-------
same general range of values. No trend is apparent from these data; both bituminous and
subbituminous coals are found at the lower and upper ends of the range. For example, a
bituminous coal from the Raton Mesa region and a subbituminous coal from the Green River
region each have an average Hg content of 6.6 Ib per 1012 Btu. At the other end of the range, a
bituminous coal from the Western Interior region has an average Hg content of 16.1 Ib per
1012 Btu and a subbituminous coal from the Wind River region has an average Hg content of
18.7 Ib per 1012 Btu. On the other hand, the Hg contents reported for the two lignite coals listed
in Table 2-2 are significantly higher than any of the bituminous and subbituminous coals (an
average of 21.8 Ib per 1012 Btu for Fort Union lignite and 36.4 Ib per 1012 Btu for Gulf Coast
lignite).

       Another key reason why the Hg content of as-mined coals cannot be related to Hg
emissions is the as-mined coal frequently is not burned in an electric utility boiler as it comes
directly from the mine. The as-mined, or raw, coal often is first processed at a coal preparation
plant to remove non-coal impurities in order to provide the coal purchaser with a uniform coal
that meets a predetermined, contractual set of specifications. These processes commonly are
collectively referred to as "coal cleaning."  Depending on the properties of the coal and the type
of process used, coal cleaning can reduce the Hg content of the coal that is ultimately  fired in the
electric utility boiler.
2.3 Coal Cleaning

2.3.1  Coal Cleaning Processes

       Raw coal from a mine contains separate rock, clay, and other minerals.  After the coal is
mined, it may first pass through a series of processes known as coal preparation or coal cleaning
before it is snipped to an electric utility power plant. The coal is processed for three main
reasons: 1) to reduce the ash content; 2) to increase the heating value; and 3) to reduce the sulfur
content to ultimately lower emissions of sulfur dioxide when the coal is burned in the utility
boiler. The removal of impurities from the coal also helps to reduce power plant maintenance
costs and to extend the service life of the boiler system.

       Coal cleaning processes currently in use separate the organic fraction of the as-mined coal
from the mineral materials according to the differences in either the density-based or surface-
based characteristics of the different materials. Physical coal cleaning typically involves a series
of process steps including: 1) size reduction and screening, 2) gravity separation of coal from
sulfur-bearing mineral impurities, and 3) dewatering and drying.

       Bituminous coals from mines in the Eastern and Midwestern United States frequently are
cleaned to meet the electric utility customer's specifications for heating value, ash content, and
sulfur.  It is estimated that about three-fourths (77 percent) of these coals are cleaned prior to
shipment to an electric utility power plant.8  The subbituminous and lignite coals from mines  in
the Western United States routinely are not cleaned before shipment to an electric utility power
                                           2-9

-------
plant, but in special cases these types of coals can be cleaned. For example, some of
subbituminous coal from mines in the Powder River coal region (a major source of coal for many
electric utilities) is cleaned for shipment to electric utility customers.

2.3.2 Mercury Removal by Coal Cleaning

      Conventional coal cleaning methods will also remove a portion of the Hg associated with
the incombustible mineral materials but not the Hg associated with the organic carbon structure
of the coal. Any reduction in Hg content of the coal shipped to an electric utility power plant
obtained from the Hg removed by coal cleaning processes transfers the removed Hg to the coal
cleaning wastes. Limited data have been gathered on the level of Hg removed by conventional
coal  cleaning methods.

       A review of test data for 26 bituminous coal samples from coal seams in four states
(Illinois, Pennsylvania, Kentucky, and Alabama) prepared for EPA's Mercury Study Report to
Congress indicates a wide range in the amount of Hg removed by coal cleaning.8 In some cases,
analysis of coal samples from the same coal seam also showed considerable variability. Analysis
of five of the coal samples showed no Hg removal associated with conventional coal cleaning
while the remaining 21 coal samples had Hg reductions ranging from approximately 3 to 64
percent. The average Hg reduction for all of the data was approximately 21 percent.

       Other studies have reported higher average Hg reductions for Eastern and Midwestern
bituminous coals. One study tested 24 samples of cleaned coal.7 These data also showed a wide
range in Hg reduction rates. The average decrease in Hg reduction on an energy basis was 37
percent, with values ranging from 12 to 78 percent.  On a mass basis, the average Hg reduction
from coal cleaning was 30 percent.  A higher Hg reduction was reported on an energy basis than
on a  mass basis because the coal cleaning raises the heating value per unit mass of the coal, as
well  as removing Hg. A second study of the effects of coal cleaning on Hg content for three
Ohio coals reported reductions in Hg content of the  coals ranging from 36 to 47 percent.9

      The variation in Hg reductions observed from the test data might be a function of the type
of process used to clean a given coal and the proportion of Hg in the coal that is present in
combination with pyrite (iron disulfide). Coal-cleaning processes that make separations
according to the density differential of particles are generally more effective in removing Hg
associated with pyrite than are surface-based processes.  The heavier pyrite is easily removed by
density-based processes, but not by surface-based processes where the similar surface
characteristics of pyrite  and the organic matter make separation of the two components difficult.
For coals that have larger portions of Hg associated with pyrite, density-based cleaning processes
are expected to have higher Hg removals. However, some coals may contain large portions of
Hg associated with the organic fraction of the coal; Hg removal in these cases would be expected
to be substantially lower since the organic fraction of coal is not removed during cleaning.
Additional reductions in Hg can probably be achieved by using more intensive coal cleaning
methods. Several advanced coal cleaning techniques being investigated to improve Hg removal
are discussed in Chapter 7.
                                          2-10

-------
2.4 Coal-fired Electric Utility Boilers

       The large steam boilers used by electric utilities are also referred to as "steam generators,"
"steam generating units," or simply "boilers." As discussed in Chapter 1, CAA Section 112(a)
defines the term "electric utility steam generating unit" to include those units that cogenerate
steam and electricity and supply more than one-third of its potential electric output capacity and
more than 25 megawatts electrical output to any utility power distribution system for sale. For
simplicity in the remainder of this report, the term "electric utility boiler" is used to mean
"electric utility steam generating unit" as defined in CAA Section 112(a)(8).

       A total of 1,143 coal-fired units meeting the CAA definition of an "electric utility steam-
generating unit" were reported in the Part II EPA ICR data to be in the United States in 1999.10
More than one boiler unit is often operated at an electric utility power plant.  The 1,143 units
were located at a total of 461 facilities. These facilities can be categorized in three facility types:
conventional coal-fired electric utility power plants, coal-fired cogeneration facilities, and
integrated coal  gasification and combined cycle (IGCC) power plants.

2,4J Conventional Coal-fired Electric Utility Power Plants ll'n

       A conventional electric utility power plant burns coal in a boiler unit solely for the
purpose of generating steam for electrical power production. A total  of 1,122 coal-fired electric
utility boilers were reported in the Part II EPA ICR data to be operating at conventional electric
utility power plants. Each of these boilers was designed to meet plant load and performance
specifications by burning coals  within a specific range of coal properties (e.g., heating value, ash
content and characteristics, and sulfur content).  While the specific equipment and design of a
coal-fired electric utility boiler will vary from plant to plant, the same basic process is used to
generate electricity. Figure 2-2 presents a simplified schematic of the major components of a
coal-fired electric utility boiler operated at a conventional electric utility power plant.

       Coal typically is delivered to a power plant by railcars, trucks, or barges. At some power
plants located near the mine supplying the coal, coal is delivered by a slurry pipeline or an
extended conveyor system. Also, a few power plants burn imported coal that is delivered to the
facility by ship. The delivered coal is unloaded and stored in outdoor storage piles or covered
storage structures such as silos or bins. Depending on how the coal is burned in the boiler (e.g.,
in a bed or burned in suspension), the coal is crushed or pulverized before being fed to the boiler.

      A conventional coal-fired electric utility boiler consists of multiple sections, each  of which
serves a specific purpose. The coal is ignited and burned in the section of the boiler called the
"furnace chamber.'" Blowing ambient air into the furnace chamber provides the oxygen required
for combustion. The carbon and hydrogen comprising the coal are oxidized at the high
temperatures produced by combustion to form the primary combustion products of carbon
dioxide (CO,) and water (H2O). Sulfur in the coal is oxidized to form SO2.  Molecular nitrogen
in the combustion air and nitrogen bound in the coal react with oxygen in certain sections of the
combustion zone in the furnace chamber to form NCv Small amounts  of other gaseous
                                           2-11

-------
                                                O
                                                o

                                                I

                                                I
                                                1

                                                1
                                                 3
                                                 Q.
                                                 O)
                                                 o
                                                A

                                                I"

                                                ••p
                                                 3
                                                 O
                                                 o

                                                "5
                                                 o
                                                V
                                                 CO
                                                TJ
                                                 0)
                                                 0.


                                                55
                                                 3
                                                 O)
2-12

-------
combustion products form from other impurities in the coal.  These hot combustion products are
vented from the furnace in a gas stream called collectively "flue gas." Additionally, most but not
all the carbon in the coal is burned in the furnace. Unburned or partially burned solid carbon
particles are entrained and vented from the furnace in the flue gas.

       The walls of the furnace chamber are lined with vertical tubes containing water.  Heat
transfer from the hot combustion gases in the furnace boils the water in the tubes to produce
high-temperature, high-pressure steam. This steam flows from the boiler to a steam turbine. In
the turbine, the thermal energy in the steam is converted to mechanical energy to drive a shaft
that spins a generator, which produces electricity. After the steam exits the turbine, it is
condensed and the water is pumped back to the boiler.

       To improve overall energy conversion efficiency, modern coal-fired electric utility boilers
contain a series of heat recovery sections. These heat recovery sections are located downstream
of the furnace  chamber and are used to extract additional heat from the flue gas. The first heat
recovery section contains a "superheater," which is used to increase the steam temperature.  The
second heat recovery section contains a "reheater," which reheats the steam exhausted from the
first stage of the turbine.  This steam is then returned for another pass thorough a second stage of
the turbine. The reheater is followed by an "economizer," which preheats feed water to the boiler
tubes in the furnace.  The final heat recovery section is the "air heater," which preheats ambient
air used for combustion of the coal.

       A portion of all coals is composed of mineral matter that is noncombustible. This matter
forms the ash that continuously must be removed from the operating utility boiler.  The ash
collection points and removal systems used for  a given boiler unit are dependent on the ash
properties and content in the coal-fired, the boiler design, and the air pollution control devices
used. The removal and handling of the coal ash is discussed further in Section 2.6.

       The flue gas exhausted from the boiler passes through air pollution control equipment and
is vented to the atmosphere through  a tall stack. The types and configurations of air pollution
controls currently used for coal-fired electric utility boilers are discussed in Chapter 3.

2.4,2 Coal-fired Cogeneration Facilities

       Approximately six percent of the boiler units are at cogeneration facilities, which are
owned and operated by independent power producers or industrial companies. Of the 1,143 total
coal-fired electric utility boilers reported in the  EPA Part IIICR data, 68 are classified as
cogeneration units. The total generating capacity of these cogeneration units is 867 MWe.  There
are more coal-fired boilers in the United States  operating as cogeneration units; however, these
units  do not meet the criteria specified in the CAA definition of a steam-generating unit (i.e., the
cogeneration unit is rated below 25 MWe or less than one-third of the unit's electrical output is
sold). These units were not surveyed for the EPA ICR database.
                                          2-13

-------
       Operation of a cogeneration facility differs from the operating configuration of the
conventional electric utility power plant shown in Figure 2-2. Two basic cogeneration unit
configurations are used: the "topping" mode or the "bottoming" mode. In the topping
cogeneration configuration, steam produced by the coal-fired electric utility boiler is used first to
generate electricity and then all or part of the exhaust heat is subsequently used for an industrial
process. The bottoming cogeneration configuration reverses this sequence using waste heat
generated by an industrial process to produce steam in a heat recovery boiler for driving a steam
turbine and generating electricity. All of the cogeneration boiler units listed in the EPA ICR data
operate using the topping mode configuration.

2.4.3 Integrated Coal Gasification Combined Cycle Power Plants

       The IGCC power plants represent a new technology and are different from conventional
electric utility power plants in two major characteristics. First, the IGCC power plants do not
bum the coal in its solid form.  Instead, the coal is first converted to a combustible gas using a
coal gasification process at the facility site.  Second, the IGCC power plants generate electricity
using two separate thermal cycles and associated turbines referred to as a "combined cycle"
operation. The coal-derived gas from the gasification process is first burned in a gas turbine that
drives an electrical generator. The exhaust gases from this gas turbine pass through a heat
recovery boiler to generate steam to power a steam turbine that drives a second electrical
generator. Three IGCC power plants have been built in the United States.  The operation of these
power plants is discussed further in Section 2.5.5.
2.S Coal-firing Configurations for Electric Utility Boilers

       Coal can be burned in a boiler using one of three basic techniques: burning coal particles
in suspension, burning large coal chunks in a fuel bed, or in a two-step process in which the coal
is first converted to a synthetic gas which is then fired in the boiler.  Five basic firing
configurations are used to burn coal for electric power generation: pulverized-coal-fired furnace,
cyclone furnace, fluidized-bed combustor, stoker-fired furnace, and gasified-coal-fired
combustor. A general comparison of the different coal-firing configurations used for electric
utility power plants is presented in Table 2-3.

      Table 2-4 shows the distribution of the  1,143 coal-fired electric utility boilers listed in the
EPA ICR data by coal-firing configuration. Pulverized-coal-fired designs account for the vast
majority of the coal-fired electric utility boilers both in terms of total number of units
(approximately 86 percent) and nationwide generating capacity. Cyclone furnaces are used to
burn coal in approximately eight percent of the units. Fluidized-bed combustors are used for
about four percent of the coal-fired electric utility boilers. Stoker-fired furnaces account for
about three percent of the total number of coal-fired electric utility boilers but provide less than
one percent of the total coal-fired megawatts.  Only three IGCC units have been built in the
United States.
                                           2-14

-------
J)

A
re
Design/Operating Characteristics
$
0
•c
M
O



Coal Combustion Process Description
w|
*?

31
o
An array of burners fire into the furnace
horizontally, and can be positioned on
one wall or opposing walls depending on
the furnace design.


•o


1


Multiple burners are positioned in
opposite corners of the furnace produdng
a fireball that moves in a cyclonic motion
and expands to fill the furnace.


0) TT"
i ^
c *
l§
1- *~

Coal is ground to a fine powder that is pneumatically
fed to a burner where it is mixed with combustion air
and then blown into the furnace. The pulverized-coal
particles burn in suspension in the furnace. Unburned
and partially burned coal particles are carried off with
the flue gas.
5 Q
•a !?
al
11
a.
sh fusion coals and retain most of the ash in
ig.
jg-55
= 1
5 E
2 ^
•a °
S>&
oi
Coal is crushed into small pieces and fed through a
burner into the cydone furnace. A portion of the
combustion air enters the burner tangentially creating a
whirling motion to the incoming coal.

0> QJ
•| |
O >2

Operates at relatively low gas stream
velocities and with coarse-bed size
particles. Air in excess of that required to
fluidize the bed passes through the bed in
the form of bubbles.

"8
a
•a
;o CD
c CO
_c
£>
m
Operates at higher gas stream velocities
and with finer-bed size particles. No
defined bed surface. Must use high-
volume, hot cydone separators to
recirculate entrained solid particles in flue
gas to maintain the bed and achieve high
combustion efficiency.


•o
O) 0>
"Sff
3.S&
C TJ «.
o ;=


Coal is crushed to fine particles. The coal particles are
suspended in a fluidized bed by upward-blowing jets of
air. The result is a turbulent mixing of combustion air
with the coal particles. Typically, the coal is mixed with
an inert material (e.g., sand, silica, alumina) and a
sorbent such as limestone (for SO, emission control).
The unit can be designed for combustion within the bed
to occur at atmospheric or elevated pressures.
Operating temperatures for FBC are in the range of 850
to 900 °C.
•a
•n 8>
81
1 °

                                               2-15

-------
               Si
               •o
               n

               2
               Q.

               ffi
j by pushing coal lumps along

ough underneath the grate.
                        O
                                = *
                                l!

                                I*
                                S?
                                «a m
                                SS



                                tj.
                                >-S-|


                                sis
                                s
                                O
 £
s
O)

oi
c
                     "O 0) .a
                     ^ (Q O)


                     c .5 JJ 1;
                     f
                       ID  . •
                        Q.E
       (0 TO u .  "55  "to
I

eo
                      OJ
                                    2-16

-------
Table 2-4. Nationwide distribution of electric utility units by coal-firing
configuration for the year 1999 as reported in the Part II EPA ICR data (source:
Reference 10).
Coal-firing
Configuration
Pulverized-coal-fired furnace
Cyclone furnace
Fluidized-bed combustor
Stoker-fired furnace
Gasified-coal-fired combustor
Nationwide Total
| Nationwide
{ Total Number of
I Units
I
I
! 979
\
I
87
•
|
|
I 32
I
I
!
I 3
!
i
I 1,143
i
Percent of
Nationwide Total
85.6 %
7.6 %
3.7 %
2.8 %
0.3 %
100 %
Percent of
Nationwide
Electricity
Generating
Capacity
90.1 %
7.6 %
1.3%
1.0%
< 0.1 %
100%
                                     2-17

-------
2.5.1  Pulverized-coal--firedFurnace

        To burn in a pulverized-coal-fired furnace, the coal must first be pulverized in a mill to
 the consistency of talcum powder (i.e.; at least 70 percent of the particles will pass through a
 200-mesh sieve). The pulverized coal is generally entrained in primary air before being fed
 through the burners to the combustion chamber, where it is fired in suspension. Pulverized-coal
 furnaces are classified as either dry or wet bottom, depending on the ash removal technique. Dry
 bottom furnaces fire coals with high ash fusion temperatures, and dry ash removal techniques are
 used.  In wet bottom (slag tap) furnaces, coal with a low ash fusion temperature is fired, and
 molten ash is drained from the bottom of the furnace.

        Pulverized-coal-fired furnaces are further classified by the firing position of the burners.
 Wall-fired boilers are characterized by rows of burners on one or more walls of the furnace. The
 two basic forms of wall-fired furnaces are single-wall (having burners on one wall) or opposed
 (having burners on walls that face each other). Circular register burners and cell burners are
 types of burner configurations used in both  single-wall and opposed-wall-fired units.  A circular
 register burner is a single burner mounted in the furnace wall, separated from other burners so
 that it has a separate, distinct flame zone. Cell burners are several circular register burners
 grouped closely together to concentrate their distinct flame zones.

        Tangential-fired boilers are based on the concept of a single flame envelope and project
 both fuel and combustion air from the corners of the furnace. The flames are directed on a line
 tangent to a small circle lying in a horizontal plane at the center of the furnace. This action
 produces a fireball that moves in a cyclonic motion and expands to fill the furnace.

 2.5.2  Cyclone Furnace

        Cyclone furnaces use burner design  and placement (i.e., several water-cooled horizontal
 burners) to produce high-temperature flames that circulate in a cyclonic pattern.  The coal is not
 pulverized but instead crushed to a 4-mesh  size. The crushed coal is fed tangentially, with
 primary air, to a horizontal cylindrical combustion chamber. In this chamber, small coal particles
 are burned in suspension, while the larger particles are forced against the outer wall.  The high
 temperatures developed in the relatively small furnace volume, combined with the low fusion
 temperature of the coal ash, causes the ash to form a molten slag, which is drained from the
 bottom of the furnace through a slag tap opening.

 2.5.5 Fluidized-bed Combustor

        Fluidized-bed combustion increasingly is being used for coal-fired electric utility power
 plants. A variety of coals, including those with high concentrations of ash, sulfur, and nitrogen,
 can be burned in a fluidized-bed combustor (FBC). The term "fluidized" refers to the state of the
 bed materials (fuel or fuel and inert material [or sorbent]) as gas passes through the bed. In a
 typical FBC, combustion occurs when coal, with inert material (e.g.,  sand, silica, alumina, or ash)
 and a sorbent such as limestone, is suspended through the action of primary combustion air
                                           2-18

-------
which is distributed below the combustor floor. The gas cushion between the solids allows the
particles to move freely, giving the bed a liquid-like characteristic (i.e., fluidized).  In an FBC,
crushed coal (between V* and 3/8 inches in diameter) is injected into a bed above a grate-like air
distributor. Air is injected upward through the grate, lifting and suspending the solid particles.
Inert materials such as sand or alumina are often mixed with the coal to maintain the bed in a
fluidized state.  Limestone particles can also be added to the bed to adsorb sulfur dioxide
produced during combustion (discussed in Chapter 3).

2.5.4 Stoker-fired Furnace

       Stoker-firing of coal is used for the oldest furnace designs in the electric utility industry,
being first introduced to the industry in the late 1800s.  Today, this design is used by only a few
of the operating power plants. New power plants are not expected to adopt this design. In stoker
furnaces, coal is burned on a bed at the bottom of the furnace.  The bed of coal burns on a grate.
Heated air passes upward through openings in the grate. Stokers are classified according to the
way coal is fed to the grate; the three general classes in use today are underfeed stokers, overfeed
stokers, and spreader stokers. Underfeed stokers feed coal by pushing it upward through the
bottom of the grate. In overfeed stokers, the coal is deposited directly on the grate from a
gravity-fed bin.  In spreader stokers, a flipping mechanism throws the coal into the furnace above
the grate; in this method, fine coal particles burn in suspension while heavier particles fall to the
grate and burn,  Additional combustion air is added above the grate to support suspension
burning. Overfeed stokers can bum every type of coal except caking bituminous coal; spreader
stokers can burn all types of coal except anthracite.

2.5.5 Gasified-coal-fired Combustor

       Unlike the four coal-firing configurations discussed above, IGCC power plants do not
burn solid coal.  In place of the coal-fired boiler used at a conventional coal-fired electric utility
power plant, at an IGCC power plant a coal gasification unit is used coupled with a gas turbine
combustor and heat recovery boiler. The solid coal  is gasified by a process in which a coal/water
slurry is reacted at high temperature and pressure with oxygen (or air) and steam in a vessel (the
gasifier) to produce a combustible gas. This combustible gas is composed of a mixture of carbon
dioxide and hydrogen and is often referred to as a synthetic gas or "syngas." Molten ash flows
out of the bottom of the gasifier into a water-filled sump where it forms a solid slag.  The syngas
is cleaned and conditioned before being bumed in a gas turbine that drives an electrical
generator. The hot combustion gases from the gas turbine are exhausted directly through a heat
recovery boiler (i.e., no combustion takes place in the boiler) to produce steam that is then
expanded through a steam turbine that  drives a second generator to produce more electrical
power.

       The generation of electricity using the IGCC process offers a number of advantages
compared to using conventional coal-fired boilers including higher thermal conversion
efficiencies (e.g., more kilowatt-hours  of electricity generated per kilogram of coal burned),
greater fuel flexibility (e.g., capability to use a wider variety of coal grades), and improved

                                          2-19

-------
control of participate matter and SC>2 emissions without the need for post-combustion control
devices (e.g., almost all of the sulfur and ash in the coal is removed during the gasification
process). Three IGCC power plant projects have been constructed in the United States as part of
the DOE's Clean Coal Technology Program, a joint government-industry cost-share technology
development program. These facilities are the 250 MWe Tampa Electric Company Polk Power
Project, the 307 MWe Wabash River Coal Gasification Repowering Project, and the 107 MWe
Sierra Pacific Pinon Pine IGCC Power Project. Two of the facilities currently are operating (the
Polk and Wabash River IGCC facilities). The Pinon Pine IGCC facility presently is shut down
because of recurring problems with particulate matter in the syngas causing premature gas
turbine blade erosion.13

       In IGCC applications, the syngas from the gasifier is cleaned and conditioned before it is
burned in the gas turbine using several different techniques. For example, at the Wabash River
IGCC facility, the syngas from the coal gasifier passes through a series of gas cleaning and
conditioning steps including a barrier filter for particulate removal, a water scrubber for gas
cooling, and an amine scrubber for removal of reduced-sulfur species. In contrast, at the Polk
IGCC facility, a hot-gas cleaning process is used and the syngas from the coal gasifier is not
cooled before it is burned in the gas turbine.
2.6 Ash from Coal Combustion

       Coal contains inorganic matter that does not burn including oxides of silicon, aluminum,
iron, and calcium. This noncombustible matter forms ash when the coal is burned. Burning of
coal in electric utility boilers generates large quantities of ash that must be removed and disposed
of.  The finer, lighter ash particles are entrained in the combustion gases and vented from the
furnace section with the flue gas. This portion of the coal ash is referred to as "fly ash." The
coarser, heavier ash particles fall to the bottom of the furnace section in the boiler unit. This
portion of the coal ash is referred to as  "bottom ash." The proportion of fly ash to bottom ash
generated in a coal combustion unit varies depending on how the coal is burned.

       In general, the fly ash is collected as a dry material at several points downstream of the
furnace section. These points  include collection hoppers beneath  the boiler economizer, air
heater, and the particulate matter control devices (other than wet scrubbers). From the collection
hopper, the fly ash is conveyed using a mechanical system, vacuum system,  pneumatic system, or
combination of these systems to a storage silo. If a wet scrubbing system is  used for air pollutant
control, fly ash is captured and removed in the scrubber wastewaters.

       For most boiler designs, the bottom ash is collected in a pit or hopper at the bottom of the
boiler furnace. The ash is collected in the form of either a dry material or a molten slag
depending on whether the furnace operating temperature is above the ash fusion temperature (i.e.,
the temperature at which the mineral compounds composing the ash melt). The ash is
continuously removed from the ash pit using a mechanical, pneumatic, or hydraulic conveyance
system.

                                          2-20

-------
       When coal is burned in a pulverized-coal furnace, on the order of 60 to 80 percent of the
total ash generated is fly ash. The high amount of fly ash results because the coal enters the
furnace in a fine powder form that bums rapidly in suspension resulting in many tiny, lightweight
ash particles that can easily be carried out of the furnace section with the flue gas. The heavier
ash particles fall to the bottom of the furnace where they are removed.  Two pulverized-coal
boiler design approaches are used to collect bottom ash. The more frequently used design
approach, commonly referred to as a "dry-bottom" furnace, collects the ash as essentially a dry
material. For the typical dry-bottom furnace, the ash and slag particles fall into a water-filled
hopper. The water serves several purposes including providing an air seal to prevent the
infiltration of ambient air into the furnace, solidifying molten slag particles, and facilitating ash
handling.  The ash is then continuously removed from the ash pit using either a mechanical or an
hydraulic conveyance system. The other design approach,  referred to as a "wet-bottom" furnace,
positions the coal burners on the furnace wall to maintain the ash that collects on the furnace
floor in a molten state.  The slag is drained through a slag tap opening into a slag tank.

       The cyclone furnace is specifically designed to burn low-ash fusion coals and retains most
of the ash in the form of a molten slag. The molten slag collects in a trough on the bottom of
furnace and is continually drained through a slag tap opening into a slag tank.  Water in the slag
tank solidifies the ash for disposal.  Only 20 to 30 percent of the ash produced  by burning coal in
a cyclone furnace is entrained as fly ash.

       By nature of the fluidized-bed combustion process, most of the ash in the coal leaves the
fluidized-bed combustor as fly ash.  Because the temperatures in the FBC remain below the ash
fusion temperature, formation of slag is avoided. Bottom ash is removed as a dry material to
maintain the fluidized bed at a constant  level. The ash removal system can be  either a
mechanical or pneumatic system.

       In stoker-fired furnaces where the coal is burned in a fuel bed, most of the ash remains on
the grate and is removed as bottom ash.  Some smaller ash particles are entrained in the upward
flow of combustion air through the grate and exit the furnace section as fly ash. The spreader
stoker has a greater proportion of the ash entrained as fly ash (up to 50  percent of the ash) than
the other stoker types (on the order of 20 percent fly ash).  This occurs  because the spreader
stoker mechanically throws the crushed coal across the  top of the grate. This allows the smaller
coal fines in the incoming coal to burn in suspension before falling to the grate. This produces
the small, lightweight ash particles that are carried out of the furnace section with the flue gas.

       No ash is produced when burning syngas derived from coal in an IGCC power plant.  The
ash contained in the coal is removed by the gasification process that is  used to  produce the
syngas. Before the syngas can be burned in the gas turbine, the gas must be precleaned to
remove all types of particulate matter in order to prevent premature wear and destruction of the
turbine blades.
                                          2-21

-------
2.7 Coals Burned by Electric Utilities In 1999

       The EPA ICR Part II survey collected data on the coal, coal wastes, and some
supplemental fuels burned in each coal-fired electric utility boiler operating in the United States
during the entire calendar year 1999.  Coal samples were analyzed for, at a minimum, the higher
heating value (HHV) and the coal sulfur, ash, Hg, moisture, and chlorine content. Samples were
collected every third to twelfth fuel shipment in each month of 1999, depending on the statistical
characteristics of initial analysis results for each boiler unit. Either the coal shipper or the power
plant operator could take the sample if the samples were collected at a point after any coal
cleaning had been completed.  Thus, "as-shipped" or "as-received" coals are considered to be
equivalent to "as-fired" coals,  and Hg analyses from such samples are assumed to represent the
quantity of Hg entering the boiler.

       In 1999, a nationwide total of approximately 786 million tons of coal and supplemental
fuels were burned in coal-fired electric utility boilers that met the CAA Section 112(a) definition
of an electric utility steam generating unit (i.e., boiler units of more than 25 megawatts that serve
a generator that produces electricity for sale). Table 2-5 shows the nationwide distribution of the
coal burned by rank as reported by the respondents to the EPA ICR (i.e., the power plant owners
and operators).

       Most electric utility power plants burn either bituminous or subbituminous coals. Half of
the coals burned by the electric utility industry in 1999 were bituminous coal (52 percent of the
total nationwide tonnage). Approximately one-third of the coals burned were subbituminous
coals (36.5 percent of the total nationwide tonnage).  Some power plants reported burning both
bituminous and subbituminous coals. At most of these facilities, the two coal types are blended
together before firing in the boiler unit.  A few of the facilities switch between the two coal types
for firing in the boiler unit to address site-specific circumstances. The vast majority of the
bituminous or subbituminous coals were supplied from mines in the United States. However,
imported coals were burned in 1999 at a few power plant locations. Ten plants, located near Gulf
of Mexico or Atlantic Ocean seaports, imported bituminous coal from South America and three
plants located in Hawaii and Florida imported subbituminous coal from Indonesia.

       In general, the burning of lignite or anthracite coals by electric utilities is limited to those
power plants that are located near the mines supplying the coal. Lignite accounted for
approximately 6.5 percent of the total coal tonnage burned at electric utility power plants in
1999. A total of 17 electric utility power plants reported burning lignite. All of these facilities
are located near the coal deposits from which the lignite is mined in Texas, Louisiana, Montana,
or North Dakota. Similarly, burning of anthracite coal in  1999 was limited to a few power plants
located close to the anthracite  coal mines in eastern Pennsylvania.  The coal-fired electric utility
boilers at these facilities burned  either newly mined anthracite coal or waste anthracite coal
reclaimed from mine waste piles.

       Table 2-5 also shows that small  amounts of supplemental fuels (e.g., petroleum coke or
tire derived fuel [TDF] chips)  also were co-fired with coal in some coal-fired electric utility
                                          2-22

-------
Table 2-5. Nationwide quantities of coals and supplemental fuels burned in
coal-fired electric utility boilers for the year 1999 as reported in the Part II EPA
ICR data (source: Reference 10).
Fuel Type
Bituminous coal
Subbituminous coal
Lignite
Bituminous/subbituminous coal mixture
Bituminous coal/petroleum coke mixture
Waste anthracite coal
Waste bituminous coal
Petroleum coke
Other (a)
Total
Total Tonnage
Burned
(million tons)
406
287
51
24
6
5
4
2
1
786
Percentage
by Weight
51.7%
36.5%
6.5%
3.0%
0.7%
0.6%
0.5%
0.3%
< 0.2%
100%
      (a) Mixes of anthracite, bituminous, and waste bituminous fuel, tires, Subbituminous coal and petroleum
         coke, or waste Subbituminous coal.
                                        2-23

-------
boilers. At these facilities, the supplemental fuels are mixed with coal before firing in the boiler
unit. These supplemental fuels typically have heating values higher than that of coal and serve to
boost the overall heating value of the ftiel mix burned in the boiler unit.  Less than 0.5 percent of
the total fuel tonnage burned in 1999 consisted of supplemental fuels.

       Selected properties of the coal and supplemental fuel burned nationwide in coal-fired
electric utility boilers in 1999, as reported in the EPA ICR Part II data, are summarized by fuel
type in Appendix A.  Table 2-6 presents a summary of the Hg content data reported for the coals
and supplemental fuels as fired in the boiler units. The EPA ICR data do not identify the coal
resource regions from which the coal burned in a given boiler  unit  was mined. However,
consistent with the Hg content data for as-mined coals presented in Table 2-2, the data presented
in Table 2-6 indicate that there is no general relationship between coal rank and Hg content of the
coal. For bituminous, subbituminous, and lignite coals, the Hg concentrations reported in the
EPA ICR data ranged from trace amounts to upper levels  of approximately 1 ppm.

       A review of the EPA ICR data suggests that there  is no direct correlation between the
sulfur content of a coal and its Hg content.  In other words, "high" sulfur coals are not necessarily
"high" Hg coals. Trace concentrations of Hg were reported for coals with high-sulfur contents.
Conversely, Hg concentrations at the upper end of the concentration ranges also were reported
for high sulfur-content coals. This observation is consistent with previous studies of the Hg
content in coal based on a much smaller database. For example, an earlier study comparing the
sulfur and Hg concentrations in 153 samples of coal shipments found no relationship between the
sulfur and Hg concentrations in these coals.14
                                          2-24

-------
U

i.*
5

§1
u r±
E
3
« BL
Eui
SI
0)
4— O)
§1
•s w
ss.
-I
  *
CM

^3 a
a 5 ™
** *^ "
c o>o
5i e ^
I'S s
" *fi Q-
o ^
= 1^
u. =.


c
o
•a
S?
u E
o 1
w &
O "— '
o>
X
,
i





O)
i

(
i





o
i
Number of
Samples
Analyzed

1
15
u.





s
CO'





CO
co
0
o
d
,
d





2
i
o
d
IV.
JM"

Bituminous coal






m'





8
CO
1
s
d
§
d




^1
d
i
o
d
o
K
CM
I
erican bituminous cc
c
1
£
5
w
in
|v.
in





9
8
d
0
d




o>
d
i
8
o
d
o
CO
CO

jbbituminous coal

10



o
CM-





cri
1
S

CO
o
d





d
S
d
CO
&
•s
in subbituminous co


1
c

§
O





IT)
i
^~
r
d




lf5
d
S
d
o

Lignite coal





8
^





S
1
s
iri
^
d




CO
CM
d
8
d
in

d




•^
in
d
o
d
1

iste anthracite coal

>



o>
i





CM
oj
CM
ri
cti
1?
d




CO

o
d
CM
h-

ste bituminous coal
CD
?



5!
£





in
-------
2.8 References
1.  American Society for Testing and Materials. 2000 Annual Book ofASTM Standards. West
   Conshohocken, PA. December 2000.

2.  Electric Power Research Institute. Evaluation of Methods for Analysis of Mercury and
   Chlorine in Coal. EPRI Report 1000287, Palo Alto, CA. September 2000.

3.  Wood, G. H., Jr., T.M. Kehn, M.D. Carter, and W.C. Culbertson.  Coal Resource
   Classification System of the U.S.  Geological Survey. U.S. Geological Survey Circular 891,
   1983.  Available at: < http://energv.er.usgs.gov/products/papers/C891 /index.htm >.

4.  U.S. Department of Energy, Energy Information Administration.  U.S. Coal Reserves: 1997
   Update. DOE/EIA-0529(97).  Office of Coal, Nuclear, Electric and Alternate Fuels, Office of
   Integrated Analysis and Forecasting, Washington, DC. February 1999.  Available at:
   < http://www.eia.doe,gov/cneaf/coal/reserves/front-l.html >.

5.  J.  Tully. Coal Resource Regions of the Conterminous United States. U.S. Geological Survey
   Open-File Report 96-279. July 6, 1996. Available at:
   < http://energv.er.usgs.gov/products/openfile/OF96-279/ >.

6.  Bragg, L.J., J.K. Oman, S.J. Tewalt, C.J.  Oman, N.H. Rega, P.M. Washington, and R.B.
   Finkelman. U.S. Geological Survey Coal Quality (COALQUAL) Database:  Version 2.0. U.S.
   Geological Survey Open-File Report 97-134. June 15, 2001. Available at:
   < http://energv.er.usgs.gov/products/databases/CoalQual/index.htni>.

7.  Toole-O'Neil, B., S.J. Tewalt, R.B. Finkleman, and R. Akers. "Mercury Concentration in
   Coal-Unraveling the Puzzle." Fuel, 78,47-54 (1999).

8.  Keating, M.H., K.R. Mahaffey, R. Schoeny, G.E. Rice, O.R. Bullock, R.B. Ambrose, Jr.,
   J. Swartout, and J.W. Nichols. Mercury Study Report to Congress, Volume II. EPA-452/R-
   97-004b. Office of Air Quality Planning  and Standards and Office of Research and
   Development, Research Triangle Park, NC. December 1997.  Available at:
   < http://www.epa.gov/airprogm/oar/mercury.html >.

9.  McDermott Technology, Inc.. Mercury Emission Results—Coal Content, Emissions and
   Control. Alliance, OH. Available at:
   <
   http://www.mtiresearch.com/aecdp/mercury.html#Coal%20Composition%20and%20Coal%2
   OCleaning >(accessed July 2001).
                                       2-26
-------
10.  U.S. Environmental Protection Agency.  Database of information collected in the Electric
   Utility Steam Generating Unit Mercury Emissions Information Collection Effort.  OMB
   Control No. 2060-0396.  Office of Air Quality Planning and Standards. Research Triangle
   Park,NC. April 2001. Available at:
   < http://www.epa.gov/ttn/atw/combust/utiltox/utoxpg.html >.

11.  Singer, J.G. (Ed.).  Combustion Fossil Power. Fourth Edition.  Combustion Engineering,
   Inc., Windsor, CT.  1991.

12. French, C.L., W.H. Maxwell, W.D. Peters, G.E. Rice, O.K. Bullock, A.B Vasu, R. Hetes,
   A. Colli, C. Nelson, and B.F. Lyons. Study of Hazardous Air Pollutant Emissions from
   Electric Utility Steam Generating Units — Final Report to Congress, Volume 1, EPA-453/R-
   98-004a.  Office of Air Quality Planning and Standards, Research Triangle Park, NC.
   February 1998. Available at: < http://www.epa.gov/ttn/atw/combust/utiltox/utoxpg.html >.

13. Cargill, P., G. DeJonghe, T. Howsley, B. Lawson, L. Leighton, and M. Woodward. Pinon
   Pine IGCC Project: Final Technical Report to the Department of Energy. DOE Award No.
   DE-FC21-92MC29309, Sierra Pacific Resources, Sparks, NV, January 2001. Available at:
   < http://www.laiil.gov/proiects/cctc/resources/pdfs/pinon/PinonFinalReport022201.pdf>.

14. Baker, S.S. EPRI Mercury in Coal Study; A Summary Report for Utilities That Submitted
   Samples Update. Prepared for EPRI Utility Air Regulatory Group by Systems Applications
   International Corporation, San Diego, CA. June  1994. pp. D-l to D-4.
                                       2-27
-------
                                       Chapter 3
                    Criteria Air Pollutant Emission Controls for
                          Coal-fired Electric Utility Boilers
3.1 Introduction

        The EPA uses "criteria pollutants" as indicators of ambient air quality.  For each criteria
 air pollutant, the EPA has established maximum concentrations for specific exposure periods
 above which adverse effects on human health may occur. Under authority of the CAA, these
 threshold concentrations for the criteria air pollutants are codified as the national ambient air
 quality standards (NAAQS). The EPA has set NAAQS for six criteria air pollutants: carbon
 monoxide (CO), lead (Pb), nitrogen dioxide (NO2), ozone (Os), particulate matter (PM), and
 sulfur dioxide (802).

        Estimates of national emissions for criteria air pollutants prepared by the EPA show that
 electric utility power plants that bum coal are significant emission sources of SO2, nitrogen
 oxides (NOX), and PM.1 Electric utility power plants are the Nation's largest source of SO2
 emissions, contributing approximately 68 percent of the estimated total national SO2 emissions in
 1998 (most recent year for which national estimates are available). Over 90 percent of these SO2
 emissions are coal-fired electric utility boilers.  Electric utilities contributed 25 percent of total
 national NOX emissions in 1998. Again coal combustion is the predominant source of NOX
 emissions from the electric utilities (almost 90 percent of the estimated NOX emissions).  Coal-
 fired electric utility power plants also are one of the largest industrial sources of PM emissions.
 In general, the high combustion efficiencies achieved by coal-fired electric utility boilers result in
 low emissions of CO and volatile organic compounds (a precursor for the photochemical
 formation of ozone in the  atmosphere). Lead is listed as a HAP in addition to being listed as a
 criteria air pollutant. Lead emissions from electric utility boilers were evaluated as part of EPA's
 report to Congress on HAP emissions from electric utility power plants (discussed in Section
 1.4.1).2 The EPA found that electric utility boilers contribute a very small percentage  of the
 nationwide Pb emissions.

        All coal-fired electric utility power plants in the United States  use control devices to
 reduce PM emissions.  Many coal-fired electric utility boilers also are required to use controls for
 SO2 and NOX emissions depending on site-specific factors such as the properties of the coal
 burned, when the power plant was built, and the area where the power plant is located. As
 discussed in Chapter 6, certain control technologies used to reduce criteria air pollutant
                                         3-1
-------
emissions from coal-fired electric utility boilers also remove some of the mercury (Hg) from the
flue gas. In addition, the existing control configuration used for a given coal-fired electric utility
boiler to meet criteria air pollutant emissions standards directly can affect the applicability,
performance, and costs of retrofitting additional Hg controls to the unit.

       The purpose of this chapter is to present a summary review of the different control
technologies currently used by coal-fired electric utility boilers to meet the applicable criteria air
pollutant emissions standards. The nationwide distribution of control configurations used at
coal-fired electric utility power plants to comply with these standards is presented using
information from the EPA ICR database. The impact or influence of these control configurations
on control of Hg emissions  is discussed in the Chapter 6.
3.2 Criteria Air Pollutants of Concern from Coal Combustion

3.2J  Particulate Matter3'4

       Dust, dirt, soot, smoke, and liquid droplets are directly emitted into the air from
anthropogenic sources as well as natural sources such as forest fires and windblown dust.  This
type of PM sometimes is called "primary particulate matter."  In addition, gaseous air pollutants
(e.g., sulfur dioxide, nitrogen oxides, and volatile organic compounds) are considered to be PM
precursors causing "secondary particulate matter" through complex transformations that occur in
the ambient environment. Human exposure to concentrations of PM at various levels results in
effects on breathing and respiratory symptoms, aggravation of existing respiratory and
cardiovascular disease, alterations in the body's defense systems against foreign materials,
damage to lung tissue, carcinogenesis, and premature death. The people most sensitive to the
effects of PM include individuals with chronic obstructive pulmonary or cardiovascular disease
or influenza, asthmatics, the elderly, and children.  Particulate matter also contributes to visibility
impairment in the United States.

       Primary PM emissions from coal-fired electric utility boilers consist primarily of fly ash.
Ash is the unburned carbon char  and the mineral portion of combusted coal.  The amount of ash
in the coal, which ultimately exits the boiler unit as fly ash, is a complex function of the coal
properties, furnace-firing configuration, and boiler operation.  For the dry-bottom, pulverized-
coal-fired boilers, approximately 80 percent of the total ash in the as-fired coal will exit the boiler
as fly ash. Wet-bottom, pulverized-coal-fired boilers emit significantly less fly ash: on the order
of 50 percent of the total ash exits the boiler as fly ash. In a cyclone furnace boiler, most of the
ash is retained as liquid slag; thus, the quantity of fly ash exiting the boiler is typically 20 to 30
percent of the total ash.  However, the high operating temperatures unique to these designs may
also promote ash vaporization and larger fractions of submicron fly ash compared to dry bottom
designs.  Fluidized-bed combustors emit high levels of fly ash since the coal is fired in
suspension and the ash is present in dry form. Spreader-stoker-fired boilers can also emit high
levels of fly ash. However, overfeed and underfeed stokers emit less fly ash than spreader
stokers, since combustion takes place in a relatively quiescent fuel bed.
                                          3-2
-------
       In addition to the fly ash, PM emissions from a coal-fired electric utility power plant
result from reactions of the SC>2 and NOX compounds as well as unburned carbon particles carried
in the flue gas from the boiler. The SO2 and NO* compounds are initially in the vapor phase
following coal combustion in the furnace chamber but can partially chemically transform in the
stack, or near plume, to form fine PM in the form of nitrates, sulfur trioxide (803), and sulfates.
Firing configuration and boiler operation can affect the fraction of carbon (from unburned coal)
contained in the fly  ash. In general, the high combustion efficiencies achieved by pulverized-
coal-fired boilers and cyclone-fired boilers result in relatively small amounts of unburned carbon
particles in the exiting combustion gases.  Those pulverized-coal-fired electric utility boilers that
use special burners for NOX control (discussed in Section 3.7) tend to burn coal less completely;
consequently, these  furnaces tend to emit a higher fraction of unbumed carbon in the combustion
gases exiting the furnace.

       Another potential source of PM in the flue gas from a coal-fired electric utility boiler is
the use of a dry sorbent-based control technology. Solid sorbent particles are injected into the
combustion gases to react with the air pollutants and then recaptured by a downstream control
device.  Sorbent particles that escape capture by the control device are emitted as PM to the
atmosphere. Control technologies using sorbent injection are discussed in Chapter 7.
3.2.2 Sulfur Dioxide
                    3,4
       Exposure of people to SC>2 concentrations above threshold levels affects their breathing
and may aggravate existing respiratory and cardiovascular disease. Sensitive populations include
asthmatics, individuals with bronchitis or emphysema, children, and the elderly. Sulfur dioxide
is also a primary contributor to acid deposition, or acid rain, which causes acidification of lakes
and streams and can damage trees, crops, historic buildings, and statues.  In addition, SOX
compounds in the air contribute to visibility impairment. In the United States, SC>2 is primarily
emitted from the combustion of fossil fuels and by metallurgical processes.

       Coal deposits contain sulfur in amounts ranging from trace quantities to as high as
eight percent or more.  Most of this sulfur is present as either pyritic sulfur (sulfur combined with
iron in the form of a mineral that occurs in the coal deposit) or organic sulfur (sulfur combined
directly in the coal structure). During combustion, sulfur compounds in coal are oxidized to
gaseous SO2 or SOs.  When firing bituminous coal, almost all of the sulfur present in coal will be
emitted as gaseous sulfur oxides (on average 98 percent). The  more alkaline nature of ash in
some subbituminous coals causes a portion of the sulfur in the  coal to react to form various
sulfate salts; these salts are emitted as fly ash or retained in the boiler bottom ash.  Generally, the
percentage of sulfur in the as-fired coal that is converted to sulfur oxides during combustion does
not vary with the utility boiler design or operation.

3.2.3 Nitrogen Oxides 4'5

       Nitrogen dioxide (NOj) is a highly reactive gas.  The major mechanism for the formation
of NO2 in the atmosphere is the oxidation of nitric oxide (NO)  when exposed to solar radiation.

                                         3-3
-------
These two chemical species are collectively referred to as nitrogen oxides (NO*). Exposure of
people to NC>2 can irritate the lungs, cause bronchitis and pneumonia, and lower resistance to
respiratory infections.  Nitrogen oxides are an important precursor together with volatile organic
compounds in the photochemical formation of ozone in the atmosphere. Ozone is a criteria
pollutant and the major component of smog.  Nitrogen dioxide is also a primary contributor to
acid rain. The major NOX emissions sources are transportation vehicles and stationary
combustion units.

       Both NO and NO2 are formed during coal combustion by oxidation of molecular nitrogen
that is present in the combustion air or nitrogen compounds contained in the coal. Overall, total
NOX formed during combustion is composed predominantly of NO mixed with small quantities
of NO2 (typically less than 10 percent of the total NOX formed).  However, once NO formed
during coal combustion is emitted to the atmosphere, the NO is oxidized to NO2.

       The NOX formed during coal combustion by oxidation of molecular nitrogen (N^) in the
combustion air is referred to as "thermal NOX." The oxidation reactions converting N2 to NO and
NO2 become very rapid once gas temperatures rise above 1,700 °C (3,100  °F).  Formation of
thermal NOX in a coal-fired electric utility boiler is dependent on two conditions occurring
simultaneously in the combustion zone: high temperature and an excess of combustion air. A
boiler design feature or operating practice that increases the gas temperature above 1,700 °C, the
gas residence time at these temperatures, and the quantity of excess combustion air will affect
thermal NOX formation.  The formation of NOX by oxidation of nitrogen compounds contained in
the coal is referred to as "fuel NOX." The nitrogen content in most coals ranges from
approximately 0.5 to 2 percent. The amount of nitrogen available  in the coal is relatively small
compared with the amount of nitrogen available in the combustion air.  However, depending on
the combustion conditions, significant quantities of fuel NOX can be formed during coal
combustion.
3.3 Existing Control Strategies Used for Coal-fired Electric Utility Boilers

       Electric utilities must comply with applicable Federal standards and programs that
specifically regulate criteria air emissions from coal-fired electric utility boilers. These
regulations and programs include New Source Performance Standards (NSPS), the CAA Title IV
Acid Rain Program, and the CAA Title V Operating Permits Program.  The EPA has delegated
authority to individual state and local agencies for implementing many of these regulatory
requirements.  In addition, individual states have established their own standards and
requirements for those power plants that operate within their jurisdictions. Electric utility
companies use one or a combination of the following three control strategies to comply with the
specific set of requirements applicable to a given coal-fired boiler.

       Pre-combustion Controls. Control measures in which fuel substitutions are made or fuel
       pre-processing is performed to reduce pollutant formation in the combustion unit.
                                         3-4
-------
       Combustion Controls.  Control measures in which operating and equipment
       modifications are made to reduce the amount of pollutants formed during the combustion
       process; or in which a material is introduced into the combustion unit along with the fuel
       to capture the pollutants formed before the combustion gases exit the unit.

       Post-combustion Controls: Control measures in which one or more air pollution control
       devices are used at a point downstream of the furnace combustion zone to remove the
       pollutants from the post-combustion gases.

       Table 3-1  shows the distribution of emissions control strategies for PM, SCh, and NOX
used for coal-fired electric utility boilers in 1999 as reported in the Part n EPA ICR data.6 All
coal-fired electric utility boilers in the United  States are controlled for PM emissions by using
some type of post-combustion controls. These particulate emission control types are discussed in
Section 3.4. Approximately two-thirds of the  total coal-fired electric utility boilers use add-on
controls for SO2 emissions. Most of these controlled units use either a pre-combustion or a post-
combustion control strategy for SO2 emissions. The  methods used for controlling SC»2 emissions
from coal-fired electric utility boilers are discussed in Section 3.5.  Although approximately two-
thirds of the coal-fired electric utility boilers are controlled for NOX emissions, these units are not
necessarily the same units controlled for SO2 emissions. The predominant strategy for
controlling NOX emissions is to use combustion controls.  Section 3.6 discusses the application of
NOX emission controls to coal-fired electric utility boilers.
3.4 Particulate Matter Emission Controls

       Four types of control devices are used to collect PM emissions from coal-fired electric
utility boilers: electrostatic precipitators, fabric filters, mechanical collectors, and particle
scrubbers. Table  3-2 presents the 1999 nationwide distribution of PM controls on coal-fired
electric utility boilers by total number of units and by percentage of nationwide electricity
generating capacity.  Electrostatic precipitators are the predominant control type used on coal-
fired electric utility boilers both in terms of number of units (84 percent) and total generating
capacity (87 percent). The second most common control device type used is a fabric filter.
Fabric filters are used on about 14 percent of the coal-fired electric utility boilers. Particle
scrubbers are used on approximately three percent of the boilers.  The least used control device
type is a mechanical collector.  Less than one percent of the coal-fired electric utility boilers use
this type of control device as the sole PM control.  Other boilers equipped with a mechanical
collector use this  control device in combination with one of the other PM control device types.

 3.4.1 Electrostatic Precipitators 4'7

       Electrostatic precipitator (ESP) control devices have been used to control PM emissions
for over 80 years. These devices can be designed to achieve high PM collection efficiencies
(greater than 99 percent), but at the cost of increased unit size. An ESP operates  by imparting an
electrical charge to incoming particles, and then attracting the particles to oppositely charged
                                          3-5

-------
Table 3-1.  Criteria air pollutant emission control strategies as applied to
coal-fired electric utility boilers in the United States for the year 1999 as reported
in the Part II EPA ICR data (source:  Reference 6).


Criteria
Air Pollutant

Participate
matter
Sulfur
dioxide
Nitrogen
oxides
Percentage of Coal-fired Electric Utility Boilers Using Control Strategy
as Reported in Phase II EPA ICR Data "
Meet Applicable
Standards
Without
Additional
Controls
0%

37%
40%

Pre-combustion
Controls

0%

40%
0%

Combustion
Controls

0%

3%
57%

Post-combustion
Controls

100%

20%
3%
    (a) Approximately 1.5 % of the boilers use a combination of pre-combustion and post-combustion SO2 controls.
    (b) Approximately 1% of the boilers using post-combustion NO, controls also use some type of combustion
       controls.
                                         3-6
-------
Table 3-2.  Nationwide distribution of existing PM emission controls used for
coal-fired electric utility boilers for the year 1999 as reported in the Part II EPA
ICR data (source: Reference 6).
PM
Control Type
Electrostatic precipitator
(Cold-side)
Electrostatic precipitator
(Hot-side)
Fabric filter
Particle scrubber
Mechanical collector (d)
Multiple control device
combinations (e)
Abbreviation
Code
CS- ESP
HS-ESP
FF
PS
MC

Nationwide Total
Phase II EPA ICR Data
Number
of Boilers
822 (a)
122
155(b)
23(c)
5
13
1,140(f)
Percent of
Nationwide
Total Number
of Units
72.1 %
10.8 %
13.6%
2.0%
0.4 %
1.1 %
100%
Percent of
Nationwide
Electricity
Generating
Capacity
74.7 %
11.3%
9.4 %
3.0 %
0.2%
1.4%
100 %
   (a) Includes 10 boilers with cold-side ESP in combination with upstream mechanical collector.
   (b) Includes eight boilers with baghouse in combination with upstream mechanical collector.
   (c) Includes two boilers with particle scrubber in combination with upstream mechanical collector.
   (d) Boilers using mechanical collector as only PM control device.
   (e) Boilers using a combination of two or more different control device types other than mechanical
      collectors. Includes two boilers that use a hot-side ESP in series with a cold-side ESP.
   (f) Does not include the three IQCC units.
                                            3-7
-------
metal plates for collection.  Periodically, the particles collected on the plates are dislodged in
sheets or agglomerates (by rapping the plates) and fall into a collection hopper. The dust
collected in the ESP hopper is a solid waste that must be disposed of.

       The effectiveness of particle capture in an ESP depends largely on the electrical resistivity
of the particles being collected.  An optimum value exists for a given ash. Above and below this
value, particles become less effectively charged and collected. Table 3-3 presents the PM
collection efficiency of an ESP compared with the other control device types. Coal that contains
a moderate to high amount of sulfur (more than approximately three percent) produces an easily
collected fly ash.  Low-sulfur coal produces a high-resistivity fly ash that is more difficult to
collect.  Resistivity of the fly ash can be changed by operating the boiler at a different
temperature or by conditioning the particles upstream of the ESP with sulfur trioxide, sulfuric
acid, water, sodium, or ammonia. In addition, collection efficiency is not uniform for all particle
sizes. For coal fly ash, particles larger than about 1 to 8 urn and smaller than about 0.3 fim (as
opposed to total PM) are typically collected with efficiencies from 95 to 99.9 percent. Particles
near the 0.3 |im size are in a poor charging region that reduces collection efficiency to 80 to 95
percent.

       An ESP can be used at one of two locations in a coal-fired electric utility boiler system.
For many years, every ESP was installed downstream of the air heater where the temperature of
the flue gas is between 130 and 180 °C (270 and 350 °F).  An ESP installed at this location is
referred is as a "cold-side" ESP.  However, to meet SO2 emission requirements, many electric
utilities switched to burning low-sulfur coal (discussed in  the Section 3.5.1). These coals have
higher electrical ash resistivities, making the fly ash more difficult to capture downstream of the
air heater.  Therefore, to take advantage of the lower fly-ash resistivities at higher temperatures,
some ESPs are installed upstream of the air heater, where the temperature of the flue gas is in the
range of 315 to 400 °C (600 to 750 °F). An ESP installed upstream of the air heater is referred to
as a "hot-side" ESP.

3.4.2 Fabric Filters4'8

       Fabric  filters (FF) have been used for fly ash control from coal-fired electric utility boilers
for about 30 years.  This type of control device collects fly ash in the combustion gas stream by
passing the gases through a porous fabric material.  The buildup of solid particles on the fabric
surface forms a thin, porous layer of solids or a filter, which further acts as a filtration medium.
Gases pass through this cake/fabric filter, but the fly ash is trapped on the cake surface. The
fabric material used is typically fabricated in the shape of long, cylindrical bags.  Hence, fabric
filters also are frequently referred to as "baghouses."

       Gas flow through a FF becomes excessively restricted if the filter cake on the bags
becomes too thick.  Therefore, the dust collected on the bags must be removed periodically. The
type of mechanism used to remove the filter cake classifies FF design types. Depending on  the
FF design type, the dust particles will be collected either on the inside or outside of the bag. For
designs in which the dust is collected on the inside of the bags, the dust is removed by either
                                          3-8
-------
Table 3-3. Comparison of PM collection efficiencies for different PM control
device types (source: Reference 4)
PM
Control Type
Electrostatic precipitator
(Cold-side)
Electrostatic precipitator
(Hot-side)
Fabric filter
Particle scrubber
Mechanical collector
Representative PM
Mass Collection Efficiency Range
Total
PM
99 to 99.7 %
99 to 99.7 %
99 to 99.9 %
95 to 99%
70 to 90 %
PM
less than 0.3 \an
80 to 95 %
80 to 95 %
99 to 99.8%
30 to 85 %
Oto15%
                                   3-9
-------
mechanically shaking the bag (called a "shaker type" FF) or by blowing air through the bag from
the opposite side (called a "reverse-air" FF). An alternate design mounts the bags over internal
frame structures, called "cages" to allow collection of the dust on the outside of the bags. A
pulsed jet of compressed air is used to cause a sudden stretching then contraction of the bag
fabric dislodging the filter cake from the bag.  This design  is referred to as a "pulse-jet" FF. The
dislodged dust particles fall into a hopper at the bottom of the baghouse. The dust collected in
the hopper is a solid waste that must be disposed of.

       An FF must be designed and operated carefully to ensure that the bags inside the collector
are not damaged or destroyed by adverse operating conditions.  The fabric material must be
compatible with the gas stream temperatures and chemical composition. Because of the
temperature limitations of the available bag fabrics, location of an FF for use in a coal-fired
electric utility boiler is restricted to downstream of the air heater. In general, fabric filtration is
the best commercially available PM  control technology for high-efficiency collection of small
particles (see Table 3-3).

       Electrostatic stimulation of fabric filtration (ESFF) involves a modified fabric filter that
uses electrostatic charging of incoming dust particles to increase collection efficiency and reduce
pressure drop compared  to fabric filters without charging.  Filter bags are specially made to
include wires or conductive threads, which produce an electrical field parallel to the fabric
surface. Conductors can also be placed as a single wire in the center of the bag.  When the bags
are mounted in the baghouse, the conductors are attached to a wiring harness that supplies
electricity.  As particles enter the field and are charged, they form a porous mass or cake of
agglomerates at the fabric surface. Greater porosity of the cake reduces pressure drop, while the
agglomeration increases efficiency of small particle collection.  Cleaning is required less
frequently, resulting in longer bag life. For felted or nonwoven bags, the field promotes
collection on the outer surface of the fabric, which also promotes longer bag life.  Filtration
velocity can be increased so that less fabric area is required in the baghouse. The amount of
reduction is based on an economic balance among desired performance, capital cost, and
operating costs. A number of variations exist on the ESFF idea of combining particle charging
with fabric filtration.

       The University of North Dakota, Energy and Environmental Research Center
(UND/EERC) has developed another type of combined control device called the Advanced
Hybrid Collector (AHC).9 A charging (and collection) section  can also be placed ahead of the
bags in a fabric filter. This  approach is used in the AHC along with the use of membrane fabrics
(woven or felted fabrics having a membrane laminated to the filtration surface of the fabric).
The membrane is typically polytetrafluoroethylene (PTFE). With about 90 percent of the mass of
particles collected in the electrostatic charging and collection section of the AHC, the load on the
fabric filter part of the system is much reduced. With a membrane fabric for the bags, it is likely
that filtration velocity can be increased significantly.
                                         3-10
-------
3.4.3 Particle Scrubbers4

       Particle scrubbers operate by shattering streams of water into small droplets that collide
with and trap solid particles contained in the flue gas or by forcing the gases into intimate contact
with water films. The particle-laden droplets or water films coalesce and are collected in a sump
at the bottom of the scrubber. The three basic types of particle scrubbers are venturi scrubbers,
preformed spray scrubbers, and moving-bed scrubbers. Venturi scrubbers are the type most
commonly used for coal-fired electric utility boilers. This scrubber design transports the particle-
laden flue gas through a constriction where violent mixing takes place. Water is introduced
either at or upstream of the constriction. Preformed spray scrubbers are usually vertical cylinders
with flue gas passing upward through droplets sprayed from nozzles near the top of the unit.
Moving-bed scrubbers have an upper chamber in which a bed of low-density spheres (often
plastic) is irrigated by streams of water from above. Gas passing upward through the bed agitates
the wetted spheres, which continually expose fresh liquid surfaces for particle transfer.
Regardless of the scrubber design, all particle scrubber systems generate wastewaters from the
scrubber blowdown that must be treated and discharged.

       Particle scrubbers are more sensitive to particle size distribution in the flue gas than either
an ESP or an FF.  In general, particle scrubbers are not as effective as these other control devices
at collecting small particles (see Table 3-3). Also, while  a venturi particle scrubber will have a
lower initial cost for a given boiler unit application than either an  ESP or an FF, the high pressure
drop required for the scrubber to achieve a high collection efficiency results in high operating
costs. These factors, in large part, account for the low use of particle scrubbers at coal-fired
utilities.

3.4.4 Mechanical Collectors4

       Mechanical collectors are the oldest, simplest, and least efficient of the four types of PM
control devices. The collectors used for utility boilers are generally in the form of groups of
cylinders with conical bottoms (multicyclones). Flue gas entering the cylinder tangentially to the
wall is imparted with a circular motion around the cylinder's axis. Particles in the gas stream are
forced toward the wall by centrifugal force,  then downward through a discharge at the bottom of
the cone.  Collection efficiency for a typical multicyclone can be about 70 to 75 percent for
10-um particles, but can drop to less than 20 percent for smaller 1-p.m particles. Mechanical
collectors can be efficient for relatively large particles because their settling velocity is high
compared to fine particles.  In a cyclone, larger particles are forced through the gas stream
towards the outer wall because of their mass and inertia, while small particles have insufficient
mass to be much affected. Electrically charging particles tends to agglomerate them, especially
small particles, with the resulting larger agglomerates having increased mass over the individual
small particles. In charged mechanical collectors, a charging section is placed ahead of a
mechanical collector, and collection efficiency for smaller particles is significantly  increased.
                                         3-11
-------
3.5 SO2 Emission Controls

       Sulfur dioxide emissions from most coal-fired electric utility boilers are controlled using
either of two basic approaches. The first approach is to use pre-combustion measures, namely,
the firing coal that contains lower amounts of sulfur. The low-sulfur coal may be naturally
occurring or the result of coal cleaning.  The other approach is to remove the sulfur compounds
from the flue gas before the gas is discharged to the atmosphere. These post-combustion
processes are collectively called "flue gas desulfurization" or "FGD" systems. All FGD systems
can be further classified as wet or dry flue gas scrubbing systems. A third control approach
available for those coal-fired electric utility boilers using a fluidized-bed combustor is to burn the
coal together with limestone. An FBC can be characterized as a boiler type with inherently lower
SC*2 emissions.  In this report, however,  combustion of coal in fluidized-bed with limestone is
also considered to be an SO2 combustion control method.  The SOz control approaches include a
number of different technology subcategories that are now commercially used in the United
States, Europe, or Pacific Rim countries.

       Table 3-4 presents the 1999 nationwide distribution of SC>2 controls used for coal-fired
electric utility boilers by total number of units and by percentage of nationwide electricity
generating capacity. For approximately  one-third of the boilers, no SOi controls were reported in
the Part II EPA ICR data. The other two-thirds of the units reported using some type of control
to meet the SC>2 emission standards applicable to the unit. Pre-combustion control by burning a
low-sulfur content coal was reported for approximately 40 percent of the boilers.  Post-
combustion control devices for 862 removal are used for approximately 20 percent of the boilers.
Wet FGD systems are the most commonly used post-combustion control technique. The newer
technologies of spray dryer systems or dry injection are limited in their application to existing
units. The remaining 3 percent of the boilers use fluidized-bed combustion with limestone.

3.5.1  Low-sulfur Coal

       A coal with sufficiently low sulfur content that when burned in the boiler meets the
applicable SOo emission standards without the use of additional controls is sometimes referred to
as "compliance coal." Coals naturally low in sulfur content may be  mined directly from the
ground. Alternatively, the sulfur content of coal fired in the boiler may be lowered first by
cleaning the coal or blending coals obtained from several sources. However, burning low-sulfur
coal may not be a technically feasible or economically  practical 862 control alternative for all
boilers. In some cases, a coal with the required sulfur content to meet  the applicable standard
may not be available or cannot be fired satisfactorily in a given boiler unit design.  Even if such a
coal is available, use of the low-sulfur coal that must be transported long distances from the mine
may not be cost-competitive with burning higher sulfur coal supplied by closer mines and using a
post-combustion control device.

       Various coal cleaning processes  may be used to reduce the sulfur content of the coal. A
significant portion of the pyritic sulfur minerals mixed with the mined coal can usually be
                                        3-12
-------
Table 3-4.  Nationwide distribution of existing SO2 emissions controls used for
coal-fired electric utility boilers for the year 1999 as reported in the Part II EPA
ICR data (source: Reference 6).
SO, Control Type
Burn low-sulfur coal
("compliance coal")
Wet FGD system
Spray dryer system
Fluldized-bed coal combustion
with limestone (a)
Dry injection
No controls reported (d)
Abbreviation
Code
LSC
FGD
SDA
FBC
Dl

Nationwide Total
Phase II EPA ICR Data
Number
of Boilers
455
173 (a)
52 (b)
37 (c)
2
421
1,140(6}
Percent of
Nationwide
Total Number
of Units
39.9 %
15.2%
4.6%
3.2%
0.2 %
36.9 %
100%
Percent of
Nationwide
Electricity
Generating
Capacity
38.2%
23.8 %
3.4 %
1.1 %
< 0.1 %
33.5%
100%
   (a) Includes one FBC boiler unit using a wet FGD system.
   (b) Includes three FBC boilers using spray dryer systems.
   (c) FBC boilers using no downstream post-combustion SO, controls.
   (d) Entry in ICR response indicated none or was left blank.
   (e) Does not include the three IGCC units.

                                        3-13

-------
removed by physical gravity separation or surface property (flotation) methods.  However,
physical coal cleaning methods are not effective for removing the organic sulfur bound in coal.
Another method of reducing the overall sulfur content of the coal burned in a given boiler unit is
to blend coals with different sulfur contents to meet a desired or target sulfur level.

3.5.2 Fluidized-bed Combustion with Limestone

       One of the features of FBC boilers is the capability to control SC>2 emissions during the
combustion process. This is accomplished by adding finely crushed limestone to the fluidized
bed. During combustion, calcination of the limestone (reduction to lime by subjecting to heat)
occurs simultaneously with the oxidation of sulfur in the coal to form SO2.  The SC>2, in the
presence of excess oxygen, reacts with the lime particles to form calcium sulfate. The sulfated
lime particles are removed with the bottom ash or collected with the fly ash by a downstream PM
control device.  Fresh limestone is continuously fed to the bed to replace the reacted limestone.

3.5.3 Wet FGD Systems

       The SO2 in flue gas can be removed by reacting the sulfur compounds  with a solution of
water and an alkaline chemical to form insoluble salts that are removed in the  scrubber effluent.
These processes are called "wet FGD systems" in this report. Most wet FGD systems for control
of SO2 emissions from coal-fired electric utility boilers are based on using either limestone or
lime as the alkaline source.  At some of these facilities, fly ash is mixed with the limestone or
lime. Several other scrubber system designs (e.g., sodium carbonate, magnesium oxide, dual
alkali) are also used by a small percentage of the total number of boilers.

       The basic wet limestone scrubbing process is simple and is the type most widely used for
control of SO2 emissions from coal-fired  electric utility boilers. Limestone sorbent is
inexpensive and generally locally available throughout the United States. In a wet limestone
scrubber, the flue gas containing SC>2 is brought into contact with a limestone/water slurry.  The
SO2 is absorbed into the slurry and reacts with limestone to form an insoluble  sludge.  The
sludge, mostly calcium sulfite hemihydrate and gypsum, is disposed of in a pond specifically
constructed for the purpose or is recovered as a salable byproduct.

       The wet lime scrubber operates in a similar manner to the wet limestone scrubber. In a
wet lime scrubber, flue gas containing SC>2 is contacted with a hydrated lime/water slurry; the
SO2 is absorbed into the slurry and reacts with hydrated lime to form an insoluble sludge. The
hydrated lime provides greater alkalinity (higher pH) and reactivity than limestone.  However,
lime-scrubbing processes require appropriate disposal of large quantities of waste sludge.

       The SO2 removal efficiencies of existing wet limestone scrubbers range from 31 to
97 percent, with an average of 78 percent. The SO2 removal efficiencies of existing wet lime
scrubbers range from 30 to 95 percent. For both types of wet scrubbers, operating parameters
affecting SO2 removal efficiency include  liquid-to-gas ratio, pH of the scrubbing medium, and
the  ratio of calcium  sorbent to SOa. Periodic maintenance is needed because of scaling, erosion,
                                        3-14
-------
and plugging problems.  Recent advancements include the use of additives or design changes to
promote SC«2 absorption or to reduce scaling and precipitation problems.

3,5.4 Spray Dryer Adsorber

       A spray dryer adsorber (sometimes referred to as wet-dry or semi-dry scrubbers) operates
by the same principle as wet lime scrubbing, except that the flue gas is contacted with a fine mist
of lime slurry instead of a bulk liquid (as in wet scrubbing). For the spray dryer absorber process,
the combustion gas containing SOi is contacted with fine spray droplets of hydrated lime slurry
in a spray dryer vessel. This vessel is located downstream of the air heater outlet where the gas
temperatures are in the range of 120 to 180 °C (250 to 350 °F). The S02 is absorbed  in the slurry
and reacts with the hydrated lime reagent to form solid calcium sulfite and calcium sulfate as in a
wet lime scrubber. The water is evaporated by the hot flue gas and forms dry, solid particles
containing the reacted sulfur.  These particles are entrained in the flue gas, along with fly ash,
and are collected in a PM collection device. Most of the SC>2 removal occurs in the spray dryer
vessel itself, although some additional SOz capture has also been observed in downstream
particulate collection devices, especially fabric filters. This process produces dry reaction waste
products for easy disposal.

      The primary operating parameters affecting SO2 removal are the calcium-reagent-to-
sulfur stoichiometric ratio and the approach to saturation in the spray dryer.  To increase overall
sorbent use, the solids collected in the spray dryer and the PM collection device may be recycled.
The SO2 removal efficiencies  of existing lime spray dryer systems range from 60 to 95 percent.

3.5.5 Dry Injection

      For the dry injection process, dry powdered lime (or another suitable sorbent) is directly
injected into the ductwork upstream of a PM control device.  Some systems use spray
humidification followed by dry injection. This dry process eliminates the slurry production and
handling equipment required for wet scrubbers and spray dryers, and produces dry reaction waste
products for easier disposal. The SC>2 is adsorbed and reacts with the powdered sorbent. The dry
solids are entrained in the combustion gas stream, along with fly ash, and then collected by the
PM control device. The SOa removal efficiencies of existing dry injection systems range from
40 to 60 percent.

3.5.6 Circulating Fluidized-bed A dsorber

      In the circulating fluidized-bed adsorber (CFBA), the flue gas flows upward through a
bed of sorbent particles to produce a fluid-like condition in the bed. This condition is obtained
by adjusting gas flow rate sufficiently to support the particles, but not carry them out of the
system.  Characteristics of the bed are high heat and mass transfer, because of high mixing rates,
and particle-to-gas contact.  These conditions allow the CFBA's bed of sorbent particles to
remove a sorbate from the gas stream with high effectiveness. In a CFBA, material is withdrawn
from the bed for treatment (such as desorption) then re-injected into the bed. Currently, CFBAs
                                         3-15
-------
are used with limestone and ash as sorbents for SC»2 control, but they also have the capability to
remove Hg from the flue gas. The SO2 removal ranges for CFBAs from 80 to 98 percent.
3.6 NOj Emission Controls

       Control techniques used to reduce NOX formation include combustion and post-
combustion control measures. Combustion measures consist of operating and equipment
modifications that reduce the peak temperature and excess air in the furnace. Post-combustion
control involves converting the NOX in the flue gas to molecular nitrogen and water using either a
process that requires a catalyst (selective catalytic reduction) or a process that does not use a
catalyst (selective noncatalytic reduction).

       Table 3-5 presents the 1999 nationwide distribution of NOX controls used for coal-fired
electric utility boilers by total number of units and by percentage of nationwide electricity
generating capacity. Approximately one-third of the boilers do not use additional NOX controls.
The other two-thirds of the units use additional controls to meet the applicable NOX standards.
The predominant control NOX strategy is to use one or more combustion control techniques.
Post-combustion NOX reduction technologies (both catalytic and noncatalytic) accounted for only
a small percentage of the NOX emission controls used in 1999 (approximately three percent of the
total units).  However, a number of electric utilities are considering the addition of these types of
controls to their coal-fired boilers to comply with new NOX emission control requirements.

3.6.1 Combustion Controls

       A variety of combustion control practices can be used including low NOX burners,
overfire air, off-stoichiometric firing, selective or biased burner firing, reburning, and
burners-out-of-service.  Control of NOx also can be achieved through staged combustion (also
called air staging).  With staged combustion, the primary combustion zone is fired with most of
the air needed for complete combustion of the coal. The remaining air needed is introduced into
the products of the partial combustion in a second combustion zone.  Air staging lowers the peak
flame temperature, thereby reducing thermal NOX, and reduces the production of fuel NOX by
reducing the oxygen available for combination with the fuel nitrogen. Staged combustion may be
achieved through methods that require modifying equipment or operating conditions so that a
fuel-rich condition exists near the burners (e.g., using specially designed low-NOx burners,
selectively removing burners from service, or diverting a portion of the combustion air). In
cyclone boilers and some other wet bottom designs, combustion occurs with a molten ash layer
and the combustion gases flow to the main furnace; this design precludes the use of low NOX
burners and air staging. Low-NOx burners may be used to lower NOX emissions by about 25 to
55 percent. Use of overfire air (OFA) as a single NOX control technique reduces NOX by 15 to
50 percent. When OFA is combined with low-NOx burners, reductions of up to 60 percent may
result.  The actual NOX reduction achieved with a given combustion control technique may vary
from boiler to boiler.
                                        3-16
-------
Table 3-5. Nationwide distribution of existing NO, emissions controls used for
coal-fired electric utility boilers for the year 1999 as reported in the Part II EPA
ICR data (source:  Reference 6).
NO, Control Type
Combustion controls -
low-NO, burners
Combustion controls -
low-NO, burners + overfire air
Combustion controls -
overfire air
Other combustion controls (a)
Selective noncatalytic reduction
Selective catalytic reduction
No controls reported (b)
Abbreviation
Code
CC-LNB
CC-LNB/OFA
CC-OFA
CC
SNCR
SCR

Nationwide Total
Phase II EPA ICR Data
Nationwide
Number
of
Boilers
404
84
79
83
32
6
452
1,140(c)
Nationwide
Percentage
of
Boilers
35.4 %
7.4 %
6.9%
7.3 %
2.8 %
0.5 %
39.7%
100%
Percent of
Nationwide
Electricity
Generating
Capacity
43.0 %
10.4 %
10.6 %
5.6 %
0.6 %
1.3%
28.5 %
100 %
    (a) Combustion controls other than low-NO, burners or overfire air. The controls include burners-out-of service,
      flue gas recirculation, off-stoichiometric firing, and fluidized-bed combustion.
   (b) Entry in ICR response indicated "none," "not applicable,* or was left blank.
   (c) Does not include the three IGCC units.

                                         3-17
-------
       Just as the combustion air to the primary combustion zone can be reduced, part of the
fuel may be diverted to create a secondary flame with fuel-rich conditions downstream of the
primary combustion zone. This combustion technique is termed reburning and involves injecting
10 to 20 percent of the fuel after the primary combustion zone and completing the combustion
with overfire air. The fuel injected downstream may not necessarily be the same as that used in
the primary combustion zone.  In most applications of reburning, the primary fuel is coal and the
reburn fuel is natural gas (methane).

      Other ways to reduce NOX formation by reducing peak flame temperature include using
flue gas recirculation (FGR), reducing boiler load, injecting steam or water into the primary
combustion zone, and increasing spacing between burners. By using FGR to return part of the
flue gas to the primary combustion zone, the flame temperature and the concentration of oxygen
in the primary combustion zone are reduced.

      Temperatures can also be reduced in the primary combustion zone by increasing the space
between burners for greater heat transfer to heat-absorbing surfaces. Another combustion control
technique involves reducing the boiler load. In this case, the formation of thermal NOX generally
decreases directly with decreases in heat release rate; however, reducing the  load may cause poor
air and fuel mixing and increase CO and soot emissions.

5.6.2  Selective Catalytic Reduction

      The selective catalytic reduction (SCR) process uses a catalyst with ammonia gas (NHj)
to reduce the NO and NOz in the flue gas to molecular nitrogen and water. The ammonia gas is
diluted with air or steam, and this mixture is injected into the flue gas upstream of a metal
catalyst  bed (composed of vanadium, titanium, platinum, or zeolite). In the reactor, the reduction
reactions occur at the catalyst surface. The SCR catalyst bed reactor is usually located between
the economizer outlet and air heater inlet, where temperatures range from 230 to 400 °C (450 to
750 °F).

3.6,3  Selective Noncatalytic Reduction

      The selective noncatalytic reduction (SNCR) process is based on the same basic
chemistry of reducing the NO and NO2 in the  flue gas to molecular nitrogen and water but does
not require the use of a catalyst to prompt these reactions. Instead, the reducing agent is injected
into the  flue gas stream at a point where the flue  gas temperature is within a  very specific
temperature range.  Currently, two SNCR processes are commercially available: the THERMAL
DeNOx7 and the NOXOUT7. The THERMAL DeNOx7 uses ammonia gas as  the reagent and
requires the gas be injected where the flue gas temperature is in the range of 870 to 1090 °C
(1,600 to 2,000 °F).  Consequently, the ammonia gas is injected at a location upstream of the
economizer. However, if the ammonia is injected above 1,090 °C (2,000 °F), the ammonia will
oxidize and form more NOX. Once the flue gas temperature drops below the optimum
temperature range, the effectiveness of the process drops significantly. By adding hydrogen gas
or other chemical enhancers, the reduction reactions can be sustained to temperatures down to
                                        3-18
-------
approximately 700 °C (1,300 °F). The NOXOUT7 is a similar process but uses an aqueous urea
solution as the reagent in place of ammonia.

       Using nitrogen-based reagents requires operators of SNCR systems to closely monitor
and control the rate of reagent injection. If injection rates are too high, NOX emissions may
increase, and stack emissions of ammonia in the range of 10 to 50 ppm may also result. A
portion (usually around 5 percent) of the NO reduction by SNCR systems results from
transformation of NO to tyO, which is a global warming gas.
3.7 Emission Control Configurations for Coal-fired Electric Utility Boilers

       Mercury can exist in several forms in the flue gas from a coal-fired electric utility boiler
(discussed in Chapter 5). The distribution of these Hg forms in the flue gas stream can be altered
when reagents for post-combustion pollutant control processes are introduced into the flue gas.
Also, as will be discussed in Chapter 6, some of the existing post-combustion control devices
already in use at coal-fired electric utility power plants to meet PM and SOz emission standards
also control Hg emissions with varying levels of effectiveness. Control measures can be
implemented that may enhance the capture of Hg by these control devices. Other Hg control
measures can be implemented in conjunction with control devices already in place at a given
facility. Therefore, understanding which types of post-combustion control devices how electric
utilities currently are implementing at their coal-fired power plants is useful when investigating
potential Hg control measures for these facilities.

       Table 3-6 presents the 1999 nationwide distribution of post-combustion control device
configurations used for coal-fired electric utility boilers. For approximately 70 percent of the
boilers, the only control device used downstream of the furnace is an ESP. If the unit is subject
to SOa and/or NOX emission limit standards, these units do burn low-sulfur coals to meet the SO2
emission limit and use some type of NOX combustion controls to meet the NOX emission limit.
Approximately 25 percent of the boilers use some combination of post-combustion control
devices. The most common configuration used is an ESP with a downstream wet scrubber for
SO2 control. Less than 2 percent of the units use a combination of PM, SO2, and NOX post-
combustion control devices.
                                        3-19


-------
Table 3-6.  Nationwide distribution of post-combustion emission control
configurations used for coal-fired electric utility boilers for the year 1999 as
reported in the Part II  EPA ICR data (source: Reference 6).
                       Post-Combustion Emission Control Device Configuration
                                                                           Phase II EPA ICR Data
  Post-combustion
  Control Strategy
                                                                                      Percent of
                                                                                      nationwide
                                                                                     total number
Number
of boilers
  Post-combustion
    PM controls
      only
  Post-combustion
   PM controls
      and
   SO, controls
  Post-combustion
   PM controls
      and
   NO, controls
  Post-combustion
   PM controls.
   SO, controls.
      and
   NO. controls
 (a) Units using hot-side ESP in series with a cold-side ESP. Counted as "multiple control device combination" in Table 3-2.
 (b) Does not include the three IGCC units.
                                            3-20
-------
3.8 References
    1.   Nizich, S.V., A.A. Pope, and L.M. Driver. National Air Pollutant Emissions Trends,
        1900-1998, U.S. EPA and the States: Working Together for Cleaner Air, EPA-454/R-
        00-002 (NTIS PB2000-108054). Office of Air Quality Planning and Standards,
        Research Triangle Park, NC. March 2000.

    2.   French, C.L., W.H. Maxwell, W.D. Peters, G.E. Rice, O.R. Bullock, A.B Vasu, R.
        Hetes, A. Colli, C. Nelson, and B.F. Lyons. Study of Hazardous Air Pollutant Emissions
       from Electric Utility Steam Generating Units — Final Report to Congress, Volume 1.
        EPA-453/R-98-004a. Office of Air Quality Planning and Standards, Research Triangle
        Park,NC. February 1998.  Available at:
       < http://www.epa.gov/ttii/atw/CQmbust/utiltox/utoxpg.html >.

    3.   U.S. Environmental Protection Agency. Air Quality Criteria for Paniculate Matter and
        Sulfur Oxides, Volunes 1-3, EPA/600/8-82/029a-c. (NTIS PB84-156777).  Office of
        Health and Environmental Assessment, Environmental Criteria and Assessment Office,
        Research Triangle Park, NC. 1982.

    4.   Buonicore, A.J., and W.T. Davis (eds.). Air Pollution Engineering Manual. Air &
        Waste Management Association. Van Nostrand Reinhold, New York, NY. 1992.

    5.   U.S. Environmental Protection Agency. Air Quality Criteria for Oxides of Nitrogen,
        Volumes 1-3, EPA/600/8-9l/049a-c (NTIS PB92-176361; 95-124525; 95-124517),
        Office of Health and Environment Assessment, Environmental Criteria and Assessment
        Office, Research Triangle Park, NC. 1991.

    6.   U.S. Environmental Protection Agency. Database of information collected in the
        Electric Utility Steam Generating Unit Mercury Emissions Information Collection
        Effort. OMB Control No. 2060-0396. Office of Air Quality Planning and Standards.
        Research Triangle Park, NC. April 2001.  Available at:
       .

    7.   Woodward, K. Stationary Source Control Techniques Document for Fine Paniculate
        Matter, EPA/425/R-97-001 (NTIS PB99-116493). Office of Air Quality Planning and
        Standards, Research Triangle Park, NC. October  1998.

    8.   Turner, J.H., and J.D. McKenna. Fabric Filter Baghouses I - Theory, Design, and
        Selection. ETS, Inc., Roanoke, VA.  1989.

    9.   Center for Air Toxic Metals (C ATM). Technical Focus - Advanced Hybrid Paniculate
        Collector, Fourth Annual Meeting. Grand Forks, ND.  September 16-17, 1997.
                                       3-21
-------
                                      Chapter 4
                             Measurement of Mercury
4.1 Introduction
       Accurate measurements of the various forms of Hg present in flue gas from a coal-fired
electric utility boiler are important: to characterize and determine facility and/or fuel-type
absolute emissions, for understanding the behavior of Hg in combustion processes and
combustion configurations, and to evaluate the removal efficiency of control technologies for Hg.
A variety of measurement techniques, both manual and continuous monitoring, are available for
measuring total Hg and select, speciated forms. It is the  latter need and ability that is most
critical to supporting the understanding of Hg behavior and its control.

       Because of the importance of these measurements, particularly speciated Hg
measurements, research on Hg measurement techniques  and performance is an integral
component of the overall Hg control research strategy. The science of speciated Hg
measurements from coal-fired electric utility boilers has  only recently been investigated, with the
majority of research on the subject occurring within the last 5  years. This research has examined
the development and performance of both manual and continuous emission monitor
measurements.  Much of this work began with examining and understanding measurement
performance under very controlled and  simplistic conditions, primarily through the use of
blended gases in a laboratory setting. This afforded the ability to investigate specific
measurement variables and issues individually. Based on this knowledge, experimentation
expanded to pilot-scale combustion systems where gases/Hg species of interest could be doped
into the combustion system, and measurement performance characterized. Though still
simplistic, this approach results in a measurement environment that more closely represents real-
world measurement scenarios. Ultimately, investigations moved to pilot-scale coal combustion
test units, and finally to full-scale, field applications.  At each step, the measurement complexity
increases.  The complexities associated with the combustion of different coal types, relative
amounts of coal combustion emissions (e.g.,  SOX, NOX,  HC1, C12, PM), and pollution control
device availability and configuration all have an impact on the ability to perform quality Hg
measurements.
       The purpose of this chapter is to provide an understanding of the principles, applications,
and limitations of Hg measurement methodologies, particularly with respect to understanding
and interpreting the Part III EPA ICR data. This chapter also serves to introduce principles and
                                          4-1
-------
issues related to Hg CEMs and their use as a valuable research tool. The following sections
provide a summary of the approaches and state-of-the art of manual and continuous emission
measurement methods and issues associated with performing Hg measurements from coal-fired
electric utility boilers.
4.2 Manual Methods for Hg Measurements

       Manual methods are well established for measuring total Hg emissions from a variety of
combustion sources. The EPA Method 101 A2 and Method 293 were developed to measure total
Hg emissions (paniculate phase and gas phase) from combustion sources such as sewage sludge
incinerators and municipal waste combustors.  These reference methods were developed and
used to support total Hg regulatory needs. A reference method for speciated Hg measurement
does not exist, essentially because there are no regulations requiring speciated Hg emissions
measurements. However, a valid, accepted methodology was needed to characterize the
emissions from coal-fired electric utility power plants to better assess the contribution from this
category as well as potential risk.  The Ontario-Hydro Method 4 (called the OH Method in this
report) presently is the method of choice for measuring Hg species in the  flue gas from coal-fired
electric utility plants. This method has been submitted to the American Society for Testing and
Materials (ASTM) for acceptance as a standard reference method.1  The Hg emission data
collected for the Part in EPA ICR were measured using the OH Method.

       Generally, all sampling trains consist of the same sampling components: a nozzle and
probe operated isokinetically for extracting a representative sample from  the stack or duct, a filter
to collect paniculate matter, and a liquid solution and/or reagent to capture gas-phase Hg. After
sampling, the filter and sorption media are prepared and analyzed for Hg  in a laboratory.
Figure 4-1 shows a diagram of the sampling train used for the OH Method.

       Several of the manual methods, including the OH Method, being developed for speciated
Hg measurements from combustion sources have been adapted/modified  from accepted test
methods for measuring total Hg. Measurement of total Hg is based on the concept that all forms
of gaseous Hg can be captured with a strong oxidizing solution such as potassium permanganate.
The speciation is accomplished relying on the solubility  and insolubility of the gaseous Hg
species. To speciate gaseous Hg into the oxidized Hg (Hg2+) and elemental Hg (Hg°) forms,
multiple solutions/reagents are used. The Hg2+ form is considered to be readily soluble in
aqueous solutions, while Hg° is essentially insoluble.1  When the aqueous solutions are
positioned immediately after the filter, the Hg2"1" is captured and the Hg° passes through to the
oxidizing solution where it is then captured. These solutions are analyzed separately to
determine the distribution of oxidized and Hg° within the sampling train.  Table 4-1 presents a
comparison of the different manual test methods, their configuration, and the solutions used that
have been investigated for measuring speciated Hg.

       The OH Method, along with the other test methods listed in Table 4-1, were thoroughly
evaluated to  determine their appropriateness for performing speciated Hg measurements from
                                          4-2
-------
       Thermocouple @ Stack

       Heated   \ B  Wa"
       Probe *-"^C
        Pitot
                                                                Thermometer
Figure 4-1. Diagram of sampling train for Ontario-Hydro Method (source:
Reference 4).
                                      4-3
-------
!i

„ „

a
!

s *
if
ft
o T
U w
% *
s> °»
e S
"5. o-
fl
"8
5
A
3
^^






g
1
1
1
u
c
D>
"5.
1




$

E


0>
03
"O
g
u
V
CO




1
£
iZ



g
U
jj I..
S 3
Back-half
(gaseo

g
d) CT?
To
-1? 2
"c ^~
^.
u_
hi
._.
1
o
o*
I

£
i"
CM
>,
-fe



o:
6"
z
X
CM


<0
|
^
"|
impinger


i
I
%
(0
•5)
EPA Method 29
1
o

1

0)
S
c

1
noneu


„
O
|
x~

CO

en
S
'a
1
impinger


1
IB
(fl
(0
0)
EPA Method 101 A'
1
o
I


i

CM
o;
1


«
1
I
g
'3)
•o

CM
V>
g
3
3
impinger


i
.Q
1C
in
CO
01
EPA Method 1 01 B
(draft)
1
O
0*
§


Ontario-Hydro Method
1
o
^^
1
3

C

o"
a[
X
CM


^
8
•43
1
in
£
f\J
CN

V)
1
3
^
impinger


1
I
S
CO
•a
Tris-Buffer Method
CVAFS *
•^5
1
3
g
c

g
8
I
(0
1
CM


0)
.E
tc
"8
(A
6
^

CM


1

£
S
1


^_
1
cc
{|JJ

MESA Method '
i
                       "8
                       I
                       is
                       0)


                       c

                   •° d = -2 fe g

                   ll-lli
                   
-------
coal-fired combustion sources.1 The University of North Dakota, Energy and Environmental
Research Center (UND/EERC) performed a thorough, parametric evaluation of these methods
under a variety of laboratory and pilot-scale test conditions, including the combustion of
multiple, representalive coal varieties. A detailed presentation of these tests and their results are
contained in two comprehensive reports.1'5

       Initial experimental work focused on EPA Method 29.  These results indicated that
Method 29 exhibited speciation measurement biases under some conditions.1 The testing
expanded to include the Mercury Speciation Adsorption (MESA) Method, Tris-Buffer Method,
draft EPA Method 10IB, and OH Method.1 Pilot-scale coal combustion experiments were then
performed in conjunction with the dynamic spiking of Hg° or mercuric chloride into the duct at
various locations within the post-combustion facility.  Samples by the respective methods were
collected at sampling locations both upstream and downstream of particulate control systems.
These tests were used to isolate the most appropriate methods for further, more definitive testing.

       It was during the initial dynamic Hg spiking tests that effects from fly ash on the quality
of speciated measurements were observed. Speciated Hg measurements using the OH Method
and Tris-Buffer Method where the gas sampling and dynamic spiking of Hg°took place at the
inlet and outlet of the PM control device indicated that significant oxidation of the Hg° occurred
as a result of reactivity with the coal fly ash (see Figures 4-2 and 4-3).

       The effects of PM on Hg speciation can be significant, particularly at sampling locations
upstream of PM control devices.  The flue gas upstream of a PM control device contains a high
concentration of PM (relative to flue gas downstream of a PM control device). When sampling
takes place upstream of a PM control device, the sampling train filter has the potential to collect
a high loading of fly ash (due to the high concentration of PM in the flue gas).  The speciated Hg
measurement can be biased in two ways. The fly ash on the filter can adsorb gaseous Hg from
the flue gas as it passes through the filter. Reactive fly ashes can also oxidize gaseous Hg°
entering the  filter. When adsorption and/or oxidation occur across the filter, they alter the
distribution of total Hg and/or gaseous Hg measured. For example, if particles on the filter
adsorb gaseous  Hg, the filter will contain a greater amount of Hgp  than if no adsorption had taken
place; in this case, the sampling-train method will overestimate the amount of Hgp in the flue gas
and underestimate the gaseous Hg, thus, the total distribution of Hg will be altered.
Alternatively, fly ash on the filter can oxidize gaseous Hg° to Hg2+ (without adsorption)
overestimating the amount of Hg2+ in the flue gas.  Thus, the distribution of gaseous Hg will be
altered. The rates of these transformations are dependent on the properties of the coal and
resulting fly ash, the amount of fly ash, the temperature, the flue gas composition, and the
sampling duration. As a result, the magnitude of these biases varies significantly and cannot be
uniformly assessed.  It is for this reason, that ICR measurements performed at  the inlet of PM
control systems possess a large degree of uncertainty.  A more detailed discussion of the
implications of fly ash speciation biases on the ICR data is presented in Chapter 6.

       A final series of pilot-scale tests were conducted to more definitively evaluate the two
most promising methods identified as a result of the initial dynamic spiking experiments
                                          4-5
-------
     20-


     18-


     16-

«
 E   14-

 o>
 a.
—'   12 —

O

<   10-1
   8    ^
   0    6-|
   at
   x

        4 —
        2-
        0 —
                                          [  | Total vapor-phas* Hg as measured by method

                                          |^| Oxidized Hg as measured by method

                                              Elemental Hg as measured by method

                                              Particle-bound Hg as measured by method
                                           Baseline * 3-run average for Test 26 (PTC run no. 550)

                                           Spiked = 3-run average for Test 28 (PTC run no. 551}
                   8.2 ug/Nm'Hg* Spiked
                                      8.2 Mg/Nm1 Hg9 Spiked
                 Baseline
                                                         8.2 jig/Nm3 Hg° Spiked
                      EPA             Tris-Buffer        Ontario-Hydro
                  Method 101A          Method             Method

                              MANUAL TEST METHOD
Figure 4-2. Comparison of Hg speciation measured by manual test methods from
UND/EERC pilot-scale evaluation tests firing Blacksville bituminous coal and
sampling and spiking Hg° at FF inlet (source: graph prepared using test data
presented in Appendix B to Reference 1).
                                         4-6
-------
       20-i
       18-
       16-
E   14-


3.
—'   12 H

O

<   10 —
   O
   u
   o>
   X
     6-
        4 —
        2-
                                          Total vapor-phase Hg as measured by method

                                          Oxidized Hg as measured by method

                                          Elemental Hg as measured by method

                                       Baseline « 3-run average for Test 27 (PTC run no. 550)

                                       Spiked " 3-run average for Test 31 (PTC run no. 552)
                                     8.2 ug/Nm* Hg° Spiked
                   8.2 ug/Nm1 Hg» Spiked
               Baseline
                                                        8.2 ug/Nm3 Ho" Spiked
                      EPA            Tris-Buffer        Ontario-Hydro
                  Method 101A          Method             Method

                              MANUAL TEST METHOD
Figure 4-3.  Comparison of gaseous Hg speciation measured by manual test
methods from UND/EERC pilot-scale evaluation tests firing Blacksviile
bituminous coal and sampling and spiking Hg° at FF outlet (source: graph
prepared using test data presented in Appendix B to Reference 1).
                                       4-7
-------
discussed above.1 Both Draft EPA Method 101B and the OH Method were selected for formal
EPA Method 301 validation testing.  Method 301 is EPA's accepted guidance for validation of
source testing methodologies.6 For these validation tests, all sampling and dynamic spiking of
Hg° and HgC^ into a flue gas stream were performed at the outlet of the high efficiency fabric
filter (FF), while burning a blend of Ohio No. 5 and Ohio No. 6 coals.1 Validation testing was
not performed at the PM control device inlet location.

      A summary of the Method 301 validation results is shown in Table 4-2. The tests
verified that both the OH Method and the draft EPA Method 101B achieved acceptable
performance as defined by Method 301.' The precision of the OH Method for total gaseous Hg
was determined to be less than 11 percent relative standard deviation (RSD) for Hg
concentrations greater than 3 ug/Nm3 and less than 34 percent RSD for Hg concentrations less
than 3 ug/Nm3.  These values were within the acceptable range, based on the criteria established
in EPA Method 301 (less than 50 percent RSD). In all cases, the laboratory bias for these tests
based on a calculated correction factor was not statistically significant, though some oxidation
(less than 15 percent) of the Hg° spike was observed even when spiking and sampling was done
at the outlet of the fabric filter. The draft EPA Method 101B also met Method 301 validation
requirements, though it did not perform as well as the OH Method.1 As a result, the OH Method
was selected as the most appropriate method for Hg speciation measurements in coal
combustion gases.1

      Final approval by the ASTM of the OH Method as an international test procedure is still
pending as of the date of this report.  The OH Method, in its current draft form, is available from
the  EPA Office of Air Quality Planning and Standards (OAQPS) Emission Measurement Center
(EMC).4  The draft version of the OH Method submitted to ASTM states that the method is
applicable for sampling elemental, oxidized, and particle-bound Hg at the inlet and outlet of
emission control devices and is suitable for measuring Hg concentrations ranging from
approximately 0.5 to 100 ug/Nm3.4 Measurement sensitivity/detection levels can be extremely
important where control technology performance is being determined in relatively low  Hg coal
content applications.

      In summary, while several manual methods for Hg speciating measurements exist, the
OH Method is the most thoroughly examined and accepted of these methods, and has met EPA
Method 301 validation requirements. Application to air pollution control device inlet locations
should be considered with caution due to the known catalytic and sorptive effects of certain coal
fly ash PM. These measurement artifacts do not affect the use of the OH Method for total Hg
measurements.
4.3 Continuous Emission Monitors for Hg Measurements

       Continuous emission monitors (CEMs) are preferable for multiple reasons to using
manual methods for measuring Hg. A CEM is capable of providing a real-time or near-real-time
response for Hg measurements.  A CEM can be used to obtain continuous Hg measurements
                                         4-8

-------
Table 4-2.  Results from EPA Method 301 evaluation tests for the Ontario-Hydro
Method (sources: References 1 and 4).
Ontario-
Hydro
Method *
Baseline
Hg" Spike
(15.0ng/Nm3)
HgCI2 Spike
(19.9 tig/Mm3)
Total Vapor-Phase Hg
Mean6,
;ig/Nm3
23.35
38.89
42.88
Standard
Deviation
2.05
2.00
2.67
RSD°,
%
8.79
5.13
6.23
Oxidized Hg
Mean6,
Mg/Nm5
21.24
23.32
40.22
Standard
Deviation
2.13
2.08
2.87
RSD,
%
10.02
8.94
7.14
Elemental Hg
Mean6,
tig/Mm'
2.11
15.57
2.66
Standard
Deviation
0.65
1.09
0.89
RSD,
%
30.69
6.97
33.31
a.  The correction factor in all cases was not statistically significant and is not shown.
b.  For each mean result, there were 12 replicate samples (four quad trains).
c.  RSD = Relative standard deviation.
                                           4-9
-------
over long periods in time. Conversely, manual methods are capable of only infrequent
"snapshot" Hg measurements over time. As a result, CEMs are able to distinguish the magnitude
and duration of short-term emission characteristics as well as perform long-term emission
measurements to truly characterize a process's temporal emissions. Again, manual methods are
not capable of performing these functions. It is for these reasons that Hg CEMs are extremely
valuable tools supporting the understanding and control of Hg emissions from coal-fired electric
utility power plants. This section discusses the state-of-the-art of using CEMs for Hg
measurements and the associated measurement issues.

       In general, Hg CEMs are a relatively new and yet unproven technology. Although CEMs
that measure total Hg only are used to support regulatory applications in several European
countries, the use of these CEMs is limited. Several total Hg CEMs are available commercially
                                 "78
and are primarily of European origin.   In the United States., Hg CEMs have been limited to
research applications with respect to coal-fired combustion emissions monitoring. As with the
manual methods, CEMs capable of Hg speciation measurement are of the most value to
supporting research on the characterization and control of Hg emissions from coal-fired electric
utility boilers. The speciating Hg CEMs currently available should be considered prototypes.

       The CEMs being developed for measuring Hg are similar to most other types of CEMs
used for combustion processes in that the combustion gas sample typically must be extracted
from the stack and then transferred to the analyzer for detection. However, continuous Hg
monitoring is complicated by the fact that Hg exists in different forms (i.e., Hg°, Hg2+, and Hgp)
and that quantitative transport of all these forms is difficult.

       Typically, Hg CEMs measure (i.e., detect) only Hg°. These CEMs measure total Hg
through the use of a conversion system that converts (reduces) the  gaseous non-elemental or Hg2+
forms to Hg° for detection. Mercuric chloride is considered to be the primary oxidized form of
Hg, though recent research suggests that other oxidized forms of Hg do indeed exist.9'10
Although particulate-bound Hg can also be reduced to the gaseous elemental form, participate
sample delivery issues make this impractical.  As  a result, for most commercially available
CEMs, the total Hg measured is in  fact total gaseous Hg (TGM).

       The conversion of gaseous, non-Hg° is commonly accomplished using a liquid reducing
agent (e.g., stannous chloride).  This technique is least preferable, though more established.  The
use of wet chemical reagents is considered to be a limitation to Hg CEM use.  The wet chemicals
typically possess corrosive properties and require  frequent replenishment.  The spent reagents
may possess hazardous properties that result in waste disposal concerns. In addition, the
reducing ability of reagents such as stannous chloride can be affected by high levels of SC«2."

       In addition to the more established wet chemistry conversion methods, dry conversion
methods are also available. These techniques use high temperature catalysts or thermal reduction
units to not only convert non-Hg°, but also condition the sample for analysis by removing
selective interferants. This approach does much to minimize the size of the conversion system as
well as maintenance requirements.  However, these systems have not been well characterized for
                                         4-10
-------
coal combustion gas Hg measurement applications.

       Because the particulate form is difficult to transfer and is also often a measurement
interferant, the particulate is typically filtered out and Hgp remains unmeasured. This could
potentially impart a negative bias to the total Hg measurements. This bias could be further
amplified as certain types of particulate may actually capture gas-phase Hg. This may not be a
significant issue for sources where Hgp is not present in appreciable quantities, but may be a
significant issue for high particulate-emitting sources (e.g., sources with minimal PM control) or
in cases where the Hg measurements are conducted upstream of PM control devices.  Therefore,
the capability of a C EM to measure Hgp is important and should not be ignored.

       Similarly, there are known complications with the quantitative transfer of mercuric
chloride.9 Mercuric chloride (HgC^) is water soluble and reactive with many surfaces.  Losses
due to adsorption are the major concern. As a result, recent emphasis has been placed on
locating the non-Hg°  conversion systems as close as possible to the source so that the elemental
form is transferred from the source to the detection unit instead of transporting the oxidized
forms long distances.

       In general, Hg CEMs can be distinguished by their Hg measurement detection principle.
Detection systems include: cold-vapor atomic absorption spectrometry (CVAAS); cold-vapor
atomic fluorescence spectrometry (CVAFS); in-situ ultraviolet differential optical absorption
spectroscopy (UVDOAS); and atomic emission spectrometry (AES).1'7'8'9

       The majority of Hg CEM systems employ CVAAS or CVAFS as the detection technique.
These detection techniques are susceptible to measurement interferences resulting from the
presence of common  combustion process emissions. Gases such as NOx, SC«2, HC1, and C\2 can
act as measurement interferants as well as degrade the performance of concentrating devices
(e.g., gold amalgams). As such, conditioning systems and/or techniques that remove or negate
the effects of these interfering gases prior to sample delivery to the detector are required. The
SC<2 is a major spectral interferant with most CVAA detection systems.  The effects of SCh are
commonly negated through the use of a gold trap.  The sample  gas is directed through a gold
trap, where the Hg amalgams with the gold surface. Once the trap is loaded, it is heated and
flushed with a SO2-free carrier gas to the detector.  The trapping also serves to improve
measurement sensitivity by concentrating the sample.  A trapping device is required of CVAFS
systems to achieve optimum sensitivity; not because of the concentrating aspect, but because the
carrier gas will enable maximum sensitivity. Oxygen and nitrogen have spectral quenching
effects that suppress measurement sensitivity. Conditioning of the sample gas prior to reaching
the gold trap is often  required. HC1 and NO\ in combination can poison the gold surface,
preventing amalgamation with the Hg. Removal of both or either of these constituents is
required.

       An alternative to  the Hg°measurement approach is AES. With this technique, the Hg is
ionized by a high-energy source (e.g., plasma) and the emission energy detected. The advantage
to this technique is that all forms of Hg, including particulate-bound Hg, are capable of being
                                          4-11
-------
ionized and detected. Although this technology is not quite as developed, another major
advantage of AES is that the ionization source and detector can be located directly at the source,
avoiding sample delivery issues. In addition, AES is not as susceptible to spectral interferences
from common flue gas constituents.

       Speciated Hg measurements are important to characterize combustion process emissions
and evaluate Hg control strategies. While there are no commercially available CEMs that
directly measure the various speciated forms of Hg,  several total gaseous Hg CEMs, both
commercial and prototype, have been enhanced to indirectly measure speciated Hg (the elemental
and oxidized forms) by determining the difference between Hg°and total gaseous Hg.  This
difference is recognized as the oxidized form. Separate Hg measurements are made before and
after the conversion step in order to calculate the oxidized form. This indirect speciation method
is referred to as "speciation by difference."  Based on the current understanding that the oxidized
species of primary interest is mercuric chloride and that mercuric chloride is the dominant form
of oxidized Hg present, the "speciation by difference" technique is considered an acceptable
approach to obtaining speciated Hg measurements.

       A key to performing the speciated Hg measurement is being able to perform reliable Hg°
measurements.  The  Hg2+ must be removed without  adding to the true amount of Hg° in the
sampled gas stream.  This is often accomplished using a liquid reagent to remove the water-
soluble Hg2+. These reagents also may serve to neutralize the effects of measurement
interferants.  The greatest concern is the reliability of the speciated Hg measurement.
Measurement artifacts exist that bias the speciation, primarily by over-reporting the level of the
oxidized species. The largest cause of this bias comes from the reactivity of certain types of PM
(as discussed in Section 4.2). The PM may possess  catalytic properties whereby, at the
conditions of Hg CEM PM filtering environments, Hg°can be oxidized across the PM surface.
This is not an issue from a TGM measurement standpoint (unless transport of oxidized Hg is an
issue). However, it may have major implications when measuring Hg in gas streams possessing
high PM loadings. This bias is minimized in low PM loading gas streams, consistent with Hg
measurements downstream of PM control devices. Another potentially significant source of
speciated Hg measurement bias takes place in the liquid phase. In combustion gases where Cb is
present, under certain conditions the Ch may react in the liquid phase to oxidize Hg0.12 There is
evidence that this problem can be mitigated.

       As stated previously, the current, primary application of Hg CEMs is as a research
tool/process monitor. Speciating Hg CEMs are integral to the DOE/EPA/EPRI Hg control
technology development and evaluation research program. These Hg CEMs are used to
characterize existing Hg emissions and distributions, including control technology performance.
More importantly, these speciating Hg CEMs are used to better understand and optimize
potential Hg control technologies so that absolute emissions can be established through OH
sampling. Ultimately, it is desired to accept the quality and performance of Hg CEMs and
measurements data so as to replace the reliance on OH measurements. Several pilot-scale and
field tests have been performed specifically to evaluate and determine the measurement
performance of both total and speciating Hg CEMs.
                                          4-12
-------
       Several tests have been conducted specifically to evaluate total Hg CEMs as a compliance
assurance tool. The first such test, sponsored by the EPA Office of Solid Waste (OSW),
evaluated the performance of three Hg CEMs to measure total Hg emissions from a cement kiln
that burned hazardous waste as a fuel.13 Measurement performance was evaluated following the
proposed "Performance Specification 12 — Specifications and Test Procedures for Total
Mercury Continuous Monitoring Systems in Stationary Sources " (PS-12).14 At the time, this was
a relatively new test procedure and had yet to be implemented. In fact, the guidance called for
Hg° and HgCk gas standards that had yet to be developed and proven. The tests were only
marginally successful.  None of the Hg CEMs tested met the performance test requirements.
Based on the test results, the EPA/OSW concluded that  Hg CEMs should not be considered as a
compliance tool for hazardous waste combustors.13 In retrospect, the harshness of the cement
kiln's exhaust gas stream was concluded as a major cause of the test program's lack of
success.8'13 The cement kiln chosen for the EPA/OSW Hg CEM testing was not equipped with
acid gas controls and had relatively high PM loading, resulting in severe interferences and
operational difficulties for the CEMs.

       The DOE Mixed Waste Focus Area (MWFA) has sponsored several tests determining the
measurement performance of a single total Hg CEM under hazardous waste incineration
conditions.15'16 Measurement performance was also evaluated following PS 12. These tests
demonstrated not only Hg CEM performance, but also that additional elements of the PS 12 test
procedures could be implemented.  A prototype Hg° compressed gas standard was used for the
first time. While these tests have been relatively successful, they are still limited in scope  and
application.

       The EPA's Environmental Technology Verification (ETV) Program, in collaboration
with the NRMRL, has completed testing of four commercially available Hg CEMs  from three
vendors using the unique capabilities  of NRMRL's pilot-scale combustion test facility. These
tests examined the measurement performance of both total and speciated Hg CEMs under two
distinct and diverse combustion conditions. Coal and chlorinated waste combustion conditions
were simulated. These verification tests used PS 12 as guidance, but also considered specific
measurement issues of interest and innovative approaches that better examined these issues.  The
pilot-scale tests were unique in that specific measurement issues were investigated as variables.
The pilot-scale combustion facility enabled independent control of Hg concentration and species.
As a result, the total Hg measurement could be challenged by the distribution of oxidized and
Hg°. Interference flue gas constituents were also independently examined.  The ETV testing
made use of several new quality assurance and quality control (QA/QC) tools. Newly developed
Hg° compressed gas standards were used to determine Hg CEM calibration drift and system bias.
As a result, not only were Hg CEMs evaluated, but also  improved techniques for evaluating Hg
CEMs were demonstrated.  Performance data for the participating Hg CEMs are not yet
available.

       The UND/EERC has evaluated the performance  of Hg CEMs during field tests at eight
different coal-fired electric utility power plants representing facilities that burn lignite,
subbituminous coal, or bituminous coal."'17   A variety of air pollution control devices and
                                         4-13
-------
configurations were encountered, including ESPs, FFs, wet FGD scrubbers, spray dryer
absorbers, and venturi scrubbers. For these tests, the Hg CEMs evaluated demonstrated the
ability to measure total gaseous Hg within +20 percent of the OH Method measurements. The
field-testing also examined the measurement performance of several Hg CEMs at low stack Hg
emissions  levels. These tests demonstrated a distinct advantage  of the AF-based systems over
the AA-based system (see Figure 4-4). Below concentrations of 5 M-g/m3» the AA-based systems
exhibited higher signal to noise ratios. At these concentrations, the  AF-based systems are a
better choice.

       The EPA/OAQPS/EMC has recently initiated a study to determine the measurement
performance of two commercially available total Hg CEMs at a coal-fired electric utility power
plant. Measurements of performance will be recorded to determine potential monitoring
applications based on measurement performance achieved. Data from this study, and future
studies of Hg CEM measurement performance at additional source categories, should aid in the
future crafting of a performance specification for application of total Hg CEMs to a variety of
different Hg emission source categories.

       Performance testing of Hg CEMs has focused primarily on total Hg CEMs; total Hg
CEMs are  the most widely available commercially.  However, with  respect to the development
and evaluation of Hg control technologies for coal-fired electric  utility power plants, the most
urgent need is for a speciating Hg monitor.  As stated previously, the primary use of speciating
Hg CEMs  is as a research tool though application as a process monitor is also appealing. Of
those speciating Hg CEMs in use, most are commercially available total or Hg° CEMs modified
for use as a speciating Hg CEM.  Very few speciating Hg CEMs are available commercially.
The major distinction among speciating Hg CEMs is not the analyzer or detection principle, but
the approach for managing potential interferants and method for converting oxidized forms of Hg
to the detectable, elemental form.

       Performance testing of speciating Hg CEMs to support Hg control technology research
has also been performed under pilot- and field-scale operations and  research continues in this
area. Work performed by the UND/EERC has also  focused on the research and development of
speciating Hg CEMs, particularly the development and evaluation of pretreatment/conversion
systems that can be used with multiple, commercially available Hg CEMs. The EERC has used
speciating Hg CEMs to support field measurement activities  in conjunction with OH Method
measurements.  Figure 4-5 compares the measurement performance of several speciating Hg
CEMs to OH Method measurements made during testing at a coal-fired electric utility power
plant.

       A key to assessing measurement performance and validating measurement data quality is
the development Quality Assurance/Quality Control (QA/QC) tools such as elemental and
oxidized Hg gas standards.  The tools are needed for instrument  calibration, continuing
calibration or drift checks, and system bias checks.  The EPA/ORD  has been active in the
development of both elemental and HgCh gas standards. A commercial compressed gas standard
for Hg°has been evaluated for stability and accuracy.  While the stability of the Hg° compressed
                                         4-14
-------
                                                                     Ontario Hydro Memod
                                                                — Semtech Hg 2000
                                                                —— Samteeh Mg 2)010
                                                                  O  PS Analytical
                                                                  *>  TeKran	
          0.01
Figure 4-4.  Comparison of total Hg results for CEMs at low Hg levels.


(Reprinted from "State-of-the-Art of Mercury Continuous Emission Monitors for Coal-Fired Systems."
Conference on Air Quality II Mercury, Trace Elements, and Particulate Matter, McLean, VA,  September
2000, by D. L. Laudal and N. B. French, with permission of the University of North Dakota Energy &
Environmental Research Center as copyright owner.)
                                          4-15
-------
         I

         o
         9
         5
         ^*
         Q)
         U
         1
3 -
             2 -
                  Stack from Unit 5
                                   PS Analytical
                                   Semtech Hg 2000
                                   Ontario Hydro (Total Hg)
                                   Ontario Hydro (Hg°)
          7-13-99
                                                          22    24   26
Figure 4-5.  Comparison of Hg speciation results for CEMs at low Hg levels.

(Reprinted from "State-of-the-Art of Mercury Continuous Emission Monitors for Coal-Fired Systems."
Conference on Air Quality II Mercury, Trace Elements, and Paniculate Matter, McLean, VA, September
2000, by D. L. Laudal and N. B. French, with permission of the University of North Dakota Energy &
Environmental Research Center as copyright owner.)
                                          4-16
-------
gas standard has been confirmed, techniques for establishing the standard's true concentration
have not. As a result, quantitative use of the standard is limited. Similarly, acceptance of a
HgCU standard is valuable: this standard is used to assess Hg conversion system effectiveness as
well as overall sampling system delivery efficiency and reactivity, parameters not challenged by
an Hg° gas standard.  This is particularly relevant in measurement applications where oxidized
Hg may be the predominant Hg form present. Moreover, several Hg CEMs vendors have
developed QA/QC capabilities to perform their own instrument calibration drift and system bias
checks from internal Hg°gas sources. These capabilities are needed for routine daily operational
performance verification.

       In summary, Hg CEMs are currently the tool of choice for evaluating the performance of
candidate Hg control  technologies. As different control technologies are evaluated, the
associated measurement issues are encountered and addressed.  Measurement issues are primarily
associated with the oxidized Hg conversion systems as well as particulate bias effects,
particularly at pollution control device inlet measurement locations. Both wet chemistry and dry
conversion/conditioning systems are used to support these control technology research programs.
It is the conversion/conditioning system that requires the most attention during operation of Hg
CEM systems.  It is also this frequent need for attention that limits their application to short
measurement intervals. As a result, consideration as a compliance assurance tool is hindered.
Clearly, in order to function as a dedicated process monitor and/or compliance tool, additional
research is needed to  develop and/or evaluate more reliable and less labor intensive Hg
conversion/sample conditioning systems. These objectives are likely to be furthered as a result
of control technology demonstration and evaluation activities.
4.4 Summary, Conclusions, and Recommendations

       Valid and reliable Hg measurements, by either manual methods or using CEMs, are
critical to the characterization and future reduction of Hg emissions from coal-fired electric
utility power plants. Although these measurement techniques are tools that support a larger
research objective, the quality, applicability, and specificity of these measurements directly
impact the ability to conduct Hg emission control research. Measurement techniques that
determine both the Hg2+ and Hg° gaseous forms of Hg are preferred over those techniques that
can measure only total gaseous Hg. Conversely, speciated Hg measurement techniques are more
complex  and more susceptible to measurement biases. Although viable measurement techniques
exist and measurement performance has been demonstrated for certain measurement situations,
acceptable measurement techniques are not available to meet all measurement needs. Additional
research and development is still needed to enable quality measurements from all necessary
measurement environments.

       The OH Method is the only manual method that is currently recognized in the United
States for speciated Hg measurements in coal combustion gases.  The OH Method appears to
provide valid speciation results at sampling locations downstream of PM control devices in
                                          4-17
-------
which most of the fly ash has been removed from the gas stream. However, measurements made
upstream of PM control devices are susceptible to measurement artifacts that bias the
measurements of the different Hg species causing potential uncertainty in results. However,
these artifacts do not affect the measurement of total Hg.

       A limited number of both private prototype and commercial Hg CEMs are available for
the measurement of total gas-phase Hg and to a lesser extent, speciated gas-phase Hg. Because
of the diversity and severity of associated measurement environments, numerous measurement
obstacles exist (e.g., PM artifacts, interferences, conversion systems, sample
conditioning/delivery) that have not been adequately addressed, particularly with respect to
speciated measurements. While Hg CEMs are used being used as a tool by researchers, these
devices are not yet suitable for routine Hg monitoring applications at coal-fired electric utility
power plants. As a research tool, Hg CEMs are suitable for short-term measurement needs.
However, the technology has not advanced to the extent that acceptable, long-term measurement
performance has been demonstrated.  This must be accomplished for Hg CEMs to be considered
suitable for any purpose beyond use as a research tool.  The primary obstacle is the lack of
sample conditioning/conversion systems suitable for long-term, minimal attention operation.

       Improved methods for the sampling and analysis are critical to support the development
of Hg emission control technologies,  for use for Hg monitoring and control (process control), and
for potential use as compliance tools. Specifically, research is needed to:

       1.  Develop improved sample conditioning/conversion systems (particularly dry, non-wet
          chemical) capable of long-term, minimal maintenance, operation,

       2.  Develop and demonstrate  improved Hg CEM measurement techniques that address
          known and potential measurement obstacles (e.g., PM artifacts, interferences/biases,
          conversion systems, sample conditioning/delivery),

       3,  Develop accepted QA/QC tools (e.g., elemental and oxidized Hg gas standards) for
          validating instrument performance and data quality,

       4.  Develop and verify a manual test method suitable for measuring total and speciated
          Hg at sampling locations upstream of PM control devices,

       5.  Develop and verify a manual test method (e.g., modified OH Method) that can
          simultaneously measure speciated Hg and other trace metals,

       6.  Develop and demonstrate  measurement techniques that are capable of directly
          identifying and quantifying trace levels of individual ionic species of Hg [e.g.,
          HgCl, HgS, HgS04, Hg (N03) 2],
                                         4-18
-------
       7.  Verify the ability of Hg CEMs to accurately measure total gas-phase Hg and speciated
          gas-phase Hg at diverse stack conditions representative of fuel type and pollution
          control device configurations (e.g., downstream of PM control devices and wet FGD
          scrubbers),

       8.  Verify the ability of Hg CEMs to accurately measure total gas-phase Hg and speciated
          gas-phase Hg at measurement locations upstream of PM control devices,

       9.  Demonstrate Hg CEM long-term monitoring performance, including operational
          requirements,

       10. Identify and evaluate alternative, cost-effective semi-continuous methods for
          measuring the stack emission of total Hg, and

       11. Demonstrate the use of Hg CEMs and semi-continuous monitoring methods as
          potential Hg emission compliance tools.
4.5 References
1.   Electric Power Research Institute. Evaluation of Flue Gas Mercury Speciation Methods,
    Final Report TR-108988, Palo Alto, CA, December 1997.

2.   U.S. Environmental Protection Agency. "Method 101A—Determination of Particulate and
    Gaseous Mercury Emissions from Stationary Sources." Code of Federal Regulations, Title
    40, Part 61, Appendix B.

3.   U.S. Environmental Protection Agency .  "Method 29--Determination of Metals Emissions
    from Stationary Sources." Code of Federal Regulations, Title 40, Part 60, Appendix A.

4.   "Standard Test Method for Elemental, Oxidized, Particle-Bound, and Total Mercury in Flue
    Gas Generated from Coal-Fired Stationary Sources (Ontario-Hydro Method), October 27,
    1999. Available at: < http://www.epa.gov/ttn/emc/prelim/pre-003.pdf >.

5.   Electric Power Research Institute. A State-of-the-Art Review of Flue Gas Mercury
    Speciation Methods,  Final Report TR-107080, Palo Alto, CA, December 1996.

6.   U.S. Environmental Protection Agency.  "Method 301 - Field Validation of Pollutant
    Measurement Methods from Various Waste Media." Code of Federal Regulations, Title 40,
    Parts 63, Appendix A.

7.   Ryan, J.V. "Development and Evaluation of Mercury CEMS for Combustion Emissions
    Monitoring." In Proceedings of 17th Annual Waste Testing and Quality Assurance
    Symposium, Arlington, VA. August 15, 2001.
                                         4-19
-------
8.   French, N., S. Priebe, and W. Haas, Jr. "State-of-the-art mercury CEMS." Analytical
    Chemistry News & Features, 470-475A (July 1, 1999).

9.   Hedges, S., J. Ryan, and R. Stevens.  Workshop on Source Emission and Ambient Air
    Monitoring of Mercury, Bloomington, MN, September 13-14, 1999.  EPA/625/R-00/002
    (NTIS PB2001-100963).  National Risk Management and National Exposure Research
    Laboratory, Cincinnati, OH. June 2000.

10.  Brown, T. D., D.N. Smith, R.A. Hargis, Jr., and W.J. O'Dowd. "1999 Critical Review:
    Mercury Measurement and Its Control:  What We Know, Have Learned, and Need to
    Further Investigate," JournaloftheAir&  Waste Management Association, June 1999.  pp.
    1-97. Available at:
    < http://www.lanl.gov/projects/cctc/resources/pdfsmisc/haps/CRIT991 .pdf >.

11.  Laudal, D. L., T. D. Brown, and P.  Chu, "Testing of a Mercury Continuous Emission
    Monitor at Three Coal-Fired Electric Utilities." Paper presented at the 92nd Annual
    Meeting and  Exposition of the Air  & Waste Management Association, St. Louis, MO, June
    1999.

12.  Linak, W. P., J. V. Ryan, B.S. Ghorishi, and J. O. L. Wendt. Issues Related to Solution
    Chemistry in Mercury Sampling Impingers. Journal of the Air & Waste Management
    Association,  51: 688-698 (2001).

13.  U. S. Environmental Protection Agency,  Draft Mercury Continuous Emissions Monitor
    System Demonstration, Volume I: Holnam, Inc., Hazardous Waste Burning Kiln, Holly Hill,
    SC.  Office of Solid Waste and Emergency Response, Washington, DC. March 1998.

14.  U. S. Environmental Protection Agency.  Draft Performance Specification 12 -
    Specifications and Test Procedures for Total Mercury Continuous Monitoring Systems in
    Stationary Sources, Office of Air Quality Planning and Standards, Emission Measurement
    Center, Research Triangle Park, NC. Proposed April 19, 1996. Available at:
    < http://www.epa.gov/ttn/emc/propperf.html >.

15.  Gibson, L. V., J. E. Dunn, R. L. Baker, W.  Sigl, and I. Skegg, "Field Evaluation of a Total
    Mercury Continuous Emission Monitor at a U. S. Department of Energy Mixed Waste
    Incinerator."  Paper presented at the 92nd Annual Meeting and Exposition of the Air and
    Waste Management Association, St. Louis, MO, June 1999.

16.  Baker, R. L.  "Are We Ready for Meeting Continuous Emission Monitoring Requirements
    for Total Mercury Combustion Sources?" Paper presented at the 93rd Annual Meeting and
    Exposition of the Air and Waste Management Association, Salt Lake  City, UT, June 2000.
                                        4-20
-------
17.  Laudal, D.L., and N.B. French, "State-of-the-Art of Mercury Continuous Emission Monitors
    for Coal-Fired Systems."  Conference on Air Quality II Mercury, Trace Elements, and
    Particulate Matter, McLean, VA, September 2000.
                                        4-21
-------
                                      Chapter 5
                         Mercury Speciation and Capture
5.1 Introduction
       The source of Hg emissions from coal-fired electric utility boilers is the Hg that naturally
exists in coal and is released during the combustion process. As discussed in Chapter 2, the Hg
content of a coal varies by coal type and where it is mined. When the coal is burned in an
electric utility boiler, most of the Hg bound in the coal is released into the combustion product
gases.  This chapter provides an introduction to Hg chemistry and behavior of Hg as it leaves the
combustion zone of the furnace and passes in the flue gas through the downstream boiler
sections, air heater, and air pollution control devices. Recent research on Hg chemistry in coal-
fired electric utility boiler flue gas is summarized.
5.2 General Behavior of Mercury in Coal-fired Electric Utility Boilers

       The majority of Hg in coal exists as sulfur-bound compounds and compounds associated
with the organic fraction in coal. Small amounts of elemental Hg may also be present in the
coal. Figure 5-1 presents a simplified schematic of the coal combustion process.  The primary
products of coal combustion are carbon dioxide (COa) and water (HjO). In addition, as
discussed in Chapter 3, significant quantities of the pollutants sulfur dioxide (SO:) and nitrogen
oxides (NOx) are also formed. When the coal is burned in an electric utility boiler, the resulting
high combustion temperatures in the vicinity of 1,500 °C (2,700 °F) vaporize the Hg in the coal
to form gaseous elemental Hg. Subsequent cooling of the combustion gases and interaction of
the gaseous elemental Hg with other combustion products result in a portion of the Hg being
converted to other forms.

       There  are three basic forms of Hg in the flue gas from a coal-fired electric utility boiler:
(1) elemental  Hg (represented by the symbol Hg° in this report); (2) compounds of oxidized Hg
(collectively represented by the symbol Hg2+ in this report); and (3) particle-bound mercury
(represented by the symbol Hgp in this report). Oxidized mercury compounds in the flue gas
from a coal-fired electric utility boiler may include mercury chloride (HgCla), mercury oxide
(HgO), and mercury sulfate (HgSO4). Some researchers refer to oxidized mercury compounds
collectively as ionic mercury. This is because, while oxidized mercury compounds may not exist
as mercuric ions in the boiler flue gas, these compounds are measured as ionic mercury by the
                                          5-1
-------
        COAL
        HgS
     Organic Hg
                   CO2   H2O  SO2   NO,  Entrained PM
       APCD
       INLET

       Major
      Mercury
140 °C  Hg9C'.2
        HgO
       HgS04
        Hgp
Figure 5-1. Mercury species distribution in coal-fired electric utility boiler flue
gas.
                                        5-2
-------
speciation test method used to measure oxidized Hg (discussed in Chapter 4).  Similarly,
particle-bound Hg is referred to as paniculate mercury by some researchers.  The term particle-
bound mercury is the preferred and is used in this report to emphasize that the mercury is bound
to a solid particle.

       The term speciation is used to denote the relative amounts of these three forms of Hg in
the flue gas of the boiler.  At present, speciation of Hg in the flue gas from a coal-fired electric
utility is not well understood. A number of laboratory and field studies have been conducted, or
are ongoing, to improve the understanding of the transformation of Hg° to the other Hg forms in
the flue gas downstream of the boiler furnace.  Data obtained to date indicate that combinations
of site-specific factors affect the speciation of Hg in the flue gas. These factors include:

       •  Type and properties of the  coal burned.
       •  Combustion conditions in the boiler furnace.
       •  Boiler flue gas temperature profile.
       •  Boiler flue gas composition.
       •  Boiler fly ash properties.
       •  Post-combustion flue gas cleaning technologies used.

       The current understanding of the mechanisms by which Hg° transforms to Hg2+ and Hgp
in the flue gas from coal-fired electric utility boilers is discussed in subsequent sections of this
chapter. It is  important to understand how Hg speciates hi the boiler flue gas because the overall
effectiveness  of different control strategies for capturing Hg often depends on the concentrations
of the different forms of Hg present in the boiler flue gas.  This topic will  be discussed in detail
in Chapters 6 and 7.

5.3 Speciation of Mercury

       As mentioned above, high temperatures generated by combustion in the boiler furnace
vaporize Hg in the coal. The resulting gaseous Hg° exiting the furnace combustion zone can
undergo subsequent oxidation in the  flue gas by several mechanisms. The predominant oxidized
Hg species in boiler flue gases is believed to be HgCli. Other possible oxidized species may
include HgO, HgSC»4, and mercuric nitrate monohydrate Hg(NOj)2»H2O.  The potential
mechanisms for oxidation of Hg° in the boiler flue gas include:

       •  Gas-phase oxidation.
       •  Fly ash mediated oxidation.
       •  Oxidation by post-combustion NOX controls.

       Each of these oxidation mechanisms is discussed in the following sections.
                                          5-3
-------
5.3.1  Gas-phase Oxidation

       As mentioned above, Hg in coal is believed to completely vaporize and convert into
gaseous Hg° in the combustion zone of a boiler system. As gaseous Hg° travels with the flue gas
in the boiler, it can undergo gas-phase oxidation to form gaseous Hg2+, most of which is believed
to be HgCk, Recent research' has speculated that the major gas-phase reaction pathway to form
gaseous HgCb is the reaction of gaseous Hg° with gaseous atomic chlorine (Cl).  The latter is
formed when chlorine in coal vaporizes during combustion.

       At the furnace exit, the temperature of the flue gas is typically in the vicinity of 1400 °C
(2552 °F). The flue gas cools as it passes through the heat exchanging equipment in the post-
combustion region. At the outlet of the air heater (the last section of heat exchanging
equipment), the temperature of the flue gas ranges from 127 to 327 °C (261 to 62 PF).  Chemical
equilibrium calculations predict that gas-phase oxidation of Hg° to Hg2+ starts at about 677 °C
(1251 °F) and is essentially complete by 427 °C (801 °F). Based on these results, Hg should exist
entirely as Hg + downstream of the air heater. However, flue-gas measurements of Hg at air
heater outlets indicate that gaseous Hg° is still present at this location, and that Hg2+ ranges from
5 to 95 percent of the gas-phase Hg.  These data suggest that, due to kinetic limitations, the
oxidation of Hg° does not reach completion.

       As mentioned previously, gas-phase oxidation of Hg° is believed to take place via
reaction with gaseous Cl. At furnace flame temperatures, a major portion of the chlorine in the
coal exists as gaseous chlorine atoms, but as gas cools in post-combustion, the chlorine atoms
combine to form primarily hydrogen chloride (HC1) and minor amounts of molecular chlorine
(Ch). The rapid decrease in Cl concentration results in  "quenched" Hg2"1"  concentrations
corresponding to equilibrium values around 527 °C (981 °F).

       Figures 5-2 and 5-3  show predicted distributions of Hg species in  coal-fired electric
utility flue gas as a function of flue gas temperature. The predicted distributions are based on
equilibrium calculations of gas-phase oxidation of Hg° in flue gas from the combustion of a
bituminous coal' and a subbituminous coal2, respectively. Figure 5-2 shows that 80 percent of
gaseous Hg° is oxidized to HgCl2 by 527 °C (981°F).  Figure 5-3 indicates no oxidation of Hg° at
or above 527 °C (981°F). As mentioned above, the gas-phase oxidation of Hg° is believed to be
kinetically limited, proceeding only to equilibrium levels around 527 °C (981 °F).

       The difference in the equilibrium oxidation levels at 527 °C (800 K) in Figures 5-2 and
5-3 is attributed to the different chlorine levels in the model coals used in the calculations. The
calculated data in Figure 5-2 are based on a bituminous coal with a relatively high chlorine
concentration of several hundred parts per million by weight (ppmw). In contrast, the calculated
data in Figure 5-3 are based on a typical western subbituminous coal with a relatively low
chlorine content of 26 ppmw.  Research indicates that coals with relatively high chlorine
contents tend to produce more Hg2+ than coals with relatively low chlorine contents.3
                                           5-4
-------
             500
600
700
800
900
1000
1100
                                    Temperature, K
Figure 5-2.  Predicted distribution of Hg species at equilibrium, as a function of
temperature for a starting composition corresponding to combustion of a
bituminous coal (Pittsburgh) in air at a stoichiometric ratio of 1.2 (source:
Reference 2).
                                     5-5
-------
        100
           500
                     600
70O
BOO
                                                900
                           1000
11OO
                                   Temperature, K
Figure 5*3.  Predicted distribution of Hg species at equilibrium, as a function of
temperature for a starting composition corresponding to combustion of a
subbituminous coal (Powder River Basin) in air at a stoichiometric ratio of 1.2
(source:  Reference 2).
                                    5-6
-------
       In addition to being kineticaily limited by Cl concentration, recent research conducted at
EPA has found that gas-phase oxidation of Hg° is also inhibited by the presence of 862 and
water vapor.4  As shown in Figure 5-1, SO2 and water vapor are constituents in the flue gas from
coal-fired electric utility boilers. Figure 5-4 shows results from bench-scale experiments
examining the effects of SO2 and water vapor on the oxidation of gaseous Hg°. These
experiments were carried out using a simulated flue gas containing a base composition of 40
parts per million by volume (ppmv) Hg°, 5 mole % carbon dioxide (CO:), 2 mole % oxygen (O2),
and a balance of nitrogen (N2); the temperature of the flue gas was 754 °C (1,389 °F).  The
effects of SC>2, water vapor, and HC1 were studied by adding these constituents to the base flue
gas.  HC1 was added to the simulated flue gas at three concentrations typical of coal combustion
flue gas (50, 100, and 200 ppmv); SO2 and water vapor were added with the HC1 at 500 ppmv
and 1.7 mole %, respectively.

       As shown in  Figure 5-4, the oxidation of Hg° was inhibited by the presence of SO2 and
water vapor. HC1 is not believed to react directly with Hg° to cause its oxidation (a chlorinating
agent such as atomic chlorine or CI2 is needed). HC1 may produce trace quantities of the
chlorinating agent in the flue gas.  It is speculated that SO2 and water vapor may inhibit gas-
phase oxidation of Hg° by scavenging  the chlorinating agent.

       In addition to experimental studies, research has also been reported on the development
of a kinetic model that is used to better understand the reaction mechanism involved in gas-phase
Hg oxidation. A detailed chemical kinetics model using a chemical mechanism consisting of 60
reactions and 21 chemical species was developed recently to predict Hg speciation in combustion
flue gas.5 The speciation model accounts for the chlorination and oxidation of key flue gas
components, including Hg°. The performance of the model is very sensitive to temperature. For
low reaction temperatures (< 630 °C), the model produced only trace amounts of Cl and Clj from
HC1, leading to a drastic under-prediction of Hg chlorination compared with experimental  data.
For higher reaction temperatures, model predictions were in good accord with experimental data.
For conditions that produce an excess of Cl and C12 relative to Hg, chlorination of Hg is
determined by the competing influences of the initiation step, Hg + Cl -» HgCl, and the
recombination reaction, 2C1 —> C12.  If the Cl recombination is faster, Hg chlorination will
eventually be determined by the slower pathway Hg + C12 -> HgCl2.

       Another attempt has been made to formulate an elementary reaction mechanism for gas-
phase Hg oxidation.6 The proposed eight-step Hg oxidation mechanism quantitatively describes
the reported extents of Hg oxidation for broad ranges of HC1 and temperature.  In the proposed
mechanism, Hg is oxidized by a Cl atom recycle process, and, therefore, the concentrations of
both Cl and C12 are important. Once a pool of Cl atoms is established, Hg° is first oxidized by Cl
into HgCl, which, in turn, is oxidized by C12 into HgCl2. The second step regenerates Cl atoms.
Since the concentrations of Hg species are small in coal combustion flue gases, independent
reactions establish and sustain the pool of Cl atoms. The pool is governed by the chemistries of
moist CO oxidation, Cl  species transformations, and nitrogen oxide (NO) production.  The model
predictions show that O2 weakly promotes homogeneous Hg oxidation, whereas moisture is a
strong inhibitor as it inhibits the decomposition of HC1 to C12. NO was identified as an effective
inhibitor for Hg° oxidation through its  effect on reducing the concentration of hydroxyl (OH)
                                          5-7
-------
                          500 ppmv SO2
                          17% H2O

                          no$O2/H2O
500 ppmvSQ2
                        50          100          200
                          MCI Concentration (ppmv)
Figure 5-4. Effects of SO2 and water vapor on the gas-phase oxidation of Hg° at
754 °C and at three different HCI concentrations.
                                  5-8
-------
in the flue gas. The formation of HOC1 from OH and Cl is essential for the oxidation of Hg,
which oxidizes HgCl into HgCl2 and OH.  The elimination of OH via OH+NO+M = HONO+M
is believed to inhibit Hg° oxidation.

5.3.2  Fly Ash Mediated Oxidation

       In fabric filtration, flue gas penetrates a layer of fly ash as it passes through the filtering
unit. The intimate contact between the flue gas and the fly ash on the filter provides an
opportunity for the latter to oxidize some of the incoming gaseous Hg°. However, this
phenomenon does not occur across ESPs because the flue gas does not pass through a collected
layer of fly ash (see Chapter 3 for a description of the operation of FFs and ESPs).

       Certain fly ashes have been shown to promote oxidation of Hg° across a FF more actively
than others. For example, fly ashes from bituminous coals tend to oxidize Hg° at higher rates
than fly ashes from subbituminous coals and lignite. Differences in oxidation appear to be
attributable to the composition of the fly ash, the presence of certain flue gas constituents, and
the operating conditions of FFs.

       Bench-scale tests were conducted at EPA to investigate the effects of fly ash composition
and flue gas parameters on the oxidation of gaseous Hg0.4'7 In these experiments, a simulated
flue gas containing Hg° (and other species) was passed through a fixed bed of simulated or actual
coal fly ash, and oxidation of Hg° was measured across the reactor. Experimental results
indicated two possible reaction pathways for fly-ash-mediated oxidation of Hg°. One possible
pathway is the oxidation of gaseous Hg° by fly ash in the presence of HC1, and the other is the
oxidation of gaseous Hg° by fly ash in the  presence of NO\.  The research also reflected that the
iron content of the ash appeared to play a key role in oxidation of Hg°.  This EPA research is
described in the ensuing paragraphs.

       Coal fly ash is a mixture of metal oxides found in both crystalline and amorphous forms.
Glasses are common ash  constituents, composed primarily of the oxides of silicon and aluminum
(known as aluminosilicate glasses) that can contain a significant amount of cations such as iron,
sodium, potassium, calcium, and magnesium. Iron oxide (in the form of magnetite or hematite)
is also as commonly found in ash as calcium oxide and calcium sulfate. In the presence of
sufficiently high flue-gas concentrations of HC1 or Cl:, metallic oxides in fly ash may be
converted to metal chlorides such as cuprous chloride (CuCl).  Three-component model fly ashes
were prepared by adding Fe2O3 or CuO at  various weights to a base mixture of A^Os and SiO2.
An additional three-component fly ash was prepared by adding CuCl to a base mixture of AhOj
and SiOi. Municipal waste combustion fly ashes contain significant amounts of copper
compared to coal combustion fly ashes that contain only trace levels of copper. Model fly ashes
were prepared and tested in order to understand the effect of differences in copper content on the
oxidation of Hg°. Four-component fly ashes were prepared by adding various weights of CaO,
and FeaOa or CuO to a base mixture of A^Os and SKV Actual coal fly ashes were obtained
from the combustion of three different coals (two subbituminous and one bituminous) from a
pilot-size, pulverized-coal-fired furnace.
                                         5-9
-------
       Model flue gas compositions were simulated to represent the temperature and
composition of coal-fired electric utility flue gas as it enters a FF. The temperature of coal
combustion flue gas as it enters a FF typically ranges from 150 °C (302 °F) to 250 °C (482 °F).
Potentially important flue gas species (in terms of Hg° oxidation) include chlorine (primarily in
the form of HC1 at FF temperatures), NOx (primarily in the form of NO at FF temperatures), SO2,
and water vapor. The base flue gas consisted of 40 ppbv Hg°, 2 mole % O2, 5 mole % CO2, and
the balance N2 at a temperature of 250 °C (482 °F). HC1 (50 ppmv), NO (200 ppmv), SO2
(500 ppmv), and/or water vapor (1.7 mole %) were added to the base gas to determine their
effect on oxidation. About 10 percent of NO2 (10 ppmv) was measured when 200 ppmv of NO
was added to the base flue gas which contains 2 mole % of O2. The mixture of NO and NO2 in
flue gas is referred lo collectively as NOx- Table 5-1 shows the simulated and actual fly ashes
and simulated flue gas tested.

       Oxidation Behavior of Model Fly Ashes. HC1 and NOx were identified as the active
components in flue gases for the oxidation of Hg°. NOx were more active than HC1. Cupric oxide
(CuO) and ferric oxide (Fe2O3) were identified as the active components in model fly ashes for
Hg° oxidation. In the presence of NOx, inert components of model fly ashes such as alumina
(AhOs) and silica (SiO2) appeared to become active in oxidation of Hg°. Steady-state oxidation
of Hg° promoted by the four-component model fly ashes (containing calcium oxide, CaO) was
reached at much slower rates compared to those obtained using the three-component model fly
ashes that contained no CaO (Figures 5-5 and 5-6). The partial removal of gas-phase HC1 by
CaO in the CaO-containing model fly ashes may have reduced the available chlorinating agent
and resulted in slower oxidation of Hg°.

       Oxidation Behavior of Actual Coal Fly Ashes.  As shown in Table 5-1, the Blacksville fly
ash (derived from a bituminous coal) completely  oxidized Hg° in the presence of NO (base +
NO), but showed little oxidation in the presence of HC1 (base + HC1).7 The Comanche fly ash
(derived from a  subbituminous coal) did not oxidize Hg° in the presence of NO or HC1. The
Absaloka coal (derived from a subbituminous coal) showed 30 to 35 percent oxidation of Hg° in
the presence of NO, but no oxidation in the presence of HC1. It is believed that the high
reactivity of the Blacksville coal in NO  is related to its relatively high Fe2O3 concentration (22
percent); this observation is in agreement to that seen for the high iron (approximately 14
percent) three- and four-component model fly ashes.

       More tests were conducted recently at EPA on actual fly ash samples with different coal
ranks and iron contents in order to get a better understanding of the effects of iron in coal fly
ashes on speciation of Hg.8 It was observed that one subbituminous (3.7 percent iron) and  three
lignite coal fly ash (1.5 to 5.0 percent iron) samples tested with low iron content did not oxidize
Hg° in the presence of NO and HC1. However, a bituminous coal fly ash sample (Valmont
Station) with a low iron content (2.3 percent iron) completely oxidized Hg° in the presence of
NO and HC1.  It was also found that, upon adding Fe2Os to the low iron content  subbituminous
and lignite fly ash samples to reach an iron content similar to that of the Blacksville sample,
significant Hg° oxidation reactivity was measured (33 to 40 percent oxidation of Hg°) for these
iron-doped samples.
                                         5-10
-------
            c
            o
            X
            o
              100
               80
60
               20
                                 4-component 3-component
                                    10     20         50

                                    exposure tim e (m in)
                                              100
200
                     3-Component: silica/alum In a (3.5/1) and 14 wt% Fe203

                     4-Component silica/alumina (3.5/1), 13 wt% Fe2O3. and 6 v\t% CaO
Figure 5-5.  Hg° oxidation in the presence of the three- and four-component model

fly ashes containing iron at a bed temperature of 250 °C (source: Reference 4).
                                      5-11
-------
                          100
                           BO
                       •    60
                       (5
                       x
                       o
                           40
                           20
                                            4-component 3-component
                                               10      20        50

                                               exposure time (min)
100    200
                                  3-Component silica/alum in a (3.5/1) and 1 wt%CuO

                                  4-Component: silicatelumina(3.5/1), 1 wt%CuO,and Bv»t%CBO
           Figure 5-6.  Hg° oxidation in the presence of the three- and four-component model

           fly ashes containing copper at a bed temperature of 250 °C (source: Reference 4).
-------
Table 5-1. Percent oxidation of Hg° by simulated and actual coal-fired electric
utility boiler fly ash (source: Reference 4).
Fly Ash Composition
(by weight percentages)
2-Component Model Fly Ash
22% AI2O3 + 78% Si02
% Oxidation of Hg" by fly ash
Base*
Base
+ HCI
Base
+
HCI,
S02
Base
+
HCI.
soz,
HP
Base
+ NO
Base
+ NO,
SO,

b
0


39
4
5-Component Model Fly Ashes
19% AI2O3, + 67% SiO2 + 14% Fe2O3
22% AI2O3 + 77% Si02 + 1 % Fe2O3
22% AI2O3 + 78% SiO, + 0,1% Fe203
22% AI2O3 •»• 77% SiO2 + 1% CuO
22% AI2O3 + 78% SiO2 + 0. 1 % CuO
22% AI203 + 72% SiO2 + 7% CaO
22% AI2O3 + 78% SiOz + 0.1% CuCI
4-Component Model Fly Ashes
21% AI2O3 + 71% Si02, + 1% CuO + 7% CaO
18% AI2O3, + 63% SiO2 + 13% Fe203 + 6% CaO
Actual Fly Ash Samples
Blacksville coal fly ash (bituminous)
22% Fe A- 6% CaO
Comanche coal fly ash (subbituminous)
5% Fe,O3. 32% CaO
Absaloka coal fly ash (subbituminous)
4% Fe203, 24% CaO
0




0
87
92
67
15
93
92
0
77
88
43

89
86
13

54
37

84
63
14
23
93
48
11
70
35
0.86

80
26
3
16
3
13




91
87
82
93
43
49








6
0







100
0
30-35



   (a) Base gas consisted of 40 ppbv Hg°, 2 mole% O2, 5 mote% C02, and balance N2 at a temperature of 523 K.
       HCI, NO, SO2, and water vapor were added to the base gas in the following concentrations SO ppmv, 200
       ppmv, 200 ppmv, and 1.7 mole%, respectively.
   (b) Blank cells mean test not conducted.
                                           5-13
-------
       The physical, chemical, and carbon properties of the Blacksville and Valmont samples
were also characterized.  It was found that the two fly ash samples have different unburned
carbon contents (3.4 percent for Valmont and 16.8 percent for Blacksville). Based on this
finding, it appears that iron content may not be the only ash-related factor that affects the Hg°
oxidation reactivity of bituminous coal fly ashes.  The effect of physical properties, such as
surface area, and the effects of chemical properties, such as sodium content and alkalinity, in the
fly ash may also determine the propensity of different fly ashes to oxidize Hg in flue gas.
       Research for obtaining a better understanding of the roles of NO\ and Fe2O3 in the
heterogeneous oxidation of Hg° was reported recently by UND/EERC.9 In UND/EERC's
reported research, the effects of NOX and hematite (a-FezOs) on Hg transformations were studied
by injecting them into actual coal combustion flue gases produced from burning bituminous
(Blacksville), subbituminous (Absaloka), and lignite (Falkirk) coals in a 7-kW combustion
system. It was found that the Blacksville fly ash has high Fe2O3 content (12.1 percent), and the
Absaloka and Falkirk fly ashes have significantly lower Fe2O3 contents (4.5 and 7.9 percent,
respectively). Portions of the FeiOs in Blacksville and Falkirk fly ashes are present as
maghemite (y-Fe2O3), and a portion of the Fe2O3 in Absaloka fly ash is present as hematite (a-
FezOs). The flue gas generated from the combustion of Blacksville coal contained Hg2+ as the
predominant Hg species (77 percent), whereas Absoloka and Falkirk flue gases contained
predominantly Hg° (84 and 78 percent, respectively).  Injections of NC>2 (80 to 190 ppm) at 440
to 880 °C and a-FesOa (6  and 15 percent) at 450 °C  into Absoloka and Falkirk coal combustion
flue gases did not change  Hg speciation. The UND/EERC researchers suggested that the lack of
transformation from Hg° to Hg2+ in the 7-kW combustion system was possibly due to
components of either Absoloka and Falkirk coal combustion flue gases, or their fly ashes,
inhibiting the (x-Fe2O3 catalyzed heterogeneous oxidation of Hg° by NO\. The researchers also
believed that an abundance of Hg2+ in Blacksville coal combustion flue gas and Y-Fe2O3 in the
corresponding fly ash, and the inertness of injected a-Fe23 with respect to heterogeneous Hg°
oxidation in Absoloka and Falkirk  flue gases, are indications that y-Fe2C>3  rather than a-Fe2O3
catalyzes Hg2+ formation.

       A study of the role of fly ash in the speciation of Hg in coal combustion flue gases was
reported by Iowa State University.10 In this study, bench-scale laboratory tests were performed
in a simulated flue gas stream using two fly ash samples obtained from the ESPs of two full-
scale coal-fired electric utility boilers. One fly ash was derived from burning a western
subbituminous coal (Powder River Basin) while the other was derived from an eastern
bituminous coal (Blacksville). Each of the two samples was separated into three subsamples
with particle sizes greater than 10,  3, and 1 um using three cyclones. The amount of sample
collected in these three size ranges was 85 to 90 percent, 10 to 15 percent, and 1 percent of the
total ash, respectively. Only the two largest sized subsamples were tested for Hg° oxidation
reactivity. The Blacksville sample was also separated into strongly magnetic (20 percent),
weakly magnetic (34 percent), and nonmagnetic (46 percent) fractions using a hand magnet for
testing Hg° oxidation reactivity of the individual fractions. Since magnetism of the fly ash
samples is contributed mainly by iron oxides in the samples, the iron oxide content of the
magnetically separated samples is in the following order: strongly magnetic > weakly magnetic >
nonmagnetic. The low iron content PRB fly ash is nonmagnetic and was not magnetically
                                          5-14
-------
separated for testing. Scanning electron microscopy with energy-dispersive x-ray analysis
(SEM-EDX) was used to examine the surface morphology and chemical composition of the fly
ash samples. X-ray diffraction (XRD) was also used to examine the mineralogical composition
of the whole and fractionated fly ash samples.  XRD identifies only crystalline components of
the samples. This is important since coal combustion fly ashes typically contain a considerably
amount of glassy, amorphous material.

       It was observed that, although the fly ashes tested were chemically and mineralogically
different, there were no large differences in the catalytic potential for oxidizing Hg°.10  The
Blacksville fly ash tended to show somewhat more catalytic reactivity (16 to 19 percent Hg°
oxidation) than the PRB fly ash (4 to 10 percent Hg° oxidation). The researchers of this project
suggested that the difference in reactivity could be due largely to the larger surface area (3.4
nr/g) of the Blacksville fly ash compared to that (1.5 m2/g) of the PRB fly ash. It was found
from the SEM-EDX analyses that the iron-rich (highly magnetic) phases in the greater than 10
um size fraction of the Blacksville sample contained about 25 percent (atomic) Fe, 10 percent
each of Al and Si, 2 percent Ca, and lesser amounts of Na, S, K, and Ti. The nonmagnetic
Blacksville fly ash fraction in the greater than 10 urn size range contained only 4 percent Fe, 10
percent Al, 20 percent Si, and  lesser amounts of Na, S, K, and Ti.  For the PRB fly ash (all
nonmagnetic), both the greater than 10 urn and greater than 3 Jim fractions contained about 3
percent Fe, 10- 20 percent Al and Si, about 10 percent Ca, and 2 percent or less of Mg,  S, K, and
Ti. The XRD results showed that the whole Blacksville ash contained primarily quartz (SiCy,
mullite (3Al2O3-2SiO2), magnetite (Fe3O4), hematite (Fe2O3), and a trace of lime (CaO). The
PRB fly ash contained mostly  quartz and lesser amounts of lime, periclase (MgO), and calcium
aluminum oxide (CajA^Oe). No magnetite or hematite was found in this ash.  It is interesting to
note that the nonmagnetic fractions  actually showed substantially higher amounts of oxidized Hg
than the magnetic fractions. The reported test results of this study indicated that the nonmagnetic
fraction resulted in 24 percent of the Hg being oxidized, while 3 percent of the Hg oxidized when
using the magnetic ash. It has been suspected that the magnetic (iron-rich) fraction in fly ash
would be more catalytic than the nonmagnetic (aluminosilicate-rich) fraction because of its
mineralogy (predominantly iron oxides), and possibly because the magnetic phase tends to be
enriched in transition metals that could also serve as Hg° oxidation catalysts. However, under
the experimental conditions employed in this study, the test results do not support this.  It was
found that the surface area of the nonmagnetic fraction is about four times that of the magnetic
fraction.  From this study it appears that surface area is a dominant factor in determining the
ash's Hg° oxidation reactivity.

       Because major differences were not observed with the two fly ashes, a set of tests
involving a full factorial design was conducted using only the Blacksville fly ash in order to
apply statistical techniques for identifying the important factors in determining Hg° oxidation.10
The statistical analysis results  indicated that the composition of the simulated flue gas used in the
tests and whether or not ash was present in the gas stream were the two most important factors.
The presence of HC1, NO, NO2, and SOj and all two-way gas interactions of the four gases listed
above were found statistically  significant for Hg° oxidation. The HC1, NO:, and SO2 appeared to
contribute to Hg° oxidation, while the presence of NO appeared to suppress Hg° oxidation. NO2
was found to be the most important  of the four reactive gases tested; next were HCl and NO.
                                          5-15
-------
However, the effect of NO depended on whether NOj was present. Although the presence of
NOz was statistically significant as a main factor, it was found more important in its interactions
with other gas components. Based on the statistical analysis results, the researchers of this
project concluded that the interactions of flue gases with fly ash to cause Hg° oxidation are
extremely complex, and the difficulty in understanding the Hg chemistry in coal combustion flue
gases is not surprising.  It is noted that the EPA study showed significant Hg oxidation reactivity
of the Blacksville ash, while studies at UND/EERC and Iowa State University show little Hg
oxidation reactivity of Blacksville ash. Since the ash samples used in the above studies were
generated at three different plant operating conditions, these conditions may play an important
role in contributing to the reactivity of the ashes.

5.3.3 Oxidation by Post-combustion NOx Controls

       There are indications that post-combustion NOX controls SCR and SNCR may oxidize
some of the Hg° in the flue gas of a coal-fired electric utility boiler. The research on this issue is
ongoing. For current understanding of this subject, the reader is referred to Chapter 6.

5.3.4 Potential Role of Deposits, Fly Ash, andSorbents on Mercury Speciation

       Gaseous Hg (both Hg° and Hg2+) can be adsorbed by the solid particles in the coal-fired
electric  utility boiler flue gas. Adsorption is the phenomenon where a vapor molecule in a gas
stream contacts the surface of a solid particle and is held there by attractive forces between the
vapor molecule and the solid. Solid particles are present in all coal-fired electric utility boiler
flue gas as a result of the ash that is generated during combustion of the coal. Ash that exits the
fiirnace with the flue gas is called fly ash. Other types of solid particles may be introduced into
the flue gas stream (e.g., lime, powdered activated carbon) for pollutant emission control.  Both
types of particles may adsorb gaseous Hg in the boiler flue gas. This section addresses the
adsorption of gaseous Hg by fly ash. Adsorption of Hg by sorbent particles introduced into the
flue gas stream and subsequently captured in a downstream PM control device is discussed in
Chapter 6 as related to specific control technologies that may be implemented to increase overall
Hg removal  from the boiler flue gas.

       Gaseous Hg can be adsorbed by fly ash in the flue  gas (sometimes called "in-flight"
adsorption).  In-flight adsorption of gaseous Hg by fly ash occurs in the post-combustion region
where the flue gas contains its highest concentration of fly ash (i.e., prior to the first PM control
device).  The type of coal from which a fly ash originates appears  to strongly influence its ability
to adsorb Hg.  Pilot-scale '' and field data12 have indicated that fly ashes from subbituminous
coals (specifically, those from the Powder River Basin in Wyoming) adsorb more gaseous Hg
than fly ash from lignite and bituminous coals. Test data show 30 percent in-flight adsorption of
gaseous Hg by fly ashes from boilers burning these subbituminous coals compared to  10 to 20
percent adsorption by the fly ashes from boilers burning lignite or bituminous coals. It has been
suggested that the measured removals of Hg by fly ash can be inflated based on the sampling
method, but  in most cases are below 15 percent. General trends indicate that in-flight  field
capture  of Hg from combustion of subbituminous coals is  higher than from combustion of
bituminous coals.13
                                          5-16
-------
       The carbon content of fly ash is another parameter that may influence adsorption of
gaseous Hg (the carbon in fly ash is unburned coal). Conditions that result in increased amounts
of carbon in fly ash tend to increase the amount and subsequent capture of particle-bound Hg.
Hg has been found to concentrate in the carbon-rich fraction of fly ash.14'15  For similar coals,
both laboratory16 and pilot- and large-scale data " have shown a positive correlation between
adsorption of gas-phase Hg and carbon content in fly ash. A research project conducted at full-
scale coal-fired electric utility boilers in Colorado indicates that certain fly ashes adsorb
significant levels of Hg from flue gas. Chapter 7 describes the methodology and results of this
study in detail. Many of these fly ashes have carbon content greater than 7 percent, but one low-
carbon content fly ash has also  been identified. This research project and the possibility of using
fly ash re-injection for Hg control is discussed in Chapter 6.

       Gaseous Hg also can be adsorbed by fly ash collected on the surface of a FF. In a FF,
there is contact of gaseous Hg in the flue gas with the collected layer of fly ash on the FF bags as
the gases flow through the FF.  Pilot-scale tests of a low-carbon fly ash (less than 0.5 percent
carbon) showed that the fly ash adsorbed 65 percent of the gaseous Hg° entering a FF; the data
indicate that fly ash properties other than just carbon content may affect adsorption.  The tested
fly ash was produced from the combustion of a subbituminous coal from the Powder River Basin
in Wyoming. Western subbituminous coals generally contain high concentrations of CaO and
tend to adsorb high levels of Hg°. At this time, the mechanisms by which these Western coals
adsorb Hg° are not known; however, the CaO content may be a factor.  It has been shown in a
pilot-scale study that combustion of western coals tends to produce relatively high particle-bound
Hg emissions.1
5.4 Capture of Mercury by Sorbent Injection

       Mercury can be captured and removed from a flue gas stream by injection of a sorbent
into the exhaust stream with subsequent collection in a PM control device such as an electrostatic
precipitator or a fabric filter.  The implementation of this type of Hg control strategy requires the
development, characterization, and evaluation of low-cost and efficient Hg sorbents.
Experimental methods for characterization and evaluation are presented below.  Further, efforts
to develop better sorbents, with greater capacity and lower cost, are also discussed.

5.4.1  Sorbent Characterization

       Sorbents are characterized by their physical and chemical properties.  The most common
physical characterization is surface area. The interior of a sorbent particle is highly porous.  The
surface area of sorbents is determined using the Brunauer, Emmett, and Teller (BET) method of
N2 adsorption.18 Nitrogen is adsorbed at the normal boiling point of-195.8 °C and the surface
area is determined based on mono-molecular coverage. Surface areas of sorbents range from 5
m2/g for Ca-based sorbents to over 2000 m2/g for highly porous activated carbons. Mercury
capture often increases with increasing surface area of the sorbent.  However, recent researchl9
has suggested that pore surface area in the micropores is more important than the total surface
area for the removal of part per billion concentrations of Hg from coal combustion flue gases.
                                          5-17
-------
       Particle size distribution is another physical characteristic that is used to describe
sorbents. Activated carbons that are used for Hg control are powdered with a size on the order of
44 ^m or less. Particle size is measured using sieves or a scanning electron microscope (SEM).
Generally, the smaller the particle size of an activated carbon, the better the access to the surface
area and the faster the rate of adsorption kinetics. Careful consideration of particle size
distribution can provide significant operating benefits, both in fabric filter applications, where
pressure drop must be considered, and in ESP (or duct injection) applications, where mass
transfer limitations in the short residence time mean that adsorption is a function of sorbent
particle size.

       Determination of the pore size distribution of an activated carbon is an extremely useful
way of understanding the performance characteristics of the material. Pore sizes are based on the
diameter of the pore and are categorized using the following IUPAC conventions: micropores
<2 ran, mesopores 2-50 nm, and macropores >50 nm. Micropore volume can be estimated from
CC>2 adsorption at 273 K using the Dubinin-Radushkevich (DR) equation. Total pore volume
can be determined using N2 adsorption.

       Some of the chemical properties of activated carbons that influence Hg capture include
sulfur content, iodine content, chlorine content, and water content.  Functional groups of a
sorbent have been shown to play an important role in adsorption behavior. Many carbon-oxygen
functional groups have been identified in activated carbon including carbonyl, carboxyl, quinone,
lactones, and phenol groups. Many methods have been used to study the  functional groups
present in carbonaceous materials including neutralization of bases, direct analysis of the oxide
layer by chemical reaction, infrared spectroscopy, and x-ray photoelectron spectroscopy. For
example, specific surface oxygen functional groups can be estimated by using the data measured
from the base titration based on the  following assumptions: NaHCOs titrates carboxyl groups;
NaOH titrates carboxyl, lactone, and phenol groups; CO2 is a decomposition product of carboxyl
and lactone groups; and CO is a decomposition product of phenol and carbonyl groups.20 The
NaOH and HC1 titration  values can estimate the acidity and basicity of a carbon, respectively.

5.4.2  Experimental Methods Used in Sorbent Evaluation

       In order to evaluate the performance of a specific Hg sorbent, several types of
experimental reactors are used. The first step is testing in a bench-scale reactor system, which
may be a fixed-bed, entrained-flow, or a fluidized-bed system.  Sorbents that perform well in
bench-scale tests are then tested in a pilot-scale system and may eventually be tested in a full-
scale system. These systems are discussed below.

5.4.2.1  Bench-scale Reactors

       Bench-scale reactors are the smallest category of reactors, hence the term "bench-scale."
There are several types of bench-scale reactors that are used to evaluate Hg sorbents. The first
type that will be discussed is a fixed-bed or packed-bed system. This type of reactor simulates
Hg° capture that would occur in a FF. Another type of bench-scale reactor is an entrained-flow
                                          5-18
-------
reactor, which simulates in-flight capture of Hg° upstream of an ESP. It is important to highlight
the major differences between these two reactors as shown in Table 5-2.

       Fixed-bed Reactor. A schematic of the experimental apparatus used by EPA to study the
capture of Hg° and HgC^ is shown in Figure 5-7. A detailed description of the apparatus can be
found elsewhere.21 In this system the Hg vapor generated is carried into a manifold by a nitrogen
stream where it is mixed with SO2, HC1, CO2,62, and water vapor (as required by each
particular experiment). The sorbent to be studied (approximately 0.02 g diluted with 2 g inert
glass beads; bed length of approximately 2 cm) is placed in the reactor and maintained at the
desired bed temperature by a temperature controller. A furnace kept at 850 °C is placed
downstream of the reactor to convert any Hg2+ (as in HgCk) to Hg .  According to
thermodynamic predictions, the only Hg species that exists at this temperature is Hg0.22 Quality
control experiments, in the absence of HC1 in the simulated flue gas, also showed that all the
HgCh could be recovered as Hg° across this furnace. The presence of the furnace enables
detection of non-adsorbed HgCh as Hg° by the on-line ultraviolet (UV) Hg° analyzer, thus
providing actual, continuous Hg° or HgCl2 capture data by the fixed bed of sorbent. The UV Hg°
analyzer used in this system responds to  862 as well as Hg°.  Signal effects due to SO2 are
corrected by placing an on-line SO2 analyzer (UV) downstream of the Hg° analyzer and
subtracting the measured SCh signal from the total response of the Hg analyzer; the SOa analyzer
is incapable of responding to Hg in the concentration range generally used.

       In each test, the fixed bed is exposed to the Hg-laden gas for 7 hours or until 100 percent
breakthrough (saturation) is achieved (whichever comes first). During this period the  exit
concentration of Hg is continuously monitored.  The instantaneous removal of Hg° or HgCb at
any time (t) is obtained as follows:

       Instantaneous removal at time t (%) = 100*[(mercury)in-(mercury)0uJ/(mercury)iR.

       The specific amount of Hg uptake (q, cumulative removal up to time t; weight Hg
species/weight sorbent) is determined by integrating and evaluating the area under the removal
curves. Selected experiments conducted using this experimental setup have been run in duplicate
and indicated a range of+10% about the mean in the experimental results.  It was found that
differences in equilibrium Hg°/HgCl2 capacities, at 200-300 mg/Nm3 inlet concentration, are
statistically significant if the Hg°/HgCl2 capacities are at least + 10 percent different from one
another.

       Entrained-flow Reactor.  An example of a bench-scale entrained-flow reactor23 is shown
in Figure 5-8. This EPA reactor is constructed of quartz and is 310.5 cm long with an inside
diameter of 4 cm. Three gas-sampling ports are located along the length of the reactor and are
labeled SP1, SP2, and SP3.  The reactor is heated with three Lindberg, 3-zone electric furnaces
in series. The baseline Hg° concentration is measured in the absence of activated carbon using
an ultraviolet (UV) analyzer (Buck Scientific, model 400A).  Once the baseline is established,
activated carbon is fed into the top of the reactor using a fluidized-bed feeder (0.2-0.5 std.
L/min). The gas-phase Hg° concentration is then measured at one of the sample ports by pulling
a gas sample (0.5 std. L/min) through a 1 urn filter to remove any particles, then through a
                                          5-19
-------
Table 5-2. Comparison of bench-scale fixed-bed with entrained-flow reactors.
Test Condition
Simulation of capture in
Sorbent exposure
Sorbent evaluation based on
Fixed-Bed Reactor
Fabric filter
Minutes/Hours/Days
Breakthrough or uptake capacity
Entrained-Flow Reactor
Upstream of an ESP
Less than 4 seconds
Reactivity
                                    5-20
-------
                     Mercury Generation Carbon Trap  Manifold
                         System
           Carrier N
             Purge N
            Carbon Trap
                    Rotameter
Figure 5-7. Schematic of bench-scale fixed-bed reactor (source: Reference 21).
                                         5-21
-------
   Fluidized Bed Feeder
CH<
Air
       Lmdberg
       3-Zone
       Furnaces
-SP2


-SP3-
                                                Hg°/Ni
Buck Hg   SOi/Oi
Analyzer   Analyzer
                                 Filter  Reducing  Nation
                                      Furnace  Drier
                                               Exhaust
                                 Carbon
                                 Trap
Figure 5-8. Schematic of bench-scale flow reactor with methane burner (source:
Reference 23).
                                           5-22
-------
reducing furnace to convert any oxidized Hg to Hg°. The reduction method is described
elsewhere.21 After the reducing furnace, the gas is dried using a Nation® gas sample dryer
(Perma Pure, Inc.) and is finally sent to a Buck analyzer.

       Initial tests are conducted using nitrogen (N2) as the carrier gas with later tests performed
in a flue gas from a methane flame. In the N2 carrier gas tests, industrial grade N2 (1 std. L/min)
flows over a Hg° permeation tube that is housed in a permeation oven (VICI Medtronic's, model
190) to generate a Hg°-laden gas stream.  The Nj/Hg stream is diluted with a second N2 stream
(12 std. L/min) to the desired concentration before entering the top of the reactor. Other gases
(SC>2, NOX, O2, water vapor) can be blended into the N2 carrier gas in the mixing manifold.

       A fluidized-bed feeder is used to inject sorbent into the reactor.  An inlet line of N2 is
used to fluidize and carry the activated carbon to the reactor. The carbon feed rate is adjusted by
varying the amount of N2 (0.2 to 0.5 std. L/min) entering the feeder.

       Because the UV analyzer used to detect Hg° is sensitive to particles, a filter is used to
remove any carbon that may have been carried with the gas. Tests have been conducted to
determine if carbon particles accumulate on the filter, as this would act like a packed bed and the
reactor's removal of Hg° would be a combination of in-flight and filter (packed-bed) capture.  In
these tests, activated carbon was injected in the absence of Hg°, and a gas sample was pulled
through the filter. After 1 minute, Hg° was added to the gas stream to see if there was a lag in
the time it takes for the baseline to return. The results were the same as for a blank filter,
suggesting that the filter does not have an effect on the results.

       The total flow through the reactor is typically 13 std. L/min, which gives residence times
of 5.2,11.5, and 17.7 s at ports SP1, SP2, and SP3, respectively.  The velocity of the particles
through the reactor is assumed to be the same as that of the gas flow since the terminal velocity
of the particles is smaller than the velocity of the gas through the reactor by a factor of 3.

       Fluidized-bed Reactor. Another type of bench-scale reactor that is used to evaluate
sorbents is a fluidized-bed reactor,24 shown in Figure 5-9. The advantage of this type of reactor
is the extended contact time between the sorbent and the Hg-laden gas.  Bench-scale Hg removal
tests can be performed on a fluidized-bed reactor apparatus, hi a typical experiment, an
Hg/NO/S02 mixture, nitrogen, and dry air are metered through rotameters to produce 12 scfh of
a dry simulated flue gas of 300 ppmv NOx, 600 ppmv SC>2, 8 percent C>2, and varying Hg
concentrations.  This gas is preheated to reaction temperature (80 °C) and humidified with
vaporized water to an average 10.5 mol % water. The resulting wet simulated flue gas is then
passed through a vertical reactor loaded with fluidized sorbent and sand, and then passed through
a filter to remove any entrained particulate to protect the downstream equipment. The reactor
and filter assembly are housed in an oven maintained at 80 °C. The test stand is equipped with a
bypass of the reactor and filter assembly to allow for bias checks.  Sorbent is exposed to
simulated flue gas for 30 minutes.  Water is removed from the  spent flue gas with a NAFION™
Dryer. Dry gas is then serially analyzed with  Hg, SO2, and NOX continuous emission monitors
(CEMs).
                                          5-23


-------
   Rotameters
   Nitrogen
   Air
Figure 5-9. Schematic of bench-scale fluidized-bed reactor system (source:
Reference 24).
                                      5-24
-------
5.4.2.2  Pilot-scale Systems

       Initial design and testing is done in bench-scale reactors. Once the fundamentals of Hg
capture have been tested in a bench-scale system, the next step is to move up to a larger or pilot-
scale system. The main difference between bench- and pilot-scale systems involves testing
sorbents in a more realistic situation involving coal combustion flue gas. This gas is generated in
a pilot-scale combustor that contains a FF or ESP for participate control. An example of this is
the pilot-scale combustor operated by DOE (see Figure 7-3). This system burns coal at a rate of
500 Ib/hr and is equipped with a FF. Sorbents, such as activated carbon, are injected upstream of
the PM control device. Mercury removal is determined by gas-phase sampling upstream of the
sorbent injection point and downstream of the PM control device.

       Pilot-scale Hg removal can also be examined using a flue gas slipstream from a full-scale
unit. An ESP or FF is attached to the slipstream and tested. A portable FF was developed by
EPRI and called a COHPAC (COmpact Hybrid PArticulate Collector) unit.26  This unit was
tested for Hg removal using activated carbon.  The URS Corporation (formerly Radian
International) also developed a reactor system that uses a slipstream of actual flue gas withdrawn
from a power plant to evaluate sorbents or catalysts in a fixed bed.27 It should be noted that the
slipstream reactor, which uses actual coal combustion flue gas, does not always produce the
same Hg captive behavior of a sorbent that a similar laboratory system does using simulated flue
gas.28  It is important to perform pilot-scale tests prior to conducting full-scale tests to eliminate
uncertainties and costly redesign of a process. With the data collected in the pilot-scale studies,
full-scale tests can be initiated.

5.4.2.3  Full-scale Tests

       Most of work to date in Hg control has been done in bench- or pilot-scale systems. These
reduced-scale systems provide insight into many issues, but cannot fully account for the impacts
that additional control technologies have on plant-wide equipment. Therefore, it is necessary to
scale up and perform full-scale tests to document  actual performance in a full-scale boiler. These
tests are based on the results obtained in bench- and pilot-scale tests.  Screening tests in bench-
and pilot-scale systems identify sorbents that are effective in capturing Hg. These sorbents are
then tested in a fiill-scale coal-fired electric utility power plant to determine full-scale
performance.

       Each full-scale unit is unique in terms of the pollution control equipment that is present
as well as the operating conditions.  Some of the factors that are evaluated include:

       •   Type of particulate control equipment that is used (ESP or FF),

       •   Impact of cake thickness and cleaning frequency in a FF, and

       •   Removal of Hg by the fly ash in the system. Subbituminous coal ashes have been
           shown to be effective in capturing Hg.
                                          5-25
-------
5.4.3  Research on Sorbent Evaluation

5.4.3.1 Sorbent Evaluation Using Enhanced-flow Reactors

       A flow reactor was designed to simulate Hg° capture through a duct or ESP and to obtain
kinetic rate constants for the adsorption of Hg° onto sorbents.  Several researchers have predicted
that, under certain conditions, dispersed-phase capture would be limited by mass transfer.29'30
Calculations were performed to determine the required operating conditions to minimize external
mass transfer effects in the flow reactor, and experimental tests were performed to verify these
calculations.23'31'32  The first test involved changing the diffusion coefficient by changing the gas
in the system from N2 to helium (He) and to argon (Ar) while holding all other parameters
constant (particle size, residence time, temperature, and Hg° concentration). The diffusion
coefficient increased by an order of magnitude by changing the gas from N2 to He.  Using a
lignite-based commercially available carbon (Norit FGD) at 100 °C and a Hg° concentration of
86 ppb, Hg° removal was 6 percent at a carbon to Hg ratio (C:Hg) of 1,500:1 and increased to 30
percent at a C:Hg of 8,000:1. Experimental results were similar when He was used as compared
to N2.  If external mass transfer were controlling, then a higher Hg° removal would have been
obtained using He, since the mass transfer coefficient increased.

       A second test involved using two commercially available activated carbons, Norit FGD
and Calgon WPL at 100 °C and 124 ppb Hg° in dry N2. Removal for the FGD carbon ranged
from 9 percent (C:Hg=2200:1) to 23 percent (C:Hg=6400:1).  Removal for the WPL carbon
ranged from 11 percent (C:Hg=340) to 94 percent (C:Hg=5000:l). If dispersed-phase capture in
the flow reactor were film-mass-transfer limited, the two activated carbons would have removed
similar amounts of Hg°at  a given C:Hg, assuming each carbon had sufficient Hg° capacity.

       The flow reactor has been used to  examine the effect of temperature, particle size,
residence time, carbon type, and gas composition on Hg° removal.3 "33 The effect of particle
size on Hg° removal for Darco FGD at 100 °C and a Hg° concentration of 86 ppb is shown in
Figure 5-10. Several particle sizes (4-8, >8-16, >16-24, and >24-44 [im) were injected into the
flow reactor at C:Hg ratios ranging from 2000 to 11,000:1.  The gas was sampled at SP2,
resulting in a gas contact time of 8.4 s.  Figure 5-11 shows that greater Hg° removal is achieved
by increasing the feed rate and by decreasing the particle size. At a C:Hg of 5000:1, a 5 percent
reduction was obtained with the >24-44 um size fraction as compared to a 20 percent reduction
with the 4-8 Jim fraction.  Thus by using a smaller particle a higher removal can be obtained at a
given C:Hg. Both external and internal mass transfers are dependent on particle size: the effect
of mass transfer increases with an increase in particle size.

5.4.3.2 Sorbent Evaluation Using Packed-bed Reactors

       Recent bench-scale studies at the University of North Dakota's Energy and
Environmental Research Center (UND/EERC) have focused on the interactions of gaseous flue
gas constituents on the adsorption capacity of activated carbon for Hg.34 Bench-scale studies
were performed using a fixed bed of carbon. The tested carbon was a commercially available
lignite-based activated carbon (LAC) commercially known as Darco FGD™ from Norit
                                          5-26
-------
                         2000      4000       6000      8000

                                    Carbon to Mercury Ratio
10000
12000
Figure 5-10.  Effect of particle size on adsorption for Darco FGD at 100 °C,
86 ppb Hg° concentration, and 8.4 s contact time (source: Reference 31).
                                      5-27
-------
Americas, Inc. A simulated flue gas containing a nominal concentration of 15 fig/Nm3 of
gaseous Hg° was passed through the fixed bed of carbon. In addition to Hg, the baseline test gas
contained 6 percent C>2, 12 percent COi, 8 percent H^O, and the balance N2.  Various flue gas
constituents (SC>2, HC1, NO, and NOa) were added individually and in combination to the
baseline test gas to determine the effects of flue gas constituents on Hg adsorption.  Temperature
effects were also examined. Table 5-3 shows the various compositions of gas tested.

       For each adsorption test, a Hg CEM was used to monitor total or elemental Hg.
Measurements were alternated between the inlet and outlet locations of the test bed. For a given
test, measurements look place primarily at the outlet location; however, occasionally the inlet
location was tested lo confirm that a constant concentration of gaseous Hg°  was entering the test
bed. For each test, the analyzer was set to measure total gaseous Hg at the outlet; however,
occasionally the analyzer was set to measure only gaseous Hg° at the outlet.  The purpose of
measuring only gaseous Hg° at the outlet was to determine if any incoming gaseous Hg° was
being oxidized by carbon in the bed (evident if the concentration of gaseous Hg° in the outlet gas
was less than the concentration of total gaseous Hg in the outlet gas).

       For adsorption to take place (assuming attractive forces exist between a particular
gaseous specie and sorbent), the adsorbing specie must have sufficient time to reach the surface
of a sorbent and diffuse into its pores (where most adsorption takes place).  If any of the
adsorbing specie in a gas stream passing through a fixed bed of sorbent cannot reach the surface
of the sorbent (mainly its  pore surfaces), the specie will pass through the bed unadsorbed.
Researchers conducted preliminary tests to show that the gaseous Hg in the test gas had
sufficient time (under the conditions tested) to contact the sorbent and to diffuse into its pores.
Proving this point was important since some of the adsorption tests showed immediate
breakthrough of Hg in the outlet gas. In these cases, immediate breakthrough was not due to
insufficient contact time but rather the carbon's inability to adsorb all of the gaseous mercury.

       Figure 5-11 shows an example of the sampling and measurements taken during testing of
the baseline test gas with HC1, NO2, and SOj (as noted in the graph, SO2 was added to the
baseline test gas 2.5 hours after the start of the test).  Except where noted, the Hg concentrations
in Figure 5-11 are those in the outlet test gas and represent concentrations of total gaseous Hg.
Mercury concentrations in the graph are quantified as a percentage of the inlet concentration of
gaseous Hg°. The percentage of Hg in the outlet test  gas is called percent breakthrough. Figure
5-11 indicates that the analyzer sampled and measured total gaseous Hg in the outlet gas at all
times during testing except at approximately 5.2 hours, at which time the analyzer sampled and
measured Hg in the inlet gas.  At approximately 5.15  hours the analyzer measured gaseous Hg°
instead of total gaseous Hg in the outlet test gas; the drop in the concentration curve at this time
from approximately 150 percent to zero percent indicates that Hg in the outlet test gas consisted
entirely of gaseous Hg2+.  Thus, while only gaseous Hg° was in the test gas entering the carbon
bed, the Hg° was oxidized to Hg2+ as it passed through the bed. (Why some of the outlet
concentrations of total gaseous Hg exceeded 100 percent of the inlet Hg concentration for this
run is explained further on in this section.)
                                          5-28
-------
Table 5-3. Composition of test gases to simulate coal combustion flue gas used
for UND/EERC bench-scale study (source:  Reference 34).
SO, ppmv
HCI ppmv
NO ppmv
NO, ppmv
Baseline test gas"
0
0
0
0
Baseline test gas plus 1 additional gas
1600
0
0
0
0
50
0
0
0
0
300
0
0
0
0
20
Baseline test gas plus 2 additional gases
1,600
1,600
1,600
0
0
0
50
0
0
50
50
0
0
300
0
0
300
300
0
0
20
20
0
20
Baseline test gas plus 3 additional gases
1,600
1,600
1,600
0
50
50
0
50
300
0
300
300
0
20
20
20
Baseline test gas plus 4 additional gases
1600
50
300
20
     (a) Prior to adding SO,, HCI, NO, and/or NO2, the baseline test gas contained 15 jig/nm3 of gaseous Hg°;
        6 percent O2; 12 percent CO2; 8 percent H2O; and the balance N2.
                                       5-29
-------
          200
               Run 770
      1
      CD
      O
       O>
       CD
      Q_
                 151.9mgLAC@225°F
                 BL-hHCI+NO,(+SO-)
                     SO2 Injection Started
Figure 5-11. Example of the sampling and measurements taken during testing of

the baseline test gas with HCI, NO2, and SO2. (source: Reference 34).
                                    5-30
-------
       Graphs of the adsorption tests with the 15 remaining gases in Table 5-3 can be found
elsewhere;   the cited graphs are similar to Figure 5-1 1 in that Hg concentrations (primarily
outlet concentrations of total gaseous Hg) are plotted versus the time of the adsorption test.

       The following summarizes the detailed test results:

       •  When the sorbent was exposed to the baseline gas only, the sorbent initially captured
          10 to 20 percent of the incoming gaseous Hg°; the rest of the Hg passed through the
          bed (i.e., was not adsorbed).

       •  When the sorbent was exposed to SOi in addition to the baseline gas, Hg capture
          improved slightly.
       •  Under exposure of the sorbent to HCI, NO, or NOa added one at a time to the baseline
          gas, the Hg capture of the sorbent improved to 90 to 100 percent.

       •  An apparently significant interaction between SO2 and NO2 gases and the sorbent
          caused a rapid breakthrough of Hg as well as conversion of the Hg to its volatile
          oxidized form. This effect occurred at both 107 and 163 °C (225 and 325 °F) and with
          or without the presence of HCI and NO.

       •  In the presence of all four acid gases (SO2, HCI, NO, and NO2), rapid breakthrough
          and oxidation of the Hg occurred at both 107 and 163 °C (225 and 325 °F). This
          suggests that the interactions between the sorbent and NO2and SO2 gases produced
          poor sorbent performance, which may be a major effect. This may be likely to occur
          over a variety of conditions typical of coal-fired electric utility boilers, and represents
          a hurdle that must be overcome to achieve effective Hg control by carbon adsorption.

       The UND/EERC is continuing to investigate the interactions of gaseous flue gas
constituents on the adsorption capacity of activated carbon for Hg.  In addition, other types of
sorbents are being developed and investigated under similar simulated flue gas conditions. Other
gaseous flue gas constituents are also being examined to assess their impact on the adsorption of
Hg.

5.4.3.3 Sorbent Evaluation Using Fluidized-bed Reactors

       Under DOE's Small Business Innovative Research (SBIR) Program, Environmental
Elements Corporation (EEC) has been developing a circulating fluidized bed (CFB)24  to promote
agglomeration of fine PM, allowing for its capture in  an ESP. In addition, a single injection of
iodide-impregnated activated carbon was added to the fluidized bed to adsorb gaseous Hg. High
residence time, as a result of particle recirculation, allows for effective utilization of the carbon
and high collection of the fine particles. Laboratory tests with heated air indicate that, with a high
density of fly ash at a 4-second residence time within  the bed, fine particle emissions are reduced
by an order of magnitude.
                                          5-31
-------
       Results from the laboratory-scale testing indicate that spiked gaseous Hg° was
significantly reduced when passed through the fluidized bed of fly ash (50 percent Hg removed)
with a further reduction to essentially zero, when activated carbon was injected into the bed
(25 fig/m3 to zero) at 110 °C (230 °F). The iodide-impregnated activated carbon was folly
utilized after greater than 2 hours within the bed. An adsorption capacity was calculated to be
770 u,g/g for the carbon and 480 jig/g for the bed of ash.  Other field tests were conducted at
Public Service Electric and Gas1 Mercer Station with similar results.24
5.5 Sorbent Development

       The implementation of an effective and efficient Hg control strategy using sorbent
injection requires the development of low-cost and efficient Hg sorbents. Of the known Hg
sorbents, activated carbon and calcium-based sorbents have been the most actively studied.
However, improved versions of these sorbents and new classes of Hg sorbents can be expected,
as this is still a very active field.

5.5.1  Powdered A ctivated Carbons

       Activated carbons have been  extensively studied for their Hg capture capability.
Activated carbon is the reference sorbent for Hg control in municipal waste combustors.  Many
factors may affect the adsorptive capability of the  activated carbon sorbent. These include the
temperature and composition of the flue gas, the concentration of Hg in the exhaust stream, and
the physical and chemical characteristics of the activated carbon (or functionalized/impregnated
carbon).  Some specific efforts at understanding these effects are given below.

5.5.1.1 Effects of Temperature, Mercury Concentration, and Acid Gases

       The effects of bed temperature, Hg concentration, presence of acid gases (HC1 and SO2),
and presence of water vapor on the capture of Hg° and HgCl2 by thermally activated carbons
(FGD and PC-100) and Ca-based sorbents [Ca (OH) 2 and a mixture of Ca(OH) 2 and fly ash]
were examined in a fixed-bed, bench-scale system.21 Sorption studies indicated an abundance of
HgCl2 adsorption sites in calcium-based sorbents.  Increasing the HgCl2 concentration increased
its uptake, and increasing the bed temperature decreased its uptake.  Gas-phase HgCl2
concentration had a very strong effect on its adsorption, while bed temperature had a small
influence on adsorption. The observed temperature and concentration trends suggest that the
process is adsorption-controlled and  that the rate of HgCl2 capture is determined by how fast
molecules in the vicinity of the active sites are being adsorbed.  Mixtures of Ca(OH)2 and fly  ash
with 7 times higher surface area than Ca(OH)2 and a totally different pore size distribution
exhibited identical HgCl2 capture to that of Ca(OH)2. The presence of acid gases (1000 ppm SO2
and 50 ppm HC1) drastically decreased the uptake of HgCl2 by Ca(OH)2. The inhibition effect of
SO2 was more drastic that HC1, and essentially controlled the HgCl2 uptake.  It was hypothesized
that the inhibition effect is due to competition between these acid gases and HgCh for the
available alkaline sites.
                                          5-32
-------
       Sorption studies further indicated that the available active sites for capturing Hg° in the
activated carbons are limited, suggesting that it is more difficult to control Hg° emissions than
HgCb emissions. Increasing the Hg° inlet concentration and decreasing the bed temperature
increased the saturation capacities of the activated carbons, the time needed to reach this
capacity, and the initial rate of Hg° uptake.  Unlike HgCl2 capture by Ca(OH)2,  bed temperature
had a very strong effect on the Hg° adsorption by the activated carbons, and gas-phase Hg°
concentration had a small influence on such adsorption. PC-100, with twice the surface area of
FGD, consistently exhibited higher saturation capacities (3-4 times higher) than FGD. The
presence of acid gases had a positive effect on the capture of Hg° by a lignite-coal-based
activated carbon (FGD) and had no influence on Hg° capture by a bituminous-coal-based
activated carbon (PC-100). This difference was related to a higher concentration of Ca (acid gas
sorbent) in  FGD.  It appears that adsorption of these acid gases by FGD creates active S and Cl
sites, which are instrumental in capturing Hg°, through formation of S-Hg and Cl-Hg bonds in
the solid phase (chemisorption). These results indicate that the optimum region for the control of
Hg° by injection of activated carbon is upstream of the acid gas removal system.

5.5.1.2 Role of Surface Functional Groups

       The content of oxygenated acidic and alkaline surface functional groups (SFGs) on the
surface of two activated carbons was manipulated to investigate their role in Hg° and HgCh
capture.35 Acidic SFGs on the surface of activated carbons were neutralized by a variety of
alkaline washes. The alkaline-treated activated carbon showed no enhancement in Hg° and
HgCh capture, thus indicating that acidic SFGs play no role in capturing Hg species. The
alkaline SFGs content was increased by a thermal treatment process. The thermally treated
activated carbons did not exhibit any improvement with regard to their Hg° and HgCh capture
capabilities as compared to the untreated ones.  The activated carbons were then treated with a
very dilute  HC1 solution to decrease their alkaline SFGs content. The HCl-treated activated
carbon showed a very significant improvement in its Hg° and HgCl2 capture capabilities. This
observation was contrary to the initial  hypothesis that alkaline sites are needed to capture acidic
HgCU from the flue gas. It was then hypothesized that HC1 treatment increases the number of
active surface chlorine sites, which subsequently enhance Hg° and HgCl2 capture.  An analytical
technique, Energy-Dispersive X-ray Spectroscopy (EDXS), was used to quantify surface Cl sites.
A strong correlation between the increased amount of surface Cl and Hg°/HgCl2 uptake
enhancement was observed. The role  of SFGs containing Cl atoms in providing Hg°/HgCl2
active sites was established. Future investigation using SEM/EDXS and Fourier Transform
Infrared (FTIR) will focus on understanding the nature of Cl bonds on the surface of carbon, so
that more effective Hg species sorbents can be manufactured.

5.5.1.3 In-flight Capture of Mercury by a Chlorine-impregnated Activated Carbon

       Activated carbon duct injection seems to be the most promising Hg control technology
for coal-fired electric utility boilers equipped with ESPs.  In this technology, the injected
activated carbon  removes Hg only while contacting the flue gas during very limited sorbent/gas
contact time (<3  seconds). Prior investigations have shown that very high, and rather costly,
carbon-to-Hg weight ratios (>50,000)  are needed to achieve adequate Hg removal. In order to
                                          5-33
-------
reduce the operating cost of the carbon injection process, either a more efficient sorbent that can
operate at a lower carbon-to-Hg weight ratio or a lower-cost activated carbon (or possibly both)
are required.  In this study33, a cost-effective Cl-impregnation process was successfully
implemented on an inexpensive virgin activated carbon.  The Cl-impregnated carbon was
produced in a 5 pound large batch, and its in-flight Hg° removal efficiency was evaluated in a
flow reactor (as previously discussed in Section 5.4.2.1) with gas/solid contact times of 3 to
4 seconds. The Hg° removal efficiency of more than 80 percent was obtained in a flue gas
containing the effluent of natural gas combustion doped with coal combustion levels of NOX and
SO2 at carbon-to-Hg weight ratios of about 3000. Hg° removal was rather insensitive to the
adsorption temperature in the range of 100-200 °C.  Cost analysis showed that this Cl-
impregnation process can produce a very active and cost-effective activated carbon that can be
used as a practical sorbent in a duct injection control technology in ESP-equipped coal-fired
electric utility boilers. Preliminary cost estimates indicated that approximately  53 percent
reduction of the total annual cost of Hg control could be possible when using Cl-impregnated
FGD in lieu of virgin activated carbon. Future investigations would be focused on evaluating the
Cl-impregnated activated carbon in a pilot-scale, 21-kW (90,000-Btu/hr) refractory-lined,
furnace fired with pulverized coal.33

5.5.2 Calcium-based Sorbents

       Work conducted by EPA and ARCADIS Geraghty & Miller, Inc. [funded by the Illinois
Clean Coal Institute (ICCI)] indicates that the injection of calcium-based sorbents into flue  gas
can result in significant removal of Hg.36'37 Researchers examined the high-temperature/short-
gas-phase residence time removal of Hg using injection of lime while burning an  Illinois #6 coal
in a pilot-scale combustor.  The lime was injected as a slurry at a calcium-to-sulfur (Ca:S) ratio
of 2.0 mol/mol at 968 °C (1775 °F). Under these conditions, 77 percent of the total Hg was
removed from the flue gas  (Table 5-4). Based on these results, they concluded, "injection of
lime in the high temperature regions of coal-fired processes upstream of air pollution control
systems can efficiently transfer Hg from the gas to the solid  phase." Summaries of work follow.

5.5.2,1 Capture of Low Concentrations of Mercury Using Calcium-based Sorbents

       The capture of Hg° and mercuric chloride (HgCl2), the Hg species identified in coal flue
gas, by three types of calcium-based sorbents differing in their internal structure, was examined
in a packed-bed, bench-scale study under simulated flue gas conditions for coal-fired electric
utility boilers.38  The results obtained were compared with Hg° and HgCl2 capture by an
activated carbon (FGD) under identical conditions.  Tests were conducted with and without SO2
to evaluate the effect of SO2 on Hg° and HgCl2 control by each of the sorbents.

       The Ca-based sorbents showed insignificant removal of Hg° in the absence of SO2.
However, in the presence of SO2, Hg° capture was enhanced for the three Ca-based sorbents.  It
was postulated that the reaction of hydrated lime with SC>2 would result in pore mouth closure as
evidenced by the sharp drop in the SO: removal rate after the initial 10 minutes of exposure.
Despite the loss of internal surface area, the relatively high uptake of Hg°, observed for these
sorbents in the presence of SO2, suggests that Hg° and SO2 do  not compete for the same active
                                          5-34
-------
Table 5-4. Mercury removal by lime sorbent injection as measured by EPA bench-
scale tests (source: Reference 36).
Test
Baseline
Lime sorbent injection
Total Hg Concentration,
tig/dscm
5.7
8.0
Total Gaseous Hg,
percent
100
23
Total Particle-bound Hg,
percent
0
77
                                   5-35
-------
sites, and that the sites for Hg° capture are influenced positively by the presence of SO2.
Moreover, the capture of Hg in the presence of 862 increased with sorbent surface area and
internal pore structure.

       Conversely, the three Ca-based sorbents showed decreased removal of HgCI2 in the
presence of SOo. In the absence of SO2, roughly 25 percent of the incoming HgQ2 was captured.
The alkaline sites in the Ca-based sorbents were postulated to be instrumental in the  capture of
acidic HgCl2. SO2 not only competed for these alkaline sites but also, as mentioned, likely
closed pores with subsequent reduction in accessibility of the interior of the Ca-based sorbent
particles to the HgCl2 molecules.

       It was hypothesized that the capture of Hg° in the presence of SO: may occur through a
chemisorption mechanism, while the nature of the adsorption of HgCl2 molecules maybe
explained through a physisorption mechanism. The effect of temperature studies further
supported this hypothesis.  Increasing the system temperature caused an increase in Hg° uptake
by the sorbents in the presence of SC<2. However, the increase in temperature resulted in a
significant decrease in the HgCl2 uptake in the absence or presence of SO2. Increased sorbent
surface area and internal pore structure had no observable effect on HgCl2 capture in the
presence of SO2.

       With the relatively large quantities of Ca needed for SO2 control at coal-fired electric
utility boilers, the above results suggest that Ca-based sorbents, modified by reaction with fly
ash, can be used to control total Hg emissions and SO2 cost effectively.  The most effective
calcium-based sorbents are  those with significant surface area (for SO2 and HgCl2 capture) and
pore volume (for Hg° capture).

5.5.2.2 Simultaneous Control of Hg°, SO2, and NOX by Oxidized-calcium-based Sorbents

       Multipollutant sorbents have been developed that can remove both Hg° and Hg+2  as
effectively as FGD activated carbon in fixed-bed simulations of coal-fired electric utility boiler
flue gas at 80 °C.39 Oxidant-enriched, calcium-based sorbents proved far superior to activated
carbon with respect to SO2 uptake on the same fixed-bed simulations. These oxidant-enriched,
calcium-based sorbents also performed better with respect to NOX and SO2 uptake than baseline
lime hydrates for fixed- and fluid-bed simulations at 80 °C.

       Preliminary economic analyses suggest that silicate sorbents with oxidants are 20 percent
of the cost of activated carbon for Hg removal, while oxidant-enriched lime hydrates offer
reduced, but significant savings. Credits for SO2 and NOx increase the savings for multipollutant
sorbents over activated carbon.

       The apparent superiority of multipollutant lime and silicate hydrates enhanced with
oxidants has been confirmed at conditions typical  of gas-cooled, semi-dry adsorption processes
on boilers; performance of sorbents at higher-temperature conditions of duct sorbent injection
technologies remains to be evaluated. Planned field evaluations of both semi-dry adsorption and
                                          5-36
-------
duct sorbent injection will allow better economic and performance comparisons of activated
carbon sorbents to that of oxidant-enriched lime and silicate hydrates.

       A technology for the efficient capture of Hg through in furnace injection of a calcium-
based sorbent has been developed by McDermott Technologies recently. A discussion of the
full-scale tests of the technology is presented in Chapter 7.

5.5,3 Development of Low-cost Sorbents

       Since 1995, EPRI has supported a sorbent development program for removal of Hg
emissions from coal-fired electric utility power plants at several research organizations including
Illinois State Geological Survey (ISGS), University of Illinois (UI), and URS Corporation. The
development of effective Hg sorbents mat can be produced at lower costs than existing
commercial activated carbons is the main objective of the program. The development efforts
were documented in three EPRI Reports.4   A significant number of sorbents were derived
from a variety of precursor materials, including coal, biomass,  waste tire, activated carbon fibers,
fly ash, and zeolite, through this work. Different preparation methods were used to determine
the effects of sorbent properties, such as pore size distribution, pore volume, surface area,
particle size, and sulfur content, on the ability to remove Hg. The effects of different processing
methods, including steam activation, grinding,  size-fractionation, and sulfur-impregnation, on
sorbent performance were also investigated in laboratory tests. Promising low-cost sorbents
were further evaluated in actual flue gas at several full-scale coal-fired electric utility power
plants.

       Results of the EPRI sorbent  development work showed that effective sorbents can be
prepared from  inexpensive precursor materials using simple activation steps.  One notable
example is that a char produced from corn fiber, a by-product from a corn-to-ethanol production
process, showed  a Hg° adsorption capacity over twice that of the commercial FGD carbon
sorbent, after the char was activated in CO2 at 865 °C for 3.5 hours.40 Inactivated corn char had
no capacity for HgCla, and only a low capacity for Hg°.  It appears that the composition of the
flue gas has a significant effect on the Hg adsorption capacities of the coal-derived activated
carbons.41  The EPRI-funded study  found that the presence of acid gases (SCh and HC1) inhibits
Hg° and HgCli adsorption for both lignite- and bituminous-coal-derived activated carbons.
However, research conducted by EPA showed that the presence of acid gases enhanced the
capture of Hg° by a lignite activated carbon and had no influence on the adsorption by a
bituminous-coal-derived activated carbon.21 In a later more extensive follow-up study funded by
EPRI and ICCI, the  effects of acid gases on the HgCb and Hg° adsorption capacities of activated
carbons were found to vary, depending on the precursor materials and characteristics of the
carbons.43  For example, carbons derived from tire and corn fiber had much higher HgCk and
Hg° adsorption capacities when they were tested in a high-SO: concentration flue gas simulating
the combustion of Eastern bituminous coals compared to those when they were tested in the low-
SC>2  concentration flue gas simulating Western subbituminous  coal combustion.  Complex
interactions occurring between the characteristics of the carbons and the acid gases may lead to
the observed varying effects of the acid gases on Hg adsorption behaviors of the carbon sorbents.
                                          5-37
-------
t
More fundamental research is needed to understand and predict the effects of acid gases on the
performance of sorbents derived from different precursor materials.

       The most effective sorbents were obtained by the sulfur-impregnation of activated
carbons derived from waste material and carbon fibers.40 Researchers at the University of
Pittsburgh demonstrated that impregnation of heteroatoms such as sulfur44 and chloride45 is an
effective method to improve the vapor-phase Hg adsorption capacities of activated carbons.  It
has been suggested that sorbent-impregnation studies should focus on highly microporous
sorbents since the presence of active surface functional groups, sulfur as an example, in the
micropores through impregnation is likely to provide the most reactive sites for Hg adsorption
from coal combustion flue gas.19  They stressed that the micropore surface area of sorbent is an
important physical property for vapor-phase Hg adsorption. Most of the commercial activated
carbons are used for liquid-phase applications and contain a large mesopore surface area, in
addition to micropores, that are less effective for adsorption of ppb levels of Hg from coal
combustion flue gases. EPA researchers46 have observed the importance of active functional
groups in the micropores for vapor-phase Hg adsorption. After treating an activated carbon with
an aqueous sulfuric solution, they found that most of the mesopores of the carbon are filled with
water due to the presence of the hydroscopic sulfuric acid, and the carbon becomes a highly
microporous sorbent.  The Hg° adsorption capacity of the sulfuric-acid-treated carbon is much
higher than that of the as-received carbon due to the presence of the active sulfuric acid
functional groups in the micropores of the treated carbon.

       The most recent research conducted by ISGS, UI, and URS Corporation showed that
relatively low surface area microporous biomass-based carbon sorbents, such as those derived
from pistachio nut shells and from corn fiber, are as effective as the commercial FGD carbon
sorbent for Hg adsorption.43  They found that the Hg adsorption capacities of the biomass-based
carbon sorbents, which contained negligible (0.09 percent) sulfur, are comparable to those of the
coal- and tire-derived carbons that have substantial sulfur contents (0.98 to 2.1 percent). The
biomass-based carbon sorbents also have very little chlorine functional groups. It appears that
more  oxygen, another heteroatom, remained in the biomass-based carbon sorbents after the
pyrolysis of the oxygen-rich biomass from the carbon-making process contributing to the
significant Hg adsorption capacities of such sorbents.  It has been suggested recently by EPA
researchers47 that the Hg° adsorption capacity of an activated carbon is correlated to  the
concentrations of the oxygen functional groups of the carbon.  They changed the oxygen
functional group concentrations of a carbon by heating the carbon sample to 900 °C in an inert
atmosphere to remove the functional groups. Also, more oxygen functional groups were
introduced to the carbon sample by oxidizing the carbon sample in an aqueous nitric acid
solution. They suggested that lactone and carbonyl groups introduced during the oxidization of
the carbon by nitric acid treatment might be the active sites for Hg° adsorption.

5.5.4  Modeling of Sorbent Performance

       The Hg adsorption data produced from bench-scale tests provide a relative indication of
performance for different sorbents; however, the actual Hg removal performance of the sorbents
in full-scale systems cannot be  predicted based on bench-scale results alone. To predict Hg
                                                      5-38
-------
removal in fiill-scale systems, mass transfer considerations have to be combined with laboratory
data.  Such an approach was applied by by EPRI recently to develop a model for predicting
sorbent performance in full-scale systems.48 The model is also capable of determining when
mass transfer limits Hg removal and when it is limited by sorbent capacity.  By incorporating the
appropriate mass transfer expressions, the model relates the adsorption characteristic data for a
given sorbent tested under a given set of flue gas conditions in the laboratory to the expected Hg
removal performance across a FF or an ESP.

       Results of the sorbent performance predicted by the model agree reasonably well with
data of the same sorbent measured by pilot-scale tests for both ESP and FF applications. The
pilot-scale facilities used for the tests consisted of an ESP with a 160-acfm wire-tube ESP, and a
FF with a 4000-acfm transportable  pulse-jet FF operating in the COHPAC configuration.
Results of the pilot-scale tests and modeling both showed that a carbon sorbent with 15 p.m
diameter and 1000 jlg/g Hg adsorption capacity achieved about 80 percent Hg removal in a FF
operated at about 140 °C (280 °F) with 3 Ib/Macf sorbent injection rate and cleaning cycle of 45
min. However, test and modeling results both showed that Hg removal decreases to less than 20
percent when the same sorbent was injected upstream of an ESP under conditions similar to the
above.

       Laboratory tests which have been conducted to evaluate the adsorption characteristics of
potential sorbents for Hg removal seem to suggest that reactivity of the sorbent might be more
important than its equilibrium adsorption capacity for sorbent injection.  Currently, an ESP is
more widely used than a FF as a PM control device for coal-fired electric utility boilers in the
United States.  Sorbent reactivity is the important parameter determining Hg removal when
injecting a powdered sorbent upstream of an ESP, where adsorption of Hg occurs mainly in-
flight with short residence times (about 2 seconds). When injecting sorbent upstream of a FF,
additional  Hg removal can occur due to the presence  of accumulated sorbent in the filter cake,
resulting in improved mass transfer and sorbent utilization.  Sorbent capacity becomes a more
important parameter than reactivity in such cases.
5.6 Capture of Mercury in Wet FGD Scrubbers

5.6,1  Wet Scrubbing

       Mercuric chloride is readily soluble in water. Thus, the oxidized fraction of Hg vapors in
flue gas is efficiently removed when a power plant is operated with a wet scrubber for removing
SOi- The elemental fraction, on the other hand, is insoluble and is not removed to any
significant degree.  A DOE-funded study49 conducted by CONSOL, Inc. showed that the nominal
Hg removal for wet FGD systems on units firing bituminous coals is approximately 55 percent,
with the removal of Hg2^ between 80 and 95 percent. Studies conducted by McDermott
Technologies, Inc. at its  10-MWe research facility suggested a possible conversion of the Hg2*
                                                  50
captured in the scrubbing media and reemisstons as Hg".   McDermott Technologies performed
follow up tests to investigate the use of additives to prevent the conversion of adsorbed Hg2+ to
gaseous Hg0.51 These tests are described in more detail in Chapter 7.
                                         5-39
-------
5.6.2 Oxidation

       The challenge to Hg removal in wet scrubbers for SO2 is to find some way to oxidize the
elemental Hg vapor before it reaches the scrubber or to modify the liquid-phase of the scrubber
to cause oxidation to occur there.

       URS Radian International has conducted various laboratory and field-test studies to
investigate adsorption and catalytic oxidation of gaseous Hg° in coal-fired electric utility flue
gas.  The results of the bench-scale testing are discussed below.  The additional pilot- and full-
scale testing conducted by URS Radian  International are discussed in Chapter 7.

       Different compositions of catalysts and fly ashes were tested in a bench-scale, fixed-bed
configuration to identify materials that adsorb and/or oxidize gaseous Hg0.52  Mixing sand with a
particular catalyst or fly ash created fixed beds of sorbents.  A simulated coal-fired electric utility
boiler flue gas containing gaseous Hg° was then passed through the bed. The flue gas was tested
at the inlet and outlet of each sorbent bed to determine Hg adsorption and/or oxidation across the
bed.  Table 5-5 lists the simulated flue gas conditions and the most active catalysts and fly ashes
identified during testing for oxidation of gaseous Hg°.

       Figure 5-12 is an example of the adsorption/oxidation of gaseous Hg° with time by one of
the iron catalysts in Table 5-5.  In this figure, the oxidation of gaseous Hg° increases as the
breakthrough of Hg from the catalyst bed increases (breakthrough is quantified as a percentage
of the incoming Hg).  At 100 percent breakthrough when the catalyst is no longer adsorbing any
of the incoming Hg (i.e., the catalyst has reached its equilibrium adsorption capacity for the
incoming Hg°), all of the Hg° passing through the bed is being oxidized to Hg  .

       Figure 5-13 shows adsorption/oxidation results for all of the catalysts in Table 5-5.
Adsorption and oxidation of gaseous Hg° was greater at  149 °C (300  °F) than at the higher
temperature of 371 °C (700 °F).  The adsorption and oxidation activity of the activated carbon
was considered the highest among the materials tested because a lower mass was utilized during
the tests compared to the other materials.

       Figure 5-14 shows the adsorption/oxidation results for the fly ashes from Table 5-5.  Like
the catalysts, the fly ashes showed higher adsorption and oxidation of gaseous Hg° at 149 °C
(300 °F) than at 371 °C (700 °F); for this reason, only the lower temperature  results are shown in
Figure 5-14. The subbituminous and bituminous coal fly ashes generally showed higher
oxidation rates than the lignite coal fly ashes. As  seen, the #2 bituminous coal fly ash had
varying adsorption and oxidation rates depending upon where the fly ash samples were collected.
Samples collected from the hoppers of the first field of the ESP indicated lower oxidation of
gaseous Hg° but a higher adsorption of Hg compared to the finer fly ash collected in the fifth and
final field of the ESP.  Although not shown, fly ash captured by a cyclone in the Hg speciation
sampling train indicated a higher adsorption but no oxidation of the gaseous  Hg°.  Fly ash from
the fifth field of the ESP indicated the highest rate of oxidation and the lowest size-fractionated
particles. This may be associated with the size differences of the fly ash and/or the surface
                                          5-40
-------
Table 5-5.  Simulated flue gas conditions with the most active catalysts and fly
ashes indicated for oxidation of gaseous Hg° to gaseous Hg^source:
Reference 52)
Parameter
Fixed-bed Temperature
Hg° Injection
Oxygen
Carbon Dioxide
Moisture
Sulfur Dioxide
HCI
Gas Flow Rate
Baseline
Conditions
300 and 700 °F
45 to 60 ng/Nm3
7 percent
12 percent
7 percent
1600ppmv
SO ppmv
1 L/min
Most Active
Catalysts
Fe #1 (1000 mg)
Pd#1(1000mg)
Fe #2 (200 mg)
Fe #3 (200 mg)
NOX Catalysts
(1000mg)
Fe#4(1000mg)
Pd #2 (1000 mg)
Carbon (20 mg)
Most Active
Fly Ashes
Subbituminous #1
Subbituminous #2
Bituminous #1
Bituminous #2-Field V
Bituminous #2-Field 5*
Bituminous #3
Lignite #1
Oil-Fired #1
      (a) Fly ash collected at the first and fifth field of the ESP at the EPRI ECTC.
                                     5-41
-------
  S »
  •
  o
  en
  i
                                                    2+
                                                  Hg
                    % Total Hg
                   200
400
600      800

 Time, hr
1000
                                                              1200
Figure 5-12. Adsorption and subsequent oxidation of gaseous Hg° in a simulated
flue gas at 149°C (300 T) (source: Reference 52).
                                    5-42
-------
  1
  ^CQ
  os
    E^3
    d
  sg
100.0
 90.0
 80.0
 70.0
 60.0
 50.0
 40.0
 30,0
 20,0
 10.0
  0.0
••'/
j
I
• W-
"I
m
33




B
:l
i


^
1

1
1
1
i






n
\
1
i

—'

M
^

%
/,
1
1
1
El





i
!
!
1




1
V,
•'/
•<
^
?
!
I
f


%Adso«toed
% Oxi<»z«l



J
1
i



'^'
^
1
*?•;
i







i
!
1


g
1
1





'•"**


!
1
3QO*f '7t»lF 30tl°F'7flO"F MO'F 700TF S»*F 703 f SOOlF 700 f MOT 70Q°* 300"* 7WF 300^
F«#1 P«d#1 F«*2 F*«3 NO,a8taty8t F*«4 P
-------
                                                    E3% Adsorbed
                                                    B% Oxidized
Figure 5-14. Adsorption and oxidation of gaseous Hg° by various coal fly ashes at
           149 °C (300 °F) and 371 °C (700 °F) (source: Reference 52).
                                    5-44
-------
chemistry of the finer fly ash being enriched in trace metals or other condensed or adsorbed
compounds from the flue gas during the combustion of the bituminous coal.

5.6.5 Gas and Liquid Oxidation Reagents

       Argonne National Laboratory has been investigating the use of oxidizing agents that
could potentially convert gaseous Hg° into more soluble species that would be absorbed in wet
FGD systems.53 Current research is focused on a process concept that involves introduction of
an oxidizing agent into the flue gas upstream of the scrubber. The oxidizing agent employed is
NOXSORB™, which is a commercial product containing chloric acid and sodium chlorate.
When a dilute solution of this agent was introduced into a gas stream containing gaseous Hg° and
other typical flue-gas species at 300 °F (149 °C), it was found that nearly 100 percent of the
gaseous Hg° was removed from the gaseous phase and recovered in process liquids.  A
significant added benefit was that approximately 80 percent of the NO was removed at the same
time. Thus, the potential exists for a process that combines removal of SO2, NO, Hg°, and,
perhaps, PM.

       Continuing laboratory research efforts  are acquiring the data needed to establish a mass
balance for the process. In addition, the effects of such process parameters as reagent
concentration, SOi concentration, NO concentration, and reaction time (residence time) are being
studied. For example, SO2 has been found to decrease slightly the amount of gaseous Hg°
oxidized while appearing to increase the removal of NO from the gaseous phase.  Preliminary
economic  projections, based on the results to date,  indicate that the chemical cost for NO
oxidation could be less than $5,000/ton NO removed; while for gaseous  Hg° oxidation, it would
be about $20,000/lb  Hg° removed.  These results will be refined as additional experimental
results are obtained.
5.7 Observations and Conclusions

       When coal is burned in an electric utility boiler, the resulting high combustion
temperatures in the vicinity of 1500 °C (2700 °F) vaporize the Hg in the coal to form gaseous Hg°.
Subsequent cooling of the combustion gases and interaction of the gaseous Hg° with other
combustion products result in a portion of the Hg being converted to other forms, viz., Hg2+ and
Hgp. The term speciation is used to denote the relative amounts of these three forms of Hg in the
flue gas of the boiler. It is important to understand how Hg speciates in the boiler flue gas
because, as discussed in Chapters 6 and 7, the overall effectiveness of different control strategies
for capturing Hg often depends on the concentrations of the different forms of Hg species present
in the boiler flue gas.

       The speciation of Hg results from oxidation of Hg° in the boiler flue gas, with the
predominant oxidized Hg species believed to be HgCh. The mechanisms for this oxidation
include gas-phase oxidation, fly-ash-mediated oxidation, and oxidation by post-combustion NOx
controls. Data reveal that gas-phase oxidation is kinetically limited and occurs due to reactions
                                         5-45
-------
-------
       Oxidized Hg is readily absorbed by alkaline solutes/slurries or adsorbed by alkaline PM
(or by sorbents). Flue gas desulfurization systems, which use alkaline materials to neutralize the
acidic SO2 gas, remove oxidized Hg effectively in the flue gas.  Current research is focusing on
optimization of the existing desulfurization systems as a retrofit technology for controlling
oxidized Hg emissions and on development of new multipollutant control technologies for
simultaneously controlling both SO: and oxidized Hg emissions.
5.8 References

1.  Senior, C.L., A.F. Sarofim, T. Zong, J.J. Helble, and R. Mamani-Paco. Gas phase
   transformation of mercury in coal-fired power plants.  Fuel Processing Technology, 63,
   (2-3): 197-214 (2000).

2.  Senior, C.L., L.E. Bool, J. Morency, F. Huggins, G.P. Huffman, N. Shah, J.O.L.Wendt, F.
   Shadman, T. Peterson, W. Seames, B. Wer, A.F. Sarofim, I. Olmeze, T. Zeng, S. Growley,
   A. Kolker, C.A. Palmer, R. Finkelman, J.J. Helble, and M.J. Wornat.  Toxic substances from
   coal combustion - a comprehensive assessment. Physical Science, Inc., Final Report
   (Contract No. DE-AC-22-95, PC 951011, U.S. Department of Energy, Federal Energy
   Technology Center). September 1997.

3.  Senior, C.L., J.J. Helble, and A.F. Sarofim. "Predicting the speciation of mercury emissions
   from coal-fired power plants." Paper presented at the Conference on Air Quality II: Mercury,
   Trace Elements, and Particulate Matter, McLean, VA. September 19-21, 2000.

4.  Ghorishi, S.B., C.W. Lee, and J.D. Kilgroe. "Mercury speciation in combustion systems:
   studies with simulated flue gases and model fly ashes." Paper presented at the 92nd Annual
   Meeting of Air & Waste Management Association, St. Louis, MO. June 20-24, 1999.

5.  Edwards, J.R., R.K. Srivastava, and J.D. Kilgroe. A study of gas-phase mercury speciation
   using detailed chemical kinetics.  Published in Journal of Air & Waste Management
   Association, 5:  869-877 (2001).

6.  Niksa, S., JJ, Helble, and N. Fujiwara. "Interpreting laboratory test data on homogeneous
   mercury oxidation in coal-derived exhausts."  Paper presented 94th Annual Meeting of the
   Air & Waste Management Association, Paper # 86, Orlando, FL. June 24 -28,2001.

7.  Lee, C. W., J.D. Kilgroe, and S.B. Ghorishi.  "Speciation of mercury in the presence of coal
   and waste combustion fly ashes." Presented at the 93rd Annual Meeting of the Air & Waste
   Management Association, Salt Lake City, UT. June 18-22, 2000.

8.  Lee, C.W., R.K. Srivastava, J.D. Kilgroe, and S.B. Ghorishi. "Effects of iron content in coal
   combustion fly ashes on speciation of mercury." Paper presented at the 94th Annual Meeting
   of the Air & Waste Management Association, Paper # 156, Orlando, FL. June 24 -28, 2001.
                                         5-47
-------
9.  Galbreath, K.C., C.J. Zygarlicke, D.L. Toman, and R.C. Schulz. "Effects of NOX and cc-
   Fe2O3 on mercury transformations in a 7-kW coal combustion system." Paper presented at
   the 94th Annual Meeting of the Air & Waste Management Association, Paper # 767,
   Orlando, FL. June 24 -28,2001.

10. Norton, G.A., H. Yang, R.C. Brown, D.L. Laudal, G.E. Dunham, J. Erjave, and J.M. Okoh.
   "Role of fly ash on mercury chemistry in simulated flue gas streams."  Paper presented at the
   94th Annual Meeting of the Air & Waste Management Association, Paper # 164, Orlando,
   FL. June 24-28, 2001.

11. Haythornthwaite, S., S. Sjostrom, T. Ebner, J. Ruhl, R. Slye, J. Smith, T. Hunt, R. Chang,
   and T.D. Brown. "Demonstration of dry carbon-based sorbent injection for mercury control
   in utility ESP's and baghouses." In Proceedings of the EPRI/DOE/EPA Combined Utility
   Air Pollutant Control Symposium, EPRITR-108683-V3; Washington, DC. August 25-29,
   1997.

12. Laudal, D.L., M.K.  Heidt, T.D. Brown, and B.R. Nott. "Mercury speciation: a comparison
   between method 29 and other sampling methods." Presented at the 89th Annual Meeting of
   the Air & Waste Management Association, Nashville, TN, Paper 96-W64A.04. June 1996.

13. Brown, T. D., D.N. Smith, R.A. Hargis, Jr., and W.J. O'Dowd. "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.
   Available at: .

14. Li, Z., and J.Y. Hwang.  "Mercury distribution in fly ash compounds."  Presented at the Air
   & Waste Management Association Annual Meeting, Toronto, Ontario, Canada. June 8-13,
   1997.

15. Huggins, F.E., N. Yap, G.P. Huffman, and J.K. Neathery.   "Investigation of mercury
   adsorption on Cherokee fly-ash using XAFS spectroscopy." Presented at the 93rd Annual
   Meeting of the Air & Waste Management Association, Salt Lake City, UT. June 18-22,2000.

16. Carey, T.R., O.W. Hargrove, Jr., C.F. Richardson, R. Chang,  F.B. Meserole. Performance of
   Activated Carbon for Mercury Control in Utility Flue Gas Using Sorbent Injection,  In
   Proceedings of the EPRI/DOE/EPA Combined Utility Air Pollutant Control Symposium,
   Washington, DC; EPRI TR-108683-V3. August 25B29, 1997.

17. Galbreath, K.C., and C.J. Zygarlicke.  Mercury transformation in coal combustion flue gas.
   Fuel Processing Technology, 65-66:  289-310 (2000).
                                        5-48
-------
18. Brunauer, S., P.H. Emmett, and E. Teller, J. Am. Chem. Soc., 60,309. 1938.

19. HsiC.,  J. Rood, M. Rostam-Abadi, S. Chen, and R. Chang. "Effects of sulfur impregnation
   temperature on the properties and mercury adsorption capacities of activated carbon fibers
   (ACFs)."  Environmental Science and Technology, 35, 2785-2791. 2001.

20. Bansal R.C., J.B. Bonnet, and F. Stoecki. Active Carbon. New York, NY, and Basel,
   Switzerland: Marcel Dekker. 1988.

21. Ghorishi, S.B., and B.K. Gullett. Sorption of mercury species by activated carbons and
   calcium-based sorbents: effect of temperature, mercury concentration and acid gases.  Waste
   Management & Research, 16: 6: 582-593. 1998.

22. Krishman, S.V., B.K. Gullett, and W. Jozewicz. Sorption of Elemental Mercury by
   Activated Carbons. Environmental Science and Technology, 28(8): 1506-1512(1994).

23. Serre, S.D., B.K. Gullett, and S.B. Ghorishi. Entrained-flow adsorption of mercury using
   activated carbon. Journal of the Air & Waste Management Association, 51: 733-741 (May
   2001).

24. Helfritch, D.G., P.L. Feldman, and  S.J. Pass. "A circulating fluid bed fine paniculate and
   mercury control concept." Presented at the EPRI/DOE/EPA Combined Utility Air Pollutant
   Control Symposium, Washington DC. August 1997.

25. Hargis, R.A., W.J. O'Dowd, and H.W. Pennline. "Sorbent injection for mercury removal in
   a pilot-scale coal combustion unit." Presented at the 93rd Annual Meeting of the Air &
   Waste Management Association, Salt Lake City, UT. June 18-22, 2000.

26. Waugh, E.G., B.K. Jenson, L.N. Lapatrick, F.X. Gibbons, S. Sjostrom, J. Ruhl, R. Slye, and
   R.A. Chang. "Mercury control in utility ESP's and FFs through dry carbon based sorbent
   injection pilot-scale demonstration."  In Proceedings of the EPRI/DOE/EPA Combined
   Utility Air Pollutant Control Symposium, Washington, DC, EPRITR-108683-V3). August
   23-29, 1997.

27. Carey, T.R., C.  Richardson,  R. Chang, and F.B. Meserole. "Assessing sorbent injection
   mercury control effectiveness."  Paper presented at the 1999 Spring National Meeting of the
   American Institute of Chemical  Engineers, Houston, TX. March 14-18, 1999.

28. Sjostrom, S., T. Ebner, T. Ley, R. Slye, C. Richardson, T. Machalek, M. Richardson, R.
   Chang, and F. Meserole. "Assessing the  performance of mercury sorbents in coal
   combustion flue gas." Paper presented at the 94th Annual Meeting of the Air & Waste
   Management Association, Orlando, FL. June 24-28,  2001.
                                         5-49
-------
29. Rostam-Adadi, M., S.G. Chen, H-C. His, M. Rood, R. Chang, T.Carey, B. Hargrove, C.
   Richardson, W. Rosenhoover, and F. Meserole. "Novel vapor phase mercury sorbents."  In
   Proceedings of the First EPRI/DOE/EPA Combined Utility Air Pollution Control Symposium
   (The Mega Symposium), Washington, DC, August 25-29,1997.

30. White, D.M., W.E. Kelly, M.J. Stucky, J.L.  Swift, and M.A. Palazzolo. Emission test report:
   field test of carbon injection for mercury control, Camden County Municipal Waste
   Combustor, EPA/600/R-93/181 (NTIS PB94-101540), U.S. EPA, Air and Energy
   Engineering Research Laboratory, Research Triangle Park, NC. September 1993.

31. Serre, S.D., B.K. Gullett, and S.B. Ghorishi. "Elemental mercury capture by activated carbon
   in a flow reactor." Paper presented at 93rd Annual Meeting of the Air & Waste Management
   Association, Salt Lake City, UT. June 18-22, 2000.

32. Serre, S.D., B.K. Gullett, and Y. H. Li. "The effect of water (vapor-phase and carbon) on
   elemental mercury removal in a flow reactor." Paper presented at 94th Annual Meeting of
   the Air & Waste Management Association, Paper # 164, Orlando, FL.  June 24 -28, 2001.

33. Ghorishi, S.B., R. Keeney, W. Jozewicz, S.  Serre, and B. Gullett. "In-flight capture of
   elemental mercury by a chlorine-impregnated activated carbon." Paper # 731 presented at
   the 94th Annual Meeting of Air & Waste Management Association, Orlando, FL. June 24-
   28,2001.

34. Brown, T. D., D.N. Smith, R.A. Hargis, Jr., and W.J. O'Dowd.  "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.

35. Ghorishi, S.B., R.M. Keeney, and B.K. Gullett. "Role of surface functional groups in the
   capture of elemental mercury and mercuric chloride by activated carbons." In Proceedings
   of the Air Quality II Conference, McLean, VA. September 19-21, 2000.

36. Gullett, B.K., S.B. Ghorishi, K. Raghunathan, and K. Ho. Removal of Coal-Based Volatile
   Trace Elements: Mercury and Selenium, Final Technical Report. September 1, 1995, through
   August 31, 1996.

37. Ghorishi, S.B., and C.B. Sedman.  "Combined Mercury and Sulfur Oxides Control Using
   Calcium-Based Sorbents." Paper presented at the EPRI/DOE/EPA Combined Utility Air
   Pollutant Control Symposium, Washington, DC. August 25-29, 1997.

38. Ghorishi, S.B., and C.B. Sedman.  Low concentration mercury sorption mechanisms and
   control by calcium-based sorbents: application in coal-fired processes, Journal of the Air &
   Waste Management Association, 48: 1191-1198, 1998.
                                        5-50
-------
39. Ghorishi, S.B., C. Singer, W. Jozewicz, C. Sectarian, and R. Srivastava. "Simultaneous control
   of Hg°, SO2, and NOX by novel oxidized calcium-based sorbents." Paper # 243, Presented at
   the 94th Annual meeting of the Air & Waste Management Association. June 24-28, Orlando,
   FL,2001.

40. EPRI Report 1000454. Development and evaluation of low cost mercury sorbents.
   November 2000.

41. EPRI Report TE-114043.  Development and evaluation of mercury sorbents. November
   1999.

42. EPRI Report TR-110532.  Development and evaluation of low-cost sorbents for removal of
   mercury emissions from coal combustion flue gas. September 1998.

43. Rostam-Abadi, M., S. Chen, A.A. Lizzio, H-C. His, C.M.B. Lehmann, M. Rood, R. Chang,
   C. Richardson, T. Machalek, and M. Richardson.  "Development of low-cost sorbents for
   mercury removal from utility flue gas."  Paper presented at U.S. EPA/DOE/EPRI Combined
   Power Plant Air Pollutant Control Symposium, and the Air & Waste Management
   Association Specialty Conference on Mercury Emissions: Fate, Effects, and Control,
   Chicago, IL. August 20-23, 2001.

44. Korpiel, J.A., and R.D. Vidic. Effect of sulfur impregnation method on activated carbon
   uptake of gas-phase mercury.  Environmental Science and Technology, 31: 2319-2326
   (1997).

45. Vidlic, R. D. and D.P. Siler. Vapor-phase elemental mercury adsorption by activated carbon
   impregnated with chloride and chelating agents.  Carbon.  3-14(2001).

46. Li, Y. H., S.D. Serre, C.W Lee, and B.K. Gullett.  "Elemental mercury adsorption by
   activated carbon treated with sulfuric acid." Presented at the U.S. EPA/DOE/EPRI
   Combined Power Plant Air Pollutant Control Symposium, and the Air & Waste Management
   Association Specialty Conference on Mercury Emissions: Fate, Effects, and Control,
   Chicago, IL. August 20 -23, 2001.

47. Li, Y. H., C.W. Lee, and B.K. Gullett. "Characterization of activated carbons' physical and
   chemical properties in relation to their mercury adsorption." Presented at the American
   Carbon Society CARBON '01, An International Conference on Carbon, University of
   Kentucky Center for Applied Energy Research, Lexington, KY. July 14-19,2001.

48. Meserole, F., C. F. Richardson, T. Machalek, M. Richardson, and R.  Chang. "Predicted
   Costs of Mercury Control at Electric Utilities Using Sorbent Injection." Presented at U. S.
   EPA/DOE/EPRI Combined Power Plant Air Pollutant Control Symposium, and the Air &
   Waste Management Association Specialty Conference on Mercury Emissions: Fate, Effects,
   and Control, Chicago, IL, August 20-23, 2001.
                                        5-51
-------
49. DeVito, M.S., and W.A. Rosenhoover. "Hg flue gas measurements from coal-fired utilities
   equipped with wet scrubbers." Presented at 92nd Annual Meeting of the Air & Waste
   Management Association, St. Louis, MO. June 20-24, 1999.

50. Redinger, K.E., A. Evans, R. Bailey, and P. Nolan.  "Mercury emissions control in FGD
   systems." Presented at the EPRI/DOE/EPA Combined Air Pollutant Control System,
   Washington, DC. August 25-29, 1997.

51. McDermott Phase HI Study Section, McDermott Technologies, Inc. Advanced Emissions
   Control Development Program Phase III - Approved Final Report, prepared for the U.S.
   Department of Energy (US DOE-FETC contract DE-FC22-94PC94251-22) and Ohio Coal
   Development Office (grant agreement CDO/D-922-13).  July 1999. Available at:
   .

52. Hargrove, O.W., Jr., T.R. Carey, C.F. Richardson, R.C. Sherupa, F.B. Meserole, R.G. Rhudy,
   and T.D. Brown. "Factors affecting control of mercury by wet FGD." Paper presented at the
   EPRI/DOE/EPA Combined Utility Air Pollutant Control Symposium. Washington DC.
   August 25-29, 1997.

53. Livengood. C.D., and M.H.  Mendelsohn. "Process  for combined control of mercury and
   nitric oxide." Presented at the EPRI/DOE/EPA Combined Utility Air Pollutant Control
   Symposium, Atlanta, GA. August 16-20, 1999.
                                        5-52
-------
                                    Chapter 6

    MERCURY CAPTURE BY EXISTING CONTROL SYSTEMS USED BY
                 COAL-FIRED ELECTRIC UTILITY BOILERS


6.1     INTRODUCTION

       Existing coal-fired electric utility boilers in the United States use a variety of emission
control technologies to meet air standards for sulfur dioxide (SOi), nitrogen oxides (NOx),
and paniculate matter (PM). The EPA's ICR data presented in Chapter 3 of this report
indicate that most electric utilities are controlling NOx emissions from their coal-fired boilers
by combustion modification techniques and controlling SOi emissions by burning low-sulfur
coal. All of the coal-fired electric utility boilers use some type of post-combustion control
device to meet PM emission standards. Of these PM controls, electrostatic precipitators
(ESPs) are the predominant control type used on coal-fired boiler units (83 percent) with the
second most common control device being a fabric filter (14 percent).  Use of post-
combustion SOa controls is less common: approximately 20 percent of the boiler units use
either wet flue gas desulfurization (FGD) systems (15 percent) or spray dryer absorber (SDA)
systems (5 percent).  While the use of either selective  non-catalytic reduction (SNCR) or
selective catalytic reduction (SCR)  on coal-fired electric utility boilers for NOX emission
control presently is very limited (less than 4 percent),  the application of these post-
combustion NOx controls is becoming more prevalent.

       The implementation of post-combustion controls is not specifically intended to
control mercury emissions from coal-fired utility boilers. However, these controls capture
mercury in varying degrees depending on the control technologies used and the mercury
speciation at the inlet to the control device(s). This chapter discusses mercury capture by
existing post-combustion control systems used by coal-fired utility boilers. An estimate of
nationwide mercury emissions from existing coal-fired utility boilers is presented. The
mechanisms by which existing post-combustion control systems capture mercury are
reviewed. The ICR mercury emission test data for mercury capture by the existing post-
combustion control systems used for coal-fired utility boilers are presented and discussed.
6.2 EPA ICR PART III DATA

       As introduced in Chapter 1 of this report, the EPA conducted a three-part data
collection effort to gather information about the coal-fired utility boilers operating in the
United States in 19991. The Part I ICR data consist of information on the coal types burned,
the boiler furnace types, and the air pollutant control devices used for the 1,143 coal-fired
utility boilers in the United States having a capacity equal to or greater than 25 MWe. These
data are summarized and discussed in Chapters 2 and 3 of this report. The Part n ICR data
                                        6-1
-------
consist of information on the quantity, mercury content, and other selected properties of coal
burned by each of the identified 1,143 boiler units during calendar year 1999. A summary and
evaluation of these data are presented in Section 2.7 and Appendix A of this report.  For Part
III of the information collection effort, the EPA selected a subset of the coal-fired electric
utility boilers for which field source testing was performed to obtain  mercury emission data
for the air pollutant control devices now being used for these units. This chapter presents a
summary and analysis of the emissions data collected by Part HI of EPA's information
collection effort.

       The EPA ICR Part in data are composed of mercury emission source test results for 80
coal-fired electric utility boilers.  These boiler units were selected by the EPA to be generally
representative of the nationwide population of coal-fired utility boilers according to the type of
boiler used, the type of coal burned, and the air emission controls used. For each of the tested
boiler units, the flue gas mercury measurements were generally made at the inlet and outlet of
control device(s). The mercury measurements were made using the OH Method for speciated
mercury (this test method is discussed in Section 4.1 of this report).  Also, samples of the coal
being burned in the boiler unit during the source test were collected and analyzed for mercury
content.

       For boiler units that use a control configuration consisting of a single PM control
device, the flue gas samples were collected at the inlet to the PM control device and in the
stack. For units using SDA systems, the flue gas measurements were made at the inlet to the
SDA and in the stack. For units using an ESP or FF followed by a wet FGD scrubber, the flue
gas measurements were taken at the inlet to the wet scrubber inlet (i.e., downstream of the PM
control device) and in the stack.  For units equipped with a PS and a  wet FGD scrubber,
measurements were made at the inlet to the PS device and in the stack.

       Of the three IGCC plants located in the United States, two of the plants (Polk Power
Station and Wabash River Repowering Project) were included as part of the Part III  ICR test
program. At both facilities, combustion gas measurements using the OH Method were made
at the exhaust stack of the gas turbines. During testing, coal  feed rates to the coal-
gasification units  were recorded. Coal samples were collected during testing and analyzed
for total mercury.

       A summary of 81 boiler and coal type configurations for which mercury emission data
were collected is given in Table 6-1.  Of these boiler units, 65 were pulverized-coal-fired
(PC-fired) boilers. Such boilers account for the vast majority of the  1,143 coal-fired electric
utility boilers operating in the United States in terms of both total units and nationwide
generating capacity as shown Table 2-4.
                                         6-2
-------
Errata Page 6-3, dated 3-21-02
                                  Table 6-1
              Distribution of ICR Mercury Emission Test Data
                     By Boiler-coal Type Configurations
Boiler Unit
Type
Pulverized-coal-
fired
Cyclone- fired
Fluidized-bed
Combustor
Stoker-fired
IGCC (b)
Total Number
of Units Tested
Number of Boiler Units Tested
Fuel Burned In Boiler Unit
Bituminous
Coal
26
3
1
2
2
34
Subbituminous
Coal
29
2
0
0
0
31
Lignite
9
2
2
0
0
13
Other(a)
1
0
2
0
0
3
Total
Number of
Units
Tested
65
7
5
2
2
81
      (a) Some units used coal wastes or a blend of fuels.
      (b) Integrated coal gasification combined cycle unit.
      A summary of the flue gas cleaning devices installed on the PC-fired test units is given
in Table 6-2 as a function of type of fuel burned in each unit in 1999. These data show that:
      *  A total of 28 test units were equipped with a CS-ESP (14), HS-ESP (8), or FF (6).
      •  The 11 dry FGD units were equipped with either a SDA/ESP (3) or SDA/FF (8).
      •  The 20 wet FGD units were equipped with a PS + Wet FGD (6), CS-ESP + Wet
         FGD (6), HS-ESP + Wet FGD (6), or FF + Wet FGD (2).
      •  Two units were equipped with a CS-ESP + FF.
      •  One was equipped with a PS.
                                     6-3
-------
Errata Page 6-4, dated 3-21-02

                                    Table 6-2
 Distribution of ICR Mercury Emission Test Data for Pulverized-coal-fired
     Boilers By Post-combustion Emission Control Device Configuration
Post-
combustion
Control
Strategy
PM Control Only
PM Control and
Dry SO2 Scrubber
System
PM Control and
Wet SO2 Scrubber
System
Post-combustion
Emission Control
Device
Configuration
CS-ESP
HS-ESP
FF
CS-ESP + FF
PS
SDA + CS-EP
SDA + FF
DI + CS-ESP
PS + wet FGD
CS-ESP + wet FGD
HS-ESP + wet FGD
FF + wet FGD
Other Control Device Configuration
Number of Units Tested
Number of Boiler Units Tested
Fuel Burned In Boiler Unit
Bituminous
Coal
7
4
4
0
0
0
3
1
1
1
1
2
2
27
Subbituminous
Coal
5
4
2
0
1
3
3
0
4
3
5
0
0
29
Lignite
1
0
0
2
0
0
2
0
1
2
0
0
0
8
Other
1
0
0
0
0
0
0
0
0
0
0
0
0
1
Total
14
8
6
2
1
3
8
1
6
6
6
2
2
65
   PM Controls
   CS-ESP = cold-side electrostatic precipitator
   HS-ESP = hot-side electrostatic precipitator
   FF = fabric filter
   PS = particle scrubber
SO? Controls
DI = dry injection
FGD = flue gas desulfurization system
SDA = spray dryer adsorber system
6.3 MERCURY CONTENT OF UTILITY COALS BURNED IN 1999

      The analysis results of more than 39,000 coal samples were reported in the Part II ICR
data. These results include the mercury content of as-fired coals and supplemental fuels
burned in electric utility boilers in 1999. A comparison of the mercury contents of the
different major coal types and supplemental fuels burned by electric utilities in  1999 and
normalized by fuel heating value is shown Figure 6-1.  Waste bituminous coal and waste
anthracite had the highest mercury contents expressed in Ib Hg/1012 Btu. The mercury content
of the bituminous coal, Subbituminous coal, and lignite (the three most commonly used fuels)
was generally less than 15 lb/1012 Btu. Statistical information on each type of fuel burned in
coal-fired utility boilers is presented in Table 6-3.
                                       6-4
-------
1
J



















"-3 U U
111'
U U U
" U V
1 x I 	




x 1 —













.
ho
ni
a
I
3
>• C
s s
3 5
D

















1
a AI
;
'



















c
^
/ n
;
i (
j 5







1



1


I




>
H
• 9u
1
' ti
L, y
3 S
H




D

X









1


.2
X



















0 Petcentile
1






1—



X






X
O








c
















X




1


\





>






1-



H
X

X



X


<
h




iin
-H x
•
-M>

X
fl


f
1
1
i-
[U

t
t



o
c
a '§
3t • - "^
1
s S
f 3 *>
m £
g «2
II - i
9
ft V
1 S J
E «"
1 6 2 §
>» •
I * S S
.2? o 01 y
•J — • 3 NN
O\
.1 2
Is 3
3- 41
S 1-
•g _ S
E SJ .M
•8 " ta
^3
3
«J
J= S

•= oo
S M
1
1
6-5
-------
                                                      Table 6-3
                            Comparison of Mercury Content Normalized By Heating Value
                      In As-fired Coals and Supplemental Fuels for Electric Utility Boilers in 1999

Fuel Type
Anthracite coal
Bituminous coal
South American bituminous coal (a)
Subbituminous coal
Indonesian subbiiuminous coal (b)
Lignite
Waste anthracite coal
Waste bituminous coal
Waste subbituminous coal
Petroleum coke
Tire-derived fuel
Number
of
Analyses
114
27,884
270
8,193
78
1,047
377
575
53
1,149
149
Ratio of Mercury to Fuel Heat Content
(Ib Hg per 10IZ Btu)
Range
5.02-35.19
0.04-103.81
0.70-66.81
0.39-71.08
0.79-4.61
0.93-75.06
2.49-73.02
2.47-172.92
5.81-30.35
0.06-32.16
0.38-19.89
Mean
15.28
8.59
5.94
5.74
2.51
10.54
29.31
60.50
11.42
23.18
3.58
Median
13.37
7.05
4.91
5.00
2.39
7.94
27.77
53.32
10.79
2.16
2.79
Standard
Deviation
6.23
6.69
5.28
3.59
0.86
9.05
11.94
44.35
4.66
3.18
2.78
                (a) Bituminous coal imported from South America and burned at one power plant in Florida and one power
                   plant in Texas.
                (b) Subbituminous coal imported from Indonesia and burned at a coal-fired power plant in Hawaii.
t
6.4  POTENTIAL MERCURY CAPTURE IN EXISTING UNITS

       Mercury capture in existing units depends on Hg speciation at the inlet to the control
device(s) and the type(s) of control technologies used.  Units that burn bituminous coals have
relatively high concentrations of Hg2+ at the inlet to the control device(s). Units that burn
subbituminous coal or lignite typically have relatively low concentrations of Hg2+ and high
concentrations of Hg°at the inlet to the control device(s).

       The effects of coal and combustion conditions are attributed primarily to the flue gas
composition and properties of fly ash that affect the speciation and capture of Hg. While OH
measurements made upstream of PM control devices do not always provide quantitatively
accurate information on Hg speciation, they do provide semi-quantitative information relative
to the amounts of Hgp, Hg2+, and Hg° in flue gas from the combustion of different types of

                                         6-6
-------
coals.  They also provide useful information on the potential for the oxidization of the Hg° and
the capture of the resulting reaction products in downstream control devices.

       The relatively high concentrations of chlorine in bituminous coals are believed to
result in the oxidization of Hg° to form Hg2+, primarily HgCl2. By contrast, both subbituminous
coal and lignite have lower amounts of chlorine and higher amounts of alkaline material
(calcium and sodium) than bituminous coals.  Chlorine from the combustion of subbituminous
coal and lignite tends to react with the alkaline materials in flue gas, and little if any chlorine
is available for the oxidization of Hg. Therefore, flue gas from combustion of subbituminous
coal and lignite tends to have relatively low concentrations of Hg 2+.

6.4.1  Units with an ESP or FF

       Approximately 77 percent of the coal-fired utility boilers currently operating in the
United States are equipped with only an ESP or an FF. Gaseous mercury (both Hg° and Hg2+)
can potentially be adsorbed on fly ash and be collected in a downstream ESP or FF. The
modern ESPs or FFs that are now used on most coal-fired units achieve very high capture
efficiencies for total particulate matter (see Table 3-3). As a consequence, these PM control
devices are also effective in capturing Hgp in the boiler flue gases.

       The degree to which mercury can be adsorbed onto fly ash for subsequent capture in
PM control is dependent on the speciation of mercury, the flue gas concentration of fly ash,
and the properties of fly ash. It is currently believed that mercury is primarily adsorbed onto
the unbumed carbon in fly ash (see Section 5.3).  Approximately 80 percent of the coal ash in
PC-fired boilers is entrained with the flue gas as fly ash.  PC-fired boilers with low-NOx
burners have higher levels of carbon in the fly ash with a correspondingly higher potential for
mercury adsorption. Cyclone and stoker boilers tend to have high levels of carbon in the fly
ash, but have lower flue gas concentrations of fly ash than PC-fired boilers. Fly ash
concentrations in fluidized-bed combustors tend to be higher than those in PC-fired boilers.
Also, the carbon content of fluidized-bed combustor fly ash is generally higher than that of
PC-boiler fly ash.

       The syngas from a coal gasifier is composed mainly of hydrogen, carbon monoxide,
carbon dioxide, and nitrogen.  This gas also contains vaporous trace elements, such as
mercury, as well as dust and aerosols containing trace elements.  The source of mercury in
syngas is the mercury that is naturally present in coal and is released during the gasification
processes, which typically takes place at 950 °C (1750 °F).  Mercury that is not retained in the
solid residue from the gasification process is released almost exclusively as Hg°.

       Gas-phase mercury in units equipped with an ESP can be adsorbed on the entrained
fly ash upstream of the ESP. The gas-phase mercury in units equipped with a FF can be
adsorbed by entrained fly ash or it  can be adsorbed as the flue gas passes through the filter
cake on the surface of the FF.  The degree to which gaseous mercury adsorbs on the filter
cake typically depends on the speciation of gaseous mercury in the flue gas; in general,
gaseous Hg2+ is easier to adsorb than gaseous Hg°(see discussion in Section 5.3.1).
                                         6-7
-------
t
6.4.2  Units with SPA Systems

       An SDA system operates by the same principle as a wet FGD system using a lime
scrubbing agent, except that the flue gas is mixed with a fine mist of lime slurry instead of a
bulk liquid (as in wet scrubbing).  The SOa is absorbed in the slurry and reacts with the
hydrated lime reagent to form solid calcium sulfite and calcium sulfate. The heat of the flue
gas, leaving dry solid particles of calcium sulfite and calcium sulfate, evaporates the water in
the mist. Entrained particles (unreacted sorbent particles, reaction products, and fly ash) are
captured in the downstream PM control device (either an ESP or FF).

       The performance of SDA systems in controlling SO2 emissions is dependent on the
difference between the SDA outlet temperature and the corresponding flue gas water vapor
saturation temperature. SDA systems on coal-fired boilers typically operate about 20 °F
(11 °C) above the saturation temperature (i.e., a 20 °F [11 °C] approach to saturation
temperature). The relatively low flue gas temperatures afforded by SDA systems increase the
potential for mercury capture.  The caking or buildup of moist fly ash deposits, which can
plug the SDA reactor and coat downstream surfaces, dictates the minimum flue gas
temperatures, which can be employed at the outlet of SDAs.

       Hgp is readily captured in SDA systems.  Both Hg° and Hg2+ can potentially be
adsorbed on fly ash, calcium sulfite, or calcium sulfate particles in the SDA. They can also
be adsorbed  and captured as the flue gas passes through the ESP or FF, whichever is used for
PM control.  In addition, gaseous Hg2+ may be absorbed in the slurry droplets and react with
the calcium-based sorbents within the droplets. Nearly all of the Hgp can be captured in the
downstream PM control device.  If the PM control device is a FF, there is the potential for
additional capture of gaseous mercury as the flue gas passes through the bag filter cake
composed of fly ash and dried slurry particles.

6.4.3  Units with Wet FGD Systems

       Approximately 15 percent of coal-fired utility boilers in the United States use wet
FGD systems to control 862 emissions. In each of these systems, a PM control device is
installed upstream of the wet FGD scrubber.  PM control devices used with wet FGD
scrubbers include particulate scrubbers (PS), CS-ESPs, HS-ESPs, and FF baghouses. As
described in Chapter 3, wet FGD systems remove gaseous SO2 from flue gas by absorption.
In wet scrubbers, gaseous species are mixed with a liquid in which they are soluble. For SOa
absorption, gaseous  SO2 is  mixed with a caustic slurry, typically water and limestone or water
and lime.

       Gaseous compounds of Hg2+ are generally water-soluble and can absorb in the
aqueous slurry of a wet FGD system. However, gaseous Hg° is insoluble in water and
therefore does not absorb in such slurries. When gaseous compounds of Hg2"1" are absorbed in
the liquid slurry of a wet FGD system, the dissolved species are believed to react with

                                         6-8
-------
dissolved sulfides from the flue gas, such as H^S, to form mercuric sulfide (HgS); the HgS
precipitates from the liquid solution as sludge. In the absence of sufficient sulfides in the
liquid solution, a competing reaction that reduces/converts dissolved Hg2+ to Hg° is believed
to take place. When this conversion takes place, the newly formed (insoluble) Hg° is
transferred to the flue gas passing through the wet FGD system. The transferred Hg°
increases the concentration of Hg° in the flue gas passing through the wet FGD (since the
incoming Hg° is not absorbed), thereby resulting in  a higher concentration of gaseous Hg" in
the flue gas exiting the wet FGD compared to that entering.  Transition metals in the slurry
(originating from the flue gas) are believed to play an active role in the conversion reaction
since they can act as catalysts and/or reactants for reducing oxidized species.

       Recent research on the capture of mercury in wet scrubber systems is discussed in
Section 5.6.

6.4.4 Units with Other Control Devices

       Some units use PS systems, primarily venturi scrubbers, to control PM emissions.
Capture of Hg in these systems is limited to soluble Hg compounds such as HgCh.  PS
systems are typically poor fine PM collectors and, if Hgp in the flue gas is associated with
fine PM, capture of Hgp by such scrubbers may be poor.  Hg° is insoluble and will not
typically be captured by the scrubber. It is possible to capture Hg2"1" in  the wet scrubbers, but
the scrubber chemistry, and the manner in which the scrubber is operated, will determine
whether it is effectively removed, or whether it is stripped, from the scrubbing liquor.
Stripping can occur if the Hg2+ is not adsorbed on the particles, or reacted chemically with
liquid-phase reactants within the scrubber.

       Mechanical collectors such as cyclones do a poor job of capturing fine PM, and
mercury capture in these control devices should be  limited to the capture of Hgp associated
with particles larger than  10 um.

6.5  EPA'S PART ffl ICR DATA EVALUATION  APPROACH

       The methods used to evaluate the Part III ICR data were based  on two interrelated
objectives. The first objective was to estimate the amount, speciation, and geographical
distribution of national mercury emissions from coal-fired power plants in 1999. The second
was to characterize the effects of coal properties, combustion conditions, and flue gas cleaning
methods on the speciation and capture of mercury.  The satisfaction of the first objective
involved the development of mercury emission factors as a function of the type of coal burned,
the type of boiler, and the air pollution control device(s) used.

6.5.1  Evaluation Method

       The development of emission factors for different classes of coal-fired units was based
on hypotheses derived from current understanding of mercury speciation and capture, as
discussed in Chapter 5. The hypotheses are:
                                         6-9
-------
t
       •  Mercury speciation and capture are dependent on the coal properties, combustion
          conditions, and flue gas cleaning methods that are used for any specific test unit,

       •  Hg2+ is more readily absorbed in aqueous media than Hg°, and therefore can be
          captured in wet scrubbers, while Hg° cannot,

       •  Gas-phase mercury can be adsorbed onto the unburned carbon in fly ash, which can
          catalyze oxidation of Hg°,

       •  Hgp can be readily captured in an ESP or an FF,

       •  The potential for mercury capture increases with decreasing flue gas temperatures, and

       •  Flue gas from combustion of bituminous coals typically has a higher fraction of
          Hg2+ than the gas from subbituminous and lignite coals.

       Combinations of coal, boiler, and control technologies that are expected to behave in
a similar manner with respect to speciation and capture of mercury can be grouped into data
sets called coal-boiler-control technology classes or bins. Many of these data sets in the ICR
database consist of tests at one or two units, and this small number of samples results in
relatively large uncertainties concerning the central values and variability of the underlying
populations. However, the mean values and statistical behavior of the classes with a large
number of test units can be investigated, and the results can be compared with the results of
classes with a small number of test sites. If the relative behavior of the large and small data
sets is consistent with our theoretical expectations, then we can have some confidence that
the speciation and capture estimates for the smaller sets are reasonable.

       The ICR Part III emission data were sorted into coal-boiler-control classes. Next, the
data in each class were evaluated for consistency, and the data between classes were evaluated
according to the postulated behavior criteria given above.  With few exceptions, the differences
in speciation and capture of mercury between the different classes were consistent with the
above-hypothesized behavior.  Based on this observation, emission factors were developed for
use in estimating the amount and speciation of mercury emissions from coal-fired electric
utility boilers in 1999.  The data in the coal-boiler-control classes were also used to conduct
further evaluations of the effects of coal properties, combustion conditions, and flue gas
cleaning conditions on the control of mercury emissions at existing coal-fired power plants.

6.5.2 Measures of Performance

       Measures used to evaluate the effect of the coal, boiler, and control device variables
on the capture of mercury included the inlet and outlet concentrations of Hgp, Hg2+, Hg°, and
HgT, and the reduction of Hgj. Emission factors, defined in this report to be the fraction of
mercury emitted to the atmosphere relative to the amount that enters the first air pollution
control device, were also calculated and used to evaluate the emission of speciated Hg and

                                         6-10
-------
HgT.

       The fraction of HgT captured in air pollution control device(s) can be used
interchangeably with the emission factor for HgT [EMF^:

       EMFT = 1 - Capture HgT

Where the fractional capture is:

       Capture HgT = [  HgT (inlet) - HgT(outlet)]/HgT(inlet) = 1 - Hg-r(outlet)/HgT(inlet)

And the percentage reduction (%Red) across the control device(s) is:

       %Red = 100 x [1 - HgT(outlet)/HgT(inlet) ]

       The %Red can be determined from either (1) the inlet  and outlet concentrations of
HgT as measured by the  OH Method, or (2) inlet concentration estimates made from Part in
coal samples and outlet concentrations obtained with the OH Method. When the OH
measurements are used to evaluate the reduction in emissions or emission factors, the inlet
and outlet concentrations must be expressed on a common basis jim/dscm at 3% Oa) or Ib of
Hg/1012 Btu of coal burned to account for air in-leakage through fans or across the air
pollution control device(s).

       The results of the OH Method emission tests for HgT are shown in Figures 6-2 and
6-3. Figure 6-2 is a scatter plot of the inlet versus the outlet concentrations of HgT. In
general, the outlet HgT concentration increases with increasing inlet HgT concentrations. The
increasing outlet HgT concentrations that appear linear with respect to HgT inlet
concentrations are indicative of a constant percentage reduction across the control device(s).
ESPs exhibit this type of performance for the control of PM.  These types of devices are
called constant reduction devices. Note that there are also a number of data points distributed
just above the x-axis; i.e., zero outlet concentration. These data points are indicative of
constant outlet devices with low emission concentrations.  FF baghouses tend to operate like
constant outlet devices.

       Figure 6-3 is a scatter plot showing inlet HgT concentration versus percent reduction
in HgT across the control device(s).  There are no discernable  trends in the capture of HgT as a
function of inlet concentration. The negative emission reductions represent cases for which
the outlet HgT concentration is higher than the inlet concentration.  This can result from one
or a combination of factors.  For  example, negative emission reductions can occur when (1)
temperature changes within the test unit increase the desorption of Hg, (2) ESP rapping
cycles result in the reentrainment of Hgp, and (3) small differences between Hg inlet and
outlet concentrations cannot be accurately quantified because of imprecision in the OH
Method.
t
                                         6-11
-------
                                  60     80     100    120    140     160

                           Inlet Hg(T) Concentration,
      Fig. 6-2.  Inlet versus outlet mercury concentration for all tests.
              150
           .2   50
           1
           u
           8
           a
0
          1
          o
              -100	
              -150
                                                       	— o
                          SO     100    120     140     liO
                            Inlet Hg(T) Concentration, ^g|'nl
Fig. 6-3.  Inlet mercury concentration versus percent reduction for all tests.
                                 6-12
-------
       Emission factors for speciated Hg can be developed by dividing or normalizing the
stack Hg species by the concentration of HgT at the inlet to the first control device. In the
development of these emission factors, it is assumed that all of the Hg in the as-burned coal is
equal to the value of HgT measured at the inlet sampling location by the OH method. The
emission factors for Hgp (EMFp), Hg2+ (EMF2+), and Hg° (EMF°) are calculated by:

                 EMFP = Hgp (outlet) / HgT (inlet),

                 EMF2+ = Hg2+ (outlet) / HgT (inlet), and

                 EMF° = Hg° (outlet) / Hgr (inlet).

For situations where HgT (outlet) is higher than HgT (inlet), the stack  emission factors are
calculated by replacing the HgT (inlet) value with the corresponding HgT (outlet) value:

                 EMFp = Hgp (outlet)/HgT (outlet),  [for HgT (outlet) > HgT (inlet)],

                 EMF2+ = Hg2+ (outlet)/HgT (outlet),  [for HgT (outlet) > HgT (inlet)], and

                 EMF° = Hg° (outlet)/HgT (outlet),   [for HgT (outlet) > HgT (inlet)].

In the latter case, it should be noted that EMFP + EMF2"1" + EMF° = 1.

       hi addition to the above emission factors, speciation factors (SPFs) are calculated and
used to characterize Hg speciation at both the inlet and outlet sampling locations. The SPFs
represent the fractions of HgT in the inlet or outlet samples that are present as Hgp, Hg2"1", or
Hg°. For the inlet sampling train:

                 SPFP = Hgp (inlet) / HgT (inlet),

                 SPF2+ = Hg2* (inlet) / HgT (inlet), and

                 SPF° = Hg° (inlet) / HgT (inlet).

For the outlet sampling train:

                 SPFp =  Hgp (outlet) / HgT (outlet),

                 SPF2+ = Hg2+ (outlet) / HgT  (outlet), and

                 SPF° = Hg° (outlet) / HgT (outlet).
In all cases:
                 SPFp + SPF2+ + SPF°=l.
                                         6-13
-------
4ft
       Emission factors and speciation factors for units equipped with an ESP, FF, PM
scrubber, mechanical collector, SDA/ESP, or SDA/FF were calculated using inlet OH
measurements for Hgj and outlet OH measurements for speciated and HgT. For units with
wet FGD scrubbing systems, emission  factors were determined by multiplying the average
emission factor for the PM control device that precedes the scrubber by the emission factors
for the scrubber as determined by OH measurements. For example, the estimated EMFs for a
PC-fired boiler burning subbituminous coal and equipped with cold-side ESP and wet FGD
system are calculated as follows:

       The class average CS-ESP EMFj for a PC-boiler firing subbituminous coal is 0.91,
       and the class average wet FGD  EMFi for a PC-boiler firing subbituminous coal is
       0.71. The EMFT across both control devices is therefore:

                    EMFT (CS-ESP  + FGD) = EMFT (CS-ESP) x  EMFT (FGD)

                    = 0.91x0.71=0.65.

       The corresponding level of control across both devices is:

                    %Reduction (CS-ESP + FGD) = 100 * [1- EMFT (CS-ESP + FGD)]

                    = 100 (1-0.65) = 35%.

       Emission factors for coal gasification units were calculated using the Hg content of
the feed coal and the OH measurements made in the stack.

6.5.3  Comparisons of Hgr (Inlet) Using OH Measurement and Coal Hg Data

       Emission factors for speciated and total Hg relative to inlet Hg concentrations can be
determined using two methods.  The first method uses the HgT inlet concentrations from OH
sampling train measurements. The second method involves the calculation of total Hg inlet
values using coal Hg data and sampling train data (flue gas flow rate, moisture concentration,
O2 concentration, and temperature).

       Emission factor estimates determined using the OH Method train data and the ICR Part
II coal data often give significantly different results. The best estimate can sometimes be
obtained by discarding outliers, by reviewing the test reports for tests conditions that can lead
to questionable results, and by comparison of the results relative to tests at other test sites.  In
some cases, it is not possible to arrive at a best estimate, and there is a significant amount of
uncertainty leading to a range of estimates.

       Mercury capture (percent reduction in emissions) and emission factors for Hgp, Hg2+,
Hg°, and Hgr were then calculated using the average stack values for each data set as
determined by both coal and OH Method sample train data. Emission factors based on the
OH Method sampling train data provided the most consistent results. The inlet

                                        6-14
-------
 concentrations and percentage reduction reflected in the body of this report correspond
 primarily to test results obtained using the OH Method.

 6.5.4  Development of Data Sets for Coal-boiler-control Classes

        As described earlier, unit classes are defined as those combinations of coal, boiler,
and control technologies that are expected to provide similar results in the speciation and
capture of Hg. Data sets for different classes of units were developed by sorting the unit tests
by coal type, boiler type, NOx control method, PM control method, and SC>2 control method.
Data sets were consolidated whenever the joint sets appeared to provide the same results as
the initial groupings.  Thus, wall- and tangentially fired PC boilers were consolidated into a
single conventional PC boiler set. Units that reported no NOx controls were consolidated with
low-NOx burners, overfire-air staging, and concentric firing systems.

 6.5.5  Questionable Nature of OH Speciation Measurements Upstream of PM Controls

        Initial evaluations of the Part HI ICR data dealt with comparisons of the coal-boiler-
 control classes using the results of OH speciation measurements at both the inlet and outlet
 sampling locations. Comparisons were also made of the results obtained using either the Part
 III ICR coal data or the inlet OH data to evaluate emission reduction trends. The comparison
 of speciation at the inlet and outlet locations produced, in some cases, results contrary to the
 expected behavior of Hg between the inlet and outlet of the control devices.

        Previous research has shown that the OH sampling method provides valid
 measurements for Hgr at both the inlet to flue  gas cleaning devices and in the stack.  Also, the
 OH Method has been shown to provide valid Hg speciation measurements when samples are
 taken  downstream of an efficient PM control device. However, the OH Method can give
 erroneous speciation measurements for locations upstream of PM control devices.

        The OH sampling train consists of a probe, a particulate filter, a series of impingers, a
 gas flow meter, and a sample pump. The filter captures particulate matter and Hgp, while the
 downstream impingers separate Hg2+ from Hg°. Fly ash captured by the sampling train filter
 can absorb gas-phase Hg (Hg2+ and Hg°) and oxidize Hg° resulting in physical and chemical
 transformations within the sampling train.  The rates of these transformations are dependent
 on the properties of fly ash, the amount of fly ash, the temperature, the  flue gas composition,
 and the sampling duration. Samples collected downstream of efficient PM control devices  do
 not contain enough fly ash to significantly alter Hg speciation within the sampling train, but
 samples collected upstream of PM control devices can give erroneous results because of fly-
 ash-induced transformations.
                                         6-15
-------
                                                                              Table 6-4
                                        ICR Mercury Emission Test Allocations by Coal-boiler-control Class
                          No.

                        Ei Group
                    Coal-boiler Control Class
                                POST-COMBUSTION CONTROLS: COLLIDE ESPS

                                Bituminous Coil, PC Boiler with CS-ESP

                                BitUBlinc- a Coil and fa Coke, PC Boiler with CS-ESP
                                Bituminous Coal, PC Boiler with SNCR and CS-ESP
                                Subbituminous Coal. PC Boiler with CS-ESP
                                Subbiluminous/ Bituminous Coal, PC Boiler with CS-ESP
                                Lignite. PC Boiler with CS-ESP
                                POST-COMBUSTION CONTROLS: HOT-SIDE ESPS
                                Bituminous Coil. PC Boiler with HS-ESP
                                lubbioiminous Coil. PC Boiler (Dry Bottom) with HS-ESP
                                Subbraiminous Coal, PC Boiler (Wet Bottom) with HS-ESP
                                Subbituminous' Biluminous Coil. PC Boiler with HS-ESP
                                POST-COMBUSTION CONTROLS: FF BAGHOUSES
                                Biniminoiia Cool. PC Boiler with FT Bsghouae
                                Bituminous CoaLTet- Coke. PC Boiler with FF Bsghouae (Measurements nor valid, disregard!
                                Birununous/SubbitiimrnousCoal, PC Boiler with FF Baghouse
                                SubbiDJianout Coal, PC Boiler with FF Baghouse
                                POST-COMBUSTION CONTROLS: MISCELLANEOUS PM CONTROLS
                                T.X Lignite, PC Boiler with CS-ESP and FT (COHPAC)
                                Sttbbirumiitous Coil. PC Boiler with PM Scrubben
                                POST-COMBUSTION CONTROLS: DRY FGD SCRUBBERS
                                Bituminous Coal, PC Boiler with DSI and CS-ESP
                                Subbraiminous Coal, PC Boiler with CS-ESP/SDA
                                Bituminous Coal, PC Boiler with SDA/FF
                                Bituminous Coal, PC Boiler with SCR and SD/WTF
                                Suhbituminoui Coal. PC Boiler with SDA/FF
                                ND Lignite, PC Boiler with SDA/FF
                                Bituminous Conl. Stoker with SDA/FF
                                POST-COMBUSTION CONTROLS: WET FGD SCRUBBERS
                                Bituminous Coal, PC Boiler with PS and Wet FGD Scrubben
                                Subbiniminous Coal, PC Boiler with PS and Wet FGD Scrubbers
                                ND Lignite. PC Soile! with PS and Wet FCD Scrubbers
                                Bituminous Coal. ?C Boiler with CS-ESP and Wet FCD Scrubbers
                                Subbituminoiis Coal, PC Boiler with CS-ESP and Wet FGD Scrubbers
                                TX Lignite. PC Boiler with CS-ESP and Wet FGD Scrubbers
                                Bituminous Coal. i>C Boiler with HS-ESP and Wet FGD Scrubbers
                                SubbituminoiU Coal, PC Boiler with HS-ESP and Wet FGD Scrubbers
                                Bilumint us Coal, PC Boiler with FF and Wet FCD Scrubber
                                CYCLONE-FIRED BOILERS
                                Lignite, Cyclone Boiler with CS-ESP
                                Subbituminous Coal/Pet. Coke, Cyclone Boiler with HS-ESP
                                Lignite. Cyclone Boiler with Mechanical Collector
                                Lignite, Cyclone Boiler with SDA/FF
                                Bituminous Coal, Cyclone Boiler with PS and Wet FCD Scrubbers
                                Bituminous Coal, Cyclone Boiler with CS-ESP and Wet FCD Scrubbers
                                FLUID11ED-BED COMBUSTORS
                                Lignite, FBC with CS-ESP
                                Anthracite Coal Waste, FBC with FF
                                Bituminous Coll Waste, FBC with FF
                                Bituminous Coal/1'cl. Coke, FBC with SNCR and FF
                                Subbiniminous Ccsl, FBC with SCR »nd FF
                                Lignite, FBC with CS-FF
No. of
 Test
Runs
(Bold numbers In pirtnllicm Indicate no.
             of lett runs)
                                                                            Brayton Point I (3), Bray ton Point 3 (3), Gibson 0300
                                                                            (3). Gibson 1099 (3), Meramcc (3). Jack Wilson (3).
                                                                            Widow Creek (3|
                                                                            Presque Isle 5 (3), Presque Isfc 6 (3)
                                                                            Salem Hurber (3)
                                                                            Montrose (3), George Neal South (3). Newlon (3)
                                                                            Si.Clsir(3>
                                                                            Slanton I (3)

                                                                            Cliflside (3), Gaston<3), Dunkirk (3)
                                                                            Cholla 3 (3), Columbia (3)
                                                                            Plane (3), Presque Isle 9 (3)
                                                                            Clifty (3)

                                                                            Samims(3). Valmont<3)
                                                                            Valley (3)
                                                                            Shawnee (3)
                                                                            Boswell 2 (3), Comanche (3)

                                                                            Bigbrown (3), Mottticcllo 1-2 (3)
                                                                            Boswell 3 (3)

                                                                            Washington (3)
                                                                            GRDA (3), Laramie 3 (3), Wyodak (3)
                                                                            Mecklenburg (3)
                                                                            Logan (3), SEI (3)
                                                                            Craig 3 (3), Rawhide (3). NSP Sherburne (3)
                                                                            Antelope Valley (3), Slanton 10 (3)
                                                                            Dwayne Collier (3)

                                                                            Bruce Mansfield (3)
                                                                            Boswell 4 (3), Cholla 2 (3). Colitnp (3), Lawrence <3)
                                                                            Lewis and Clark (3)
                                                                            AES Cayuga (SK Big Bend (3)
                                                                            Jim Bridger(3). Uramit River I (3). Sam Seymore (3)
                                                                            Monrieello 3 (3). Limestone (3)
                                                                            Charles Lowtnan (3). Morrow (3)
                                                                            Coronado (3), Craig I (3). Navajo 13). San Juan (3)
                                                                            Clover (3), Intennountain (3)

                                                                            Leland Olds (2)
                                                                            Nelson Dewey (3)
                                                                            Bay From (3)
                                                                            Coyote (I)
                                                                            Badly (3)

                                                                            R.M Heskett(3)
                                                                            Kline Township (3)
                                                                            Sciubgnm (3)
                                                                            SlocWon Cogen (3)
                                                                            AES Hawnii (3)
                                                                            TNP(3)
t
          The effects of filtered solids on a filter in the OH sampling train are  shown in Figure 6-4.
These test results were obtained from pilot-scale coal  combustion experiments conducted by the
DOE Federal Energy Technology Center (FETC) [now the National Energy Technology
Laboratory (NETL)]. The OH sampling train speciation data shown in Figure 6-4 were
collected simultaneously in two different manners.  In the  first, tests designated by the symbols
OH-n (n=l, 2,3...), samples were collected by running the sampling train in the prescribed
method by collecting an  isokinetic sample with the  probe nozzle facing upstream.  In the second
manner, tests designated by MOH-n (n=l, 2, 3,...) were run with the probe nozzle facing
downstream so that the PM entering the train would be minimal2.
                                                                                  6-16
-------
          12
          10
s  6
2
o
14
                                                                         •Elemental
                                                                         B Oxidized
                                                                         D Participate
                                                      V
              Figure 6-4. Effect of OH sample filter solids on Hg speciation.

    The results of these experiments show that, for each of the simultaneous runs, the values of
Hgj can be considered to be equal when taking into account sample variations resulting from the
imprecision of the OH Method. However, the samples taken with the probe facing upstream
indicated higher concentrations of Hgp and Hg2+ than the samples with the nozzles facing
downstream.  This provides evidence that PM collected on the filter of the train facing upstream
resulted in the oxidization and adsorption of Hg as flue gas passed through the sampling train.
This and other evidence indicate that in some cases the use of the OH Method to collect
speciation samples upstream of PM control devices provides questionable results3.

6.6 FUEL, BOILER, AND CONTROL TECHNOLOGY EFFECTS

       Based on current understanding of speciation and capture of mercury, it is believed that
the ICR data represent a number of subpopulations corresponding to fuel-boiler-control
combinations. Sections 6.6 and 6.7 provide an interpretation of physical and chemical
phenomena that can be used to characterize the roles that coal, combustion, and flue gas
cleaning variables play in the speciation and capture of Hg.  Section 6.8 provides a summary
of national emission estimates that were based on data described in Sections 6.6 and 6.7.
Conclusions are provided in Section 6.9.

       The interpretations in Sections 6.6  and 6.7 are based on previous bench-, pilot-, and
full-scale tests, plus  a number of different modeling efforts related to speciation and capture of
Hg in coal-fired boilers. While we have attempted to provide an internally consistent
interpretation of the  data, some of the observed results are inconsistent with the current
theories on the behavior of Hg. In these instances, either our interpretations may be incorrect
and other factors may account for the apparent discrepancies in results, or the data may be
incorrect.  It is believed that some discrepancies result from questionable OH Method or from
                                        6-17
-------
t
errors in reporting test results.

       The evaluation of ICR Phase HI data indicates that air pollution control technologies
now used on coal-fired utility boilers exhibit levels of control that range from 0 to 99 percent
reduction of Hgr. The level of control varies with the coal, combustion conditions, and flue
gas cleaning methods used at individual sites. In some instances, there is substantial variation
in the three tests conducted at individual sites.  The run-to-run variations at any given site can
result from actual variations in emissions or with problems associated with the measurement
method.

       The OH Method is relatively complex, and measurement method problems can result
from errors that occur:
    •   during the collection of samples,
    •   in extracting samples from the sampling train,
    •   from the chemical extraction of Hg from the nozzle and probe wash, from the sample
       train filter, and from the different impingers,
    •   from Hg analysis, and
    •   from data reduction and transcription.

Some errors are inevitable in spite of the best efforts of everyone involved in the measurement
process.

       In statistical terms, the OH data represent a very small number of samples of the
underlying population.  Each individual test represents the average of flue gas concentration of
speciated Hg during a short "snapshot" in time.  Run-to-run variations at any given site result
from temporal variations in coal properties, combustion conditions, and emission control
technology process conditions. There are also site-to-site variations within a given coal-
boiler-control class and variations between classes. Even considering these sample population
variations, the ICR data provide a great deal of information, when evaluated in the context of
current knowledge on the behavior of Hg in coal-fired electrical generating units.

       Table  6-5 shows differences in the average reduction in Hgr emissions for coal-boiler-
control classes that bum pulverized coal. Plants that employ only post-combustion PM
controls display class average Hgy emission reductions ranging from 1 to 90 percent.  Units
with FFs obtained the highest average levels of control. Decreasing average levels of control
were generally observed for units equipped with a CS-ESP, HS-ESP, and PS. For units
equipped with dry scrubbers, the class average HgT emission reductions ranged from 2 to 98
percent.  The estimated class average reductions for wet FGD scrubbers were similar and
ranged from 10 to 98 percent.

       For PC-fired boilers, the amount of Hg captured by a given control technology is
greater for bituminous coal than for either subbituminous coal or lignite.  For example, the
average capture of Hg based on OH inlet measurements in PC-fired plants equipped with a
CS-ESP is 36 percent for bituminous coal, 9 percent for subbituminous coal, and 1 percent for
lignite.

                                        6-18
-------
Errata Page 6-19, dated 3-21-02
                                     Table 6-5
       Average Mercury Capture by Existing Post-combustion Control
                    Configurations Used for PC-fired Boilers
Post-
combustion
Control
Strategy
PM Control Only
PM Control and
Spray Dryer
Adsorber
PM Control and
Wet FGD
System(a)
Post-combustion
Emission
Control Device
Configuration
CS-ESP
HS-ESP
FF
PS
SDA+ESP
SDA+FF
SDA+FF+SCR
PS+FGD
CS-ESP+FGD
HS-ESP+FGD
FF+FGD
Average Mercury Capture by Control Configuration
Coal Burned in Pulverized-coal-fired Boiler Unit
Bituminous Coal
36%
9%
90%
not tested
not tested
98%
98%
12%
74%
50%
98%
Subbituminous
Coal
3%
6%
72%
9%
35%
24%
not tested
-8%
29%
29%
not tested
Lignite
-4%
not tested
not tested
not tested
not tested
0%
not tested
33%
44%
not tested
not tested
       (a) Estimated capture across both control devices

       CS-ESP = cold-side electrostatic precipitator
       FF = fabric filter
       SDA = spray dryer adsorber system
HS-ESP = hot-side electrostatic precipitator
PS = particle scrubber
6.6.1 Coal Effects

       While OH speciation measurements may not provide an accurate characterization of
the speciation at the inlet sampling location, transformations within the sampling train provide
an indication of the fly ash reactivity, and potential for Hg adsorption. SPFs for selected coal-
boiler-control classes are summarized in Table 6-6. The data in Table 6-6 are class average
SPFs for PC-fired boilers at the inlet and outlet sampling locations. Data are shown for
bituminous, subbituminous, ND lignite, and TX (ignite.  Relatively high levels of SPFP at the
inlet indicate that the Hg was either present as Hgp in the flue gas, or it was readily absorbed
by fly ash on the sampling train filter. Relatively high levels of Hg2+  at the inlet indicate that
Hg at the inlet sampling location was either already oxidized or oxidized as the flue gas passed
through the sampling train.  Relatively high levels of measured Hg° indicate that there were
relatively high levels of Hg° in the inlet flue gas.

       The units burning bituminous coal exhibited relatively high levels of SPFP and SPF 2+
in the inlet samples. It is hypothesized that high levels of SPFP + SPF2+, or alternatively low
SPF°, in the inlet sampling train indicates a high probability that Hg can be readily captured in
downstream APCD(s).  For the biruminous-coal-fired units, values of SPFP and SPF 2+ ranged
from 0.03 to 0.92, while values of SPF° ranged from 0.01 to 0.37. The HS-ESP unit exhibited
the highest level of Hg° followed by units equipped with SDA/FF systems. HS-ESP units
                                        6-19
-------
t
                operate at temperatures where Hg° is not easily oxidized or captured. The SDA/FF units
                exhibited a 98 percent capture of Hgr, and the relative concentrations of the SPF2+ and SPF°
                measurements at the stack sampling location were 0.22 and 0.77, respectively. This could
                result from the efficient capture of Hg2+ in these units.

                      The PC-fired units burning subbituminous coal exhibit inlet SPF° values ranging from
                0.44 to 0.84. The summed SPFP + SPF2+ values for the CS-ESP and HS-ESP units were
                similar.  Both of these classes of units exhibited Hgi captures of 9 percent.  The moderately
                low HgT captures for the SDA/ESP (38 percent) and SDA/FF (25 percent) are reflected by the
                summed inlet SPFP + SPF 2+ values for these units. The units with FF systems (72 percent
                average capture) had measured average inlet SPF° values of 44 percent.

                      There were a limited number of tests for units firing lignite. The units burning ND
                lignites tend to have a higher SPF° values than units burning TX lignites. The CS-ESP units
                burning ND lignite exhibited an average inlet SPF° value of 0.98.  While there was no
                comparable test unit that fired TX lignite, a unit equipped with a CS-ESP + FF exhibited an
                average inlet SPF° of 0.60. While the inlet measurements for the CS-ESP + FF unit were
                taken downstream of the CS-ESP, a higher SPF° would have been expected if the TX lignite
                were to provide similar speciation results as the ND lignite. Moderate to average SPF° values
                (0.47) were also noted for the CS-ESP + wet FGD units using TX lignite. Inlet measurements
                for these units were also made downstream of a CS-ESP.

                      The similarities between inlet and outlet SPF values can also be used to identify
                instances where the measured inlet speciation values provide a good estimate of the true Hg
                speciation  in the flue gas at the inlet sampling location.  Units with similar inlet and outlet
                SPFs are identified by an (a) in Table 6-6. These cases correspond to tests in which the
                capture of Hgj is < 25 percent for many of the units firing subbituminous coals and ND
                lignite (e.g., comparison of the respective inlet and outlet values for SPFP).
                                                        6-20
-------
 Errata Page 6-21, dated 3-21-02
                                    Table 6-6
        Effects of Coal and Control Technology Inlet and Outlet SPF
                       and Capture for PC-fired Boilers
Coal-Control Class
Bituminous
CS-ESP
SNCR and CS-ESP
HS-ESP (a)
FF
SDA/FF
SCR and SDA/FF

Subbituminous
CS-ESP (a)
HS-ESP (a)
FF
SDA/ESP
SDA/FF (a)

ND Lignite
CS-ESP (a)
SDA/FF (a)

TX Lignite
CS-ESP + FF
CS-ESP + Wet FGD
Inlet
SPFD

0.35
0.92
0.09
0.92
0.59
0.82


0.05
0.02
0.33
0.13
0.01


0.01
0.03


0.09
0.00
SPF2+

0.58
0.03
0.53
0.04
0.28
0.17


0.25
0.15
0.23
0.26
0.06


0.01
0.04


0.31
0.52
SPF

0.07
0.05
0.37
0.04
0.15
0.01


0.70
0.83
0.44
0.61
0.84


0.98
0.93


0.60
0.47
Outlet
SPFD

0.02
0.20
0.04
0.01
0.01
0.05


0.00
0.00
0.01
0.00
0.01


0.00
0.00


0.00
0.01
SPF2+

0.78
0.35
0.59
0.52
0.22
0.46


0.31
0.17
0.87
0.05
0.05


0.04
0.03


0.70
0.14
SPF°

0.20
0.45
0.37
0.47
0.77
0.48


0.69
0.83
0.12
0.94
0.94


0.%
0.97


0.30
0.85
% Red
HgT

36
91
9
90
98
98


3
6
72
35
24


-4
0


NA
44
(a) Units with similar inlet and outlet SPF values.

6.6.2  Control Technology Effects

       Control technology effects are inseparable from coal and boiler effects.  In the
following sections, post-combustion control technology effects will be evaluated in terms of
the three major types of controls currently used for coal-fired utility boilers: PM controls, dry
FGD scrubbing controls, and wet FGD controls. These evaluations will be discussed initially
in terms of control technology and coal effects on PC-fired boilers. The speciation and capture
of Hg from cyclone-fired combustors, FBCs, and IGCC units will then be discussed.

       A summary of test results for each of the coal-boiler-control classes for which ICR Hg
emission data were collected is given in Table 6-7. The data include information on the
number of tests for each class, the average emission factors for Hgp, Hg2+, Hg°,  and Hgr, and
the average and range of HgT emission reductions.
                                       6-21
-------
   J
   u
    2
   I

    o
   •c
    u
 ea  S3
H P«
   •o
    s
   Ed
   I
    &
    2
    Si

*
X
s
I


^^
t
w
ffi
1
*



4
J

"8
6






\
I
!
1
li
l
I



S
<

1

S


V

*
*
.
J.
s
0
Ji

„
*






j
J
J
:
!
i
D.
I
U
S


























POST-COMBUSTION CONTROLS: COLD-SIDE ESP>



o
«
8
0
5
oo
3
0

—

«
c>
5

*•"!

00


f*






Bituminous Coal, PC Boiler with CS-ESP

' '

*
s
R
f\
*
§
O

ri

0-
d
o
0


^
•*








Biluminous Coal and Pet Coke, PC Boiler and CS-ESP



g
g
r-
i
s
s
0

0

o
o
„
0

d










Bituminous Coal, PC Boiler with S NCR and CS-ESP




00
o

i
s
o

4

_
0
g
o


s









Subbituminous Coal, PC Boikr wilh CS-ESP



R
^
-
M
S
"
*
0

in

g
0
8
O


P









Subbituminous/ Bituminous Coal, PC Boiler with CS-ESP



5

S

oo
-
_
0

&>

O
d
S
0


oo









Subbiiuminous Coal, PC Boiler wilh PM Scrubbers

























































£
u
POST-COMBUSTION CONTROLS: DRY FGD SCRUBS



<*
5
S
*
si
^
0



ft
0
s
0

0
o
d
oo

3









Subbituminous Coal. PC Boiler with CS-ESP/SDA



_
*
5
£
8
fM
O
0

y

s
0
§


d
s









Bituminous Coal, PC Boiler withSDA/FF



w
.
S
r-
s
CM
o
0

o

0
o
s
o
=0
d
a









Bituminous Coal. PC Boilerwith SCR and SDA/FF



§
a
§
o
•*
*
d

r<:

0
o
0


r-










Subbituminous Coal PC Boiler withSDA/FF



S
""
s
d
oo
g
~—



,
o
8
O
3!

a









ND Lignite. PC Boiler with SDA/FF



VI
3
3
ff.
$
S
0

o

S
o
0
o
„
0










BituminoiE Coal, Stoker with SDA/FF



                                                                                   c
                                                                                  •a
                                                                                   B
                                              6-22
-------
   I
   IA
   «
   u
   8
   u
   I

   i
S
IS
U

I

i
00

I

SS
Red in He



u
S

*
B
g
"

£


B

£
£

•3
i


.
(
i
.
*
t
B

t
X
.
EC



MM
s
»

i
o
S
a

e
H


I
i
•
i
•i
^
D
1 Ea Group






















V.
!•
B
B
e
g
1
c
s-
z
z
i
S
S

0
fl
r
CM
^
r*-
3

iy
C
S
c



!_,
o
s
c
r-

-
"




J
c
£
1
1
g
h
I
y
a
-
^
c
c
r-
*
K


..



<*
d
IP-
c
c


s





Siibhituininn.M CnaL PC Boiler with PS and Wet FGD Scmhhe«
-
r-
•*
r*
5

c
v
c



c
c
s
e
2
w*
"
2





NT) Lienite PC Hoi If r with PS and Wrt FGD Scrubber;
r*
£
"
.06-64.01
^

o
.
c



e
c
§
c
1

r-




t
Bituminous Coal PC Boiler wilh CS-ESP and Wet FGD Scruhne
-
P
r
f
v

c:
r
p-
c



c
c
8
o
s

f
~



1
SubbituminniK Cnal PC Boiler with CS-ESP and Wet FGD Scru
Wl
r-
••a
£
'

e
„
c




c
i
c








TX Li anile PC Boiler with CS-ESP and Wei FGD Scrubber*
,
r*
3
^0-27^6 	
V

O
a
c




c
c
c
ot






?
Rituminoua Coal PC Boiler with tK-FSP and Wet FGD Srnihhe
-
r


.
(X
S


C
w
c



,4
c
K
c
0
5







I.ienite. FBC with CS-RSP
-
f
S
P

SI
d
g
c



*
c
g
c
-

s





Anthracite Cnal Waste. FBC with FF
-
o
3
i
s
g

c
a




^
c
s
a
:







RituminniN Coal Waste FBC with FF
-
V
•
i
i

4
d
r
c



c
c
c
=
r

C"





Bituminous Coal/Ptt Coke. FBC with SNCR and FF
"
t~
r-
V
V
5
2

C
_
e



c
c
s
c
P

P





SuhhituminniM Coal FBC with SCR and FF
^
g




0
^
c



^
c
8
C


tt
0





It
1
5
U
'c
e
,
                                                     S2
                                                     2


                                                     I
                                                     o
                                                     u
                                                     (4-1
                                                     O
                                                     O
                                                     cfl
                                                     .S
                                                     cs
                                                     o
                                                     1/3

                                                     £
                                                     00
                                 6-23
-------
6.6.3  Post-combustion PM Controls
       In 1999, 72 percent of the coal-fired electric utility boilers in the U.S. used post-
combustion controls that consisted only of PM controls. The Phase IICR revealed that there
were 890 units that used only post-combustion PM controls. This included 791 units using
either CS- or HS-ESPs and 80 units that used FF baghouses. The number of boiler units in the
U.S. equipped only with PM controls is shown in Table 6-8 along with the number of test
units in each PM control category.

                                      Table 6-8
 Number of Coal-fired Utility Boilers Equipped with Particulate Matter Controls Only
Particulate Matter
Control

CS- or HS-ESP (a)
Two ESPs in series
Fabric Filter
ESP w/ Fabric Filter
Particulate Scrubber
ESP w/ Particulate
Scrubber
Mechanical Collector
Number of Units
Utility Industry
791
2
80
6
5
4
2
Test Units
25
2
12
2
1
0
1
                 (a) 14 CS-ESPs and 9 HS-ESPs were tested

6.6.3.1 Cold-side ESPs

       A total of 14 PC-fired units equipped with CS-ESPs were tested.  The types of fuels
that were used in these tests are given in Table 6-9.

                                     Table 6-9
             Type of Fuel Used in PC-fired Units Equipped with CS-ESP
Type of Fuel
Bituminous
Bituminous & Pet. Coke
Subbituminous
Subbituminous/Bituminous
Total
No. of Test Units
8
2
3
1
14
       One of the units burning bituminous coal was also equipped with an SNCR system for
NOx control.  One cyclone-fired unit that burned lignite was also tested. The results of Hg
emission tests on PC-fired units equipped with a CS-ESP are given in Table 6-10.
                                       6-24
-------
                                               Table 6-10
                            Post-combustion Controls: Cold-side ESPs
RltiiminniK Oial. Pr Roller
Bravton foot I     !     2.01
Brayton Point I   |  2  I   2.61
Braylon Point I   I  3  i   2.17
Bravton Petal 3     3  I   1.40
 	Average;  _
 libwn 0300       1     1.94
 itbson0300     131   1.75
                   m
                   5.53
                                                                                  3.35  |  69.?7
                                                                                  2.30     71.96
                                                                                          29.M

                                                                                          44.75   I
Maximum
STDEV   	
Bitaminaui Tail and Pet Coke. PC Boiler with CS-ESP
                                                                                                                     t
                                                                             (continued)
                                                   6-25
-------
                                   Table 6-10 (cont'd)
                       Post-combustion Controls: Cold-side ESPs
                                                                                   820l_
                                                                                   83.54..
                                                                                   8322
                                                                                 ssssagSSi
    	 Average
George Neal So.
GewEeNealSo
George Neal So.
    	verage
Newton__
Newton
Newton
      Averaee
                                             15.18   :   -6.90
                                               ?	-47.37
                                             -52.29      -936

                                             -0.10   I  -14.00
                                             16.33    . . -9.28
                                             8.46   I   -7.8J
                                                                           -52.29    -47.37
                                                                           17.46    83.54
                                                                           21.75    50.89
.with CS-ESP I
1.97    6.79	[_ J6.26
 S	  «.39  '  14.36
Siibhituminaus/ Bituminous CoA]»
SlCbff	_ i  3
 	 .Average
Lignite, PC Boiler

       The test units with a CS-ESP display significant run-to-run differences (variations) in
the Hgt (inlet), Hgj (outlet), and % Hgy reduction. These differences may result from the
changing HgT inlet concentrations, changing boiler and control device operating conditions, or
sampling and analysis problems. Two important variables that affect Hg capture are changes
in Hg inlet concentration and unit  operating temperatures.

       Run-to-run variations for test units burning bituminous coal in PC-fired boilers
equipped with CS-ESPs are shown in Figure 6-5.  While the class average HgT reduction for
these units was 36 percent, the run-to-run emission reductions in Hgj range from 0 to 81
percent.  All  inlet and outlet Hgy concentrations for the Widow Creek, Jack Watson, Brayton
3, and Brayton 1 were similar.  The Meramec plant exhibited relatively high HgT reductions as
did run 2 on Gibson 1099. Gibson 0300 exhibited high stack gas concentrations of Hgj, and
run 1 on Gibson 0300 had a higher outlet HgT concentration than at the inlet. The unit-to-unit
variations in  HgT emission reductions for these same units are shown in Figure 6-6. The
average emission reduction for the seven 3-run tests  shown in Figure 6-6 is  still 37 percent,
but unit-to-unit emission reductions  range from 3 percent for Gibson 0300 to 74 percent for
Meramec.  The speciation of Hg for the bituminous coals is predominantly Kg2*.
                                          6-26
-------
       In Figure 6-6, there are two unit test averages given for Gibson. Both averages are for
the same unit, Gibson 0300.  The unit average for Gibson 1099 is for tests conducted in
October 1999, while the average for Gibson 0300 is for tests conducted in March 2000. The
tests in October and March used coal from the same source. Average unit reductions in Hgj
for the October and March tests were 35 and 3 percent, respectively. The apparent
discrepancy in the test results led plant engineers to investigate.  The investigation indicated
that steam-cleaning of the air preheater during the collection of OH samples was the probable
cause of these inconsistencies.

       The Hg speciation and Hgi reductions for PC-fired units equipped with CS-ESPs and
burning subbituminous coal and lignite are shown in Figure 6-7. Hg emission reductions for
the units range from -4 to 12 percent, exhibiting little if any Hg capture.  The relative
concentrations of Hg° in the stack gas are higher than those observed for units firing
bituminous coal.
     Widow Creek
      Jack Watson
        Mcramcc
      Gibson 1099
      Gibson 0300
        Brayton 3
        Brayton 1
                         10           20           30          40          50
                           Total Mercury Concentration, ug/dscm @ 3% Qz
Figure 6-5. Inlet and outlet mercury concentrations for bituminous PC-fired boilers with
                                       CS-ESP.
                                         6-27
-------
   Widow Creek
      [53%]

   Jack Watson
_    [30%]
^
o
.2
      Meramec
       [74%]
•a   Gibson 1099
OS      [35%]
Sf
•—  Gibson 0300
»•
              m
      Brayton 3
       [28%]

      Brayton 1
       [28%]
                                                      0Hg(2+)Out DHg(0)Out
             0
                                                                           40
                                                                                   45
                    5       10      15      20      25      30      35
                             Mercury Concentration, jig/dscm @ 3% O2

Figure 6-6. Mercury emissions from bituminous-coal-fired PC boilers with CS-ESP.
        Stanton 1 [-4%]
   J     Newton [8%]
   1
      George Neal South
           [12%]
         Montrose [9%]
                                                    ul SHg(2+)Out PHg(0)Oul
                     0       2       4       6       8       10      12      14

                              Mercury Concentration, jig/dscm @ 3% O2

  Figure 6-7. Mercury emissions for subbituminous- and lignite-fired PC boilers with
                                       CS-ESP.
                                        6-28
-------
       Run-to-run variations on a given unit can be attributed to operating variables such as
inlet Hg concentrations, operating temperature, soot blowing, reentrainment losses within an
ESP, or the imprecision of the OH Method.

       Mercury outlet concentrations can be expressed by:

       HgT (outlet) = HgT (inlet) - Hgp (captured in the control device)

              + Hgp (reentrained and escapes the control device)

              - Hg° or Hg2+ (adsorbed and captured within the control device)

              + Hgp, Hg2+, or Hg° (desorbed or is reentrained and escapes capture)

       Deposits or captured fly ash between the inlet and outlet sampling location (the stack)
can adsorb or desorb gas-phase Hg, depending on time-dependent changes in the inlet Hg
concentration and operating temperatures downstream of the inlet sampling location.
Temperature effects can be understood by considering the deposits and collected fly ash
between the inlet and stack locations to be a complex system that adsorbs and desorbs Hg.  If
the system has reached equilibrium in terms of operating conditions, there will be a constant
relationship between the inlet and outlet concentrations of Hg. Increases in operating
temperatures within the system can increase the rate at which Hg is desorbed, resulting in
increased outlet concentrations relative to the inlet concentrations. Temperature decreases  can
increase Hg adsorption within the system. This can cause a decrease in the Hg outlet
concentrations relative to the inlet concentrations.

       Temporal changes in inlet and outlet Hg concentrations are the result of hysteresis or
history effects.  Hypothetical changes in Hg reduction for three tests on a single unit that could
occur because of the time lag between changing inlet and outlet Hg concentrations are
illustrated in Figure 6-8. In this illustration, Hg emission reductions during runs  1,2, and 3
averaged 30, -15, and 40 percent, respectively. The -15 percent indicates that the measured
outlet Hg concentrations were higher than the inlet concentrations.
                                         6-29
-------
   HgT
                                           71 ire
 Figure 6-8. Hypothetical effect of inlet and outlet HgT concentration changes on run-to-
                                  run HgT capture.

       Changes in the fly ash carbon content, changes in unit operating conditions such as
load, and diurnal changes in temperature may also result in hysteresis effects. The ICR tests
for each unit represent a snapshot in time. Additional OH Method tests or tests with Hg
CEMs are needed over an extended period of time to more fully characterize the effects of
coal, combustion, and control technology variables on stack emissions of Hg.
6.6.3.2 Hot-side ESPs

       Eight ICR units that burn pulverized coal and that were equipped with an HS-ESP
were tested. Three of these units burned bituminous coal; four burned subbituminous coal;
and one burned subbituminous and bituminous coal. A ninth, a cyclone-fired unit equipped
with an HS-ESP, burned subbituminous coal and petroleum coke. Hg test data for the eight
PC-fired units are given in Table 6-11.
                                       6-30
-------
                                        Table 6-11
                        Post-combustion Controls: Hot-side ESPs
                                           3.8*  i   0.10  .  2.27
                                                        3.97
4.3I...   38.00
6.61  ,   42.51
Suhhltuminout C
               f CBollerXBry Bottom) with HS=ESfl
                nfl?  !   nv7   \  101   I  517
                                                  0.01   I  0.01	.  IJO
                                                  0.01     0.39     1.27
                                           10.30  j   0.00  ,  2,16   ; J1JB
                                           10.35  i   0.00  i  2.65
                            ,  14.« _  19.24  i.. 54.43
                           J	Hi.  I 8.26  I  21.77
                    __.  4.15  _;	9.82
                                           ZJ4_.. .   0.07  :  5.50
                                           8.02     0.70    3.60
       As shown in Figure 6-9, the units that fired bituminous coal exhibited average
emission reductions of 18 percent (Dunkirk), -17 percent (Gaston), and 27 percent (Cliffside).
In Figure 6-10, the HS-ESP units that burned subbituminous coal and lignite exhibit Hg
emission reductions of 2 percent (Cholla), -1 percent (Columbia), -3 percent (Platte), and -6
percent (Presque Isle). Stack concentrations of Hg° were substantially higher for the units
burning subbituminous coal than for those burning bituminous coal.

        Hot-side ESPs tend to exhibit poor capture because they operate over a temperature
range where the oxidization and adsorption of Hg° is limited.
                                          6-31
-------
                                                                                                           1
               Dunkirk [18%]
            e
            .2
            BI
            a
            •*•*
               Gaston [-17%]
• Hg(p)Out HHg(2+)Out DHg(0)Out
               Cliffside [27%]
                          0          2          4         6          8         10

                                   Mercury Concentration, jig/dscm @ 3% Oj





     Figure 6-9. Mercury emissions from bituminous-fired PC boilers with HS-ESP.
      Prcsque Isle 9 [-6%}

           Plane [-3%]
    BC
   EC
   —    Columbia [-1 %]
   •SS*
    B
          Cholla 3 [2%]
                               IHg(p)Out BHg(2-t-)Out  DHg(0)Out
                    0     2     4      6      8      10     12     14     16


                            Mercury Concentration, H-g/dscm @ 3% O2
Figure 6-10. Mercury emissions for subbituminous- and lignite-fired PC boilers with HS-

                                           ESP.
                                           6-32
-------
6.6.3.3  FFBaghouses
       Six PC-fired units with FF baghouses were tested. The results of one test unit (Valley)
were omitted from the results because of data quality problems. The unit name, type of coal
burned, and reduction in HgT are given in Table 6-12 for the five units with valid test data.

                                     Table 6-12
             Mercury (HgT) Reduction at PC-fired Units with FF Baghouses
Unit
Sammis
Valmont
Shawnee
Boswell 2
Comanche
Coal
Bituminous
Bituminous
Bituminous/subbituminous
Subbituminous
Subbituminous
Reduction in HgT,
%
92
87
70
83
62
       Detailed test results for the five units listed in Table 6-12 are given in Table 6-13. The
average run-to-run HgT reductions for the FF units ranged from 53 to 92 percent.  The class
average emission reductions for the two bituminous-coal-fired units was 90 percent, the
average for the single unit that fired bituminous and subbituminous coals was 70 percent, and
the class average for the two units that fired subbituminous coal was 72 percent. There were
generally high stack concentrations of Hg2+ for all FF units.  Hg° can be oxidized as it passes
through the FF, either from reactions with fly ash on the filter cake or from reactions with bag
filter material. This can lead to relatively low concentrations of Hg° in the stack gas. These
observations may not apply to all bag filter materials, or units that burn either lignite or
subbituminous coal.

6.6.3.4  Comparison of ESPs andFFs

       The average unit-to-unit reductions in HgT in the inlet and outlet of PC-fired units
equipped with a CS-ESP, HS-ESP, or FF baghouse are shown in Figure 6-11. Stack
concentrations and speciation  results are shown in Figure 6-12. SPF results are shown in
Figure 6-13.

       The best Hg capture is exhibited for units equipped with a FF (72 to 90 percent
average reductions). This is followed by units that are equipped with a CS-ESP and that burn
bituminous coal or bituminous coal and petroleum coke (35 to 54 percent average reductions).
Poor capture (-4 to 9 percent average reductions) is shown for all units that are equipped with
a HS-ESP and for units that are equipped with a CS-ESP and burn either subbituminous coal
or lignite. Units, which exhibit poor HgT capture, display higher SPF° values than units that
have good HgT capture. In units that bum bituminous coal or bituminous coal and petroleum
                                        6-33
-------
coke, Hg2* constitutes more than half of the total Hg in the stack gas. This is also true for the
unit that is equipped with a FF and burns subbituminous coal.
                                          Table 6-13
                          Post-combustion Controls: FF Baghouses
           o«]/P£t-
 VaUey		  I.. __SM
 Valley        I 2   0.05
with FE Ba&houstXMeasurtments noLyaud^di

 Ritnmlnnus/Siilihltilminom Coal. PC Boiler with FF Baehoust
 SuhhitinninniH Pnal PC Rnllor with FF Ravhittltp
                                                                                 76.06     82.34
                                             6-34
-------
          Sub(wet)/HS-ESP
              [-3%]
      I
      s

      £
      o
      U
              Bit/CS-ESF
                [36%]

          [72%]

          Bit/FF
          [90%]
                                                 J
                                                               Inlet H Outlet
                              2      4     6      8     10     12

                            Mercury Concentration, Hg/dscm @ 3% O2
                                                             14
16
Figure 6-11. Mercury emission reductions for PC-fired boilers with ESPs and FFs.
    Sub(wet)/HS-ESP
        [-3%]

       Sub/HS-ESP
ft
w
i
       Bit/HS-ESP
Lig/CS-ESP
   [-4H]
Sub/CS-ESP
   [3%]

Bit/CS-ESP
  [36%]

   Sub/FF
   [72%]

   Bit/FF
   [90%]
                 s:
                                                 I Hg(p) Out S Hg(2+K>ut D Hg(0) Out
                0        2        4        6        8        10       12       14

                          Stack Concentrations of Mercury, ng/dscm @ 3% O2


     Figure 6-12. Mercury speciation for PC-fired boilers with ESPs and FFs.
                                       6-35
-------
                                               • Hg(p) Out HHg(2+) Out DHg<0) Out
     Sub{wet)/HS-ESP
         [-3%]

        Sub/HS-ESP
 .
 •w
 w
E
Q
K
 3
         Bit/HS-ESP
         Lig/CS-ESP
           t-4%]

        Sub/CS-ESP
        Bit/CS-ELSP
         [36%]

          Sub/FF
          [72%]

          Bit/FF
          [90%]
                             20           40           60           80

                                Relative Mercury Concentration in Stack, %
                                                                              100
       Figure 6-13. Relative mercury speciation for PC-fired boilers with ESPs and FFs.
6.6.3.5 Other PM Controls

       Other PM control methods that were tested included two units firing TX lignite and
equipped with a CS-ESP followed by a pulse-jet FF baghouse, and one PC-fired unit burning
subbituminous coal and equipped with a PM scrubber (see Table 6-14).  The three-run average
Hgr reduction across the PM scrubber on this latter unit was 9 percent.

       At the Bigbrown and Monticello units, the inlet and outlet Hg measurements were
made across the baghouse. There is little consistency between three runs for the Monticello
unit, and the data may not be valid. Bigbrown exhibited negligible Hgy capture across the FF.
While some Hgp and Hg2+ may have been captured in the upstream ESP, the low amounts of
fly ash captured in the downstream FF  probably account for the lack of Hgj capture in the
baghouse.
                                        6-36
-------
                                     Table 6-14
                Post-combustion Controls: Miscellaneous PM Controls
                            8,82_ _ I 36,21. .
                            46.29  ! 74.TO
                            14.07  i 15.16
 6.6.4 He Capture in Units with Dry FGD Scrubbers

       Thirteen units with dry scrubbing systems were tested.  One unit uses dry sorbent
injection in combination with a CS-ESP, three units use SDA/ESP systems, and the
remaining nine units are equipped with SDA/FF systems. Two of the units equipped with
SDA/FFs were also equipped with a SCR system.  Hg emission test results for the dry
scrubber units are summarized in Table 6-15.

       At the Port Washington unit, sorbent is injected downstream of the air preheater. OH
inlet measurements were made upstream of the preheater, and outlet measurements were
made in the duct downstream of the CS-ESP. The average capture of HgT for the Port
Washington dry sorbent injection unit was 45 percent.  The SPF2+ and SPF° values for this
unit fell within the range of values exhibited by PC-fired boilers that are equipped with a CS-
ESP and burn bituminous coal. The three pulverized subbituminous-coal-fired units
equipped with a SDA/ESP system exhibited  average HgT captures of 25 percent (GRDA), 40
percent (Laramie 3), and 41 percent (Wyodak).

       As mentioned above, nine  units equipped with a SD/FF system were tested. One unit
firing bituminous coal had a Hgt capture of 98 percent.  The two units firing bituminous coal
and also equipped with an SCR system had a class average HgT capture of 99 percent. Three
SDA/FF units fired with subbituminous coal had HgT captures of 36, 32, and 5 percent.
                                       6-37
-------
-------
                                  Table 6-15 (cont'd)
                    Post-combustion Controls: Dry FGD Scrubbers
                             7.80.	;  g.34    13.85 .j	0.01
                             7.82    B.45    I6.C3  ,  0.02
Bituminous, Stoker with SBAJEE
       The average Hgj captures in two units firing lignite were 1 and -1 percent, A single
stoker-fired boiler burning bituminous coal had a total average Hg capture of 94 percent.

       The reduction in emissions for each SDA test class is shown in Figure 6-14.  The
stack concentrations of Hgp, Hg2+, Hg°, and Hgt are shown in Figure 6-15 along with the
average total Hgr capture for each SDA class. The relative Hg speciation for the same coal-
fired boiler classes is shown in Figure 6-16.  The predominance of Hg° in the stack emissions
from units fired with subbituminous coal and lignite is attributed to low levels of Hg°
oxidization and the relative ease of Hg2+ capture.
                                         6-39
-------
*.
e"
•a

I
B
Q
U

5
Sub, SDA/ESP
   [35%]
 Lig, SDA/FF
 Sub, SDA/FF
   [24%]
  Bit, SDA/FF
    [98%]
                            4        6        8        10       12

                            Mercury Concentration, iig/dscm @ 3% O2
                Figure 6-14. Mercury control for dry FGD scrubbers.


e
3
X
a
U

U

Sub, SDA/ESP
[35%]

Lig, SDA/FF
[OH]

Sub, SDA/FF
[24H]

Bit, SDA/FF
[98%]

§

S

I

I j • Hg(p) Out H Hg(2+) Out D Hg(0) Out ,
                      0       2       4       6      8       10      12

                            Mercury Concentration, ug/dscm @ 3% O2


           Figure 6-15. Mercury speciation for PC boilers with SDA.
                                    6-40
-------
  e
•2
oc
M

Q
£
  a
        Sub, SDA/ESP
           [35%]
         Lig, SDA/FF
                                      • Hg(p)0ut  HHg(2+)OutnHg(0)Out |
         Sub, SDA/FF
           [24%)
         Bit, SDA/FF
           [98%]
                  0     10     20    30     40    50    60     70    80    90     100

                               Relative Mercury Concentration in Stack, %


          Figure 6-16. Relative mercury speciation for PC boilers with SDA.


6.6.5 He Capture in Units with Wet FGD Scrubbers

       The wet FGD scrubber systems that were tested consisted of units equipped with four
PM control device configurations. These different configurations are expected to have
different effects on the speciation and capture of Hg. These different configurations included
units equipped with a PS, a CS-ESP, an HS-ESP, or a FF baghouse. Inlet and outlet
measurements on the PS + wet FGD units were made across both control devices. Inlet
measurements on the systems with an ESP or FF were made between the PM control device
and the FGD scrubber. Outlet measurements were made in the stack.

       A total of 23  units with wet FGD systems were tested. Seven units used PM scrubber
systems to control particulate emissions, eight used CS-ESPs, six used HS-ESPs, and two
used FF baghouses.  Twenty-one of the test units burned pulverized coal.  The other two test
units burned bituminous coal  in cyclone-fired boilers. One unit was equipped with a PM
scrubber, and the other had a CS-ESP. The number of PC-fired test units in each coal-control
class is shown in Table 6-16.  (Also see Tables 6-4 and 6-6.)
                                       6-41
-------
                                     Table 6-16
                     PC-fired Boiler PM Controls for Wet FGD Systems
PM
Control
PS
CS-ESP
HS-ESP
FF

Number of Test Units
Bit.
1
2
2
2

Subbit.
4
3
4
0

Lignite
1
2
0
0

Totals
6
7
6
2
21
       The results of emission tests on wet FGD systems are summarized in Table 6-17. The
next to last column in Table 6-17 shows the percent reduction in HgT across the wet FGD
scrubber as determined by the OH sampling train measurements. The last column is an
estimate of the reduction in Hgj across the PM control device and wet FGD scrubber. These
estimates were made using the class PM average coal-boiler-control EMF that is applicable to
each test unit (see Section 6.5.2).

       Class average emission test results for the PC-fired boilers with wet FGD units are
shown in Figures 6-17, 6-18, and 6-19. Each of these figures is based on capture estimates
across the PM control device and wet FGD scrubber combinations.  Figure 6-17 shows the
class average stack concentrations of Hgp, Hg2+, and Hg°. Figure 6-18 shows the average
inlet and outlet concentrations of Hgj and percent reduction for each class. Figure 6-19
shows the relative mercury speciation for PC-boilers with wet FGD scrubbers.
                                        6-42
-------
Errata  Page 6-43, dated  3-21-02
                                                  Table 6-17
                       Post-combustion  Controls: Wet FGD Scrubbers
Bituminous Coal, PC Boiler with PS and Wet FGD Scrubber
Bruce Mansfield        1      0.27     8.65     1.58     10.50    10.93    0.04
Bruce Mansfield        2      0.73     9.84     2.08     12.65    8.93     0.06
Bruce Mansfield        3      0.27     8.34     1.70     10.31    11.82    0.04
Subbituminous Coal, PC Boiler with PS and Wet FGD Scrubber
BoswelU             1      0.11     0.33     5.05     5.48    6.98     0.02
BoswelH             2      2.98     1.07     1.47     5.53    6.63     0.20
BosweIN             3      2.75     0.55     1.16     4.45    7.93     0.28
Cholla 2
Cholla 2
Cholla 2
Colstrip
Colstrip
Colstrip
 Lawrence
 Lawrence
 Lawrence
1      0.42
2      1.11
3      0.41
       1.78
       1.94
       1.63
       0.23
       0.53
       0.24
0.97
0.93
2.06
2.29
2.37
2.86
1.65
0.63
0.65
4.66
2.62
2.99
6.07
4.66
5.46
6.99     0.15
6.37     0.19
5.09     0.11
1.08     5.15     7.63    0.05
6.37     10.68    7.98    0.02
5.39     9.88     7.93    0.02
4.99     6.86    6.24    0.01
4.41     5.58    5.47    0.08
4.96     5.86    6.03    0.09
                                                        1.89
                                                        2.73
                                                        1.22
                                                        0.10
                                                        0.44
                                                        0.59
0.21
0.14
0.14
                         0.42
                         0.45
                         0.39
                         0.49
                         0.53
                         0.51
                                                 7.01
                                                 7.96
                                                 8.29
                                                 5.53
                                                 5.89
                                                 5.57
3.93
4.67
4.22
                         9.13
                        11.03
                         2.13
                         6.37
                         6.71
                         6.20
                                                 8.95     14.81    18.11
                                                 10.76    14.94   -20.57
                                                 9.55     7.42    19.25
                                                 5.65
                                                 6.53
                                                 6.43
4.29
5.01
4.46
                9.60
                11.51
                2.54
                6.87
                7.32
                6.81
                                                 -3.08
                                                -18.25
                                                -44.40
29.30
-7.51
18.29
                -86.54
                -7.74
                74.27
                -0.07
                -31.14
                -16.21
Minimum
Maximum
STDEV
       0.11     0.33
       2.98     2.86
       1.02     0.85
        1.08
        6.37
        1.82
        4.45
        10.68
        1.96
        5.09    0.01
        7.98    0.28
        0.98    0.09
 ND Lignite, PC Boiler with PS and Wet FGD Scrubber
 Lewis and Clark        1       1.15     16.47    11.65    29.27    15.33    0.06
 Lewis and Clark        2       1.68     13.64    8.43     23.75    15.54    0.00
 Lewis and Clark        3       1.41     6.28     10.20    17.89    18.96    0.00
                                                 19.00
                                                 1.41
                                                 18.91
38.59
21.38
12.27
                -25.89
                -44.19
                67.94
                -10.01
                -33.75
                -12.96
                0.10     2.13     2.54    -86.54   -44.19
                0.59     11.03   11.51    74.27   67.94
                0.17     2.34     2.40    39.47   31.96
                                                        0.50     13.86    14.42    50.75     5.98
                                                        0.35     14.19    14.55    38.74     6.41
                                                        0.50     15.81    16.31    8.81    13.94
                                                                         a
                                                                          CONTINUED
                                                   6-43a
-------
Errata Page 6-43b, dated  3-21-02
                                        Table6-17(cont'd)
                    Post-combustion Controls: Wet FGD Scrubbers
 Bituminous Coal, PC Boiler with CS-ESP and Wet FGD Scrubber
 AESCayuga         2    0.00    6.40    2.58     8.98     11.87    0.00
 AESCayuga         1    0.00    5.87    2.24     8.11     10.70    0.00
 AESCayuga         3    0.00    5.55    2.95     8.50     10.80    0.00
 Big Bend
 Big Bend
 Big Bend
1    0.09    4.86   2.40    7.34    17.52    0.05
2    0.05    4.92   2.31     7.29    11.25    0.00
3    0.02    4.26   2.13    6.41    12.01    0.03
                                                  0.18
                                                  0.36
                                                  0.18
                                     0.21
                                     0.12
                                     0.23
                                             2.70
                                             2.73
                                             3.08
                                                     2.18
                                                     1.75
                                                     2.05
                                                             2.88    67.91    76.06
                                                             3.09    61.88    71.56
                                                             3.26    61.63    71.38
                                        2.44
                                        1.87
                                        2.31
                               66.70    75.16
                               74.37    80.88
                               64.01    73.15
 Minimum
 Maximum
 STDEV
0.00    4.26    2.13    6.41
0.09    6.40    2.95    8.98
0.03    0.78    0.30    094
                     10.70    0.00
                     17.52    0.05
                      2.59     0.02
                                                  0.12
                                                  0.36
                                                  0.08
                                1.75     1.87
                                3.08     3.26
                                0.50     0.53
                               61.63
                               74.37
                               4.78
                               71.38
                               80.88
                               3.56
 Subbituminous Coal, PC Boiler with CS-ESP and Wet FGD Scrubber
 Jim Bridget          1     0.05    2.49    5.21    7.74  \o coal flov  0.06     0.25
 Jim Bridget          2     0.44    2.04    5.64    8.12  10 coal flov  0.05     0.29
 Jim Bridger          3     0.07    1.78    4.50    6.35  10 coal flov  0.03     0.20
 Laramie River 1
 Laramie River 1
 Laramie River 1
 Sam Seymour
 Sam Seymour
 Sam Seymour
1    0.25
2    0.04
3    0.02
3.14
2.16
3.08
              7.52
              8.35
              7.53
10.91    13.52
10.55    15.45
10.63    15.71
0.02
0.00
0.01
0.29
0.12
0.03
     0.03    3.00
     0.01     4.08
     0.01     5.39
                                                          6.63     6.95    10.32    14.60
                                                          6.51     6.85    15.64    19.67
                                                          5.92     6.15     3.06    7.69
4.86     5.18    52.57   54.83
5.73     5.85    44.54   47.18
4.48     4.52    57.53   59.56
                                                             1.51     1.51
                                                             23.90    23.90
                                                             31,99    31.99
 Minimum
 Maximum
 STDEV

 •Note the column title changes from coal to Wet FGD and PM+FGD
                                                                          1.51     1.51
                                                                         57.53    59.56
                                                                         21.09    20.83

                                                                        CONTINUED
                                             6-43b
-------
Errata Page 6-43c, dated 3-21-02
                            (Intentionally Blank)
                                  6-43c
-------
                                               Table 6-17 (cont'd)
                           Post-combustion Controls: Wet FGD Scrubbers
TX Lignite, PC Boiler with, CS-ESP and wetEGB Scrubbei)   _
                                       29.3.9. _.,  46.07   I  61.96   ;  0.31
                                       28.15     48.03   ',  63.13   .  pj.8
                                       27.21   ,  53.16   •  76.52   .  0.24
                     U. U    i   i3.BJ   !   ^/.^L   i  3J.Lb   <   'O.^   .   U.^4       /./ft   j
                                                37.68     20.84     0.33  . ;
                                                42.29   j  15.29     0.12
Ritiiminnuc Pnal PC Rnilrr with HS-F.SP ind wet FC.n Krruhhfr
               1J	2,64
               2J	1.55
              . 3 I   3.4J
Charles Lowman
Charles Lowman
Diaries Lowman
         YfiSffi,
                                        4.41    I  13.27... '   5.48      0.05   I   2.06
                                                                                                               38.03.
                                                                                                           _   59.20_
                                                                                                               7.66
                    PC Boiler with HS-F.SP and we! FGD Scrubber
                                        2,i9_  .  3.20   ,   4.4S
                                                2,71   j   4_.7§
                     0.03       l.0»       1.87      2.99   ]   3.86

                     D.04       0.33       3.61  __L  3.97   :   2.4
                                                2.85
                                        l.W   .  2.19      2.30
Rttinnlnoiis rnal PC Bailer with FF and Viet FCD urubber
Ctover   	
Clova-     	
         :aus..
InlemipmiUiil
                                                        6-44
-------
9
w
M
I
u
Sub, HS-ESP+wet FGD {20%]

 Bit, HS-ESP+wet FGD [42%]

 Lig, CS-ESP+wet FGD [44%]

Sub, CS-ESP+wet FGD [27%]

 Bit, CS-ESP+wet FGD [66%]

    Bit, FF+wet FGD [72%]

   Lig, PM+wet FGD [33%]

   Sub, PM+wet FGD [-8%]

   Bit, PM+wet FGD [I2%]
                                         IHg(p)Out BHg(2+)Out DHg(0)Out
                          0       5      10      15     20      25
                                Mercury Concentration, m>/dscm @ 3% O2


          Figure 6-17. Mercury speciation for PC boilers with wet FGD.
                                                                   30
Sub, HS-ESP+wet FGD [20%]
•^ Bit, HS-ESP+wet FGD [42%]
^i.
fi
o Lig, CS-ESP+wet FGD [44%]
-

"g Sub, CS-ESP+wet FGD [27%]
04

£f Bit, CS-ESP+wet FGD [66%]
£
3 Bit, FF+wet FGD [72%]
o
^5 Lig, PM+wet FGD [33%]
Q'

Sub, PM+wet FGD [-8%]

Rit PM+ux>t rnn ri'j'Wii

•'•C^'a.^
1

~°*V. V'V^-"''"^ ~ ''^^"'-'^X^.X^X'^'^ * '-:"'^-» '•- '•/-i'.]
)

,V.V.V-.V.-,H


;•-;•. ••-,! . n Inlet Q Outlet ;

b
\ \V\X\vVv\ 'V\^"V'.\ '-.^1
1

vN-Ov'Al
1

-•.xv :-..w->:.-,i
I
                    0
                          10    15    20    25   30    35    40   45
                           Mercury Concentration, uc/dscm @ 3% Oi
          Figure 6-18. Mercury emissions for PC boilers with wet FGD.
50
                                     6-45
-------
 B
 .2
 1
 M
 a
 Q
 U
 a.
Sub, HS-ESP+wel FGD [20%]

Bit, HS-ESP+wet FGD [42%]

Lig, CS-ESP+wet FGD [44%]

Sub, CS-ESP+wet FGD [27%]
Bit, CS-ESP+wet FGD [66%] |>J

    Bit, FF+wet FGD [72%]

   Lig, PM+wet FGD [33%]

   Sub, PM+wet FGD [-8%]

   Bit PM+wet FGD [ 12%]
                          ssi.
                                               DHg(p)Out HHg(2+)Out  DHg(0)Out
                   0     10    20    30    40    50    60    70    80

                               Relative Mercury Concentration in Stack, %
                                                                             90    100
              Figure 6-19. Relative mercury speciation for PC boilers with wet FGD.
       The best levels of HgT capture are exhibited by units burning bituminous coal and
equipped with a FF (98 percent), CS-ESP (15 percent), or HS-ESP (50 percent). The higher
capture levels for bituminous-fired boilers equipped with the CS-ESP, HS-ESP, or FF control
devices are consistent with the high levels of Hg° oxidization associated with these coal-boiler
control classes (see Figures 6-12 and 6-13).  The very high levels of Hg capture exhibited by
the bituminous-coal-fired boiler units with a FF and wet FGD system can be attributed to high
levels of Hg° oxidization and to the capture or conversion of Hgp and Hg2+ as flue gas passes
through the FF cake. Estimates of Hgj capture across the wet FGD and PM + wet FGD
combinations are shown in Table 6-18 for units burning bituminous coal.  Detailed data for
these units are given in Table 6-17. The best control is exhibited by wet FGD systems
equipped with a FF followed by units equipped with a CS-ESP and a HS-ESP.

       The HgT capture in one test unit burning bituminous coal and equipped  with a PM
scrubber + wet FGD system averaged 12 percent. Hg at the outlet of the scrubber was
predominantly Hg°.
                                         6-46
-------
                                      Table 6-18
                    Wet FGD Scrubbers Burning Bituminous Coal
Controls and Test Unit
FF + Wet FGD
Clover
Intermountain
Average
CS-ESP + Wet FGD
Big Bend
AES Cayuga
Average
HS-ESP + Wet FGD
Charles R. Lowman
Morrow
Average
Reduction in HgT, %
FGD
76
68
72

68
64
66

36
49
43
PM + FGD
98
97
98

76
73
75

44
55
50
       The estimated capture of HgT in wet FGD units burning subbituminous coals is given
in Table 6-19. The four PS units were Boswell 4, Cholla 2, Colstrip, and Lawrence. The inlet
and outlet HgT data appeared reasonable except for runs 1 and 2 on Colstrip. All tests on
Lawrence and Boswell 4 had HgT outlet concentrations higher than the corresponding HgT
inlet concentrations. Cholla 2, which had HgT emission reductions ranging from -8 to 29
percent, appeared to exhibit hysterisis effects. One unit, Lewis and Clark, burned a ND
lignite. This unit also appeared to exhibit hysterisis effects, with successive HgT reductions
for the three tests of 51, 39, and 9 percent. The declining reductions in HgT capture were
mirrored by inlet reductions of HgT and Hg2+.

       The erratic nature and differences in capture for the CS-ESP units are probably due to
differences in the subbituminous coals being burned and the differences in the scrubber
operating conditions. Except for the Coronado tests, the test results on HS units were fairly
consistent. It is not known whether the sampling and analysis results from the Coronado unit
are incorrect or whether differences in the coal and operating conditions caused the lower HgT
capture results.
                                        6-47
-------
Errata Page 6-48, dated 3-21-02
                                  Table 6-19
             Wet FGD Scrubbers Burning Subbituminous Coal
Controls
and Test Unit
PS + Wet FGD
Boswell 4
Cholla 2
Colstrip
Lawrence
Average
HS-ESP + Wet FGD
Coronado
Craig 1
Navajo
San Juan
Average
CS-ESP + WetFGD
Jim Bridger
Laramie 1
Sam Seymour
Average
Reduction
in HgT, %
FGD
NA
NA
NA
NA


1
23
21
37
20

10
52
19
27
PM + FGD
-22
13
-7
-16
-8

11
31
29
44
29

14
54
19
29
      Two units, burning TX lignite and equipped with a CS-ESP, exhibited average HgT
captures of 46 percent (see Table 6-20). The SPF2+ for limestone was 0.65 and the SPF2* for
Monticello 3 was 0.42, indicating moderately high relative concentrations of Hg2+ at the
scrubber inlets of these two units. TX lignites appear to have a higher oxidization and capture
potential than ND lignites.
                                     6-48
-------
Errata Page 6-49, dated 3-21-02
                                     Table 6-20
                       Wet FGD Scrubbers Burning TX Lignite
Controls and Test Unit
CS-ESP + Wet FGD
Limestone
Monticello 3
Average
Reduction in HgT, %
FGD
51
36
44
PM + FGD*
51
36
44
*Estimated
6.6.6 Potential Effects of Post-combustion NO* Controls

       Post-combustion NO\ controls convert NO\ in the boiler flue gases to molecular
nitrogen and water using a catalytic process (selective catalytic reduction) or a noncatalytic
process (selective noncatalytic reduction). For both processes, a reducing agent (usually
ammonia) is injected into the boiler flue gas at a point upstream of any post-combustion PM
or SO2 control device.  A limited amount of data is available in the ICR Hg emission database
regarding the potential effects of these post-combustion NOx controls on Hg capture. These
data are presented in Table 6-21.  Test results for pulverized-coal boilers burning bituminous
coal with either SNCR or SCR systems are compared to the results of tests on similarly
controlled units that do not use post-combustion NOx controls.

                                     Table 6-21
 Potential Effects of Post-combustion NOx Control Technologies on Mercury Capture in
                      PC-fired Boilers Burning Bituminous Coal
Post-combustion
Controls
CS-ESP
SDA + FF
Post-
combustion
NOX Control
none
SNCR
none
SCR
Number of
Pulverized-
coal-fired
Boiler Units
Tested
6
1
2
1
Average Mercury
Capture by Control
Configuration
36%
91%
98%
98%
       Tests on the single pulverized-coal boiler unit using a CS-ESP with SNCR shows an
average Hg capture that is significantly higher than the six units tested with a CS-ESP using
no post-combustion NOx controls (91 percent with SNCR versus 36 percent without SNCR).
It was reported that the fly ash from the boiler unit using SNCR contained unusually high
levels of carbon.  Because data are available only for this one test, it is not known whether
                                       6-49
-------
the high levels of Hg capture indicated by the test results are attributable to the high fly ash
carbon content, (he use of an SNCR system, a combination of both, or some other site-
specific  factor.

      A comparison of tests for pulverized-coal boiler units using an SDA with an FF
shows no discernable difference in Hg capture with or without the use of an SCR for post-
combustion NOx control. An average Hg capture of 98 percent was measured by the tests on
the one unit equipped with an SCR compared to 98 percent Hg capture for the two similar
units without SCR systems. Because of the very high levels of Hg capture by all of the tested
control configurations, it is not possible to determine the effect of SCR on Hg capture.

      Recent tests on a pilot-scale, pulverized-coal combustor, which was equipped with an
SCR and a CS-ESP, showed increased Hg capture when bituminous coals were burned but
not when a subbituminous coal was burned.  Mercury emission reductions were observed
when the SCR system was operated normally with the injection of ammonia upstream of the
SCR catalyst. Improvement of Hg capture was also noted when ammonia was injected, but
the SCR catalyst was bypassed.  These tests provide evidence that SNCR and SCR systems
may enhance Hg capture under some conditions.

6.7     COMBUSTION SYSTEM EFFECTS

      LNBs and combustion modification techniques are believed to increase the unburned
carbon in fly ash and increase the adsorption of Hg onto collectable fly ash.  Since neither the
fly ash carbon content nor the LOI was measured during the ICR field test, it is not possible
to evaluate Hg capture performance benefits that accrue from the use of NOx control
combustion modification techniques.  The ICR field test program included tests on six
different unit classes using cyclone-fired boilers and six unit classes with FBCs. The results
of ICR tests on units with cyclone-fired boilers and FBCs are shown in Tables 6-22 and 6-23,
respectively.
                                       6-50
-------
Errata Page  6-51,  dated 3-21-02
                                                    Table  6-22
                                          Cyclone-fired Boilers
    Lignite, Cyclone Boiler with CS-ESP
LelandOlds       1     0.56       0.23      3.30      4.09      5.63
LelandOlds       2     0.26       0.46      8.80      9.51      10.18
LelandOlds       3     2.85       0,81      4.77      8.43      7.94
       Average
Subbitumlnous/Pet. Coke, Cyclone Boiler with HS-ESP
Nelson Oewey     1     0.01       0.49      3.20      3.69      6.62
Nelson Dewey     2     0.01       0.24      2.19      2.43      6.47
Nelson Dewey     3     0.01       0.12      2.06      2.18      6.09
       Average
Lignite, Cyclone Boiler with Mechanical Collector
                     0.76
                     1.08
                     0.09
0.78
0.67
0.77
2.17
1.94
1.74
3.70
3.69
2.60
3.S8
3.01
3.36
                                       13.68
                                       13.90
                                       14.91
                   1S.99
                   18.06
                   19.66
                   10.51
                   18.55
                   11.39
Bay Front
Bay Front
Bay Front
       Average
Lignite, Cyclone Boiler with SDA/FF
Coyote           1     0.69       1.62
Coyote           2     1.18       2.98
Coyote           3     1.69       3.07
       Average
Bituminous, Cyclone Boiler with PS and Wet FGD Scrubbers
Lacygne          1     6.70       3.99      1.30      12.00   no inlet flow
Lacygne          2     6.52       3.34      0.60      10.46   no inlet flow
Lacygne          3     5.98       °J2L.^J-£jL~---^Ji!^
       Average     l|j|^||jill|jjljlpB^
Bituminous, Cyclone Boiler with CS-ESP and wet FGD Scrubber
Ballly            1     0.04       3.18      2.57       5.79      4.41
Baifly            2     0.04       2.37      2.9S       5.36      5.20
Baily            3     0.09       3.01      2.58       5.68      4.08
                                      0.00
                                      0.00
                                      0.00
                                      0.10
                                      0.04
                                      0.04
1.19
0.86
0.48
                   0.06
                   0.14
                   0.06
                                                                    0.04
                                                                    0.05
                                                                    0.09
                                      0.82
                                      1.09
                                      1.60
                                      0.26
                                      0.16
                                      0.25
0.60
2.75
3.57
                  0.04
                  0.24
                  0.44
                                               0.44
                                               0.43
                                               0.41
                                     4.04
                                     5.26
                                      LS
                                     3.33
                                     2.40
                                     2.44
1.91
1.80
1.78
                  13.97
                   LS
                  18.06
                                               8.74
                                               7.41
                                               5.10
                                                                    0.00
                                                                    0.00
                                                                    0.00
                                               0.36
                                               0.31
                                               0.39
                                               2.65
                                               2.62
                                               2.76
                                     4.86
                                     6.35
                                      NA
                                     3.69
                                     2.60
                                     2.73
3.69
5.40
5.64
                  14.10
                   NA
                  18.58
                                               9.22
                                               7.B9
                                               5.59
                                               3.22
                                               2.93
                                               3.17
                                                                                                         -18.68
                                                                                                         33.26
                                                                                                          NA
                                               13.66
                                               37.64
                                               NA
                                                                                                         0.13      44.27
                                                                                                         -6.90      59.63
                                                                                                         -24.95      55.22
 0.34       -2.95
-46.54     -79.21
-125.00    -73.79
                  11.81
                   NA
                  5.48
                                              54.24
                                              54.95
                                              54.11
                  -34.23
                    NA
                  -63.12
                                              23.18   no inlet flow
                                              24.53   no inlet flow
                                              22.17   no Met flow
                                               27.09
                                               43.53
                                               22.31
                                                       6-51
-------
Errata Page 6-52, dated 3-21-02
                                                Table 6-23
                                      Fluidized Bed Combustors
 Lignite, FBC with CS-ESP
 R.M. Heskatl     1     4.73     5.39      3.83     13.95    13.54     1.06      1.44
 R.M. Heskell     2     2.93     0.96      2.61     6.50     12.68     0.07      0.41
 R.M. Heskelt     3     7,43     0.4_4_      3._08_ ^  10.94    11.11     0.05      0.18

 Anthracite Waste. FBC with FF
 Kline Township    1    44.54     0.12      045     45.11    148.68    0.00      0.06
 Kline Township    2    43.12     0.06      040     43.58    212.95    0.00      0.06
 Kline Township    3    44.97     0.06      0.34     45.37    153.77    0.00      0.06
       Average
 Bituminous Waste, FBC with FF
 Scrubgrass       1    184.04     0.68      0.19
 Scrubgrass       2    124.11     0.42      0.09
 Scrubgrass       3
       Average
                                                   4.57      7.07     49.29    47.76
                                                   5.31      5.78     11.09    54.40
                                                   4.74      4.98     54.49    55.19
                                                               mm
                                                   0.06      0.12     99.74
                                                   0.06      0.12     99.73
                                                   0.06      0.12     99.74
0.22	    0.07
                                                   o.oe
                                                   0.07
                                  0.15
                                  0.12
                                           0.04      0.07      0.11
                                  99.92
                                  99.91
                                  99.85
                                                   99.92
                                                   99.95
                                  99.85
                                  99.89
                                  99.89
 Bituminous/Pet. Coke, FBC with SNCR and FF
 Stockton Cogen   1     2.71      0.06      0.06
 Stockton Cogen   2     1.56      0.07      0.06
 Stockton Cogen   3     2.08      0.06      0.06
       Average
 Subbitumlnous, FBC with SCR and FF
 AES Hawaii      1     0.26      0.04      1.29
 AES Hawaii      2     0.35      0.17      1.38
 AiES Hawaii      3     0.36      0.11      1.18
       Average
 Lignite, FBC with CS-FF
 TNP            1    21.65     8.68      7.42     37.74
 TNP            2    10.65     4.51      609     21.25
 TNP            3    28.12     13.78     7.04     48.94
       Average
                 283
                 1.69
                 2.20
                 1.59
                 1.90
                 1.64
1.68
144
1.66
3.77
3.72
2.51
                         63.81
                         44.22
                         95.04
0.02
0.03
0.03
0.00
0.00
0.00
        0.04
        0.03
        0.04
0.04
0.05
0.05
0.02
0.02
0.02
        12.13
        6.78
        13.54
0.05
0.05
0.05
0.68
0.90
0.55
        4.74
        2.94
        5.07
0.11
0.13
0.12
96.09     93.39
92.16     9080
94.48     9267
0.70     55.84     81.39
0.92     51.35     75.16
0.58     64.91     77.06
        16.91     55.20    73.50
        9.76     54.07    77.93
        18.66     61.88    80.37
               '-Mi
6.7.1 Cyclone-fired Boilers
        Mercury capture and stack gas speciation for cyclone-fired boilers are shown in
Figures 6-20 and 6-21. The percentage of total Hg capture in these units appears to be similar
to the Hg captured in pulverized-coal-fired units burning similar fuels and equipped with
comparable air pollution control devices (see Table 6-24). Except for the unit equipped with
a mechanized collector, the Hg in flue gas  consisted primarily of Hg°.
                                                 6-52
-------
      18
      16 T-
   I.of-
   I   I
   I 8-	


   I .1-
                            I Hg(p) Out   B Hg(2+) Out  D Hg(0) Out
u 	
£-4- 	


a
* L J
S 2 f- 	

^






         Bil/CS-ESP/Wet- Lignite^CS-ESP  Lig/MC(-57)    Bit/PS (23)    Sub-Pet/HS-ESP Lignite/SDA/FF
           FGD(S4)        (7)                                (-I1)         (9)

                             Coal/APCD (Hg Reduction, %)


             Figure 6-20. Mercury speciation for cyclone-fired boilers.
   100   T
S
x
e
 e
U
 >,
w

S
01
           Bit/CS-    Lignite/CS-ESP  Ug'MC (-57)   Bil/PS (23)   Sub-Pet/HS-ESP Lignite/SDA/FF
         ESP/Wct-FOD      (7)                               (-]])         (9)

            (54)            Coal/APCD (Hg Reduction, %)


        Figure 6-21. Relative mercury speciation for cyclone-fired boilers.
                                          6-53
-------
                                    Table 6-24
                           Comparison of Class Average
                  HgT Reductions for PC- and Cyclone-fired Boilers
Unit Class
Lignite, CS-ESP
Subbituminous/Pet Coke, HS-ESP
Lignite, Multicyclone
ND Lignite, SDA/FF
Bituminous, PM scrubber + wet FGD
Bituminous, CS-ESP + wet FGD
Reduction in HgT, %
Cyclone
9
0
0
7
23
54
PC-Fired
36
7
NA
2
12
81
6.7.2  Fluidized-bed Combustors

       Six fluidized-bed combustors were tested on the ICR program. Test results for the
fluidized-bed units are shown in Figures 6-22 and 6-23. All of the units injected limestone
into the FBC to control SOa emissions.  One unit was equipped with a CS-ESP while the
remaining five units were equipped with a FF. One of the FF units was also equipped with an
SNCR system. The unit equipped with the CS-ESP burned lignite. The capture of HgT for
this unit averaged 38 percent. The reduction in Hgj for units equipped with FF systems
depended primarily on the type of fuel that was burned. The one unit that burned
subbituminous coal was equipped with an SCR system and a FF. Inlet and outlet HgT
concentrations for the two valid runs on this unit were 1.7 and 0.7 p.g/dscm, respectively,
resulting in a 57 percent capture efficiency.  One unit that burned waste anthracite had an
average Hgi reduction efficiency of 99.7 percent, while another unit burning bituminous coal
and petroleum coke had an average reduction of 94 percent.

       The best performance for any unit tested during the Part III ICR program exhibited
average Hgj inlet concentrations  of 185 ug/dscm, outlet concentrations of 0.15 p.g/dscm, and
an average HgT reduction of 99.9 percent.
                                       6-54
-------
Mercury Concentration, |ig/dscm @ 3% O2
> K* £b &>. OO O K
^
—
n
i
—
-
• Hg(p)0ut !
S Hg(2-t-) Out1
1
DHg(0)Out
       Ligniie/CS-ESP  Ant(waste)/FF  Bit-Pel/FF (94)   Bil (waste)/FF Sub/SCR/FF (57)   Lig/FF(57)
           (38)         (99)                     (99)
                           Coal/APCD (Hg Reduction, %)
                   Figure 6-22. Mercury speciation for FBCs.
   90
1
55
.£
1
•c
70
   60
   so
   30
   I0
                                                                       j iOHg(0)Out '
                                                                             2+) Out
       Lignitc/CS-ESP  fM (waste)/FF  Bit-Pct/FF (94) Bit (waste)/FF  Sub/SCR/FF
           (38)        (99)                    (99)         (57)
                          Coal/APCD (Hg Reduction, %)
                                                              Lig/FF (57)
              Figure 6-23. Relative mercury speciation for FBCs.
                                         6-55
-------
6.7.3  IGCC Facilities

        Table 6-25 summarizes the emission source test data and coal analysis data for the
Tampa Electric Company Polk Power Project and Wabash River Coal Gasification
Repowering Project.  A more detailed presentation of the test data is included in Appendix C
of this report.  Coal data were used to calculate inlet feed rates of total Hg to the coal-
gasification units. The total Hg in the exhaust gas from the gas turbine was determined by
summing the three Hg species obtained using the OH Method during each test run (i.e., Hgp,
Hg2+,andHg°).

                                           Table 6-25
       Calculated Mercury Removal in IGCC Power Plants Using Bituminous Coa!
IGCC
Facility
Tampa
Electric
Company
Polk Power
Project
Wabash River
Coal
Gasification
Repowering
Project
Test
Run
Run 1
Run 2
Run 3
3 -Run
Average
Run 1
Run 2
Run 3
3 -Run
Average
Coal Fed to Gasifer
Coal
Flow
Rate
(kg/hr,
dry)
91,454
88,707
71,373
83,845
90,663
89,629
89,493
89,928
Total Hg
Content
(ppm,
dry)
NDa
ND"
NDa
	
0.064
0.068
0.070
0.067
Total Hg
Feed
Rate
(kg/hr)
0.0091 b
0.0089 b
0.007 lb
0.0084
0.0058
0.0061
0.0063
0.0061
Gas Turbine Exhaust Gas Stream
Gas Stream
Flow Rate
(dscm/hr)
1,430,191
1,453,617
1,414,052
1,432,620
1,372,064
1,385,884
1,352,458
1,370.135
Total Hg
Content'
(Hg/dscm)
3.94d
3.86d
3.68"
3.83d
2.57*
2.60'
2.76 =
2.64 "
Total Hg
Emission
Rate
-------
stream to begin with, the elemental Hg released during the coal gasification process has very
few opportunities to be adsorbed on solid particles to form particle-bound Hg.

       The OH Method test results show that elemental Hg is the predominant species in the
gas turbine exhaust gas.  For the Polk IGCC facility, the measured distribution of gaseous Hg
species was approximately 90 percent elemental Hg and 10 percent Hg2+.  For the Wabash
River IGCC facility, no Hg2+ was detected by the OH Method (i.e., 100 percent of the HgT in
the exhaust gas stream was in the form of Hg°). One possible explanation for these results is
the different gas cleaning processes used at the two IGCC facilities.  The syngas from the
coal gasifier at the Wabash River IGCC facility is cleaned and conditioned using a system
that includes a water scrubber for gas cooling and an amine scrubber for removal of reduced-
sulfur species. Oxidized Hg is water-soluble and is readily absorbed by a wet scrubbing
system. However, Hg° is insoluble and passes through a wet scrubbing system. Thus, it is
reasonable to expect that the water and amine scrubbers used at the Wabash River IGCC
facility effectively remove the oxidized Hg in the syngas before it is burned in the gas
turbine.

       The Polk  IGCC facility uses a hot gas-cleaning system. There is no wet scrubbing
process to remove any Hg2+ from the syngas before it is burned. The syngas is not cooled and
remains at elevated temperatures until it is fed to the gas turbine. It cannot be determined
from the test data how the elevated syngas temperatures and combustion process in the gas
turbine combustors affect Hg speciation. However, it is believed that any Hg2+ in the syngas
will be converted back to Hg° when the syngas is burned. The degree of oxidization will
probably be limited by the combustion gas composition and the rate at which it is cooled
before it is emitted to the atmosphere.

       The last column in Table 6-25 provides an estimate of the overall amount of Hg in the
coal removed by  the IGCC process. Based on these two tests, approximately one-third of the
Hg in the coal is removed. The Hg that remains in the combustion gas is primarily Hg°.
 6.8  NATIONAL AND REGIONAL EMISSION ESTIMATES

       Estimates of the nationwide Hg emissions provide an indication of the overall level of
Hg capture being achieved by existing control systems used by coal-fired utility boilers in the
United States.  A number of different approaches can be used for these estimates. The EPA
evaluated four different methods for estimating nationwide Hg emissions using information
from the ICR database.  The method selected as being the best is outlined below:

       •  ICR Part II coal data were used to determine the average Hg content and the
          amount of coal burned in each of 1143 units supplying data for 1999.

       •  Mercury in the flue gas from coal burned in each boiler unit in 1999 was
          calculated assuming that all of the Hg in the coal was in the flue gas leaving the
          furnace.

                                        6-57
-------
       •  Each unit was assigned a coal-boiler-control class that best met the characteristics
          of the unit.

       *  Total Hg in the boiler flue gas for each unit was multiplied by the class emission
          factors for speciated and total Hg that had been assigned to the unit.

       •  Total and speciated Hg emissions for each unit were added to provide estimates of
          national Hg emissions from coal-fired utility boilers in 1999.

       Computer runs using this procedure resulted in estimated national Hg emissions in
1999 of 43.5 tons.

       Using the EPA's ICR database, EPRI independently estimated the nationwide Hg
emissions from existing coal-fired utility boilers in the United States to be in the range of 45
to 48 tons in 1999. EPRI selected a different estimation methodology than the one used by
EPA.  EPRPs method is based on a model that correlates the level of Hg emissions with the
amount of chlorine in coal and the ratio of chlorine to sulfur in the coal for the case of units
with cold-side  ESPs.  Both the EPA and EPRI estimate that approximately 75 tons of Hg was
in coals burned in 1999.

       After EPA announced its decision to develop the NESHAP,  the transfer of data from
the field test reports to the emission databases was rechecked for errors. It was found that
several test units had been assigned to the wrong coal-boiler-control classes. Also, the results
of a number of tests failed data quality requirements and were removed from the analysis set.
Subsequent computer evaluations resulted in the following estimates:

       •  48 tons of Hg was emitted to the atmosphere from coal-fired utility boilers in
          1999, and

       •  27 tons of Hg was captured by existing flue gas cleaning devices.

       Nationwide, approximately 25 tons (52 percent) of Hg was emitted from the
combustion of bituminous coal, followed by 17 tons (36 percent)  from the combustion of
subbituminous coals, and 4 tons (8 percent)  from the combustion  of lignite. The total
amounts of Hg emitted compared to the tonnage and types of coal burned in 1999 are
presented in Table 6-26.
                                        6-58
-------
                                     Table 6-26
                    Nationwide Coal Burned and Mercury Emitted
                 From Electric Utility Coal-fired Power Plants in 1999
Coal Type
Bituminous
Subbituminous
Lignite
Other
Total
Nationwide
Total Coal
Tonnage
Burned In 1999
(dry tons) ("
427,572,000
279,227,000
50,932,000
10,756,000
768,487,000
Percent of
Total Coal
Burned
56
36
7
1
100
Nationwide
Total Mercury
Emitted in 1999
(tons)
25
17
4
2
48
Percent of
Total Mercury
Emitted
52
36
8
4
100

  (a) For wet tons (as received), nationwide total is 928,398,000 tons in 1999.
     Percentages for wet tons are 50% bituminous, 41% Subbituminous, and 8% lignite.

6.9  SUMMARY AND CONCLUSIONS

       Previous research has shown that the capture of Hg by flue gas cleaning devices is
dependent on Hg speciation.  Both Hg° and Hg2+ are in a vapor phase at flue gas cleaning
temperatures.  Hg° is insoluble in water and cannot be captured in wet scrubbers. The
predominant Hg2+compounds in coal flue gas are weakly-to-strongly soluble and can be
generally captured in wet FGD scrubbers.  Both Hg° and Hg2+can be adsorbed onto porous
solids such as fly ash, PAC, or calcium-based acid gas sorbents for subsequent collection in a
PM control device. Hg2+ is generally easier to capture by adsorption than Hg°. Hgp is attached
to solids that can be readily captured in ESPs and FFs.

       The evaluation of ICR data provides valuable insights into relationships between the
speciation and capture of Hg, the type of coal burned, the types of boilers used, and the types
of post-combustion technologies used for flue gas cleaning.  The evaluation of ICR data
indicates that the behavior of Hg in conventional PC-fired utility boilers is primarily
dependent on the type of coal burned and the control technologies used at each site. This
behavior is consistent with the ensuing interpretations.

                                        6-59
-------
Bituminous Coals

       The Hg° in flue gas from the combustion of bituminous coal is readily oxidized and
converted to Hgp or Hg2*. The best technologies for controlling corresponding Hg emissions
are dry or wet FGD scrubbers along with post-combustion PM controls.  Dry scrubbing
systems that use a SDA/FF are superior in performance to those that use a SDA/ESP. In
SDA/FF systems, Hg can be absorbed on PM in the SDA, and particulate- and gas-phase Hg
can be captured as it passes through the FF and its associated filter cake. SDA/ESP systems
depend on the in-fight capture of Hg.

       A PM control device always precedes wet FGD scrubbers. Four types of PM control
devices are commonly used: FFs, CS-ESPs, HS-ESPs, and PM scrubbers. Units equipped
with a FF exhibit the best capture followed by units equipped with a CS-ESP, HS-ESP, and
PM scrubbers.  Units that are equipped with FF + wet FGDs can capture Hg in FF and can
convert Hg° to Hg2+ for subsequent capture in the scrubber.  Hg capture in CS-ESP + wet FGD
systems depends on the degree of Hg capture and oxidization as the flue gas passes through
the CS-ESP. Hg capture in units equipped with HS-ESPs is generally lower than the capture
in CS-ESPs because HS-ESPs operate at temperatures where the oxidization and capture of
Hg is  limited. The single test unit equipped with a PS + wet FGD system exhibited an average
HgT capture of 12 percent.

Subbituminous Coals

       Some subbituminous coals exhibit little, if any, Hg° oxidization in PC-fired boilers.
Others display moderate amounts of Hg° oxidization.  The use of low NOx burners tends to
increase the amount of unburned carbon and the potential for capturing gas-phase Hg.  The
ICR data show that the oxidization of Hg° can occur from gas-phase reactions or gas/solid
reactions with fly ash or surface deposits in power plants. The unburned carbon in fly ash can
oxidize Hg° or adsorb gas-phase Hg. Hg2+is believed to be more readily captured by
adsorption than Hg°. Because of temperature considerations, the adsorption of Hg onto fly ash
in units equipped with CS-ESPs is believed to occur as the flue gas flows through the air
preheater and the ducting that leads to the ESP. Additional adsorption can also occur within
the ESP.

       Flue gas from the combustion of bituminous coal contains moderate to high levels of
Hg2*,  primarily in the form of HgCl2.

    The EPA ICR database provides a massive amount of information that can be mined for
additional information. However, its usefulness is limited by the uncertainty of some of the
measurements and by information that the data set does not contain.  Some of the uses and
limitations of the ICR data are summarized below. The data provide:

•      Reasonable estimates of National and Regional emissions for Hgp, Hg2+, Hg°, and
       HgT. They cannot be used to predict the total and speciated Hg emissions of
       individual plants.
                                       6-60
-------
       Information against which hypotheses and models of the speciation and capture of Hg
       in coal-fired boilers can be tested.  It cannot be used to identify or confirm specific
       mechanisms that control the speciation and capture of Hg.

       Information needed to guide the development of control technologies and identify
       effective strategies for the control of Hg emissions.
Cautions:
       Mercury speciation measurements made with the OH Method upstream of the PM
       control devices should be used with caution. PM collected on the sampling train filter
       can result in physical and chemical transformations with the sampling train - with the
       result that OH Method speciation results do not accurately characterize the different
       forms of Hg in the flue gas where the samples were collected. The OH Method
       samples for Hgj accurately reflect the concentration of Hgr in the flue gas where the
       sample was collected.  Also the samples collected at the inlet to air pollution control
       devices may not accurately represent the average Hg concentration because of flow
       stratifications near the sampling location.

       At low inlet and outlet concentrations, the precision of the OH Method can obscure
       real differences between these concentrations.  When the capture across the control
       devices is being evaluated, the underlying imprecision of the measurements can show
       dramatic positive or negative reductions in emissions.

       It is believed that the positive variations in flue gas temperature can result in de-
       sorption of Hgp collected within PM control devices, resulting in flue gas
       concentrations of Hg that are higher at the outlet than at the inlet.  Reentramment of
       Hgp during rapping cycles of an ESP can also result in outlet concentrations that are
       higher than the inlet.

       There is uncertainty in the central values and statistical characteristics of the OH
       measurements.  The samples represent a short snapshot in time, and the effects of
       long-term variations in coal properties and plant operating conditions are unknown.

       The adsorption of Hg onto fly ash is highly dependent on fly ash properties and
       particularly on the fly ash carbon content.  The lack of information on coal and fly
       ash properties limits the ability to evaluate the effects of LNBs on the capture of Hg.
                                        6-61
-------
6.10 REFERENCES

    1.  U.S. Environmental Protection Agency.  Database of information collected in the
      Electric Utility Steam Generating Unit Mercury Emissions Information Collection
      Effort. OMB Control No. 2060-0396. Office of Air Quality Planning and Standards.
      Research Triangle Park, NC. Available at:
      < http://www.epa.gov/ttn/atw/combust/utiltox/utoxpg.html >.

    2.  Hargis, R., W. O'Dowd, and H. Pennline. "Sorbent Injection for Mercury Removal in
      a Pilot-Scale Coal Combustion Unit." Presented at the 93rd Annual Meeting of the
      Air & Waste Management Association, Salt Lake City, UT.  June 18-22,2000.

    3.  Haythomthwaite, S., T. Hunt, M. Fox, J. Smith, G. Anderson, and C. Graver.
      "Investigation and Demonstration of Dry Carbon-Based Sorbent Injection for
      Mercury Control," Quarterly Report under DOE Contract No. DE-AC-22-
      96PC95256, December  1998.
                                      6-62
-------
                                      Chapter 7
                       Research and Development Status of
                Potential Retrofit Mercury Control Technologies
7.1 Introduction

       The Part III EPA ICR data show that ESP and FF control devices currently used to meet
PM emission standards do capture particle-bound mercury (Hgp) from coal-fired electric utility
boilers (see Chapter 6). The data also suggest that SDA and wet FGD scrubbers in place to meet
SC»2 emission standards do capture oxidized mercury (Hg2+). However, these data also show that
the air pollution control devices presently used at most electric utility power plants are not very
effective in capturing elemental mercury (Hg°). Consequently, to achieve further reductions in
Hg emissions from existing coal-fired electric utility power plants, additional Hg reduction
strategies must be implemented.

       Potential Hg control strategies may be technology based (e.g., adding Hg emissions
control devices), economics based (e.g., Hg emissions trading programs), or national energy
policy based (e.g., switching from coal to alternative energy sources for electrical power
production). This chapter discusses technology-based control strategies available for reducing
Hg emissions from existing coal-fired electric utility power plants (Section 7.2). Current
research and development is focused on retrofitting Hg control technologies to the coal-fired
electric utility power plant's existing air pollution control systems (Section 7.3). This retrofit
approach offers the potential for reduced costs of implementing Hg controls by enhancing the
capability of the air pollution control equipment already in place to capture more Hg.

       Building on the results of laboratory- and bench-scale research studies (discussed in
Chapter 5), additional studies have been, and currently are being, conducted using pilot-scale test
facilities to further investigate the more promising retrofit Hg control technologies (Section 7.4).
For the many existing coal-fired electric utility boilers that are equipped with only ESPs or FFs,
retrofit technologies under development are based on injecting sorbents into the flue gas
upstream of the control device (Section 7.5). Retrofit technologies  to improve wet FGD scrubber
performance in capturing Hg are based on promoting oxidization of Hg° to soluble species by the
addition of oxidizing agents or the installation of fixed oxidizing catalysts upstream of the
scrubber (Section 7.6). The high levels of Hg control already achieved by the few existing
boilers using SDA for control of PM and SC«2 may be further enhanced by coinjection of a
second sorbent (Section 7.7). From a long-term perspective, the most cost-effective Hg controls
may be those implemented under a multipollutant emission control  strategy. New
                                         7-1
-------
multipollutant control technologies, which potentially are effective in controlling Hg emissions,
are under development (Section 7.8).
7.2 Technology-based Mercury Control Strategies for Existing Coal-fired Electric Utility
     Boilers

7.2.1 Remove Mercury Prior to Burning by Coal Cleaning

       Reducing the amount of Hg in the coal burned in electric utility boilers would reduce the
level of Hg emissions from these boilers without the need for additional post-combustion Hg
controls.  Switching coal suppliers to obtain coals with lower Hg contents raises complex
economic and national energy policy issues that are beyond the scope of this report.

       Physical cleaning of coal (discussed in Chapter 2) has traditionally been used at coal
preparation plants to remove mineral matter (i.e., a source of coal combustion ash) and mineral-
bound sulfur (pyrite) from the mined coal. Mercury and other trace metals are also removed by
this cleaning depending on whether these metals are associated with the organic carbon structure
of coal or coal mineral inclusions.  However, the existing commercially available coal-cleaning
methods remove only a portion of the Hg associated with the non-combustible mineral matter in
the coal and none of the Hg associated with the organic carbon structure of the coal.
Consequently, conventional physical coal cleaning can remove only a limited portion of the Hg
content of specific coals and may not be applicable to all coals.

       There is the potential for additional Hg reductions in the coal from several advanced
physical coal-cleaning processes using selective agglomeration or column froth flotation now
being developed. For example, Microcel™ is a type of column froth flotation available through
ICF  Kaiser and Control International. The company is the exclusive licensee for use of the
technology for coal deposits east of the Mississippi River and has sold units for commercial
operation in Virginia, West Virginia, and Kentucky. Ken-Flote™ is another type of column
froth flotation cell coal-cleaning technology that is commercially available. Results of bench-
scale studies indicate that the combination of conventional with advanced coal-cleaning
techniques removes from 40 to 82 percent of the Hg contained in samples of raw coal.l>2

       Advanced coal-cleaning processes using naturally occurring microbes and mild chemical
treatments to  reduce the Hg content in coal have been investigated under DOE-funded bench-
scale studies.  The results of these tests indicate  that these chemical  and biological coal-cleaning
processes have the potential for further reduction in the Hg content of coals.  However, DOE
viewed the processes as potentially high-cost control technologies, and DOE currently is not
funding development of these types of coal-cleaning technologies.3

       From  a near-term perspective, some reduction of the Hg content in certain coals burning
at existing coal-fired electric utility power plants can be achieved by physical coal-cleaning
processes. However, there are no easily identifiable coal deposits or coal types that will reliably
benefit from cleaning, with respect to reducing Hg content. In addition, even with
                                         7-2
-------
implementation of widespread coal cleaning for Hg emissions control, significant quantities of
Hg will remain in the coal after cleaning; this will require that other control techniques be used
to achieve additional reductions in Hg emissions.

7.2.2 Retrofit Mercury Controls to Existing Air Pollution Control Systems

       In addition to reducing the amount of Hg in the coal before it is burned in a coal-fired
electric utility boiler, a second technology-based alternative is to remove more of the Hg in the
boiler flue gas before it is vented out the stack. One strategy is to retrofit or adapt  control
technologies to the facility's existing air pollution control systems to increase the amount of Hg
captured by these systems rather than install new, separate Hg control devices. The strategy
offers the potential advantage of reducing the costs of implementing the Hg controls by
enhancing the capability of the air pollution control equipment already in place to  capture more
Hg.

       The existing air pollution controls used for a given coal-fired electric utility boiler
depends on site-specific factors including the properties of the coal burned, age and size of the
boilers, the geographic location of the  facility, individual state regulatory requirements, and
preferences of the facility owner or operator. For approximately 70 percent of the existing coal-
fired electric utility boilers in the United States, the control device used is an ESP (see Table 3-6
in Chapter 3).  These power plants typically burn low-sulfur coals to control  SO2 emissions and
use combustion modifications for NO\ emissions control. Most boilers use a "cold-side" ESP
where the control device is installed downstream  of the boiler air heater (discussed in Section                     •
3.4.1). Some of the boilers use a "hot-side" ESP where the control device is installed upstream                   I
of the boiler air heater. A  small number of existing boilers (7 percent) that fire low-sulfur coal
use FFs instead of ESPs. In general, FFs are being used at these coal-fired electric utility power
plants to obtain better PM  control or to solve ESP performance problems associated with high-
resistivity fly ash. A FF can be used only downstream of the boiler air heater because of
temperature limitations of the fabric filter bags.

       Post-combustion SC>2 emissions controls are used at approximately 27 percent of existing
coal-fired electric utility boilers.  The SO2 control technology most commonly used for these
boilers is a wet FGD scrubber. In all cases,  a PM control device, usually an ESP, precedes a
scrubber. Wet FGD scrubbers remove gaseous SC>2 from flue gas by absorption. In absorption,
gaseous species are contacted with a liquid in which they are soluble. For SOi absorption,
gaseous SO2 is contacted with a caustic slurry, typically water and limestone or water and lime.
The newer semi-dry SCh scrubber technologies currently are used at small number of the existing
coal-fired utility boilers (about 5 percent). However, for retrofit Hg control, these semi-dry
scrubbers have the advantage of an existing sorbent delivery system coupled with, in most cases,
a downstream FF to collect the reacted sorbent already in place. A detailed discussion of
potential retrofit options and current research and development status is presented  in following
sections.
                                         7-3
-------
7.2.3 Integrate Mercury Control Under a Multipollutant Control Strategy

       The most cost-effective, long-term Hg controls may be those implemented as part of a
multipollutant control strategy. Selection and deployment of new SO2, NOx, and fine PM
controls, which also control or contribute to the control of gaseous Hg in coal combustion flue
gas, may reduce or eliminate the need for Hg-specific controls. For example, installation of a
wet or semi-dry FGD unit should reduce oxidized Hg emissions by 90 to 95 percent over
previous levels; adding upstream NOx controls, which assist in oxidation of Hg°, would further
reduce total Hg emissions. Replacing or supplementing existing ESPs with FFs will likely
remove additional Hg, especially for bituminous coal applications.

       The remaining majority, Sections 7.3 through 7.7, discusses control technologies that
reduce Hg emissions by improving the performance of existing air pollution control devices to
capture the Hg in coal combustion flue gas. Section  7.8 discusses new multipollutant control
technologies (other than serial SOx-NOx-PM control devices), which are under development and
are potentially applicable to electric utility coal-fired electric utility power plants.

7.3 Post-combustion Mercury Control Retrofit

       Retrofits that reduce Hg emissions from existing electric utility coal-fired electric utility
power plants are implemented by modifying existing post-combustion emission control
techniques already in place. Potential retrofit options are identified and discussed below.
Options that are discussed may not be technically feasible or economically practical to install and
operate at all facilities.

7.3.1  Cold-side ESP Retrofit Options

       Add Flue Gas Cooling.  Lowering the flue gas temperature entering the ESP assists
natural fly ash sorption of Hg as well as improves the performance of any sorbents injected
upstream for Hg control.  However, the acid dew point temperature limits gas cooling when the
flue gas has significant HC1 or H2SO4 formation potential.

       Add Sorbent Injection. Gaseous Hg can be converted to Hgp  by  adsorption onto solid
particles in flue gas. Injecting sorbents into the flue gas upstream of the ESP increases the
amount of Hg captured in the form of Hgp.  This modification may require adding ducting
between the sorbent injection point and the ESP, and adding a gas absorber/humidifier upstream
of the ESP. This approach also may be limited by the ability of the ESP to collect high-
resistivity sorbents. For coal-fired electric utility boilers with marginally performing ESPs that
have difficulty meeting opacity requirements and may not be candidates for a sorbent injection
retrofit, the following option may be preferred.

       Add Downstream FF with Sorbent Injection.  Adding a FF downstream of the existing
ESP, while a more expensive retrofit option, allows a significant portion of the fly ash to be
collected without reacted sorbent and enhances overall PM control efficiency where ESP
performance is marginal.  Further, because the FF would have a much lower paniculate loading,
                                         7-4
-------
the collecting surface can be smaller (higher air-to-cloth ratio) or have longer cleaning cycles
(good for sorbent performance and bag life).

       ESP Modifications.  Potential ESP modifications include converting the last field of the
ESP to a wet ESP or a very compact pulsejet FF. These conversions would likely be made
because of PM collection improvements needed, rather than Hg control considerations;
nonetheless, associated Hg control benefits would also be likely.

7.3.2 Hot-side ESP Retrofit Options

       Convert to Cold-side ESP with Sorbent Injection. Adding flue gas cooling is not an
option for a hot-side ESP because of its location upstream of the air preheater. The only
potential retrofit option for Hg capture without adding a new downstream PM control device is
to convert the existing ESP from a hot-side configuration to a cold-side configuration.
Depending on the plant layout and ESP design, this may be possible by reconfiguring the ducting
and retuning the ESP to operate at the lower temperature. Adding sorbent injection with the
modification would further improve Hg capture. The lower flue gas temperature entering the
ESP enhances the adsorption of gaseous Hg onto fly ash or sorbent (if injected upstream) and
subsequent collection of the particulate Hg in the ESP.

       Add Downstream FF with Sorbent Injection. The same as for a cold-side ESP, adding a
FF downstream of the existing ESP, while a more expensive retrofit option, allows a significant
portion of the fly ash to be collected without reacted sorbent.

7.3.3 Fabric Filter Retrofit Options

       Add Flue Gas Cooling. As is the case for ESPs, lowering the flue gas temperatures
entering the FF enhances the adsorption of gaseous Hg onto fly ash or sorbent (if injected
upstream). Again, the acid dew point temperature limits gas cooling when the flue gas has
significant HC1 or H2SO4 formation potential.

       Add Sorbent Injection. Use of sorbent injection may require some internal FF
modifications to ensure good sorbent performance.  In general, existing FFs were not designed as
adsorbers, so some modifications may be in order to ensure that sorbent particles stay entrained
and become part of the filter cake. This may be accomplished by removing baffles, changing the
point of gas entry, increasing gas velocity, or using smaller sorbent particles. Operating
requirements of the FF may require more frequent cleaning with the additional sorbent loading.

       FF Modifications. Potential FF retrofit options include replacing fabric bags with
catalytic bags that oxidize Hg° to Hg++ and Hgp or adding electrostatic augmentation to increase
the bag cleaning cycle interval time and hence increase sorbent/gas contact time.  This last
improvement would be especially beneficial with higher-cost, high-capacity sorbents.

                                         1-5
-------
 s
 t
t
7.3.4 Spray Dryer Absorber Retrofit Options

       Use Oxidation Additives. Existing SDA systems already achieve very high Hg removal
on certain coals but show poor performance on other coals. Possible causes are low oxidation
potential resulting from high alkaline fly ash content as well as low effective carbon content in
fly ash.  Therefore possible performance improvements include producing a higher carbon
content fly ash by NOx combustion control modifications, direct addition of activated carbon to
the absorber with lime, and addition of oxidants to the absorber.

      Replace Existing ESP with FF Control Device.  Where the PM control device used for
the absorber is an ESP, replacement of the unit with a FF would likely improve Hg removal as a
result of enhanced PM control as well as greater conversion of Hg2+ to Hgp.

7.3.5  Wet FGD Scrubber Retrofit Options

       Use Oxidation Additives. Oxidation of the gaseous Hg° to gaseous Hg2+ can potentially
increase the total Hg removed by wet scrubbing since gaseous Hg2+ is more readily captured by
these systems than gaseous Hg°. Several flue gas additives and scrubbing additives are being
developed to increase the conversion of Hg° to Hg** prior to the scrubber inlet. Flue gas and
scrubber additives are also being developed for use in preventing the conversion of absorbed
Hg2* to gaseous Hg° in wet FGD systems.  The one caution is that increasing oxidants upstream
or within the scrubber may also oxidize other species such as SO3 and NO/NO2 to sulfuric and
nitric acid aerosols.

      Add Fixed Oxidizing Catalysts Upstream of Scrubber.  Improvements in wet scrubber
performance in capturing Hg may be accomplished by installation of fixed oxidizing catalysts
upstream of the scrubber to promote oxidization of Hg° to soluble species. Potential catalysts
currently are being tested.

       Wet FGD Scrubber Modifications.  Several studies of pilot-scale wet FGD systems
suggest that modifying the scrubber operation and design (as well as the control and design of
upstream ESPs) may  improve the capture of gaseous Hg2* and reduce the conversion of absorbed
Hg2"1" to Hg°. Specifically, these studies have found that the liquid-to-gas ratio  and tower design
of a wet FGD unit affect the absorption of gaseous Hg2+, while the oxidation air influences the
conversion of absorbed Hg2+ back to Hg° which is then emitted to the atmosphere in the scrubber
exhaust gas.

7.3.6 Particle Scrubber Retrofit Options

      A few existing power plants use wet scrubbers exclusively for control of PM emissions.
Knowledge gained in the enhancing control of Hg emissions from wet FGD scrubbers by
operating modifications also may be useful in improving the Hg removal performance of these
particle scrubbers.
                                                     7-6
-------
                                                                                                 t
7.4 Retrofit Control Technology Research and Development Programs

       None of the retrofit options discussed in Section 7.3 are routinely being used by the
electric utility industry at this time. In addition, the Hg emissions control technologies that are
successfully used for municipal waste combustors (MWCs) in the United States and Europe
cannot be directly retrofitted to existing coal-fired electric utility boilers.  Differences in flue gas
properties, combustion unit design, and other factors (discussed in Section 7.4.1) prevent the Hg
control devices now used for MWCs  to be directly installed at coal-fired electric utility power
plants. Consequently, development of effective retrofit control technologies for coal-fired
electric utility boilers is the subject of bench-scale, pilot-scale, and full-scale test programs.
Chapter 5 discusses laboratory studies investigating potential Hg control techniques for coal-
fired electric utility boilers. To further develop the most promising of these control techniques
for full-scale application to coal-fired electric utility boilers, pilot-scale and full-scale research
studies are being funded by the EPA, DOE, EPRI, state agencies, and private companies.
Section 7.4.2 describes several pilot-scale test units that are being used  for research and
development programs. Building upon the results obtain using these test facilities, a number of
full-scale test programs currently are  being conducted to provide a more thorough
characterization of the performance and potential for widespread commercial application of
specific retrofit Hg control technologies.

7.4.1  MWC Mercury  Control Technology

       Injection of activated carbon into the flue gas from a MWC and collecting the reacted
sorbent in a downstream FF is one Hg control method widely used for MWCs.4'5 Mercury
removal levels in excess of 90 percent are achieved.  However, the level of Hg control achieved
by adding sorbents into the flue gas from a particular combustion unit is influenced by the
particular characteristics of the flue gas from that unit including flue gas temperature, flow rate,
Hg content, and chloride Hg content.  Table 7-1 compares selected properties of the flue gas
from a coal-fired utility boiler with those for a MWC flue gas. As shown in this table, Hg
concentrations in MWC flue gas streams may be up to several orders of magnitude greater than
those seen in utility flue gas streams.  In addition, MWC flue gas contains mostly Hg2+, while
flue gas from coal-fired electric utility boilers can have substantial amounts of Hg°, which
generally is less likely to be adsorbed. Additionally the flue gas ductwork for a coal-fired utility
boiler is substantially larger and more complex (multiple passes) than for a MWC, therefore duct
injection of a sorbent is more complicated and its performance more difficult to predict for a
coal-fired utility boiler due to variations in temperatures, residence time, and other factors.

       Similarly,  the wet scrubber technology used by European MWCs is not directly
applicable to controlling emissions from coal combustion. European MWCs typically have two-
stage scrubbers consisting of a low-pH water scrubber to control hydrochloric acid (HC1)
emissions, produced as a result of the large quantities of plastics in the garbage burned, followed
by an alkaline scrubber to  control SC>2 emissions. In contrast, wet scrubbing systems typically
used by the electric utility industry in the United States to control SCh emissions resulting from
burning high sulfur coal consist of a single-stage wet scrubber using a limestone or lime
scrubbing agent. As a consequence, there are significant differences  in the underlying chemistry
                                         7-7
s
-------
 s
             Table 7-1. Comparisons of typical uncontrolled flue gas parameters for coal-fired
             utility boiler versus municipal waste combustor (MWC).
S
                           Flue Gas
                          Parameter
                         Temperature
                          Hg Content
                          (ug/dscm)
                        Chloride Content
                          (ug/dscm)
Flow Rate
(dscm/min)
                   Coal-fired Electric Utility Boiler
                                                      12110177
                             1to25
                         1.000 to 140,000
11,000 to 4,000,000
                       Municipal Waste Combustor
                                                       177 to 299'
                              400 to 1,400'
                            200,000 to 400,00'
80,000 to 200,000'
                   (a) Temperature, chloride content, and flow rate data taken or determined from Reference 6
                   (b) Mercury content data taken from Reference 4.
S
                                                       7-8
-------
                                                                                               s
of the scrubbing systems used for MWCs compared to those currently in use at coal-fired electric
utility power plants.

7.4.2 Pilot-scale Coal-fired Test Facilities

       To date, most of the retrofit control technology development has been conducted using
pilot-scale test units that simulate full-scale coal-fired electric utility boiler combustion
conditions. The DOE Federal Energy Technology Center, the Ohio Coal Development Office
(OCDO), and McDermott Technology, Inc., jointly funded one program titled the Advanced
Emissions Control Development Program (AECDP).  This test program was conducted in three
phases using a 10  MW coal-fired test faculty.7'8'9  The test facility is capable of testing a full-flow
ESP, a partial-flow pulsejet FF, and a wet FGD scrubber.  All testing under the AECDP was
performed firing Ohio bituminous coals.  Figure 7-1 shows a schematic of the test facility.
Specific AECDP test results related to specific retrofit options are discussed later in this chapter
under the relevant topic headings.

       For a DOE cooperative agreement test program, the project team of Public Service
Company of Colorado (PSCO), ADA Technologies, and EPRI fabricated a pilot-scale
paniculate control module (PCM) to investigate Hg control in actual coal combustion flue gas by                 I
different dry sorbents.10  Figure 7-2 shows a schematic of the PCM. The PCM draws a                 ^^     |
slipstream of flue  gas (600 actual cubic feet per minute) from the 350-MWe coal-fired electric
utility boiler (Unit 2) at PSCO's Comanche Station power plant.  This boiler is an opposed-fired,
pulverized-coal boiler firing Powder River Basin (PRB) subbituminous coal.  Flue gas can be
drawn either from the inlet (high paniculate loading) or the outlet (essentially particle free) of the
full-size Unit 2 reverse-gas FF.  In addition, the PCM can be configured as an ESP, a reverse-gas
or pulse-jet FF, and as EPRI's TOXICON pulse-jet FF. Gaseous Hg is injected into the flue gas
to the PCM along  with recycled fly ash and/or sorbent; the solids can be injected at various
locations upstream of the PCM  to investigate the effects of Hg adsorption at different in-flight
residence times (0.5 to 3 seconds). The PCM is also equipped with in-duct heating and water
spraying to investigate the effects of Hg adsorption at different temperatures. Specific results
from testing using the PCM are discussed later in this chapter under the relevant topic headings.

       The DOE National Energy Technology Laboratory (NETL) is conducting in-house
research studies using a 500-lb/hr coal combustion unit to simulate a pulverized-coal-fired
electric utility boiler. 1>12 Figure 7-3 shows a schematic of the DOE/NETL coal combustion test
facility. The system consists of a wall-fired, pulverized-coal furnace equipped with a water-
cooled convection system, a recuperative air heater, spray dryer, sorbent injection duct (SID) test
section, and FF. Sorbent can be injected at numerous locations along the SID test section; this
allows  for a wide range of sorbent in-duct residence times relative to the FF and to the SID flue-
gas sampling locations.
S
                                                                                               t
                                        7-9
-------
 s
                    QMWrCEOF

                    • ACCOP T»« eHMlpnwr*
                                       . Slip ttitttn
S
Figure 7-1. Schematic of 10-MWe coal-fired Babcock & Wilcox (B&W) Clean
Environment Development Facility (CEDF) as used for Advanced Emissions
Control Development Program (AECDP) (source: Reference 9).
S
                                            7-10
-------
                                                                              t
              ,t
Low Ash
              't
H—
Ast
Inlet
.Sample^^Duct Heater
\ ^
Hg
Doping
i
^, Carbon
ir Injection

Particulate Control Module
m^^m
1
Outlet
Sample
              Flue
              Gas
                          V
                                                                              t
Figure 7-2. Schematic of Particulate Control Module (PCM) at Public Service
Company of Colorado (PSCO) Comanche Station (source: Reference 10).
                                                                              t
                                 7-11
-------
 t
t
t
           Figure 7-3.  Schematic of DOE/NETL in-house 500-lb/hr coal combustion test
           facility (source: Reference 12).
                                            7-12
-------
                                                                                                 t
7.5 Mercury Control Retrofits for Existing Coal-fired Electric Utility Boilers Using ESP or
    FF Only

       The focus of research and development for existing coal-fired electric utility boilers
equipped only with an ESP or FF has been the use of dry sorbent injection. As discussed in
Chapter 5, gaseous Hg can be adsorbed onto solid particles in the flue gas.  A solid particle that
absorbs gaseous species is called a "sorbent."  The flue gas from every electric utility boiler that
directly burns coal (i.e., all boilers except for IGCC units) contains fly ash particles that adsorb
gaseous Hg in the flue gas to various degrees.  Other types of solid particles can be injected into
the flue gas for the purpose of adsorbing gaseous Hg.  Materials being investigated as possible
sorbents for Hg control include activated carbon, calcium-based and sodium-based (trona)
sorbents, various clays and zeolites, alkaline-earth sulfides, and lime and lime-silica
multipollutant sorbents.  An alternative sorbent-based Hg control approach that has been
investigated is passing the flue gas through a fixed bed of a noble-metal-based sorbent.

7.5.1  Sorbent Injection Configurations

       In general, four retrofit configurations are  possible for injecting dry sorbent particles into
the flue gas from a coal-fired utility boiler. It may not be technically feasible to implement one
or more of these configurations at a given existing coal-fired power plant because of site-specific
factors such as the existing flue gas duct configuration, availability of space to add additional
ducting or new control device, use of a wet FGD scrubber, or other plant layout and operation
considerations.

       Configuration A • Sorbent injection into the flue gas duct upstream of existing ESP or FF.
       Cooling of the flue gas upstream of the sorbent injection point or modifications to the
       ducting may be needed.

       Configuration B - Sorbent injection into the flue gas duct downstream of the existing PM
       control device followed by a new FF (to collect the reacted sorbent), with or without flue
       gas cooling upstream of the injection point. This configuration requires higher capital
       costs but reduces sorbent costs  compared to Configuration A. The configuration also
       allows the fly ash collected by the upstream PM control device to be sold without being
       mixed with the injected sorbent.

       Configuration C - Sorbent injection into a circulating fluidized-bed absorber (CFA)
       upstream of the existing ESP or FF, with or without flue gas cooling upstream of the
       CFA.  The advantage to using a CFA is that it  recirculates reacted materials with fresh
       sorbent to create an entrained bed with a high number of reaction sites resulting in higher
       sorbent utilization and enhanced capture of Hg and other pollutants.

       Configuration D - Sorbent injection into a CFA downstream of the existing PM control
       device and followed by a new FF (to collect the reacted sorbent). Like Configuration B,
       this configuration allows the fly ash collected by the upstream PM control device to be
       sold without being mixed with  the injected sorbent.
t
I
                                         7-13
-------
       The level of Hg capture using sorbent injection with a downstream ESP depends on in-
flight adsorption of Hg by entrained sorbent particles. Mercury capture in a downstream FF
occurs by this same in-flight adsorption process as well as a second mechanism when flue gas
must pass through the filter cake collected on the FF bags. This filter cake contains a mixture of
previously captured fly ash and sorbent particles, and provides good contact between gaseous Hg
and captured particles. Filter cake retention times between bag cleaning cycles may be as long
as 60 minutes, greatly increasing the adsorption of Hg on the sorbent particles.  This compares
with the relatively short time that in-flight adsorption occurs upstream of the control device
(nominal times for in-flight adsorption are 0.5 to 1.5 seconds).  In addition, FFs generally are
more efficient than ESPs in collecting fine particles and any associated Hgp (see Table 3-3).  The
extra contact time and higher collection efficiency provided by a FF reduces the amount of
sorbent needed for adsorption compared to what is needed for an ESP to achieve a given level of
control.

       Cooling the flue gas before the sorbent injection point can improve Hg adsorption by the
sorbent, which in turn may reduce the amount of sorbent needed for a given level of control.
However, the temperature to which the flue gas may be cooled is limited because sulfuric acid
(and perhaps hydrochloric acid) mists may be formed if the flue gas temperature drops below the
acid dew point(s) of the flue gas. For all four configurations, sorbent capacity may be
maximized by recycling and reinjecting sorbent and fly ash collected in the PM control device(s)
located downstream of the injection point.

7.5.2 Sorbent Adsorption Theory

       Gas-phase adsorption occurs when a gaseous specie contacts the surface of a solid and is
held there by attractive forces between the gaseous specie and the solid. In adsorption
terminology, the gaseous specie being adsorbed is called the "adsorbate," and the solid is called
the "adsorbent" or "sorbent."  While all solids have the potential to adsorb gaseous species,
adsorption is not very pronounced unless a solid has a large surface area. As a result, most solids
for gas-phase adsorption are highly porous and in the form of particles or granules.  The porosity
of the solids provides large amounts of internal surface area where most adsorption takes place.
When a gaseous  specie is adsorbed onto the surface of a solid particle, the gaseous specie
becomes a particle-bound specie.

       Gas-phase adsorption may be classified as chemisorption or physical adsorption
depending on the nature of the attractive force between the adsorbate and sorbent. In
chemisorption, the adsorbate reacts with the surface of the sorbent, thus, the attractive force
between die adsorbate and sorbent is similar to a chemical bond.  Chemisorption often involves
the use of sorbents impregnated with compounds that are reactive with the adsorbate. In physical
adsorption, the attractive force between an adsorbate and  sorbent is electrostatic in nature
(similar to the attraction between metal filings and a magnet, where the metal filings are
analogous to the adsorbate and the magnet is analogous to the sorbent).  Different adsorbates
have different attractive forces for a given sorbent due to  differences in molecular weight,
normal boiling point  (or vapor pressure), degree of unsaturation,  polarity, and structural
configuration. When a sorbent is exposed to more than one adsorbate, preferential adsorption
                                         7-14
-------
tends to take place due to differences in the attractive forces between the different adsorbates and
the sorbent particles.

       Equilibrium adsorption capacity is the maximum amount of adsorbate a given mass of
sorbent can hold at a given temperature and adsorbate gas concentration.  Generally, the
adsorption capacity of a sorbent for a given adsorbate increases with increased adsorbate
concentration and decreases with increases adsorption temperature.

       In a dynamic adsorption system (i.e., an adsorption system involving a moving gas
stream), a gas stream containing one or more adsorbates is passed through a fixed or fluidized
bed of sorbent particles or the sorbent particles are injected directly into the gas stream.  In
dynamic adsorption systems, the contact time between the sorbent particles and the adsorbate in
the gas stream is critical. While contact time does not affect the equilibrium adsorption capacity
of the sorbent, it directly affects the sorbent's ability to capture the adsorbate from the gas
stream.  Maximum capture of adsorbate from the gas stream will not take place unless the
adsorbate has sufficient time to contact the sorbent and diffuse into its pores. Thus, increasing
the contact time increases Hg capture by the sorbent.

7.5.5 Pilot-scale and Full-scale Research and Development Status

       The laboratory studies of using dry sorbents for Hg control based on bench-scale reactor
testing are discussed in Section 5.4. This section discusses the results  from field studies testing
different sorbents in pilot-scale or full-scale systems.

7.5.3.1 Coal Fly Ash Reinjection

       As discussed in Chapter 5, fly ash generated naturally when burning certain coals in a
utility boiler adsorbs some of the gaseous Hg in the flue gas.  The adsorption of gaseous Hg by
the fly ash vented in the  flue gas from the boiler, referred to by some researchers as "native fly
ash," is believed to occur at active sites on the ash surface similar to those on sorbent (e.g., fly
ash carbon analogous to activated  carbon or fly ash alkaline species akin to injected lime).  As
part of the DOE cooperative agreement test program to investigate dry sorbents, the project team
of PSCO, ADA Technologies, and EPRI evaluated Hg removal rates by the fly ash in the flue
gas from burning two types of Western coals and the potential for Hg removal by reinjection of
low levels of collected fly ash back into the flue gas upstream of the particulate control device.10
The use of reinjected fly ash for Hg control avoids the potentially adverse impact on the
commercial viability of selling the fly ash collected in the downstream particulate control
devices.  The use of activated carbon as a Hg sorbent may increase the level of carbon in the
collected fly ash/activated carbon mixture above allowable maximum  levels for some
commercial fly ash applications (e.g., sale of fly ash for use as a concrete additive).

       Full-scale testing was conducted at three PSCO coal-fired electric utility power plants to
characterize gaseous Hg removal by the native fly ash in flue gas at each facility; a boiler using a
FF for PM control was tested.  At one facility, a second boiler using an ESP was also tested.
Two of the three power plants burned subbituminous coal from the Powder River Basin (PRB),
                                         7-15
-------
4ft
            and the other burned a Colorado-mined bituminous coal. Flue gas measurements were taken
            concurrently at the inlet and outlet of each particulate control device.  At two of the power
            plants, testing was conducted in both the summer and winter in order to investigate the effect of
            ambient temperature on the adsorption of Hg on the fly ash.

                  Results of the full-scale tests are summarized in Table 7-2. Mercury removal measured
            across the three FFs ranged from 61 to 99 percent.  Mercury removal across the ESP was
            significantly lower at 28 percent.  The two boilers units demonstrating Hg removals above
            80 percent (Arapahoe 4 and  Cherokee 3) were equipped with low-NOx burner retrofits. The use
            of these burners often causes elevated levels of unburned carbon in the fly ash. Measuring
            unburned carbon by the "loss-on-ignition" (LOI) test, the fly ashes from Arapahoe 4 and
            Cherokee 3 had LOI contents approximately 7 to 14 times higher than the fly ashes from the
            other two boilers. The  Hg levels measured for the Cherokee 3 unit was essentially the same in
            both summer and winter, indicating no adverse temperature effects on adsorption. In contrast,
            the Arapahoe 4 tests showed better adsorption at cooler test conditions (i.e., winter versus
            summer).

                  To examine the use of fly ash reinjection for Hg emissions controls, a series of pilot-scale
            tests were conducted by collecting the fly ash samples from the three power plants and injecting
            the collected fly ash into the  PCM located at the Comanche Station (discussed in Section 7.4.2).
            For the recycled fly ash tests, the PCM was configured as a reverse-gas FF and drew fly-ash-free
            flue gas from the  outlet side of the FF serving the coal-fired boiler. The flue gas was spiked with
            gaseous Hg to produce a Hg  concentration of approximately 10 ug/Nm3. The gaseous Hg
            concentration was sampled at the inlet and outlet of the PCM using a Hg continuous emissions
            monitor (Perkin Elmer  MERCEM). Recycled fly ash was injected into the flue gas just
            downstream of the inlet sampling port. Except during one test, the injected fly ash samples were
            not treated in any way to enhance their Hg-adsorbing properties.  For one test, a sample of fly
            ash from the Comanche 2 unit was treated with a hot nitrogen purge in an attempt to desorb any
            Hg on the ash particles.

                  Table 7-3  summarizes Hg removal data for the fly ashes tested. Reinjected
            subbituminous coal fly ash removed 84 to 86 percent of the gaseous Hg across the PCM. In
            contrast, reinjecting fly ash from the boiler burning bituminous coal showed only a 10 percent
            removal of gaseous Hg. The removal efficiency for bituminous coal fly ash was increased to 31
            percent when this ash was thermally pretreated to desorb Hg before injection into the PCM.  The
            results in Table 7-3 show that the recycled fly ashes from the Cherokee and Arapahoe boilers had
            additional capacity to adsorb gaseous Hg (beyond what they had adsorbed from their source flue
            gas),  while the untreated recycled fly ash from the Comanche 2 boiler appeared to be saturated or
            no longer reactive. The LOI  contents of the Cherokee 3 and Arapahoe 4 fly ash samples were 8
            and 14 percent, respectively. The  LOI contents of the Comanche 2 fly ash samples were 0.3 to
            0.4 percent. As was observed during the full-scale testing, fly ashes with the highest LOI
            contents (those from the Arapahoe 4 and Cherokee 3 boilers) adsorbed more Hg than fly ashes
            with lower LOI contents (those from the Comanche 2 boiler).
                                                   7-16
-------
Table 7-2. Hg removal by native fly ashes measured across PM control devices at
PSCO power plants burning selected western coals (source: Reference 10).
Power Plant
PSCO*
Cherokee
PSCO
Arapahoe
PSCO
Comanche
Type of Coal
Burned
Bituminous
(Colorado)
Subbituminous
(Powder River
Basin)
Subbituminous
(Powder River
Basin)
PM Control
Device
Reverse-gas FF
(Boiler Unit #3)
ESP
(Boiler Unit #1)
Reverse-gas FF
(Boiler Unit #4)
Reverse-gas FF
(Boiler Unit #2)
Ash Carbon
Content
(% LOI°)
7.6
<1
14.4
0.4
Gaseous Hg
Removal
(%)
98 (summer)
99 (winter)
28
62 (summer)
82 (winter)
61
      (a) PSCO = Public Service Company of Colorado
      (b) LOI = Loss on ignition
                                   7-17
-------
Table 7-3. Hg removals by fly ash reinjection measured across PCM at PSCO
Comanche power plant for selected western coals (source: Reference 10).
Reinjected Fly Ash
Coal Source
(PSCO power plant)
PRB Subbituminous coal
(Arapahoe 4)
PRB Subbituminous coal
(Cherokee 3)
Colorado
Bituminous coal
(Comanche 2)
Flue Gas
Temperature
320
320
280
280
Ash
Reinfection
Rate
(grains/act)
0.13
0.33
1.13
1.21
Ash
Carbon
Content
(% LOO
14.4
7.6
0.42
0.26
Gaseous Hg
Removal
84
86
10
31"
      (a) LOI = Loss on ignition
      (b) Deadsorbeb ash.
                                  7-18
-------
       In addition to evaluating the adsorption capacity of recycled fly ashes, several tests were
made using the PCM to evaluate the effects of temperature on fly ash adsorption.  For the
temperature tests, fly-ash-laden flue gas was extracted from the inlet of the FF serving the
Comanche 2 boiler and passed through the PCM; gaseous Hg was injected upstream of the PCM.
Hg adsorption across the PCM was monitored as the temperature of the flue gas through the
PCM was varied. Table 7-4 summarizes the results of the temperature tests. For the baseline
tests (no heating or cooling), the temperature of the flue gas through the PCM was in the range of
135 °C (275 °F); at this temperature, the Comanche 2 fly ash removed 20 to 40 percent of the
gaseous Hg present in the flue gas. When the flue gas was heated to around 152 °C (305 °F), the
fly-ash Hg removal dropped to zero, while spray cooling to reduce the flue gas temperature to
about 110 °C (230 °F) increased the Hg removal to around 60 percent. As expected, the data from
these tests show that adsorption is greatly affected by temperature, with adsorption increasing
with decreasing flue gas temperature.

7.5.3.2 Activated Carbon Sorbent Injection

       The most frequently tested activated carbon for Hg removal from coal combustion gases
has been a commercially available carbon manufactured by Norit Americas, Inc. (trade name
Darco FGD™).  The Darco FGD™ carbon is produced from lignite specifically for the removal
of heavy metals and other contaminants from MWC flue gas streams.  Other commercially
available activated carbons and experimental carbons also have been tested.

       A full-scale test program jointly funded by EPRI and Public Service Electric and Gas
(PSE&G) evaluated the potential of activated carbon injection for Hg control.11 The tests were
performed at the PSE&G Hudson Generating Station, which fires low-sulfur bituminous coal and
uses an ESP for PM control. Two types of activated carbon were tested, the Darco FGD™
carbon and  an experimental carbon identified as AC-1.  Results from these tests are shown in
Table 7-5. The data indicate a distinct reduction in total Hg removal efficiency with increased
temperature.  The maximum Hg removal measured was 83 percent using the Darco FGD™
carbon at a C:Hg ratio of 45,000:1 and an ESP operating temperature of 221 °F. Full-scale ESP
operation at this low temperature is not practical, however, due to potential problems with acid
condensation.

       Sorbent injection using Darco FGD™ carbon and an ESP was also tested as part of the
AECDP Phase III studies.9  For this test, the coal burned was an Ohio bituminous coal.  The
carbon was injected upstream of the ESP, with an approximate in-flight particle residence time
of 1 second. The injection temperature was approximately 204 °C (400 °F) and the ESP inlet
temperature was about 174 °C (345 °F). The carbon flow rate was approximately 14 Ib/hr, which
is equivalent to a C:Hg mass ratio of 9,000:1. Both particulate and gaseous Hg species were
measured at the inlet and outlet of the ESP during the carbon injection test. The test results are
presented in Figure 7-4. Also shown in this figure are baseline Hg concentrations measured
before any injection tests. Compared to the baseline condition,  injection of the activated carbon
resulted in a total Hg removal of 53 percent. Carbon injection at the test conditions had no effect
on the removal of gaseous Hg°, suggesting that Hg removal appears to be a result of the capture
                                        7-19
-------
Table 7-4. Effect of flue gas temperature on fly ash Hg adsorption measured
across PCM at PSCO Comanche power plant burning PRB subbituminous coal
(source: Reference 10).
Test Condition
Baseline
Heated flue gas
Cooled flue gas
Flue Gas
Temperature
CC)
135
152
110
Gaseous
Hg Removal
(%)
20 to 40
0
60
                                7-20
-------
Table 7-5.  Hg removal by activated carbon injection measured at PSE&G Hudson
Station burning low-sulfur bituminous coal and using ESP (source: Reference
13).
Sorbent
Tested
Baseline
(no sorbent injection)
Darco FGD™
Activated Carbon
Experimental
Activated Carbon
AC-1
ESP Operating
Temperature
IT)
255
268 -278
240 - 255
240 - 255
220 -235
275 -280
270 -275
240 -250
240 - 250
280
Sorbent Injection
Ratio
-------
     25.0
     20.0
     10.0
      5.0
      0.0 —
                                       • Partkubte
                                       D Elemental
                                       • OxWbed
iurauce Exit Before-
 Sorbent Injection
                                 Furnace Exit
                             Afler Sorbent l^ectbn
RSPExll
Figure 7-4. Hg removal by activated carbon injection measured at AECDP test
facility burning Ohio bituminous coal and using ESP (source: Reference 9).
                                   7-22
-------
of gaseous Hg2+ (onto or into the particulate phase) and then the subsequent removal of the
participate in the ESP.

       The DOE/NETL also tested injecting Darco FGD™ carbon for Hg control using the
DOE/NETL in-house coal combustion test facility.11 For these tests, low-sulfur bituminous coal
was burned based on the rationale that this is a coal-type likely be burned in utility power plants
that do not have flue gas desulfurization systems. Throughout testing, the furnace was operated
to achieve high combustion efficiency with low levels of unbumed carbon in the fly ash.
Unburned carbon levels in the fly ash under baseline conditions were  generally less than two
percent. Flue gas measurements of Hg were conducted at the FF inlet using the OH Method, and
a Modified Ontario-Hydro Method (MOH Method). The modified method samples the flue gas
non-isokinetically whereas the former samples the flue gas isokinetically.  Stack measurements
downstream of the FF were made for speciated Hg using the OH Method and total Hg using EPA
Method 101 A.  Analysis of coal and ash deposits was made using ASTM D3684.  The MOH
Method was used at the inlet to minimize PM collection during sampling.  Eliminating entrained
PM in the sample flue gas allowed researchers to determine in-duct Hg removals.  In addition,
the effect of filtered solids on Hg speciation was deduced by comparison with the  Hg speciation
measured with the OH  Method.

       Test results measured using the DOE/NETL test facility for sorbent injection upstream of
a FF using the Darco FGD™ carbon are presented in Table 7-6.  Total Hg removals measured
ranged from 39 to 86 percent at  injection C:Hg ratios of 2, 600:1 to 10, 300:1. The test results
show a general trend where the total Hg removal increased with increasing C:Hg ratios.  A
second commercially available activated carbon has also been tested for possible Hg control
using the NETL test facility.12 Mercury removals of 30 to 40 percent were measured injecting
Calgon FluePac ™ activated carbon at C:Hg injection ratios of 2,500:1 to 5,100:1.  The
DOE/NETL in-house research also shows no significant in-duct removals of Hg under the test
conditions, and Hg° appears to be oxidized by the filter cake.  On-going research on activated
carbon injection using the DOE/NETL test facility includes tests to quantify the effects of
humidification and FF pressure drop on Hg removal, evaluating novel sorbents, determining
sorbent effectiveness downstream of a FF  with and without recycle, and comparing Hg removals
using sorbent injection with ESP versus FF.12

       A multiple-site, full-scale field test program is currently being conducted under a
DOE/NETL cooperative agreement to obtain performance and cost data for using  activated
carbon injection to reduce Hg emissions from existing coal-fired electric utility power plants
equipped only with an ESP or FF for post-combustion air pollution controls.14  The DOE/NETL
is working in partnership with ADA-ES, PG&E National Energy Group (NEG), Wisconsin
Electric, a subsidiary of Wisconsin Energy Corp., Alabama Power Company, a subsidiary of
Southern Company, EPRI, and Ontario Power Generation on  a field evaluation program at four
power plant  facilities. Other organizations participating in this test program as team members
include EPRI, Apogee  Scientific, URS Radian, Energy & Environmental Strategies, Physical
Sciences, Inc., Southern Research Institute, Hamon Research-Cottrell, Environmental Elements
Corporation, Norit Americas, and EnviroCare International. The first test site is a boiler unit at
the Alabama Power Gaston facility that burns various low-sulfur bituminous coals and is
                                       7-23
-------
Table 7-6. Hg removal by activated carbon injection measured at DOE/NETL in-
house test facility burning low-sulfur bituminous coal and using FF (Source:
Reference 11).
Test
Run ID
9907-1
(baseline)
9907-2
9907-3
9907-4
9908-1
(baseline)
9908-2
9908-3
9908-4
Fabric Filter
Temperature
(T)
294
294
265
268
296
296
296
270
Sorbent
Injection
Ratio
(C:Hg)
0
9.500:1
10,300:1
6,200:1
0
2,600:1
5,400:1
2,900:1
Total
Hg
Removal
(%)
2.7
86.0
82.3
75.1
35.0
38.8
64.0
54.2
Mass Balance (%)
Fabric Filter
103.2
77.4
130.1
80.0
84.4
100.6
94.7
103.2
Overall
79.4
78.6
76.7
98.1
67.1
90.8
89.1
86.8
                                 7-24
-------
equipped with a hot-side ESP followed by a COHPAC FF. Testing at this site was conducted in
the spring of 2001.15 The next test site being tested is a boiler unit at the Wisconsin Electric
Pleasant Prairie facility that burns PRB subbituminous coal and uses a cold-side ESP for PM
control. The other two sites are scheduled to be tested in 2002, and are the PG&E NEG Salem
Harbor and Brayton Point facilities that burn low-sulfur bituminous coals and are equipped with
cold-side ESPs.

7.5.3.3  Calcium-based Sorbent Injection

       An alternative to using activated carbon is to use a calcium-based sorbent. Laboratory
studies conducted by the EPA and Acurex Environmental Corporation (funded by the State of
Illinois, ICCI) indicated that the injection of calcium-based sorbents into flue gas could result in
significant removal of Hg (discussed in Section 5.3).  Other benefits associated with the use of
limestone injection for Hg control include an incremental amount of SC«2 removal and a high
probability for SOs removal. Flue gas Hg removal using furnace limestone injection was
evaluated as part of a study conducted by McDermott Technology, Inc. titled Combustion 2000
Project/Low Emission Boiler System Program.16  In this study, limestone was injected into the
upper furnace firing Ohio bituminous coal at a temperature of about 1,204 °C (2,200 °F).  The
Ca:S ratio was set at 1.40 mol/mol.  An 80 percent efficient cyclone was then used to collect the
fly ash and calcined lime. At this location the flue gas temperature was approximately 163 °C
(325 °F). The Hg concentration in the flue gas was measured downstream of the cyclone using
the OH Method.  The measured Hg concentrations for the baseline (no limestone injection) and
the six limestone injection tests are shown in Figure 7-5. The data show that the Hg
concentration in the flue gas was significant lower when limestone was injected compared to the
baseline. The overall average Hg reduction for the six limestone injection runs was 82 percent.
The researchers note that using more efficient ESP or FF PM control devices with collection
efficiencies of greater than 99 percent  in place of a cyclone (see Table 3-3) is expected to
provide an additional increase in Hg removal.

       Based on the test results from the EPA/Acurex ICCI studies and the Combustion 2000
Project/Low Emission Boiler System Program, McDermott Technology, Inc. conducted
additional limestone injection tests during Phase  HI of the AECDP.9  The same limestone
previously tested in the Combustion 2000 program was used for the Phase in tests.  Two
limestone flow rates were tested. The flow rates chosen for the limestone injection tests were
200 Ib/hr (Ca:S = 0.35 mol/mol) and 25 Ib/hr (Ca:S = 0.04 mol/mol). An injection temperature
target of 1,149 °C to 1260 °C (2,100 °F to 2,300 °F) was chosen as the optimum range to
calcine the limestone (CaCO3) into lime (CaO).  It was assumed that CaO would be more
reactive with Hg, as it is with SO2, because of the increased surface area and reactivity.
Limestone was injected upstream of an ESP. The ESP inlet flue gas temperature was 177 °C
(350 °F). Mercury  concentrations were determined at the inlet and outlet of the ESP with
triplicate Ontario Hydro measurements. One set of triplicate measurements was performed prior
to sorbent injection to provide a baseline set of comparison data.

       Figure 7-6 shows the Hg partitioning and speciation for three sets of Hg measurement
locations: 1) at the ESP inlet without limestone injection (baseline); 2) at the ESP inlet with
                                        7-25
-------
~  14

I  12
31
r;  10


I   s
3   6
s

4*

S
   £    2
        0
                                           G Baseline Without
                                             Limestone Injection

                                           • Limestone Injection
                                             with Cyclone
                                             Particulate Removal
                                           O Average with
                                             Injection
            Base     #t      #2      #3     #4

                                 Injection Tests
                                                       #6   Average
Figure 7-5. Hg removal by limestone injection measured in Combustion 2000
furnace using mechanical cyclone separator (source: Reference 9).
                                   7-26
-------
                                                    • Particular;
                                                    OBertentat
                                                     Oxidized
              Furmce ExK
             Before Sorbext
               Iqjectkm
    Furnace Exit
After Sot-tent Injection
ESP Kvlt
Figure 7-6. Hg removal by limestone injection measured at AECDP test facility
burning Ohio bituminous coal and using ESP (Source: Reference 9)
                                   7-27
-------
 limestone injection of 200 Ib/hr; and 3) at the ESP outlet with limestone injection of 200 Ib/hr.
 As shown in Figure 7-6, the total Hg in the flue gas at the ESP inlet with and without limestone
 injection is about the same. Limestone injection substantially increases the Hgp, thereby
 substantially reducing gaseous Hg2+. The Hgp is then removed by the ESP, providing an overall
 Hg removal of 53 percent compared to the baseline condition. Reducing the limestone feed rate
 to 25 Ib/hr showed the same Hg partitioning trends observed for 200 Ib/hr but with a reduction in
 total Hg removal. An overall Hg removal of 41 percent compared to the baseline condition was
 measured.  The increased removal provided by limestone injection compared to the baseline
 appears to be a result of the capture of Hg2+ by the  CaO particulate (onto or into the particulate
 phase) and the subsequent removal of the particulate  in the ESP.  Limestone injection had no
 apparent effect on the Hg°.

        Table 7-7 presents a summary comparison of limestone sorbent injection test results with
 the activated carbon injection results from the AECDP Phase III studies (discussed in
 Section 7.5.3.2).  The table shows that limestone sorbent injection at 200 Ib/hr achieved an
 equivalent level of total Hg removal with activated carbon injection. The difference in sorbent-
 to-Hg ratios for these two tests is about a factor of  15. Based on the test results, the researchers
 concluded that activated carbon is a more effective sorbent than limestone on a mass basis;
 however, because the cost of activated carbon typically is an order of magnitude more than the
 cost for limestone, limestone is more effective on a sorbent cost basis.

 7.5.3.4 Multipollutant Sorbent Injection

        The EPRI/PSE&G Hudson sorbent injection study discussed in section 7.5.3.2 included
measurement of Hg removal by coinjection of activated carbon with calcium-based sorbents for
SCh control.13 The calcium-based sorbents tested were sodium bicarbonate and hydrated lime.
With the coinjection of either of the calcium-based sorbents, the researchers reported
improvement in the adsorption of gaseous Hg by the activated carbon.

        A study of the coinjection of a  sodium-based sorbent with activated carbon showed  that
 the removal of gaseous Hg by the native fly ash and the activated carbon was impeded
 when the sodium sesquicarbonate was  coinjected.  As part of the AECDP Phase III studies using
 the PCM at the PSCO Comanche Station, tests were conducted to investigate whether any
 synergistic removal of Hg or impairment of SO2 removal occurs when injecting both activated
 carbon for Hg control and sodium sesquicarbonate  for SC>2 control into the flue gas and collected
 in a FF.17'18 The activated carbon tested was Darco FGD™.

        When no  sorbent (carbon or sodium) was injected into the flue gas, the measured Hg
 removal across the PCM by the native  fly ash ranged from 41 to 76 percent at the respective
 temperatures of 162 °C (324 °F) and 138 °C (280 °F). When activated carbon was injected into
 the flue gas with  no sodium sesquicarbonate, measured Hg removal across the PCM was
 74 percent at 162 °C (324 °F).  When sodium sesquicarbonate was injected into the flue gas with
 no activated carbon injection, gaseous  Hg removal  percentages were in the negative range (i.e.,
 test measurements indicated an increase  in Hg concentrations at the PCM outlet compared to the
 inlet). When both activated carbon and sodium sesquicarbonate were injected into the flue  gas,
                                         7-28
-------
Table 7-7.  Comparison of Hg removals for activated carbon injection versus
limestone injection measured at AECDP test facility burning Ohio bituminous
coal and using ESP (Source: Reference 9).
Parameter
Sorbent injection rate
Sorbent:Hg mass ratio
Sorbent injection
temperature (°F)
ESP operating
temperature (°F)
Total Hg removal (%)
Sorbent Injected Upstream of ESP
Activated
Carbon
14 Ib/hr
9,000:1
400
345
53
Limestone
0.35 Ca:S mass ratio
200 Ib/hr
125.000:1
2,200
350
53
0.04 Ca:S mass ratio
25 Ib/hr
16,000:1
2,200
350
41
                                 7-29
-------
o
            Hg removal percentages ranged from -104 to 22 percent.  The SO2 removal percentages did not
            appear to be either impeded or improved with the coinjection of the activated carbon.

                  Based on the limited data, the researchers  speculated that the impediment of Hg capture
            occurred either because of inhibition of the sorbent mechanism or because the addition of sodium
            increased the level of NO2 in the flue gas. During the sodium sesquicarbonate tests, NO2 in the
            flue gas  increased from 5 to 41 ppmv, with the higher values associated with the higher
            temperatures tested. If the increase in the NOj levels was real, researchers are questioning
            whether NO2 had a negative impact on Hg removal and subsequent Hg desorption in the flue gas.
            Nitrogen dioxide is a strong oxidizer, which may have stripped Hg from the internal surfaces of
            the PCM, resulting in higher Hg measured at the outlet than the inlet (thus explaining the
            negative removal efficiencies for Hg). If this were the case, the effect would diminish over time
            as the Hg on the walls of the pilot unit came into equilibrium with the flue gas.  No tests were
            run with sufficient time to observe this effect, and credible Hg data were not available in real
            time.

                  The negative impact of the sodium sesquicarbonate injection on Hg removal by activated
            carbon injection is contrary to the results reported for the Hudson Station power plant tests  where
            injecting either sodium bicarbonate or hydrated lime with activated carbon improved the
            activated carbon's Hg adsorption capability. The  Hudson data were taken over a single test day,
            and the two power plants tested burned different coal types with different fly ash properties and
            flue gas  compositions (eastern bituminous coal at  Hudson versus PRB subbituminous at
            Comanche). Drawing any definite conclusions regarding coinjection of alkaline materials and
            activated carbon based on these two tests would be conjecture.

            7.5.3.5 Noble-metal-based Sorbent in Fixed-bed Configuration

                  ADA Technologies Inc. (ADA) has patented a sorbent process for Hg control in coal
            combustion flue gas, trade name Mercu-RE M.  Unlike the dry sorbent injection processes
            previously discussed, the Mercu-RE™ process is  based on the adsorption of the Hg by noble
            metals in a fixed-bed, regeneration of the sorbents by thermal means, and recovering the
            desorbed Hg for commercial recycle or disposal.19'20 Laboratory testing of the noble-metal
            sorbent showed that the sorbent captured virtually all of the Hg° and mercuric chloride  injected
            into a simulated coal combustion flue gas.  During 1999, the noble-metal sorbent was tested for
            6 months using  a flue gas  slipstream from the PSE&G Hudson Station. The acid gases in the
            flue gas  degraded the performance of the noble-metal sorbent. The field data suggested that
            there are limitations on the commercial application of using noble-metal sorbents for removal of
            Hg from coal combustion  flue gas without upstream acid gas controls installed.  Laboratory
            testing indicated that sorbent capacity can be recovered by scrubbing acid gases from flue gas
            prior to the sorbent bed. Additional  testing is being conducted to determine if noble-metal
            sorbents can be  used effectively on scrubbed flue  gas.
                                                    7-30
-------
7.6 Mercury Control Retrofits for Existing Coal-fired Electric Utility Boilers Using Semi-
Dry Absorbers

7.6.1  Retrofit Options

       Spray dryer absorber systems are the most common semi-dry scrubbers currently being
used at electric utility coal-fired electric utility power plants. With this control technology, a
slurry of hydrated or slaked lime is sprayed into an absorber vessel where the flue gas reacts with
the drying slurry droplets. The resulting particle-laden dry flue gas then flows to an ESP or an
FF where fly ash and 862 reaction products are collected.  In some cases, water-soluble sodium-
based sorbents are used instead of calcium-based sorbents. SDA systems can also provide
opportunities for injection of other dry sorbents for Hg or multipollutant control schemes.

       In a dry sorbent injection (DSI) system, a sorbent is injected into a flue gas duct upstream
of the PM collector. In many cases water is injected upstream of the sorbent injection location to
increase flue gas moisture content. This water spray, called spray humidification, reduces the
flue gas temperature and increases the sorbent reactivity. DSI systems can also provide
opportunities for injection of Hg or multipollutant sorbents. A circulating fluid-bed absorber
(CFA) is effectively a "vertical duct absorber" that allow simultaneous gas cooling, sorbent
injection and recycle, and gas sorption by flash drying of wet lime reagents.  It is believed that
CFAs can potentially control Hg emissions at costs lower than those associated with use of spray
dryers. With these absorbers, opportunities for use of advanced sorbents appear to be more
favorable than for DSI, due to the improved sorbent utilization by re-circulation, recycle, and                     •
flash evaporative cooling.                                                                                 I

7.6*2  Pilot-scale and Full-scale Research and Development Status

      Full-scale tests on eastern bituminous coals (i.e., a 180 MWe boiler with a SDA-FF
control system and a 55 MWe boiler with CFA-FF controls) were conducted in September
2000.21  The EPA Method 101A was used for absorber inlet Hg measurements and the OH
Method for the boiler stacks. Both units averaged over 97 percent Hg removal in the respective
control systems based on outlet and inlet flue gas measurements. Using the raw coal analysis
and the stack OH Method measurements, each system removed about 95 percent of total Hg.
Further Hg/multipollutant testing of SDA and CFA units are planned in DOE-EPRI-EPA pilot
and field test programs.

7.7 Mercury Control Retrofits for Existing Coal-fired Electric Utility Boilers Using Wet
    FGD Scrubbers

7.7.1 Retrofit Options

       Wet FGD scrubbers are typically installed downstream of an ESP or FF. Removal of PM
from the flue gas before it enters the wet scrubber reduces solids in the scrubbing solution and
avoids chemistry problems that may be associated with fly ash.  In the United States, plants that
use wet limestone scrubbers for SO2 control generally capture more than 90 percent of the Hg2+
                                        7-31
-------
in the flue gas entering the scrubber. Consequently these FGD scrubbers may lower Hg
emissions by about 20 to more than 80 percent, depending on the speciation of Hg in the inlet
flue gas.

       Improvements in wet scrubber performance in capturing Hg depend primarily on the
oxidation of Hg° to Hg2*.  This may be accomplished by the injection of appropriate oxidizing
agents or installation of fixed oxidizing catalysts to promote oxidization of Hg to soluble
species. Oxidation of gaseous Hg° to gaseous Hg2"1" can potentially increase the total Hg removed
by wet scrubbing and sorbent systems since gaseous Hg * is more readily captured by these
systems than gaseous Hg°. Several flue gas additives and scrubbing additives are being
developed for this purpose. Flue gas and scrubber additives are also being developed for use in
preventing the conversion of absorbed Hg2+ to gaseous Hg° in wet FGD systems.

       An alternative strategy for controlling Hg emissions from wet FGD scrubbing systems is
to inject sorbents upstream of the PM control device. In units equipped with FFs this allows for
increased Hg capture and oxidization of Hg° as the flue gas flows through the filter cake.
Increased oxidization afforded by FFs results in increased Hg removal in the downstream
scrubber.  In FGD units equipped with ESPs, performance gains are limited by sorbent injection
and Hg adsorption rates.

7.7.2 Mercury Absorption Theory

       Gaseous Hg° is insoluble in water and therefore does not absorb in  the aqueous slurry of a
wet FGD system. Gaseous compounds of Hg2+ are water-soluble and do absorb in such slurries.
When gaseous compounds of Hg2+ are absorbed in the liquid slurry of a wet FGD system, the
dissolved species are believed to react with dissolved sulfides to form mercuric sulfide (HgS);
the mercuric sulfide precipitates from the liquid solution as a sludge. In the absence of sufficient
sulfides in the liquid solution, a competing reaction that reduces/converts dissolved Hg2+ to Hg
is believed to take place.  When this conversion takes place, the newly formed (insoluble) Hg° is
transferred to the flue gas passing through the wet FGD unit. The transferred Hg° increases the
concentration of Hg° in the flue gas passing through the wet FGD unit (since the incoming Hg° is
not absorbed) giving rise to a higher concentration of gaseous Hg° in the flue gas exiting the wet
FGD than entering it. Transition metals in the slurry (originating from  the flue gas) are
suspected to play an active role in the conversion reaction since they can act as catalysts and/or
reactants for reducing oxidized species

7.7.3 Pilot-scale and Full-scale Research and Development Status

7.7.3.1 Oxidation Additives

       As part of the AECDP Phase III studies, tests were conducted to investigate two potential
chemical additives for controlling the conversion of oxidized Hg to the  elemental form, and
enhancing the control of Hg in a pilot-scale wet FGD system.9 The first additive was gaseous
H2S.  The selection of H2S as a potential additive was based on the possibility that a sulfide-
donating species could assist in capturing Hg2+. A HjS gas stream at a  concentration of about 2
                                        7-32
-------
ppm was injected into the flue gas entering the scrubber. The Hg concentrations of gaseous Hg2*
and gaseous Hg° measured at the wet scrubber inlet and outlet for the baseline and HaS injection
tests are shown in Figure 7-7,  Gaseous Hg removal by the wet scrubber increased with the
addition of HiS (at about 2 ppm) from 46 to 71 percent. This increase was attributed mainly to a
decrease in the conversion of Hg2+ to gaseous Hg°.

       The second additive tested was ethylenediaminetetraacetic acid (abbreviated EDTA).
This chemical was selected because EDTA is strong chelating agent.  Chelating agents react with
metallic ions to form soluble nonionic compounds. Because, transition metals may act as a
catalyst in the conversion of Hg2+ to gaseous Hg° in wet FGD scrubbers, their chemical binding
may reduce the conversion.  For the test, EDTA was added to the scrubbing slurry. The Hg
concentration of gaseous Hg2+ and gaseous Hg° measured at the wet scrubber inlet and outlet for
the ESP baseline and EDTA additive tests is shown in Figure 7-8. The total Hg removal
increased to  73 percent with the addition of EDTA. Under a new cooperative agreement with
DOE/NERL, McDermott Technologies, Inc. is conducting a full-scale test program of using
scrubber additives to achieve increased Hg removal at two power plants burning high-sulfur
Ohio bituminous coal: 1) Michigan South Central Power Agency's (MSCPA) 55-MWe Endicott
Station located in Litchfield, MI, and 2) Cinergy's 1300-MWe Zimmer Station located near
Cincinnati, OH.22

7.7.3.2 Mercury Oxidation Catalysts

       Under a DOE/NETL cooperative agreement, laboratory and field tests were conducted to                  I
investigate catalytic oxidation of gaseous Hg° in coal-fired electric utility boiler flue gas.  The                   I
project tested the actual rate to convert gaseous Hg° to a soluble form using different candidate
catalysts under simulated and actual coal combustion  flue gas conditions.  The results of the
bench-scale  studies are discussed in Chapter 5.  Additional extended tests with the most-active
catalysts and fly ash were conducted in the field to assess their adsorption and/or oxidation of Hg
in an actual coal-fired boiler flue gas.24 These tests were conducted in a fixed-catalyst-bed test
rig using a flue gas slipstream from a electric utility boiler firing a Texas lignite. Total Hg
concentrations in the flue gas slipstream varied from 7 to 35 ug/Nm3, with the gaseous Hg°
concentrations varying from 4 to 18 Jig/Nm3. The inlet  gaseous Hg2+ also was variable, ranging
from 30 to 80 percent of the total, and the concentrations of SO2 and NOx varied considerably
during the testing period. The catalysts and fly ash were exposed to flue over periods ranging
from 3,480 to 3,490 hours. Table 7-8 presents the oxidation results over the 5-month-plus period
of testing. For the values of the catalyst field measurements shown in the table, the Hg°
oxidation measured across the sand "blank" was subtracted from the actual measured Hg°
oxidation for each catalyst.  In general, the field test results indicate that while the initial Hg°
oxidation percentages achieved by the catalysts  matched the percentages measured in the
laboratory tests, the metal-based and some carbon-based catalysts were deactivated after a
relatively short time exposure to the actual coal  combustion flue gas. The researchers identified
sulfur trioxide and selenium (or selenium compounds) as possible flue gas constituents that
rapidly deactivate the iron-based and other metal catalysts. Additional bench-scale laboratory
tests conducted as part of the study indicate that regeneration of spent catalysts should be
possible.
                                        7-33
-------
            ESP Railing Tret
                                                      HjSAddMioa
           WS Intel
WSOuila
WS Met
Figure 7-7.  Effect of using H2S as an oxidation additive on wet FGD scrubber Hg
removal measured at AECDP test facility burning Ohio bituminous coal (source:
Reference 9).
                                  7-34
-------
             ESP Baseline Test
                                                       EDTA Addition
           WS Inlet
WS inlet
WS Outkt
Figure 7-8. Effect of using EDTA as an oxidation additive on wet FGD scrubber
Hg removal measured at AECDP test facility burning Ohio bituminous coal
(source: Reference 9).
                                  7-35
-------
Table 7-8.  Comparison of field test results using flue gas from electric utility
boiler firing Texas lignite versus bench-scale results using simulated flue gas for
selected candidate Hg oxidation catalysts (Source: adapted from Reference 24).
Test Parameters
Catalyst Type
Sand (non-catalyst blank)
Activated carbon #1 (1" Bed)
Activated carbon #1(2" Bed)
Activated carbon #2
Pd#1
SB #5 (fly ash)
Laboratory
Bench-Scale
Results
Field Test Results "
at hour
24
at hour
1,000
at hour
2,400
at hour
3,055
at hour
3,477
Percent of Hg° Oxidized Across Catalyst Bed
3%
100%
100 %
96%
91%
4/70 % '
3.3-8.1%
100%
100%
97%
90%
100%
7%
66%
81 %
not
recorded
not
recorded
36%
9-12 %
45%
42 - 59 %
76%
82%
82%
23%
0%
0%
0%
0%
73%
0%
89%
0%
76%
0%
0%
Test Conditions
Catalyst Bed Temp. "C (°F)
Inlet Hg° tug/Mm5)
Total Hg (ng/Nm3)
149 (300)
50
50
149(300) 149(300) 149(300) 104(220) 149(300)
3.7-16.2 5.4 8.3-9.3 17.8 3.7
7.0-26.1 9.8 15-27 31-35 27
   * All catalyst oxidation values corrected for the sand blank oxidation values.
   " Number of hours passing flue gas through the catalyst materials
   "Laboratory tests using SB#5 (fly ash) were conducted in a simulated flue gas with HCi (70 percent oxidation
     with 1 ppmv of HCI) and without HCI (4 percent oxidation).
                                        7-36
-------
       A pilot-scale field test program is currently being conducted under a DOE/NETL
cooperative agreement to obtain addition data on the potential commercial application of Hg
oxidation catalysts to enhance Hg capture by an existing wet FGD system downstream of high-
efficiency ESP.25  This study is testing selected catalysts previously identified as being effective
by the DOE-sponsored studies in a commercial form in larger pilot-scale units for longer periods.
The DOE/NETL is working in partnership with URS Group, Inc., EPRI, and two electric utility
companies, Great River Energy and City Public Service of San Antonio, TX. The first test site is
the Great River Energy Coal Creek Station, which fires North Dakota lignite.  The second site
the City Public Service of San Antonio's J.K.. Spruce Plant, which fires a PRB subbiruminous
coal. The pilot-scale tests will continue for over a year at each of two sites.

 7.7.3.3 Wet FGD Scrubber Design and Operating Modifications

       Several studies of pilot-scale wet FGD scrubbers suggest that modifying the operation
and design of the scrubber unit as well as the upstream ESP  may improve the capture of gaseous
Hg2+ and reduce the conversion of absorbed Hg2+ to Hg°.  Specifically, these studies have found
that the liquid-to-gas ratio and tower design of a wet FGD unit affect the absorption of gaseous
Hg2+, while the oxidation air influences the conversion of absorbed Hg2+.  The operating voltage
of ESPs upstream of wet FGD systems has also been shown to influence the latter. The
remainder of this section summarizes these findings.

       Scrubber Liquid-to-gas Ratio. The liquid-to-gas ratio (L/G ratio) of a wet FGD system is
dictated by the desired removal efficiency to control SO2 emissions.  The  selected L/G ratio also
can impact the removal efficiency of gaseous Hg2+. In general, high efficiency FGD systems
(95+ percent SO2 removal) are designed with L/G ratios in the range  of 120 to 150 gallons (gal.)
of aqueous slurry per 1,000 actual cubic feet (acf) of gas flow. In two separate pilot-scale
studies26 increasing the L/G ratio from approximately 40 to 125 gal./1,000 acf increased the
removal efficiency of gaseous Kg2"1" from 90 to 99 percent. However, increasing the L/G ratio
did not affect the removal of gaseous Hg°, which was close to zero percent. Similar studies were
conducted prior to these studies and produced similar findings.23'27

       Scrubber Tower Design. Most of the existing wet FGD systems in the United  States use
either an open-spray tower or tray tower design.  In one study of wet  FGD systems, where the
composition of the flue gas was mostly gaseous Hg2"1",  the tray tower design removed from 85 to
95 percent of the total Hg, whereas the open spray tower design removed  from 70 to 85 percent
of the total Hg.28  This study suggests that a tray tower design is more effective in removing
gaseous Hg2+ from boiler flue gas than an open spray tower  design for a given SO2 removal
level.

       Scrubber Oxidation Air. When SO2 is absorbed in the scrubbing slurry of a wet FGD
system, the dissolved SO2 reacts with lime or limestone to form insoluble sulfate/sulfite sludge;
the sulfate reaction consumes oxygen, which is present in the flue gas.  Some wet FGD systems
add air to the system to increase the amount of oxygen available for the reaction; the additional
oxygen accelerates the reaction between SOa and lime or limestone.
                                        7-37
-------
       The effect of oxidation air on FGD Hg removal was investigated as part of the AECDP
Phase III studies by conducting test runs at baseline, intermediate and low levels of oxidation
air.9 Figure 7-9 compares wet scrubber inlet and outlet Hg concentration measured for the base
case and the runs at a mid- and low-level of oxidation air. The bars include the elemental and
oxidized fractions of the total gaseous Hg. The relative amounts of Hg° at the inlet and outlet did
not change significantly for the three tests. However, the amount of absorbed Hg2+ converted to
Hg° decreased as the oxidation air decreased.  This point is further illustrated in Figure 7-10 that
shows only the gaseous Hg° for the three tests. For the baseline test, gaseous Hg° increased by
265 percent across  the wet scrubber.  This improved to a 76 percent increase for the second test,
and only two percent for the low oxidation air test.  Total gaseous-phase Hg removal improved
from 46 percent for the base case to 80 percent for the low oxidation air case. These normalized
oxidation air stoichiometry results show a strong relationship between oxidation air and wet
scrubber Hg removal for a wet FGD system. The researchers of this study hypothesize that low
oxidation air must somehow inhibit the reduction of absorbed Hg2+, or provide a species needed
to sequester the absorbed Hg2+ in the slurry. The researchers also note that the level to which the
scrubber oxidation  air can be reduced at a given coal-fired electric utility power plant is highly
site-specific specific and depends on several factors such as scaling considerations and gypsum
purity requirements.

       Voltage of ESP Upstream of Scrubber.  The effect of ESP operating power on wet
scrubber Hg removal was investigated as part of the AECDP Phase III studies.9 Concentrations
of gaseous Hg2f and gaseous Hg° were measured at the inlet and outlet of the wet FGD system
for three different ESP operating conditions. For the first operating condition (the baseline
operation), the pilot-scale ESP was operated with three of its four fields in service, and the power
was set to maintain an outlet particulate  loading of 0.02 to 0.03 Ib/MBtu (below the PM limit of
the New Source Performance Standard for utility boilers). In the second operating condition, the
ESP voltage was increased by 60 percent above the baseline voltage.  In the third operating
condition, the ESP  power was turned off and an FF was used for PM control upstream of the wet
FGD system.  For all three operating conditions, triplicate measurements of Hg were made at the
inlet and outlet of the pilot-scale wet FGD system.

       Figure 7-11 compares the concentrations of gaseous Hg2"1" and gaseous Hg° measured at
the inlet and outlet  of the wet FGD system for the three different ESP operating conditions.
Since the Hg measurements  were taken downstream of the ESP and FF, very little Hgp was
measured; thus, Hgp measurements are not shown in the Figure 7-11.  Figure 7-12 presents only
gaseous Hg° for the same three ESP conditions as those in Figure 7-11.  The figures clearly show
that the operating voltage  of the ESP has a direct, negative impact on the wet scrubber Hg
control performance. The proportion of gaseous Hg2"1" and gaseous Hg° at the wet scrubber inlet
is the same for all three tests.  However, for the high-power test, the amount of gaseous Hg°
significantly increased across the wet scrubber. The gaseous Hg° remains constant for the
no-power test, which is the observed behavior when the scrubber is preceded by the FF.  This
indicates that the electric field affects some component of the flue gas, which, in turn, has a
negative impact on wet scrubber chemistry.
                                        7-38
-------
        WS Ink*  WS GutW
WSInl«  WSGuUei
                                                      WSInkl  WSUtftfe
Figure 7-9.  Effect of oxidation air on wet FGD scrubber Hg removal as measured
at AECDP test facility burning Ohio bituminous coal (source: Reference 9).
                                  7-39
-------
    12
    10
 •ft*
     2
              Qi Air
                                  O* Air MM
                                                          O* AirL»w
                W&OuMu
WSInki  \VSOmlui
WS (nl«i  WS (kulci
Figure 7-10. Effect of oxidation air on Hg° in wet FGD scrubber flue gas as
measured at AECDP test facility burning Ohio bituminous coal (source:
Reference 9).
                                   7-40
-------
       W5 Mil
                              WS htkrl   WS Uflkl
                                                     W S kikt   WS ttflkt
Figure 7-11.  Effect of ESP operating voltage on wet FGD scrubber Hg removal as
measured at AECDP test facility burning Ohio bituminous coal (source:
Reference 9).
                                  7-41
-------
     16




     14
   i

   1
   a
•-4
S

    2



    0
                                                             Ilk HI;
                                                          fltT ws
Figure 7-12.  Effect of ESP operating voltage on Hg° in wet FGD scrubber flue gas

as measured at AECDP test facility burning Ohio bituminous coal (Source:

Reference 9).
                                  7-42
-------
7.8 Multipollutant Control Technologies

       This section presents a summary of control systems being commercially offered or
developed for multipollutant emissions control. The current status of many systems is based
upon reports that targeted one or two pollutants.  A caution here is that, when evaluating the best
system for a specific application, it is important to consider both:  1) how a given system affects
the emissions of all pollutants, and 2) how that system affects the long-term performance,
operation, and cost of other downstream systems, including ductwork, heat exchangers, stacks,
and other emission control equipment. To date no comprehensive long-term evaluations of the
multipollutant systems described below have been conducted.

7.8.1  Corona Discharge

       Generation of an intense corona discharge (ionization of air by a high voltage electrical
discharge) in the boiler flue gas upstream of an ESP and wet scrubber is being investigated with
respect to improving PM control by oxidation of a portion of the entering SO2 to SO3. 9  The
corona discharge creates oxygen-carrying reactive species, which, in turn, oxidize the Hg° in the
flue gas (i.e., convert Hg° to Hg2+). The increased SO3 both improves  ESP collection of PM and
acts to convert Hg° to Hg2+which  may then be captured by an alkaline FGD scrubber
downstream. Representative reactions for SO2 oxidation by corona discharge include:

                         02 + e- -->2O + e-
                         O2  +O ->O3
                        SO2 + O3 --> SO3 + O2
                        SO3 + H2O-->H2SO4

Similarly, for NO,

                          NO + e-  -> NO-
                       NO + NO-  ~>NO2 + N + e-
                           O2 + e- ->2O-t-e-
                           O2  +O ->O3
                         H2O + O3 -->2OH + O2
                        N02 + OH --> HNO3

       Environmental Elements Corporation is developing a process based on corona discharge
that recovers the oxidized sulfur and nitrogen compounds as marketable sulfuric and nitric acids
in wet ESP sections and or/absorbers.  A slipstream pilot plant has been installed at Alabama
Power Miller Plant (Unit 3). Initial tests indicated 80 percent Hg removal and complete
oxidation of Hg° at 10 and 20 W/cfm, respectively.

       Powerspan Corporation is developing a single, integrated pollution control device that
uses a proprietary technology called Electro-Catalytic Oxidation™ or ECO™ to control SO2,
NOx, Hg, and fine PM in coal-fired boiler flue gas.30  The first stage of the device uses a
dielectric barrier discharge to convert NOX and SO2 to acids and to oxidize Hg°. A condensing,
                                        7-43
-------
wet ESP is used to collect acid mists, fine PM, and Hg. The effluent from the wet ESP is
processed to produce salable byproducts (e.g., concentrated acids, gypsum for wallboard
manufacture, and ammonia for fertilizer). Before entering the ECO™ unit, flue gas passes
through a conventional ESP to remove the majority of the ash particles. In partnership with
FirstEnergy Corporation, Powerspan has built a pilot-scale ECO test facility at FirstEnergy's R.E.
Burger Plant near Shadyside, OH.31 This test facility processes a slipstream of flue gas from a
150-MW boiler unit burning high-sulfur eastern bituminous coal. The test results showed a Hg
emission reduction of 68 percent. Under a new DOE cooperative agreement, Powerspan and
FirstEnergy are conducting a research project using the ECO™ pilot test facility to optimize the
technology's Hg removal capability while maintaining the performance of the ECO™ unit for
removal of nitrogen oxides,  sulfur dioxide, and fine PM.32  In addition, Powerspan and
FirstEnergy are currently constructing an $ 11.9 million ECO commercial demonstration unit at
FirstEnergy's Eastlake Plant near Cleveland, OH. The project is being cofunded by a $3.5
million  grant from the  Ohio  Coal Development Office.

7.5.2 Electron Beam Irradiation

       The E-Beam Process has been offered commercially since the 1980s and is now used in
Japan and China.33 The chemical reactions are identical to  corona discharge, except that the
power source is a battery of irradiating electron "guns" and the oxidation products then enter a
semi-dry absorption system  with ammonia reagent and are converted to ammonium sulfate and
nitrate salts suitable for use as a fertilizer. It is presumed that the Hg solids would also be
present  in the fertilizer as contaminants. The polishing reactions for E-Beam are:

                      NH4OH + HNO3 --> NH4NO3 + H2O

                     2 NH4OH + H2SO4 ~> (NH4)2SO4 + 2 H2O
 7.8.3 Oxidant Injection in Flue Gas

       A number of proposed schemes would add an oxidant such as chlorine, peroxide, or
 ozone to the flue gas upstream of an absorber. Again the reaction products would be similar to
 corona or electron beam, and the recovered products could range from weak acids to
 sulfate/nitrate fertilizers or lower-value soil amendments; trace Hg salts would likely be
 contained within these products. An example of ozone injection is the Lo-TOx.34 The ISCA is a
 chlorine-based system producing byproduct acids. Hydrogen peroxide and other chlorine-based
 oxidation schemes have been investigated but have not been proposed for commercial use.35
 Typical oxidation reactions are:
                                        7-44
-------
          Hydrogen Peroxide:                       Ozone:
           H2O2 -->  2 OH                  NO + O3 »>N02 + O2
      H2O2 + OH -->  HO2 + H2O         2 NO2 +  O3 --> N2O5 + O2
       NO + OH  --> NO2 + H             SO2  + O3 --> SO3 + O2
       NO + OH  --> HNO2                N205 + H2O --> 2 HNO3
      NO + HO2 »>HNO3                SO2 + N2O5 ->SO3 +
      NO2 + OH -> HNO3

7.8.4  Catalytic Oxidation

       Catalysts can be employed in higher temperature regimes  to speed up oxidation of SO2
and NOx, but not Hg°.  However, increasing the SO3 and NO:/N2O4/N2O5 concentrations will
likely result in increased conversion of Hg° to Hg2"*" downstream, as acid gases and PM are
removed in control devices.  Lower temperature catalysis (less than 500 °F) would likely directly
oxidize Hg° to Hg2"1". Thus, any number of catalytic oxidation schemes that produce byproduct
acids would likely remove a substantial portion of total Hg with the acids as a Hg salt — chloride,
sulfate, or nitrate.  A number of catalytic technologies are under commercial development; an
example of this class - SNOx - has been evaluated under DOE's Clean Coal Technology
Program. At least one current DOE-sponsored project is examining the effectiveness of an
oxidation catalyst upstream of wet FGD scrubber to decrease total Hg emissions.36

7.8.5  Oxidant Addition to Scrubber

       One current DOE test program is measuring the effectiveness of a Hg oxidant added to
the liquor of commercial wet scrubbers. The EPA is sponsoring similar research, which will
culminate in a pilot-scale slipstream evaluation of oxidant addition.37  Another DOE-sponsored
project is investigating the use of oxidated-lime and lime-silica sorbents to a semi-dry circulating
bed absorber for combined SO2, NO\, and Hg control.38 Other combinations of sorbents injected
upstream of an efficient PM collector such as the EPRI Toxecon™ process may be used for a
multiple pollutant control strategy centered around PM control.

7. 8. 6  Catalytic Fabric Filters

       Some pilot-scale efforts have reported substantial oxidation of Hg within a FF,
presumably by catalytic action of certain fibers or residual fly ash imbedded within the fabric.39
Several investigations are being made into woven carbon fibers or other catalytic materials
integrated into the bag filters for a combined Hg/PM control device.

7.5.7  Carbon-fiber FFs and ESPs

       Carbon-fiber FFs are commercially available. Carbon-fiber ESP piates are being
investigated under a study sponsored the Ohio Coal Development Office. While combined
Hg/PM control using this approach would be initially effective, the Hg capacity would be
                                       7-45
-------
realized in a relatively short time period; therefore, means of regenerating the carbon active sites
without replacing the fabric filter bags or ESP plates have to be devised.

7.9  Summary

       A practical approach to controlling Hg emissions at existing utility plants is to minimize
capital costs by adapting or retrofitting existing equipment to capture Hg. Potential retrofit
options for control of Hg were investigated for units that currently use the following post
combustion emission control methods: (1) ESPs or FFs for control of PM, (2) dry FGD scrubbers
for control of PM and SO2, and (3) wet FGD scrubbers for the control of PM and SO2.

Hg Control Retrofits for ESP and FF

       ESPs and FFs are either cold-side or hot-side devices.  Hot-side devices are installed
upstream of the air heater while cold-side devices are installed downstream. Flue gas
temperatures in hot-side devices typically range from 350 to 450 °C while cold-side devices
typically operate at temperatures ranging from 140 to 160 °C.  Based on current information, it
appears that little Hg can be captured in hot-side ESPs or FFs.

       Least-cost retrofit options for the control of Hg emissions from units with ESP or FF are
believed to include:

       •   Injection of a sorbent upstream of the ESP or FF. Cooling of the stack gas or
           modifications to the ducting may be needed to keep sorbent requirements at
           acceptable levels.

       •   Injection of a sorbent between the ESP and a pulsejet FF retrofitted downstream of
           the ESP.  This approach will increase capital costs but reduce sorbent costs.

       •   Installation of a semi-dry CF A upstream of an existing ESP used in conjunction with
           sorbent injection.  The CFA recirculates both fly ash and sorbent to create an
           entrained bed with a large number of reaction sites. This leads to higher sorbent
           utilization and enhanced fly ash capture of Hg and other pollutants.

       Units equipped with a FF require less sorbent than units equipped with an ESP. ESP
systems depend on in-flight adsorption of Hg by entrained fly ash or sorbent particles. The FFs
obtain in-flight capture and capture as the flue gas passes through the FF.

       In general, the successful application of cost-effective sorbent injection technologies for
ESP and FF units will depend on: (1) the development of lower cost and/or higher performing
sorbents, and (2) appropriate modifications to the operating conditions or equipment being
currently used to control emission of PM, NO\, and SOz.
                                         7-46
-------
Mercury Control Retrofits for Wet FGD Scrubbers

       Wet FGD scrubbers are typically installed downstream of an ESP or FF. Wet limestone
FGD scrubbers are the most commonly used scrubbers on coal-fired electric utility boilers.
These FGD units generally capture more than 90 percent of the Hg2"1" in the flue gas entering the
scrubber. Consequently, existing wet FGD scrubbers may lower Hg emissions by about 20
percent to more than 80 percent, depending on the speciation of Hg in the inlet flue gas.

       Improvements in wet scrubber performance in capturing mercury depend primarily on the
oxidation of Hg° to Hg2+.  This may be accomplished by 1) the injection of appropriate oxidizing
agents, or 2) the installation of fixed oxidizing catalysts upstream of the scrubber to promote
oxidization of Hg° to soluble species.

       An alternative strategy for controlling Hg emissions from wet FGD scrubbers is to inject                 I
sorbents upstream of the PM control device.  In wet FGD systems equipped with ESPs,                         I
performance gains are limited by the in-flight oxidization of Hg°, and the in-flight capture of
Hg2+ and Hg°. In systems equipped with FFs, increased oxidization and capture of Hg can be
achieved as the flue gas flows through the FF. Increased oxidization of Hg° in the FF will result
in increased Hg removal in the downstream scrubber.

Mercury Control Retrofits for Semi-dry FGD Systems

       SDA systems that use calcium-based sorbents are the most common dry FGD systems
used in the utility industry. An aqueous slurry containing the sorbent is sprayed into an absorber
vessel where the flue gas reacts with the drying slurry droplets. The resulting, particle-laden, dry
flue gas then flows to an ESP or a FF where fly ash and SOj reaction products are collected.

       CFAs are "vertical duct absorbers" that allow simultaneous gas cooling, sorbent injection
and recycle, and gas absorption by flash drying of wet lime reagents. It is believed that CFAs
can potentially control Hg emissions at costs lower than those associated with use of spray
dryers.

       Dry FGD systems are already equipped to control emissions of SOi and PM. The
modification of these units by the use of appropriate sorbents for the capture of Hg and other air
toxics is considered to be the easiest retrofit problem to solve.


7.10 References
1.   Smit, F. J., G.L. Shields,  and C.J. Mahesh. "Reduction of Toxic Trace Elements in Coal By
    Advanced Cleaning." Presented at the Thirteenth Annual International Pittsburgh Coal
    Conference, September 3-7,1996.
                                        7-47
-------
2.  Topical Report No. 5 Trace Element Removal Study." Prepared for U.S. Department of
   Energy's Pittsburgh Technology Center by ICF Kaiser Engineers, Fairfax, VA. March 1995.

3.  Brown, T. D., D,N. Smith, R.A. Hargis, Jr., and W.J. O'Dowd.  "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.

4.  Nebel, K. L., D.M. White, W.H. Stevenson, and M.G. Johnston. A Summary of Mercury
   Emissions and Applicable Control Technologies for Municipal Solid Waste Combustors.
   U.S. EPA, Office of Air Quality Planning and Standards, Research Triangle Park, NC.
   September 1991.

5.  Getz, N. P., B.T. Ian, and  C.K. Amos.  "Demonstrated and Innovative Control Technologies
   for Lead, Cadmium and Mercury for Municipal Waste Combustors," Proceedings of the Air
   & Waste Management Association 85th Annual Meeting and Exhibition, Kansas City, MO.
   1992.

6.  Brown, B., and K. Felsvang. "Control of Mercury and Dioxin Emissions from United States
   and European Municipal Waste Incinerators by Spray Dryer Absorption Systems," in
   Proceedings of the Municipal Waste Combustion International Specialty Conference, Air
   and Waste Management Association, VIP-19, Tampa, FL, pp 685-705, April 1991.

7.  Babcock & Wilcox Alliance Research Center.  Advanced Emissions Control Development
   Program Phase I - Approved Final Report prepared for the U.S. Department of Energy
   (U.S. DOE-FETC contract DE-FC22-94PC94251) and Ohio Coal Development Office
   (grant agreement CDO/D-922-13), July 1996.

8.  McDermott Technologies, Inc. Advanced Emissions Control Development Program
   Phase II - Approved Final Report, prepared for the U.S. Department of Energy (U.S. DOE-
   FETC contract DE-FC22-94PC94251) and Ohio Coal Development Office (grant agreement
   CDO/D-922-13), RDD:98:43509-500-200:01R, April 1998. Available at:
   .

9.  McDermott Technologies, Inc. Advanced Emissions Control Development Program
   Phase III - Approved Final Report, prepared for the U.S. Department of Energy (U.S. DOE-
   FETC contract DE-FC22-94PC94251—22) and Ohio Coal Development Office (grant
   agreement CDO/D-922-13). July 1999. Available at:
   < http://www.osti.gov/dublincore/servlets/puriy756595-LACvcL/webviewabIe/756595.pdf>.

10. Grover, C., J.  Butz, S. Haythornthwaite, J. Smith, M. Fox, T. Hunt, R. Chang, T. Brown, and
   E. Prestbo. "Mercury Measurements Across Particulate Collectors of PSCO Coal-fired
   Electric Utility Boilers," EPRI/DOE/EPA Mega-Symposium, Atlanta, GA. August 1999.
                                      7-48
-------
11. Hargis, R. A., WJ. O'Dowd, and H.W. Pennline. "Sorbent Injection for Mercury Removal
   in a Pilot-scale Coal Combustion Unit," presented at the 93th Annual Meeting & Exhibition
   of the Air & Waste Management Association, Salt Lake City, UT. June 2000.

12. U.S. Department of Energy, National Energy Technology Laboratory. In-House Research
   on Mercury Measurement and Control at NETL. Pittsburgh, PA . November 2001. Available
   at: .

13. Waugh, E.G., B.K. Jensen, L.N. Lapatnick, F.X. Gibbons, S. Sjostrom, J. Ruhl, R. Slye, and
   R. Chang. "Mercury control in utility ESPs and baghouses through dry carbon-based sorbent
   injection pilot-scale demonstration," In Proceedings of the EPRI/DOE/EPA Combined
   Utility Air Pollutant Control Symposium, EPRITR-108683-V3; Washington, DC, August
   25-29, 1997.

14. Durham, M.D, C.J. Bustard, R. Schlager, C. Martin, S. Johnson, and S. Renninger. "Field
   Test Program to Develop Comprehensive Design, Operating and Cost Data for Mercury
   Control Systems on Non-Scrubbed Coal-Fired Boilers," presented at the Air & Waste
   Management Association 2001 Annual Conference and Exhibition, Orlando, FL. June 24-28,
   2001.

15. Bustard, C. J., M. Durham, C. Lindsey, T. Starns, K. Baldrey, C. Martin, S. Sjostrom, R.
   Slye, S. Renninger, and L. Monroe, "Full-Scale Evaluation of Mercury Control with Sorbent
   Injection and COHPAC at Alabama Power E.C. Gaston," presented at the A&WMA
   Specialty Conference on Mercury Emissions: Fate, Effects, and Control and the U.S.
   EPA/DOE/EPRI Combined Power Plant Air Pollutant Control Symposium: The Mega
   Symposium, Chicago, IL. August 20-23, 2001.

16. Madden, D.A., and M.J. Holmes. "B&W's E-LIDS TM Process - Advanced SOx,
   Particulate, and Air Toxics Control for the Year 2000," presented at the 1998 EPRI-DOE-
   EPA Combined Utility Air Pollutant Control Symposium, Washington, DC. August 25-29,
   1997.

17. Sjostrom, S., J. Smith, T. Hunt, R. Chang, and T. D. Brown. "Demonstration of Dry Carbon-
   Based Sorbent Injection for Mercury Control in Utility ESPs and Baghouses."  Presented at
   the Air & Waste Management Association's 90th Annual Meeting & Exhibition, Toronto,
   Ontario, Canada. June 8-13,1997.

18. Haythornthwaite, S., S. Sjostrom, T. Ebner, J. Ruhl, R. Slye, J. Smith, T. Hunt, R. Chang,
   and T.D. Brown. "Demonstration of Dry Carbon-Based Sorbent Injection for Mercury
   Control in Utility ESPs and FFs," in Proceedings of the EPRI/DOE/EPA Combined Utility
   Air Pollutant Control Symposium; Washington, DC; EPRI TR-108683-V3. August 25-29,
   1997.
                                      7-49
-------
19. Roberts, D.L., J. Albiston, T, Broderick, C. Greenwell, and R. Stewart. Novel Process for
   Removal and Recovery of Vapor Phase Mercury, Phase I Final Report under Contract DE-
   AC22-95PC95257 to DOE Federal Energy Technology Center, Pittsburgh, PA. September
    1997.

20. Turchi, C.S., J. Albiston, I.E. Broderick, and R.M. Stewart. "Removal of Mercury from
   Coal-Combustion Flue Gas Using Regenerable Sorbents," presented at the 92nd Annual
   Meeting of the Air & Waste Management Association, St. Louis, MO. June 1999.

21. ARCADIS Geraghty & Miller. Roanoke Valley Energy Facility Mercury Testing. Research
   Triangle Park, NC.  November 6, 2000.

22. U.S. Department of Energy, National Energy Technology Laboratory. "Full-Scale Testing of
   Enhanced Mercury Control in Wet FGD,"  November 2001. Available at
   .

23. Hargrove, O.W., Jr., T.R. Carey, C.F. Richardson, R.C. Skarupa, F.B. Meserole, R.G. Rhudy,
   and T.D. Brown. "Factors Affecting Control of Mercury by Wet FGD," Presented at the
   EPRI/DOE/EPA Combined Utility Air Pollutant Control Symposium, Washington, DC.
   August 1997.

24. Blythe, G.M, T.R.Carey, C.F. Richardson , F.B. Meserole, R.G. Rhudy,  and T.D. Brown.
   "Enhanced Control of Mercury by Wet Flue Gas Desulfurization Systems," Presented at the
   92nd Annual Meeting & Exhibition of the Air & Waste Management Association, St. Louis,
   MO. June 1999.

25. U.S. Department of Energy, National Energy Technology Laboratory. "Pilot Testing of
   Mercury Oxidation Catalysts," Pittsburgh, PA. November 2001.  Available at:
   < http://www. fetc.doc. gov/coalpower/environment/mercurv/index.html >.

26. Redinger, K. E., A. P. Evans, R. T. Bailey, and P. S. Nolan. "Mercury Emissions Control in
   FGD Systems," presented at the EPRI/DOE/EPA Combined Air Pollutant Control
   Symposium, Washington, DC. August 25-29, 1997.

27. Hargrove, O.W., Jr., J.R. Peterson, D.M. Seeger, R.C. Skarupa, and R.E. Moser.  "Update of
   EPRI Wet FGD Pilot-Scale Mercury Emissions Control Research," presented at the
   EPRI/DOE International Conference on Managing Hazardous and Particulate Pollutants,
   Toronto, Canada. August 15-17,1995.

28. Electric Power Research Institute. Electric Utility Trace Substances Synthesis Report -
    Volume 3: Appendix O, Mercury in the Environment. EPRI TR-104614-V3, Project
   3081,3297. November 1994.
                                      7-50
-------
29. Helfritch, D.J., and P.L. Feldman. "Flue Gas Mercury Control by Means of Corona
    Discharge," Paper 99-157, Air & Waste Management Association 92nd Annual Meeting,
    St. Louis, MO. June 20-24, 1999.

30. McLarnon, C. R., M. L. Horvath, and P. D. Boyle. "Electro-Catalytic Oxidation Technology
    Applied to Mercury and Trace Elements Removal from Flue Gas," presented at Conference
    on Air Quality II, McLean, VA. September 20, 2000.

31. McLarnon, C. R, and M. D. Jones. "Electro-Catalytic Oxidation Process for Multi-Pollutant
    Control at FirstEnergy's R.E. Burger Generating Station," presented at Electric Power 2000,
    Cincinnati,  OH. April  5, 2000.

32. U.S. Department of Energy, National Energy Technology Laboratory. "Non-thermal Plasma
    Based Removal of Mercury,"  November 2001. Available at
    .

33. Hirona, S. "Simultaneous SO2, SOj and NOX Removal by Commercial Application of the
    EBA Process," presented at the EPRI/DOE/EPA Combined Utility Air Pollution Control
    Symposium, Atlanta, GA, EPRITR-113187-V2, pp 8-1 through 8-14. August 1999.

34. Anderson, M.H., A.P. Skelley, E. Goren, and J. Cavello.  "A Low Temperature Oxidation
    System for the Control of NO\ Emissions Using Ozone Injection," presented at the Institute
    of Clean Air Companies Forum 98: Cutting NOx Emissions, Durham, NC. March 18-20,
    1998.

35. Livengood, C.D., and M.H. Mendelsohn. "Process for Combined Control of Mercury and
   Nitric Oxide," presented at the EPRI/DOE/EPA Combined Utility Air Pollution Control
    Symposium, Atlanta, GA, EPRI TR-113I87-V2, pp 19-30 through 19-41. August 1999.

36. Richardson, C.F., G.M. Blythe, T.R. Carey, R.G. Rhudy, and T.D. Brown. "Enhanced
    Control of Mercury by Wet FGD Systems," EPRI/DOE/EPA Combined Utility Air Pollution
    Control Symposium, Atlanta, Georgia, EPRI TR-113187-V3, pp 20-41 through 20-54,
    August 1999.

37. Roy, S., and G.T. Rochelle. Chlorine Absorption in S (IV) Solutions. EPA-600/R-01-054
    (NTIS PB2001-107826), National Risk Management Research Laboratory, Research
    Triangle Park, NC . August 2001.

38. Ghorishi, S.B., C.F. Singer,  W.S. Jozewicz, R.K. Srivastava, and C.B. Sedman.
    "Simultaneous Control of Hg°, SOa, and NO\ by Novel Oxidized Calcium-Based Sorbents,"
    Paper # 243, presented at the 94th AWMA Annual Meeting, Orlando, FL. June 2001.

                                      7-51
-------
39.  McManus, T.J., R.O. Agbede, and R.P. Khosah. "Conversion of Elemental Mercury to the
    Oxidized Form in a Baghouse," Paper 98-WP79A.07, presented at the A&WMA 91st
    Annual Meeting, San Diego, CA. June 14-18, 1998.
                                      7-52
-------
                                       Chapter 8
                 Cost Evaluation of Retrofit Mercury Controls for
                          Coal-fired Electric Utility Boilers
8.1 Introduction

       A practical approach to controlling Hg emissions at existing coal-fired electric utility
power plants is to minimize control costs by adapting or retrofitting existing air pollution control
equipment to capture Hg. As discussed in Chapter 3, coal-fired electric utility power plants
currently use a wide variety of technologies to control the emission of criteria air pollutants (e.g.,
PM, SC>2, and NOx emissions). Generally, the air pollution control methods and configurations
used for a given coal-fired electric utility boiler depend on the type of coal burned, age and size
of the boiler unit, and the power plant location.

       Potential retrofit technologies for the control of Hg emissions from existing coal-fired
electric utility boilers are  discussed in Chapter 7. Control technologies using injection of
powdered activated carbon (PAC) into the flue gas have been applied successfully on municipal
waste combustors to reduce Hg emissions. Pilot-scale testing indicates that these technologies
offer the potential to provide significant Hg removal from the flue gas of coal-fired electric utility
boilers. This chapter discusses an initial evaluation of annual Hg control costs based on the
retrofit of PAC injection-based control technologies to a series of model plant scenarios (not
actual full-scale applications) representative of the coal-fired electric utility power plants
operating in the United States.  It is worth noting that, while performance and cost of only PAC-
related technologies were evaluated, other non-PAC-based Hg control technologies are expected
to be available in the future. For example, enhanced Hg oxidation using oxidants or catalysts
followed by wet scrubbing may become available.  Also, the role of an SCR-FGD combination
may become more cost effective and attractive. The information presented in this chapter was
used in the EPA's recent  regulatory determination regarding Hg and other air toxics.

       The cost estimates of the PAC injection-based Hg control technologies presented in this
chapter are based on relatively few data points from pilot-scale tests  and, therefore, are
considered to be preliminary estimates. As discussed in Section 8.2, factors that are known to
affect adsorption of Hg on activated carbon include speciation of Hg in the flue gas, flue gas and
ash characteristics, and the degree of mixing between the flue gas and activated carbon. The
effects of these factors may not be entirely accounted for in the relatively few pilot-scale data
points available for this evaluation. Successful testing of a control approach at small pilot plants
                                         8-1
-------
does not necessarily guarantee successful implementation of the approach in full-scale systems.
Temporary wall effects at small scale will generally not be realized at full scale.  Appropriate
mass transfer associated with mixing and the number, placement, and design of reagent and
sorbent injection equipment may also need to be determined.  Further, potential longer-term
problems such as deposits, fouling, and corrosion of the control equipment are frequently not
addressed by pilot-scale tests because of shorter-term, non-continuous operation. Ongoing
research is expected to address these issues to improve the potential of using sorbents for Hg
control in coal-fired boilers.

       Coal-fired electric utility power plants are currently required to reduce emissions of NOx,
SO2, and PM.  The EPA has also revised the National Ambient Air Quality Standards (NAAQS)
for PM and ozone. These revisions may require electric utility sources to adopt control measures
aimed at reducing concentrations of fine PM in the atmosphere. In addition, as discussed above,
the EPA has recently expressed its intent to regulate Hg emissions from these sources. Adding to
these environmental requirements and activities, Congress is introducing bills aimed at
developing legislation requiring simultaneous reductions  in emissions of multiple emissions.
Improved sorbents and other methods for controlling Hg and multipollutant (e.g., Hg and NOx)
emissions are also under development by DOE, EPA, EPRI, the electric industry, and equipment
vendors.  These development activities include large demonstration programs that are underway
under the sponsorship of DOE/NETL and industrial participants. The demonstrations are
focused on full-scale testing of powdered activated carbon injection and modifications to flue gas
cleaning systems aimed at improving Hg capture.

       It is expected that, when the research and development activities being conducted by
DOE, EPA, EPRI, and others are completed, there will be many more control options for Hg and
multipollutants with attendant benefits in improved cost effectiveness.
8.2 Cost Estimate Methodology

       The methodology used for the Hg control cost evaluation consists of the following six
steps:
          First step, a set of model plant and Hg control scenarios was defined;
          Second step, cost estimates were made for selected scenarios using a cost model
          developed collaboratively by the DOE and the EPA;
          Third step, the cost impacts of selected variables were examined;
          Fourth step, the cost model results were used to develop indications of costs for those
          model plant scenarios for which data on PAC use are currently not available;
          Fifth step, potential future  improvements in the cost estimates were examined; and
          Sixth step, in order to place Hg control costs in perspective, these costs were
          compared to current costs of applying NOx controls  to coal-fired electric utility
          boilers.
                                         8-2
-------
8.2.1 Mercury Control Technologies Evaluated

       The cost evaluation is based primarily on the application of potential PAC injection-based
control technologies. These technologies were selected because sufficient pilot-scale data are
available to make reasonable estimates of the Hg capture efficiency of the technologies. Mercury
capture performance data are currently not available for other potential Hg control technologies
(e.g., use of catalysts to oxidize Hg° in wet scrubber systems) that conceivably could be applied
to coal-fired electric utility boilers at this time. Table 8-1 lists the PAC injection-based Hg control
technologies defined for this study. Pilot-scale applications of most of these technologies have
been reported in published literature.1'2'3'4'5'6

       PAC injection-based retrofit control technologies ESP-1, ESP-3, ESP-4, ESP-6, and
ESP-7 are applicable to coal-fired electric utility boilers equipped with a cold-side ESP.

       In ESP-1, PAC is injected between the air preheater and the cold-side ESP (CS-ESP, i.e.,
an ESP located downstream of the boiler's air preheater). This configuration is the simplest to                   I
install, requiring only PAC injection equipment upstream of the ESP. Activated carbon                         I
consumption is expected to be relatively high because the high temperature of the flue gas would
inhibit adsorption of Hg onto PAC.

       In ESP-3, PAC is injected downstream of the CS-ESP and is collected using a polishing
fabric filter (PFF). This technology permits recycling of the PAC sorbent to increase its
utilization. Typically, this recycling is  achieved by transferring a portion of used sorbent from
the PM control device (e.g., PFF) to the sorbent injection location using a chain or a belt
conveyor, mixing the used sorbent with fresh sorbent, and injecting the resulting sorbent mixture
into the flue gas. Further, the technology provides a contact bed (i.e., filter cake on PFF) for
increased adsorption of Hg.

       ESP-4 is similar to ESP-1, but adds spray cooling (SC) upstream of the PAC  injection
location.  Cooling the flue gas aids adsorption and reduces  PAC injection requirements.
However, adding too much water to the flue gas could cause acid condensation, which  would
corrode ductwork and equipment.  In the cost modeling conducted for this work, flue-gas
temperatures are not allowed to reach the acid dewpoint (i.e., the temperature at which  the acidic
components in the flue gas would condense).

       ESP-6 is similar to ESP-3, but provides SC upstream of PAC injection. Cooling the flue
gas aids adsorption and reduces PAC injection requirements. Also, use of PFF permits sorbent
recycling, leading to improved sorbent utilization.

       ESP-7 is the same as ESP-6 except for the addition of a second sorbent, lime. In addition
to Hg removal, this technology would remove acid gases from the flue gas. Pilot-scale results
have indicated that this may result in significant lowering of PAC injection rates.
                                         8-3
-------
Table 8-1. Mercury control technologies.
Existing Post-combustion
Control Devices
Used for
Coal-fired Boiler Unit"
CS-ESP
HS-ESP
FF
SDA + FF
SDA + CS-ESP
Mercury Control Technologies "
Identification
Code
ESP-1
ESP-3
ESP-4
ESP-6
ESP-7
HESP-1
FF-1
FF-2
SD/FF-1
SD/ESP-1
Additional Control Equipment Installed
PAC injection
PAC injection + PFF
SC + PAC injection
SC + PAC injection + PFF
SC + PAC injection + lime injection + PFF
SC + PAC injection + PFF
PAC injection
SC + PAC injection
PAC injection
PAC injection
 (a) Existing controls may include wet FGD scrubber system or post-combustion NOX controls such as selective
    catalytic reduction (SCR) and selective noncatalytic reduction (SNCR).
 (b) CS-ESP = cold-side electrostatic precipitator
    HS-ESP = hot-side electrostatic precipitator
    FF = fabric filter
    PAC = powdered activated carbon
    PFF = polishing fabric filter
    SC = spray cooling
    SDA = spray dryer adsorber system
                                          8-4
-------
       In HESP-1, SC, PAC injection, and a PFF are added downstream from a hot-side ESP (an
ESP located upstream of the boiler's air preheater).  This configuration is identical to ESP-6, only
the location of the ESP is different.

       Two PAC injection-based retrofit controls are applicable to coal-fired electric utility
boilers equipped with a fabric filter. FF-1  is the fabric filter analogue of ESP-1. However, Hg
collection should be better than that in ESP-1 because the FF provides added residence time and
a contact bed (filter cake on the bags) for increased adsorption of Hg. FF-2 is the fabric filter
analogue of ESP-4; spray cooling and PAC injection are installed upstream of an existing fabric
filter. As with ESP-4, cooling reduces PAC requirements, which reduces total annual PAC costs
for FF-2 compared to FF-1.

       Finally, use of a PAC injection in combination with an existing spray dryer adsorber
system for SOa control was evaluated. In SD/FF-1, PAC is injected into the flue gas of a boiler
that uses  a SDA + FF combination.  In this configuration, only PAC injection equipment is added
to the existing air pollution control system, with the SDA providing flue gas cooling.  SD/ESP-1
is similar to SD/FF-1 except that an ESP is used in place of an FF for particulate collection. The
advantages are similar to those of SD/FF-1; however, larger amounts of PAC may be needed to
achieve performance levels comparable to those achieved by SD/FF-1.

8.2.2 Model Plant Descriptions

       Costs for installing and operating the Hg control technologies described in Table 8-1 are
estimated by combining these control configurations with appropriate model plant descriptions
representing plants firing different types of coal on varying boiler sizes. Eighteen different
model plant descriptions or "scenarios" were defined for the cost evaluation. Table 8-2 lists these
scenarios.

       Approximately 75 percent of the existing coal-fired electric utility boilers in the United
States are equipped with an ESP for the control of PM.7 The remaining boilers employ fabric
filters, particulate scrubbers, or other equipment for control of PM. Additionally, units firing
medium-to-high  sulfur coals may use FGD technologies to meet their SC"2 control requirements.
Generally, larger units firing high-sulfur coals employ wet FGD, and smaller units firing
medium-sulfur coals use SDAs. While developing the model plant scenarios, these PM and SOi
control possibilities were taken into account. It may be worth noting that, since the majority of
boilers use  an ESP for PM control, most Hg control technology applications would likely take
place on such boilers and would reflect pertinent performance and costs.

       The two coal-fired boiler sizes (expressed as gross electricity output), used for the model
plant scenarios listed in Table 8-2, were  selected to approximately span the range of typical
electric utility boiler sizes, and to be consistent with the model plant sizes used in previous cost
studies.1  It was also envisioned that the use of post-combustion NOx controls (i.e., SCR or
SNCR) may enhance oxidation of Hg in flue gas and result in the "cobenefit" of
                                         8-5
-------
c
*
6<
u
Applicable
rcury Control
W
S
§
to
Post-combu
0,1
1
1
CO


J
g ;
® .
i

f^ ^


8g
IA
3 s

Keirorii
nfiguration(s)
ee Table 8-1)
o «
o —


(0
fl


Sulfur content

h-


5 =
5S-

u
o
e k'
n c
'1
IT

CO

CO
LU
d
CO
LU


n
SP + wet FG
UJ
CO
o
s?
co

o
c
I
m

ID
2
in
N.



CO





CD
LU
Q.
CO
LU



CS-ESP


sS
CO
d
CO
3
O
C
I
s?

0)

K



.q-





LL.
LL




LL
LL


CD
d
§
m


|
in



in





0.
CO
LU
I




CL
CO
LU
CO


CO
d
1
|
ha


|
in



CD





2
CO
LU
f
0.
CO
LU



CS-ESP


in
d
V)
3
3
CO

1
in
fe



j>^





LL




U_


in
d
1
15
CO
CD
|
fe



oo





D.
CO
HI
I




Q.
CO
LU
CO
X


in
d
minous
3
-Q
CO
0)
1
in
fe



en





SD/ESP-1




A + CS-ESP
Q
CO
as
oo
1
E
m

CO

8



o





SD/FF-1




LL
li.
f/1


ss
co
(0
E
CD

V

§



*-





d.
CO
LU



n
SP + wet FG
UJ
X
CO
(O
E
m

CO

§



CM





Q.
CO
LU
I
UJ



CS-ESP


CO
en
0
S

«
|
8



CO





LL
LL




t


se
CO
d
0
E
im

ID
5
8



t





CO
LU
X




UJ
X


CD
d
S
0
.ti
m


|
8



in





2
CO
LU
2

Ul



Q.
CO
LU


S?
in
d
minous
3
1
CO
CO

8



CD





U-




Ll.
LL


in
d
minous
3
1
CO

§
8



!*•





d
CO
LU




HS-ESP


as
in
d
minous
i
3
CO
-------
increased Hg removal in wet FGD systems. This is especially relevant since many SCR
applications are expected to take place in the next few years and, in response to SO2 reduction
requirements, more wet FGD systems may be installed.  However, at the time of this study, some
data on this co-benefit were available for SCR applications only. Since SCR is a capital-
intensive technology, generally its use is more cost-effective for larger boilers. Accordingly, in
this work, the Hg co-benefit resulting from SCR use was evaluated for model plant scenarios 1,
2, and 3, utilizing large (975 MWe) boilers and wet FGD.

8,2.3  Computer Cost Model

       The DOE/NETL developed a cost model for estimating the costs of Hg control options
for coal-fired electric utility boilers. This cost model, called the NETL Mercury Control Cost
Model, can provide capital and operating and maintenance (O&M) costs estimated in year 2000
constant dollars for the application of selected Hg control configurations to coal-fired electric
utility boilers.  The model has been used for other studies conducted to characterize  the costs
associated with using PAC injection on coal-fired electric utility boilers.8  For this evaluation, the
EPA collaborated with the DOE to modify this cost model to incorporate the PAC injection rate
algorithms described in the following section. An overview of the modified version of the NETL
Mercury Control Cost Model used for this cost evaluation is presented in Appendix  D to this
report. This model is hereafter referred to simply as the cost model.

8.2.4  PAC Injection Rate Algorithms

       The current understanding is that Hgp is well collected in PM or SOz control systems, Hg°
is not so well collected, and Hg2+ is collected to a greater or lesser degree depending on
characteristics of the control device and conditions within it. Therefore, for a specified Hg
removal requirement, the rate of PAC injection needed will depend, in part, on the ability of
existing controls to remove the three  forms of Hg. The major factor affecting the cost of PAC
injection-based technologies is the rate of PAC injection needed for the required Hg removal
efficiency. In general, this rate depends on the time of contact between carbon particle and flue
gas, the properties of the carbon (particle size, micropore surface area, pore size distribution, and
Hg adsorption capacity), the temperature of the flue gas, and the type of coal-fired in the boiler.
For this work,  PAC injection rates at specific flue gas temperatures and Hg removal efficiencies
achieved in pilot-scale tests were  fitted to the form of Equation (8-1) with curve-fit parameters a,
b, and c (see Attachment 2 in Appendix D). For each technology for which pilot-scale test data
are available, separate correlations of Hg removal efficiency and PAC injection rate were
determined for bituminous and subbituminous coals. These coals are predominantly used at
electric utility boilers and, therefore, were chosen for this work.


  Mercury Removal Efficiency (%) = 100	(Eq. 8-1)
                                         \PAC  Injection Rate (lb/lQ6acf}+b\ C
                                         8-7
-------
       Equation 8-1 can be used to calculate the PAC injection rate (lb/106 acf) needed to
achieve a specified Hg removal efficiency (percent) for the control technology of interest. Note
that Hg removal efficiency (percent) is based on total Hg (the sum of Hg°, Hg2+, and Hgp)
removed from the flue gas and is defined as
   Mercury Removal Efficiency (%) = lOOx
                                          (Emissionin — Emissiono
                                                 Emission,*
                                                                              (Eq.8-2)
    where:     Emission;,, = total flue gas Hg concentration at the inlet to the first air pollution
              control device; and
              Emissionout = total flue gas Hg concentration at the outlet of the last air pollution
              control device.

       Preliminary analysis of the Pat III EPA ICR data 9 reflected that, at boilers firing
bituminous coals and using a CS-ESP for PM capture, higher levels (more than 50 percent) of Hg
were being removed with fly ash than were found in earlier pilot-scale tests (see Attachment 2 in
Appendix D).  Accordingly, for each of technologies ESP-1, ESP-3, ESP-4, and ESP-6, two
separate sets of correlations, relating PAC injection rate (lb/106 acf) to Hg removal efficiency
(percent), were created for use with bituminous-coal-fired boilers.  The first of these sets,
hereafter referred to as the pilot-scale PAC injection rate, was derived using presently available
pilot-scale test data. The other set, hereafter referred to as the ICR/pilot-scale PAC injection rate,
was derived using preliminary ICR results for fly ash capture of Hg (i.e., no PAC injection) and
pilot-scale results for PAC injection.

       Note that the above data-fitting procedure resulted in correlations of PAC injection rate
(lb/106 acf) versus Hg removal efficiency (percent), as a function of flue gas temperature, for all
of the technologies except:  (1) FF-1, FF-2, and SD/FF-1, applied on boilers firing bituminous
coals, for which no data are available; (2) HESP-1, applied on boilers firing either bituminous or
subbituminous coals, for which no data are available; and (3) ESP-7, applied on boilers firing
either bituminous or subbituminous coals. The only available data on ESP-7 are from a pilot-
scale application on a boiler firing a bituminous coal.10 Since these data reflect that more than 90
percent of the Hg can be removed by injecting relatively  small amounts of PAC with lime, in this
work, application of ESP-7 was evaluated at 90 percent Hg removal efficiency in a sensitivity
analysis.

       The algorithms describing sorbent injection rates  for various technologies can be found in
Attachment 2 in Appendix D. The PAC injection rate algorithms could not be determined for the
retrofit configurations defined for model plant scenarios 2, 3, 5, 6, 9, 11, 12,14, 15, and 18.  As
such, costs  for these model plant configurations cannot be estimated using the cost model.
                                         8-8
-------
8.2.5  Cost Estimate Assumptions

       To estimate the costs for the model plant configurations using the cost model, the
following specifications were used.

   (1) Mercury concentration in the flue gas for each model plant scenario is 10 (ig/Nm3. This
       concentration has been used in previous cost studies1'8 and is in the range of mean
       concentrations (1.7-50.1 p.g/dscm) determined from ICR data for pulverized-coal-fired
       electric utility boilers equipped with different air pollution controls.9  Note also that the
       corresponding median and mean concentrations are 9.1 and 11.4 (ig/dscm, respectively.

   (2) For each of retrofit configurations ESP-1, ESP-3, ESP-4,  and ESP-6, two separate sets of
       correlations, relating PAC injection rate (lb/106 acf) to Hg removal efficiency (percent),
       were created for use with bituminous-coal-fired boilers. The first of these sets, hereafter
       referred to as the pilot-scale PAC injection rate, was derived using presently available
       pilot-scale test data.  The other set, hereafter referred to as the ICR/pilot-scale PAC
       injection rate, was derived using preliminary EPA ICR results for fly ash capture of Hg
       (i.e., no PAC injection) and pilot-scale results for PAC injection. Accordingly, two sets
       of cost estimates for applying retrofit configurations ESP-1, ESP-3, ESP-4, and ESP-6
       were made: one estimate used the pilot-scale PAC injection rate, and the other used the
       ICR/pilot-scale PAC injection rate.

   (3) PAC injection rate correlations generally reflect that PAC  injection requirements increase
       nonlinearly with increases in Hg removal efficiency. To characterize the impact of this
       behavior, wherever possible, model plant costs were estimated for Hg removal
       efficiencies of 60, 70, 80, and 90 percent.

   (4) In general, for any given Hg removal requirement, the PAC injection rate decreases if the
       temperature of the flue gas  is lowered.  For this reason, the flue gas is cooled by water
       injection in some of the retrofit configurations (see Table 8-1). However, injecting water
       into an acidic flue gas can lead potentially to corrosion of downstream equipment. To
       avoid this corrosion, an approach to acid dew point (ADP) of 18 °F was used for the
       retrofit configurations with spray cooling (i.e., ESP-4, ESP-6, ESP-7, and FF-2).11 For
       these retrofit configurations, the extent  of SC provided was determined based on the
       temperature of the flue gas  before cooling and the temperature nearest to the above
       approach to ADP for which a PAC injection rate correlation was available. Note that, in
       the high-sulfur coal applications  with relatively high ADPs, this constraint resulted in no
       SC if the SC>2 control technology was wet FGD. However, in applications using SDAs
       for SO2 control, SC is inherent and acid gases are removed prior to PAC injection;
       therefore, this constraint was not applied.

   (5) No data are currently available for recycling of sorbent in technology applications
       utilizing PAC injection and PFF.  Accordingly, no sorbent recycle was used in retrofit
       configurations ESP-3 and ESP-6.
                                         8-9
-------
   (6) Mercury speciation in the flue gas from bituminous-coal-fired boilers is assumed to be 70
       percent of the total Hg being oxidized, with 30 percent being Hg°.  The corresponding
       assumption for boilers firing subbituminous coals is 25 percent oxidized with 75 percent
       Hg°. These Hg speciation percentages were determined from a preliminary analysis of
       ICR data (see Attachment 2 in Appendix D).

   (7) Wet FGD systems are assumed to remove 100 percent of Hg2+ and no Hg°.  This is based
       on the fact that mercuric chloride (the assumed major oxidized species) is soluble in
       water, while Hg° is  insoluble. It is anticipated that ongoing research on wet scrubbers
       will result in improved performance through the use of reagents or catalysts to convert Hg
       to chemical compounds that are soluble in aqueous-based scrubbers.

   (8)  Use of SCR is assumed to increase Hg2+ content in flue gas by 35 percent for both
       bituminous- and subbituminous-coal-fired boilers. This increase in mercury oxidation
       was determined from a preliminary analysis of ICR data as follows. As explained above,
       oxidized mercury content in flue gas from bituminous-coal-fired boilers is assumed to be
       70 percent.  Also, ICR data revealed that SCR application with SDA at one plant firing
       bituminous  coal resulted in greater than 95 percent mercury removal. It is hypothesized
       that virtually all of the mercury removed at this plant was oxidized mercury.  Based on
       these considerations, it is assumed that SCR increases oxidized mercury content by
       35 percent (also see Attachment 2 in Appendix D). Currently, research and development
       efforts are underway to investigate the effects of SCR on Hg oxidation. A more mature
       set of findings regarding SCR impacts are expected from these efforts.

   (9) For each of the model plant scenarios, a plant capacity factor of 65 percent was used.
                                                        12
    (10) The cost of PAC is assumed to be $1.00 per kilogram.

    Other specifications are described in Attachments  1, 2, and 3 in Appendix D.
8.3 Estimated Costs of Reducing Mercury Emissions

       This section describes the estimates of total annual cost determined using the cost model
for application of Hg controls to those model plant scenarios for which PAC injection rate
algorithms could be determined (i.e., model plant scenarios 1, 4, 7, 8, 10, 13, 16, and 17). It is
important to note that cost estimates presented in this section are based on currently available
data and, as explained later, may be improved with R&D efforts and as long-term operating data
from full-scale demonstrations become available.

       In general, capital costs of PAC injection-based Hg control technologies comprise a
relatively minor fraction of the total annual costs of these technologies; the major fraction is
associated with the costs related to the use of PAC.12 As an example, for application of SC+PAC
injection (ESP-4) to achieve 80 percent Hg reduction on a  975-MWe boiler firing bituminous
                                        8-10
-------
coal and using an ESP, the capital cost contributes about 23 percent of the total annual cost.
Therefore, for such technologies, the cost assessment should be based on total annual costs.
Accordingly, total annual costs of controlling Hg emissions from coal-fired electric utility boilers
are examined in this section. These costs include annualized capital charge, annual fixed
operation and maintenance (O&M) costs, and annual variable O&M costs.  Note that Reference
12 provides an examination of the contribution of various cost elements, including cost of PAC,
to total annual cost of Hg controls.

8.3.1 Bituminous-coal-fired Boiler Using CS-ESP

       Several of the Hg control technologies listed in Table 8-1 are potential options for
reducing Hg emissions from a electric utility boiler that fires bituminous coal and already is
using an ESP for PM control.  For boilers firing low-sulfur bituminous coals, these options
include configurations ESP-4 (SC + PAC injection) and ESP-6 (SC + PAC injection + PFF).  For
large boilers firing high-sulfur bituminous coals, the options include configurations ESP-1 (PAC
injection + wet FGD) and ESP-3 (PAC injection + PFF + wet FGD). For smaller boilers
(typically less than 300 MW), these options include configuration SD/ESP-1 (SDA + PAC
injection + ESP). For each of these cases, cost estimates were determined using the cost model.

      Table 8-3 presents the estimated total annual Hg control costs for a bituminous-coal-fired
boiler with existing CS-ESP. The table presents two sets of cost estimates. The first set of
estimates was made based on levels of Hg capture on fly ash using PAC injection rates derived
from the available pilot-scale test data. A subsequent review of the Part III EPA ICR data
(discussed in Section 6.2), however, suggests that levels of Hg capture higher than those
measured in the pilot-scale tests may be occurring.  Consequently, the cost estimates based solely
on pilot test data for Hg control technologies applied to bituminous-coal-fired boilers using ESP
may be overstating the costs. Therefore, a second set of estimates is presented based on the
preliminary ICR results for fly ash capture of Hg (i.e., no PAC injection) in combination with the
pilot-scale results for PAC injection.

       For ESP-4 applied to low-sulfur (0.6 percent) bituminous coal and using pilot-scale PAC
injection rates, the estimated total annual cost ranges from 2.81 mills/kWh for a 100-MWe boiler
removing 90 percent of the total Hg to 0.53 mill/kWh for a 975-MWe boiler removing 60 percent
of the total Hg. The corresponding costs with ICR/pilot-scale PAC injection rates are 1.65
mills/kWh for the 100-MWe boiler and 0.24 mill/kWh for the 975-MWe boiler.

      In general, these results reflect that, for a given boiler, the total annual cost increases non-
linearly with increases in the Hg reduction requirement in concert with the behavior of the PAC
injection rate algorithms (see Attachment 2  in Appendix D). A comparison of results obtained
with pilot-scale and ICR/pilot-scale PAC injection rates also indicates that research and
development efforts aimed at ensuring broad availability of relatively high levels of fly ash
capture of Hg have the potential of providing significant reductions in Hg control costs.
                                        8-11
-------