United States
Environmental Protection
Agency
Research and Development
Air and Energy Engineering
Research Laboratory
Research Triangle Park, NC 27711
EPA/600/SR-92/192 January 1993
i§rEPA Project Summary
Emission Test Report:
OMSS Field Test on Carbon
Injection for Mercury Control
K L. Nebel, D. M. White, C. R. Parrish, T. G. Zirkle, M. A. Palazzolo, and
M. W. Hartman
In 1991, the U.S. Environmental Pro-
tection Agency conducted a paramet-
ric evaluation of powdered activated
carbon for control of mercury (Hg) emis-
sion from a municipal waste combus-
tor (MWC) equipped with a lime spray
dryer absorber/fabric filter (SD/FF). The
primary test objectives were to evalu-
ate the effect of carbon type, feed rate,
and feed location on Hg emissions and
control efficiency. Secondary process
parameters studied included the impact
of ammonia injection for nitrogen ox-
ides control, SD outlet temperature, and
SD/FF acid gas control efficiency on
Hg removal. The time stability of Hg
collected with ash was also studied.
Conducted at the Odgen Martin Sys-
tems of Stanislaus, Inc. MWC, near
Modesto, CA, testing covered 16 sys-
tem operating conditions, including
normal unit operation (no carbon injec-
tion) and operation without ammonia
injection.
Test results showed that the two pri-
mary variables affecting both Hg emis-
sion and control efficiency were car-
bon feed rate and uncontrolled Hg lev-
els. The results also indicated that Hg
emissions were reduced by over 80%
at high carbon addition rates. At low
carbon feed rates, both the average Hg
emissions rate and the variability in Hg
levels during individual tests were sig-
nificantly higher. The secondary param-
eters did not affect Hg control over the
range of values tested, nor did the mass
of Hg collected with ash change over a
28-day period.
This Project Summary was developed
by EPA's Air and Energy Engineering
Research Laboratory, Research Tri-
angle Park, NC, to announce key find-
ings of the research project that is fully
documented in a separate report of the
same title (see Project Report ordering
information at back).
Introduction
Section 129 of the 1990 Clean Air Act
requires the U.S. Environmental Protec-
tion Agency (EPA) to promulgate mercury
(Hg) emission limits for municipal waste
combustors (MWCs). Data from existing
MWCs equipped with lime spray dryer ab-
sorber (SD) and fabric filter (FF) emis-
sions control systems indicate highly vari-
able Hg collection efficiencies ranging from
zero to over 95%. To help develop a bet-
ter understanding of factors influencing
the control of Hg emissions from MWCs,
EPA's Air and Energy Engineering Re-
search Laboratory (AEERL) contracted
with Radian Corporation to conduct a se-
ries of tests to evaluate the injection of
powdered activated carbon as a Hg con-
trol technique. These tests were conducted
during the Summer of 1991 on Unit 2 of
the Ogden Martin Systems of Stanislaus,
Inc. (OMSS) MWC facility near Modesto,
California. The objectives of these tests
were to evaluate:
• the effectiveness of injecting activated
carbon into the flue gas upstream of an
SD/FF system and into the lime slurry
feed to the SD to control Hg emissions;
• the impact of ammonia (NH3) injection
in the Thermal DeNOx® process on the
control of Hg^emissions;
• the effect of SD/FF system operating
conditions on Hg control and emissions;
and
Printed on Recycled Paper
-------
• the stability (permanence of capture) of
Hg collected on ash/residue over time.
Facility Description
The OMSS MWC consists of two identi-
cal Martin GmbH mass burn waterwall
(MB/WW) combustors, each capable of
combusting 400 tons* per day (tpd) of
municipal solid waste. Each unit is
equipped with an Exxon Thermal DeNOx*
NH, injection system for reducing emis-
sions of nitrogen oxides (NOX) and a Flakt
SD/FF system for reducing emissions of
acid gases, particulate matter (PM), met-
als, and organics.
Test Matrix
Table 1 is the test matrix for the 16
conditions examined. Primary variables
examined were carbon type, carbon feed
rate, and carbon feed location. Secondary
variables were NH3 injection, FF gas tem-
perature, and acid gas control efficiency.
Most of the conditions consisted of three
1-hrtest runs. During Condition 15, only
two 1-hrtest runs were conducted.
To evaluate baseline Hg levels, Condi-
tions 4 and 5 runs were conducted with-
out activated carbon injection. The remain-
ing tests examined Hg removal efficiency
when injecting activated carbon. To ex-
amine the effect of injecting carbon in the
absence of NH, injection, Condition 7 was
run with the Thermal DeNOx® system off
while injecting carbon at the SD inlet.
Three carbon types were used to inves-
tigate the impact of carbon characteristics
on Hg control. Activated carbon made from
coal was used in 10 of the test conditions.
During two test conditions, a lignite-based
carbon was used. The third carbon type,
which was also used during two test con-
ditions, was wood-based. The wood car-
bon was chemically activated, while the
coal and lignite carbons were thermally
activated. Of these three, the lignite-based
carbon had the lowest specific surface
area and average particle size, and the
highest average pore radius. The wood-
based carbon had the highest specific sur-
face area, and the coal-based carbon had
the smallest average pore radius and
tamped density.
The effect of carbon feed rate using the
coal-based carbon was examined during
Conditions 3, 6, and 8. During these tests,
carbon was injected at the SD inlet, down-
stream of the inlet sampling location. Dur-
ing these three conditions, the average
feed rates were 2.8, 12.1, and 6.1 Ib/hr,
respectively, corresponding to approxi-
mately 17, 73, and 37 mg/dscm corrected
M Ion a 907 kg.
to 7% oxygen (O2). Additional testing ex-
amining the effect of carbon feed rate was
conducted at the SD inlet during Condi-
tions 9 and 10 while injecting lignite-based
carbon and during Conditions 13 and 14
while injecting wood-based carbon.
The effect of injecting carbon at alter-
nate locations was also examined. During
Conditions 1 and 2, carbon was injected
through three ports in the horizontal duct
just downstream of the economizer outlet,
but upstream of the sampling location for
the SD inlet. Conditions 15 and 16 exam-
ined the effect of mixing carbon with the
lime slurry used in the SD.
Condition 11 was conducted to assess
the effect of reduced FF temperature.
Condition 12 was conducted to study the
effect of reduced lime stoichiometry (i.e.,
reduced sulfur dioxide [SOJ and hydro-
gen chloride [HCI] control) on Hg emis-
sions.
Impact of Carbon Feed
Location, Type, and Feed Rate
on Mercury Emissions
Carbon Feed Location
Figure 1 shows the relationship between
carbon injection location and Hg removal
for the low and high feed rates of the
coal-based carbon. At the low feed rate of
approximately 3 Ib/hr, removal efficiencies
were 66-85% with carbon injection at the
economizer outlet feed location (Condi-
tion 1), and 53-77% with carbon injection
at the SD inlet feed location (Condition 3).
At the high feed rate of approximately 12
Ib/hr, Hg removal was 88-92% with car-
bon injection at the economizer outlet feed
location (Condition 2) and 91-98% with
carbon injection at the SD inlet feed loca-
tion (Condition 6). When the carbon was
injected into the SD with the lime slurry,
Hg removal was 88-96% (Condition 16).
At both carbon feed rates, statistical analy-
sis using the t-statistic at the 95% confi-
dence level indicated that the differences
in Hg reductions as a function of feed
location were not statistically significant.
Carbon Type
Figure 2 presents Hg removal efficiency
as a function of carbon type for the low,
medium, and high carbon feed rates.
Based on statistical analysis of the data,
carbon type did not significantly influence
Hg emissions or Hg removal at any of the
feed rates.
Carbon Feed Rate
As indicated by Figures 1 and 2, carbon
feed rate had a significant impact on Hg
removal for all carbon feed locations and
types. To better define the impact of car-
bon feed rate on Hg reduction efficiency
and outlet concentration, stepwise multi-
variate regression analysis was used. In
this analysis, Hg reduction efficiency and
outlet concentration were the dependent
variables, and uncontrolled Hg levels, car-
bon feed rate, carbon type, carbon feed
location, and NH3 injection rate were the
independent variables.
For the analysis, Hg reduction efficiency
values were converted to emissivity val-
ues (100 minus percent reduction). Both
the actual values and natural log trans-
form of Hg emissivity and outlet Hg con-
centrations were evaluated as dependent
variables. Because of the tendency of sor-
bent to have diminishing effectiveness as
carbon feed rate increases (i.e., decreas-
ing sorbent utilization), three formats for
carbon feed rate were examined: the feed
rate as measured, the measured feed rate
raised to the 0.5 power (i.e., square root),
and the feed rate raised to the 0.7 power.
The best predictive model identified for
Hg percent reduction was based on the
square root of the carbon feed rate and
the uncontrolled Hg level. The regression
equation was:
In(IOO-PRED) = 4.81 - 0.639*(CFR)°-5-
0.000776*HGIN (1)
where PRED is Hg percent reduction, CFR
is carbon feed rate (in Ib/hr), and HGIN is
the uncontrolled (inlet) Hg level (in ug/
dscm at 7% O2). The "goodness of fit" (R2)
of this model is 0.762. Figure 3 shows the
measured and predicted values of Hg re-
duction versus inlet Hg level. The three
curves for predicted reduction are based
on carbon feed rates of 3, 6, and 12 Ib/hr.
These carbon mass feed rates correspond
to roughly 18, 36, and 72 mg of carbon
per dscm of flue gas at 7% O2. Although
there is significant scatter in the data,
particularly at the low carbon feed rates,
the model predicts the expected increase
in Hg reduction at both higher carbon feed
rates and higher uncontrolled Hg levels.
Based on this model, a carbon feed rate
of 12 Ib/hr would achieve an average Hg
reduction of at least 90% over the entire
range of uncontrolled Hg levels shown in
Figure 3. However, note that, because of
scatter in inlet and outlet Hg concentra-
tions caused by variations in process op-
eration and measurement imprecision, cal-
culated Hg reductions during individual
runs will be higher and lower than the
levels indicated by the predicted curves.
The best predictive model for outlet Hg
concentration was also based on the
square root of carbon feed rate and un-
controlled Hg level. This regression equa-
tion was:
-------
Table 1. Test Matrix for OMSS Emissions Control Field Test (1991)
Condition
No.
1
2
3
4 (BL)"
5(BL)
6
7
8
9
10
11
12
13
14
15
16
Test
Date
(1991)
7/22=
7/23
7/24
7/25
7/29
7/26
7/30
7/31
8/1
8/7
8/5'
8/5 <
8/2
8/6<>
8/10
8/10
Number of
Test Runs
3
3
3
3
3
3
3
3
3
3
3
3
3
3
2
3
Operating Parameters
Thermal
DeNOx
Normal
Normal
Normal
Normal
Off
Normal
Off
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Carbon Feed
Rate (Ib/hr)
2.8
12.0
2.8
0
0
12.1
2.9
6.1
2.8
12.3
2.9
2.8
3.2
6.6
18.3
12.2
Carbon *
(Raw Material)
Coal
Coal
Coal
None
None
Coal
Coal
Coal
Lignite
Lignite
Coal
Coal
Wood
Wood
Coal
Coal
Fabric Filter
Temperature
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Low
Normal
Normal
Normal
Normal
Normal
Carbon *
Injection Location
Econ. Outlet
Econ. Outlet
S.D. Inlet
None
None
S.D. Inlet
S.D. Inlet
S.D. Inlet
S.D. Inlet
S.D. Inlet
S.D. Inlet
S.D. Inlet
S.D. Inlet
S.D. Inlet
w/lime slurry
w/lime slurry
Lime
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Low
Normal
Normal
Normal
Normal
'Lignite Raw Material = DARCO FGD, Surface Area = 600 rrf/g.
Coal Raw Material = DARCO PC100, Surface Area = 950 rrf/g.
Wood Raw Material = DARCO KB, Surface Area = 1500 rrf/g.
"S.D. Inlet = Spray Dryer Inlet, Econ. Outlet = Economizer Outlet.
"One run conducted on 7/23/91.
"BL = Baseline
'One run conducted on 8/6/91.
'One mn conducted on 8/2/91.
9One run conducted on 8/7/91.
I
u
100
90
80
70
60
50
I
CO
LFR
HFR
Carbon Feed Rate
(Coal-Based)
LFR = Low Feed Rate (approximately 3 Ib/hr)
HFR = High Feed Rate (approximately 12 Ib/hr)
Econ = Economizer Outlet
SDi = Spray Dryer Inlet
SL = Slurry
Figure 1. Effect of carbon feed location on mercury removal.
In(HGOUT) = 5.66 - 0.649*(CFR)°-S +
0.000724*HGIN (2)
where HGOUT is outlet Hg concentration
in jig/dscm at 7% O2, and CFR and HGIN
are as defined in Equation 1. The "good-
ness of fit" of this model is 0.777. Figure 4
shows the measured values of outlet Hg
level versus inlet Hg level, and the regres-
sion model curves at carbon feed rates of
3, 6, and 12 Ib/hr. As with percent reduc-
tion, there is significant scatter in the data
at low carbon feed rates. As expected,
the modeled regression curves indicate
that Hg outlet levels increase with in-
creased uncontrolled Hg levels and de-
creased carbon feed rate. The increased
slope of the regression curves at high
uncontrolled Hg levels indicates a ten-
dency toward decreased carbon utiliza-
tion. This suggests greater saturation of
individual carbon-to-Hg adsorption sites.
Based on this model, a carbon feed rate
of 12 Ib/hr is predicted to achieve an aver-
age outlet Hg concentration of less than
80 jig/dscm at 7% O., over the entire range
of uncontrolled Hg levels. Again, however,
because of the variability in outlet Hg lev-
els caused by changes in process opera-
-------
100
90 -
eo
70
60
SO
LFR
MFR
Carbon Feed Rate
HFR
/.F/7 =*Low Feed Rate (approximately 3 Ib/hr)
MFR - Medium Feed Rate (approximately 6 Ib/hr)
HFR = High Feed Rate (approximately 12 Ib/hr).
Figure 2. Effect of carbon type on mercury removal.
100
90
80
70
60
50
40
30
300
700
1100
500 '" 900 1300
Inlet Hg Concentration (jig/dscm at 7% C>2, Dry Basis)
• LFR = Low Feed Rate (approximately 3 Ib/hr)
t MFR ~ Medium Feed Rate (approximately 6 Ib/hr)
...* HFR * High Feed Rate (approximately 12 Ib/hr)
Figure 3. Effect of carbon feed rate on mercury removal.
tion and measurement imprecision, indi-
vidual measurements of stack Hg levels
will be higher and lower than the levels
indicated by the curves in Figure 4.
Impact of Other Operating
Parameters on Mercury
Emissions
Ammonia Injection
To study the effect of NH3 injection on
Hg removal, NH3 injection was shut off
during Conditions 5 and 7. In addition,
during Condition 5, no carbon was in-
jected. Condition 4 was the baseline con-
dition without carbon injection, but with
NH3 injection. As shown on the left half of
Figure 5, Hg removal during Conditions 4
and 5 are comparable, with removal effi-
ciencies of 15-36% for Condition 4 and
18-46% for Condition 5.
The effect of NH3 injection on Hg re-
moval while injecting carbon can be ex-
amined by comparing Conditions 3 and 7.
During both conditions, coal-based car-
bon was injected at the SD inlet at an
average rate of approximately 2.8 Ib/hr.
During Condition 3, NH3was injected, while
during Condition 7, no NH3 was fed. As
shown on the right half of Figure 5, Hg
removals were 53-77% during Condition
3, and 48-56% during Condition 7. Given
the limited number of samples, it is not
possible to conclude that the values mea-
sured at these two conditions are statisti-
cally different.
It is noteworthy, however, that the NK
levels in the flue gas measured at the SD
inlet were consistently low (less than 5
ppmv) during all but two runs and were
not significantly affected by the Thermal
DeNOx® system's being on or off. As a
result, it is not possible to clearly establish
the impact of NH3 level in the flue gas on
Hg collection efficiency.
Lime Feed Rate/Acid Gas
Removal
A potential concern identified with Hg
control was whether the conversion of va-
porous elemental Hg (Hg°) to paniculate
mercuric chloride (HgCI2) could be reduced
at high lime feed rates. To examine this
potential, the lime feed rate was lowered
during Condition 12. Because of problems
with direct measurement of lime feed rate,
SO2 control efficiency was used as a sur-
rogate indicator. The SO removal during
Condition 12 was 66-77%, and averaged
73%. The average SO2 removal during
most other conditions was at least 90%.
Condition 3 was similar to Condition 12,
but with normal SO2 control. Comparing
the results from Conditions 3 and 12, the
range of Hg removals is similar — Condi-
-------
tion 3 at 53-77% (average 66%) and
Condition 12 at 36-83% (average 59%).
As shown in Figure 6, there does not
appear to be a correlation between Hg
and SO2 control efficiency over the limited
SO2 range of these tests. During both test
conditions, HCI control efficiencies were
greater than 96%.
Fabric Filter Temperature
Unlike most other metals (e.g., cadmium,
lead), Hg can exist as a vapor at normal
SD/FF temperatures and, therefore, does
not readily condense onto PM as do other
metals. During the testing, average stack
temperatures generally were 275-295°F.*
The stack temperatures during Condition
11 averaged 282°F. By comparison, FF
temperatures during Condition 3 averaged
294°F. For both conditions, coal-based
carbon was injected at the SD inlet at an
average rate of roughly 2.8 Ib/hr. Compar-
ing average Hg removal levels, there is
essentially no difference between
Conditions 3 (66%) and 11 (64%).
Mercury Stability in Ash
Streams
The stability of Hg captured in ash has
important consequences with regard to
storage, transport, and disposal of the ash.
Therefore, studies were conducted to de-
termine the stability of Hg on combined fly
ash and bottom ash and on FF ash as a
function of time and temperature. (Mois-
ture, carbon, and loss on ignition analy-
ses were also conducted on FF and SD
ash samples.)
The results of the combined ash time
stability study indicate that Hg did not
volatilize from the ash over the 28-day
period of study. The samples were held in
a heated environment, in a refrigerated
environment, and at room temperature.
These results are of significant importance
since these samples represent the ash
that is normally landfilled, and concern
had been raised over the fate and stability
of Hg in the landfilled ash over time.
The results of the studies on the FF ash
samples are somewhat inconclusive. All
of the FF ash was collected dry, prior to
the usual quenching the FF ash experi-
ences. The dry FF ash is extremely hy-
groscopic and, because of the difficulty of
removing hydrated water from the samples
and the rapid rehydration of ash when the
samples were removed from the oven,
precise measurements were not possible.
The data for the samples held in the re-
528
300
\
500
700
I
900
1100
1300
Inlet Hg Concentration (Mg/dscm at 7% O?, Dry Basis)
—5 LFR =Low Feed Rate (approximately 3 Ib/hr)
—t MFR = Medium Feed Rate (approximately 6 Ib/hr)
„ _t HFR = High Feed Rate (approximately 12 Ib/hr)
Figure 4. Effect of carbon feed rate on outlet mercury emissions.
g
c:
iS
£
70
60
50
40
30
20
m
-
•S
CO X
Q I AT
.(o S
" 1 ? S A
E
• ^
-
.
:
'= ' O
o to
1 !
Off
Carbon Injection
On
Figure 5. Effect of ammonia injection on mercury removal.
*°C = 5/9(°F-32)
-------
K. L Nebel, D. M. White, C. R. Parrish, T. G. Zirkle, M. A. Palazzolo, and M. W.
Hartman are with Radian Corp., Research Triangle Park, NC 27709.
Theodore G. Brna is the EPA Project Officer (see below).
The complete report, entitled "Emission Test Report: QMSS Field Test on Carbon
Injection forMercuryControl"(OrderNo.PB93-105518/AS;Cost: $27.00, subject
to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22181
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Air and Energy Engineering Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
United States
Environmental Protection Agency
Center for Environmental Research Information
Cincinnati, OH 45268
BULK RATE
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
Official Business
Penally for Private Use
$300
EPA/600/SR-92/192
------- |