United States
Environmental Protection
Agency
Air and Energy Engineering
Research Laboratory
Research Triangle Park, NC 27711
Research and Development
EPA/600/SR-93/181 January 1994
&EPA Project Summary
Emission Test Report, Field Test
of Carbon Injection for Mercury
Control, Camden County
Municipal Waste Combustor
D. M. White, W. E. Kelly, M. J. Stucky, J. L. Swift, and M. A Palazzolo
In 1992, the U.S. Environmental Pro-
tection Agency (EPA) conducted a para-
metric testing to evaluate the injection
of powdered activated carbon to con-
trol volatile pollutants in municipal
waste combustor (MWC) flue gas. This
testing was conducted at a spray dryer
absorber/electrostatic preciprtator (SD/
ESP)-equipped MWC in Camden
County, New Jersey. The primary test
objectives were to evaluate the effect
of carbon type, feed rate, feed method,
and ESP operating temperature on
emissions of mercury (Hg) and chlori-
nated dioxins and furans (CDD/CDF),
and to assess the impact of carbon
injection on the particulate matter con-
trol performance of the ESP. Second-
ary objectives were to examine the
impact of carbon injection on emissions
of other metals and volatile organic com-
pounds (VOCs). This testing included
operation of three different carbon in-
jection systems and examined 16 dif-
ferent SD/ESP and carbon injection
system operating conditions. This test
was conducted as a follow-on to an EPA-
funded test program at a SD/fabric filter-
equipped MWC that focused on the
performance of carbon injection for con-
trolling Hg emissions.
The test results indicate that carbon in-
jection upstream of a SD/ESP can achieve
high levels (greater than 90%) of Hg and
CDD/CDF reduction. Key system operat-
ing parameters are carbon feed rate,
carbon feed method, and ESP tempera-
ture. No detrimental impacts on ESP
performance were identified. The study
also found that carbon injection does
not have a significant impact on emis-
sions of the other metals sought or of
VOCs.
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
In June 1987, the U.S. EPA's Office of
Air Quality Planning and Standards
(OAQPS) announced its intention to de-
velop new air pollution rules for MWC
facilities. The following fall, EPA's Air and
Energy Engineering Research Laboratory
(AEERL) and OAQPS began a compre-
hensive field test and engineering assess-
ment program to evaluate technologies
for controlling MWC emissions. The field
test program focused on the use of a lime
spray dryer absorber and fabric filter
baghouse (SD/FF) or a lime SD/ESP sys-
tem for controlling MWC emissions. Regu-
lations for large MWCs were proposed in
late 1989 and finalized in 1991.
In November 1990, the Clean Air Act
Amendment (CAAA) required EPA to add
Hg emission limits to the MWC standards.
Under the direction of EPA, Radian Cor-
poration prepared a technical memoran-
dum summarizing the effectiveness of
various technologies in controlling Hg
emissions. Based on U.S. field tests and
research in Europe and Canada, it was
concluded that the control of Hg emis-
sions in MWC SD/particulate matter (PM)
Printed on Recycled Paper
-------
control systems depends on the amount
of carbon in the fly ash. Modern mass
burn combustors, which typically contain
relatively low amounts of carbon in their
fly ash, exhibited Hg control efficiencies
with acid gas scrubbers ranging from 0 to
40%. Refuse-derived fuel combustors,
which contain substantially more carbon
in their fly ash, had Hg control efficiencies
exceeding 95%. In these applications Hg
vapor does not condense but is adsorbed
on carbonaceous particles that can be
subsequently collected in PM control de-
vices.
Research and field tests in Europe indi-
cated that powdered activated carbon can
be used to enhance control of Hg emis-
sions in SD/PM control systems. EPA-
sponsored tests in 1991 at the Ogden
Martin Systems of Stanislaus, Inc. (OMSS)
MWC indicated that powdered activated
carbon was effective in controlling Hg
emissions in mass burn MWCs with SD/
FF systems. It was found that carbon feed
rate is the primary factor affecting Hg con-
trol. The OMSS testing also indicated that
there are no significant differences in Hg
control performance as a function of the
physical characteristics of the carbon (origi-
nal material, particle size, pore size, and
density), the method of injection (as a dry
powder or mixed with the SD's lime slurry),
or the location of injection (economizer
exit, SD inlet, and into the SD). However,
it was not possible to determine how much
of the Hg is removed by adsorption onto
carbon particulate in the free stream with
subsequent PM collection and how much
Hg is removed by adsorption as the flue
gas flows through the FF cake containing
captured carbon.
Because of uncertainties regarding the
mechanisms of Hg capture by carbon in
SD/FF systems, it is not possible to di-
rectly translate the data collected at SD/
FF-equipped MWCs to units equipped with
SD/ESP systems. Available U.S. data on
SD/ESP-equipped MWCs indicated incon-
sistent and often poor Hg control. Data on
the collection of Hg by SD/ESPs is of
interest because of the number of such
systems already operating commercially,
as well as the potential for retrofitting ex-
isting ESP-only systems with SDs as
needed for control of acid gas and or-
ganic emissions.
In addition, very little data are avail-
able from either SD/FF- or SD/ESP-
equipped MWCs on the effectiveness of
carbon injection for reducing emissions
of CDD/CDF and various VOCs.
In 1992 EPA sponsored tests at the
Camden County, New Jersey, MWC facil-
ity to evaluate the use of activated carbon
for controlling emissions in SD/ESP sys-
tems. The Camden County and the OMSS
tests were directed by AEERL. The report
summarizes the objectives and results of
the Camden County tests.
Purpose of Tests
The purpose of the Camden County
tests was to evaluate the effectiveness of
powdered activated carbon for reducing
emissions of Hg, "other" metals, and trace
organic emissions from mass burn MWCs
equipped with SD/ESP systems. Specific
objectives of the tests were to evaluate:
The relationship of carbon feed rate,
the inherent carbon in fly ash, and
Hg capture;
The effects of carbon injection on
the control of "other" metals, CDD/
CDF, and VOCs;
The effect of carbon injection rate
and method (dry powder or slurried
with lime) on Hg and CDD/CDF
control;
The effect of ESP operating tem-
perature on Hg collection, and
Whether there are long-term im-
pacts of carbon injection on ESP
performance in collecting PM and
"other" metals.
Test Design
Description of Facility
The Camden County MWC is owned
and operated by Camden County Energy
Recovery Associates, a subsidiary of Fos-
ter Wheeler Power Systems, Inc. Located
in Camden, New Jersey, it began operat-
ing in 1991. The facility contains three
identical mass-burn waterwall combustion
units, designated as Units A, B, and C.
Each unit is capable of burning 317 tonnes
(350 tons) per day of municipal solid waste
(MSW), and collectively they provide steam
for two 17-MW turbine generators. A gen-
eral schematic of a test unit showing car-
bon injection and sampling locations is
given in Figure 1.
The air pollution control system on each
combustor consists of a Deutsche Babcock
SD and a Belco five-field ESP. The flue
gas leaves the economizer, passes down
through a vertical circular duct, through a
90-degree elbow, and through a horizon-
tal circular duct before entering a cyclone.
The cyclone separates coarse PM from
the flue gas and distributes flue gas to six
vertical flow tubes that connect to the base
of the SD vessel. A two-fluid nozzle lo-
cated in the top of each flow tube injects
lime slurry up into concurrently flowing
flue gas. The lime slurry flow can be con-
trolled by stack sulfur dioxide (SO2) con-
centration or it can be set to provide a
fixed lime flow rate. Dilution water flow is
controlled by the SD's exit gas tempera-
ture. After leaving the SD, flue gas passes
through an inverted U-shaped circular duct
before entering a five-field ESP. During
normal operation, only four of the ESP
fields are in operation. Flue gas from each
ESP is ducted into a separate flue in the
stack. The stack contains four circular
flues, one per operating unit, and one for
a future unit.
The process control systems include a
Bailey Net 90 for the boiler, a separate
control and data display system for the
SD/ESP, and two separate continuous
emission monitoring (CEM) data acquisi-
tion systems. The CEM systems include
extractive monitors for SO2 and oxygen
(O2) at the economizer exit and monitors
for O2, carbon dioxide (CO2), water (H2O),
carbon monoxide (CO), total hydrocarbons
(THCs), methane (CH4), SO2, hydrogen
chloride (HCI), nitrogen oxides (NOX), and
opacity in the stack.
Test Matrix
The Camden County MWC test project
encompassed three distinct testing efforts
and was conducted in two phases (see
Table 1). Phase I concerned only Unit B.
These tests, defined as Phase I-B tests,
were used to select the carbon type and
carbon feed rates to be used for the Phase
II performance tests. The Phase II tests
included two distinct efforts. The primary
effort, conducted on Unit B (II-B tests),
investigated the effects of key carbon in-
jection system operating variables on Hg
control efficiency. The other Phase II tests,
conducted on Unit A (II-A tests), investi-
gated the potential long-term impacts of
extended carbon injection on ESP perfor-
mance.
Triplicate sampling runs were conducted
for each test condition and one test condi-
tion was completed during each test day
Plant process instruments and CEM equip-
ment were used to monitor combustor and
SD/ESP operating conditions during each
run.
Phase I-B Tests. During Phase I-B
tests, carbon was injected as a dry pow
der into the flue gas duct just upstream or
the SD (see Figure 1). The tests evalu
ated Hg control levels for two carbon types
and two carbon feed rates. Both of the
tested carbons, Darco PC-100 and Darco
FGD from the American Norit Company,
were used during the OMSS tests. The
first carbon (PC-100) was a thermally ac
tivated, bituminous-coal-based carbon with
medium surface area and high tamped
density. The second carbon (FGD) was
thermally activated from lignite and had a
lower surface area, smaller average par
tide size, and lower tamped density than
-------
Legend
A : Inlet Sampling Location
B: Dry Carbon Injection Location
C : Slurry Carbon Injection Location
D : Outlet Sampling Location
Stack
I.D. Fan Electrostatic Spray Ash Conveyors
Predpitator Dryer
Absorber
Figure 1. Schematic of the Camden County Municipal Waste Combustor.
Table 1. Test Matrix-Camden County MWC, Spring 1992
Phase-
Condition
I-B1
I-B2
I-B3
I-B4
I-B5
II-B6
II-B7
II-B8
II-B9
II-B10
II-B11
II-B12
II-B13
II-A1
II-A2
II-A3
II-A4
II-A5
ESP Inlet
Temperature
°C(°F)
132(270)
132(270)
132 (270)
132 (270)
132(270)
177(350)
177(350)
132 (270)
132(270)
132 (270)
132 (270)
132 (270)
132 (270)
132(270)
132 (270)
132(270)
132 (270)
132 (270)
Number of
Operating
ESP Fields
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
3
Carbon
Type
None
FGD
PC-100
PC-100
FGD
None
FGD
FGD
FGD
None
FGD
FGD
FGD
None
FGD
FGD
FGD
FGD
Carbon Feed
Method
None
Dry
Dry
Dry
Dry
None
Dry
Dry
Dry
None
Dry
Slurry
Slurry
None
Slurry
Slurry
Slurry
Slurry
Nominal
Carbon Feed Rate
kg/hr (Ib/hr)
None
4.5(10)
4.5(10)
27(60)
27(60)
None
23 (50)
11(25)
2.3(5)
None
23 (50)
23 (50)
11 (25)
None
23 (50)
23 (50)
23 (50)
23 (50)
Sample Analytes'
Hg, PM, %C
Metals, PM, %C
Hg, PM, %C
Hg, PM, %C
Hg, PM, %C
Hg, PM, %C
Metals, PM, %C
Hg, PM, %C
Hg, PM, %C
Metals, PM, %C, CDD/CDF, VOC
Metals, PM, %C, CDD/CDF, VOC
Metals, PM, %C, CDD/CDF
Metals, PM, %C
Hg, Cd, Pb, PM, %C, PSD"
Hg, Cd, Pb, PM, %C, PSD
Hg, Cd, Pb, PM, %C, PSD
Hg, Cd, Pb, PM, %C, PSD
Hg, Cd, Pb, PM, %C, PSD
• Metals: Ag, As, Ba, Be, Cd, Co, Cr, Cu, Hg, Mn, Mo, Ni, Pb, Sb, Se, 77, V
" PSD: Particle size distribution
-------
the coal-based carbon. The lower target
carbon feed rate was 4.5 kg/hr (10 Ib/hr),
which equates to a flue gas concentration
of approximately 60 mg of carbon per
dscm of flue gas1 and is approximately
equal to the high carbon feed rate for
tests conducted at OMSS. The higher tar-
get carbon feed rate of 27 kg/hr (60 Ib/hr)
is about 360 mg/dscm and was expected
to achieve Hg removals in excess of 90%.
During each run, simultaneous sampling
was conducted at the economizer exit and
in the stack for total PM and Hg using the
multiple metals sampling train. The ports
at the economizer outlet are located ap-
proximately 2 equivalent duct diameters
from the nearest upstream flow disturbance
and 1.6 diameters from the nearest down-
stream disturbance. The stack ports are
located near the 62-m (205-ft) elevation in
the 112-m (366-ft) stack. Each sampling
run was 1 hour long.2 In addition, an EPA
Method 5 type sampling train was oper-
ated at the economizer exit to collect a
daily composite sample of PM. The com-
posite sample was then used to deter-
mine the percent carbon in the fly ash
caused by incomplete combustion.
Both carbon types indicated similar lev-
els of Hg control during the I-B tests.
Thus, the cheaper Darco FGD was se-
lected as the carbon for use in the Phase
II tests.
Phase II-B Tests. The Phase I I-B para-
metric tests included eight test conditions
for evaluating the impact of carbon feed
rate, carbon feed method, and flue gas
temperature on Hg control (see Table 1).
During each run, simultaneous sampling
was conducted at the economizer exit and
in the stack for total PM and Hg using the
multiple metals sampling train. During six
of the test conditions (five from II-B and
one from I-B), the sampling fractions col-
lected by the multiple metals train were
analyzed for 16 other metals. These met-
als were antimony (Sb), arsenic (As), barium
(Ba), beryllium (Be), cadmium (Cd), chro-
mium (Cr), cobalt (Co), copper (Cu), lead
(Pb), manganese (Mn), molybdenum (Mo),
nickel (Ni), selenium (Se), silver (Ag), thal-
lium (Tl), and vanadium (V). In addition, a
Method 5 type sampling train was oper-
ated at the economizer exit to collect a
daily composite sample of PM for deter-
mination of percent carbon in the fly ash.
1 Based onafluegas flow rate of75,000 dscm/hr. Unless
otherwise noted, all flue gas flow rates used in this
paper are corrected to standard conditions [20°C (68°F),
101.3 kPa (14.7 psia)] and all concentrations to 7% O2
in dry gas.
2 All run durations in this paper are actual sampling times
and exclude times for port changes and resolving
equipment problems.
Except for the three test conditions dis-
cussed below, each sampling run was 1
hour long.
The testing also included sampling for
CDD/CDF emissions during Conditions
B10, B11, and B12, and for VOCs during
Conditions B10 and B11. Each sampling
run during these three test conditions was
2 hours in duration.
Phase II-A Tests. Phase II-A tests
were conducted to evaluate potential
detrimental impacts on ESP performance
due to carbon injection over an extended
time period, and to assess the relation-
ship between PM collection efficiency
and Hg control. To satisfy these objec-
tives, 5 days of sampling were conducted
over a 12-day period on Unit A. Follow-
ing an initial day of testing without car-
bon injection to establish baseline
performance, Darco FGD carbon was
added to the lime slurry feed tank and
continuously pumped into the SD.
As shown in Table 1, the first 4 days of
sampling were conducted with four ESP
fields in service. These tests were run on
the day prior to the start of carbon injec-
tion and on the first, third, and eighth days
after the start of carbon injection. After
completion of testing on the eighth day,
the fourth ESP field was turned off (the
fifth field was off for all tests), thus result-
ing in operation with only three fields. The
fifth day of sampling was conducted after
the unit had been operating with three
fields for 4 days. This sampling schedule
was designed to allow the ESP to reach
equilibrium operation with three fields.
These tests were conducted to evaluate
the effects of carbon injection on smaller
three-field ESPs.
During each run, simultaneous sampling
was conducted at the economizer exit and
in the stack for total PM, Hg, Cd, and Pb
using the multiple metals sampling train.
Also, during each run, two eight-stage
Andersen impactors were used to mea-
sure the particle size distribution (PSD) of
PM in the stack gas. The two PSD trains
were run throughout each test day to col-
lect sufficient PM for quantitative mea-
surement of the weight gain by each
impactor stage. In addition, a Method 5
type sampling train was operated at the
economizer exit to collect a daily com-
posite sample of PM for determination
of percent carbon in the fly ash. The PM
and PSD data provided a direct method of
evaluating degradations in ESP perfor-
mance that might be associated with car-
bon injection.
Carbon Feed Systems
Carbon was fed to Units A and B by
two different methods using three differ-
ent injection systems. The testing on Unit
B included injection of dry carbon and
addition of carbon into a slurry mix tank
installed just prior to the SD. Carbon was
injected into Unit A by addition of carbon
to lime slurry in the feed tank.
The dry injection system consisted of a
screw feeder and a pneumatic transport
system. The carbon injection probe con-
sisted of a 0.025-m (1-in.) pipe inserted in.
the flue gas duct upstream of the SD (see
Figure 1). The end of the probe was cut at
a 45° angle, which faced downstream.
Experiments conducted prior to the Phase
I tests suggested that the cyclone removed
little, if any, of the dry injected powdered
carbon.
Prior to the start of the test project, the
carbon feed system was calibrated. The
carbon feed rate was confirmed for each
test by recording the amount of each car-
bon addition and the time between each
pair of refillings.
At the end of each testing day, the
carbon feed rate was adjusted to the tar-
get level for the next day of testing. The
feeder was then operated overnight at this
feed rate to condition the SD/ESP prior to
the start of the next day of testing.
The second carbon feed system for Unit
B involved addition of carbon to the lime
slurry in a feed tank installed near the SD
inlet. Estimated carbon retention time in
the slurry with this system was about 10
minutes. Lime slurry was supplied to this
small feed tank from the plant's existing
slurry system. Carbon feed rates were
determined in the same manner as the
dry injection system, and the carbon was
carried in the lime slurry to the SD.
During the Phase II-A tests, carbon was
added to the plant's lime feed tank during
each slaking cycle. The average carbon
injection rates were calculated from the
amount of carbon added during each timed
slaking cycle. The lime/carbon slurry mix-
ture was injected into the reactor with the
existing slurry feed through the atomiza-
tion system. The carbon retention time in
the slurry during these tests is estimated
at 3 to 8 hours, with an average of ap-
proximately 5 hours.
Interpretation of Results
Mercury
A summary of key operating data and
Hg test results is presented in Table 2
-------
Table 2. Summary of Test Conditions and Mercury Test Results
Phase-
Condition
I-B1
I-B2
I-B3
I-B4
I-B5
II-B6
II-B7
II-B8
II-B9
II-B10
I-B11
II-B12
II-B13
H-A1
Run
1
2
3
AVG
4
5
6
AVG
7
8
9
AVG
10
11
12
AVG
13
14
15
AVG
10
11
12
AVG
13
14
15R
AVG
16
17
18
AVG
19
20
21
AVG
25
26
27
AVG
28
29
30
AVG
34
35
36
AVG
37
38
39
AVG
1
2
3
AVG
Carbon Type
None
FGD
PC-100
PC-100
FGD
None
FGD
FGD
FGD
None
FGD
FGD
FGD
None
Injection
Method
None
Dry
Dry
Dry
Dry
None
Dry
Dry
Dry
None
Dry
Slurry
Slurry
None
ESP Inlet
Temperature
°C(°F)
132 (269)
131 (267)
128 (262)
130 (266)
134 (274)
130 (266)
135 (275)
133 (272)
129 (264)
133 (272)
134 (273)
132 (270)
129 (265)
143 (290)
144 (291)
139 (282)
135 (275)
136 (277)
128 (262)
133 (271)
176(348)
177(350)
176(349)
176(349)
178(352)
178(352)
173(344)
176(349)
131 (267)
128 (263)
128 (262)
129 (264)
130 (266)
130 (266)
129 (265)
130 (266)
132 (269)
130 (266)
126 (258)
129 (264)
133 (271)
134 (273)
132 (269)
133 (271)
132 (269)
132 (269)
135 (275)
133 (271)
130 (266) '
128 (263)
129 (264)
129 (264)
136 (277)
132 (270)
134 (273)
134 (273)
Carbon
Injection
Concentration
mg/dscm
0
0
0
0
73
79
78
77
89
73
88
83
477
456
418
450
430
444
450
441
0
0
0
0
313
329
324
322
173
149
190
171
30
46
43
40
0
0
0
0
357
342
387
362
324
325
336
328
183
194
200
192
0
0
0
0
Total Carbon
at Cyclone
Inlet
mg/dscm
79
79
79
79
154
160
159
158
154
138
153
148
579
558
520
552
534
548
554
546
74
101
55
77
387
429
418
411
305
276
306
295
111
141
129
127
83
98
91
91
506
504
505
505
385
368
403
385
233
265
248
249
100
86
198
128
Mercury
Concentration
at Inlet
u.g/dscm
356
1363
711
810
972
593
835
800
593
639
586
606
491
440
512
481
680
820
644
715
365
249
349
321
964
506
778
749
545
455
525
508
485
957
463
635
663
433
384
493
626
635
664
642
299
521
300
373
382
377
974
578
268
430
610
436
Mercury
Concentration
at Outlet
ng/dscm
175
210
54
146
296
63
149
169
134
29
102
88
21
14
17
17
9
13
12
12
301
177
261
246
107
22
59
63
40
23
24
29
103
170
124
132
388
279
207
291
20
16
16
17
50
77
69
65
78
81
158
106
121
290
322
244
Mercury
Removal
Efficiency
%
50.8
84.6
92.4
75.9
69.5
89.4
82.2
80.4
77.4
95.5
82.6
85.2
95.7
96.8
96.6
96.4
98.6
98.4
98.2
98.4
17.5
28.9
25.2
23.9
88.9
95.6
92.4
92.3
92.7
95.0
95.4
94.4
78.8
82.2
73.2
78.1
41.5
35.6
46.1
41.0
96.8
97.4
97.7
97.3
83.2
85.3
77.0
81.8
79.7
78.5
83.8
80.7
54.9
32.6
47.2
44.9
(continued)
-------
Table 2. (Continued)
Phase-
Condition
II-A2
II-A3
II-A4
H-A5
Run Carbon Type
4 FGD
5
6
AVG
7 FGD
8
9
AVG
22 FGD
23
AVG
31 FGD
32
33
AVG
Injection
Method
Slurry
Slurry
Slurry
Slurry
ESP Inlet
Temperature
°C(°F)
129 (265)
129 (265)
130 (266)
129 (265)
137(278)
132 (269)
142 (288)
137(278)
141 (285)
139 (283)
140 (284)
139 (283)
139 (283)
140 (284)
139 (283)
Carbon
Injection
Concentration
mg/dscm
344
346
343
344
402
356
386
381
442
391
417
269
280
249
266
Total Carbon
at Cyclone
Inlet
mg/dscm
427
468
450
448
579
412
629
540
640
567
604
381
404
356
380
Mercury
Concentration
at Inlet
ug/dscm
302
403
1412
706
530
458
690
559
643
816
730
335
295
364
331
Mercury
Concentration
at Outlet
ug/dscm
55
78
261
131
43
108
156
102
49
90
70
40
51
52
48
Mercury
Removal
Efficiency
%
81.9
80.7
81.5
81.4
91.9
76.4
77.4
81.9
92.3
89.0
90.7
88.0
82.6
85.6
85.4
Note: Paniculate matter was not measured at the Inlet during Phase I; therefore, carbon concentrations were estimated using the average PM inlet
concentration for all Phase II-B tests. All concentrations are referenced to 7% O2 in dry gas at 20°C (68°F) and 101.3 kPa (14.7 psia). Test II-B7-15R
was conducted due to possible problems caused by interruption of carbon feed toward the end of Test 11-87-15.
The table includes data on carbon type,
injection rate and method, ESP inlet tem-
perature, total carbon concentration at the
cyclone inlet, inlet and stack Hg concen-
trations, and Hg removal efficiency.
Impact of Carbon Type. The influence
of carbon type was examined during Phase
I-B tests with dry injection of lignite-based
carbon (Conditions B2 and B5) and coal-
based carbon (Conditions B3 and B4).
Conditions B2 and B3 were conducted at
a low carbon feed rate corresponding to
approximately 80 mg/dscm of flue gas.
Conditions B4 and B5 were conducted at
a high carbon feed concentration of ap-
proximately 450 mg/dscm.
At the low carbon feed rate, the calcu-
lated removal efficiency was 70 to 89%
with the lignite-based carbon and 78 to
96% with the coal-based carbon. At the
high carbon feed rate, the removal effi-
ciency was 98 to 99% with the lignite-
based carbon and 95 to 97% with the
coal-based carbon. While the coal-based
carbon appeared to be marginally better
at the low feed rate, both carbons re-
sulted in similar performances at the high
injection rate. Because there was no clear
distinction in the removal efficiency of
these two carbons, the remaining tests
were conducted using the more eco-
nomical lignite-based carbon. The simi-
larity in performance of these two
carbons when injected as a dry powder
is consistent with the results of the
OMSS testing.
Impact of Carbon Injection Rate. Fig-
ures 2 and 3, respectively, present Hg
removal efficiency and stack gas Hg con-
centration as a function of carbon injec-
tion rate for all Unit B tests with lignrte-
and coal-based carbons. These figures
also include all tests for Unit A and Unit B
without carbon injection. The rectangular
data points show the removal efficiencies
and stack gas concentrations measured
during individual runs conducted at an ESP
temperature of approximately 132°C
(270°F) and while injecting either dry lig-
nite-based carbon or no carbon. The spe-
cific test conditions are A1, B1, B2, B5,
B8, B9, B10, and B11. Carbon injection
concentrations for these tests ranged from
40 to 450 mg/dscm.
These figures show that increased car-
bon injection concentrations increase Hg
removal and decrease stack gas Hg con-
centration. Increased injection concentra-
tions also decrease the variability in Hg
reduction and stack gas Hg concentration
between individual runs of the same test
condition. These tendencies were also
observed during the OMSS test program.
However, the carbon feed rates at Camden
County were significantly higher than at
OMSS, where the highest feed rate of dry
carbon corresponded to an injection con-
centration of about 70 mg/dscm.10'12 (Note
that the pulse-jet-cleaned FF for the OMSS
test unit had a cleaning cycle of about 12
minutes; thus the carbon in the filter cake
had a much longer average contact time
with "free stream" Hg than carbon col-
lected in an ESP.)
For a 132°C (270°F) ESP inlet tem-
perature and dry carbon injection concen-
trations above 150 mg/dscm, Hg removals
were 93% or greater and stack gas Hg
concentrations were less than 50 (ig/dscm
At higher injection rates, there were rela-
tively small incremental increases in Hg
reduction and decreases in stack gas Hg
concentration. At these feed rates, the
variability in Hg reduction between runs of
a given test condition was 3% or less. At
carbon feed rates of less than 150 mg/
dscm, the Hg removal efficiencies were
noticeably lower and the run-to-run vari-
ability between individual runs was as
much as 20% during a single test condi
tion.
The greatest variation in Hg reduction
and stack gas Hg concentration was ob-
served during tests without carbon injec-
tion, Conditions A1, B1, and B10. In
particular, during Runs 2 and 3 of Condi-
tion B1, removal efficiencies were 85 and
93%, nearly 40% higher than other runs
with no carbon injection. Stack gas Hg
concentrations during Runs 2 and 3 were
210 and 54 u,g/dscm, respectively. It was
initially believed that these high Hg re-
moval efficiencies reflected poor combus-
tion conditions caused by high waste
moisture content (it rained heavily for sev
era! days preceding the B1 tests) and in
the case of Run 2, high inlet Hg concen-
trations. However, similar "wet waste" con
drtions during the B10 tests did not result
-------
700-r
-^ 80
^ 70
1
I-
30
10
0 50 100 150 200 250 300 350 400 450 500
Injected Carbon (mg/dscm)
Dry Carbon @ 270F A Wet Carbon @ 270F * Dry Carbon @ 350F
Figure 2. Dependence of mercury reduction on carbon injection rate and method at 132°C (270°F)
and177°C(350°F).
400
I
200-
100
50
*.-.-
50 JOO 750 200 250 300 350 400
Injected Carbon (mg/dscm)
450 500
| • Dry Carbon @ 270F A Wet Carbon @ 270F * Dry Carbon @ 350F
Figure 3. Dependence of mercury concentration on carbon injection rate and method at 132°C
(270°F) and 177°C (350°F)
in abnormally high Hg captures. Review
of the three previous quarterly Hg emis
sion tests from Unit B shows reductions
during three-run tests of 41 to 43%, 41 to
55%, and 30 to 73% (all based on EPA
Method 101 A). These data suggest that
Hg removals without carbon injection at
Camden are typically between 30 and
55%, but removals can be either higher or
lower.
Impact of Inherent Carbon. Hg cap-
ture in SD/PM control systems is believed
to depend on the amount of inherent car-
bon contained in combustor fly ash. To
quantify the amount of unburned carbon
present in the flue gas, a composite PM
sample was collected at the economizer
exit during each test condition and ana-
lyzed for carbon content. The carbon lev-
els measured during each day were
between 1.1 and 2.2% of the dried sample
weight. The percent carbon found in each
daily sample was then multiplied by the
measured PM loading at the economizer
exit for each run on that day. The result-
ing estimate of inherent carbon was then
added to the concentration at which acti-
vated carbon was injected to estimate the
total carbon level in the flue gas (see
Figure 4). Note that this approach pro-
vides only a rough estimate of the total
entrained carbon. Specifically, this ap-
proach provides a single estimate of the
PM's carbon content for each day, and
any run-to-run variations in combustion
conditions that could result in increased
carbon levels during an individual run are
not measured. Also the amount of carbon
available for Hg absorption at location;
downstream of the SD's inlet cyclone is
overestimated. Visual inspection of cyclone
catches showed a substantial number of
large inherent carbon particles in the col-
lected fly ash.
As shown in Figure 4, the 30-55% re-
duction in emitted Hg at 132°C (270°F) in
the absence of carbon injection could be
explained by the presence of approxi-
mately 100 mg/dscm of unburned carbon
associated with the combustor fly ash.
The high levels of Hg capture obtained
during test conditions B11 and B5 can be
partially explained by the high levels of
inherent carbon during these two condi-
tions (2.2% during B11 and 1.9% during
B5). At OMSS, the carbon content of the
fly ash from the combustor (0.5-1.0%) was
approximately half the level at Camden
County, and the Hg reduction without car-
bon injection was also approximately half
the average Camden level (25%).
Impact of Carbon Injection Method.
The relationship between the carbon in-
jection method and Hg removal and stack
-------
700-
90
80
70
60
QC 50-
| 40
I 30
20
10
—sir
200 300 400
Total Carbon (mg/dscm)
500
600
Dry Carbon @ 270F A Wet Carbon @ 270F * Dry Carbon @ 350F
Figure 4. Mercury reduction as a function of total carbon concentration in flue gas and carbon injection
method at 132°C (270°F) and 177°C (350°F).
gas Hg concentrations is shown in Fig-
ures 2 and 3, respectively. At the medium
carbon injection concentration (149 to 200
mg/dscm), Hg removal efficiencies were
92 to 95% with dry carbon injection (Con-
dition B8) and 79 to 84% when the carbon
was injected with the lime slurry (Condi-
tion B13). For medium carbon injection
rates, dry injection resulted in stack gas
Hg concentrations ranging from 23 to 40
(ig/dscm, while slurry injection resulted in
stack Hg concentrations ranging from 78
to 158 u,g/dscm. At the carbon injection
concentrations >324 mg/dscm, Hg removal
efficiencies were 97 to 98% with dry injec-
tion (Condition B11) and 77 to 85% with
slurried carbon (Condition B12). For these
tests, the dry carbon injection test stack
gas Hg concentrations ranged from 16 to
20 u,g/dscm, while the slurry carbon injec-
tion test stack gas Hg concentrations
ranged from 50 to 77 u,g/dscm.
This observation is in contrast to the
OMSS results, which found that feed
method did not have a significant impact
on Hg emissions and Hg removal. The
cause of this difference is uncertain, but
may be due to the different carbon type
used or the type of PM control device.
The carbon used at OMSS during slurry
testing was coal-based, rather than lig-
nite-based as used during slurry testing at
Camden County. Also, the carbon col-
lected in the filter cake during the OMSS
test was retained on the bags for about
12 minutes and thus was probably dry
during most of its contact time with flue
gas, thus effectively functioning like dry
carbon relative to Hg capture.
In contrast, a very short period (on the
order of 10 seconds) was available for the
carbon to dry out in the ESP. Carbon
retention time in the slurry on Unit A was
estimated to be 3 to 8 hours, but only 8 to
10 minutes for Unit B. The carbon feed
concentration during each of these condi-
tions was approximately 360 mg/dscm.
The Hg removal efficiency for Units A and
B was very similar, with both units aver-
aging 82%. It can be concluded that the
decreased Hg adsorbency of carbon when
mixed with slurry occurred in less than 8
to 10 minutes.
Impact of ESP Temperature. Figure 5
shows the effects of ESP inlet tempera-
ture and total carbon concentration on Hg
removal efficiency and stack gas Hg con-
centration. When operating without car-
bon injection and an ESP inlet temperature
near 132°C (270°F) [Conditions B1 and
B10], Hg removals ranged from 36 to 92%
and stack gas Hg concentrations ranged
from 54 to 388 (ig/dscm. At the higher
ESP temperature of 177°C (350°F) [Con-
dition B6], the Hg removals were 18 to
29% and the stack gas concentrations
were 180 to 300 |ig/dscm. At high total
carbon concentrations (360 mg/dscm) and
132°C (270°F) ESP temperature (Condi-
tion B11), Hg removals were 97 to 98%.
At similar carbon feed rates, but an ESP
inlet temperature of 177°C (350°F) [Con-
dition B7], Hg removals were 89 to 96%
and stack gas Hg concentrations were 22
to 107 u.g/dscm. These data suggest that
the ability of carbon to absorb Hg is di-
rectly related to flue gas temperature, but
that, even at a relatively high ESP inlet
temperature of 177°C (350°F), activated
carbon injection results in high Hg reduc-
tions.
Impact of PM Control Efficiency. The
PM control efficiency of the ESP aver-
aged 99.8% or more for all test condi-
tions. The PM concentrations in the stack
ranged from 1.1 to 8.9 mg/dscm (0.0005
to 0.004 gr/dscf). These levels of PM con
trol and emission did not exhibit any ap-
parent relationship between PM control
and Hg removal efficiency for the tests
conducted at the Camden facility.
Multlvariate Regression
Analysis of Hg Control
A stepwise multivariate regression analy
sis was used to assess the statistical sig-
nificance of individual process variables
and to develop predictive equations for
Hg removal efficiency and stack gas Hg
concentration. These analyses identified
three statistically significant process vari-
ables influencing Hg control efficiency: car-
bon feed rate, ESP inlet temperature, and
carbon injection method. The analyses also
identified four statistically significant pro-
cess variables affecting stack gas Hg con
centration: carbon feed rate, carbon
injection method, ESP inlet temperature
and inlet Hg concentration.
Figure 6 shows the predicted stack gas
Hg concentrations for the injected carbon
concentrations for the best regression
model based on an ESP operating tern
perature of 132°C (270°F) and inlet Hg
concentrations of 200, 500, 800, and 1,100
|ig/dscm. Note that most of the reduction
in stack Hg concentration occurs at in
jected carbon concentrations below about
100 mg/dscm. At injected carbon concen
trations above this level, the stack Hg
concentration decreases, but the reduc
tion is much more gradual. Considering
the degree of control by inherent fly ash
carbon, the variations in the inlet flue gas
Hg concentration, and the variation in re
duction efficiency of the carbon injection
process, complying with an Hg emission
limit of 100 u,m/dscm at the Camden f acil
ity would probably require injected carbon
concentrations in the range of 150 to 200
mg/dscm.
The absence of inlet Hg concentration
as a statistically significant variable for
predicting Hg removal efficiency is in con
trast to the OMSS data and is believed tc
reflect the difference in control capability
-------
Mercury Reduction (%)
100
90
80
70
60
60
40 ~
30 ~
20
10 -
n
-•*!"•
•m
—
J
•
* m
*
100
200 300 400
Total Carbon (mg/dscm)
500
600
Dry Carbon @ 270F * Dry Carbon <§> 350F |
Figun 5. Mercury reduction as a function of total carbon concentration at 132°C (270°F) and 177°C
(350°F).
450
400
Inlet Mercury Concentration
1100\ig/dscm
800 \i.g/dscm
500 \jig/dscm
. 200\ig/dscm
100
200 300
Injected Carbon (mg/dscm)
400
500
Figure 6. Regression lines for stack mercury concentration as a function of inlet mercury
concentration and dry carbon injection rate at 132°C (270°F).
of systems equipped with a FF versus an
ESP. With a FF, carbon will adsorb Hg
both while entrained in the flue gas and
after it is collected in the filter cake. When
inlet Hg levels vary (e.g., due to a short-
duration spike in Hg concentration), the
unsaturated carbon on the filter cake is
able to adsorb additional Hg and to mod-
erate the spike at the FF outlet. In this
situation, the efficiency of the control sys
tem (i.e., entrained carbon and filter cake)
can increase for short periods. During
these periods the outlet concentration is
not strongly dependent on inlet concen
tration until there is "breakthrough" of the
filter cake carbon. The ability of the filter
cake to buffer outlet Hg spikes in inlet Hg
levels is similar to the ability of the excess
sorbent in the filter cake to moderate fluc-
tuations in inlet acid gas levels. With an
ESP, most of the Hg reduction occurs
while the carbon is entrained in the flue
gas and is controlled by the likelihood of
contact between carbon particles and Hg
prior to the collection of carbon with the
fly ash on the ESP plates. Once a carbon
particle is collected on an ESP plate, the
potential for contact with Hg is greatly
reduced, because of both the removal of
carbon from flue gas and the limited op-
portunity for Hg in flue gas to contact
carbon collected on plates of the ESP. In
this case the Hg capture is limited by the
total concentration of suspended "free
stream" carbon, and the outlet Hg con
centrations for a given carbon loading are
highly dependent on inlet concentrations.
Other Metals
Flue gas concentrations of the 16 other
metals in addition to Hg were determined
during six test conditions (see Table 1)
Five of these test conditions were con
ducted at 132°C (270°F): no carbon injec-
tion (B10), dry carbon injection at a low
and a high feed rate (B2 and 611, respec
tively), and slurry injection of carbon at a
medium and a high feed rate (B13 and
B12, respectively). The sixth test condi
tion was conducted at 177°C (350°F) with
dry carbon injection (B7).
For Cd, Pb, As, Ba, and Cu, metals
removal efficiencies exceeded 99% dur
ing all test conditions. For Cr and Mn
removal efficiencies exceeded 99% ex
cept during the high temperature run (B7)
and for Mn during the medium feed rate
carbon-in-slurry test condition (B13). For
Mo and Ni, removal efficiencies showed
significant variability, ranging from a low
of 72% for Mo during the high tempera-
ture test condition up to 98%. Removal
efficiencies for Sb, Be, Co, and V could
not be precisely determined because their
concentrations in the stack were below
-------
their analytical detection limits. Removal
efficiencies for Ag and Tl could not be
estimated because concentrations of these
metals were below their analytical detec-
tion limits at both the inlet and outlet sam-
pling locations. Considering the poor
recovery of matrix spikes used for analyti-
cal quality assurance, the data for Se were
unacceptable and are not reported here.
These data indicate that the 13 detected
metals, with the possible exception of Mo,
are emitted from the combustor primarily
as PM and that emissions of these metals
are controlled predominantly by the PM
control device. There also appears to be
an effect of ESP temperature on the con-
trol of Cr, Mn, and Ni but, given the small
size of the data set, this relationship may
be due to random chance. Injection of
activated carbon did not have a quantifi-
able impact on emissions of any of the 13
detected metals, which were generally re-
moved at high levels.
Organic Compounds
Economizer outlet and stack gas con-
centrations of CDD/CDF were measured
during Conditions B10 (no carbon injec-
tion), B11 (dry carbon at 132°C [270°F]),
and B12 (slurry carbon at 132°C [270°F]).
During Condition B10 without carbon in-
jection, the total CDD/CDF removal effi-
ciency across the SD/ESP was 78 to 80%.
During Condition B11 with a high injection
concentration of dry carbon (360 mg/
dscm), the removal efficiency was 95 to
98%. During Condition B12 with a high
injection concentration of carbon in slur-
ried lime (330 mg/dscm), the removal effi-
ciency was 96 to 97%. These data suggest
that, unlike Hg, the CDD/CDF removal
efficiency of dry and slurried carbon injec-
tion was similar.
As shown in Figure 7, stack gas con-
centration of CDD/CDF was reduced from
40 to 60 ng/dscm without carbon injection
to less than 7 ng/dscm for dry carbon
injection and less than 13 ng/dscm for
slurry injection. The higher stack CDD/
CDF levels during slurry injection of car-
bon reflect the higher concentration of
CDD/CDF measured at the economizer
outlet during two of the Condition B12
runs of approximately 375 ng/dscm, com-
pared to 130 to 220 ng/dscm for the other
seven runs. The carbon injection concen-
tration into slurry also averaged nearly
10% below that for dry carbon injection.
The reduction of CDD/CDF emissions by
carbon injection is consistent with Euro-
pean field test results.
Sampling for VOCs was conducted dur-
ing Conditions B10 (no carbon injection)
and B11 (dry carbon injection). There ap-
CDD/CDF Concentration (ng/dscm)
ou
50
40
30
20
10
0
' '
*
1
1
1 1
No Carbon Dry Carbon Wet Carbon
(II-B10) (II-B11) (II-B12)
Figure 7. Stack CDD/CDF concentration with and without carbon injection into flue gas at 132°C
(270°F).
peared to be a reduction in the level
of some compounds (carbon disulfide,
benzene, chlorobenzene) across the
SD/ESP and an increase in others
(trichloro-fluoromethane, methylene
chloride, toluene). Of significance to this
study, there was no apparent impact of
carbon injection on reduction of any of
these compounds in the SD/ESP system.
Acid Gases
As noted in Description of Facility,
above, various gaseous components in
the flue gas were monitored during each
test condition. Emissions of SO2, HCI, and
NOx were monitored using the plant CEM
system.
The SCX data displayed a general in-
crease in SO2 removal with increasing car-
bon injection rate. However, the size of
the data set and the scatter in the data
are such that this apparent relationship
may be due to random chance. No rela-
tionship between carbon feed rate and
HCI or NOX emissions was apparent.
Impact of Carbon Injection on
ESP Performance
ESP performance test results from Unit
A were evaluated using average PM, Cd,
and Pb removal efficiencies and the per-
cent of total PM less than 2 u.m during
each test condition. There was no consis-
tent change in any of these parameters
during the first four test conditions, indi-
cating that carbon injection did not alter
ESP performance. During Condition A5,
with the fourth ESP field out of service,
there was no apparent change in PM re-
moval efficiency. However, the removal
efficiency for Cd and Pb decreased, and
the percent of emitted PM less than 2 jim
increased. These changes are consistent
with the expected enrichment of volatile
metals onto fine paniculate and the re-
duced ability of the ESP to collect fine
particulate when the fourth ESP field was
out of service. Stack opacity, ESP volt-
age, and ESP current remained within nor-
mal operating ranges during the entire
Unit A test period.
Conclusions
The collected data and results calcu-
lated for the Camden County tests led to
several conclusions:
Hg reductions in excess of 90%
were achieved by injection of dry
carbon at both of the ESP operat-
ing temperatures examined (132°C
[270°F]and 177°C [350°F]).
The most important process vari-
ables affecting Hg emissions were
carbon feed rate, carbon injection
method, and ESP operating tem-
perature.
• The amount of unburned carbon
present in fly ash played a signifi-
cant beneficial role in controlling
Hg emissions.
The two carbons tested (Darco FGD
and Darco PC-100) were similarly
effective in controlling Hg emissions
when injected as a dry powder. (No
slurry tests were conducted with
the coal-based PC-100.)
• Injecting carbon (the lignite-based
Darco FGD) with lime slurry was
10
-------
less effective in reducing Hg emis-
sions than dry injection. This con-
trasts with the results of the OMSS
tests and may be due to the perfor-
mance characteristics of an ESP
versus a FF or to differences in
carbon properties (Darco PC-100
was used in the OMSS tests).
Multivariate regression analyses
identified three statistically signifi-
cant process variables influencing
Hg control efficiency: carbon feed
concentration, ESP inlet tempera-
ture, and carbon injection method.
The four significant process vari-
ables which influence stack gas Hg
concentrations are: carbon feed
concentration, ESP inlet tempera-
ture, carbon feed method, and inlet
flue gas Hg concentration.
Injecting carbon reduced stack
emissions of CDD/CDF by over
75% so that CDD/CDF removals
increased to 95% or more. How-
ever, there was no apparent effect
of carbon injection on emissions of
VOCs.
Emissions of metals other than Hg
were primarily associated with PM,
and their control was determined
mainly by the efficiency of PM re-
moval. A possible exception to this
relationship was Mo. Carbon injec-
tion had no apparent benefit on the
emission control of these metals.
Carbon injection had no discernible
impact on the ESP efficiency in con-
trolling PM.
11
&U.S. GOVERNMENT PRINTING OFFICE: 1994 - 550-0«7/80137
-------
D. White, W. Kelly, M. Stocky, J. Swift, and M. Palazzolo are with Radian Corp.,
Research Triangle Park, NC 27709.
James D. Kilgore is the EPA Project Officer (see below).
The complete report, entitled "Emission Test Report, Field Test of Carbon Injection
for Mercury Control, Camden County Municipal Waste Combustor," (Order No.
PB94-101540; Cost: $27.00; 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:
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
Official Business
Penalty for Private Use
$300
BULK RATE
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
EPA/600/SR-93/181
------- |