This is the accessible text file for GAO report number GAO-09-860T 
entitled 'Clean Air Act: Preliminary Observations on the Effectiveness 
and Costs of Mercury Control Technologies at Coal-Fired Power Plants' 
which was released on July 9, 2009. 

This text file was formatted by the U.S. Government Accountability 
Office (GAO) to be accessible to users with visual impairments, as part 
of a longer term project to improve GAO products' accessibility. Every 
attempt has been made to maintain the structural and data integrity of 
the original printed product. Accessibility features, such as text 
descriptions of tables, consecutively numbered footnotes placed at the 
end of the file, and the text of agency comment letters, are provided 
but may not exactly duplicate the presentation or format of the printed 
version. The portable document format (PDF) file is an exact electronic 
replica of the printed version. We welcome your feedback. Please E-mail 
your comments regarding the contents or accessibility features of this 
document to Webmaster@gao.gov. 

This is a work of the U.S. government and is not subject to copyright 
protection in the United States. It may be reproduced and distributed 
in its entirety without further permission from GAO. Because this work 
may contain copyrighted images or other material, permission from the 
copyright holder may be necessary if you wish to reproduce this 
material separately. 

Testimony: 

Before the Subcommittee on Clean Air and Nuclear Safety, Committee on 
Environment and Public Works, U.S. Senate: 

United States Government Accountability Office: 
GAO: 

For Release on Delivery: 
Expected at 10:00 a.m. EDT:
Thursday, July 9, 2009: 

Clean Air Act: 

Preliminary Observations on the Effectiveness and Costs of Mercury 
Control Technologies at Coal-Fired Power Plants: 

Statement of John B. Stephenson, Director: 
Natural Resources and Environment: 

GAO-09-860T: 

GAO Highlights: 

Highlights of GAO-09-860T, testimony before the Subcommittee on Clean 
Air and Nuclear Safety, Committee on Environment and Public Works, U.S. 
Senate. 

Why GAO Did This Study: 

The 491 U.S. coal-fired power plants are the largest unregulated 
industrial source of mercury emissions nationwide, annually emitting 
about 48 tons of mercury—a toxic element that poses health threats, 
including neurological disorders in children. In 2000, the 
Environmental Protection Agency (EPA) determined that mercury emissions 
from these sources should be regulated, but the agency has not set a 
maximum achievable control technology (MACT) standard, as the Clean Air 
Act requires. Some power plants, however, must reduce mercury emissions 
to comply with state laws or consent decrees. 

After managing a long-term mercury control research and development 
program, the Department of Energy (DOE) reported in 2008 that systems 
that inject sorbents—powdery substances to which mercury binds—into the 
exhaust from boilers of coal-fired power plants were ready for 
commercial deployment. Tests of sorbent injection systems, the most 
mature mercury control technology, were conducted on a variety of coal 
types and boiler configurations—that is, on boilers using different air 
pollution control devices. 

This testimony provides preliminary data from GAO’s ongoing work on (1) 
reductions achieved by mercury control technologies and the extent of 
their use at coal-fired power plants, (2) the cost of mercury control 
technologies in use at these plants, and (3) key issues EPA faces in 
regulating mercury emissions from power plants. GAO obtained data from 
power plants operating sorbent injection systems. 

What GAO Found: 

Commercial deployments and 50 DOE and industry tests of sorbent 
injection systems have achieved, on average, 90 percent reductions in 
mercury emissions. These systems are being used on 25 boilers at 14 
coal-fired plants, enabling them to meet state or other mercury 
emission requirements—generally 80 to 90 percent reductions. The 
effectiveness of sorbent injection is largely affected by coal type and 
boiler configuration. Importantly, the substantial mercury reductions 
using these systems commercially and in tests were achieved with all 
three main types of coal and on boiler configurations that exist at 
nearly three-fourths of U.S. coal-fired power plants. While sorbent 
injection has been shown to be widely effective, DOE tests suggest that 
other strategies, such as blending coals or using other technologies, 
may be needed to achieve substantial reductions at some plants. 
Finally, sorbent injection has not been tested on a small number of 
boiler configurations, some of which achieve high mercury removal with 
other pollution control devices. 

The cost of the mercury control technologies in use at power plants has 
varied, depending in large part on decisions regarding compliance with 
other pollution reduction requirements. The costs of purchasing and 
installing sorbent injection systems and monitoring equipment have 
averaged about $3.6 million for the 14 coal-fired boilers operating 
sorbent systems alone to meet state requirements. This cost is a 
fraction of the cost of other pollution control devices. When plants 
also installed a fabric filter device primarily to assist the sorbent 
injection system in mercury reduction, the average cost of $16 million 
is still relatively low compared with that of other air pollution 
control devices. Annual operating costs of sorbent injection systems, 
which often consist almost entirely of the cost of the sorbent itself, 
have been, on average, about $640,000. In addition, some plants have 
incurred other costs, primarily due to lost sales of a coal combustion 
byproduct—fly ash—that plants have sold for commercial use. The carbon 
in sorbents can render fly ash unusable for certain purposes. Advances 
in sorbent technologies that have reduced sorbent costs at some plants 
offer the potential to preserve the market value of fly ash. 

EPA’s decisions on key regulatory issues will have implications for the 
effectiveness of its mercury emissions standard. For example, the data 
EPA decides to use will impact (1) the emissions reductions it starts 
with in developing its regulation, (2) whether it will establish 
varying standards for the three main coal types, and (3) how the 
standard will take into account a full range of operating conditions at 
the plants. These issues can affect the stringency of the MACT standard 
EPA proposes. Data from EPA’s 1999 power plant survey do not reflect 
commercial deployments or DOE tests of sorbent injection systems and 
could support a standard well below what has recently been broadly 
achieved. Moreover, the time frame for proposing the standard may be 
compressed because of a pending lawsuit. On July 2, 2009, EPA announced 
that it planned to conduct an information collection request to update 
existing emission data, among other things, from power plants. 

To view the full product, including the scope and methodology, click on 
[hyperlink, http://www.gao.gov/products/GAO-09-860T]. For more 
information, contact John Stephenson at (202) 512-3841 or 
stephensonj@gao.gov. 

[End of section] 

Mr. Chairman and Members of the Subcommittee: 

I am pleased to be here today to discuss our preliminary findings on 
the effectiveness and costs of mercury control technologies, as well as 
key issues the Environmental Protection Agency (EPA) faces in 
developing a regulation for mercury emissions from coal-fired power 
plants. Mercury is a toxic element that poses human health threats-- 
including neurological disorders in children that impair their 
cognitive abilities. Coal-fired power plants, the nation's largest 
electricity producers, represent the largest unregulated industrial 
source of mercury emissions in the United States.[Footnote 1] 

EPA determined in 2000 that it was "appropriate and necessary" to 
regulate mercury emissions from coal-fired power plants under section 
112 of the Clean Air Act. Subsequently, in 2005, EPA chose to 
promulgate a cap-and-trade program,[Footnote 2] rather than 
establishing a maximum achievable control technology (MACT) standard to 
control mercury emissions--as required under section 112. However, the 
cap-and-trade program was vacated by the D.C. Circuit Court of Appeals 
in February 2008 before EPA could implement it. EPA must now develop a 
MACT standard to regulate mercury emissions from coal-fired power 
plants[Footnote 3]--which will require most existing coal-fired boilers 
to reduce mercury emissions to at least the average level achieved by 
the best performing 12 percent of boilers.[Footnote 4] While developing 
MACT standards for hazardous air pollutants can take up to 3 years, EPA 
may be required to promulgate these standards in a shorter period of 
time to fulfill a negotiated settlement with litigants or comply with a 
court decision. Specifically, EPA has until July 27, 2009, to settle or 
respond to a lawsuit filed by several environmental groups requesting 
an order requiring the EPA Administrator to promulgate final mercury 
emissions standards for coal-fired power plants by a date certain no 
later than December 2010. 

The Department of Energy's (DOE) National Energy Technology Lab has 
worked with EPA and the Electric Power Research Institute (EPRI), among 
others, during the past 10 years on a comprehensive mercury control 
technology test program. Mercury is emitted in such low concentrations 
that its removal and measurement are particularly difficult, and it is 
emitted in several forms, some of which are harder to capture than 
others.[Footnote 5] The DOE program has focused largely on testing 
sorbent injection systems on all coal types and at a variety of boiler 
configurations at operating power plants.[Footnote 6] Testing at a 
variety of boiler configurations using different types of coal was 
important because the type of coal burned and the variety of air 
pollution control devices for other pollutants already installed at 
power plants can impact the effectiveness of sorbent injection systems. 
Further, some power plants achieve mercury reductions as a "co-benefit" 
of using controls designed to reduce other pollutants, such as sulfur 
dioxide, nitrogen oxides, and particulate matter. 

According to a 2008 DOE report describing its mercury technology 
testing program, "DOE successfully brought mercury control technologies 
to the point of commercial-deployment readiness." Nonetheless, the 
report stated that while the results achieved during DOE's field tests 
met or exceeded program goals, the only way to truly know the 
effectiveness--and associated costs--of mercury control technologies is 
through their continuous operation in commercial applications at a 
variety of configurations. In recent years, at least 18 states have 
passed laws or regulations requiring mercury emission reductions at 
coal-fired power plants. The compliance time frames for the state 
requirements vary, and four states--Connecticut, Delaware, 
Massachusetts, and New Jersey--require reductions currently. In this 
context, you asked us to examine (1) what mercury reductions have been 
achieved by existing mercury control technologies and the extent to 
which they are being used at coal-fired power plants; (2) the costs 
associated with mercury control technologies currently in use; and (3) 
key issues EPA faces in developing a new regulation for mercury 
emissions from coal-fired power plants. 

We are currently responding to these objectives. To do this, we are 
identifying power plants with coal-fired boilers that are currently 
operating sorbent injection systems--the most mature, mercury-specific 
control technology--to reduce mercury emissions. Using a structured 
interview tool, we are obtaining data from plant managers and engineers 
on the effectiveness of sorbent injection systems at reducing mercury 
emissions and the costs of doing so. We are also obtaining information 
on the engineering challenges plant officials have encountered in 
installing and operating sorbent injection systems and actions taken to 
mitigate them.[Footnote 7] In addition, we are examining DOE National 
Energy Technology Lab, EPRI, and academic reports on the effectiveness 
and costs of sorbent injection systems over time and reviewing 
literature from recent technical conferences that addressed strategies 
to overcome challenges that some plants have experienced with sorbent 
injection systems. We are also reviewing EPA's requirements for 
establishing MACT standards under the Clean Air Act and recent court 
cases with implications for how EPA establishes such standards. 
Finally, we have met with EPA officials in the Office of Air and 
Radiation regarding the agency's plans for regulating mercury at power 
plants. EPA officials in the Offices of Air and Radiation and Research 
and Development provided comments on the information provided in this 
testimony, and we have made technical clarification where appropriate. 

Background: 

Mercury enters the environment in various ways, such as through 
volcanic activity, coal combustion, and chemical manufacturing. As a 
toxic element, mercury poses ecological threats when it enters water 
bodies, where small aquatic organisms convert it into its highly toxic 
form--methylmercury. This form of mercury may then migrate up the food 
chain as predator species consume the smaller organisms. Fish 
contaminated with methylmercury may pose health threats to people who 
rely on fish as part of their diet. Mercury can harm fetuses and cause 
neurological disorders in children, resulting in, among other things, 
impaired cognitive abilities. The Food and Drug Administration and EPA 
recommend that expectant or nursing mothers and young children avoid 
eating swordfish, king mackerel, shark, and tilefish and limit 
consumption of other potentially contaminated fish. These agencies also 
recommend checking local advisories about recreationally caught 
freshwater and saltwater fish. In recent years, most states have issued 
advisories informing the public that concentrations of mercury have 
been found in local fish at levels of public health concern. 

Coal-fired power plants burn at least one of three primary coal types--
bituminous, subbituminous, and lignite--and some plants burn a blend of 
these coals. Of all coal burned by power plants in the United States in 
2004, DOE estimates that about 46 percent was bituminous, 46 percent 
was subbituminous, and 8 percent was lignite. The amount of mercury in 
coal and the relative ease of its removal depend on a number of 
factors, including the geographic location where it was mined and the 
chemical variation within and among coal types. Coal combustion 
releases mercury in oxidized, elemental, or particulate-bound form. 
Oxidized mercury is more prevalent in the flue gas from bituminous coal 
combustion, and it is relatively easy to capture using some sulfur 
dioxide controls, such as wet scrubbers. Elemental mercury, more 
prevalent in the flue gas from combustion of lignite and subbituminous 
coal, is more difficult to capture with existing pollution controls. 
Particulate-bound mercury is relatively easy to capture in particulate 
matter control devices. In addition to mercury, coal combustion 
releases other harmful air pollutants, including sulfur dioxide and 
nitrogen oxides.[Footnote 8] EPA has regulated these pollutants since 
1995 and 1996, respectively, through its program intended to control 
acid rain. Figure 1 shows various pollution controls that may be used 
at coal-fired power plants: selective catalytic reduction to control 
nitrogen oxides, wet or dry scrubbers to reduce sulfur dioxide, 
electrostatic precipitators and fabric filters to control particulate 
matter, and sorbent injection to reduce mercury emissions. 

Figure 1: Sample Layout of Air Pollution Controls, Including Sorbent 
Injection to Control Mercury, at a Coal-Fired Power Plant: 

[Refer to PDF for image: illustration] 

The illustration depicts the following entities: 
Coal supply; 
Stack; 
Fixed adsorption device; 
Supplemental fabric filter; 
Sorbent injection; 
Scrubber fabric filter or electrostatic precipitator; 
Selective catalytic reduction; 
Boiler. 

Source: Electric Power Research Institute. 

[End of figure] 

From 2000 to 2009, DOE's National Energy Technology Lab conducted field 
tests at operating power plants with different boiler configurations to 
develop mercury-specific control technologies capable of achieving high 
mercury emission reductions at the diverse fleet of U.S. coal-fired 
power plants. As a result, DOE now has comprehensive information on the 
effectiveness of sorbent injection systems using all coal types at a 
wide variety of boiler configurations. Most of these tests were 
designed to achieve mercury reductions of 50 to 70 percent while 
decreasing mercury reduction costs--primarily the cost of the sorbent. 
Thus, the results from the DOE test program may understate the mercury 
reductions that can be achieved by sorbent injection systems to some 
extent. For example, while a number of short-term tests achieved 
mercury reductions in excess of 90 percent, the amount of sorbent 
injection that achieved the reductions was often decreased during long- 
term tests to determine the minimum cost of achieving, on average, 70 
percent mercury emission reductions. 

Under its mercury testing program, DOE initially tested the 
effectiveness of untreated carbon sorbents. On the basis of these 
results, we reported in 2005 that sorbent injection systems showed 
promising results but that they were not effective when used at boilers 
burning lignite and subbituminous coals.[Footnote 9] DOE went on to 
test the effectiveness of chemically treated sorbents--which can help 
convert the more difficult-to-capture mercury common in lignite and 
subbituminous coals to a more easily captured form--and achieved high 
mercury reduction across all coal types.[Footnote 10] Finally, DOE 
continued to test sorbent injection systems and to assess solutions to 
impacts on plant devices, structures, or operations that may result 
from operating these systems--called "balance-of-plant 
impacts."[Footnote 11] In 2008, DOE reported that the high performance 
observed during many of its field tests at a variety of configurations 
has given coal-fired power plant operators the confidence to begin 
deploying these technologies. 

Bills have been introduced in the prior and current Congress addressing 
mercury emissions from power plants. The bills have proposed specific 
limits on mercury emissions, such as not less than 90 percent 
reductions, and some have specified time frames for EPA to promulgate a 
MACT regulation limiting mercury emissions from power plants. For 
example, a bill introduced in this Congress would require EPA to 
promulgate a MACT standard for mercury from coal-fired power plants 
within a year of the bill's enactment. In addition, some bills 
introduced the past few years--termed multipollutant bills--would have 
regulated sulfur dioxide, nitrogen oxides, and carbon dioxide 
emissions, in addition to mercury, from coal-fired power plants. Most 
would have required a 90 percent reduction--or similarly stringent 
limit--of mercury emissions, with the compliance deadlines varying from 
2011 to 2015. One such bill currently before Congress would prohibit 
existing coal-fired power plants from exceeding an emission limit of 
0.6 pounds of mercury per trillion British thermal units (BTUs), a 
standard measure of the mercury content in coal--equivalent to 
approximately a 90 percent reduction--by January 2013. 

Substantial Mercury Reductions Have Been Achieved Using Sorbent 
Injection Technology at 14 Plants and in Many DOE Tests, but Some 
Plants May Require Alternative Strategies to Achieve Comparable 
Results: 

The managers of 14 coal-fired power plants reported to us they 
currently operate sorbent injection systems on 25 boilers to meet the 
mercury emission reduction requirements of 4 states and several consent 
decrees and construction permits.[Footnote 12] Preliminary data show 
that these boilers have achieved, on average, reductions in mercury 
emissions of about 90 percent.[Footnote 13] Of note, all 25 boilers 
currently operating sorbent injection systems have met or surpassed 
their relevant regulatory mercury requirements, according to plant 
managers. For example: 

* A 164 megawatt bituminous-fired boiler, built in the 1960s and 
operating a cold-side electrostatic precipitator and wet scrubber, 
exceeds its 90 percent reduction requirement--achieving more than 95 
percent mercury emission reductions using chemically treated carbon 
sorbent. 

* A 400 megawatt subbituminous-fired boiler, built in the 1960s and 
operating a cold-side electrostatic precipitator and a fabric filter, 
achieves a 99 percent mercury reduction using untreated carbon sorbent, 
exceeding its 90 percent reduction regulatory requirement. 

* A recently constructed 600 megawatt subbituminous-fired boiler 
operating a fabric filter, dry scrubber, and selective catalytic 
reduction system achieves an 85 percent mercury emission reduction 
using chemically treated carbon sorbent, exceeding its 83 percent 
reduction regulatory requirement. 

While mercury emissions reductions achieved with sorbent injection on a 
particular boiler configuration do not guarantee similar results at 
other boilers with the same configuration, the reductions achieved in 
deployments and tests provide important information for plant managers 
who must make decisions about pollution controls to reduce mercury 
emissions as more states' mercury regulations become effective and as 
EPA develops its national mercury regulation.[Footnote 14] The sorbent 
injection systems currently used at power plants to reduce mercury 
emissions are operating on boiler configurations that are used at 57 
percent of U.S. coal-fired power boilers.[Footnote 15] Further, when 
the results of 50 tests of sorbent injection systems at power plants 
conducted primarily as part of DOE's or EPRI's mercury control research 
and development programs are factored in, mercury reductions of at 
least 90 percent have been achieved at boiler configurations used at 
nearly three-fourths of coal-fired power boilers nationally.[Footnote 
16] Some boiler configurations tested in the DOE program that are not 
yet included in commercial deployments follow: 

* A 360 megawatt subbituminous-fired boiler with a fabric filter and a 
dry scrubber using a chemically treated carbon sorbent achieved a 93 
percent mercury reduction. 

* A 220 megawatt boiler burning lignite, equipped with a cold-side 
electrostatic precipitator, increased mercury reduction from 58 percent 
to 90 percent by changing from a combination of untreated carbon 
sorbent and a boiler additive to a chemically treated carbon sorbent. 

* A 565 megawatt subbituminous-fired boiler with a fabric filter 
achieved mercury reductions ranging from 95 percent to 98 percent by 
varying the amount of chemically treated carbon sorbent injected into 
the system.[Footnote 17] 

As these examples of deployed and tested injection systems show, plants 
are using chemically treated sorbents and sorbent enhancement 
additives, as well as untreated sorbents. The DOE program initially 
used untreated sorbents, but during the past 6 years, the focus shifted 
to chemically treated sorbents and enhancement additives that were 
being developed. These more recent tests showed that using chemically 
treated sorbents and enhancement additives could achieve substantial 
mercury reductions for coal types that had not achieved these results 
in earlier tests with untreated sorbents. For example, injecting 
untreated sorbent reduced mercury by an average of 55 percent during a 
2003 DOE test at a subbituminous-fired boiler. Recent tests using 
chemically treated sorbents and enhancement additives, however, have 
resulted in average mercury reductions of 90 percent for boilers using 
subbituminous coals.[Footnote 18] Similarly, recent tests on boilers 
using lignite reduced mercury emissions by roughly 80 percent, on 
average. 

The examples of substantial mercury reductions highlighted above also 
show that sorbent injection can be successful with both types of air 
pollution control devices that power plants use to reduce emissions of 
particulate matter. Specifically, regulated coal-fired power plants 
typically use either electrostatic precipitators or fabric filters for 
particulate matter control. The use of fabric filters--which are more 
effective at mercury emission reductions than electrostatic 
precipitators--at coal-fired power plants to reduce emissions of 
particulate matter and other pollutants is increasing, but currently 
less than 20 percent have them. Plant officials told us that they chose 
to install fabric filters along with 10 of the sorbent injection 
systems currently deployed to assist with mercury control--but that 
some of the fabric filters were installed primarily to comply with 
other air pollution control requirements. One plant manager, for 
example, told us that the fabric filter installed at the plant helps 
the sorbent injection system achieve higher levels of mercury emission 
reductions but that the driving force behind the fabric filter 
installation was to comply with particulate matter emission limits. 
Further, as another plant manager noted, fabric filters may provide 
additional benefits by limiting emissions of acid gases and trace 
metals, as well as by preserving fly ash--fine powder resulting from 
coal combustion--for sale for reuse.[Footnote 19] 

The successful deployments of sorbent injection technologies at power 
plants occurred around the time DOE concluded, on the basis of its 
tests, that these technologies were ready for commercial deployment. 
Funding for the DOE testing program has been eliminated.[Footnote 20] 
Regarding deployments to meet state requirements that will become 
effective in the near future, the Institute of Clean Air Companies 
reported that power plants had 121 sorbent injection systems on order 
as of February 2009.[Footnote 21] 

Importantly, mercury control technologies will not have to be installed 
on a number of coal-fired boilers to meet mercury emission reduction 
requirements because they already achieve high mercury reductions from 
their existing pollution control devices.[Footnote 22] EPA data 
indicate that about one-fourth of the industry may be currently 
achieving mercury reductions of 90 percent or more as a co-benefit of 
other pollution control devices.[Footnote 23] We found that of the 36 
boilers currently subject to mercury regulation, 11 are relying on 
existing pollution controls to meet their mercury reduction 
requirements.[Footnote 24] One plant manager told us their plant 
achieves 95 percent mercury reduction with a fabric filter for 
particulate matter control, a scrubber for sulfur dioxide control, and 
a selective catalytic reduction system for nitrogen oxides control. 
Other plants may also be able to achieve high mercury reduction with 
their existing pollution control devices. For example, according to EPA 
data, a bituminous-fired boiler with a fabric filter may reduce mercury 
emissions by more than 90 percent. 

While sorbent injection technology has been shown to be effective with 
all coal types and on boiler configurations at more than three-fourths 
of U.S. coal-fired power plants, DOE tests show that some plants may 
not be able to achieve mercury reductions of 90 percent or more with 
sorbent injection systems alone. For example: 

* Sulfur trioxide--which can form under certain operating conditions or 
from using high sulfur bituminous coal--may limit mercury reductions 
because it prevents mercury from binding to carbon sorbents. 

* Hot-side electrostatic precipitators reduce the effectiveness of 
sorbent injection systems. Installed on 6 percent of boilers 
nationwide, these particulate matter control devices operate at very 
high temperatures, which reduces the ability of mercury to bind to 
sorbents and be collected in the devices. 

* Lignite, used by roughly 3 percent of boilers nationwide, has 
relatively high levels of elemental mercury--the most difficult form to 
capture. Lignite is found primarily in North Dakota and the Gulf Coast, 
the latter called Texas lignite. Mercury reduction using chemically 
treated sorbents and sorbent enhancement additives on North Dakota 
lignite has averaged about 75 percent--less than reductions using 
bituminous and subbituminous coals. Less is known about Texas lignite 
because few tests have been performed using it. However, a recent test 
at a plant burning Texas lignite achieved an 83 percent mercury 
reduction. 

Boilers that may not be able to achieve 90 percent emissions reductions 
with sorbent injection alone, and some promising solutions to the 
challenges they pose, are discussed in appendix I. Further, EPRI is 
continuing research on mercury controls at power plants that should 
help to address these challenges. 

In some cases, however, plants may need to pursue a strategy other than 
sorbent injection to achieve high mercury reductions. For example, 
officials at one plant decided to install a sulfur dioxide scrubber-- 
designed to reduce both mercury and sulfur dioxide--after sorbent 
injection was found to be ineffective. This approach may become more 
typical as power plants comply with the Clean Air Interstate Rule and 
court-ordered revisions to it, which EPA is currently developing, and 
as some plants add air pollution control technologies required under 
consent decrees. EPA air strategies group officials told us that many 
power plants will be installing devices--fabric filters, scrubbers, and 
selective catalytic reduction systems--that are typically associated 
with high levels of mercury reduction, which will likely reduce the 
number of plants requiring alternative strategies for mercury control. 
Finally, mercury controls have been tested on about 90 percent of the 
boiler configurations at coal-fired power plants. The remaining 10 
percent include several with devices, such as selective catalytic 
reduction devices for nitrogen oxides control and wet scrubbers for 
sulfur dioxide control, which are often associated with high levels of 
mercury emission reductions. 

Mercury Control Technologies Are Often Relatively Inexpensive, but 
Costs Depend Largely on How Plants Comply with Requirements for 
Reducing Other Pollutants: 

The cost to meet current regulatory requirements for mercury reductions 
has varied depending in large part on decisions regarding compliance 
with other pollution reduction requirements. For example, while sorbent 
injection systems alone have been installed on most boilers that must 
meet mercury reduction requirements--at a fraction of the cost of other 
pollution control devices--fabric filters have also been installed on 
some boilers to assist in mercury capture or to comply with particulate 
matter requirements, according to plant officials we interviewed. 

The costs of purchasing and installing sorbent injection systems and 
monitoring equipment have averaged about $3.6 million for the 14 coal- 
fired boilers that use sorbent injection systems alone to reduce 
mercury emissions (see table 1).[Footnote 25] For these boilers, the 
cost ranged from $1.2 to $6.2 million.[Footnote 26] By comparison, on 
the basis of EPA estimates, the average cost to purchase and install a 
wet scrubber for sulfur dioxide control, absent monitoring system 
costs, is $86.4 million per boiler--the estimates range from $32.6 to 
$137.1 million.[Footnote 27] EPA's estimate of the average cost to 
purchase and install a selective catalytic reduction device to control 
nitrogen oxides is $66.1 million, ranging from $12.7 to $127.1 million. 

Capital costs can increase significantly if fabric filters are also 
purchased to assist in mercury emission reductions or as part of 
broader emission reduction requirements. For example, plants installed 
fabric filters at another 10 boilers for these purposes. On the five 
boilers where plant officials reported also installing a fabric filter 
specifically designed to assist the sorbent injection system in mercury 
emission reductions, the average reported capital cost for both the 
sorbent injection system and fabric filter was $15.8 million per 
boiler--the costs ranged from $12.7 million to $24.5 million. 
Importantly, these boilers have uncommon configurations--ones that, as 
discussed earlier, DOE tests showed would need additional control 
devices to achieve high mercury reductions.[Footnote 28] Table 1 shows 
the per-boiler capital costs of sorbent injections systems depending on 
whether fabric filters are also installed primarily to reduce mercury 
emissions. 

Table 1: Average Cost to Purchase and Install Mercury Control 
Technologies and Monitoring Equipment, per Boiler (2008 dollars): 

Mercury control technology: Sorbent injection system; 
Number of boilers[A]: 14; 
Sorbent injection system: $2,723,277; 
Mercury emissions monitoring system: $559,592; 
Consulting and engineering: $381,535; 
Fabric filter: [B]; 
Total: $3,594,023[C]. 

Mercury control technology: Sorbent injection system and fabric filter 
to assist in mercury removal; 
Number of boilers[A]: 5; 
Sorbent injection system: $1,334,971; 
Mercury emissions monitoring system: $119,544; 
Consulting and engineering: $1,444,179; 
Fabric filter: $19,009,986; 
Total: $15,785,997[D]. 

Source: GAO analysis of data from power plants operating sorbent 
injections systems. 

[A] We identified 25 boilers with sorbent injection systems to reduce 
mercury emissions, for which power companies provided cost data on 24. 
Cost data for 19 of the 24 are provided in the table. Costs for the 
remaining 5 are discussed further below because much of the cost 
incurred for fabric filters in these cases is not related to mercury 
removal. 

[B] Not applicable. 

[C] Numbers do not add to total. Total capital costs data were provided 
for 14 boilers in this category, and these totals were used to provide 
the average total capital cost. However, the average cost for the 
individual cost categories include data on only 12 of the 14 boilers in 
this category for which we were provided data. 

[D] Numbers do not add to total. Total capital cost data were provided 
for five boilers with fabric filters, and these totals were used to 
provide the average total capital cost. However, the average cost for 
the individual cost categories only include data on two of the five 
boilers for which we were provided data. 

[End of table] 

For the five boilers where plant officials reported installing fabric 
filters along with sorbent injection systems largely to comply with 
requirements to control other forms of air pollution, the average 
reported capital cost for both the sorbent injection system and fabric 
filter was $105.9 million per boiler, ranging from $38.2 million to 
$156.2 million per boiler.[Footnote 29] We did not determine what 
portion of these costs would appropriately be allocated to the cost of 
reducing mercury emissions. Decisions to purchase such fabric filters 
will likely be driven by the broader regulatory landscape affecting 
plants in the near future, such as requirements for particulate matter, 
sulfur dioxide, and nitrogen oxides reductions, as well as EPA's 
upcoming MACT regulation for coal-fired power plants that, according to 
EPA officials, will regulate mercury as well as other air toxics 
emitted from these plants. 

Regarding operating costs, plant managers said that annual operating 
costs associated with sorbent injection systems consist almost entirely 
of the cost of the sorbent itself. In operating sorbent injection 
systems, sorbent is injected continuously into the boiler exhaust gas 
to bind to mercury passing through the gas. The rate of injection is 
related to, among other things, the level of mercury emission reduction 
required to meet regulatory requirements and to the amount of mercury 
in the coal used. For the 18 boilers with sorbent injection systems for 
which power plants provided sorbent cost data, the average annualized 
cost of sorbent was $674,000.[Footnote 30] 

Plant engineers often adjust the injection rate of the sorbent to 
capture more or less mercury--the more sorbent in the exhaust gas, for 
example, the higher the likelihood that more mercury will bind to it. 
Some plant managers told us that they have recently been able to 
decrease their sorbent injection rates, thereby reducing costs, while 
still complying with relevant requirements. Specifically, a recently 
constructed plant burning subbituminous coal successfully used sorbent 
enhancement additives to considerably reduce its rate of sorbent 
injection--resulting in significant savings in operating costs when 
compared with its original expectations. Plant managers at other plants 
reported that they have injected sorbent at relatively higher rates 
because of regulatory requirements that mandate a specific injection 
rate. One state's consent decree, for example, requires plants to 
operate their sorbent injection systems at an injection rate of 5 
pounds per million actual cubic feet.[Footnote 31] Among the 19 boilers 
for which plant managers provided operating data, the average injection 
rate was 4 pounds per million actual cubic feet; rates ranged from 0.5 
to 11.0 pounds per million actual cubic feet. 

For those plants that installed a sorbent injection system alone--at an 
average cost of $3.6 million--to meet mercury emissions requirements, 
the cost to purchase, install, and operate sorbent injection and 
monitoring systems represents 0.12 cents per kilowatt hour, or a 
potential 97 cent increase in the average residential consumer's 
monthly electricity bill. How, when, and to what extent consumers' 
electric bills will reflect the capital and operating costs power 
companies incur for mercury controls depends in large measure on market 
conditions and the regulatory framework in which the plants operate. 
Power companies in the United States are generally divided into two 
broad categories: (1) those that operate in traditionally regulated 
jurisdictions where cost-based rate setting still applies (rate- 
regulated) and (2) those that operate in jurisdictions where companies 
compete to sell electricity at prices that are largely determined by 
supply and demand (deregulated). Rate-regulated power companies are 
generally allowed by regulators to set rates that will recover 
allowable costs, including a return on invested capital.[Footnote 32] 
Minnesota, for example, passed a law in 2006 allowing power companies 
to seek regulatory approval for recovering the cost of anticipated 
state-required reductions in mercury emissions in advance of the 
regulatory schedule for rate increase requests. One utility in the 
state submitted a plan for the installation of sorbent injection 
systems to reduce mercury emissions at two of its plants at a cost of 
$4.4 and $4.5, respectively, estimating a rate increase of 6 to 10 
cents per month for customers of both plants.[Footnote 33] 

For power companies operating in competitive markets where wholesale 
electricity prices are not regulated, prices are largely determined by 
supply and demand.[Footnote 34] Generally speaking, market pricing does 
not guarantee full cost recovery to suppliers, especially in the short 
run. Of the 25 boilers using sorbent injection systems to comply with a 
requirement to control mercury emissions, 21 are in jurisdictions where 
full cost recovery is not guaranteed through regulated rates. 

In addition to the costs discussed above, some plant managers told us 
they have incurred costs associated with balance-of-plant impacts. The 
issue of particular concern relates to fly ash--fine particulate ash 
resulting from coal combustion that some power plants sell for 
commercial uses, including concrete production, or donate for 
beneficial purposes, such as backfill. According to DOE, about 30 
percent of the fly ash generated by coal-fired power plants was sold in 
2005; 216 plants sold some portion of their fly ash. Most sorbents 
increase the carbon content of fly ash, which may render it unsuitable 
for some commercial uses. Specifically, some plant managers told us 
that they have incurred additional costs because of lost fly ash sales 
and additional costs to store fly ash that was previously either sold 
or donated for beneficial re-use. For the eight boilers with installed 
sorbent injection systems to meet mercury emissions requirements for 
which plants reported actual or estimated fly-ash related costs, the 
average net cost reported by plants was $1.1 million per year.[Footnote 
35] 

Advances in sorbent technologies that have reduced costs at some plants 
also offer the potential to preserve the market value of fly ash. For 
example, at least one manufacturer offers a concrete-friendly sorbent 
to help preserve fly ash sales--thus reducing potential fly ash storage 
and disposal costs. Additionally, a recently constructed plant burning 
subbituminous coal reported that it had successfully used sorbent 
enhancement additives to reduce its rate of sorbent injection from 2 
pounds to less than one-half pound per million actual cubic feet-- 
resulting in significant savings in operating costs and enabling it to 
preserve the quality of its fly ash for reuse. Other potential advances 
include refining sorbents through milling and changing the sorbent 
injection sites. Specifically, in testing, milling of sorbents has, for 
some configurations, improved their efficiency in reducing mercury 
emissions--that is, reduced the amount of sorbent needed--and also 
helped minimize negative impact on fly ash re-use. Also, in testing, 
some vendors have found that injecting sorbents on the hot side of air 
preheaters[Footnote 36] can decrease the amount of sorbent needed to 
achieve desired levels of mercury control. 

Some plant managers reported other balance-of-plant impacts associated 
with sorbent injection systems, such as ductwork corrosion and small 
fires in the particulate matter control devices. Plant engineers told 
us these issues were generally minor and have been resolved. For 
example, two plants experienced corrosion in the ductwork following the 
installation of their sorbent injection systems. One plant manager 
resolved the problem by purchasing replacement parts at a cost of 
$4,500. The other plant manager told us the corrosion problem remains 
unresolved but that it is primarily a minor engineering challenge not 
impacting plant operations. Four plant managers reported fires in the 
particulate matter control devices; plant engineers have generally 
solved this problem by emptying the ash from the collection devices 
more frequently. Overall, despite minor balance-of-plant impacts, most 
plant managers said that the sorbent injection systems at their plants 
are more effective than they originally expected. 

Decisions EPA Faces on Key Regulatory Issues Will Have Implications for 
the Effectiveness of its Mercury Emission Standard for Coal-Fired Power 
Plants and the Availability of Monitoring Data: 

EPA's decisions on key regulatory issues will impact the overall 
stringency of its mercury emissions limit. Specifically, the data EPA 
decides to use will affect (1) the mercury emission reductions 
calculated for "best performers," from which a proposed emission limit 
is derived, (2) whether EPA will establish varying standards for the 
three coal types, and (3) how EPA's standard will take into account 
varying operating conditions. Each of these issues could affect the 
stringency of the MACT standard the agency proposes. In addition, the 
format of the standard--whether it limits the mercury content of coal 
being burned (an input standard) or of emissions from the stack (an 
output standard)--may affect the stringency of the MACT standard the 
agency proposes. Finally, the vacatur of the Clean Air Mercury Rule has 
delayed for a number of years the continuous emissions monitoring that 
would have started in 2009 at most coal-fired power plants. 
Consequently, data on mercury emissions from coal-fired power plants 
and the resolution of some technical issues with monitoring systems 
have both been delayed. 

Current Data from Commercial Deployments and DOE Tests Could Be Used to 
Support a More Stringent Standard for Mercury Emissions from Power 
Plants Than Was Last Proposed by EPA: 

Obtaining data on mercury emissions and identifying the "best 
performers"--defined as the 12 percent of coal-fired power plant 
boilers with the lowest mercury emissions[Footnote 37]--is a critical 
initial step in the development of a MACT standard for mercury. EPA may 
set one standard for all power plants, or it may establish 
subcategories to distinguish among classes, types, and sizes of plants. 
For example, in its 2004 proposed mercury MACT,[Footnote 38] EPA 
established subcategories for the types of coal most commonly used by 
power plants.[Footnote 39] Once the average mercury emissions of the 
best performers are established for power plants--or for subcategories 
of power plants--EPA accounts for variability in the emissions of the 
best performers in its MACT standard(s). EPA's method for accounting 
for variability has generally resulted in MACT standards that are less 
stringent than the average emission reductions achieved by the best 
performers. 

To identify the best performers, EPA typically collects emissions data 
from a sample of plants representative of the U.S. coal-fired power 
industry through a process known as an information collection request. 
Information collection requests are required when an agency collects 
data from 10 or more nongovernmental parties. According to EPA 
officials, this data collection process, which requires Office of 
Management and Budget (OMB) review and approval, typically takes from 8 
months to 1 year. EPA's schedule for issuing a proposed rule and a 
final rule has not yet been established as the agency is currently in 
negotiations with litigants about these time frames. In developing the 
rule, EPA told us it could decide to use data from its 1999 information 
collection request, data from commercial deployments and DOE tests to 
augment its 1999 data, or implement a new information collection 
request for mercury emissions. On July 2, 2009, EPA published a draft 
information collection request in the Federal Register, providing a 60- 
day public comment period on the draft questionnaire to industry prior 
to submitting this information collection request to OMB for review and 
approval. 

Our analysis of EPA's 1999 data, as well as more current data from 
deployments and DOE tests, shows that newer data may have several 
implications for the stringency of the standard. First, the average 
emissions of the best performers, from which the standard is derived, 
may be higher. Our analysis of EPA's 1999 data shows an average mercury 
emission reduction of nearly 91 percent for the best performers. 
[Footnote 40] In contrast, using more current commercial deployment and 
DOE test data, as well as data on co-benefit mercury reductions 
collected in 1999, an average mercury emission reduction of nearly 96 
percent for best performers is demonstrated. The 1999 data do not 
reflect the significant and widespread mercury reductions achieved by 
sorbent injection systems. Further, EPA's 2004 proposed MACT standards 
for mercury were substantially lower than the 1999 average emission 
reduction of the best performers because of variability in mercury 
emissions among the top performers, as discussed in more detail below. 

Second, more current information that reflects mercury control 
deployments and DOE tests may make the rationale EPA used to create 
MACT standards for different subcategories less compelling to the 
agency now. In its 2004 proposed MACT, using 1999 data, EPA proposed 
separate standards for three subcategories of coal used at power 
plants, largely because the co-benefit capture of mercury from 
subbituminous-and lignite-fired boilers was substantially less than 
from bituminous-fired boilers and resulted in higher average mercury 
emissions for best performers using these coal types. Specifically, the 
1999 data EPA used for its 2004 MACT proposal showed that best 
performers achieved average emission reductions of 97 percent for 
bituminous, 71 percent for subbituminous, and 45 percent for lignite. 
In contrast, more current data show that using sorbent injection 
systems with all coal types has achieved at least 90 percent mercury 
emission reductions in most cases. 

Finally, using more current emissions data in setting the mercury 
standard, may mean that accounting for variability in emissions will 
not have as significant an effect as it did in the 2004 proposed MACT-
-thereby lowering the MACT standard--because the current data already 
reflect variability. In its 2004 proposed MACT, EPA explained that its 
1999 data, obtained from the average of short-term tests (three samples 
taken over a 1-to 2-day period), did not necessarily reveal the range 
of emissions that would be found over extended periods of time or under 
a full range of operating conditions they could reasonably anticipate. 
EPA thus extrapolated longer-term variability data from the short-term 
data, and on the basis of these calculations, proposed MACT standards 
equivalent to a 76 percent reduction in mercury emissions for 
bituminous coal, a 25 percent reduction for lignite, and a 5 percent 
reduction for subbituminous coal--20 to 66 percentage points lower than 
the average of what the best performers achieved for each coal type. 

However, current data may eliminate the need for such extrapolation. 
Data from commercial applications of sorbent injection systems, DOE 
field tests, and co-benefit mercury reductions show that mercury 
reductions well in excess of 90 percent have been achieved over periods 
ranging from more than 30 days in field tests to more than a year in 
commercial applications. Mercury emissions measured over these periods 
may more accurately reflect the variability in mercury emissions that 
plants would encounter over the range of operating conditions. Along 
these lines, at least 15 states with mercury emission limits require 
long-term averaging--ranging from 1 month to 1 year--to account for 
variability. According to the manager of a power plant operating a 
sorbent injection system, long-term averaging of mercury emissions 
takes into account the "dramatic swings" in mercury emissions from coal 
that may occur. He told us that while mercury emissions can vary on a 
day-to-day basis, this plant has achieved 94 percent mercury reduction, 
on average, over the last year.[Footnote 41] Similarly, another manager 
of a power plant operating a sorbent injection system told us the 
amount of mercury in the coal they use "varies widely, even from the 
same mine." Nonetheless, the plant manager reported that this plant 
achieves its required 85 percent mercury reduction because the state 
allows averaging mercury emissions on a monthly basis to take into 
account the natural variability of mercury in the coal. 

The Type of Standard EPA Chooses May Also Affect the Stringency of the 
Regulation: 

In 2004, EPA's proposed mercury MACT included two types of standards to 
limit mercury emissions: (1) an output-based standard for new coal- 
fired power plants and (2) a choice between an input-or output-based 
standard for existing plants. Input-based standards establish emission 
limits on the basis of pounds of mercury per trillion British thermal 
units (BTUs) of heat input; output-based standards, on the other hand, 
establish emission limits on the basis of pounds of mercury per 
megawatt hour of electricity produced. These standards are referred to 
as absolute limits. For the purposes of setting a standard, absolute 
emissions limits can be correlated to percent reductions. For example, 
EPA's 2004 proposed standards for bituminous, lignite, and 
subbituminous coal (2, 9.2, and 5.8 pounds per trillion BTUs, 
respectively) are equivalent with mercury emissions reductions of 76, 
25, and 5 percent, respectively, based on nationwide averages of the 
mercury content in coal. During EPA's 2004 MACT development process, 
state and local agency stakeholders, as well as environmental 
stakeholders, generally supported output-based emission limits; 
industry stakeholders generally supported having a choice between an 
emission limit and a percent reduction. EPA must now decide in what 
format it will set its mercury MACT standard(s). 

Input-based limits can have some advantages for coal-fired power 
plants. For example, input-based limits can provide more flexibility to 
older, less efficient plants because they allow boilers to burn as much 
coal as needed to produce a given amount of electricity, as long as the 
amount of mercury per trillion BTUs does not exceed the level specified 
by the standard.[Footnote 42] However, input-based limits may allow 
some power plants to emit more mercury per megawatt hour than output- 
based limits. Under an output-based standard, mercury emissions cannot 
exceed a specific level per megawatt-hour of electricity produced-- 
efficient boilers, which use less coal, will be able to produce more 
electricity than inefficient boilers under an output-based standard. 
Moreover, under an output-based limit, less efficient boilers may have 
to, for example, increase boiler efficiency or switch to a lower 
mercury coal. Thus, output-based limits provide a regulatory incentive 
to enhance both operating efficiency and mercury emission reductions. 

We found that at least 16 states have established a format for 
regulating mercury emissions from coal-fired power plants. Eight states 
allow plants to meet either an emission limit or a percent reduction, 
three require an emission limit, four require percent reductions, and 
one state requires plants to achieve whatever mercury emissions 
reductions--percent reduction or emission limit--are greater.[Footnote 
43] On the basis of our review of these varying regulatory formats, we 
conclude that to be meaningful, a standard specifying a percent 
reduction should be correlated to an absolute limit. When used alone, 
percent reduction standards can limit mercury emissions reductions. For 
example, in one state, mercury reductions are measured against 
"historical" coal-mercury content data, rather than current coal- 
mercury content data. If plants are required to reduce mercury by, for 
example, 90 percent compared to historical coal data, but coal used in 
the past had higher levels of mercury than the plants have been using 
more recently, then actual mercury emission reductions would be less 
than 90 percent. In addition, percent reduction requirements do not 
provide an incentive for plants burning high mercury coal to switch 
coals or pursue more effective mercury control strategies because it is 
easier to achieve a percent reduction requirement with high mercury 
coal than with lower mercury coals. 

Similarly, a combination standard that gives regulated entities the 
option to choose either a specified emission limit or a percent 
reduction might limit actual mercury emission reductions. For example, 
a plant burning coal with a mercury content of 15 pounds per trillion 
BTUs that may choose between meeting an absolute limit of 0.7 pounds of 
mercury per trillion BTUs or a 90 percent reduction could achieve the 
percent reduction while emitting twice the mercury that would be 
allowed under the specified absolute limit. As discussed above, for the 
purposes of setting a standard, a required absolute limit, which 
provides a consistent benchmark for plants to meet, can be correlated 
to a percent reduction. For example, according to EPA's Utility Air 
Toxic MACT working group, a 90 percent mercury reduction based on 
national averages of mercury in coal equates to an emission limit of 
approximately 0.7 pounds per trillion BTUs.[Footnote 44] For bituminous 
coal, a 90 percent reduction equates to a limit of 0.8 pounds per 
trillion BTUs; for subbituminous coal, a 90 percent reduction equates 
to a limit of 0.6 pounds per trillion BTUs; and for lignite, a 90 
percent reduction equates to a limit of 1.2 pounds per trillion BTUs. 

Continuous Monitoring of Mercury Emissions at Most Power Plants Has 
Been Delayed, as Has Resolution of Emissions Monitoring Challenges: 

EPA's now-vacated Clean Air Mercury Rule required most coal-fired power 
plants to conduct continuous emissions monitoring for mercury--and a 
small percentage of plants with low mercury emissions to conduct 
periodic testing--beginning in 2009. State and federal government and 
nongovernmental organization stakeholders told us they support 
reinstating the monitoring requirements of the Clean Air Mercury Rule. 
In fact, in a June 2, 2008, letter to EPA, the National Association of 
Clean Air Agencies requested that EPA reinstate the mercury monitoring 
provisions that were vacated in February 2008 because, among other 
things, the monitoring requirements are important to state agencies 
with mercury reduction requirements. This association for state clean 
air agencies also said the need for federal continuous emissions 
monitoring requirements is especially important in states that cannot 
adopt air quality regulations more stringent than those of the federal 
government. However, EPA officials told us the agency has not 
determined how to reinstate continuous emissions monitoring 
requirements for mercury at coal-fired power plants outside of the MACT 
rulemaking process. As a result, continuous monitoring of mercury 
emissions from coal-fired power plants may continue to be delayed for 
years. 

Under the Clean Air Mercury Rule, the selected monitoring methodology 
for each power plant was to be approved by EPA through a certification 
process. For its part, EPA was to develop a continuous emissions 
monitoring systems (CEMS) certification process and approve protocols 
for quality control and assurance. However, when the Clean Air Mercury 
Rule was vacated, EPA put its CEMS certification process on hold. 

Effective emissions monitoring assists facilities and regulators in 
ensuring compliance with regulations and can also help facilities 
identify ways to better understand the efficiency of their processes 
and the efficiency of their operations. Monitoring mercury emissions is 
more complex than monitoring other pollutants, such as nitrogen oxides 
and sulfur dioxide, which are measured in parts per million. Mercury, 
for example, is emitted at lower levels of concentration than other 
pollutants and is measured in parts per billion--it is like "trying to 
find a needle in a haystack," according to one plant engineer. 
Consequently, mercury CEMS require more time to install and setup than 
CEMS for other pollutants, and, according to plant engineers using 
them, they involve a steeper learning curve in getting these relatively 
complex monitoring systems up and running properly. 

EPA plans to release interim quality control protocols for mercury CEMS 
in July 2009. In our work, we found that these systems are installed on 
16 boilers at power plants for monitoring operations or for compliance 
reporting.[Footnote 45] Our preliminary data shows that for regulated 
coal-fired boilers, plant managers reported that their mercury CEMS 
were online from 62 percent to 99 percent of the time. When these 
systems were offline, it was mainly because of failed system integrity 
checks or routine parts failure. Some plant engineers told us that CEMS 
are accurate at measuring mercury, but others said that these systems 
are "several years away" from commercial readiness. However, according 
to an EPA Clean Air Markets Division official, while some technical 
monitoring issues remain, mercury CEMS are sufficiently reliable to 
determine whether plants are complying with their relevant state 
mercury emissions regulations. 

Concluding Observations: 

Data from commercially deployed sorbent injection systems show that 
substantial mercury reductions have been achieved at a relatively low 
cost. Importantly, these results, along with test results from DOE's 
comprehensive research and development program, suggest that 
substantial mercury emission reductions can likely be achieved at most 
coal-fired power plants in the United States. Other strategies, 
including blending coal and using other technologies, exist for the 
small number of plants with configuration types that were not able to 
achieve significant mercury emissions reductions with sorbent injection 
alone. 

Whether power plants will install sorbent injection systems or pursue 
multipollutant control strategies will likely be driven by the broader 
regulatory context in which they operate, such as requirements for 
sulfur dioxide and nitrogen oxides reductions in addition to mercury, 
and the associated costs to comply with all pollution reduction 
requirements. Nonetheless, for many plants, sorbent injection systems 
appear to be a cost-effective technology for reducing mercury 
emissions. For other plants, sorbent injection may represent a 
relatively inexpensive bridging technology--that is, one that is 
available for immediate use to reduce only mercury emissions but that 
may be phased out--over time--with the addition of multipollutant 
controls, which are more costly. Moreover, some plants emit small 
amounts of mercury without mercury-specific controls because their 
existing controls for other air pollutants also effectively reduce 
mercury emissions. In fact, while many power companies currently 
subject to mercury regulation have installed sorbent injection systems 
to achieve required reductions, about one-third of them are relying on 
existing pollution control devices to meet the requirements. 

As EPA proceeds with its rulemaking process to regulate hazardous air 
pollutants from coal-fired power plants, including mercury, it will 
likely find that current data on commercially deployed sorbent 
injection systems and plants that achieve high mercury reductions from 
their existing pollution control devices justify a more stringent 
mercury emission standard than was last proposed in 2004. More 
significant mercury emission reductions are actually being achieved by 
the current best performers than was the case in 1999 when such 
information was last collected--and similar results can likely be 
achieved by most plants across the country at relatively low cost. 

Mr. Chairman, this concludes my prepared statement. We expect to 
complete our ongoing work by October 2009. I would be happy to respond 
to any questions that you or other Members of the Subcommittee may have 
at this time. 

GAO Contact and Staff Acknowledgments: 

Contact points for our Offices of Congressional Relations and Public 
Affairs may be found on the last page of this statement. For further 
information about this testimony, please contact me at (202) 512-3841 
or stephensonj@gao.gov. Key contributors to this statement were 
Christine Fishkin (Assistant Director), Nathan Anderson, Mark Braza, 
Antoinette Capaccio, Nancy Crothers, Philip Farah, Mick Ray, and Katy 
Trenholme. 

[End of section] 

Appendix I: Potential Solutions to Challenges Associated with Achieving 
Mercury Emissions Reductions of 90 Percent or More Using Sorbent 
Injection Systems: 

DOE tests show that some plants may not be able to achieve mercury 
reductions of 90 percent or more with sorbent injections alone. 
Specifically, the tests identified three factors that can impact the 
effectiveness of sorbent injection systems: sulfur trioxide 
interference, using hot-side precipitators, and using lignite. These 
factors are discussed below, along with some promising solutions to the 
challenges they pose. 

Sulfur trioxide interference. High levels of sulfur trioxide gas may 
limit mercury emission reductions by preventing some mercury from 
binding to carbon sorbents. Using an alkali injection system in 
conjunction with sorbent injection can effectively lessen sulfur 
trioxide interference. Depending on the cause of the sulfur trioxide 
interference--which can stem from using a flue gas conditioning system, 
a selective catalytic reduction system, or high sulfur bituminous coal--
additional strategies may be available to ensure high mercury 
reductions: 

* Flue gas conditioning systems, used on 13 percent of boilers 
nationwide, improve the performance of electrostatic precipitators by 
injecting a conditioning agent, typically sulfur trioxide, into the 
flue gas to make the gas more conducive to capture in electrostatic 
precipitators. Mercury control vendors are working to develop 
alternative conditioning agents that could be used instead of sulfur 
trioxide in the conditioning system to improve the performance of 
electrostatic precipitators without jeopardizing mercury emission 
reductions using sorbent injection. 

* Selective catalytic reduction systems, a common control device for 
nitrogen oxides, are used by about 20 percent of boilers nationwide. 
Although selective catalytic reduction systems often improve mercury 
capture, in some instances these devices may lead to sulfur trioxide 
interference when sulfur in the coal is converted to sulfur trioxide 
gas. Newer selective catalytic reduction systems often have improved 
catalytic controls, which can minimize the conversion of sulfur to 
sulfur trioxide gas. 

* High sulfur bituminous coal--defined as having a sulfur content of at 
least 1.7 percent sulfur by weight--may also lead to sulfur trioxide 
interference in some cases. As many as 20 percent of boilers nationwide 
may use high sulfur coal, according to 2005 DOE data; however, the 
number of coal boilers using high sulfur bituminous coal is likely to 
decline in the future as more stringent sulfur dioxide regulations take 
effect. Plants can consider using alkali-based sorbents, such as Trona, 
which adsorb sulfur trioxide gas before it can interfere with the 
performance of sorbent injection systems. Plants that burn high sulfur 
coal can also consider blending their fuel to include some portion of 
low sulfur coal. In addition, according to EPA, power companies are 
likely to have or to install scrubbers for controlling sulfur dioxide 
at plants burning high sulfur coal and are more likely to use the 
scrubbers, rather than sorbent injection systems, to also reduce 
mercury emissions. 

Hot-side electrostatic precipitators. Installed on 6 percent of boilers 
nationwide, these particulate matter control devices operate at very 
high temperatures, which reduce the incidence of mercury binding to 
sorbents for collection in particulate matter control devices. However, 
at least two promising techniques have been identified in tests and 
commercial deployments at configuration types with hot-side 
electrostatic precipitators. First, 70 percent mercury emission 
reductions were achieved with specialized heat-resistant sorbents 
during DOE testing. Moreover, one of the 25 boilers currently using a 
sorbent injection system has a hot-side electrostatic precipitator and 
uses a heat-resistant sorbent. Although plant officials are not 
currently measuring mercury emissions for this boiler, the plant will 
soon be required to achieve mercury emission reductions equivalent to 
90 percent.[Footnote 46] Second, in another DOE test, three 90 megawatt 
boilers--each with a hot-side electrostatic precipitator--achieved more 
than 90 percent mercury emission reductions by installing a shared 
fabric filter in addition to a sorbent injection system, a system 
called TOXECONTM. According to plant officials, these three units 
currently use this system to comply with a consent decree and achieved 
94 percent mercury emission reductions during the third quarter of 
2008, the most recent compliance reporting period when the boiler was 
operating under normal conditions. 

Lignite. North Dakota and Texas lignite, the fuel source for roughly 3 
percent of boilers nationwide, have relatively high levels of elemental 
mercury--the most difficult form to capture. Overall, tests on boilers 
using lignite reduced mercury emissions by roughly 80 percent, on 
average. For example, four long-term DOE tests were conducted at coal 
units burning North Dakota lignite using chemically-treated sorbents. 
Mercury emission reductions averaged 75 percent across the tests. The 
best result was achieved at a 450 megawatt boiler burning North Dakota 
lignite and having a fabric filter and a dry scrubber--mercury 
reductions of 92 percent were achieved when chemically-treated sorbents 
were used. In addition, two long-term tests were conducted at plants 
burning Texas lignite with a 30 percent blend of subbituminous coal. 
With coal blending, these boilers achieved average mercury emission 
reductions of 82 percent. Specifically, one boiler, with an 
electrostatic precipitator and a wet scrubber, achieved mercury 
reductions in excess of 90 percent when burning the blended fuel. The 
second boiler achieved 74 percent reduction in long-term testing. 
However, 90 percent was achieved in short term tests using a higher 
sorbent injection rate. Although DOE conducted no tests on plants 
burning purely Texas lignite, one power company is currently conducting 
sorbent injection tests at a plant burning 100 percent Texas lignite 
and is achieving promising results. In the most recent round of 
testing, this boiler achieved mercury removal of 83 percent using 
untreated carbon and a boiler additive in conjunction with the existing 
electrostatic precipitator and wet scrubber. 

[End of section] 

Footnotes: 

[1] EPA's 1999 data, the agency's most recent available data on mercury 
emissions, show that the 491 U.S. coal-fired power plants annually emit 
48 tons of mercury into the air. 

[2] EPA's cap-and-trade program, known as the Clean Air Mercury Rule, 
was established under Clean Air Act section 111 and was to establish a 
cap on mercury emissions of 38 tons for 2010 and a second phase cap of 
15 tons for 2018. 

[3] According to EPA, its MACT will also cover the other hazardous air 
pollutants listed in the Clean Air Act as well as emissions from oil- 
fired power plants. 

[4] For categories with fewer than 30 sources, the MACT standard must 
be set, at least, at the average level achieved by the top five 
performing units. 

[5] Mercury can be emitted in particulate, oxidized, or elemental form. 

[6] Sorbent injection systems inject sorbents--powdery substances, 
typically activated carbon, to which mercury binds--into the exhaust 
from boilers before it is emitted from the stack. 

[7] To date, we have visited seven plants using sorbent injection 
systems, and we have interviewed plant managers at five other plants 
that are meeting state mercury emissions requirements with existing 
pollution control devices for other pollutants. 

[8] Pollution controls that may be used at coal-fired power plants 
include selective catalytic reduction to control nitrogen oxides, wet 
or dry scrubbers to reduce sulfur dioxide, electrostatic precipitators 
and fabric filters to control particulate matter, and sorbent injection 
to reduce mercury emissions. 

[9] GAO, Clean Air Act: Emerging Mercury Control Technologies Have 
Shown Promising Results, but Data on Long-Term Performance Are Limited, 
[hyperlink, http://www.gao.gov/products/GAO-05-612] (Washington, D.C.: 
May 31, 2005). 

[10] DOE injected sorbents that were treated with halogens such as 
chlorine or bromine, which help convert mercury from an elemental form 
into an oxidized form. 

[11] Near the end of the research program, DOE continued field tests of 
advanced mercury control technologies but aimed to achieve 90 percent 
or greater mercury capture at low costs and to have them available for 
commercial demonstration by 2010. According to a DOE official, federal 
funding for DOE tests was eliminated before the final phase of tests 
was completed. 

[12] To date, we have interviewed managers at plants with 24 of the 25 
sorbent injection systems. We do not have mercury emissions reduction 
data for 5 of the 24 sorbent injection systems because the power 
company running these systems is not required to measure emissions 
under its regulatory framework. 

[13] This number reflects 9 boilers that were required to achieve 90 
percent mercury emission reduction--which seven surpassed--and 10 
boilers that were required to achieve reductions between 80 percent and 
89 percent. Plant officials did not provide data on mercury reductions 
achieved by sorbent injection systems for 5 boilers. Data for another 
boiler are pending. 

[14] For example see EPRI's 2006 Mercury Control Technology Selection 
Guide, which summarized tests by DOE and other organizations to provide 
the coal-fired power industry with a process to select the most 
promising mercury control technologies. EPRI assessed the applicability 
of technologies to various coal types and power plant configurations 
and developed decision trees to facilitate decision making. 

[15] We used EPA's 2006 National Electric Energy Data System database 
for calculating the percentage of coal-fired boilers with particular 
configuration types. We excluded coal-fired boilers under 25 megawatts 
from our analysis because the Clean Air Act does not apply to smaller 
units such as these. 

[16] We identified 56 field tests conducted by DOE during its mercury 
control technology testing program. Of these tests, we examined mercury 
reduction data of 41 tests conducted at power plants. The majority of 
these tests were long-term tests (30 days or more). We did not include 
mercury reduction data associated with the other 15 tests in our 
analysis either because they reflected mercury reduction associated 
with mercury oxidation catalysts--an emerging mercury control 
technology--or because test result data were not reported. We also 
analyzed results of 9 tests conducted by industry, primarily by EPRI. 

[17] The rate of sorbent injection varied between 1.0 lbs per million 
actual cubic feet and 3.0 lbs per million actual cubic feet. 

[18] On subbituminous coal units, eight long-term tests were conducted 
using chemically treated sorbents. The average mercury emission 
reduction was 90 percent, with mercury reductions ranging from 81 
percent to 93 percent. 

[19] Properties of fly ash vary significantly with coal composition and 
plant-operating conditions. Some power plants sell fly ash for use in 
Portland cement and to meet other construction needs. 

[20] The DOE mercury testing program has not received new funding since 
fiscal year 2008. 

[21] Illinois, Maryland, Minnesota, Montana, New Mexico, New York, and 
Wisconsin require compliance by the end of 2010. Arizona, Colorado, New 
Hampshire, Oregon and Utah require compliance in 2012 or beyond. 
Georgia and North Carolina require installation of other pollution 
control devices between 2008 and 2018 that capture sulfur dioxide, 
nitrogen oxides, and mercury as a side benefit. North Carolina requires 
the submission of specific mercury reduction plans for certain plants 
by 2013. 

[22] Nationwide, mercury reductions achieved as a co-benefit of other 
pollution control devices reduces mercury emissions from about 75 tons 
(inlet coal) to approximately 48 tons. Mercury reductions achieved as a 
co-benefit range from zero to nearly 100 percent, depending on control 
device configuration and coal type. For example, a boiler using 
bituminous coal and having a fabric filter can achieve mercury 
reductions in excess of 90 percent. In contrast, a boiler using 
subbituminous coal and having only a cold-side electrostatic 
precipitator might achieve little, if any, co-benefit mercury capture. 

[23] This estimate is based on data from EPA's 1999 information 
collection request, which EPA air toxics program officials believe to 
be representative of the current coal-fired power industry. 

[24] Two of these plants will face increasingly stringent limits in the 
next 3 to 4 years. One plant manager, facing a mercury reduction 
requirement that will increase from 80 percent to 90 percent, told us 
that the plant is currently installing a sorbent injection system in 
anticipation of the more stringent standard. The other plant manager, 
facing a mercury reduction requirement that will increase from 85 
percent to 95 percent, told us that his plant will likely need to 
install a sorbent injection system in the future to supplement the co- 
benefit mercury capture the plant currently achieves with existing 
pollution controls. 

[25] All reported cost data have been adjusted for inflation and are 
reported in 2008 dollars. 

[26] The total cost to purchase and install a sorbent injection system 
reflects the costs of (1) sorbent injection equipment, (2) an 
associated mercury emissions monitoring system, and (3) associated 
engineering and consulting services. 

[27] EPA cost estimates reported in 2006 have been adjusted for 
inflation and are reported in 2008 dollars. 

[28] Three of the five boilers with fabric filters designed 
specifically to assist in mercury reduction, for instance, have hot- 
side electrostatic precipitators--a relatively rare particulate matter 
control device that inhibits high mercury removal when sorbent 
injection systems are used without fabric filters. 

[29] The average cost of the sorbent injection system for these boilers 
was $2.9 million and for the monitoring systems, $500,000. The average 
cost for the fabric filters was $84 million and for the engineering 
studies, $11 million. 

[30] Sorbent costs ranged from $76,500 to $2.4 million. 

[31] Pounds per million actual cubic feet is the standard metric for 
measuring the rate at which sorbent is injected into a boiler's exhaust 
gas. 

[32] Under traditional cost-based rate regulations, utility companies 
submit to regulators the costs they seek to cover through the rates 
they charge their customers. Regulators examine the utility's request 
and decide what costs are allowable under the relevant rules. 

[33] The rate increase request will be submitted in conjunction with 
requests for rate increases for the utility's other plants. 

[34] If demand for electricity is elastic (that is, consumers have some 
flexibility in adjusting the quantities that they purchase in response 
to price changes), suppliers may not be able to raise prices in order 
to fully recover the incremental cost of mercury emissions control. For 
instance, if pollution controls add 5 percent to the cost of generating 
electricity, the generating company may be able to raise its prices by 
only 3 percent. 

[35] Technologies to mitigate balance-of-plant costs associated with 
fly ash are available. For example, one plant installed a polishing 
fabric filter using TOXECONTM system, which preserves the plant's 
ability to sell its fly ash. Another plant had previously installed as 
ash reduction device that removes excess carbon in fly ash and enables 
the plant to sell the vast majority of its fly ash when operating its 
sorbent injection system. 

[36] An air preheater is a device designed to preheat the combustion 
air used in a fuel-burning furnace for the purpose of increasing the 
thermal efficiency of the furnace. 

[37] This is how section 112 of the Clean Air Act, as amended, defines 
best performers for the largest categories of sources when establishing 
MACT standards. 

[38] Prior to finalizing the Clean Air Mercury Rule, EPA also proposed 
a MACT standard for mercury emissions from coal-fired power plants. EPA 
chose not to finalize the MACT rule. 

[39] Under the Clean Air Act Amendments of 1990, EPA had 10 years from 
the enactment of the amendments, or two years from the listing of 
electric steam generating units as sources of hazardous air pollutants 
subject to regulation, whichever was later, to promulgate a MACT 
standard. Because EPA did not list electric steam generating units 
until 2000, it originally had two years, or until 2002, to promulgate a 
MACT standard. 

[40] Our analysis of EPA's data includes the three primary coal ranks: 
bituminous, subbituminous, and lignite. 

[41] The requirement for this plant, which the plant manager reported 
it has met, is for a 90 percent reduction averaged over a 3-month 
period. 

[42] The main types of coal burned, in decreasing order of rank, are 
bituminous, subbituminous, and lignite. Rank is the coal classification 
system based on factors such as the heating value of the coal. High- 
rank coal generally has relatively high heating values (i.e., heat per 
unit of mass when burned) compared with low rank coal, which has 
relatively low heating values. 

[43] Colorado, Connecticut, Delaware, Illinois, Massachusetts, New 
Jersey, Oregon, and Utah allow either an emission limit or a percent 
reduction; Montana, New Mexico, and New York require an emission limit; 
Maryland, Minnesota, New Hampshire, and Wisconsin require percent 
reductions; and Arizona requires the more stringent option. 

[44] Presentation on "Recommendations on the Utility Air Toxics MACT, 
Final Working Group Report, October 2002." The Working Group on the 
Utility MACT was formed under the Clean Air Act Advisory Committee, 
Subcommittee for Permits/New Source Reviews/Toxics. 

[45] At least 14 states have enacted mercury emission standards that 
include a mercury monitoring requirement. Six states require monitoring 
to be conducted in accordance with the monitoring provisions of the 
Clean Air Mercury Rule. Four states require sole use of CEMS. Three 
states allow periodic stack tests--a method not approved under the 
Clean Air Mercury Rule--until CEMS can be used at a later date. One 
state requires use of CEMS or other method approved by the state 
environmental protection agency. 

[46] Plant officials did not provide us with mercury emission reduction 
data for this boiler. 

[End of section] 

GAO's Mission: 

The Government Accountability Office, the audit, evaluation and 
investigative arm of Congress, exists to support Congress in meeting 
its constitutional responsibilities and to help improve the performance 
and accountability of the federal government for the American people. 
GAO examines the use of public funds; evaluates federal programs and 
policies; and provides analyses, recommendations, and other assistance 
to help Congress make informed oversight, policy, and funding 
decisions. GAO's commitment to good government is reflected in its core 
values of accountability, integrity, and reliability. 

Obtaining Copies of GAO Reports and Testimony: 

The fastest and easiest way to obtain copies of GAO documents at no 
cost is through GAO's Web site [hyperlink, http://www.gao.gov]. Each 
weekday, GAO posts newly released reports, testimony, and 
correspondence on its Web site. To have GAO e-mail you a list of newly 
posted products every afternoon, go to [hyperlink, http://www.gao.gov] 
and select "E-mail Updates." 

Order by Phone: 

The price of each GAO publication reflects GAO’s actual cost of
production and distribution and depends on the number of pages in the
publication and whether the publication is printed in color or black and
white. Pricing and ordering information is posted on GAO’s Web site, 
[hyperlink, http://www.gao.gov/ordering.htm]. 

Place orders by calling (202) 512-6000, toll free (866) 801-7077, or
TDD (202) 512-2537. 

Orders may be paid for using American Express, Discover Card,
MasterCard, Visa, check, or money order. Call for additional 
information. 

To Report Fraud, Waste, and Abuse in Federal Programs: 

Contact: 

Web site: [hyperlink, http://www.gao.gov/fraudnet/fraudnet.htm]: 
E-mail: fraudnet@gao.gov: 
Automated answering system: (800) 424-5454 or (202) 512-7470: 

Congressional Relations: 

Ralph Dawn, Managing Director, dawnr@gao.gov: 
(202) 512-4400: 
U.S. Government Accountability Office: 
441 G Street NW, Room 7125: 
Washington, D.C. 20548: 

Public Affairs: 

Chuck Young, Managing Director, youngc1@gao.gov: 
(202) 512-4800: 
U.S. Government Accountability Office: 
441 G Street NW, Room 7149: 
Washington, D.C. 20548: