Mercury Emissions from Electric Generating Units: A Review of EPA Analysis and MACT Determination

Congressional research reportJan 21, 2005

Ask Donna

What actually matters in this document.

Text

Order Code RL32744

CRS Report for Congress

Received through the CRS Web

Mercury Emissions from Electric Generating

Units: A Review of EPA Analysis and

MACT Determination

January 21, 2005

Dana A. Shea, Larry Parker, James E. McCarthy,

and Thomas Chapman

Resources, Science, and Industry Division

Congressional Research Service ˜ The Library of Congress

Mercury Emissions from Electric Generating Units: A

Review of EPA Analysis and MACT Determination

Summary

The Environmental Protection Agency (EPA) has identified mercury as the

hazardous air pollutant emitted from electric generating units that is of greatest public

health concern. As a result, EPA proposed regulations limiting electric generating

unit (EGU) mercury emissions. In December 2000, EPA committed to promulgating

Maximum Achievable Control Technology (MACT) emissions limits under Section

112 of the Clean Air Act.

Section 112 sets specific requirements for MACT standards. For new facilities,

the MACT standard must be at least as stringent as the degree of emissions control

achieved at the best controlled similar source. For existing facilities, the MACT

standard must generally achieve limits equal to the average performance of the best

12% of comparable sources. Determination of this performance is complicated by

the lack of installed commercial technology specifically for capture of EGU mercury

emissions, thereby necessitating data collection on other existing technologies and

extensive analysis of potential control levels. To determine the level of allowable

mercury emissions, EPA collected data regarding coal composition and mercury

emissions from an 80 EGU sample. Analysis of these data led EPA to subcategorize

EGUs and propose MACT standards for each subcategory in January 2004.

The proposed MACT standards have been criticized by a wide range of

stakeholders on several criteria, including EPA methodology to determine the

allowable emissions threshold. The EPA methodology incorporates two statistical

treatments, the use of, first, a 97.5% upper confidence limit to account for the

variability in input coal, and, second, another 97.5% upper confidence interval to

account for variance in plant operation. The result is proposed standards that are

substantially less stringent than the average emissions rate of the top 12% of the 80

unit sample. While EPA justifies these statistical treatments, others assert they

unnecessarily weaken the proposed regulation. Indeed, the proposed MACT standard

allows a greater amount of mercury emission than most stakeholders, from all

viewpoints, had recommended prior to the EPA proposal. The proposed MACT

standards (pounds of mercury per trillion Btu) are shown, along with the average

mercury emission and the results of the first statistical treatment, in the table below.

Coal

Subcategory

Bituminous

Subbituminous

Lignite

IGCC

Waste Coal

Average Measured

Mercury Emission

0.1

0.8

5.0

0.09

Average 97.5% Upper

Confidence Limit

1.1

3.1

7.8

0.14

Proposed MACT

Level

2.0

5.8

9.2

0.38

The EPA has agreed to promulgate final mercury emission standards by March

15, 2005 for EGUs. Whether these standards should follow the proposed regulation

in its current form, an alternate form, or through an alternate mechanism is a topic

of likely congressional interest. This report will not be updated.

Contents

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Maximum Achievable Control Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

EPA Methodology for Generating MACT Levels . . . . . . . . . . . . . . . . . . . . . . . . . 5

Alternate Interpretations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Sampling Concerns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

Subcategories and MACT Floor Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Variability and MACT Floor Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Appendix A: Effects of Statistical Treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

List of Figures

Figure 1. Effect of Statistical Treatment on Proposed MACT Standard

for Bituminous- and Subbituminous-fired Facilities . . . . . . . . . . . . . . . . . . . 8

Figure 2. Proposed Mercury MACT Relative to Projected Mercury Emissions

for Facilities Used to Set MACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

List of Tables

Table 1. Proposed MACT Emission Limits for Coal-Fired Electric Generating

Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

Table 2. Existing Coal-Fired Electric Generating Units Compared to MACT

Sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

Table 3. Comparison of Measured Mercury Emissions, 97.5% Upper

Confidence Limits, and Proposed MACT Standard . . . . . . . . . . . . . . . . . . . 7

Table 4: Pre-proposal Stakeholder Group Recommendations . . . . . . . . . . . . . . . 11

Note: Thomas Chapman is on detail with the Congressional Research Service from

the National Science Foundation.

Mercury Emissions from Electric Generating

Units: A Review of EPA Analysis and

MACT Determination

Introduction

Mercury is a heavy metal and potent neurotoxin that bioaccumulates and

concentrates as it passes through the food chain. The Environmental Protection

Agency (EPA) considers it “both a public health concern and a concern in the

environment.”1 These concerns have grown in recent years as research has indicated

mercury’s presence at significant levels in numerous species of fish. Analyses of

human dietary intake and resulting levels of mercury in blood have pointed to

potential health risks from mercury ingestion, particularly for women of child-bearing

age.2

To limit mercury exposure, 45 states have issued fish consumption advisories.3

At the same time, EPA and the states have promulgated a series of regulations to

reduce the amount of mercury used in products4 and limit emissions for most sources

of mercury emitted to the atmosphere.5

U.S. emissions of mercury to the air come from eight principal sources. Of

these, the largest source, and the only major source for which emission standards

have not yet been promulgated, is coal-fired electric generating units. Coal-fired

U.S. EPA, “Regulatory Finding on the Emissions of Hazardous Air Pollutants from

Electric Utility Steam Generating Units,” 65 Federal Register 79830, December 20, 2000.

For specific information see U.S. EPA, “Office of Children’s Health Protection, America’s

Children and the Environment: Measures of Contaminants, Body Burdens, and Illnesses,

2nd edition, February 2003. See also Kathryn R. Mahaffey, “Methylmercury: Epidemiology

Update,” presented at U.S. EPA National Forum on Contaminants in Fish, San Diego, CA,

January 26, 2004, p. 5, available online at

[http://www.epa.gov/waterscience/fish/forum/2004/presentations/monday/mahaffey.pdf].

Twenty-one states (primarily in the Midwest and Northeast) have issued statewide

advisories for mercury in all their freshwater lakes and/or rivers. In all, these advisories

cover more than 13 million acres of lakes and roughly 767,000 river miles. Eleven states,

primarily in the South, have statewide advisories for mercury in their coastal waters. In

addition, Hawaii has a statewide advisory for mercury in marine fish. For more information,

see U.S. EPA, Office of Water, “National Listing of Fish Advisories,” Fact Sheet, August

2004, p. 4, available at [http://www.epa.gov/waterscience/fish/advisories/factsheet.pdf].

For more information, see Mercury in Products and Waste: Legislative and Regulatory

Activities to Control Mercury, CRS Report RL31908 by James E. McCarthy.

For more information, see Mercury Emissions to the Air: Regulatory and Legislative

Proposals, CRS Report RL31881, pp. 2-4 (federal regulations) and p. 7 (state regulations).

CRS-2

electric generating units account for between one-third and one-half of total U.S.

mercury emissions. In January 2004, EPA proposed regulations to limit emissions

from these units, positing two possible control mechanisms. One is to establish

Maximum Achievable Control Technology (MACT) standards for emissions of

mercury from coal-fired electric power plants; the other is to develop a cap-and-trade

mechanism for those same emissions. This report focuses on the proposed EPA

MACT standards, including critiques thereof, and does not treat the proposed capand-trade alternative model.

A number of controversies have arisen concerning the proposed MACT

standards. In general, critics argue that the proposed standards are not sufficiently

stringent to meet the statutory requirements for MACT. The EPA responds that

technology to meet more stringent standards is not currently available, and that the

standards reflect the best performance achieved by currently installed control

equipment.6

This report examines this controversy, i.e., whether the proposed MACT

standards do reflect the best emission control achieved at existing coal-fired electric

generating units as required by statute. This report begins by describing the statutory

requirements and the standards proposed by EPA. It then proceeds to describe the

sampling methods used by EPA to generate data on existing mercury control levels.

The methodology used by EPA to determine the proposed MACT standard, based on

the sampling data, is described. Finally, some of the comments received by EPA on

the proposed standards are reviewed.

Maximum Achievable Control Technology

On January 30, 2004, under Section 112(d) of the Clean Air Act, the EPA

proposed the establishment of Maximum Achievable Control Technology (MACT)

standards for emissions of mercury from coal-fired electric power plants.7 The

standards rely on the statutory requirements in Section 112(d)(3) of the Clean Air

Act, which states that for new sources, “the degree of reduction ... shall be not less

stringent than the emission control that is achieved in practice by the best controlled

similar source.” For existing sources, the degree of reduction shall be (with some

qualifications) “not less stringent, and may be more stringent than ... the average

emission limitation achieved by the best performing 12 percent of the existing

sources (for which the Administrator has emissions information)...,” or, if there are

fewer than 30 sources in a category or subcategory, it shall be not less stringent than

“the average emission limitation achieved by the best performing 5 sources (for

U.S. EPA, Proposed National Emission Standards for Hazardous Air Pollutants; and, in

the Alternative, Proposed Standards of Performance for New and Existing Stationary

Sources: Electric Utility Steam Generating Units, Preamble, Section IV.D.2, 69 Federal

Standards were also proposed for nickel emissions from oil-fired power plants, but these

standards have not proven as controversial, and are not discussed in this report.

CRS-3

which the [EPA] Administrator has or could reasonably obtain emissions

information)....”8

The statute establishes a minimum requirement (a “MACT floor”), but it

explicitly permits the Administrator to impose more stringent requirements. It is

unclear how often EPA has used this authority to go “beyond the floor,” although two

examples of the Agency’s doing so involve mercury controls imposed on municipal

and medical waste incinerators in the mid-1990s.9 This report focuses on EPA’s

calculation of the current MACT floor for mercury emissions from coal-fired

powerplants, not whether the Agency should have gone beyond it.10 Thus, this report

will not discuss at any length beyond-the-floor technologies or standards.

No U.S. power plant has, as yet, installed emission control technologies

specifically designed to capture mercury emissions. Therefore, the currently

proposed MACT floor for these sources relies on calculations of mercury removal

rates (“co-benefits”) achieved by other technology that was installed to control sulfur

dioxide, particulates, or other regulated pollutants. The EPA’s objective was to

identify the technologies designed for other pollution control objectives that have the

maximum level of these co-benefits for mercury emissions.

The proposed MACT standards for mercury emissions from coal-fired

powerplants are shown in Table 1. Using the authorities provided in Section 112(c)

and (d), EPA divided coal-fired plants into five subcategories — four based on coal

type (bituminous, subbituminous, lignite, and coal refuse) and the fifth based on

combustion technology (integrated gasification combined cycle, also known as

IGCC). Each type of coal has differing properties and mercury contents.

Additionally, the number of facilities burning each type of coal varies, with the

majority of facilities burning bituminous or subbituminous coal. The EPA set

standards that vary by as much as a factor of 30, depending on the coal type. The

most stringent standards apply to coal-refuse-fired and bituminous-fired facilities,

while the other categories have less stringent standards. As shown in Table 2, the

size of the five subcategories also varies widely: 95% of the existing units fall into

42 U.S.C. 7412

At the time these regulations were promulgated, these technologies were considered

“beyond the floor,” although now they are standard controls on incinerators. The most

common of the technologies is activated carbon injection (ACI), which, when used with a

fabric filter (or “baghouse”), captures more than 90% of mercury emissions from

incinerators. Similar technology has been the subject of successful full-scale field tests at

at least four coal-fired power plants, but has not yet been commercially installed.

Considerable controversy has arisen over whether EPA is correct in deciding not to go

“beyond the floor.” This determination is generally supported by industry groups, who

agree that mercury-specific controls are not commercially available. Others, including

environmental groups and some state and local agencies, contend that because technologies

such as ACI exist, EPA’s MACT standards for mercury from electric utilities should have

been more stringent by imposing standards that would be achieved by the use of ACI or

some other promising technology. For an example, see comments from State and Territorial

Air Pollution Program Administrators/Association of Local Air Pollution Control Officials,

Docket ID No. OAR-2002-0056 (June 29, 2004), p. 3.

CRS-4

the bituminous or subbituminous groups while the other three subcategories have

relatively few units.

The proposed standards would apply on a unit-by-unit basis for each unit with

a capacity greater than 25 megawatts electric (MW); but the proposal allows facilities

three forms of flexibility in achieving compliance. First, emissions can be averaged

among all of the units at a facility.11 A large power plant commonly has several

units, so this provision might allow facilities to comply by installing stringent

controls on one or more of their units, while leaving other units uncontrolled.

Second, compliance will be determined not by whether a facility is within the

standard continuously, but by whether the rolling 12-month average of emissions is

within the standard. This would allow a facility to exceed the standard for days or

weeks, provided that it lowers emissions sufficiently at other times during the 12

months to offset peak emissions. The EPA states that use of an averaging period is

appropriate because mercury emissions from power plants are not an acute health

hazard.

Table 1. Proposed MACT Emission Limits for Coal-Fired

Electric Generating Units

Emission Limit for Existing Sources

Output Basis

Emission Limit for

New Sources

(lb Hg/TBtu)a

(10-6 lb Hg/MWh)b

Bituminous-fired

2.0

6.0

Subbituminous-fired

5.8

Lignite-fired

9.2

IGCC

200

Coal-refuse-fired

0.38

4.1

1.1

Input Basis

Unit Type

Source: 69 Federal Register 4662, January 30, 2004. Limits for all sources are based on a 12-month

rolling average.

Pounds of mercury per trillion British thermal units.

Millionths of a pound of mercury per megawatt-hour.

Third, the standards for existing facilities are expressed both in input-based

(lbs/trillion Btu) and output-based (lbs/megawatt-hour) form. Output-based

standards provide an incentive for more efficient generation: a plant that can exceed

the efficiency factor used by EPA to correlate the input-based and output-based

standards would be able to burn more coal before reaching the mercury emissions

limit.

A facility is defined as “a contiguous plant site where one or more electric utility steam

generating units are located.”

CRS-5

The proposed standards do not include a percentage emission reduction

requirement. The EPA estimates that the standards would result in emissions of 34

tons of mercury from the affected sources, a 29% reduction as compared to the 1999

emission level of 48 tons.12

Sampling

The MACT standard is based on data obtained during an information collection

request (ICR) initiated by the EPA in April, 1998 to gather data on mercury

emissions from electric generating units.13 The industry was canvassed and those

data were used to determine the amount of mercury emitted with respect to the coal

burned. The collection of comprehensive data is key to understanding the range of

possible plant configurations, coal compositions, and subsequent mercury emissions.

EPA recognized that competing considerations, the cost of the data collection

effort versus the comprehensiveness of the data obtained, were important factors in

determining the final number of sites sampled.14 The EPA decided that, due to the

factors of cost and effectiveness, 3 units would be selected for each combination of

mechanical sulfur dioxide control, coal type, and particulate matter control.15 This

led to a maximum of 135 units considered, assuming that at least three units

populated each combination.16 In practice, no site existed for several of the

combinations. Data was subsequently obtained from 80 sites and considered to

establish the MACT floors.17 Table 2 provides a comparison of the number of

existing coal-fired power plants to the number of plants used to determine the

proposed MACT standard.

EPA Methodology for Generating MACT Levels

EPA chose to subcategorize the mercury-emissions data by unit type because of

the different mercury collection characteristics following combustion. A different

MACT floor was determined for each of these categories by examining the mercuryemission rates. In the case of coal-refuse-fired, lignite-fired, and IGCC units, fewer

than 30 units exist. For these subcategories, the MACT floor was to be established

by the average emission limitation achieved by the best performing 5 sources. In the

case of subbituminous-fired and bituminous-fired units, more than 30 units exist, and

69 Federal Register 4661, January 30, 2004.

63 Federal Register, 17406, April 9, 1998.

63 Federal Register, 17408, April 9, 1998.

The EPA referred to the last category as electrostatic precipitator control, though it

included other forms of particulate control, such as fabric filters.

Office of Air Quality Planning and Standards, Standard Form 83-I Supporting Statement

for OMB Review of EPA ICR No. 1858.01: Information Collection Request for Electric

Utility Steam Generating Unit Mercury Emissions Information Collection Request, U.S.

EPA, Research Triangle Park, NC, p. 13, November 16, 1998.

69 Federal Register 4661, January 30, 2004.

CRS-6

so the MACT floor was to be determined by the best performing 12% of existing

units.

Table 2. Existing Coal-Fired Electric Generating Units

Compared to MACT Sample

Unit Type

Number of Existing Units

Number of Units in Sample

Bituminous-fired

701

Subbituminous-fired

236

Lignite-fired

IGCC

Coal-refuse-fired

Source: 69 Federal Register 4662, January 30, 2004. Memorandum from Clark Allen, RTI

International to Jeffrey Cole, RTI International, December 2003.

The EPA found it necessary to statistically treat the data received from the

information collection request. Because the mercury-emission measurements were

limited in time, often described as a snapshot, EPA determined that it was necessary

to project the probable mercury emissions from coal burned at the facility over the

course of a year. Coal mercury-content and heating-value data were combined with

estimates of control efficiencies, based on the observed data, of various air-pollution

control equipment, in some cases taking into account the effect of coal chlorine level,

to estimate the distribution of mercury emitted.18

Rather than use the directly measured mercury emissions, EPA chose instead

to use the range of coal compositions reported by each plant to estimate an emissions

value which would be exceeded no more than 2.5% of the operating time, the 97.5%

upper confidence limit. The EPA considers this value to represent the operation of

the unit under conditions reasonably expected to occur at each unit.19

Table 3 presents the average measured mercury emissions for the facilities

included in the MACT determination, the results of the first statistical treatment, and

the proposed MACT level for each coal subcategory.

In cases where a sufficient correlation was established between the coal chlorine content

and the emitted mercury, the coal chlorine content was used to predict mercury emissions

from coal data. In cases where this correlation was insufficient, the estimated control

efficiencies of the existing air-pollution control equipment was used to predict mercury

emissions from coal data.

69 Federal Register 4673, January 30, 2004.

CRS-7

Table 3. Comparison of Measured Mercury Emissions, 97.5%

Upper Confidence Limits, and Proposed MACT Standard

(in pounds of mercury per trillion Btu coal)

Coal

Subcategory

Average Measured

Mercury Emissiona

Average 97.5% Upper

Confidence Limit

Proposed

MACT Level

Bituminous

0.1

1.1

2.0

Subbituminous

0.8

3.1

5.8

Lignite

5.0

7.8

9.2

IGCC

Waste Coal

0.09

0.14

0.38

Source: CRS analysis from EPA data.

a. The average measured mercury emission is for those facilities directly included in the MACT

determination, not for all facilities within the subcategory.

For bituminous and subbituminous-fired units, there are more than 30

established units; thus the MACT floor is established by the best performing 12% of

units. The EPA determined that the sample obtained from the information collection

request was representative of existing facilities, and therefore the best performing

12% of the sample would suffice to establish the MACT floor. In both the

bituminous- and subbituminous-fired cases, EPA took the four units which had the

lowest measured mercury emissions and combined the observed collection

efficiencies with coal-composition values reported by those units over the course of

a year to calculate the expected mercury emission. From these estimated emission

rates, EPA determined, for each unit, the projected mercury-emission level that

would not be exceeded more than 2.5% of the time. These four values were

averaged, and their average was then statistically adjusted to a level which would not

be exceeded more than 2.5% of the operating time. This second statistical

adjustment is intended to reflect the possible variability in operation of different units

burning the same fuel type.20 This final value was assigned as the MACT standard.

For a graphical representation of the effects of the statistical treatments on the MACT

standard, see Figure 1.

Ibid.

CRS-8

Figure 1. Effect of Statistical Treatment on Proposed MACT

Standard for Bituminous- and Subbituminous-fired Facilities

Source: Figure developed by CRS from EPA data regarding coal composition for electric generating

units chosen by EPA for determination of proposed MACT.

Note: The above figures display the EPA projected mercury emissions based on coal analyses of those

facilities included in the MACT determination. The horizontal axis is displayed in units of pounds of

emitted mercury per trillion Btu coal. The vertical axis is displayed in relative frequency, the number

of times a projected mercury emission was predicted for a given value divided by all predictions for

that facility. The data has been normalized by CRS so as to appropriately weight uneven reporting

between facilities. The average 97.5% cutoff is the average of the upper-97.5% confidence limit

determined from the calculated distributions for each set of facilities. The measurement average refers

to the average mercury emissions measured for the facilities included in the MACT standard

determination, not the average mercury emissions measured for all facilities of that subcategory. The

incorporation of each statistical treatment serves to make the proposed MACT standard less stringent.

Similarly, for lignite-fired units, the observed mercury emissions from the five

best performing units were adjusted to reflect the coal variability, being adjusted to

a value which would be exceeded no more than 2.5% of the operating time, and then

averaged. This average value was then adjusted so that it would not be exceeded

more than 2.5% of the operating time. This final value was assigned as the MACT

floor.

For IGCC and coal-refuse-fired units, EPA had mercury-emissions data from

only two units, even though the MACT floor is to be set from the five best

performing units.21 The EPA used the same method as was used to determine the

MACT floor for lignite-fired units, except that they only used the data for the two

units.

While there are only 2 existing IGCC units, there are 17 existing coal-refuse-fired units.

The information collection request only gathered data from 2 coal-refuse-fired plants.

CRS-9

The outcome of the EPA efforts to account for variations with fuel composition

within the performance of a single unit and between different units burning the same

types of coal is to significantly increase the MACT floor. The resulting MACT floor

is higher than the average of the measured emissions and higher than the emissions

average corrected for coal variability in all cases. For a graphical example of the

effect of the statistical treatments on the MACT analysis for other MACT

subcategories, see Appendix A.

Alternate Interpretations

EPA received over 680,000 public comments on its 2004 proposed mercury

rule, including almost 5,000 unique comments.22 The comments represent another

phase in a regulatory process that began in earnest with EPA’s 1998 identification of

mercury as the hazardous air pollutant emitted by electric utilities that was of the

“greatest potential concern.”23 Many of these comments build on recommendations

provided by forums and stakeholder groups that met before release of the 2004

proposal.

During the formulation of the proposed mercury rule process, EPA convened

a stakeholder forum called the Utility MACT Working Group. Consisting of

industry, environmental groups, and state and local agencies, the working group

attempted to develop recommendations to assist EPA in setting a utility mercury

MACT. Unable to develop consensus, the working group’s efforts resulted in

separate sets of recommendations from four groups of stakeholders: (1) an

environmental group;24 (2) a state and local agencies group;25 (3) a major industry

group;26 and (4) a minority industry group (the Clean Energy Group, CEG).27 Many

Environmental Protection Agency, Proposed National Emission Standards for Hazardous

Air Pollutants; Notice of Data Availability, p. 10.

U.S. EPA, Study of Hazardous Air Pollutant Emissions from Electric Utility Steam

Generating Units — Final Report to Congress, 1998.

Environmental Stakeholders included Natural Resources Defense Council, National

Wildlife Federation, National Environmental Trust, Clean Air Task Force, and

Environmental Defense.

State and Local Agencies included The Northeast States for Coordinated Air Use

Management (NESCAUM), State and Territorial Air Pollution Program Administrators /

Association of Local Air Pollution Control Officials (STAPPA/ALAPCO), State of New

Jersey, and the Regional Air Pollution Control Agency (Dayton, Ohio).

Majority Industry Group included Cinergy, National Mining Association, West

Associates, Latham & Watkins, Southern Company Generation, the United Mine Workers

of America, the Utility Air Regulatory Group, the Class of 85 Regulatory Response Group,

American Public Power Association, and the National Rural Electric Cooperative

Association.

CEG members included PG&E National Energy Group, Consolidated Edison, Public

Service Enterprise Group, Conectiv, Exelon Corporation, KeySpan, Northeast Utility, and

Sempra Energy.

CRS-10

of the comments subsequently received by EPA referred to recommendations

developed during this process.

As analyzed and summarized by the Northeast States for Coordinated Air Use

Management (NESCAUM) in Table 4, the four stakeholder groups developed

recommendations considerably different from each other and from EPA’s proposal.28

The following observations are among those that can be made:

Unlike EPA’s proposal, all the stakeholder groups determined that

fluidized-bed combustion (FBC) units could achieve a more

stringent mercury MACT than other facilities.

Like EPA’s proposal, the industry groups used some form of

subcategorization, generally based on coal type, while the

environmental group and state and local agencies group did not.

EPA’s variability analysis resulted in less stringent MACT floors

than those discussed or recommended by any of the four stakeholder

groups.

The rationale behind these recommendations intersect in interesting ways with

the rationale of EPA’s proposals. These interactions are evident in the comments

various members of these groups submitted in response to the EPA proposal. The

following Table 4 examines these intersections in three categories: the recommended

mercury emission standard, the relevant annual mercury baseline (in tons), and the

post-MACT annual mercury emissions (in tons).

See NESCAUM, Mercury MACT Under the Clean Air Act: An Assessment of the

Mercury Emissions Outcomes of Stakeholder Group Recommendations (May 8,

2003). The National Mining Association (NMA) characterized the MACT floor attributed

to the majority industry group as “illustrative,” not definitive recommendations. See NMA,

Docket ID No. OAR-2002-0056 (May 14, 2004), p. 9.

CRS-11

Table 4: Pre-proposal Stakeholder Group Recommendations

Stakeholder

Group

Environmental

Stakeholders

State and

Local Agencies

Clean Energy

Group (CEG)

Majority

Industry

Groupa

Environmental

Protection

Agency

Proposed

MACT

Recommended Mercury Emission Standard

C 0.19 lbs/TBtu for FBC facilities

C 0.21 lbs/TBtu for all other facilities for all coal types

C 0.4-0.6 lbs/TBtu for bituminous and subbituminous coal

C 0.320 lbs/TBtu for bituminous or subbituminous FBC

facilities

C 11.984 lbs/TBtu for lignite FBC facilities

C 1.223 lbs/TBtu for all other bituminous or subbituminous

boilers

C 9.091 lbs/TBtu for all other lignite boiler types

C 3.7 lbs/TBtu for hot stack bituminous facilities

C 2.2 lbs/TBtu for saturated stack facilities burning

bituminous coal

C 3.2 lbs/TBtu for wet stack facilities burning bituminous coal

C 4.2 lbs/TBtu for subbituminous facilities

C 6.5 lbs/TBtu for facilities burning lignite

C 2.0 lbs/TBtu for FBC facilities

C 2.0 lbs/TBtu for bituminous coal facilities

C 5.8 lbs/TBtu for subbituminous coal facilities

C 9.2 lbs/TBtu for lignite coal facilities

C 19 lbs/TBtu for IGCC facilities

C 0.38 lbs/TBtu for waste coal facilities

Relevant Annual

Baseline (tons)

44.6

Post-MACT Annual

Emissions (tons)

1.9

41.5

6.3-6.7

44.6

13.1

44.6

25.0-30.0

Source: NESCAUM, Mercury MACT Under the Clean Air Act: An Assessment of the Mercury Emissions Outcomes of Stakeholder Group

Recommendations (May 8, 2003), Appendix A, and 69 Federal Register 4662, January 30, 2004.

Note: NESCAUM’s analysis generally uses a 44.6 ton annual mercury baseline (instead of EPA’s 48 ton estimate) because a small subset of units

in EPA’s Utility Air Toxic Study database lack either the coal supply or mercury emissions data used to calculate projected mercury reductions.

NESCAUM further reduced the baseline for the State and Local Agency Stakeholder group as that group’s recommendations did not cover the 11

lignite units (3.1 tons mercury) contained in the database.

a. The National Mining Association (NMA) characterized the MACT floor attributed to the majority industry group as “illustrative,” not definitive

recommendations. See NMA, Docket ID No. OAR-2002-0056 (May 14, 2004), p. 9.

CRS-12

Sampling Concerns

While the EPA is confident the data sample is representative of the industry,29

others have raised concerns whether the method used to select both the number of

sites sampled and the specific sites included in the dataset provides sample data

representative of those units already employed in industry settings. For example, the

National Mining Association (NMA) states:

The ICR emissions data represent, at best, a limited ‘snap shot’ of emissions

from a few units, taken over a very short period of time, with a limited number

of fuels, and cannot account for the wide variability of coals and process

conditions encompassed by the full fleet of U.S. utility boilers. In addition, the

units chosen by EPA for mercury emissions sampling in the ICR program is [sic]

unrepresentative of U.S. coal-fueled power plants.30

Based on an analysis conducted for NMA by AEMS, LLC (a consulting firm), that

found inadequacies and biases in the EPA 80 unit sample, NMA states:

As a result of these inadequacies the ICR Part III data cannot be used to identify

and evaluate the emissions performance of specific units that would rank in the

top 12 percent of their subcategory (or among the 5 best) on an annual basis.31

Indeed, NMA argues that the data and related analysis are so flawed that EPA should

abandon a MACT-based form of mercury regulation, stating that if EPA proceeds:

“NMA and its members reserve all potential objections to such standards arising

under the Data Quality Act and other applicable provisions of Federal law.”32

The State of Missouri, Department of Natural Resources, contending that the

MACT standard, at least with respect to subbituminous coal, is too lenient, states:

The Missouri source calculations indicate that the proposed MACT emission

limitation for existing subbituminous coal-fired units is too lenient. This

leniency was likely caused by a sampling error resulting from basing the

determinations on a sample population too small to be representative of the norm.

In this case, the sample size of 32 units used to determine the floor greatly

increased the likelihood of sampling errors.33

The EPA determined that an appropriate experimental design for the ICR was

to vary three parameters, coal type, sulfur dioxide control technology, and particulate

69 Federal Register 4670, January 30, 2004.

National Mining Association, Docket ID No. OAR-2002-0056 (May 14, 2004), p. 9.

NMA, Docket Number OAR-2002-0056 (May 14, 2004) pp. 16-17.

NMA, Docket Number OAR-2002-0056 (May 14, 2004) p. 43.

Department of Natural Resources, State of Missouri, Docket ID No. OAR-2002-0056, p.

9-10.

CRS-13

matter control technology, and include one facility for each combination.34 Some

groups have questioned whether an experimental strategy including a more

proportionate distribution of facilities or including technologies that are in

development, but not yet implemented, would have provided for a more stringent

mercury standard. For example, NESCAUM asserts that EPA’s attempts to treat

variability in the ICR data is flawed, in part, due to the lack of any data from mercury

specific controls, such as activated carbon injection.35 The EPA has stated that

mercury-specific control technologies are not yet commercially available for electric

generating units, and will likely not be commercially available prior to 2010, and

therefore consideration of these technologies in determining the MACT floor would

be incorrect.

Subcategories and MACT Floor Levels

A primary source of disagreement about EPA’s proposal is the use of

subcategories. While Section 112(d) provides the EPA Administrator with the

authority to subcategorize MACT standards, the criteria for determining these

subcategories are not defined. The EPA used a combination of coal types

(bituminous, subbituminous, lignite) and firing configurations (IGCC, boilers) to

develop subcategories within the utility mercury MACT. This decision has proven

highly controversial, both with respect to the subcategories chosen, and the MACT

floor determination for each subcategory.

Environmental groups have attacked the proposed subcategories as “without

rational basis, arbitrary and capricious.”36 The groups note that many sources,

including some in EPA’s data sample, use a blend of different coal types and that

combustion technologies vary little with respect to coal type burned. In addition,

these groups state that many boilers have some flexibility to adapt to different coal

feedstocks. Also criticizing the use of subcategories, the State of New Jersey

suggests in its comments that because the proposed standard for subbituminous coal

is nearly three times that for bituminous coal, utilities may be able to comply with the

proposed MACT standards by simply switching coal supplies, emitting more mercury

than before.37 That the weaker subbituminous proposal could encourage western

subbituminous coal production at the expense of eastern bituminous production has

been expressed by eastern coal interests.38

Office of Air Quality Planning and Standards, Standard Form 83-I Supporting Statement

for OMB Review of EPA ICR No. 1858.01: Information Collection Request for Electric

Utility Steam Generating Unit Mercury Emissions Information Collection Request, U.S.

EPA, Research Triangle Park, NC, p. 28, November 16, 1998.

NESCAUM, Docket Number OAR-2002-0056 (June 29, 2004) p. 5.

Comments of Eleven Environmental and Public Health Organizations, Docket ID No.

OAR-2002-0056 (June 29, 2004) p. II-7.

Department of Environmental Protection, State of New Jersey, EPA’s Proposed MACT

Floor Standards for Mercury Emissions from Coal-fired Utility Units: A Statistical and

Analytical Assessment, (July, 2004) pp. 11-12.

For example, see “EPA’s Leavitt Promises Mercury Rule Will Treat Coal Types Equally,”

(continued...)

CRS-14

These groups believe that the MACT floor should not be stratified by coal type,

but should be a simple average of the best performing facilities.39 By eliminating

coal type and EPA’s variability adjustment, the environmental groups’ joint

comments developed a MACT floor of 1.0 (10-6lbs/MWh output basis) — higher

than the environmental stakeholders recommended levels in Table 4, but generally

more stringent than EPA’s proposal. In a similar vein, NESCAUM found that if the

average observed mercury emissions from the top-performing 12% of EPA’s 80 unit

data sample, regardless of coal type, was used as the emissions standard, an

emissions reduction of 91.1% would be achieved, in comparison with the predicted

29% reduction in mercury emissions expected under EPA’s proposed MACT

standard.40

Most industry groups agreed with EPA’s determination that subcategories were

appropriate, but disagreed on their specification. In particular, several commenters

argued that North Dakota and Gulf Coast lignite should be regulated separately, not

combined. Stating that mercury content of Gulf Coast lignite was at least double that

of North Dakota lignite, the Edison Electric Institute (EEI) argued that the proposed

lignite standard would be unachievable for units using Gulf Coast lignite.41 Other

concerns were expressed about the absence of any units burning Powder River Basin

subbituminous coal in EPA sample data.42

Variability and MACT Floor Levels

A major source of disagreement surrounding the proposed mercury rule is

EPA’s variability analysis. To determine the proposed standard, EPA used a

methodology to identify and quantify expected variability in mercury emissions

because of the variability of mercury in coal. The result is proposed standards that

are substantially less stringent than the simple average emissions rate of the top 12%

of its 80 unit sample.

This variability analysis has come under particular scrutiny by all parties.

Environmental groups label the analysis as “statistical gimmicks” that result in

MACT floors that represent “virtually the worst short-term emissions from the worst

performing of the best units, and assumes that these pollution levels will persist

through the year.”43 One of the more detailed analyses of EPA’s methodology has

been done by the State of New Jersey. In a white paper by its Department of

(...continued)

Energy Washington Week (December 8, 2004).

The environmental groups would allow MACT levels stratified by firing configuration.

As noted earlier, both groups believe that “beyond the floor” regulation should be

promulgated.

NESCAUM, Docket Number OAR-2002-0056 (June 29, 2004) p. 3.

EEI, Docket Number OAR-2002-0056 (June 29, 2004) pp. 54-55.

NMA, Docket Number OAR-2002-0056 (May 14, 2004) p. 14.

Comments of Eleven Environmental and Public Health Organizations, Docket ID No.

OAR-2002-0056 (June 29, 2004) p. II-25.

CRS-15

Environmental Protection, New Jersey attacks EPA’s approach that uses variability

in short-term peak emissions to set a standard that would be based on a 12-month

rolling average,44 stating:

If it were essential to identify and quantify variability in emissions, the relevant

measure of variability would be what was experienced over 12 months, not what

was experienced in any single stack testing event. A transient peak in emissions

during the 12-month period will not cause noncompliance with the standard,

because lower emissions during the rest of the period will average out the effect

of the peak. For that reason, using the short-term worst-case scenario to set an

annual standard immediately makes the standard unnecessarily lax. If compliance

is to be based on a long-term average, EPA should be estimating long-term

averages, not worst case emissions, to determine the MACT floor.45

New Jersey proceeds to argue that EPA’s proposed standards are indeed less stringent

as a result of their variability analysis. Eliminating EPA’s variability analysis,

including subcategorization, would reduce the MACT floor to 0.18 lbs/TBtu.

Alternately, if the data were only stratified by coal type, the average of the observed

mercury emissions would be 0.12 lbs/TBtu for bituminous units, in contrast with the

proposed MACT standard of 2.0 lbs/TBtu.

In contrast, while industry generally supported EPA’s efforts, some industry

analysis found that EPA insufficiently dealt with the mercury variability issue. In

arguing for MACT floors substantially higher than EPA’s proposal, NMA asserted

that a combination of data quality and variability concerns require an increase in

MACT floor estimates for new sources to 5.1 lbs/TBtu for bituminous units and 7.4

lbs/TBtu for subbituminous units.46 According to analysis by AEMS, LLC, the EPA

MACT standards for new sources, when converted from an output basis to an input

basis, are equivalent to 0.63 lbs/TBtu for bituminous units and 2.1 lbs/TBtu for

subbituminous units.47 As stated by NMA when comparing the converted standards

developed by AEMS, LLC with those standards suggested by NMA: “These

estimates of appropriate emission limits for new units are an order of magnitude

higher than those estimated by EPA, reflecting AEMS’ more comprehensive

Department of Environmental Protection, EPA’s Proposed MACT Floor Standards for

Mercury Emissions from Coal-fired Utility Units: A Statistical and Analytical Assessment,

(July, 2004).

Department of Environmental Protection, EPA’s Proposed MACT Floor Standards for

Mercury Emissions from Coal-fired Utility Units: A Statistical and Analytical Assessment,

(July, 2004), pp. 3-4.

NMA also argued in its comments that “the combination of uncertainty in the emissions

analysis and repeat test variability means that the MACT floors for new units must admit

the possibility of measured emission levels in excess of the mercury content of the coals

(negative removals).” See NMA, Docket Number OAR-2002-0056 (May 14, 2004) p. 34.

In converting from an output basis to an input basis, AEMS LLC assumed a heat rate of

9,500 Btu/kwh. AEMS, LLC, Review and Critique of Data and Methodologies Used in EPA

Proposed Utility Mercury MACT Rulemaking, (April 2004) p. 79.

CRS-16

approach to variability analyses, as well as the use of available data on mercury

content of U.S. coal.”48

Conclusion

The proposed regulations to limit mercury emissions from electric generating

units have been criticized by a wide range of stakeholders based on several criteria,

including the method used by EPA to determine the allowed emissions threshold.

The proposed emissions threshold incorporates two statistical treatments. This serves

to make the proposed MACT standard less stringent than the average of the actual

emission measurements, and less stringent than the recommendations of

stakeholders, including industry and environmental groups, prior to the MACT

proposal. While EPA justifies these statistical treatments as necessary to account for

the variability in input coal and plant operation, others assert that they represent a

significant, unnecessary weakness of the proposed regulation. Whether mercury

emissions from electric generating units should be controlled via the proposed

regulation in its current form, an alternate form, or through an alternate mechanism

is a topic of likely congressional interest.

NMA, Docket Number OAR-2002-0056 (May 14, 2004) p. 35.

CRS-17

Appendix A: Effects of Statistical Treatment

The effect of each statistical treatment performed by EPA serves to make the

proposed MACT standard less stringent than the average of the measured mercuryemissions data. The EPA found it necessary to statistically treat the data received

from the information collection request for several reasons.

First, the

mercury-emission measurements were limited in time, often described as a snapshot,

and may not represent the actual emissions performance of the facility. As a

consequence, EPA projected the probable mercury emissions from coal burned at the

facility over the course of a year.

The EPA developed these projections through one of two methods. In the case

of some equipment configurations, it was determined that a correlation between the

mercury emissions and the coal chlorine-content existed. In these situations, the coal

chlorine content was used to predict the mercury emissions. In other cases, a suitable

correlation did not exist. In these situations, the mercury emissions for that facility

were determined by using an empirical estimate of the control efficiency for mercury

of existing emissions-control equipment. By obtaining these projected mercury

emissions, EPA was able to determine the expected mercury emissions level over a

twelve-month period.

The EPA took the projected mercury emissions, developed from the coal

compositions reported by each plant, and determined the emissions value which

would be exceeded no more than 2.5% of the operating time, the upper 97.5%

confidence limit.49 The EPA considers this value to represent the operation of the

unit under conditions reasonably expected to occur at each unit.

The proposed MACT standard was then determined from the average of these

upper 97.5% confidence limits. Once the average of the upper 97.5% confidence

limits was determined, a second statistical treatment was performed to determine the

variation between the units. From this variation, a calculation was made to determine

what the upper 97.5% confidence limit would be for these averaged units.50 This

value was then set as the MACT standard.

The number of facilities averaged depended on the subcategory considered. In

the case of bituminous- and subbituminous-fired facilities, 4 facilities were included

in the MACT standard determination. For lignite-fired facilities, 5 facilities were

included. For waste-coal-fired facilities and integrated gasification combined cycle

(IGCC) facilities, 2 facilities were included.

The data at the extremes of the distribution, i.e. the highest and lowest measurements, play

a predominant role in determining the confidence interval. Therefore, the quality of these

measurements is crucial, since any inaccuracies in this data will be magnified in the

calculation of the confidence interval.

The number of facilities averaged influences the width of this confidence interval. The

greater the number of facilities averaged together, the lower the uncertainty regarding the

distribution of these values. This lower uncertainty translates to a narrower confidence

interval.

CRS-18

The effects of EPA’s statistical treatment of the data is shown graphically in

Figure 2.

Figure 2. Proposed Mercury MACT Relative to Projected Mercury

Emissions for Facilities Used to Set MACT

0.2

0.18

0.16

0.14

0.12

0.1

0.08

0.06

0.04

0.02

10 11 12 13 14 15 16 17 18 19 20

0.25

0.2

0.15

0.1

0.05

0.1

0.2

0.3

0.4