Interim Report of the Committee on Changes in New Source Review Programs for Stationary Sources of Air Pollutants (2005)

In 1977, Congress amended the Clean Air Act (CAA) to include a pair of programs that together are known as New Source Review (NSR). These programs set out requirements that must be met before a large “stationary source” (source with a fixed location) of pollution may be constructed or modified. One of these programs, known as Prevention of Significant Deterioration (PSD), applies to the construction or modification of any “major emitting facility”¹ in “attainment areas”—those classified as attaining the National Ambient Air Quality Standards (NAAQS) established by the U.S. Environmental Protection Agency (EPA) (CAA § 165(a), 42 USC § 7465(a)). The other, commonly called Part D New Source Review (Part D NSR), applies to construction or modification of “major stationary sources” (not the same as a major emitting facility) in “nonat-

tainment areas”—those classified as not meeting the NAAQS² (CAA § 172(b)(5), 42 USC § 7502 (b)(5)). An overview of both NSR programs is provided later in this chapter.

NSR covers modifications of existing sources as well as construction of new sources. The CAA defines a modification as “any physical change, or change in the method of operation of a stationary source” that significantly increases the emissions of air pollution from the source (CAA §111(a)(4), 42 USC § 7411(a)(4)).³ This definition explicitly applies to both the PSD (CAA §169(2)(C), 42 USC § 7479(2)(C)) and Part D NSR (CAA § 171(4), 42 USC § 7501(4)) programs.

EPA regulations governing NSR (40 CFR 51.166 and 40 CFR 52.21)⁴ elaborate on the statutory definition of a modification. In 2002 and 2003, EPA altered those regulations. The first of those revisions, published December 31, 2002 (67 Fed. Reg. 80186 [2002]), altered the rules in five ways:

• Nonutility sources may now calculate annual emissions before the physical change by averaging annual emissions in any 24-month period within the 10 years before the change. This average must be reduced to reflect any tightening of the source’s emission limit since that 24-month period. The previous rules had required those sources to average their annual emissions over the past 2 years

unless the operator could show that a different period was more representative of normal operations.

• Nonutility sources may now calculate annual emissions after a physical change by projecting actual emissions at the source after the change. (This is known as the actual-to-projected-actual approach. Utility sources were already permitted to use this approach.) Increases in emissions that result from increases in demand, rather than from the change, can be subtracted from calculating postchange annual emissions. If using the new methodology would exempt the project from NSR, and it is reasonable to believe that the project might result in a significant emissions increase, the source must track its postchange emissions for 5 years (10 years in some cases). The previous rules had used the source’s postchange potential to emit (its maximum emissions if operated as designed, reduced by any legal limit on the source’s emissions) as the measure of its postchange emissions. (This is known as the actual-to-potential test.) This method is now required only for new units of existing facilities. If a source’s postchange potential to emit significantly exceeds its prechange actual emissions, then the source could escape NSR only by making a binding commitment to never significantly increase actual emissions over prechange levels.⁵

• Sources may seek permission to establish plant-wide applicability limitations (PALs)—a limit on emissions from the plant as a whole-to determine whether a given physical change increases emissions and therefore constitutes a modification. A PAL runs for 10 years and is renewable. Physical changes within the plant do not require an NSR permit so long as the PAL is not exceeded. The PAL’s size for each pollutant is calculated by adding a “de minimis” level to the source’s current emissions (calculated in the same way as prechange emissions). The previous rules had no similar provisions.

• Physical changes at “clean units” need not obtain an NSR permit so long as the changed unit continues to meet its maximum

allowable emissions rate. A clean unit is one that meets emission limits that are equal in severity to those required under the NSR programs. The unit need not have actually passed through NSR so long as its operator presents analyses of its air-quality impacts that are like those required by NSR. The exemption lasts for 10 years after the pollution controls are brought into operation and is renewable if the pollution controls continue to be as strict as those required by NSR. The previous rules had no similar provisions.

• The existing exemption from NSR for pollution control projects (PCPs) is expanded. This exemption applies when a permitting authority deems the project to be environmentally beneficial, even though it would significantly increase emissions of an air pollutant other than the pollutant(s) reduced by the project. Sources no longer must show that reducing air pollution is the primary purpose of the PCP. In addition, the rule lists a number of projects presumed to be eligible for the exemption.⁶ Those projects, along with the pollutant(s) that is controlled, are listed in Box 2-1. Other presumed environmentally beneficial PCPs include activities or projects undertaken to accommodate the following: (1) switching to different ozone-depleting substances with a less damaging ozone-depleting effect (factoring in the ozone-depletion potential and projected usage), and (2) switching to an inherently less-polluting fuel, to be limited to the following:

• Switching from a heavier grade of fuel oil to a lighter fuel oil or any grade of oil to 0.05% sulfur diesel (that is, from a higher sulfur content no. 2 fuel, or from no. 6 fuel, to about 0.05% sulfur no. 2 diesel).

• Switching from coal, oil, or any solid fuel to natural gas, propane, or gasified coal.

• Switching from coal to wood, excluding construction or demolition waste, chemical- or pesticide-treated wood, and other forms of “unclean” wood.

• Switching from coal to no. 2 fuel oil (0.5% maximum sulfur content).

• Switching from high-sulfur coal to low-sulfur coal (maximum 1.2% sulfur content).

In proposing a predecessor version of these rules in 1996, EPA stated that it would allow states to choose between following the old or the new rules. In contrast, EPA’s final rule requires that all states adopt these or more stringent changes. The agency considers its new rules to be more environmentally beneficial than the old rules, and thus states do not have the option of continuing to follow the old rules (67 Fed. Reg. 80241 [2002]). Some state and local air program officials strongly disagree with this position (STAPPA/ALAPCO 2002, 2003). EPA has given the states until January 2, 2006, to submit state implementation plan (SIP)⁷ revisions adopting the new or the more stringent measures. Few such revisions have been submitted so far. (This is not surprising given the many procedural steps involved in developing a SIP revision.) The changes went into effect March 3, 2003, in areas with no approved NSR programs in their SIPs.

In response to petitions for reconsideration, EPA announced in mid-2003 that it would take further comments on the agency’s conclusion (EPA 2002a) that the new rules would benefit air quality (68 Fed. Reg. 44620 [2003]). Later that year, EPA decided to clarify some portions of the 2002 rule change but otherwise to leave them in place (68 Fed. Reg. 63021 [2003]).

EPA promulgated another rule change in October 2003 (68 Fed. Reg. 61248 [2003]). This rule change—known as the equipment replacement rule—defines certain kinds of equipment replacements as “routine maintenance, repair and replacement” and therefore not constituting “physical changes or changes in the method of operation.” Hence these replacements do not need NSR permits even if a significant emissions increase (as calculated under the actual-to-projected-actual technique) can be expected to occur and as long as the source does not exceed its maximum level of allowable emissions (see Box 2-2).⁸ Under the promulgated rule, replace-

ment of components of a process unit with identical components (or their functional equivalents) is exempt from NSR if

• The cost of replacing the component (including the fixed capital cost and costs of the replacement activity, such as construction) is less than 20% of the replacement value of the process unit.

• The replacement does not change the unit’s basic design parameters.

• The unit continues to meet enforceable emission and operational limitations—that is, the unit, while possibly emitting more than in the past because of greater utilization, does not emit more than is legally allowed.

Under the previous rules, EPA used a case-by-case approach in determining which equipment replacements constituted routine maintenance, repair, or replacement. The case-by-case approach continues to be available for a source whose project does not qualify under the categorical exception.

The CAA requires EPA to set NAAQS for pollutants that may reasonably be anticipated to endanger public health or welfare and that come from numerous or diverse sources (CAA § 108(a)(1), 42

USC § 7408(a)(1)). Each NAAQS limits the maximum permissible concentration of a pollutant in the ambient air—that is, the outside air to which the public has access (40 CFR § 50.1(e)). Under section 109 (42 USC § 7409), primary NAAQS must protect the public health with an adequate margin of safety, and secondary NAAQS must protect the public welfare (e.g., effects on crops, other vegetation, and animals). As detailed in Chapter 3, EPA has set NAAQS for carbon monoxide (CO), lead (Pb), ozone (O₃), nitrogen dioxide (NO₂), particulate matter (PM), and sulfur dioxide (SO₂) (40 CFR Part 50).

Much of the CAA consists of mechanisms, including NSR, to control emissions of the NAAQS pollutants and their precursors (e.g., volatile organic compounds [VOCs] are regulated because they contribute to the formation of O₃). The following mechanisms apply to emissions from stationary sources:

• Under section 110 (42 U.S.C. § 7410), each state must prepare and enforce a SIP for the NAAQS pollutants. The SIP must demonstrate that it will result in attainment of the primary NAAQS by a given deadline and of the secondary standards as expeditiously as practicable. SIPs invariably regulate major stationary sources among other sources of air pollution.

• Under section 111 (42 U.S.C. § 7411), EPA establishes new source performance standards (NSPS) for categories of stationary sources that emit air pollution. NSPS emission limits are based on the degree of emission limitation that can be achieved by a source in a category through use of the best demonstrated technology. NSPS (with an exception not relevant here) apply only to sources that commence construction or modification after the NSPS for the category is proposed (CAA § 111(a)(2), 42 USC § 7411(a)(2)).

• Section 112 (42 USC § 7412) requires EPA to establish category-wide standards to limit the emission of hazardous air pollutants (HAPs). Because some HAPs are subsets of PM and VOCs, this requirement results in regulation of pollutants covered by the NAAQS.

• Section 169A (42 USC § 7491) requires EPA to regulate sources whose emissions degrade visibility in mandatory Class I areas (national parks and similar areas) where visibility is an air-quality-related value. (Mandatory Class I areas are more fully discussed later.) EPA has established rules for regulating “plume

blight”—visibility degradation that can be reasonably attributed to one or a few large stationary sources (40 CFR § 51.302). The program includes regulation of major new sources that are located in nonattainment areas (exempt from PSD) but might damage visibility in Class I areas (40 CFR § 51.307).

• Two “cap-and-trade” programs have been established to limit emissions from utility sources. These programs assign a reduction target to a region. Regulated sources are granted allowances in proportion to their historic emissions and are allowed to trade allowances so that the cap level can be achieved at the least cost.

—Utility sources are subject to the acid rain program of Title IV. By 2010, this program phases in an annual cap of 8.97 million tons of emissions of SO₂ from these sources. The program also includes requirements for controlling nitrogen oxide (NO_x) emissions from utilities. Generally, utility sources are issued allowances that equal 1.2 pounds per million British thermal units multiplied by the source’s annual average heat input (a measure of the source’s usage) for the period from 1985 to 1987.

—Section 110(a)(2)(D) requires each state’s SIP to prevent emissions that “contribute significantly” to nonattainment of the NAAQS in other states or that interfere with another state’s program to prevent significant deterioration of its air quality. In 1998, EPA found that NO_x emissions in 22 states interfered with attainment of the O₃ NAAQS in other states. (The number of states was later reduced to 19.) EPA ordered that the states cut back NO_x emissions by 28%, or more than a million tons a year, during the April-October period when O₃ concentrations are at their highest. This target represents the reduction that EPA found could be made through cost-effective measures (those costing less than $2,000 per ton of emissions eliminated) (Michigan v. EPA, 213 F.3d 663 [D.C. Cir. 2000]) (the court’s decision upheld the program). The states may achieve the required reduction as they wish, but because existing utility plants are the most cost-effective sources to control, the brunt of the reductions falls on them. The states have followed EPA’s suggestion to establish cap-and-trade programs to accomplish the reduction at the least cost.

EPA and the Bush Administration have proposed several new programs:

• EPA has proposed rules for regulating existing sources that contribute to regional haze (69 Fed. Reg. 25814 [2004]). The proposed rules would not apply in states covered by the proposed Clean Air Interstate Rule (CAIR) described below.

• In 2002 and 2003, the Bush Administration proposed “Clear Skies” legislation to establish lower caps on utility emissions of SO₂ and NO_x than do the other stationary source programs that currently exist. SO₂ emissions would be capped at 4.5 million tons in 2010—half of what would be allowed by the acid rain program—and at 3 million tons in 2018. This represents a 73% decrease from year 2000 levels. Clear Skies also would reduce NO_x emissions from 5 million tons annually in 2000 to 2.1 million tons in 2008 and 1.7 million tons in 2018—a two-thirds reduction.

Clear Skies has attracted opposition from utility and industrial groups, who say the goals are too stringent, and from environmental groups, who say the goals are too lax. Environmentalists have tended to favor Senator James Jeffords’s proposed Clean Power Act. This rival proposal calls for more extensive and quicker emissions cuts. For instance, the Clean Power Act would cap SO₂ emissions at 2.25 million tons in 2007, compared with the Bush Administration’s proposal for a 4.5-million-ton cap in 2010.

Neither of these proposals has been brought up for formal committee consideration in the Senate or the House of Representatives, and their futures are unclear. In January 2004, EPA proposed (but, at this writing, has not promulgated) the Interstate Air Quality Rule also known as the Clean Air Interstate Rule (69 Fed. Reg. 4566 [2004]). The proposal would adopt elements of Clear Skies by administrative means. EPA proposes to find that SO₂ and NO_x emissions from 29 eastern and midwestern states are interfering with attainment and maintenance of the air-quality standards for O₃ and PM. Emissions of SO₂ from power plants in these states would be capped at 3.9 million tons in 2010 and at 2.7 million tons in 2015. EPA (2004a) estimated that those caps would result in a reduction of 3.6 million tons of SO₂ emissions in 2010 and an additional reduction of 2 million tons per year when the rules are fully implemented (approximately 70% below 2002 levels). Emissions of NO_x from electricity-generating units would be capped at 1.6 million tons in 2010 and at 1.3 million tons in 2015. EPA estimated that NO_x emission would be reduced by 1.5 million tons in 2010 and by 1.8 million tons

in 2015 (about 65% below 2002 levels). As with the 1998 NO_x cut-back, the emissions goals are based on what can be done through relatively cost-effective technology. This proposal has attracted many of the same concerns as Clear Skies, and, like Clear Skies, its outlook is uncertain at this writing.

In general, NSR provides more stringent emission limits for new and modified major sources than EPA provides in other existing programs. The proposed programs are, in their current form, unlikely to change this. The visibility proposal would not affect the current treatment of new and modified sources. Clear Skies would exempt modifications at existing utilities from NSR and would exempt new utility plants from most NSR requirements.⁹ In exchange, the NSPS for power plants would be tightened. But the new NSPS would not apply to all modifications currently covered by NSR. Moreover, as time passes, NSR-control-technology determinations become more stringent than NSPS because of advances in control technology. It is therefore unlikely that Clear Skies would result in emission limits at individual sources that are tighter than those achieved when NSR is triggered at the same sources.

The Clean Air Interstate Rule also is unlikely to require sources to do more than is required by NSR. This would occur only if the emission cap is so tight, and the cost of allowances is so great, that new sources would prefer to reduce their emissions beyond NSR levels rather than purchase allowances. EPA, though, has not projected such an effect of the Clean Air Interstate Rule. It should also be kept in mind that NSR programs require special protection for local areas (e.g., the increment system and the offset requirement) that are not included in cap-and-trade programs.

In general, NSR provides more stringent emission limits for new and modified major sources than do the programs listed here. A very stringent emission cap could provide an extra incentive for future new and modified sources to decrease emissions as well.

In 2001, EPA estimated that PSD control-technology determinations from 1997 to 2001 eliminated a potential increase of 1.4 mil-

lion tons of air pollution annually. About 822,000 tons of these reductions, or approximately 60%, were in NO_x, emissions and about 420,000 tons, or about 35%, were in SO₂ emissions. This estimate does not take into account benefits that occur when source operators limit emissions so that the PSD permit process will not apply. Ninety percent of these benefits are believed to have occurred at new electricity-generating units (EPA 2001).

The NSR programs consist of the PSD program for areas that attain the NAAQS and the Part D NSR program for areas that do not. Each program goes beyond NSPS in two respects. First, each program requires an evaluation of whether a proposed new or modified source can control emissions beyond the requirements of NSPS. This often results in lower emissions because some individual sources can do more than the ordinary source in a category and because technology often advances more quickly than does EPA in revising an NSPS. Second, unlike NSPS, each program gives special attention to sources in or near sensitive areas; the Part D NSR program emphasizes areas that violate air quality standards, while PSD seeks to give special protection to national parklands and areas that are experiencing rapid growth in concentrations of SO_x, NO_x, and PM.

The PSD program was born in the early 1970s. The CAA Amendments of 1970 (Pub. L. 91-604) did not expressly stipulate whether states were required to include in their SIPs measures to prevent the deterioration of air quality that is superior to the NAAQS. EPA Administrator William Ruckelshaus ruled that states had no obligation to do so. The District Court for the District of Columbia overturned this interpretation of the Act (Sierra Club v. Ruckelshaus, 344 F. Supp. 253 [1972]). An equally divided U.S. Supreme Court eventually affirmed this decision without opinion (Fri v. Sierra Club, 412 U.S. 541 ([1973]). As a result, EPA created the PSD program in 1974 to impose requirements for the construction or modification of

major sources in clean air areas (39 Fed. Reg. 42510 [1974]). Congress altered and codified PSD in the CAA Amendments of 1977 (Pub. L. 95-95, § 127). In section 160 of the Act (42 USC § 7460), Congress articulated several goals for the PSD program, such as protection of national parks and the prevention of health and welfare effects that can occur at levels allowed by the NAAQS (see Box 2-3). Later, by technical amendment, Congress made clear that the PSD program, as in EPA’s 1974 rules, covers modifications (Pub. L. 95-190, § 14(a)(54) (adding CAA § 169(2)(C), 42 USC § 7479(2)(C)) (defining the term “construction” to include “modifications”).

The PSD program requires a permit for the construction or modification of a “major emitting facility” that is located in an area to which the program applies (CAA § 165(a), 42 USC § 7475(a)) (see Box 2-4). The program applies in every area that attains at least one of the NAAQS (Alabama Power v. Costle, 636 F.2d 323, 364-368 (D.C. Cir. 1980), (CAA § 165(a), 42 USC § 7475(a)). Because no area violates all the NAAQS, PSD therefore applies nationwide. (It does not apply, however, to emissions of those pollutants for which the source’s locale does not attain the NAAQS.) If a source is in 1 of 28 named categories, it is a major emitting facility if its “potential to emit” any regulated air pollutant is 100 tons per year or more. Otherwise, the source is covered if its potential to emit is 250 tons per year or more (CAA § 169(1), 42 USC § 7479(1)). Potential to emit generally represents the source’s emissions if operated at maximum design capacity. A source may reduce its potential to emit by agreeing to a legally binding limit on its emissions (40 CFR § 51.166(b)(4)). If the source agrees to a limit that reduces its potential to emit below the coverage thresholds, it is no longer a major emitting facility and therefore is exempt from the program.

A modification of a major emitting facility is covered if it would increase the net emissions of a pollutant by a “significant amount” (see Box 2-2). Significance levels are based on the levels at which, in EPA’s view, the benefits from regulation would be de minimis.

An applicant for a PSD permit must show that the new or modified facility will, for each regulated pollutant emitted in significant amounts, limit emissions to the level achievable through use of the best available control technology (BACT). The BACT determination is made on a case-by-case basis and must be at least as stringent

as whatever NSPS exists for the source’s category (CAA § 169(3), 42 USC § 7479(3)). Since 1987, EPA has stated that BACT must be set at a level at least as stringent as a top-down approach: that is, BACT should be set at the most stringent level achieved by a source in the same category unless the applicant can show that level to be unachievable (61 Fed. Reg. 38250, 38272-38273 [1996] [proposing to formally incorporate this approach into EPA’s rules]).

The applicant also must show that the new or modified source, in combination with emission increases from other sources, will comply with a system of increments that limit permissible growth in air pollutant concentrations over the baseline concentration—the

concentration that existed in an area when the first application was filed for a PSD permit (CAA § 169(4), 42 USC § 7479(4)). Increments currently exist for NO₂, SO₂, and PM (see Box 2-5). The size of the increments varies with the area’s classification as Class I, II, or III. The loosest increments apply in Class III areas. The tightest increments apply in Class I areas, thus tending to shift new sources away from those areas. Congress designated 158 large national parks and wilderness areas existing in 1977 as mandatory Class I areas. Most of those areas are west of the Mississippi River; nearly one-quarter of them are in Utah, Arizona, New Mexico, or Colorado (Oren 1989). The remainder of the nation initially was classified as Class II. States and Indian tribes are allowed to redesignate areas as Class I or III. To date, a few Indian tribes have designated their reservations as Class I; no areas have been designated as Class III. The many national parks and wilderness areas created since 1977 (e.g., those created by Congress in 1980 in Alaska) have remained Class II areas.

As of June 2003, 39 states have obtained EPA approval for their SIP submissions incorporating the PSD program. In some of the remainder (e.g., New York), EPA runs the program itself, but delegates its responsibility to the state for most of the day-by-day

decisions. There are, though, jurisdictions where EPA has not delegated authority.

The 1977 amendments also included a NSR program for nonattainment areas—those whose air quality does not meet the NAAQS. This program also applies to major sources of VOCs in the ozone transport region in the northeast, even if located in attainment areas (CAA § 184(b)(2), 42 USC § 7511c(b)(2)).

In 1970, Congress required attainment of the primary standards no later than 3 years after approval of the state’s SIP, with a possible 2-year extension. It became clear that this timetable would not be met. In late 1976, EPA published an interpretive ruling that outlined conditions under which new and modified major sources would be allowed in areas that failed to attain the air-quality standards on schedule (41 Fed. Reg. 55524 [1976]). Congress codified this program (known as Part D NSR because it is included in that part of the Act) in 1977 as part of a renewed effort to bring about attainment of the NAAQS (Pub. L. 95-95, § 129).

As with PSD, Part D NSR requires that the operator obtain a permit before construction or modification of a major stationary source (see Box 2-6). Generally, a major stationary source is defined as one that emits or has the potential to emit more than 100 tons per year of any air pollutant (CAA § 302(j), 42 USC § 7602(j)). These size cutoffs are lower for NO_x and VOCs in O₃ nonattainment areas. For instance, in areas designated as “extreme” (currently, the Los Angeles area), the threshold is set at 10 tons per year. A major source is covered only to the extent that its emissions could contribute to nonattainment of a NAAQS. Thus, if the proposed source would emit two pollutants, and the area violates the NAAQS for only one, then the source is covered by Part D NSR for that pollutant and by PSD for the other.

A modification is covered under the Part D NSR program if it is “major”—that is, it would in itself require a permit if it were newly constructed (40 CFR § 51.165(a)(2)(I)) and if it would contribute to nonattainment of the NAAQS.

To obtain a permit to construct or modify, the operator must show that the new or modified source will emit at the lowest achiev-

able emission rate (LAER), defined in the statute as the more stringent of the tightest emission limit achieved in practice or the tightest SIP limit for that category of source (unless the operator can show that level not to be achievable) (Section §173 (a)(2), 42 USC § 7503(a)(2)). The applicant must also (except in limited cases) obtain emissions offsets—emission reductions from other sources that are enforceable and not otherwise required, so that the construction or modification will not disrupt progress toward attaining the NAAQS. In areas that do not attain the ozone standard, the offset must be greater than one to one, with the exact magnitude depending on the severity of the nonattainment.

The controversy about modifications stems from the CAA’s differentiation between new and existing stationary sources. As mentioned previously, new sources must meet technology-based standards in addition to showing that they will not damage air qual-

ity. In contrast, most existing sources need accomplish only as much emission reduction as is necessary to enable their locales to meet and maintain the levels of the NAAQS set by EPA.

This differentiation has attracted controversy. Supporters assert that it is justified because new sources can most easily incorporate the latest pollution control technology. In addition, supporters argue, tight regulation of new sources is the best way to ensure against future air pollution problems and that the turnover of capital stock results in reduced emissions. Critics argue that the differentiation between old and new sources encourages industry to keep older, heavily polluting sources on line longer instead of building new sources. Emissions trading advocates urge that it would be preferable to allow trading between sources, whether new or existing, to achieve the needed emission reductions. This approach is reflected in the Bush Administration’s proposal, so far unsuccessful in Congress, that the emissions from electric utility plants be capped and that NSR requirements be relaxed for new and modified utility plants. Opponents of this proposal argue that a trading approach by itself would not be sufficient to protect especially vulnerable areas from large new sources.

Existing plants pose an even more difficult question. Plants where changes are occurring are on the boundary between new and existing sources. For instance, inserting state-of-the-art technology when a source experiences a change is, at least sometimes, more problematic than including such controls at a new plant. Nonetheless, plants where changes are occurring often may be better targets for regulation than unaltered existing sources. For instance, changes at existing plants, if unregulated by NSR, might keep such plants on line longer and slow their replacement by new, cleaner facilities. Slowing replacement of existing plants may give a competitive advantage to those plants over new plants, thus perpetuating high emission levels. Also, adding control technology to an existing source when it is undergoing modification may well be easier than installing such controls at an existing source that is not undergoing modification.

These issues are reflected in the different viewpoints about how the term “modification” should be defined.

Environmental groups argue that a broad definition is needed because of the following:

• Health and the environment are endangered when existing sources increase actual emissions.

• Narrowing the definition would interfere with enforcement actions that are permanently lowering emissions and thus bettering air quality.

• Congress intended a broad definition of modification as a way of ensuring that older sources eventually would have to install the best available control technology.

• A narrow definition of the term “modification” would cover renovations that allow existing sources, particularly electricity-generating units, to remain in operation indefinitely through renovation. A broader interpretation would discourage those renovations and instead would lead to replacing those plants with new capacity that would be far cleaner than existing plants.

Industry groups counter by saying that a narrower definition is appropriate because

• Many projects that would be covered under a broad view of “modification” do not increase emissions and in fact reduce them by replacing older equipment with less-polluting replacements.

• The programs are quite complex, and it is difficult to determine whether an NSR permit is required for a given change.

• Preparing a permit application, obtaining any needed offsets, waiting for EPA or state officials to process the application, and complying with BACT for the modification may be costly and burdensome. The process of reviewing the application takes additional time.

• Other programs, such as caps on utility emissions, can constrain emissions at a lower cost than a stringent NSR program.

Congress’s 1977 codification of PSD and Part D NSR made it necessary for EPA to revise its rules governing the programs. This was done via a rulemaking in 1978. The following year, the D.C. Circuit Court of Appeals in the Alabama Power decision overturned several important portions of the rules. EPA then promulgated new rules in 1980 (45 Fed. Reg. 52676 [1980]).

The 1980 rules cover both the “physical or operational change” aspect and the “any increase in emissions” aspect of the congressional definition of modification. The term physical or operational change was defined to exclude routine maintenance, repair, and replacement, a term that the regulations did not elaborate upon. An increase in emissions is defined in terms of an increase in actual emissions, taking into account contemporaneous increases and decreases in emissions. Emissions before the change were specified to mean the average emissions at the source over the previous 24-month period unless the source could show that a different consecutive 24-month period was more representative of normal source operation.

The 1980 regulations subjected postchange emissions to a special meaning of the term “actual emissions.” If a unit had not yet entered normal operations, then the term actual emissions was defined as equal to the unit’s potential to emit. Thus, for units that had not entered normal operations, the 1980 rules covered a physical change as a modification if the source’s postchange potential to emit exceeded the source’s prechange actual emissions by a significant (or, in the case of Part D NSR, a major) amount. This is commonly referred to as the “actual/potential” test. Because a source’s potential to emit is often significantly greater than its actual emissions, an actual/potential test tends to lead to coverage of a project by NSR. The plant can escape coverage only by making a binding promise to never increase actual emissions significantly over prechange levels.

The 1989 case of Puerto Rican Cement v. EPA illustrates the workings of the test. Puerto Rican Cement had been running its kilns at 60% capacity and emitting 1,100 tons of NO_x and 1,340 tons of SO₂ annually. The company planned to build a new cement kiln. If operated at the same 60% capacity as the older unit, the new kiln would emit 578 tons of NO_x and 850 tons of SO₂ annually. But if the unit operated at full allowable capacity, it would emit 1,250 tons of NO_x and 1,927 tons of SO₂ annually. EPA compared the latter numbers with the prechange annual tonnages of 1,100 and 1,340, respectively, and ruled that construction of the new kiln would increase emissions within the meaning of its 1980 regulations and therefore a PSD permit was required.

The U.S. Court of Appeals for the First Circuit upheld this approach in a decision written by now-Supreme Court Justice Stephen Breyer (Puerto Rican Cement v. EPA, 889 F.2d 292 [1st Cir. 1989]). The court rejected the company’s argument that EPA’s approach was arbitrary because it would discourage modernizations that decreased

emissions at the present rate of use. Instead, the court wrote, EPA had decided to focus on the possibility that the introduction of new, more efficient equipment would lead a company to produce at higher levels and therefore increase emissions. Hence, the company needed a PSD permit for the new kiln unless it was willing to reduce the unit’s potential to emit by making a binding commitment never to increase emissions by more than a de minimis amount over the prechange levels.

The court acknowledged, though, that in some situations EPA’s actual/potential test might be unreasonable. For instance, the court suggested, it might be irrational to assume that a replaced peak-load generator would run at its full potential to emit.

This observation became important in the 1990 Wisconsin Electric Power (WEPCO) decision (Wisconsin Electric Power v. Reilly, 893 F.2d 901 (7th Cir. 1991)), which involved a so-called “life-extension project” at WEPCO’s Port Washington plant. The plant consisted of five coal-fired steam generating units placed in service between 1935 and 1960. Over time, each had deteriorated from its design capacity of 80 megawatts, and one unit had been shut down because of the risk of catastrophic failure. The aim of the project was to keep the units operating beyond their original 1992 retirement date until 2010. As part of the project, WEPCO planned to replace air heaters, steam drums, and other major components on four units.

EPA held that a PSD permit was required on the grounds that the project did not constitute routine maintenance, repair, and replacement, and that the project would increase emissions using the actual/potential test of EPA’s NSR regulations. Hence, the dispute involved the physical change as well as the emissions increase aspect of modification. The utility appealed to the U.S. Court of Appeals for the Seventh Circuit, which decided for EPA on the physical change issue, while holding that the project would not increase emissions for NSR purposes.

WEPCO argued that “like-kind” changes—the replacement of existing equipment—do not constitute a physical change because they do not alter the plant. EPA argued that the WEPCO project was unprecedented: “WEPCO did not identify, and EPA did not find, even a single instance of renovation work at any electric utility generating station that approached the Port Washington life extension project in nature, scope or extent” (Wisconsin Electric Power v. Reilly, 893. F.2d 901 [7th Cir. 1991], p. 911). The court agreed with EPA, saying that a contrary reading contravened Congress’s intent to stimulate the advance of pollution

control techniques by requiring controls when generating systems are extensively replaced.

The court also rejected WEPCO’s argument that its project qualified as routine maintenance, repair, and replacement under EPA’s rules. The court noted that EPA stated that it “makes a case-by-case determination by weighing the nature, extent, purpose, frequency and cost of the work, as well as other relevant factors, to arrive at a common-sense finding.” In this case, EPA had relied on the magnitude of the project, the exclusion from the project of repetitive maintenance normally performed during outages, and the lack of similar projects at other power plants. The court held that EPA could use these factors and that EPA had reasonably applied them to the facts of the case. The court noted that WEPCO had stated that its project involved a life extension and that the project would involve items that normally would occur only once or twice during a unit’s expected life cycle; this, according to the court, supported EPA’s finding that the project did not constitute routine maintenance, repair, and replacement.

But the court did not accept EPA’s argument that the project required a permit because it would lead to an increase in emissions. The court distinguished between the NSPS program and the NSR programs. In the NSPS program, EPA’s regulations call for a comparison of prechange and postchange emission rates, as expressed in kilograms per hour, at maximum physical capacity (40 CFR § 60.14 (b); 57 Fed. Reg. 32314, 32316 [1992]). Because such an increase had occurred at three of the five units, the project needed to comply with the NSPS for the source’s category for those units.

The court held that, by contrast, a project is subject to NSR if annual emissions increase. The court overturned EPA’s use of the actual/potential methodology to determine whether an increase would occur. The court found it unreasonable for EPA to disregard past operating conditions at the plant and to regard the units as having never entered normal operation. The Puerto Rican Cement case was distinguished as involving a new unit at an existing site, unlike a “like-kind replacement” of equipment at an existing unit. The latter, the WEPCO court ruled, resembled the peak operating unit example in the Puerto Rican Cement decision.

EPA responded to the decision with what is known as the WEPCO rule (57 Fed. Reg. 32314 [1992]). Under this rule, utility units are excluded from the actual/potential test so long as the proposed project nei-

ther adds a new unit nor replaces an existing one. Instead, utility units may compare prechange actual emissions with postchange projected annual emissions. If the utility concludes that there would be no significant increase in emissions, thereby exempting the project from NSR, then the utility must submit documentation of its emissions for 5 years after the change to confirm that a significant increase in emissions did not occur as a result of the project. In addition, the calculation of postchange emissions may exclude emission increases attributable to increased market demand rather than to the physical change. This exclusion could apply to increases that legally and physically would have been feasible without the change.

EPA also altered the definition of prechange emissions for utility sources. Before the alteration, prechange emissions were calculated for any source by averaging emissions over the 2 years before the change unless the source could show that a different 2-year period was more representative. EPA changed this rule to allow utility sources to use any consecutive 2-year period in the past 5. In the preamble to the WEPCO rule, EPA promised guidance on what is “routine maintenance, repair, and replacement” (57 Fed. Reg. 32326 [1992]).

Finally, EPA excluded from the definition of physical change, and hence from NSR review, pollution control projects that “do not render the unit less environmentally beneficial.” In this way, EPA asserted, the prospect of NSR review would not influence a utility’s choice of how to cut emissions to comply with the acid rain control program that had been adopted by Congress in 1990.

These changes were confined to electricity-generating units because EPA believed that it did not have enough knowledge of other source categories to allow the changes to be extended to them. EPA later, however, issued guidance that extended the pollution control project exemption to nonutility source categories (67 Fed. Reg. 80232 [2002]).

In 1992, EPA also launched an effort to simplify and streamline its NSR rules. EPA formed a subcommittee of its existing Clean Air Act Advisory Committee, composed of representatives of states, environmental groups, and industries. For several years, the task force members discussed possible changes in the rules. The task force did not achieve

consensus, but, in EPA’s view, its existence helped to make clear which issues were important.

In 1996, EPA published in the Federal Register a package of proposed changes to the NSR rules (61 Fed. Reg. 38250 [1996]). This proposal discussed the areas covered in the 2002 rule (e.g., expanded use of the actual-to-projected-actual methodology), although the 2002 rule differs in important respects. The proposed changes also included elements sought by environmental groups, such as greater protection for national parks and codification of the long-standing top-down method of determining what constitutes the best available control technology.

The 1996 proposal generated a great deal of comment. In 1998, EPA issued a “notice of availability” in which EPA expressed reservations about some of its proposed changes (e.g., the enforceability of the actual-to-projected-actual approach and of its proposal to allow sources to exclude emission increases due to demand growth in projecting future emissions) and solicited comment on possible alternatives (63 Fed. Reg. 39857 [1998]). EPA did not complete the rule-making process before the end of the Clinton Administration.

EPA’s 1996 proposal did not discuss the issue of which changes constitute routine maintenance, repair, and replacement. This issue became increasingly important in the late 1990s. EPA, frequently joined by environmental groups and northeastern states, asserted that some large utility plants had been undertaking modifications without obtaining NSR permits. According to EPA, these projects allow utilities to run the altered plants at higher capacity levels and therefore increase emissions from them. In addition, EPA claimed that the projects allow the plants to remain on line longer instead of being replaced by new, cleaner plants that would decrease emissions substantially from present levels. Utilities, on the other hand, contend that these projects should be considered to constitute routine maintenance, repair, and replacement and therefore exempt from NSR. The projects, according to utilities, have always been undertaken in the industry and are necessary to ensure adequate reliable generating capacity. A report by the National Coal Council states that coal-fired power plants more than 20 years old—a category that accounts for two-thirds of electricity generation from coal—have been derated—

reduced in power-generating capacity—and that a substantial amount of generation capacity (approximately 20,000 megawatts) could be regained by addressing the causes of derating (EPA 2001).

The cases brought against utilities by EPA and their status are listed in Table 2-1, as of July 2, 2004. In addition, EPA brought administrative enforcement actions against the Tennessee Valley Authority (TVA) and WEPCO. (The action against TVA eventually was judicially invalidated on procedural grounds [Tennessee Valley Authority v. Whitman, 336 F.3d 1236 {11th Cir. 2003}]); environmental groups have since brought actions against TVA on the same substantive grounds.) The actions regarding TVA, WEPCO, and those listed in Table 2-1 alleged that the companies had undertaken major modifications without obtaining NSR permits.

Several other actions have been settled, as shown in Table 2-2 from EPA.¹⁰ As the table shows, the settlements, when fully implemented in 2010-2015, will reduce SO₂ emissions by more than 440,000 tons per year. (It is possible that some of these reductions might have been required by other programs under the CAA.) These reductions are permanent; that is, they reduce emissions to levels lower than the 8.95-million-ton national cap. Each settlement agreement includes a provision requiring the source to annually surrender SO₂ emission allowances. Surrendered allowances are retired by EPA, and so they become unavailable for use by other sources. In this way, the settlements will reduce SO₂ emissions below the present 8.95-million-ton cap described above. Similarly, the required reductions in NO_x may not be used to generate NO_x credits that can be sold to other sources.

Generally, the number of allowances surrendered annually equals the number of tons by which the settlement reduces the source’s allowable SO₂ emissions. There are two exceptions. First, if the source already had more allowances than its presettlement emissions, then the source must surrender these surplus allowances in addition to the allowances represented by the settlement. If, on the other hand, the settlement would reduce a source’s allowable emissions below its annual allowances, then the company need surrender only the difference between presettlement emissions and its allowances. (The Virginia Electric Power Company [VEPCO] settlement summarized below is an example of this.)

The number of allowances surrendered might well be reduced if Congress or EPA lowers the SO₂ allowance cap of 8.95 million tons (e.g., through the CAIR or Clear Skies initiatives).

Table 2-2 also shows estimates of the capital cost of the required reductions. These costs are stated in today’s dollars and so are not discounted for the value of postponing an investment for several years. Moreover, these cost figures are not annualized. Therefore, they cannot be used to generate cost-effectiveness numbers. According to EPA enforcement personnel, the costs of the reductions are approximately $500 per ton for SO2 and $900 to $3,000 per ton for NO_x. These costs are comparable to those for installing BACT generally.

The settlement between EPA and the VEPCO illustrates the kind of steps that an agreement to end the enforcement litigation may involve. There, the company has committed itself, among other things, to install scrubbers on 70% of its coal-fired generation capacity. That step will reduce SO₂ emissions by 176,500 tons per year by 2012. The company also will install by 2013 selective catalytic reduction technology on 67% of its coal-fired generation capacity, thereby reducing annual NO_x emis-

sions by 66,000 tons more than current EPA rules require. In addition, the company will surrender to EPA 45,000 allowances per year beginning in 2012 (EPA 2003a). Settlements can affect emissions significantly in an area. A settlement by EPA and New Jersey, with a utility plant on the outskirts of Trenton and another in Jersey City, will reduce SO₂ and NO_x emissions in the state by 19% and 5%, respectively, by 2012 (EPA 2002b).

EPA enforcement officials have asserted that the cases they are currently pursuing could reduce annual SO₂ and NO_x emissions in 10 years by 1,750,000 tons and 629,000 tons, respectively (EPA 2004b). It is not

clear what the actual reductions would be. Of the pending actions, so far two have resulted in decisions by U.S. district courts on the merits (U.S. v. Ohio Edison Co. 276 F. Supp. 2d 819 (S.D. Ohio 2003); U.S. v. Duke Energy Corp, 278 F.Supp. 2d 619 (M.D. N.C. 2003)). These cases, although decided only 2 months apart, came to opposite conclusions on

• How to determine whether an increase in emissions has occurred. Ohio Edison allows the use of the actual-to-projected-actual method for determining whether a utility unit has increased emissions, while Duke Energy holds that an increase in the hourly emission rate also must have occurred.

• The test for deciding whether a particular change constitutes routine maintenance, repair, and replacement. Ohio Edison holds that the test hinges in part on whether a particular activity is routinely conducted at an individual plant, while Duke Energy looks to whether the activity is routine in an industry. Thus, the former decision, but not the latter, covers an activity that is performed only once or twice in the life of a plant but is performed at most plants in an industry.

• Who has the burden of proof in showing whether a change constitutes routine maintenance, repair, and replacement. Ohio Edison holds that the burden is on the source, while Duke Energy places it on the government.

Because both of these decisions were rendered by district courts, neither has superior status as a statement of the law, and so the eventual resolution cannot be predicted. Even if EPA’s enforcement theory is correct, the change in regulation may well make it more difficult for EPA to persuade companies to settle or to obtain favorable judgments. As EPA has pointed out, the new rule is not retroactive, and therefore enforcement actions based on the previous rules may proceed. But, as a practical matter, a court might be reluctant to find liability, or to impose a substantial penalty, for violating rules that are no longer in force. Hence, the rule changes have the potential to diminish the size of future settlements. Again, estimating the magnitude of this change is difficult.

It is important to note that only enforcement actions against violators result in the surrender of allowances. If a utility complies without the need for an enforcement action (the utility requests a permit for changes that EPA would not consider to be routine maintenance, repair, and replacement), then it need not surrender allowances. Instead compliance by the utility would reduce the utility’s need to find additional al-

lowances to accommodate the emission increase that would have been caused by the change. These allowances would then be available to other sources, thus perhaps resulting in increases in emissions from those sources that might offset, in part or in whole, the effect of compliance.

Shortly after his inauguration, President Bush directed Vice President Cheney to develop a national energy policy. In May 2001, the Vice President issued a report on behalf of the National Energy Policy Group (2001).

A report from that group in 2001 recommended that legislation be introduced to cap utility emissions of NO_x and SO₂ at utility plants.¹¹ The report also recommended that EPA, in conjunction with the Department of Energy and other federal agencies, examine EPA’s NSR regulations (including their administrative interpretation and implementation) and report on the impact of NSR on investment in new utility and refinery generation, energy efficiency, and environmental protection.

EPA undertook a 90-day review of the NSR program and in June 2001 issued a background report (EPA 2001) that became the basis for a report to President Bush in May 2002 (EPA 2002c). These reports concluded that the overall benefits of NSR are significant and that NSR has not substantially impeded the construction of new electricity-generating units or refineries. In contrast, the reports found that NSR “has impeded or resulted in the cancellation of projects which would maintain and improve reliability, efficiency and safety of existing energy capacity” (emphasis added) (EPA 2002c). In some cases involving refineries and other industries, the reports found, the impeded projects either would not have increased air pollution or would have decreased air pollution. These conclusions were based largely on anecdotal reports from companies that lack specifics about the projects (GAO 2003).

The report to the President recommended that changes in EPA’s NSR regulations be made. EPA followed up on this recommendation by promulgating the December 2002 rule summarized above. This rule has

been challenged by some states and environmental groups before the D.C. Circuit Court of Appeals.

On the same day that EPA promulgated the NSR revisions summarized above, EPA proposed changes in its rules defining what projects constitute routine maintenance and are exempt from the NSR program. A modified version of the proposal, summarized above, was promulgated in August 2003 and published in October 2003. This rule has also been challenged before the D.C. Circuit Court of Appeals by some states and environmental groups. In December 2003, 2 days before the rule was to become effective in areas where EPA administers the PSD program, the D.C. Circuit Court issued a stay of the rule on the grounds that the petitioners had shown irreparable harm and the likelihood of success on the merits. This stay remains in place and so the new rule has not gone into effect.

On June 30, 2004, EPA announced a 180-day period for reconsideration of the rule. The agency requested comment on the rule’s legality and on the choice of the 20% threshold (69 Fed. Reg. 40278 [2004]).

Following the regulatory overview of the NSR programs presented in this chapter, Chapter 3 examines contributions that emission sources subject to NSR may make to ambient air quality and relationships between specific air pollutants and health effects.