EPA/600/A-92/134
REDUCING MERCURY EMISSION FROM MUNICIPAL WASTE COMBUSTION
WITH CARBON INJECTION INTO FLUE GAS
T.G. Brna, J.D. Kilgroe, arid C.A. Miller
U.S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC
Abstract
The Clean Air Act Amendments of 1990 require the U.S. Environmental
Protection Agency (EPA) to set emission limits for cadmium (Cd), lead (Pb), and
mercury (Hg) for municipal waste combustors (MWCs). To aid developing Hg
emission rules, tests were conducted in July and August 1991 on a 360-tonne (400-
ton)/day mass burn MWC at the Ogden Martin Systems of Stanislaus, Inc. (OMSS)
facility near Crows Landing, CA. The primary objective was to evaluate the
effectiveness of powdered activated carbon (C) in controlling Hg emission. The C
was injected into flue gas at both the economizer outlet and spray dryer absorber
(SDA) inlet, as well as into the lime slurry fed to the SDA, during separate test
conditions.
Secondary test objectives were to the evaluate (1) the impact of ammonia
(NH3) slip from the selective noncatalytic reduction (SNCR) system on Hg control,
(2) the effect of lime stoichiometry in the SDA/fabric filter (FF) system on Hg
emission, (3) the effect of FF gas temperature on Hg emission, and (4) the
time/temperature stability of Hg residing with the ash (residue) collected at several
locations from the unit tested (Unit 2):
Results of the tests indicated that C addition was effective in improving Hg
removal, the removal increasing with increasing C feed rate. Hg removal improved
from about 30% without C addition to over 90% at the highest C feed rate tested.
The test data obtained for evaluation of the secondary test objectives suggested that
(1) NH3 addition (which occurred with low slip) had no apparent adverse effect on
Hg control, (2) lime stoichiometry had no effect on Hg emissions over the narrow
range studied, (3) the FF gas temperature over the range tested did not affect Hg
control, and (4) the Hg content of ash/residue stream samples remained relatively
constant over a period of 28 days in samples held at 54°C (130°F).
For Presentation at
ECO WORLD '92
Washington, DC
June 1992

-------
INTRODUCTION
Section 129 of the 1990 Clean Air Act Amendments requires the U.S.
Environmental Protection Agency to set emission limits for mercury (Hg) for
municipal waste combustors (MWCs).l Data collected during the regulatory
development process for MWCs show highly variable Hg emissions. Tests of
recently built MWCs, frequently equipped with spray dryer absorber (SDA) and fabric
filter (FF) flue gas cleaning systems to control air pollutant emissions, have
indicated Hg removals ranging from zero to over 95%.2 Modern mass burn
waterwall (energy recovery) MWCs having high combustion efficiencies [and also
low carbon (C) content in their flyash] appear prone to low Hg removals even when
equipped with SDA/FF systems. Thus, the C content of the flyash is suspected of
having a key role in the control of Hg emissions from MWCs.
MWCs with dry scrubbing systems [dry sorbent injection (DSI) or SDA
followed by particulate matter (PM) collection] in Canada and Europe have injected
either sodium sulfide (Na2S) or activated powdered C before sorbent addition for
acid gas control to supplement Hg control.2-4 The products of the Na2S-Hg or -Hg-
compound reactions or the Hg or Hg compounds physically or chemically absorbed
on the C particles were subsequently removed with the PM. Hg removals exceeding
80% were reported using these supplemental control technologies.2-4
Many recently built MWCs in the U.S. (and those being planned) have lime
SDA/FF systems for flue gas cleaning. Few MWCs employ DSI/FF systems in the
U.S., and use of these systems has been restricted to units with capacities under 225
tonnes (250 tons)/day. Outside of the U.S., activated C for supplemental Hg control
has been successfully used with SDAs coupled with either FFs or electrostatic
precipitators (ESPs), whereas Na2S injection has normally been used in conjunction
with DSI/FF or ESP systems. Safety considerations [since crystalline Na2S is
frequently mixed with water on-site to the desired Na2S concentration (for injection
into flue gas) with potential liberation of hydrogen sulfide (H2S) as a consequence]
and the trend in the U.S. of using SDA/FF systems were advantages considered by
EPA in choosing to evaluate activated C instead of Na2S for supplemental Hg
control. EPA also determined that its evaluation should be on a modern mass burn
waterwall MWC equipped with a lime SDA/FF system because of its history of low
Hg removal but high control of other air pollutant emissions.
EPA's Air and Energy Engineering Research Laboratory (AEERL) selected the
Ogden Martin Systems of Stanislaus, Inc. (OMSS) facility near Crows Landing, CA, as
meeting the criteria for its supplemental Hg control study. This facility, which
began operation in 1988, has two identically designed Martin GmbH mass burn
waterwall MWCs with selective noncatalytic reduction (SNCR) and lime SDA/FF
2

-------
systems for flue gas cleaning. Each unit has a capacity of 360 tonnes (400 tons)/day
and normally operates continuously at full capacity.
AEERL developed the test plan in coordination with OMSS. The major
objective of the test project was to evaluate the addition of activated powdered C to
flue gas on reducing stack Hg emission. Secondary objectives included investigating
the effect on Hg emission from the stack of (1) ammonia (NH3) addition in the
furnace (SNCR) which is used to control nitrogen oxide (NOx) emissions, (2) lime
stoichiometry in the SDA, and (3) gas temperature into the FF. Earlier tests at the
OMSS facility had indicated that increasing NH3 and lime feed rates correlated with
increasing Hg emission,6 and tests elsewhere noted increasing Hg emission with
increasing lime stoichiometry.7 Another objective was to determine the stability of
Hg collected with ash* over time.
Prior to the test program, AEERL coordinated the proposed test plan with EPA
Region 9 (San Francisco) and the Stanislaus County (Modesto) air pollution control
authorities. Since some test conditions (e.g., operation without NH3 injection)
would or could result in exceeding emission limits for the unit being tested, the test
program conducted conformed to agreements with these regulatory authorities.
Hg was sampled using EPA Method 101A. This method was modified to
include Hg collected on the laboratory filter in determining Hg collected in each
sample. [EPA is expected to modify this method to include analysis of the laboratory
(analytical) filter for Hg in 1992].**
DESCRIPTION OF UNIT TESTED
The test program was performed using Unit 2 of the OMSS facility. A
schematic diagram of the 360-tonne/day unit is shown in Figure 1, with the NH3
injection in the furnace corresponding to the SNCR (Exxon Thermal DeNOx®)
system and the lime SDA/FF (ABB Flakt) system being the flue gas cleaning system.
Flue gas with a design flow rate of 1890 standard cubic meters per minute
(scmm) [66,700 standard cubic feet per minute (scfm)]*** leaves the economizer. It is
divided into three equal streams at the SDA inlet before contacting atomized lime
slurry from the SDA's two-fluid nozzles. The lime slurry feed rate is regulated
according to the sulfur dioxide (SO2) concentration in the stack, while the dilution
* The term "ash" is used here for "residue," which consists of flyash, reaction products, and unreacted
sorbent.
" Hg sampling validation studies and comparison of Hg data obtained using EPA Methods 101A and 29
(multiple metals train) were also performed in conjunction with this program. This work is reported in
"Evaluation of Two Methods for the Measurement of Mercury Emissions in Exhaust Gases from
Municipal Waste Combustors," EPA-450/4-92-013, U.S. Environmental Protection Agency, Office of Air
Quality Planning and Standards, Research Triangle Park, NC.
**" Standard conditions correspond to 20°C (68°F) and 101.3 kPa (14.7 psi).
3

-------
Activated
Carbon
Injection
Locations
Slurry Feed
Ammonia
injection
Slurry Atomizer
Penthouse
Convective Section
Feed
Chute
Sample
Poits
MS
Spray
Dryer
Absorber
Furnace
Stack
fabric
Filter
Grate
v7
Ash Conveyors -J
Ash Sample Locations
Ash Conveyor
Ash Discharge
Induced
Draft
Fan
Figure 1 Schematic Diagram for a Unit of the Ogden Martin Systems of Stanislaus, Inc. (OMSS) Facility.
\

-------
water added to slaked lime in the slurry feed tank is controlled by the gas
temperature leaving the SDA. The flue gas transit time through the SDA is about 15
seconds. The FF has a net air-to-cloth ratio of 0.98 m^/min.-m2 (3.2 cfm/ft2) and six
compartments of Teflon™-coated glass fiber bags. The pulse-jet cleaning is
continuous, and the cleaning cycle is about 2 minutes per compartment, or
approximately 12 minutes for the entire baghouse.
A covered conveyor carries ash from the FF to the SDA where ash from the
SDA is added. Flyash from the convective section of the boiler is added to the SDA
and FF ashes and transported for combination with bottom ash. After quenching
with water, the combined ash is discharged to a shaker conveyor which carries the
ash to a large-material separator. Ash from this separator moves on a belt conveyor
to a magnetic separator where ferrous materials are removed. The ferrous
materials and remaining ash are retained in their separation building until they are
removed for recycling and landfilling, respectively.
TEST PROGRAM
Table I summarizes the test parameters for the 16 conditions evaluated. The
activated C type, feed rate, and feed location (see also Figure 1) were the primary
independent variables. Secondary variables evaluated for their effect on Hg
emission in the stack were NH3 injection (either zero or at rates corresponding to
normal unit operation), flue gas temperature in the FF [136 to 151°C (277 to 303°F)],
and acid gas control efficiency [using SO2 outlet concentration (up to about 30
ppmv)* as a surrogate]. With the exception of Condition 15 where two 1-hour tests
were run, three 1-hour tests were normally conducted for all conditions. During
Conditions 4, 5, 6, 8, and 9, dual Method 101A sampling trains were operated in the
stack. Thus, two stack Hg concentrations were obtained for each test run for these
conditions, except for Runs 2 and 3 of Condition 4.
Conditions 4 and 5 were performed without C addition to establish baseline
conditions. Condition 4 corresponded to normal unit operation, while no NH3
injection occurred during Condition 5 so that results for these conditions could be
used to evaluate the effect of NH3 on Hg removal by the SDA/FF system. Condition
7 was conducted to study the impact of C injection without NH3 addition and
followed Condition 4 to minimize the time without operation of the SNCR system.
To allow time for the combustor and SDA/FF system to be purged of NH3, the
injection of NH3 was terminated (with regulatory approval) 12 hours prior to
starting the Condition 5 tests. NH3 injection was resumed immediately after
completion of the third (final) test at Condition 7, Hg was also sampled at the SDA
outlet for Conditions 5 and 7 to study Hg control in the SDA and FF separately.
* Unless otherwise noted, all concentrations are referenced to 7% O2 in dry gas
5

-------
TABLE I. MERCURY EMISSIONS CONTROL FIELD TEST PARAMETERS FOR THE OMSS MWC FACILITY
CONDITION
NUMBER
TEST
DATE
NUMBER
OF TEST
RUNS
OPERATING PARAMETERS
THERMAL
DENOx
CARBON
FEED RATEa
kg/hrdb/hr)
CARBONb
(Raw
Material)
FABRIC FILTER
TEMPERATURE
CARBON
INJECTION
LOCATION*
LIME
FEED
RATE
1
7/22/91d
3
Normal
1.3 (2.8)
Coal
Normal
EO
Normal
2
7/23/91
3
Normal
5.4 (12)
Coal
Normal
EO
Normal
3
7/24/91
3
Normal
1.3 (2.8)
Coal
Normal
SDAI
Normal
4 (BL)e
7/25/91
3
Normal
Off
None
Normal
None
Normal
5 (BL)
7/29/91
3
Off
Off
None
Normal
None
Normal
6
7/26/91
3
Normal
5.5 (12.1)
Coal
Normal
SDAI
Normal
7
7/30/91
3
Off
1.3 (2.9)
Coal
Normal
SDAI
Normal
8
7/31/91
3
Normal
2.8 (6.1)
Coal
Normal
SDAI
Normal
9
8/1/91
3
Normal
1.3 (2.8)
Lignite
Normal
SDAI
Normal
10
8/7/91
3
Normal
5.6 (12.3)
Lignite
Normal
SDAI
Normal
11
8/5/91f
3
Normal
1.3 (2.9)
Coal
Low
SDAI
Normal
12
8/5/918
3
Normal
1.3 (2.8)
Coal
Normal
SDAI
Low
13
8/2/91
3
Normal
1.5 (3.2)
Wood
Normal
SDAI
Normal
14
8/6/91h
3
Normal
3.0 (6.6)
Wood
Normal
SDAI
Normal
15
8/10/91
2
Normal
8.3 (18.3)
Coal
Normal
w/lime slurry
Normal
16
8/10/91
3
Normal
5.5 (12.2)
Coal
Normal
w/lime slurry
Normal
Numbers may not agree because of rounding of English units to two significant (metric) figures.
Lignite = Darco FGD, Surface Area = 600 mVg, Average Pore Radius = 3,0 nm. Tamped Density = 475 kg/m3
Coal = Darco PC-100, Surface Area = 950 /g, Average Pore Radius = 15 nm. Tamped Density = 685 kg/m3
Wood = Darco KB, Surface Area = 1500 /g, Average Pore Radius = 25 nm. Tamped Density = 450 kg/m3
EO = Economizer Outlet, SDA1 = Spray Dryer Absorber Inlet
One run conducted on 7/23/91
BL = Baseline
One run conducted on 8/6/91
One run conducted on 8/2/91
One run conducted on 8/7/91
c
d
e
f
S
h

-------
The remaining test conditions examined Hg removal as a function of C type,
feed rate, or feed location. C from coal was used in 10 test conditions, for three feed
rates, and three feed locations (see Figure 1). C from lignite and wood was studied at
two feed rates and two feed locations, both into the flue gas duct. All C types were
injected at the low feed rate [1.3 kg/hr (2.8 lb/hr)] into the flue gas entering the SDA.
Table I decribes the three C types tested.
During Conditions 3, 6, and 8, the effect of C feed rate with coal-based C was
studied with the C injected into the SDA inlet duct. As shown in Figure 1, this
injection site was downstream of the SDA inlet sample ports. Condition 3 had an
average C feed rate of 1.3 kg/hr (2.8 lb/hr); the feed rates for Conditions 6 and 8
averaged 5.5 and 2.8 kg/hr (12.1 and 6.1 lb/hr), respectively. These feed rates
corresponded to C concentrations in flue gas of about 17, 73, and 37 mg/dscm (0.0074,
0.032, and 0.016 gr/dscf).
The effect of injecting C at different locations was also investigated. Injecting
the C at the economizer outlet provided about a second longer contact time between
the C and flue gas compared to C injection at the SDA inlet. When the C was mixed
in the lime slurry and injected with the slurry, the C flue gas contact time was
slightly less than when the C was injected in the SDA inlet duct; however, the
major factors expected to influence Hg removal by C carried in the slurry were the
wetting of the C, the potential agglomeration of particles, and reduction in carbon
surface area through coating of the C particles with lime. Since these factors were
expected to decrease the effectiveness of the C in capturing Hg, the C feed rate for
Condition 15 was increased by about 50% over the high rate for powdered C
injection into flue gas.
The effect of reduced gas temperature in the FF was examined during
Condition 11. Because the SDA/FF system was designed to shut off the lime slurry
feed pump when the flue gas temperature at the FF reached 135°C (275°F), this bag
protection feature limited the minimum temperature for the tests.
Condition 12 was performed to investigate the effect of lime stoichiometry on
Hg emission. Since the maximum permitted stack SO2 emission was 30 ppmv, the
tests at this condition were run with manual slurry feed control. Because of varying
SO2 concentration with time at the SDA inlet and the finite response time required
for the changed slurry flow to counter the change in inlet SO2 concentration,
controlling the stack SO2 steadily at a set point over a 1-hour test was not possible.
Ash samples collected daily for each test location and condition were
aggregated and thoroughly mixed before obtaining a representative sample for
analysis. In addition to determining Hg retention in the combined ash over time,
FF ash samples were analyzed for moisture, C, and loss-on-ignition (LOI).
7

-------
DISCUSSION OF RESULTS
Effect of Carbon
Coal-based C was studied in 10 test conditions encompassing three feed rates
and three feed locations. The effect of feed location at the high and low feed rates on
Hg removal by the lime SDA/FF system is shown in Figure 2. The data scatter is
greater at the low feed than at the high feed rate, and the Hg reduction values for the
feed locations overlap. At the low feed rate, Hg reductions were 66-85% with C fed
at the economizer outlet (Condition 1) and 53-77% for C fed at the SDA inlet
(Condition 3). With the high feed rate, Hg reductions were 88-92% (Condition 2)
and 91-98% (Condition 6) with C fed at the economizer outlet and SDA inlet,
respectively. During conditions when the C was fed at the medium rate of
approximately 2.8 kg/hr (6 lb/hr) at the SDA inlet, Hg reductions were 73-92% (not
shown in Figure 2).
The differences in Hg reductions as a function of C feed location were not
statistically significant. The C feed rate (or C concentration in flue gas) did have a
significant effect on Hg reduction, and this result is consistent with that reported
earlier.^
The trend of increasing Hg reduction with increasing C feed rate, regardless of
C type, is indicated in Figure 3. This figure also suggests that the type of C had no
effect on Hg reduction for the three types investigated.
The data points in Figure 4 show that Hg reduction generally increased with
increasing Hg concentration in flue gas entering the SDA as well as with increasing
C feed rate. The curves on this figure are plotted using Equation 1, which is
discussed below.
Considering 57 valid data points for Hg reduction and 60 points for outlet
(stack) Hg concentrations, these variables were correlated with C feed rate raised to
the 0.5, 0.7, and 1.0 power, respectively, using multiple regression analysis.® The
best predictive equations for Hg reduction (PRED) and outlet Hg concentration
(HGOUT) were:
In (100-PRED) = 4.81 - 0.948 (CFR) °-5 - 0.000776 (HGEN)	(1)
and
In (HGOUT) = 5.66 - 0.963 (CFR) 0 5 + 0.000724 (HGIN)	• (2)
where: In is the natural (Naperian) logarithm, PRED is the Hg reduction in percent,
CFR is the C feed rate in kg/hr, and HGEN and HGOUT are Hg concentrations in
(ig/dscm at the SDA inlet and in the stack, respectively. The "goodness of fit" for
these predictive models for PRED and HGOUT given by Equations 1 and 2 are 0.762
and 0.777, respectively.
8

-------
00
Econ = Economize* Outlet
SDAl = Spray Dryer Absorber Inlet
SL = Slurry
IFH . lwj Rale » -< * < **•»)
Iff
HFR
Carbon Feed Rale
Figure 2. Effect of Carbon (Darco PC-100) Feed Location on
Mercury Reduction.
a
X
LFfl . Few)Ran » -i « hgnw 13 Mu)
uf ft « Meflium Feeo Rate ¦ -? 7 tgf* (6 ttfv]
mFR . Hgti Feed Rale « -S « (1? tirtv)
LFR
MFR
MFR
Carbon Feed Rale
Figure 3. Dependence of Mercury Reduction on Carbon Type Under
Normal Operation and Carbon Injection at the Spray Dryer
Absorber Inlet.
O
r
	? « Low Feco Rale ¦ -i < hp** 13 Krtv)
/ • Uftitn Rax • -2 ? kQtv {6 brtw}
~ > H«gh F«»d Rate * -M (i? ttfwj
300 400 SOO 600 700 600 9C0 1 000 MOO »?O0 iJOO
Intel Hg, ng/dscm
Figure 4. Effect of Carbon Feed Rate on Mercury Reduction.
9

-------
The curves (from Equation 1) shown in Figure 4 show that Hg reduction
increases with C feed rate and inlet Hg concentration. As noted earlier, the data
scatter is greater at the low than at the high C feed rate. At the high feed rate,
Equation 1 predicts Hg reductions of 90% or more for inlet Hg concentrations above
300 (ig/dscm. The slope of each curve decreases slightly with increasing Hg
concentration and suggests that C utilization for adsorption of Hg decreases as the
Hg concentration rises. The convergence of the curves indicates that increasing the
C feed rate to reduce Hg emission becomes less effective as the inlet Hg
concentration increases.
Using statistical methods and all data for the high C feed rate, it has been
estimated that about 95% of individual tests will attain Hg reductions of at least
80%.8 Using similar procedures, approximately 95% of individual tests would be
expected to yield outlet Hg concentrations of 112 M-g/dscm or less.
Hg reductions across the SDA and FF individually were obtained during
Conditions 3 and 7, both conditions with the low C feed rate at the SDA inlet. With
NH3 injection during Condition 3, the Hg removal in the SDA averaged 17%.
However without NH3 injection during Condition 7, the average Hg concentration
in the flue gas increased by 19% (a negative Hg reduction) in the SDA. Noting that
the Hg reduction between the SDA outlet (FF inlet) and the stack averaged 60% for
both conditions, the cause for this apparent anomaly has not yet been determined.
While the average overall Hg reduction of 66% (53-77% range) for Condition 3 is
consistent with the 74% (66-85% range) obtained at Condition 1 (noting that
injection location did not significantly impact Hg reduction), the average Hg
reduction for Condition 7 with two valid data sets was 52% (average of 48 and 56%),
While greater Hg removal was expected in the FF than in the SDA because of the
greater sorbent residence time in the FF, an increase in Hg concentration across the
SDA was unexpected. More on this finding will be said later in discussing the effect
of NH3 on Hg emission.
Effect of NH3 on Hg Control
As noted earlier, Conditions 4 and 5 were conducted to determine if NH3 had
an impact on Hg emission in the absence of C addition. The results of tests at
Condition 4 (normal NH3 feed) gave an average Hg removal of 28% in the SDA/FF
system and those at Condition 5 (no NH3 feed) indicated an average removal of
31%. For both conditions dual (Method 101 A) trains were used in the stack. The
average Hg removals were based on four valid stack values (16-36%) for Condition 4
and six values (18-39%) for Condition 5. Thus, these Hg reductions are comparable.
The NH3 concentrations in the flue gas entering the SDA (NH3 slips)* were
below 5 ppmv for all but two runs (9.4 ppmv for Run 2 of Condition 3 and 23 ppmv
* NH3 slip is defined as NH3 which does not react with NOx in the SNCR process.
10

-------
for Run 2 of Condition 12) and included the runs (Conditions 3 and 7) without NH3
feed. These low values of NH3 slip suggest that the SNCR system and NH3
injection rates were "well-tuned," which may not have been the case for the 1988
tests on Unit 2 when Hg emissions appeared to increase with increasing NH3 feed
rate (no NH3 slip data were reported).6
Earlier it was stated that Hg concentration increased in the SDA for both runs
during Condition 7, which was conducted as the last of three consecutive test
conditions without NH3 feed. Thus, no residual effects of NH3 were expected. The
Hg concentration at the SDA inlet for Condition 7 averaged 389 |ig/dscm, the lowest
average obtained for all 16 test conditions and 50 ng/dscm lower than the next
lowest average value (Condition 8). However, no basis has yet been found for
invalidating these data of Condition 7. As noted earlier, the Hg removals in the FF
were equal for both Conditions 3 and 7 and the Hg concentrations entering the FF
and in the stack were reasonably consistent relative to similar tests.
In view of the above discussion, no firm conclusion can be made regarding
the impact of NH3 concentration in flue gas on Hg removal. The low NH3 slip
values measured suggest that the NH3 concentrations in flue gas entering the SDA
were independent of whether the SNCR system was or was not operated during this
test program. The small difference (10%) between the average Hg reductions for
Conditions 3 and 7 indicates that NH3 had a minor, if any, effect on Hg removal in
these tests.
Effect of Gas Temperature
Temperature can affect Hg removal from flue gas in several ways. Decreasing
temperature supports condensation of vapors of Hg and Hg compounds present in
flue gas and their adsorption on PM. Gas temperature can affect the reaction rate
between Hg species and chemical reactants in flue gas.
The temperatures of flue gas leaving the SDA and entering the FF were 136-
151°C (277-303°F) and about 1-3°C (2-5°F) lower in the stack. Since the Hg removal
from flue gas in the FF will usually require the Hg species to be in PM form, gas
temperature entering the FF was selected as the appropriate temperature for study.
Because the vapor pressure (or concentration)/temperature data for Hg
components expected in flue gas of MWCs equipped with lime SDA/FF systems
show that these components [e.g., elemental Hg, mercuric chloride (HgCl2),
mercurous chloride (HgCl), and mercuric oxide (HgO)] are predominantly in the
vapor phase even below operating temperatures [about 140-145°C (284-293°F)],
removal of these vapors through condensation is not expected to be significant. The
average Hg removal for Condition 3 was 66% when the temperature was 147°C
(297°F), while Hg removal for Condition 11 was 64% when the temperature
averaged 140°C (284°F) for otherwise equivalent MWC operating conditions. This
11

-------
finding suggests that the SDA outlet temperature over the limited range permitted
for this test program did not affect Hg removal.
Effect of Lime Stoichiometry
Earlier tests on Unit 2 showed a linear correlation between lime slurry feed
rate (proportional to stoichiometry) and Hg emission, with emission increasing
with feed rate. 6 This increasing Hg emission may have resulted from the reduction
of HgCl2 solids through reaction of calcium hydroxide [Ca(OH)2] to form HgO and
calcium chloride (CaCl2), and subsequent conversion of HgO yielding Hg vapor.
Since the lime stoichiometry could not be measured directly, the SO2 outlet
concentration provided a continuously measured surrogate for lime stoichiometry
and was controlled manually in the attempt to maintain a set point of 30 ppmv over
each 1-hour test at Condition 12.* Because of the varying SO2 concentration at the
SDA inlet and the finite response time required to adjust lime feed slurry flow and
lime solids concentration in the slurry, maintaining the desired SO2 set point
continuously was not possible. With both inlet and outlet SO2 concentrations
measured continuously, the corresponding SO2 removal was calculated, with high
removal suggesting high lime stoichiometry. With Equation 3 defining
stoichiometric ratio (SR), which is the ratio of the actual to the stoichiometric lime
requirement for flue gas entering the SDA, it is seen that the lime supplied is
inversely proportional to both the SO2 and hydrogen chloride (HC1) present in the
flue gas to be cleaned.
	[mol lime supplied]	
~ [mol SO2 + 0.5 (mol HCD] in flue gas into SDA
If removal efficiencies for SO2 (ESO2 ) anc* HC1 (EhCI) are expressed as
fractions, Equation 3 becomes Equation 3a.
cR	:	[mol lime supplied)	 , .
[ESo2 (mol SO2) + 0.5 (EHClKmol HCl)]in»o SDA + [(mol S02 ) + 0.5 (mol HCDlfrom SDA w '
The average SO2 removal during Condition 12 was 73% (66-77% range), when
the SO2 inlet concentration averaged 82 ppmv. The average HC1 reduction was 97%
and the inlet HC1 concentration averaged 661 ppmv at this condition. An average
Hg reduction of 59% was obtained during Condition 12 when the inlet Hg
concentration averaged 682 |ig/dscm. Condition 3 was similar to Condition 12
except normal SO2 control yielded an average SO2 removal of 92% when the inlet
SO2 concentration averaged i08 ppmv. The.average HC1 inlet concentration and
' The SO2 outlet concentration is normally controlled so that either 80% SO2 removal or an outlet SO2
concentration of 30 ppmv at 12% CC>2; whichever is less restrictive, is maintained. Although
regulatory concurrence was received to vary SO2 concentration, an effort was made to stay within the 30
ppmv limit during Condition 12.
12

-------
removal were 517 ppmv and 98%, respectively. The average Hg reduction for
Condition 3 was 66% for removals in the 53-77% range when the Hg concentration
averaged 534 p.g/dscm. Comparison of the HC1 and Hg reductions for Condition 12
and 3 indicates similar control levels, while the lower SO2 removal during
Condition 12 relative to Condition 3 suggests a lower lime stoichiometry. However,
the different acid gas concentrations and removals during these conditions do not
permit assessing the related lime stoichiometries for these conditions (see Equation
3a). Thus, the limited test data appear inconclusive regarding the existence of a
correlation between SO2 control efficiency and Hg emission control.
Ash Characteristics and Retention of Hg by Ash
It was noted earlier that the C content of flyash is believed to impact control of
Hg emission, and the temperature at which combined ash is held may affect Hg
retention in this ash over time. Several studies were made to study these factors.
Flyash samples collected from flue gas entering the SDA were analyzed for
dry C content using a modification of ASTM Method D3174 for an additional weight
measurement at an intermediate temperature to account for the water of hydration
associated with calcium chloride hydrate. The LOI was the percent weight lost
between samples heated and held for given periods first at the intermediate
temperature of 210°C (410°F) and then at the final temperature of 750°C (1380°F).
Ten of the 11 samples analyzed had a dry C content of 1% or less by weight. The
other samples had a dry C value of 2.2%. The highest dry C value (2.2%) occurred
during Condition 2 when C powder was injected at 5.4 kg/hr at the economizer
outlet, just upstream of the flyash sampling ports. Thus, C injection was believed
responsible for this high dry C value. All LOIs exceeded 4% but were below 6.5%.
As noted earlier, the average Hg removal for tests without C injection was
about 30%. This corresponded to a flyash carbon content of 1% or less. Tests at the
Mid-Connecticut facility showed Hg removals exceeding 95% and probable flyash C
contents of several percent/ This comparison supports the premise that the C
content of flyash affects Hg emission control.
Time stability of Hg retention in ash was studied for the fabric filter and
combined ashes. The samples evaluated for Hg content were weighed initially and
after 14 and 28 days in storage, respectively, in an oven at 54°C (130°F), a refrigerator
at 7°C (45°F), and room air at 16°C (60°F). At the highest temperature, the change in
weight in five samples (one without C addition, two with coal-based C, and one each
with lignite- and wood-based C addition to flue gas) indicated a maximum weight
change of about 10% (one coal-based C sample after 14 days) and values below 10%
for the other samples. However, the highly hygroscopic nature of the samples
stored at the lower temperatures led to absorption of water when weighed in air.
Laboratory analysts noted while weighing these samples that the samples gained
weight so rapidly it was difficult to record a weight. Whether the gain of weight via
13

-------
absorption of water by the sample during its weighing accounts for all the
differences in weights over time cannot be determined, but a method is being
developed to study this effect.
The combined ash samples are representative of the ash being landfilled. The
Hg concentrations before storing the ash samples in the three temperature
environments noted above (oven, refrigerator, and room air) were less than 2 |ig/g
for Condition 4 (no C injection) and about 9 Jig/g for Condition 6. The results
indicate that Hg was not volatilized from the combined ash after its capture over the
28-day period of this study. The highest ash exposure temperature [54°C (130°F)] was
greater than expected in combined ash landfills and was expected to result in greater
Hg emission than at the lower temperatures tested.
SUMMARY
The field evaluation on Unit 2 of the OMSS MWC facility near Crows
Landing, CA, showed that C feed rate affected the control of Hg emission but C type
and feed location did not. Increasing the feed rate from zero to 5.4 kg/hr improved
Hg reduction from about 30% to over 90% and, correspondingly, lowered Hg
emission in the stack. The low C feed rate (1.3 kg/hr) resulted in lower Hg
reductions and greater variability in the reductions than did the medium (2.8 kg/hr)
and high (5.4 kg/hr) feed rates when comparing test runs.
NH3 injection in the SNCR system appeared to have little or no effect on Hg
removal: the NH3 concentration (NH3 slip) in flue gas entering the SDA seemed to
be independent of whether the SNCR was operated or not. Additional data are
needed to verify if NH3 injection negatively impacts Hg emission control because
the limited test data preclude a firm conclusion. No correlation between Hg
removal and lime feed rate (or SO2 removal) was possible with the limited test data
obtained. Gas temperature entering the FF over the temperature range studied (136-
151 °C) did not affect Hg emission control.
Test data indicated that the dry C content of flyash entering the SDA was 1%
or less and that the LOI of the flyash was under 7%. The low C content of the flyash
was consistent with the low Hg removal obtained without C injection. The time
stability study of combined ash which is landfilled indicated that Hg captured with
this ash did not volatilize from this ash even when held in an oven at 54°C for the
28-day study.
ACKNOWLEDGMENTS
The authors express their appreciation to Fred Engelhardt and Anthony Velez
of the OMSS facility; Jeffrey Hahn of Ogden Martin Systems; Bert Brown of Joy
Environmental Technologies, Inc.; and Robert Edwards of American Norit
Company, Inc. for their assistance and cooperation in planning and executing the
test program. Foston Curtis of the Emission Measurements Branch of EPA's Office
14

-------
of Air Quality Planning and Standards (OAQPS) provided valuable assistance in
monitoring the sampling and measurement procedures, while the Industrial
Studies Branch of OAQPS provided significant support to the program. Their
assistance is gratefully acknowledged.
REFERENCES
1.	Clean Air Act Amendments of 1990, P.L. 101-549, U.S. Congress, Washington,
DC, November 15, 1990.
2.	Nebel, K.L. and D.M. White, "A Summary of Mercury Emissions and
Applicable Control Technologies for Municipal Waste Combustors," Report
prepared by the Radian Corporation for the Office of Air Quality Planning and
Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC,
September 1991. Docket No. A-89-08, U.S. Environmental Protection Agency,
Washington, DC.
3.	Guest, T.L. and O. Knizek, "Mercury Control at Burnaby's Municipal Waste
Incinerator," 84th Annual Meeting & Exhibition, Air & Waste Management
Association, Vancouver, B.C., June 1991.
4.	Brown, B. and K.S. Felsvang, "Control of Mercury and Dioxin Emissions from
United States and European Municipal Solid Waste Incinerators by Spray Dryer
Absorption Systems," Second International Conference on Municipal Waste
Combustion, Tampa, FL, April 1991.
5.	Moller, J.T. et al., "Process for Removal or Mercury Vapor and/or Vapor of
Noxious Organic Compounds and/or Nitrogen Oxides from Flue Gas from an
Incinerator Plant," U.S. Patent No. 4,889,689, issued December 26, 1989.
6.	Test Report for Air Compliance Tests at Stanislaus Waste-to-Energy Company.
Prepared for Ogden Projects, Inc., Fairfield, NJ, for submittal to Stanislaus Air
Pollution Control District, Modesto, CA, by Energy Systems Associates, Tustin,
CA, February 1989.
7.	Brna, T.G., J.D. Kilgroe, and A. Finkelstein, "The Joint EC/EPA Mid-
Connecticut Test Program: A Summary," Second International Conference
on Municipal Waste Combustion, Tampa, FL, April 1991.
8.	White, D.M. et al., "Parametric Evaluation of Activated Carbon Injection for
Control of Mercury Emissions from a Municipal Waste Combustor," Paper
No. 92-40.06, 1992 Annual Meeting, Air & Waste Management Association,
Kansas City, MO, June 1992.
15

-------
AFFRJ _ p-QQQ ni TECHNICAL REPORT DATA
¦" (Phase readIsmvctions on (he reverse before complet'
1. REPORT NO,
EPA/600/A-92/134
2,
3.
4, TITLE ANO SUBTITLE
Reducing Mercury Emission from Municipal Waste
5. REPORT DATE
Combustion with Carbon Injection into Flue Gas
6, PERFORMING ORGANIZATION COOE
7. AUTHOR(S)
T.G. Brna, J. D. Kilgroe,
and C. A. Miller
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME ANO ADDRESS

10. PROGRAM ELEMENT NO.
See Block 12






11. CONTRACT/GRANT NO.
NA (Inhouse)
12. SPONSORING AGENCY NAME ANO ADDRESS
EPA, Office of Research and Development

13. TYPE OF REPORT AND PERIOD COVERED
Published paper; 7/91-5 ^92
Air and Energy Engineering Research Laboratory
Research Triangle Park, North Carolina 27711
14. SPONSORING AGENCY CODE
EPA/600/13
15, SUPPLEMENTARY NOTES AEERL
541-2683. Presented at EC
project officer is Theodore G. Brna, Mail Drop 65, 919/
C World '92, Washington, DC, 6/14-17/92.
is. abstractpaper gives results of tests (to aid in developing emission rules) in JuLy
and August 1991 on a 360-tonne/day mass burn municipal waste combustor (MWC) at
the Ogden Martin Systems of Stanislaus, Inc. facility near Crows Landing, CA. Test
results indicated that carbon (C) addition was effective in improving mercury (Hg) re'
moval: removal increased with increasing C feed rate. Hg removal improved from
about 30% without C addition to over 90% at the highest C feed rate tested. The test
data obtained for evaluation of the secondary test objectives suggested that (1) ammo-
nia (NH3) addition (which occured with low slip) had no apparent adverse effect on Hg
control, (2) lime stoichiometry had no effect on Hg emissions over the narrow range
studied, (3) the fabric filter (FF) gas temperature over the range tested did not
affect Hg control, and (4) the Hg content of ash/residue stream samples remained
relatively constant over a period of 28 days in samples held at 54 C.
17.
KEY WORDS ANO DOCUMENT ANALYSIS


a. DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. cos ATI Field/Group
Pollution
Mercury
Carbon
Flue Gases
Wastes
Combustion
Pollution Control
Stationary Sources
Municipal Waste Com-
bustion
13 B
07B
21B
14G
18. DISTRIBUTION STATEMENT
Release to Public

19, SECURITY CLASS (This Reportj
Unclassified
21. NO. OF PAGES
16


20 SECURITY CLASS (This page J
Unclassified
22. PRICE
EPA Form 2220-1 <9 73J

-------