Preliminary Performance and Cost Estimates of Mercury
   Emission Control Options for Electric Utility Boilers
         Ravi K. Srivastava, Charles B. Sedman, and James D. Kilgroe
                  U.S. Environmental Protection Agency
              National Risk Management Research Laboratory
               Air Pollution Prevention and Control Division
                    Research Triangle Park, NC 27711
                       Prepared for presentation at

               A&WMA 2000 Annual Conference & Exhibition
                          Salt Lake City, Utah
                           June 18-22,2000

-------
Preliminary Performance and Cost Estimates of Mercury

Emission Control Options for Electric Utility Boilers

Ravi K. Srivastava, Charles B. Sedman, and James D. Kilgroe
U.S. Environmental Protection Agency
National Risk Management Research Laboratory
Air Pollution Prevention and Control Division
Research Triangle Park, NC 27711


ABSTRACT

Under the Clean Air Act Amendments of 1990, the Environmental Protection Agency (EPA) has
to determine whether mercury emissions from coal-fired power plants should be regulated. To
aid in this determination, preliminary estimates of the performance and cost of activated-carbon-
based mercury control technologies have been developed. This paper presents the methodology
used in arriving at these estimates, estimated capital and annual operation and maintenance
(O&M) costs, and results of sensitivity analyses conducted on these cost estimates.

Results reveal that for the majority of applications capital costs of activated carbon injection
(ACI) mercury control technologies may be comparable to those associated with low nitrogen
oxide (NOX) burners (LNBs), Also, total annual costs may be comparable to, or in some cases
slightly higher than, corresponding costs for LNBs. Moreover, ACI requirements appear to
dominate the total annual costs of ACI-based technologies.

The performance and cost estimates of the ACI mercury control technologies presented in this
paper are based on relatively few data points from pilot-scale tests and, therefore, are considered
to be preliminary. Factors that are known to affect the adsorption of mercury on activated carbon
include speciation of mercury in flue gas, available residence time (duct length), effect of flue gas
and ash characteristics, and degree of mixing between flue gas and activated carbon. The effect
of these factors may not be entirely accounted for in the relatively few pilot-scale data points that
comprised the basis for this work. Ongoing research is expected to address these issues.

INTRODUCTION

Since mercury is an element, it cannot be created or destroyed. In the atmosphere, mercury exists
in two forms; elemental mercury vapor (Hg°) and ionic mercury (Hg**). Hg can circulate in the
atmosphere for up to one year and, consequently, can undergo dispersion over regional and
global scales. Hg"1"1" in the atmosphere either is bound to airborne particles or exists in gaseous
form. This form of mercury is readily removed from the atmosphere by wet and dry deposition.
After deposition mercury is commonly emitted back to the atmosphere as either a gas or a
constituent of particles and redeposited elsewhere. In this fashion,, mercury cycles in the
environment.

-------
A number of human health and environmental impacts appear to be associated with exposure to
mercury. Mercury is known to bioaccumulate in fish and animal tissue in its most toxic form,
methylmercury. Human exposure to methylmercury has been associated with serious
neurological and developmental effects. Adults exposed to methylmercury show symptoms of
tremors, loss of coordination, and memory and sensory difficulties. Offspring exposed during
pregnancy show atrophy of the brain with delayed mental development. The incidence and extent
of such effects depend on the level of exposure to methylmercury. Hg°is readily absorbed
through lungs and, being fat-soluble, is rapidly distributed throughout the body. Subsequently, it
slowly oxidizes to Hg"1"1", which accumulates in the brain and can lead to tremors, memory
disturbances, sensory loss, and personality changes. Hg** is absorbed through the digestive tract,
accumulates in the kidneys, and can lead to immune-mediated kidney toxicity. Adverse effects of
mercury on fish, birds, and mammals include reduced reproductive success, impaired growth,
behavioral abnormalities, and even death. Details of the risks associated with exposure to
mercury are discussed in the literature.1

The Clean Air Act Amendments of 1990 (CAAA) require EPA to examine the potential risks
associated with mercury emissions from all air pollution sources in the U.S. Under this
requirement, EPA's Office of Air Quality Planning and Standards developed a Mercury Study
Report to Congress.1 The report concluded that in 1994-95 U.S. coal-fired electric utility boilers
emitted 51.7 tons (46,944 kg) of mercury and comprised the largest anthropogenic source of
mercury emissions in the U.S. Under the CAAA, EPA must make a determination regarding the
regulation of mercury emissions from these sources. As an aid for this determination, EPA's
Office of Air and Radiation has conducted preliminary analyses examining potential pollution
control options for the electric power industry to lower the emissions of its most significant air
pollutants, including mercury,2 These analyses were conducted using the Integrated Planning
Model (IPM)3, which had to be supplemented with information on performance and cost of
mercury emission control technologies.

The development of preliminary estimates of performance and cost of mercury emission control
technologies for electric utility boilers is described in this paper. These estimates were used in
conducting  the above-mentioned analyses. The paper also presents analyses that evaluate the
impacts of important cost components.

MERCURY SPECIATION AND CAPTURE

Mercury is volatilized and converted to Hg° in the high temperature regions of combustion
devices. As the flue gas cools, Hg° is oxidized to Hg"1"1". The rate of oxidization is dependent on
the temperature, flue gas composition, properties, and amount of fly ash and any entrained
sorbents. In coal-fired combustors, where the concentrations of hydrogen chloride (HC1) are low,
and where equilibrium conditions are not achieved, Hg° may be oxidized to mercuric oxide
(HgO), mercuric sulfate (HgSO4), mercuric chloride (HgGb), or some other mercury compound.4
The oxidization of Hg° to HgCla and to other ionic forms of mercury is abetted by catalytic
reactions on the surface of fly ash or sorbents that may be present in the flue gas.

-------
Hg°, HgCla, and HgO are primarily in the vapor phase at flue gas cleaning temperatures.
Therefore, each of these forms of mercury can potentially be adsorbed onto porous solids such as
fly ash, powdered activated carbon (AC), and calcium-based acid gas sorbents for subsequent
collection in a paniculate matter (PM) control device. These mercury forms may also be captured
in carbon bed filters or other reactors containing appropriate sorbents.
Some mercury removal with wet scrubbers also appears to be possible. HgCla is water-soluble
and reacts readily with alkali metal oxides in an acid-base reaction; therefore, conventional acid
gas scrubbers used for sulfur dioxide (SOa) control can also effectively capture HgCli. The total
mercury removal efficiency of wet scrubbers has been reported to range from 30 to 90%.5
However, Hg° is insoluble in water and must be adsorbed onto a sorbent, or converted to a
soluble form of mercury that can be collected by wet scrubbing. HgO has low solubility and
probably has to be collected by methods similar to those used for Hg°.

MERCURY EMISSION CONTROL TECHNOLOGIES

A number of air pollution control technologies are now used commercially to control mercury
emissions from municipal waste combustors (MWCs). In Europe these technologies include the
use of ACI followed by collection in a PM control device, the use of wet scrubbers, the use of
carbon filter beds, and the use of selenium filters. In the U.S., ACI is being used to control
mercury emissions from MWCs.1  "

At present, the control of mercury emissions from coal-fired boilers is not commercially
practiced in the U.S. The direct application of MWC mercury control technologies to coal-fired
boilers is difficult because of factors that make mercury more difficult, and expensive, to capture
in boiler applications. Typically, flue gas from coal-fired boilers contains a higher fraction of Hg°
compared to flue gas from MWCs. This requires that more effective sorbents must be developed
for Hg° capture or that Hg° must be converted to more a easily captured form such as HgCli.
Moreover, the concentration of mercury in coal-fired utility boiler flue gas is typically less that
10 p,g/dscm, while the concentration of mercury in MWCs may range from less than 200 to more
than 1500 jig/dscm. The combination of low mercury concentrations and large flue gas volumes
increases the difficulty and cost of controlling mercury emissions from coal-fired utility boilers.4

Research and development (R&D) efforts examining control of mercury emissions from coal-
fired boilers are currently underway. Some of these efforts are investigating the factors that affect
mercury speciation in flue gas. Development of effective sorbents and reagents for mercury
capture is also being pursued. This development can result in lower sorbent or reagent unit costs
($/kg), increased sorbent collection efficiency or reagent reactivity, increased sorbent or reagent
utilization rates, and cost-effective multipollutant control. Finally, additional R&D efforts are
aimed at evaluating and developing improved control technologies based on more effective
equipment or process conditions. Examples include the use of additional ducting upstream of the
PM control device to improve sorbent utilization through increased residence time, recycling of
collected fly ash and sorbent to increase sorbent utilization, and equipment modifications made
to inject a reagent into the flue gas to increase conversion of Hg° to HgCla for collection in wet
scrubbers. It is believed that these R&D efforts will develop improved technologies for

-------
controlling mercury emissions from coal- fired boilers. While substantial progress is expected
over the next several years, it is not possible at this time to quantify the improvements in
performance and costs that will accrue from these efforts.

Control Technology Selections

Based on published literature,1'5"9 control technologies using injection of AC into the flue gas
appear to hold promise for reducing mercury emissions from utility boilers. This determination
considered: (1) ACI-based technologies have been applied successfully on MWCs; (2) results
achieved in pilot-scale ACI tests, conducted on coal-fired utility boilers, are available in
published literature; and (3) limited cost data are available for applications of ACI-based
technologies on utility boilers. Accordingly, this work focussed on control of mercury emissions
using ACI.

Approximately 75% of the existing coal-fired utility boilers in the U.S. are equipped only with
electrostatic precipitators (ESPs) for the control of PM.4 The remaining boilers employ fabric
filters (FFs), paniculate scrubbers, or other equipment for control of PM. While developing the
ACI-based technology applications, these PM collection choices were taken into account. Model
plants with eight different flue gas cleaning equipment configurations and firing either
bituminous or subbituminous coal were used in this work. The resulting 16 ACI-based
technology application cases, reflecting differences in flue gas cleaning equipment and type of
coal burned, are shown in Table 1. In general, the operating costs associated with use of AC
comprise a significant portion of total costs for ACI-based technologies. Since spray cooling
(SC) of flue gas results in significantly reduced requirements of AC , this cooling was included
in each of the technologies shown in Table 1. In developing the ACI-based technology
application cases, the available information described below was also taken into consideration.

Pilot-scale test data indicate that mercury emission control on units equipped only with ESPs
(cold or hot) may be possible using SC followed by:  (1) ACI upstream of the existing cold ESP,
or (2) ACI upstream of a new (relatively small) FF installed downstream of the existing hot ESP.
As such, these control options were used for units with ESPs.

AC in the FF cake provides better gas-particle contact than an ESP and results in better sorbent
utilization. This has been validated in full-scale tests on MWCs and pilot-scale tests on coal-fired
combustors. Field tests have shown that it takes two to three times more AC to achieve the same
performance on MWCs equipped with dry scrubbers and ESPs than with dry scrubbers and
FFs.10

It is anticipated that current research on wet scrubbers will result in improved performance
through the use of reagents or catalysts to convert mercury to chemical compounds that are
soluble in aqueous-based scrubbers. However, since no published pilot-scale test data confirm
this hypothesis, this study assumes that ACI is needed for facilities with existing wet scrubbers.

Performance Estimates

-------
Two performance parameters were estimated for each of the mercury control technologies shown
in Table 1. These parameters were: (1) the mercury removal efficiency across the air pollution
control system, and (2) the ratio of the required carbon injection rate (g/s) to mercury flow rate
(g/s) in the flue gas at the inlet to the first air pollution control device, expressed as C/Hg ratio.
As discussed below and summarized in Table 1, the estimates of mercury emission reduction
performance and associated C/Hg ratio are based on reported data from pilot-scale tests.6"9

-------
Table 1. Mercury emission control technologies for utility boilers.
Case

1A
IB

2A
2B

3A
3B

4A
4B

5A
5B

6A
6B

7A
7B

8A
8B
Existing
Equipment

Cold ESP
Cold ESP

Hot ESP
Hot ESP

Cold ESP + FGD
Cold ESP + FGD

HotESP + FGD
Hot ESP + FGD

FF
FF

FF + wet FGD
FF + wetFGD

DS + FF
DS + FF

DS + ESP
DS + ESP
Coal Type

bituminous
subbituminous

bituminous
subbituminous

bituminous
subbituminous

bituminous
subbituminous

bituminous
subbituminous

bituminous
subbituminous

bituminous
subbituminous

bituminous
subbituminous
Mercury Emission
Control Technology

SC + ACI
SC + ACI

SC + ACI + FF
SC + ACI + FF

SC + ACI
SC + ACI

SC + ACI + FF
SC + ACI + FF

SC + ACI
SC + ACI

SC + ACI
SC + ACI

ACI
ACI

ACI
ACI
Reduction
(%)

80
'65

85
85

90
90

90
90

85
85

90
90

85
85

85
85
C/Hg
(g carbon /gHg)

15,000
7,500

10,000
6,000

15,000
7,500

10,000
6,000

10,000
6,000

10,000
6,000

6,000
3,000

10,000
6,000
Note that, in the table above, FGD refers to wet scrubbing and DS refers to dry scrubbing (spray
drying). In arriving at the contents of Table 1, six assumptions were made with respect to
mercury control system hardware and performance. These assumptions are described below.

1. The performance estimates presented in Table 1 reflect the mercury emission reduction
   percent calculated using
    Reduction(%) = lOOx
                        (Emissionin - Emission
                                             slack
                                Emission,.
(1)
   where:  "

    Emission^ = flue gas mercury concentration at the inlet to the first air pollution control
   device; and
    Emissionstack = flue gas mercury concentration at the stack.

-------
2. It is known that the type of coal fired in the boiler can influence the performance of mercury
   emission reduction technologies. Therefore, as shown in Table 1, separate estimates of
   mercury emission reduction performance and carbon injection rate were determined for
   bituminous and subbituminous coals.

3. The capital costs associated with SC equipment are independent of the degree of cooling
   needed, within the range of referenced applications. Also SC is assumed to exist at boiler
   sites with DSs and, therefore, additional costs for SC are not needed in Cases 7A, 7B, 8A,
   and 8B.

4, The performance estimates for boiler sites with cold ESPs (Cases 1A and IB), FFs (Cases
   5A and 5B), DSs + FFs (Cases 7A and 7B), and DSs + ESPs (Cases 8A and 8B) were
   determined from published data6"9 as described below.

   Waugh et al.6 describe ACI tests on  a low-sulfur (<1%), high-chlorine (0.1%), Eastern
   bituminous-coal-fired boiler served by parallel pilot-scale ESP and FF units. In these tests:
   (1) with an ESP, 80% mercury control was achieved at a C/Hg ratio of about 55,000:1 and (2)
   with a FF, 90% mercury control was achieved at a C/Hg ratio of about 45,000:1. The high
   chlorine content of the coal was thought to be responsible for the poor carbon utilization.
   Waugh et al.7 describe subsequent tests in which lime was added with ACI upstream of the
   FF, which resulted in improved capture of Hg with ACI. In these tests using a FF, 90%
   mercury control was achieved at a C/Hg ratio of about 12,000:1. Interpolation of the results
   in Waugh et al.7 suggests that 85% mercury reduction may be achieved with a C/Hg ratio of
   about 10,000:1 using lime. Therefore, for this work a C/Hg ratio of 10,000:1 was selected for
   Case 5A, as shown in Table 1. Since no comparable data are available for Case 1A, the 73%
   improvement in C/Hg ratio in FF tests (from a C/Hg of 45,000:1 needed without lime
   addition to a C/Hg of 12,000:1 needed with lime addition), is assumed to hold for ESPs. This
   suggests that ESPs (Case 1A) can achieve 80% Hg control at a C/Hg ratio of about 15,000:1,
   using hydrated lime or with no HC1 present in the flue gas.

   It is observed that the supplemental use of hydrated lime with ACI to enhance mercury
   capture will not negatively impact the estimated costs of control. Capital costs associated
   with lime injection would be covered in the sorbent storage and injection system costs, and
   lime is relatively cheap compared to the AC that it may  displace (hydrated lime costs about
   $85/ton compared with $909/ton for the AC used in this study). In addition, the use of
   hydrated lime would generate additional SOa emission credits. Considering these factors, the
   cost associated with use of lime to assist in mercury control was not included in this work

   Haythornthwaite et al.8 describe similar pilot-scale ACI testing with an ESP or a FF on a low-
   sulfur, subbituminous coal. In the  tests with an ESP, up  to about 75% mercury removal was
   achieved at a C/Hg ratio about 8,000:1. Further, the data from these tests suggest that an
   average of 65% mercury removal may be achieved at a C/Hg ratio of about 7500:1 in ESPs
   with gas cooling (Case IB). Note that the combination of lower performance requirements
   (65% collection efficiency) and lower C/Hg ratio (7,500:1) in Case IB, compared to Case
   1A, is consistent with potentially lower mercury flue gas inlet concentrations associated with

-------
   the combustion of subbituminous coals. Again in the tests with a FF, 86-90% mercury was
   removed at a C/Hg ratio of about 6,000:1, Therefore, this work assumed that an average of
   85% mercury can be removed at a C/Hg ratio of 6,000:1 with gas cooling (Case 5B). The
   improved carbon utilization achieved in subbituminous coal tests over the bituminous coal
   results provided above is thought to be directly related to high fly ash alkalinity and the low
   chlorine content of subbituminous coal.

   Redinger et al.9 describe tests using wet and dry scrubbers on flue gas resulting from firing of
   midwestern coal. The dominant form of mercury in this flue gas is known to be HgCli. In
   these tests mercury removals of 75-85% were observed. No experience is currently known
   where AC has been added upstream of a spray dryer receiving coal-fired flue gas. Therefore,
   performance estimates for Cases 7 and 8 were determined by using the performance reported
   in Reference 10 and the performance estimates of Cases 5 and 1, respectively. For the cases
   with FF, it was assumed that absorption of mercury by spray dryer would result in a 40 to
   50% reduction in AC consumption for a specific level of mercury control (compare Cases  5
   and 7). Similarly, for the cases with ESPs, it was assumed that use of spray dryers would
   result in 20 to 33% reduction in AC consumption (compare Cases 1 and 8),

5. The performance and carbon requirements associated with Cases 2A and 2B are assumed to
   be identical to Cases 5A and 5B, respectively. This assumption is made because a FF is the
   PM control device in each of these cases. However, whereas in Cases 5A and 5B FFs exist at
   boiler sites, in Cases 2A and 2B small FFs are considered to be needed to collect AC.

6. In the EPA Mercury Study Report to Congress1, the required mercury control efficiency for
   each retrofit option was assumed to be 90%. In this study 90% control was assumed to be
   required only for those units whose existing equipment configuration includes wet scrubbers.
   Consequently, if wet FGD is available at a boiler site in addition to an ACI-based technology,
   then 90% mercury reduction is assumed to be available without any change in cost over the
   corresponding case without wet FGD. Thus, for example, costs for Cases 1A and 3A of Table
   1 would be identical but 90% reduction is available in Case 3A in contrast to 80% in Case
   1A. Thus costs for Cases 3A, 3B, 4A, 4B, 6A, and 6B are identical to Cases 1A, IB, 2A, 2B,
   5A, and 5B, respectively, but 90% mercury removal is available in cases with wet FGD.

Based on the C/Hg ratios shown in Table 1, estimates of costs associated with the use of each of
the ACI-based technologies were developed. Described in the ensuing paragraphs is the
methodology used in arriving at these estimates.

Capital Cost Estimates

ACI systems, FFs, and SC are used in the ACI-based technologies shown in Table  1. Some cost
information for hardware items used in ACI-based technology applications on model boilers of
sizes 100 and 975 MW is available in Reference 1 and Brown et al.11 This cost information is
shown in Table 2.

-------
Table 2. Cost (1993 U.S. $) of major equipment used in ACI-based technologies.
Source

Reference 1
Brown et al.11
FF
100 MW
1,813,479
Same as
above
975 MW
12,978,750
Same as
above
SC
100 MW
258,627
Same as
above
975 MW
2,993,796
Same as '
above
ACI System
100 MW
109,448
257,000
975 MW
109,448
2,231,000
Note that, as discussed in Appendix B of Volume VIH of the Mercury Study Report to Congress1,
purchased equipment costs for SC and ACI systems were based on vendor contacts reported in
1993 and FF costs were based on 1992 data. Accordingly, EPA's costs for SC and FF shown in
Table 2 are considered to be in 1993 dollars. The ACI system costs reported by Brown et al.,11
and shown in Table 2, are in 1989 dollars. For this work, these costs were escalated to 1993
dollars using the inflation factor of 1.144003567, derived from the Economic Report of the
                                                   1 *?
President, Council of Economic Advisers, February 1998.

Brown et al.11 agree with the cost estimates for FF and SC provided in Reference 1 but have
conducted more detailed cost estimates for ACI systems. Accordingly, costs for FFs and SCs
reported in Reference 1 and costs for ACI systems reported in Brown et al.11 were used to
develop the capital cost functions fof each of the 16 ACI-based technologies shown in Table 1.
Note, however, that Brown et al. used a C/Hg ratio of 30,000;I.11 Therefore, the ACI system
costs of Brown et al.11 were adjusted to correspond to the injection rates shown in Table 1.

The procedure used in calculating the capital cost in $/kW for an ACI-based technology
application  on a boiler of size PI is shown in Table 3. The values of the installation cost factor, x,
and the indirect cost factor, y, used in this work are identical to those used in the Mercury Study
Report to Congress.1 The values of x used are 0.34 for SC equipment, 0.15 for ACI systems, and
0.72 for FFs, The values of y used are 0.45 for SC equipment, 0.30 for ACI systems, and 0.45 for
FFs. As shown in Table 3, a retrofit factor of 1.15 was included in the cost estimation procedure
to account for more difficult retrofits.

  Table 3. Capital cost calculation procedure.
Description
Boiler size (MW)
Purchased equipment cost ($)
Installation cost ($)
Indirect cost ($)
Total capital cost ($)
Retrofit factor
Total capital cost w/ retrofit ($)
Capital cost ($/kW)
Cost or Data Symbol
Pi
PE
Inst = .x*PE
Ind = y*PE
TCC = PE -f Inst + Ind
1.15
TC=1.15*TCC
CC = TC/(Pi*1000)
                                          10

-------
Using the procedure shown in Table 3, capital costs ($/kW) were calculated for each of the ACI-
based technology cases shown in Table 1 for model boilers1'n of sizes 100 and 975 MW. For
each of these cases, the capital costs ($/kW) in 1993 dollars thus obtained were then used to
develop the capital cost equation relating ($/kW) to boiler size (MW). This function has the form
                                                                                (2)
where:
 Q =  capital cost ($/kW) of ACI-based technology installation at the first model boiler;
 C2 =  capital cost ($/kW) of ACI-based technology installation at the second model boiler;
 Pl = first model boiler size in MW;
 F2 = second model boiler size in MW; and
 a = a scaling factor reflecting  economy-of-scale.

For technologies whose capital costs ($/kW) reflect an economy-of-scale, the value of the scaling
factor a can range from approximately 0.3 to O.6.3

Annual Fixed O&M Cost Estimates

Fixed O&M costs include costs related to maintenance requirements (labor and materials).
Following the guidelines in the Electric Power Research Institute's Technical Assessment
Guide13, the annual cost of maintenance labor and materials, expressed in ($/kW-yr) is assumed
to be 1.5% of the capital cost ($/kW).

Annual Variable O&M Costs Estimates

Variable O&M costs include operating labor and supervision charges, costs related to
consumables (e.g., activated carbon), cost of energy used, any incremental waste disposal cost,
cost of operating materials (e.g., water for SC), and overhead. Calculation of each of these is
described below. Note that, although the Mercury Study Report to Congress1 was released in
December 1997, the equipment-related cost data in this study appear to be in 1993  dollars. As a
conservative measure, for this study all of the cost data presented in Reference 1 were assumed to
be in 1993 dollars.

Labor cost:   Both operating and supervisory labor charges are accounted for in the cost
estimates. Following the information provided in the Mercury Study Report to Congress1, the
labor rate is taken to be $12/hr and the supervisory labor costs total 15% of the operating labor
costs.  Then, for a boiler of size PI operating with a capacity factor (CF) labor cost was computed
using
                                          11

-------
Activated carbon cost:       Based on recent information14, the cost of AC is taken to be $ 1.00
per kg of activated carbon. Using this cost along with the flue gas flow, concentration of mercury
in the flue gas, CF, and C/Hg ratio, the cost of AC consumption (mills/kWh) is determined for
each model boiler application. An average of the costs for two model boiler applications was
taken to be the carbon consumption cost (mills/kWh) for each of the ACI-based technologies.

Energy consumption cost:   Energy costs provided in Reference 1 for model boiler applications
were based on a unit energy cost of 46 mills/kWh and on a carbon injection ratio of 460 g of
carbon per g of mercury. For this work, the annual energy consumption costs ($/yr) provided in
Reference 1 were adjusted to reflect any change in carbon injection ratio and an energy cost of
19.4 mills/kWh in 1993. The adjusted energy consumption costs ($/yr) were used to arrive at the
energy consumption cost (mills/kWh) for model boiler applications. The average of the costs for
two model boiler applications was taken to be the carbon consumption cost (mills/kWh) for each
of the ACI-based technologies.

Disposal cost:       Annual disposal costs ($/yr) for the model boiler applications provided in
Reference 1 were adjusted to reflect apy change in carbon injection ratio, and then were used to
arrive at disposal costs expressed in mills/kWh. The average of the costs for two model boiler
applications was taken to be the disposal cost (mills/kWh) for each of the ACI-based
technologies.

Operating materials cost:    Annual operating materials costs ($/yr) for the model boiler
applications provided in Reference 1 were used to arrive at costs expressed in mills/kWh. The
average of the costs for two model boiler applications was taken to be the annual operating
materials cost (mills/kWh) for each of the ACI-based technologies. Note that in calculation of
these costs, it was assumed that the amount of water used for SC would not depend on C/Hg
ratio.

Overhead:    Following the information provided in Reference 1, the overhead cost for each of
the ACI-based technologies was taken to be 60% of the operating labor and maintenance costs.

Annual Carrying Charge

Annual carrying charge (ACC) includes costs related to capital recovery, property taxes,
insurance, and administration. The annual carrying charge factor of 10,4% used in the EPM
model3 accounts for these cost elements. Therefore, in this work annual carrying charge is
expressed by

    ACC($/kW - yr) = 0.104x Capital Cost($/kW)                                (4)
                                          12

-------
Total Annual Cost

Total annual cost includes ACC, annual fixed O&M, and annual variable O&M.

RESULTS

Capital and total annual costs for the mercury control cases without FGD are shown in Figures 1
and 2, respectively. As discussed before, incremental costs for cases with existing wet FGD are
identical to corresponding cases without wet FGD, but 90% mercury removal is considered to be
possible in the cases with wet FGD. As such, costs for cases with wet FGD (Cases 3A, 3B, 4A,
4B, 6A, and 6B) are not shown in these figures. From Figure 1, it is evident that relatively higher
capital costs, about 40-52 $/kW, are estimated for installations of mercury controls on boilers
using a hot ESP, viz., Cases 2A and 2B. In each of these cases, a FF is installed to capture spent
AC, raising the capital cost. For other technologies, these costs are lower than 10 $/kW. It is
interesting to note that capital costs for cases other than 2A and 2B do not exhibit significant
economies-of-scale that  would generally be expected. This lack of economies-of-scale is a direct
result of the data used in the determination of capital cost functions.

Figure 1. Capital costs of ACI-based technologies for mercury control.
              in
             £30

             Jj
             '5.
              CO
             O 20
                  100
300         500         700
      Boiler Size (MW)
1000
                                          13

-------
Figure 2. Total annual costs of ACI-based technologies for mercury control.
                 100
300         500         700
      Boiler Size (MW)
1000
Total annual costs of ACI-based technologies range from about 0.2 to 1.8 mills/kWh. Total
annual costs are relatively higher for applications utilizing FFs for mercury control (Cases 2A
and 2B) due to higher values of capital recovery. For other cases, these costs are estimated to be
about 1.0 mill/kWh or less. It is interesting to note that, in contrast to capital costs for cases other
than 2A and 2B, total annual costs for these cases do exhibit economies-of-scale. This is due to
the effect of the labor cost, which decreases with increasing boiler size [see equation (3)],

An understanding of the mercury control costs may be gained by comparing these with costs of
currently used controls for nitrogen oxides (NOX). Shown in Table 4 are the ranges of capital and
total annual costs for the mercury controls examined in this work and for two of the currently
used NOX control technologies, viz., Low NOX Burner (LNB) and Selective Catalytic Reduction
(SCR). These cost ranges reflect costs for technology applications on dry-bottom, wall-fired
boilers ranging in size from  100 to 1200 MW. In general, costs associated with LNB and SCR
are expected to span the costs of currently used NOX controls; therefore these costs were chosen
for comparison with mercury control costs. These costs were derived from the information
available in Reference 3.
                                           14

-------
     Table 4. Comparison of mercury control costs with NOX control costs
Control
Mercury Control Costs
LNB Costs
SCR Costs
Capital Costs
($/kW)
0.43-52.21
7.31 - 35.89
40.88-91.51
Total Annual Cost
(mills/kWh)
0.17-1.76
.0.15-0.54
1.30-2.41
As seen in Table 4, capital and total annual costs for mercury controls are estimated to lie mostly
between applicable costs for LNB and SCR. However, Figures 1 and 2 show capital and total
annual costs of mercury controls to be higher for Cases 2A and 2B that are applicable to the
minority of plants using hot ESPs. Therefore, for many applications, capital costs of mercury
controls may be comparable to LNB capital costs, and total annual costs may be comparable to,
or in some cases higher than, those for LNBs.

To gain further understanding of costs, contributions of total annual cost components were
calculated for each of the mercury control cases applied to a 500 MW boiler. The results of these
calculations are shown in Figure 3. As discussed before, costs for cases with wet FGD are
identical to corresponding cases without wet FGD, but 90% mercury removal is considered to be
possible in the cases with wet FGD. As  such, cases with wet FGD (Cases 3A, 3B, 4A, 4B, 6A,
and 6B) are not shown in Figure 3.

It is evident from Figure 3 that AC costs comprise the dominant cost component and range from
approximately 40 to 70% of the total annual costs for all cases, except ones in which FFs are
installed as part of mercury control (i.e., Cases 2A and 2B). The cost of FF results in increased
capital cost, as discussed before. Accordingly, annual carrying charge contributes about 40% of
total annual cost for each of Cases 2A and 2B. For these cases, AC costs range from
approximately 20 to 25% of the total annual costs. Moreover, annual carrying charge
contributions in cases other than 2A and 2B are relatively small, and range from approximately 3
to 15%. These results indicate that changes in the costs associated with AC may have a
significant impact on cost of ACI-based mercury controls and that, except for retrofits requiring
installation of FFs (Cases 2A and 2B), changes in capital cost are not likely to significantly affect
total annual cost. To investigate these indications, sensitivity analyses were conducted to assess
the impact of changes in capital and AC-related costs for each of the mercury control cases
without wet FGD. As mentioned before, within the assumptions of this work, costs of wet FGD
cases will be identical to corresponding cases without wet FGD.
                                          15

-------
Figure 3. Contributions of total annual cost components for technology applications on a 500
MW boiler.
              1A    1B    2A    2B    5A    5B    7 A    7B    8A    8B
                                                                            ^Overhead
                                                                            BOp, Mails.
                                                                            • Disposal
                                                                            H Carbon
                                                                            El Labor
                                                                            pFixed O&M
                                                                            OACC
The impact of a 50% increase or decrease in capital cost is shown in Table 5 for each of the
mercury control cases applied on a 500 MW boiler. As seen in this table, for all cases except 2A
and 2B, significant changes in capital costs result in only modest changes (about 10% or less) in
total annual costs. As mentioned above, in both 2A and 2B a FF is installed as part of mercury
control and the capital cost associated with the FF results in a relatively dominant ACC cost
component (see in Figure 3). Consequently, changes in this component result in significant
changes in total annual cost. Considering that only a relatively small percentage of boilers with
existing hotlSSPs would use Cases 2A and 2B, capital costs would not be expected to heavily
influence the total annual costs of ACI-based mercury controls, in most cases.

ACI-related costs are influenced by price of AC and by AC injection requirements. In general,
the price of AC would be influenced by the market factors of supply and demand while
technological improvements would influence AC injection requirements. Note that AC injection
requirements affect both variable and fixed O&M costs as sizing of an ACI system is based on
                                           16

-------
these requirements. In this sense, AC injection requirements play a significant role in total annual
cost of any ACI-based mercury control.
Table 5. Impact of changes in capital cost of mercury controls applied on a 500 MW boiler.
Case
1A
IB

2A
2B

5A
5B

7A
7B

8A
8B
Base Total
Annual Cost
(mllls/kWh)
0.92
0.58

1.44
1.24

0.74
0.54

0.35
0.19

0.55
0.35
50% Increase in Capital Cost
Total Annual
Cost
(mills/kWh)
0.98
0.63

1.70
1.49

0.79
0.59

0.35
0.20
ix
0.56
0.35
Percent Change
(%)
6.5
8.6

18.1
20.2

6.8
9.3

0.0
5.3

1.8
0.0
50% Decrease in Capital Cost
Total Annual Cost
(mills/kWh)
0.86
0.52

1.09
0.90

0.68
0.49

0.34
0.19

0.54
0.34
Percent Change
(%)
-6.5
-10.3

-24.3
-27.4

-8.1
-9.3

-2.9
0.0

-1.8
-2.9
The impact of a 50% increase or decrease in AC injection requirements is shown in Table 6 for
each of the mercury control cases without FGD, applied on a 500 MW boiler. As seen in Table 6,
for all cases except those requiring installation of FFs (Cases 2A and 2B), significant changes in
AC injection requirements result in significant changes, ranging between 30 and 45%, in total
annual costs. Again considering that only relatively few boilers with existing hot ESP would use
Cases 2A or 2B, the AC injection requirement appears to drive the total annual costs of ACI-
based mercury controls. Since utilization of AC is intimately related to physical and chemical
mechanisms affecting mercury's speciation, mass transfer, and adsorption, ongoing R&D efforts
in these areas may have a profound impact on development of more cost-effective mercury
controls.
                                           17

-------
Table 6. Impact of changes in activated carbon injection requirements for a 500 MW boiler.
Case
1A
IB

2A
2B

5A
5B

7A
7B

8A
8B
Base Total
Annual Cost
(mllls/kWh)
0.92
0.58

1.44
1.24

0.74
0.54

0.35
0.19

0.55
0.35
50% Increase in AC Injection
Requirement
Total Annual
Cost
(mills/kWh)
1.26
0.75

1.69
1.39

0.99
0.69

0.50
0.27

0.80
0.50
Percent Change
(%)
36.96
29.31

17.36
12.10

33.78
27.78

42.86
42.11

45.45
42.86
50% Decrease in AC Injection
Requirement
Total Annual
Cost
(mills/kWh)
0.58
0.41

1.19
1.08

0.49
0.39

0.19
0.12

0.30
0.19
Percent Change
(%)
-36.96
-29.31

-17.36
-12.90

-33.78
-27.78

-45.71
-36.84

-45.45
-45.71
CONCLUSIONS

A methodology for estimating costs of ACI-based mercury control technologies for coal-fired
electric utility boilers is presented in this paper. Using this methodology and available data,
preliminary estimates of costs of these technologies applicable to the existing boiler population
have been determined. Examination of these costs reveals that for the majority of applications
capital costs of mercury controls would be comparable to LNB capital costs, and total annual
costs would be comparable to, or in some cases higher than, those for LNBs.

Results of sensitivity analyses conducted on cost estimates generally reflect that total  annual
costs of ACI-based mercury controls are relatively insensitive to capital costs. The AC injection
requirement appears to dominate the total annual cost. Since utilization of AC is intimately
related to physical and chemical mechanisms affecting mercury's speciation, mass transfer, and
adsorption, ongoing R&D efforts in these areas may have a profound impact on  development of
more cost-effective mercury controls.

The performance and cost estimates of the ACI mercury control technologies presented in this
paper are based on relatively few data points from pilot-scale tests and, therefore, are considered
to be preliminary. Factors that are known to affect adsorption of mercury on activated carbon
include speciation of mercury in flue gas, available residence time (duct length), effect of flue gas
and ash characteristics, and degree of mixing between flue gas and activated carbon. The effect
of these factors may not be entirely accounted for in the relatively few pilot-scale data points that
comprised the basis for this work. Ongoing research is expected to address these issues.
                                           18

-------
DISCLAIMER

The material presented in this paper is of a preliminary nature and does not reflect any EPA
positions. Furthermore, the paper has no bearing on any future EPA actions.

REFERENCES

1.  Mercury Study Report to Congress, Office of Air Quality Planning and Standards and Office
   of Research and Development, U.S. Environmental Protection Agency, Research Triangle
   Park, NC, December 1997, EPA-452/R-97-003 (NTIS PB98-124738).

2.  Analysis of Emissions Reduction Options for the Electric Power Industry, Office of Air and
   Radiation, U.S. Environmental Protection Agency, Washington, D.C., March 1999. Available
   at the web site www.epa.gov/capi.

3.  Analyzing Electric Power Generation Under CAAA; Office of Air and Radiation, U.S.
   Environmental Protection Agency, Washington, D.C., March 1998. Available at the web site
   www .epa.gov/capi.

4.  Brown, T.D.; Smith, D.N.; Hargis, R.A.; O'Dowd, WJ. "1999 Critical Review, Mercury
   Measurement and its Control: What We Know, Have Learned, and Need to Further
   Investigate," Journal of the Air & Waste Management Association, June 1999, pp. 1-97.

5.  Study of Hazardous Air Pollutant Emissions from Electric Utility Steam Generating Units -
   Final Report to Congress, Volume 1, Office of Air Quality Planning and Standards, U.S.
   Environmental Protection Agency, Research Triangle Park, NC, February 1998, EPA-453R-
   98-004a (NTIS PB98-131774).

6.  Waugh, E.G.; Jensen, B.K.; Lapatnick, L.N.; Gibbons, EX.; Sjostrom, S.; Ruhl, J.; Slye, R.;
   Chang, R. "Mercury Control in Utility ESPs and Baghouses through Dry Carbon-Based
   Sorbent Injection Pilot-Scale Demonstration," EPRI-DOE-EPA Combined Air Pollutant
   Control Symposium, Particulates and Air Toxics, Volume 3, EPRITR-108683-V3, Electric
   Power Research Institute, Palo Alto, CA, August 1997, pp. 1-15.

7.  Waugh, E.G.; Jensen, B.K.; Lapatnick, L.N.; Gibbons, F.X.; Haythornthwaite, S.; Sjostrom,
   S.; Ruhl, J.; Slye, R.; Chang, R. "Mercury Control on Coal-Fired Flue Gas Using Dry
   Carbon-Based Sorbent Injection: Pilot-Scale Demonstration," presented at the 1998 Air &
   Waste Management Association Annual Meeting and Exhibition, San Diego, CA, June 1998,
   pp.1-15.

8.  Haythornthwaite, S.; Sjostrom, S.; Ebner, T.; Ruhl, J.; Slye, R.; Smith, J.; Hunt, T.; Chang,
   R.; Brown, T.D. "Demonstration of Dry Carbon-Based Sorbent Injection for Mercury Control
   in Utility ESPs and Baghouses," EPRI-DOE-EPA Combined Air Pollutant Control
   Symposium, Particulates and Air Toxics, Volume 3, EPRI TR-108683-V3, Electric Power
   Research Institute, Palo Alto, CA, August 1997.
                                         19

-------
9.  Redinger, K.E.; Evans, A.P.; Bailey, R.T.; Nolan, P.S. "Mercury Emissions Control in FGD
   Systems," EPRI-DOE-EPA Combined Air Pollutant Control Symposium, Particulates and
   Air Toxics, Volume 3, EPRITR-108683-V3, Electric Power Research Institute, Palo Alto,
   CA, August 1997.

10. Kilgroe, J.D. Journal of Hazardous Materials, 1966,47, pp. 163-194.

11. Brown T.; O'Dowd, W.; Reuther, R.; Smith, D. "Control of Mercury Emissions from Coal-
   Fired Power Plants: A Preliminary Cost Assessment," in Proceedings of the Conference on
   Air Quality, Mercury, Trace Elements, and Particulate Matter, Energy & Environmental
   Research Center, McLean, VA, December 1998, pp. 1-18.

12. Economic Report of the President, Council of Economic Advisers, February 1998. Available
   at the web site www.access.gpo.gov/eop/,

13. Technical Assessment Guide, Volume 1: Electricity Supply —1993 (Revision 7), EPRI TR-
   102276s Vol. 1 Rev. 7, EPRI, Palo Alto, CA, 1993.

14. Personal communication between Charles Sedman and James Kilgroe of EPA and Anthony
   Licata of Licata Energy & Environmental Consultants, Inc., Yonkers, NY, December 14,
   1998.
                                        20

-------
 N RM RL-RT P-P-4 72
       TECHNICAL REPORT DATA
(f lease read Instructions on the reverse before completing)
1. REPORT NO.
   1PA/600/A-00/037
  2.
                                                      3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
 Preliminary Performance and Cost Estimates of
  Mercury Emission Control Options for Electric
  Utility Boilers
                                                      S. REPORT DATE
                             6. PERFORMING ORGANIZATION CODE
7. AUTHORiS)
 Ravi K. Srivastava, Charles B.  Sedman, and
 James  D.  Kilgroe
                             8. PERFORMING ORGANIZATION REPORT NO,
9. PERFORMING ORGANIZATION NAME AND ADDRESS
 See Block 12
                                                      10. PROGRAM ELEMENT NO.
                             11. CONTRACT/GRANT NO,

                               NA (Inhouse)
12. SPONSORING AGENCY NAME AND ADDRESS
 EPA, Office of Research and Development
 Air Pollution Prevention and Control Division
 Research Triangle Park, NC 27711
                             13. TYPE OF REPORT AND PERIOD COVERED
                              Published paper; 10-12/99
                             14. SPONSORING AGENCY CODE
                              EPA/600/13
is.SUPPLEMENTARYNOTES^PPCD pr0ject officeris Ravi R. Srivastava, Mail Drop 65, 919/
 541-3444. For presentation at AWMA Conference,  Salt Lake City, UT,  6/18-22/00.
16. ABSTRACT
              paper discusses preliminary performance and cost estimates of mer-
 cury emission control options for electric utility boilers.  Under the Clean Air Act
 Amendments of 1990, EPA had to determine whether mercury emissions from coal-
 fired power plants  should be regulated. To aid in this determination, preliminary
 estimates of performance  and cost of activated- carbon-based mercury control tech-
 nologies were developed, based on relatively  few data points from pilot- scale tests.
 The paper presents the methodology used in arriving at these estimates,  estimated
 capital and annual operation and maintenance  costs,  and results of sensitivity analy-
 ses conducted on these cost estimates. Results reveal that,  for most applications,
 capital costs of activated carbon injection (ACI) mercury  control technologies may
 be comparable to those associated with low nitrogen oxide burners (LNBs). Also
 total annual costs may be comparable  to, or in some cases slightly higher than,
 corresponding costs for LNBs. Moreover,  AQ requirements appear to dominate
 the total annual costs of ACI-based technologies.  The effect of several factors, . .
 known to affect the adsorption of mercury on  activated carbon,  may  not be entirely
 accounted for in the relatively few pilot- scale data points  that comprised  the  basis
 for this work. Ongoing research is expected to address these issues.
17.
                             KEY WORDS AND DOCUMENT ANALYSIS
                DESCRIPTORS
                                          b.lDENTIFIERS/OPEN ENDED TERMS
                                          c.  COSATI Field/Group
 Pollution           Performance
 Mercury            Cost Estimates
 Emission           Activated Carbon
 Steam Electric Power Generation
 Coal
 Combustion
                  Pollution Control
                  Stationary Sources
13B  14G
07B  05A.14A
14G  11G
10A
21D
21B
18. DISTRIBUTION STATEMENT
 Release to Public
                                           19. SECURITY CLASS {This Report)
                                           Unclassified
                                          21. NO. OF PAGES
                                              20
                 2O. SECURITY CLASS (This page)
                 Unclassified
                                          22. PRICE
EPA Form 2220-1 (9-73)

-------