffusion of S02 to the slurry droplets
during this period. From this assump-
tion the following mathematical rela-
tionship was developed:
r tP i
Tin-Tsat
0.62
(4)
where E = SO2 removal efficiency, frac-
tion
Tsat =Adiabatic saturation tem-
perature of the flue gas, °F*
Tin = Inlet temperature of the
flue gas, °F
tp = Approach to the adiabatic
saturation temperature at
the spray dryer outlet (Tout-
Tsat), °F
Tout = Spray dryer outlet tempera-
ture of the flue gas, °F
Test data obtained by Buell2 at the
Martin Drake Station seem to show that
the model is valid for relatively large
spray dryers (greater than 8,500 acfm),
but it underpredicted S02 removal at
B&W's Alliance Research Center 1500
acfm spray dryer.4 These conflicting
results may be due to the different spray
dryer designs used by Buell and B&W,
or they may be due to "wall effects."
These wall effects involve chemical
reactions between the S02 and the lime
that coats the walls of the spray dryer.
This coating provides sites for additional
S02 removal. There is more wall surface
area per molecule of gas in a small spray
dryer, so that more reaction sites are
available per SOa molecule. Therefore,
under the same test conditions greater
S02 removal would be expected in a
small dryer.
Another mathematical model devel-
oped by B&W correlates S02 removal
efficiency with the stoichiometric ratio
The model states that
E = 1-e"
(5)
'English Engineering units are used, Metric-
English conversion factors are given in the
Appendix
where: E = SO2 removal efficiency,
fraction
SR = Stoichiometric ratio =
moles CaO in
moles SO2 in
K' = Correlation coefficient
The variable K' is a function of the
specific surface of the lime (in ft2 of
surface per Ib of lime) and the residence
time of the flue gas in the spray dryer.
Test data collected by B&W4 and Buell2,
utilizing a close approach to saturation
temperature and S02 concentrations
less than 2000 ppm, fit the above
relationship extremely well. Therefore,
it seems that this general correlation
can be used to predict SO2 removal
efficiency for these operating conditions.
Process Parameters Affecting
the Lime Spray Drying
Reactions
It has been postulated that the spray
drying reaction rate is controlled initially
by the diffusion of S02 into the droplet
and then becomes controlled by the
dissolution of Ca{OH)2 into ions. The
rate of dissolution of Ca(OH)2 changes
as the slurry droplet dries. Initially, Ca++
and OH~ ions saturate the slurry droplet
and are easily dissoluted, but as the
droplet dries dissolution becomes
limited by the amount of moisture left in
the droplet. Therefore, three major
processes can be postulated:
1) Gas-phase SO2 absorption into
the slurry droplet.
2) Ca(OH)2 dissolution in the slurry
droplet.
3) Ca(OH)2 dissolution in the spray-
dried solid.
The parameters that affect these
processes are presented below
Both the approach to saturation
temperatures and the size of the initial
slurry droplet affect mass transfer
during these three processes and are
therefore crucial process variables. The
process variables can be categorized as1
1) Variables affecting SO2 mass
transfer into the slurry droplet—
Approach to saturation temper-
ature.
Size of the slurry-droplet method
of atomization.
Mixing of the gas and droplets.
Residence time of flue gas in the
dryer.
Inlet SCb concentration.
Inlet flue gas temperature.
2) Variables affecting Ca++ mass
transfer in the slurry droplet—
Stoichiometric ratio.
Approach to saturation temper-
ature.
Size of the slurry droplet.
Residence time of flue gas in the
dryer.
3) Variables affecting Ca++ mass
transfer in the spray-dried solids—
Approach to saturation temper-
ature.
Size of the initial slurry droplet.
Slaking.
Stoichiometric ratio.
Off-product recycle.
These variables and their effects are
discussed briefly in the following
sections.
Approach to Saturation
Operating the spray dryer outlet at a
close approach to saturation tempera-
ture increases the residual moisture
level in the spray-dried solids and
affects all three mass transfer processes.
The closer the approach to saturation
temperature, the cooler and more
humid the flue gas becomes. Increased
humidity lengthens the evaporation
period and also increases the amount of
moisture retained by the solids. By
reducing the outlet temperature from
the spray dryer by 10°F (from a 30°F
approach to 20°F), S02 removal ef-
ficiency can be increased by approxi-
mately 10 percent5.
Slurry Droplet Size -
Method of Atomization
The size of the slurry droplets also
affects all three mass transfer processes.
The spray of slurry droplets must be
finely atomized to provide the large
surface area required for rapid S02
absorption. However, the droplets must
be large enough so that they do not dry
out before a satisfactory degree of
reaction has occurred, the reaction
being highly dependent on the presence
of water. (The drying time varies with
the square of the initial droplet
diameter.)4 Typically, 1 ft3 of reagent
properly atomized has a surface area of
18,000 ft2.6 Proper atomization is
extremely important for mass transfer
during the pseudo-equilibrium period,
where the spray-dried solids exit the
spray dryer with an equilibrium moisture
content, because the size of the final
agglomerate is directly related to the
initial droplet size.
-------
Residence Time
The effect of residence time is most
important during the first two mass
transfer processes, SO2 mass transfer
to the slurry droplet and Ca++ mass
transfer in the slurry droplet, because it
determines the amount of time the flue
gas and slurry droplets will be in contact
in the spray dryer. The flow rate of gas
through the absorber must be high
enough for efficient mixing of the gas
and spray, but not so high that a proper
degree of drying cannot occur in the
chamber. The residence time, defined
as the chamber volume divided by the
absorber outlet gas flow volume, must
be controlled to an optimum value so
that high SOa removal rates are ob-
tained. The slower the drying process,
the longer surface moisturewill existon
the droplet, allowing for more reaction.
Stoichiometric Ratio
The Stoichiometric ratio is most
important during the last two mass
transfer processes, those that involve
Ca++ mass transfer. The stoichiometry
determines the amount of lime that will
be present to react with the S02 and
significantly affects SOa removal in
spray drying. The relationship between
SOa removal efficiency andthe Stoichio-
metric ratio (SR) is expressed by the
proposed mathematical model.4
E = 1 -e"
(5)
This model proposesthat SOa removal is
a function of Stoichiometric ratio. Pilot
spray dryer test data is well correlated to
the mathematical model; thus, it would
appear that the model is valid for the
operating conditions used in these
tests.
The obvious method for increasing
SOa removal would be to increase the
lime stoichiometry, but there are
several factors which limit the amount
of lime which can be added. First of all,
the amount of slurry that can be added
to the flue gas is set by energy balance
considerations to ensure the solids will
be properly dried Secondly, the amount
of lime that can be added to the slurry is
limited, because an upper limit is
reached on the weight percent of lime in
the slurry. Finally, the effect of the
Stoichiometric ratio on SOa removal
efficiency begins to level off at SR
values greater than 3.0
Droplet - Gas Mixing
Mixing of the slurry droplets and the
flue gas affectsthe mass transfer of S02
to the droplet. To provide for good
contact between the flue gas and the
slurry droplets, intimate mixing must be
achieved. This requires fineatomization
of the slurry droplets, as well as proper
gas flow patterns. Some designs rely on
gas dispersers placed around the
atomizer that impart a swirling down-
ward motion to the gas. These gas
dispersers cause the slurry droplets and
spray-dried solids to be carried along m
the helical pattern of the entering flue
gas. This helps to ensure more complete
mixing of the slurry droplets and flue
gas as it passes through the chamber.7
Another design also includes a "central
gasdisperser" that ducts part of the flue
gas flow up toward the atomizer
Lime Slaking
The quality of the slaking process
determines how large, porous, and
reactive the lime particles are. The
amount of pore surface area is important
for reaction to proceed in the spray-
dried solids. Therefore, slaking has the
greatest effect on the mass transfer of
Ca++ in the spray-dried solids. The
techniques used in the slaking of lime
can significantly affect its reactivity and
surface area. Thus, slaking methods are
among the first important design con-
siderations in a dry scrubbing system.
Slaking involves hydratmg the lime to
form calcium hydroxide in the presence
of excess water, as described by
Equation 6.2
CaOisi + HaOm - Ca(OH)2ISi +
27,500 Btu/lb-mole (6)
When high calcium, soft-burned pebble
lime is slaked with clean water at a
water-to-lime ratto of 3:1 or 4:1, the
lime pebbles rapidly disintegrate in an
explosive, chain-slaking reaction. This
produces a si urry of extremely fine (0.5 -
4.0 /jm), porous slaked lime particles
suspended in water with a large total
surface area, which is ideal for use in
the spray absorption process.2 The
water used for slaking must be of good
quality. The water should not only be
softened to prevent scaling problems,
but it must be low in sulfates and other
chemicals which can cause a reduction
in the reactivity of the lime. For best
results, slaking should take place at a
relatively constant temperature of 190-
200° F.2
droplet. The inlet SOa concentration has
only a moderate affect on S02 removal
over the range of concentrations which
have been investigated and for which
data are available.
Flue Gas Inlet Temperature
The inlet temperature of the flue gas
affects S02 diffusion and, therefore,
influences the mass transfer of S02 to
the slurry droplet. Contradicting results
have been obtained regarding the effect
of the flue gas inlet temperature on S02
removal efficiency. In a test performed
by Babcock and Wilcox, the inlet
temperature was reduced from 280°Fto
230°F at a constant approach to
saturation of 20°F.5 The decrease in
temperature resulted in a 10-percent
decrease in S02 removal efficiency.
However, a series of tests performed by
Buell indicate that the inlet gas temper-
ature has a negligible effect on SOa
removal.2
The mathematical model developed
by B&W,
=1- r tp i
Tin-Tsat
0.62
(4)
Inlet SOa Concentration
The concentration of SOa in the flue
gas that enters the spray dryer affects
the mass transfer of SO2 to the slurry
predicts that higher flue gas inlet
temperature will result in higher SOa
removal efficiency. An attempt was
made to numerically evaluate the effect
of changing the inlet flue gas tempera-
ture by using this mathematical model.
Since it is difficult to assess the
adiabatic saturation temperature with-
out knowing the flue gas humidity, this
temperature was assumed to be 125°F.
A reduction in the flue gas inlet
temperature from 280 to 230°F, at a
constant approach of 20°F, results in a
decrease in SOa removal efficiency of 8
percent. This seems to support the
results of the test performed by B&W.
Further testing may be required to
clarify the effect of inlet flue gas
temperature on SOa collection efficiency.
Spray-Dried Off-Product Recycle
The use of off-product (reacted and
unreacted sorbent and fly ash) recycle
may provide more available lime surface
which allows for increased mass
transfer of Ca++ in the spray-dried solids.
Recycling of the spray dryer off-product
permits the reuse of unreacted lime
reagent, and if the fly ash is high in
available alkalinity, it may reduce the
amount of required lime by substituting
fly ash alkalinity for lime.6 Fly ash
recycle may improve spray dryer per-
formance in two ways: (1) by providing
-------
available alkalinity, and (2) by acting as a
surface catalyst to enhance the lime-
SO2 reaction. The fly ash acts as a
surface catalyst by providing an alternate
site for CaSO3 to precipitate, thus
decreasing the solid deposits on the
lime particles which cause increased
resistance to mass transfer.
Since making additional alkaline
material available for reaction with S02
is the primary objective of recycling part
of the off-product, care must be taken in
the design of the recycle system so that
the maximum benefit is realized. The pH
of the slurry to which the recycle
material is added must be sufficiently
low so that a reasonable amount of the
recycle alkali will go into solution. Tests
performed by Buell showed that if off-
product was added to the lime slurry,
which was saturated with Ca(OH)2 at a
pH of 12, no measurable benefit
resulted, regardless of the amount of
process off-product added. When a
separate slurry of process off-product
and water alone was prepared in a
separate tank and mixed with the lime
slurry at the point of injection to the
absorber, however, considerable bene-
fit resulted.2
Fabric Filter versus ESP
Paniculate Collection
For the spray drying system either an
ESP or a fabric filter can be used to
remove particulates from the spray
dryer exhaust to meet environmental
emission requirements The type of
paniculate collection device used will
influence mass transfer in the spray-
dried solids. Some tests have shown
that the downstream collection device
can increase the total system SO2
removal by approximately 10 percent.8
Initially, it was felt that a fabric filter
would provide better SC>2 removal than
an ESP,6 and only one commercial
system purchased to date uses an ESP;
the rest use fabric filters. As the fabric
filter operates, a cake of the spray-dried
solids builds up on the filter. If this cake
contains unreactedsorbent, itcanactto
remove S02 as the flue gas passes
through the filter. However, test results
obtained by B&W at the Jim Bndger
Station have shown that both ESPs and
fabric filters provide about the same
degree of S02 removal.5
One of the advantages of using an
ESP instead of a fabric filter is that the
ESP allows the spray dryer to be
operated at a closer approach to the
saturation temperature without causing
maintenance problems in the collection
device. A closer approach to the
saturation temperature will provide
greater SO2 removal in the spray dryer.
ESPs have been found to perform very
well in collecting the particulates from
the spray dryer exhaust. The spray-dried
product does not build up on the
discharge and collecting electrodes.
Due to the humidity of the flue gas, the
fly ash becomes conditioned, which
results in higher particulate migration
velocities than would be expected. Also,
the low outlet temperature from the
spray dryer reduces the resistivity of the
fly ash. In addition, ESPs can be
operated at lower system pressure
drops than fabric filters.
Fabric filters have also been found to
collect the spray-dried solids very well.
The main advantage of a baghouse is
that it is a bulk collection device
whereas an ESP is a percentage
collector. Tests on recirculation have at
times produced dust loading as high as
25 gr/ft3 as an extreme condition. If
environmental restrictions were to call
for a maximum outlet loading of 0.01
gr/ft3, the precipitation efficiency would
have to be 99.96 percent - a number
that may be difficult to achieve con-
sistently with a precipitator in a power
plant. Properly designed baghouses, on
the other hand, should be able to meet a
0.01 gr/ft3 outlet emission regardless of
inlet loading. In addition, the cost of an
ESP for the above duty could be far more
than the comparable baghouse, and the
baghouse will generally be more reliable
in producing low opacity.6
Data Gaps in Literature
To test the proposed mathematical
models and the effects of the variables
listed above, detailed test data are
required. Much of the published test
data lack detailed and complete infor-
mation on inlet SC>2 concentration,
stoichiometric ratio, or approach to
saturation.
There is a particular lack of informa-
tion on spray dryer performance when
the SO2 concentration exceeds 2500 -
3000 ppm S02. This information is
necessary to test the applicability of
lime spray drying for use in units firing
coals containing more than 3 percent
sulfur. Additional analysis mightalsobe
required to resolve conflicting theories
on the effects of certain variables (e g.,
inlet gas temperature, off-product
recycle) on SO2 removal efficiency.
Dry Injection of
Sodium Compounds
Dry injection involves contacting SO2-
laden flue gas with a dry sorbent. The
sorbent is pneumatically injected into
the duct carrying the flue gas and/or
precoated on fabric filter bags. The SC>2
is removed from the flue gas by
adsorption by the sorbent.
Information on dry injection of sodium
compounds was evaluated to charac-
terize the reactions and reaction mech-
anisms involved in the desulfurization
process. The sodium sorbents that were
investigated were nahcolite ore (pre-
dominantly NaHCOs) and trona ore
(Na2CO3 • NaHCOs • 2H2O). The results
of laboratory- and pilot-scale dry
injection tests were evaluated to
establish the reaction mechanism and
rate controlling steps. There was a great
deal of inconsistency in the experi-
mental results, which madeitdifficultto
draw conclusions about the dry injec-
tion process However, qualitative and
quantitative models describing the dry
injection process have been proposed '
and are presented in this report. In
addition, important process parameters
which influence the dry injection
reactions were qualitatively evaluated.
Finally, data gaps and conflicting results
identified in published literature are
discussed to identify areas that may
require further investigation.
Qualitative and Quantitative
Reaction Models for Sodium
Compound Dry Injection
The dry injection reactions that occur
when the SO2 is contacted with the
sorbent are strictly gas-solid phase
reactions. The overall chemical reac-
tions that occur during dry injection for
nahcolite and trona, respectively, are:
(7)
2CO2(gl+H2O,g,
2(Na2C03 • NaHC03 • 2H20) (s> + 3S02(g)
+ 3/2 O2(g) - 3Na2SO4(S> + 4CO2(g) +
5H2O,g, (8)
These reactions do not occur in the
single steps written above, but instead
involve a series of three steps: (1)
thermal decomposition of NaHCOs to
Na2C03, (2) reaction between Na2CO3
and S02 to form Na2S03, and (3)
oxidation of Na2S03 to Na2S04 Several
changes in the physical characteristics
of the sorbent accompany these
reactions.
-------
When the NaHCO3 is thermally de-
composed to Na2C03, small pores
(approximately 0.3 /jm in diameter)
develop in the sorbent particle as H2O
and 062 gas are evolved. These pores
provide an increased surface area for
chemical reaction, thus increasing the
reactivity of the sorbent. The rate of
decomposition of NaHCOs is extremely
temperature dependent, and temper-
atures between 300 and 600°F are
required for adequate pore develop-
ment.
Test results have shown that trona is
less reactive than nahcolite.10 It is
postulated that trona is less reactive
because it already contains 1 mole of
Na2CC>3 for every mole of NaHCOs
present. Therefore, trona does not
contain as much NaHCOa as nahcolite,
and the decomposition of the sorbent
does not generate as much pore volume
per particle of trona. It has been found
that the sorbent with the greatest
amount of sodium bicarbonate (NaHCOa)
parts for total sodium parts present will
generally be the most effective for S02
removal.
The reaction of the SO2 with the
sodium sorbent proceeds from the
exterior of the sorbent particle inward,
leaving behind an inert layer of Na2S04
on the surface of the sorbent particle.
The molar volume of this reacted Na2
S04 layer is approximately 26 percent
greater than the molar volume of the
unreacted Na2C03.7This increase in the
molar volume tends to plug the internal
pores that were generated by the
decomposition of the sorbent. The
plugging of the pores tends to reduce
the reactivity of the sorbent. Scanning
electron micrographs (SEMs) of halved
sorbent particles recovered after reac-
tion with SO2 showed a surface zone of
Na2SO4, an intermediate reaction zone,
and a central unreacted core of Na2COa.
The SEMs of the reacted sorbent
particles warrant the application of the
classical unreacted-core model. This
model proposes three major steps
involved in the reaction of S02 with the
sorbent:
1) Diffusion of gaseous SO2 to the
surface of the sorbent particle
through the stagnant gas film
surrounding the particle.
2) Penetration and diffusion of gas-
eous SO2 through the Na2S04 ash
layer to the unreacted core.
3) Chemical reaction of gaseous S02
with the sorbent at the reaction
surface
Each of these steps has a different
resistance to the rate of reaction, but the
step with the highest resistance is
considered to be rate controlling. Test
results indicate that the diffusion of
gaseous SO2 through the stagnant gas
film is not a rate controlling step.9 It has
been proposed9 that initially the rate of
reaction is controlled by chemical
reaction, and as the ash layer builds up,
the rate of reaction becomes controlled
by the diffusion of S02 through the ash
layer.
The rate of chemical reaction may be
controlled by either the rate of thermal
decomposition of NaHCOs to Na2C03 or
the rate of uptake of S02 by the Na2C03
to form Na2SC>3. However, most experi-
mental results 7'9'10 seem to indicate
that thermal decomposition is the rate
controlling chemical reaction.
Mathematical relationships have
been developed11 for the case where
each of the three reaction steps (gas film
diffusion, ash layer diffusion, and
chemical reaction) controls the rate of
reaction. However, for the dry injection
process the rate of SO2 removal is
controlled by either the rate of chemical
reaction or diffusion through the ash
layer. Expressions describing SO2
removal efficiency as a function of
reaction time for these two cases are:
Ash Layer Diffusion Controls
- = 1 -3(1-E/SR)23
T 2(1-E/SR)
(9)
where:
T =
PR2
6DvCso2-g(MW) (10)
(The above equation cannot be solved
directly for E.)
Chemical Reaction Controls
where:
SR (11)
(12)
ks Cso -g(MW)
The variables included in
these equations are defined
as:
t = Time, sec
T =Time needed for complete
conversion of the sorbent,
sec
E = SO2 removal efficiency,
mole SO2 removed
mole S02 in flue gas
SR' = Stoichipmetric ratio,
mole Na2 O injected
mole SO2 in flue gas
R = Initial particle radius, ft
p = Density of the sorbent, Ib/ft3
Cso -g = Concentration of SO2 in the
gas phase, Ib-mole/ft3
(MW) = Molecular weight of sorbent,
Ib/lb-mole
Dv =Mass diffusivity of SO2
through the ash layer, ft2/sec
ks = First-order rate constant for
chemical reaction, ft/sec
These mathematical models were not
evaluated because test data published
thus far have been incomplete, and
many of the variables required for the
determination of T (i.e., Dv, ks, R, andp)
were not provided and could not be
easily estimated from the available
literature.
Process Parameters Affecting
the Dry Injection Process
Many process parameters affect S02
removal efficiency in the dry injection
process. These parameters were evalu-
ated and were ranked in the approxi-
mate order of their relative importance.
In general, the parameters that may
affect SO2 removal in dry injection are
those that affect the diffusion of SO2 to
the surface of the sorbent particle, diffu-
sion of SOa through the inert ash layer
on the particle surface, or chemical
reaction. Parameters that influence
these three reaction steps are charac-
terized below.
1) Variables affecting SO2 mass
transfer to the sorbent particles:
Particle size.
Residence time of the reactants
in the duct.
Inlet SO2 concentration.
Gas velocity and air-to-cloth
ratio.
Mode of sorbent injection.
Gas temperature.
2) Variables affecting SO2 mass
transfer through the inert ash
layer:
Sorbent preparation.
Particle size.
Residence time of the reactants
in the duct.
Filter cleaning intervals.
Stoichiometric ratio.
3) Variables affecting chemical re-
action:
Sorbent type.
Sorbent preparation.
Temperature.
Stoichiometric ratio.
-------
These variables and their effects are
discussed briefly in the following
sections.
Sorbent Type
The type of sorbent used for dry
injection will directly affect the chemical
reactions that occur. Tests with a wide
variety of sorbents have shown that only
sodium compounds produce significant
S02 removal at typical air preheater
outlet temperatures (approximately
300°F). Lime and limestone have been
demonstrated to achieve S02 removal
only at much higher gas temperatures
(600°F+).1 Sorbents that have been
shown to have good potential for S02
removal are sodium bicarbonate,
nahcolite ore, and trona ore. Nahcolite
ore is a naturally occurring mineral
which typically contains 70 - 90 percent
sodium bicarbonate. Trona ore is also a
naturally occurring mineral which
contains approximately 40 - 50 percent
sodium carbonate (Na2CO3) and 20 -30
percent sodium bicarbonate (NaHC03).
Experimental results from tests using
sodium carbonates and sodium bi-
carbonates indicate that the bicarbonate
form is much more reactive. The
increased reactivity of NaHC03 is
thought to result from the pores that
develop when NaHCOs particles are
thermally decomposed to NaaCCh. In
general, the sorbent with the highest
percentage of "bicarbonate parts" for
"total sodium parts" present would be
expected to perform the best.10
Sorbent Preparation
Treatment of the sorbent prior to
injection influences both the chemical
composition and porosity of the sorbent.
Any changes in porosity will influence
the ability of SOa to diffuse through the
ash layer that surrounds the sorbent
particles, and changes in chemical
composition will affect the rate of
chemical reaction. Sorbent preparation
generally involves grinding the particles
to increase the surface area available
for reaction and heating to thermally
decompose the NaHCO3 to Na2CC>3,
causing an increase in the pore volume
of the particles. The effects of thermal
decomposition will be discussed in this
section; particle size effects are dis-
cussed m the following section.
The thermal decomposition of NaHCOs
to Na2CO3 increases the porosity of the
sorbent, which increases the surface
area available for contact with S02.
Howatson'2 investigated the pore devel-
opment of nahcolite at 300°F with a
scanning electron microscope (SEM)
and found scattered development of 0.1
- 0.7 /urn pores after 10 minutes. After
20 minutes the surface was completely
covered with pores averaging about 0.3
/urn. The Na2CO3 particles produced in
this manner have a much larger void
space than the parent NaHCOs particle.13
These pores provide a greater surface
area for chemical reaction and a greater
volume for S02 diffusion. The reactivity
of the sorbent particles has been found
to increase by up to a factor of 60
because of thermal decomposition.14
Investigations have been conducted15'16
to determine if decomposition of the
sorbent prior to injection enhances SO2
removal and to determine the optimum
decomposition temperature. These
experiments have often produced con-
flicting results. However, a majority of
the tests show that predecomposition of
the sorbent at high temperatures
reduces S02 removal efficiency.
Carson15 reported that predecomposi-
tion of NaHCO3 to Na2C03 significantly
reduced the desulfurization capacity of
the sorbent, however, the decomposi-
tion temperature was not specified. A
' bench-scale study by Genco et al.14 with
a fluidized-bed reactor also showed a
reduction in S02 removal with nahcolite
that had been predecomposed at 600°F.
Particle Size
The size of the sorbent particles
affects the mass transfer of SO2 through
the gas film that surrounds the particle,
because it determines the surface area
through which the SO2 must diffuse.
Particle size also influences diffusion
through the inert ash layer because the
rate of Decomposition of NaHC03 to
Na2CO3 is somewhat dependent on
particle size, and the decomposition rate
influences the porosity of the sorbent
particle. Results characterizing the role
of particle size in S02 removal efficiency
are often conflicting. However, the
general consensus is that better re-
moval is obtained with smaller particles.
Temperature
The temperature of the flue gas
influences the diffusivity of the SOs
molecules and the mass transfer of SO2
to the sorbent particles. However, SO2
mass transfer properties will vary only
slightly in the typical flue gas temper-
ature range (300° - 400°F). The tem-
perature at which the sorbent is injected
into the flue gas affects the thermal
decomposition of NaHCO3 to NaaCOs.
This in turn influences the reactivity of
the sorbent. Many tests have been
performed trying to characterize the
effect of temperature on S02 removal.
Again, these tests have yielded con-
flicting results. However, it appears that
the best SO2 removal occurs when the
sorbent is injected into the duct at
approximately 600°F, and the baghouse
is operated at approximately 300°F. The
high injection temperature allows for
rapid decomposition of NaHCO3 to
Na2C03, and a temperature of 300°F is
optimum for the reaction between
Na2C03 and S02 to form NaaSOs.
Two methods of elevating theflue gas
temperature above air preheater outlet
temperatures have been employed in
pilot-scale systems. One method in-
volves using a stream of hot flue gas
which bypasses the air preheater; the
other method utilizes heaters or heat
exchangers at the air preheater outlet
duct.
Stoichiometric Ratio
Since the Stoichiometric ratio deter-
mines the quantity of sodium available
to react with a given quantity of SO2, it
affects the chemical utilization of the
sorbent particles and the degree of
completion of the chemical reaction.
Also, at lower stoichiometries the
utilization of the sorbent should be
greater than at higher stoichiometries.
That is, the individual sorbent particles
will be more highly reacted at lower
stoichiometries because fewer particles
are injected into the flue gas. Therefore,
at low stoichiometries the ash layer
surrounding the particles should be
thicker and tend to limit S02 mass
transfer.
The Stoichiometric ratio in dry injec-
tion is defined as.
SR = Me/MSO2 (13)
where: Me = Mole equivalent of
Na2O in the sorbent
MSO2 = Moles of SO2 in the
inlet flue gas
Work by several organizations (EPRI,
University of Tennessee, Battelle-
Columbus, Air Preheater, American Air
Filter, Wheelabrator-Frye, and Buell)
has shown an increase in S02 removal
with increased Stoichiometric ratio.
Mode of Injection
The mode of sorbent injection affects
the external mass transfer of SO2 to the
entrained sorbent particles in the
ductwork and through the filter cake in
the fabric filter. The sorbent can be
injected into the system in three ways:
-------
1) Continuous. After the bag is
cleaned, sorbent is continuously
supplied to the flue gas from
injection points in the ductwork,
upstream of the baghouse. The
sorbent them accumulates on the
bags to produce a filter cake.
2) Batch. After the bag is cleaned, all
sorbent is added to the bag as a
precoat before flue gas isresumed.
3) Semi-batch. This feeding method
is a combination of methods 1 and
2. After bag cleaning, some sorbent
is precoated onto the bags and the
remainder is added continuously
upstream.
In a typical semi-batch system, 20
percent of the ground ore is used to
precoat the filter bags. The additional
sorbent is added continuously in the
duct. A maximum grain loading of 10
gr/acf for continuous injection was
used by Genco et al. to prevent particle
settling in the duct.14
The effects of precoat on SO2 removal
have been analyzed by KVB for EPRI.
They found that precoating the bags
with nahcolite increased S02 removal
from 42 to 66 percent, with all other
conditions held constant.13
Tests have also been conducted in an
attempt to characterize where SO2
removal occurs, in the ductwork or in
the filter. These tests indicate that S02
removal in the ductwork is variable and
is a strong function of the flue gas
temperature in the duct. In general, very
little S02 removal will occur in the duct
at temperatures around SOOT; how-
ever, substantial SO2 removal will be
achieved in the duct at temperatures
around 600°F.
Residence Time of the
Reactants in the Duct
The residence time determines the
amount of time that the flue gas and
sorbent will be in contact in the duct.
Therefore, it affects the extent of mass
transfer through both the gas film and
the ash layer that surround the sorbent
particles. Residence time is defined as
the volume of the duct divided by the
volumetric flow of gas through the duct
and is normally expressed in seconds. In
general, S02 removal in the duct
increases with increased residence
time 7
Filter Cleaning Intervals
The longer the baghouse operates
between cleanings, the greater the
sorbent utilization. An increase in
sorbent utilization causes the ash layer
surrounding the particles to become
thicker. Thie increase in ash layer
thickness decreases the mass transfer
of SC>2 through the ash. When the
sorbent being used is nahcolite, the
filter cells should be operated for at least
40 minutes between cleanings, be-
cause it takes 40 minutes to achieve
maximum steady state SO2 removal
with nahcolite.1 However, maximum
steady state removal with trona is
achieved very rapidly.1 Genco et al.
operated the filter cells for approxi-
mately 70 minutes between cleanings
for a nahcolite sorbent.14 The filter
cleaning interval must be balanced
against the rate of SO2 removal. After a
certain amount of sorbent has been
utilized, the rate of S02 removal begins
to level off, and a longer filter cleaning
interval will result in very little addi-
tional SO2 removal.
Inlet S02 Concentration
The inlet SO2 concentration affects
the mass transfer of SO2 to the sorbent
particles, because it determines the
driving force for S02 mass transfer to
the particle surface. Several studies
measuring the effect of S02 concen-
tration on removal efficiency showed
that the inlet SO2 concentration had
only a small effect on removal for a
given stoichiometric ratio. Work by
Wheelabrfctor-Frye, with a nahcolite
sorbent and inlet S02 concentrations of
800 - 2800 ppm, showed that removal
was slightly higher for higher inlet SOa
levels. It was felt that the slight increase
in removal resulted from a higher
driving force for the diffusion of SO2
from the bulk gas to the solid particle.
However, the increase in removal was
somewhat offset because at higher inlet
SO2 concentrations more nahcolite had
to be fed to maintain stoichiometric
ratios, which necessitated more fre-
quent bag cleaning and resulted in
lower nahcolite residence time in the
system.1 Note t hat Wheelabrator-Frye's
reasoning is not consistent with the
reaction models developed in this
report, because the diffusion of S02
from the gas to the particle was not
considered to be a rate controlling step.
The dry injection system has not been
tested with S02 concentrations greater
than approximately 3500 ppm. There-
fore the applicability of dry injection
with coals containing greater than 3.5
percent sulfur is not known.
Gas Velocity and
Air-to-Cloth Ratio
The velocity of the flue gas relative to
the sorbent particles affects the mass
transfer of S02 through the stagnant
gas film to the surface of the sorbent
particle. Gas velocity affects the dif-
fusional properties of the SOa and the
time the sorbent and flue gas are in
contact. The lower the velocity of the
flue gas, the longer it remains in the dry
injection system to react with the
sorbent. However, studies by several
sources (Buell, EPRI, American Air
Filter, University of Tennessee, and
Wheelabrator-Frye) showed that the
velocity of the gas had no significant
effect on SO2 removal. The air-to-cloth
ratio, which represents the velocity of
the gas through the filter cloth and is
described as the volumetric flow of the
gas through a specified filter area, was
varied from 1 .4 to 5.0 ft/min with no
observable effect on SO2 removal.10'14'1
Data Gaps in Literature
It is proposed that the un reacted core
model can be applied to the reactions
that occur during dry injection.10
However, this model was not evaluated
against experimental data; therefore, it
may not be valid.
To evaluate the applicability of the
unreacted core model, experimental
data describing sorbent conversion
versus time will be required. The
mechanism by which the desulfuriza-
tion reaction
Na2CO3(sl + SO2(Qi - Nas-SOscsi + COzigi (14)
proceeds also needs to be established,
and the influence of the flue gas
moisture content on desulfurization
should be examined.
It would be useful to characterize SO2
removal at inlet SO2 concentrations
greater than 2000 ppm to determine if
dry injection is applicable to systems
that burn coal with a higher sulfur
content In addition, parametric testing
should be performed to clear up conflict-
ing experimental results concerning the
effects of particle size, injection tem-
perature, and thermal predecomposition
of the sorbent on S02 removal. Finally, it
may be beneficial to test dry injection on
a flue gas produced from combustion
under low excess air conditions. Mini-
mizing the oxygen content of the flue
gas could reduce the oxidation of
Na2SO3 to Na2SC>4. If in fact the
oxidation of Na2SOs causes the pores in
the sorbent particle to plug, then a
reduction in the amount of Na2SC>4
8
-------
13.
14.
15.
16.
generated should reduce plugging and 12.
allow for better S02 removal.
Literature Cited
1. Kelly, M.E. and S.A. Shareef. Third
Survey of Dry S02 Control Systems.
EPA-600/7-81-097 (NTIS PB81-
218976) June 1981.
2. Parsons, E.L. et al. "S02 Removal
by Dry FGD." In: Proceedings
Symposium on Flue Gas Desulfun-
zation—Houston, October 1980;
Vol. 2. EPA-600/9-81-019b (NTIS
PB81-243164), pp 801-852. April
1981.
3. Getler, J.L. et al. "Modelling the
Spray Absorption Process for SO2
Removal." In: Journal of the Air
Pollution Control Association, 29:12,
p. 1270. December 1979.
4. Downs, W., W.J. Sanders, and C.E.
Miller. "Control of SO? Emissions
by Dry Scrubbing." (Presented at
the American Power Conference.
Chicago, IL. April 21-23, 1980.)
5. Hurst, T.B. and G.T. Bielawski. "Dry
Scrubber Demonstration Plant -
Operating Results". In: Proceedings:
Symposium on Flue Gas Desulfuri-
zation—Houston, October 1980;
Vol. 2. EPA-600/9-81-019b (NTIS
PB81-243164), pp. 853-860. April
1981.
6. Meyler, J.A. "Dry Flue Gas Scrub-
bing: A Technique for the 1980's"
(Presented at 1980 Joint Power
Conference. Phoenix, AZ. Septem-
ber 1980.)
7. Dickerman, J.C. et al. "Evaluation
of Dry Alkali FGD Systems." Draft.
EPA Contract No. 68-02-3190.
March 1978.
8. Stevens, NJ. "Dry S02 Scrubbing
Pilot Test Results." In: Proceedings:
Symposium on Flue Gas Desulfuri-
zation—Houston, October 1980;
Vol. 2 EPA-600/9-81-O19b (NTIS
PB81-243164), pp. 777-800 April
1981.
9. Davis, WT. andT.C. Keener. "Phase
I: Chemical Kinetic Studies on Dry
Sorbents, Literature Review." Uni-
versity of Tennessee. August 10, ft2-hr-°F
1980.
10. Samuel, E.A. and D.E. Lapp. "Test- 1 Btu
ing and Assessment of a Baghouse lb-°F
for Dry SO2 Removal, Task 4: SO2
Removal Using Dry Sodium Com- 1 ft2
pounds." September 1980. Draft sec
report prepared for U.S. EPA,
Contract No. 68-02-3119. 1 Jb_
11. Levenspiel, 0. Chemical Reaction ft3
Engineering, 2nd edition. New
York: John Wiley and Sons, Inc., pp. 1 Ib
357-377. 1972. ft-hr
Howatson, J., J.W. Smith, D.A.
Outka, and H.D. Dewald. "Nahcolite
Properties Affecting Stack Gas
Pollutant Absorption." In: Proceed-
ings of 5th National Conference on
Energy and the Environment, Cin-
cinnati, OH. November 1977.
"SO2 Control by Dry Sorbent Injec-
tion." In: EPRI Journal. June 1980.
p. 52.
Genco, J.M. et al. "The Use of
Nahcolite Ore and Bag Filters for
S02 Emission Control." In: Journal
of Air Pollution Control Association,
25:12, p. 1244. December 1975.
Carson, J.R. "Removal of Sulfur
Dioxide and Nitric Oxide from a Flue
Gas Stream by Two Sodium Alkalis
of Various Sizes." Master's Thesis,
University of Tennessee, Knoxville.
August 1980.
Davis, W.T. and T.C. Keener. "Re-
search on the Removal of S02 by
Additive Injection Techniques on a
Stoker-Fired Boiler." (Presented at
the 71 st Annual Meeting of the Air
Pollution Control Association, June
25, 1980.)
Appendix
Conversion
British
1 ft
1 Ib
1 000 cf m
1 gal./1000ft
1 Btu
1 gal
1 ft3
1 Ib-mole
1 atm
1 Btu
Factors
Metric
= 0.3048m
= 0.454 kg
= 0.5 mVs
= 0.13 liters/m
= 1 .055 kJ
= 3.79 liters
= 28.32 liters
= 453.6 g-mole
= 1.013x 105Pa
= 5.678 J
m2-s-K
kg-K
= 9.29x10'2mi
s
= 16.018Jia-
m3
= 4.133x 10"4Pa-s
-------
C. Apple and M. E. Kelly are with Radian Corporation. Durham. NC 27705.
Theodore G. Brna is the EPA Project Officer (see below).
The complete report, entitled "Mechanisms of Dry S02 Control Processes,"
(Order No. PB 82-196 924; Cost: $ 13.50, subject to change) will be available
only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
10
-------
Q) O
a
c
V)
(D
CO
8
n *-«
J>W
T> m
-af< i i3
TJ(Q
W O O O
W< tt 3
w o g
D ®
3
•-»
0)
------- |