In these consolidated petitions for review, we address approximately thirty challenges to three regulations promulgated by the United States Environmental Protection Agency (EPA or Agency): (1) the “Major Boilers Rule,” 1 (2) the “Area Boilers Rule,” 2 and (3) the “Commercial/Industrial Solid Waste Incinerators (CISWI) Rule.” 3 Collectively, these rules — all promulgated under the Clean Air Act (CAA or Act), 42 U.S.C. §§ 7401 et seq. — set emissions limits on certain combustion machinery known to release hazardous air pollutants (HAPs). Roughly one-half of the challenges are advanced by a group of municipal-electric organizations, industrial-trade associations, oil-and-gas industry representatives, and other entities that own and operate boilers, process heaters, and incinerators (Industry Petitioners). The other one-half are pressed by organizations interested in safeguarding the environment (Environmental Petitioners).
The three rules at issue address a common phenomenon: when combustion occurs, emissions result. The emissions include numerous materials, some of which pose risks to the environment in general and to human health in particular. Because combustion is an inevitable occurrence in the machinery that helps to power modern society, the Congress has authorized the EPA to provide for a regulatory framework that minimizes the deleterious effects of the incineration industry while simultaneously allowing it to operate.
In 2013, the EPA finalized its efforts to do so for discrete types of combustion machinery: boilers, process heaters, and incinerators. Two of the three rules at issue — the Major Boilers Rule and the Area Boilers Rule — govern boilers and process heaters. The former are enclosed devices that use a controlled flame to heat water and convert it into steam or hot-water. 40 C.F.R. § 63.11237. The latter are also enclosed devices that use a controlled flame but, instead of generating steam, they indirectly heat a “process material,” whether liquid, gas, or solid, or a “heat transfer material” like glycol or a mixture of glycol and water. Id. For simplicity, our use of “boilers” covers both machinery types.
The two boiler-specific rules further divide the machinery into three categories: industrial, commercial, and institutional.
See
2011 Area Boilers Rule,
The third rule that we address — the CI-SWI Rule — governs combustion machinery known as “solid waste incineration unit[s].” 42 U.S.C. § 7429. The Act defines an incinerator as a “distinct operating unit of any facility” that burns solid waste from either commercial establishments, industrial establishments, or the general public.
Id.
§ 7429(g)(1). An incinerator subjects “waste material” to “high temperatures until it is reduced to ash.” Incinerator, New OxfoRD AmericaN Dictionary 853 (2d ed. 2005). Incinerators fall into different subcategories and, in the past, the EPA has issued rules governing many of them, including,
e.g.,
municipal solid-waste incinerators, medical-waste incinerators, and sewage-sludge incinerators.
4
At issue in the CISWI Rule are incinerators located in commercial or industrial facilities that combust solid waste as defined in the Resource Conservation Recovery Act (RCRA), 42 U.S.C. §§ 6901
et seq. See
2011 CISWI Rule,
A. The Clean Air Act, 42 U.S.C. §§ 7401 ET SEQ.
Enacted “to protect and enhance the quality of the Nation’s air resources so as to promote the public health and welfare and the productive capacity of its population,” 42 U.S.C. § 7401(b)(1), the Act has been amended several times since the Congress first attempted to control air pollution via legislation in 1963. In 1970, the Congress required the EPA to identify and publish a list of HAPs, which the CAA defined as substances that increase “mor
The risk-focused approach to capping HAP emissions left something to be desired. “In light of unrealistic time frames and scientific uncertain[t]y over which substances posed a threat to public health,” the EPA “only listed eight pollutants as hazardous between 1970 and 1990,”
Nat Res. Def. Council v. EPA (NRDCII),
1. 42 U.S.C. § 7412 — “Hazardous Air Pollutants”
The 1990 CAA Amendments overhauled the Act’s “Hazardous Air Pollutants” provision, codified at 42 U.S.C. § 7412. Although earlier iterations of the Act had assigned HAPs-identification responsibility to the EPA, the slow pace at which the EPA discharged its duty prompted the Congress to create a list of pollutants itself.
5
See Sierra Club I,
a. Identifying and Categorizing HAP Sources
The EPA’s first task is to create HAP-source categories and subcategories. See 42 U.S.C. § 7412(c). The Act distinguishes “major” from “area” sources, defining the former as “any stationary source or group of stationary sources” that neighbor each other, share common control, and emit (or have the potential to emit) either ten tons per year or more of any single HAP or twenty-five tons per year or more of any HAP combination. 6 Id. § 7412(a)(1). The latter are sources that do not emit enough HAPs to qualify as “major.” Id. § 7412(a)(2). Although the EPA must set stringent restrictions on major sources, it has discretion to set more lenient emissions caps on area sources. See id. § 7412(d)(5).
Apart from the statutory distinction between major and area sources, the EPA has discretion to differentiate “among
b. Setting Emission Standards for Major Sources — the “MACT” Standard
After the EPA identifies HAP-source categories and subcategories, it then sets emissions limits for each.
See id.
§ 7412(d)(2). “[Whenever ... feasible,” the caps must use numeric HAPs limits.
Id.
§ 7412(h)(4). The size of the source— either “major” or “area” — dictates whether the EPA must set the numeric limit at the most stringent level that current technology allows or at the level set by “generally available control technologies.”
Id.
§ 7412(d)(5). For major sources, the CAA directs the EPA to establish emissions caps that result in the “the maximum degree of reduction in emissions” that the EPA determines is “achievable.”
Id.
§ 7412(d)(2). We refer to an emissions cap that reflects the current “maximum achievable control technology” as a “MACT” standard.
See NRDCII,
First, the EPA establishes a “MACT floor” for each category or subcategory.
Sierra Club I,
Second, the EPA must determine whether current technology makes it possible for a source to perform even better than the best performing similar source or sources. In other words, the CAA directs the EPA to consider whether it should set a “beyond-the-floor” MACT standard.
Nat’l Lime Ass’n,
c. Setting Emission Standards for Area Sources — the “GACT” Standard
Although the EPA must cap HAP emissions from major sources at the “maximum degree of reduction,” see 42 U.S.C. § 7412(d)(2), it has discretion to set less stringent caps on emissions from area sources. Indeed, the EPA need not list categories of area sources at all unless: (A) it finds that the sources in that category or subcategory “present[] a threat of adverse effects” to the environment or human health, see id. § 7412(c)(1), (3); or (B) control of a particular area source category or subcategory is necessary to ensure that sources accounting for at least 90 per cent of the aggregate emissions of the thirty HAPs the EPA believes “present the greatest threat to public health in the largest number of urban areas” are subject to CAA control, id. § 7412(c)(3), (k)(3)(B). If it finds that controlling emissions from a particular area source subcategory is necessary to achieve a 90 per cent reduction in the aggregate emissions of any of seven CAA-enumerated HAPs, section 7412(c)(6) requires the Agency to impose MACT caps on that subcategory. See id. § 7412(c)(6).
With the exception of section 7412(c)(6)’s MACT-standard requirement, the EPA need not cap emissions from area sources at the MACT level. Instead, it may set more lenient emissions limits based on “generally available control technologies.” Id. § 7412(d)(5). We refer to these caps as GACT standards. The Act provides no guidance for setting GACT standards but the legislative history of the 1990 CAA Amendments describes GACT “as methods, practices and techniques [that] are commercially available and appropriate for application by the sources in the category considering economic impacts and the technical capabilities of the firms to operate and maintain the emissions control systems.” S. Rep. No. 101-228, at 171 (1989). According to the EPA, it can and will consider the following in setting a GACT standard:
• “costs and economic impacts ..., which [are] particularly important when developing regulations for source categories that may have many small businesses ... ”;
• “the control technologies and management practices that are generally available to the area sources in the source category”;
• “the standards applicable to major sources in the analogous source category to determine if the control technologies and management practices are transferable and generally available to area sources”; and
• “technologies and practices at area and major sources in similar categories to determine whether such technologies and practices could be considered generally available for the area source categories at issue.”
2011 Area Boilers Rule,
d. Work-Practice and Management-Practice Standards
Although the CAA requires numeric emission standards where possible, the EPA can “promulgate a design, equipment, work practice, or operational standard, or combination thereof’ if it determines that a numeric limit is “not feasible.” 42 U.S.C. § 7412(h)(1).. In oth
(A) a hazardous air pollutant or pollutants cannot be emitted through a conveyance designed and constructed to emit or capture such pollutant, or that any requirement for, or use of, such a conveyance would be inconsistent with any Federal, State or local law, or
(B) the application of measurement methodology to a particular class of sources is not practicable due to technological and economic limitations.
Id. § 7412(h)(2).
Similarly, for area sources, the EPA can impose a “management-practice standard” in lieu of a numeric GACT standard. See id. § 7412(d)(5). A management-practice GACT standard is like a work-practice MACT standard in all ways but one — the EPA need not consider feasibility when setting management-practice standards. Compare id. § 7412(d)(2), with id. § 7412(d)(5).
2. 42 U.S.C. § 742 — “Solid Waste Combustion”
In addition to amending the Act’s “Hazardous Air Pollutants” provision,
see id.
§ 7412, the 1990 CAA Amendments added to the U.S. Code section 7429, titled “Solid Waste Combustion.” Section 7429 regulates “solid waste incineration units” generally,
see id.
§ 7429(a)(1)(A), and CI-SWI specifically,
see id.
§ 7429(a)(1)(D). Although section 7412 requires the EPA to control emissions of nearly two hundred HAPs,
see id.
§ 7412(d)(1), section 7429 mandates that the EPA control emissions from only nine specific'pollutants (as well as opacity, where appropriate), none of which the Congress included on its initial section 7412 list,
see id.
§ 7429(a)(4);
see also Nat. Res. Def. Council v. EPA (NRDC I),
Whether a source falls under section 7412 or section 7429, “the statutory directive on setting MACT standards is virtually identical.”
Id.; see also Nat’l Lime Ass’n,
Finally, the 1990 CAA Amendments added a provision to Title V of the Act that requires all owners and operators of HAP sources to obtain operating permits.
See id.
§ 7661a. Title V does no more than consolidate “existing air pollution requirements into a single document, the Title V permit, to facilitate compliance monitoring” without imposing any new substantive requirements. Sierra
Club v. Leavitt,
B. The Major Boilers, Area Boilers, and CISWI Rules
On March 21, 2011, the.EPA issued the first iteration of all three rules under review. That same day, however, the EPA announced that it intended to reconsider certain aspects of each rule. Not long after, multiple parties filed the petitions for review that we now address. Earlier, the EPA had concluded its reconsideration and issued the most recent iteration of the three rules. Because of this procedural quirk, each “rule” we address is in fact two separate rules — the EPA’s “final” 2011 version and its “final” 2013 version. The EPA’s analyses remained mostly consistent from 2011 to 2013 and we indicate, where necessary, the instances in which the EPA changed course in a significant way.
1. The Major Boilers Rule
The Major Boilers Rule sets HAPs emission caps for all industrial, commercial, and institutional boilers that emit a large volume of HAPs.
See
2011 Major Boilers Rule,
a. The “Upper Prediction Limit”
Several factors complicate the process of setting MACT floors. The first is the CAA itself, which mandates that all MACT floors (1) must be achievable, see 42 U.S.C. § 7412(d)(2); (2) must ensure continuous regulation of the covered sources, see id. § 7602(k); and (3) must be no less stringent than the emissions levels being achieved by the best-controlled sources, see id. § 7412(d)(3). The second is that no source emits any HAP at a constant level; rather, HAP emissions fluctuate over time and for many reasons, including, e.g., “operation of control technologies, variation in combustion materials and combustion conditions, variation in operation of the unit itself, and variation associated with the emission measurement techniques.” Memorandum from Stephen D. Page, EPA Director of Air Quality Planning and Standards, EPA’s Response to Remand of the Record for Commercial and Industrial Solid Waste Incineration Units (Page Mem.) (July 14, 2014), at 3 (No. 11-1125 J.A. 1316). Finally, most sources do not measure their HAP emissions at all times and under all conditions. 8 Id. at 6. Instead, data are usually gathered when a source conducts a “three-run stack test.” Id. This test provides three “snapshots” of a source’s emissions in a limited set of conditions and, accordingly, it fails to demonstrate accurately a source’s emissions during all times and under all conditions. Id.
To compensate for the lack of adequate emissions data, the EPA uses a statistical tool known as the “upper prediction limit” (UPL) to account for the expected variability in emissions levels.
See
2011 Major Boilers Rule,
b. The “Pollutant-By-Pollutant” Approach
In identifying the best performing sources in a given category, often the EPA could not identify a single source that controlled
all
HAPs better than
all
other sources. Instead, the EPA found that one source effectively controlled emissions from one HAP but was nonetheless one of the worst-performing sources at controlling emissions from a different HAP. For this reason, the EPA adopted a “pollutant-by-pollutant” approach in setting MACT floors for major boiler subcategories.
See
2011 Major Boilers Rule,
c. Startups, Shutdowns, and Malfunctions
The EPA found it difficult to account for HAP emissions when sources start up, shut down, and malfunction. All three occurrences alter HAP emissions and, historically, the EPA exempted sources from normal numeric MACT-standard compliance when these events occurred.
See, e.g.,
Standards of Performance for New Stationary Sources, 42 Fed. Reg. 57,125 (Nov. 1, 1977). Nevertheless, concluding that the Act “require[s] that there must be
continuous
section [7412]-compliant standards” and observing that the exemption meant that “no section [7412] standard governs these events,” in 2008 we vacated the exemption for startups, shutdowns,- and malfunctions when the issue arose in a case challenging a different rule.
Sierra Club v. EPA (Sierra Club III),
In response to the
Sierra Club III
vaca-tur, the EPA established a work-practice standard in lieu of a numeric MACT standard during startup and shutdown periods (but not during malfunctions) when it promulgated the Major Boilers Rule.
See
2011 Major Boilers Rule,
But because a malfunction is “sudden, infrequent, and not reasonably preventable,”
id.
(quoting 40 C.F.R. § 63.2), the EPA declined to treat a malfunction as a “distinct operating mode,”
id.
As a result, the EPA did not account for malfunctions when it set the MACT floors and it required sources to comply with all MACT floors even during periods of malfunction.
Id.
At the same time and recognizing that even the best equipment can fail and that such failure can spike emissions, the EPA added to the Major Boilers Rule “an affirmative defense to civil penalties for excee-dances of numerical emission limits that are caused by malfunctions.”
Id.
In reviewing a challenge to a different EPA rule, however, we vacated a materially identical affirmative-defense provision and held that the EPA has no power under the CAA to create a defense to civil liability.
See Natural Res. Def. Council v. EPA (NRDC III),
d. The One-Time Energy Assessment The EPA also promulgated a “beyond-the-floor” requirement for all facilities with existing major boilers.
See
2011 Major Boilers Rule,
e. The Health-Based Emissions Limits for HC1
Although the EPA set numeric MACT standards to control HC1 emissions,
see
2013 Major Boilers Rule,
2. The Area Boilers Rule
In the Area Boilers Rule, the EPA set ■emissions limits for the same three boiler categories it controlled in the Major Boilers Rule,
see supra
§ T.B.l: industrial, commercial, and institutional boilers.
See
2013 Area Boilers Rule,
For these subcategories, the EPA set emissions limits for three HAPs: Hg, PM, and CO, with PM functioning as a surrogate for non-Hg urban metals and CO functioning as a surrogate for polycyclic organic matter (POM).
See
2011 Area Boilers Rule,
Thus, with the exception of Hg and CO emissions from coal-fired boilers, the EPA had discretion to promulgate GACT standards for all other HAPs in all other source subcategories. See 42 U.S.C. § 7412(d)(5). Exercising this discretion resulted in the following standards:
[[Image here]]
[[Image here]]
2013 Area Boilers Rule,
The Area Boilers Rule shares many of the same features as the Major Boilers Rule; for example, the Area Boilers Rule treats startups, shutdowns, and malfunctions in the same fashion as the Major Boilers Rule,
see supra
§
I.B.l.c
— i.e., the Area Boilers Rule creates work-practice (or management-practice) standards for startup and shutdown periods but does not account for malfunctions at all, save for the Agency’s commitment to consider malfunctions on a case-by-case basis.
See
2013 Area Boilers Rule,
a. Exclusion of “Temporary Boilers”
After the EPA promulgated the 2011 Area Boilers Rule but before it promulgated the 2018 version, it proposed an amendment to 40 C.F.R. § 63.11195 that added temporary boilers to the list of those boilers
not
regulated by section 7412.
See
National Emission Standards for Hazardous Air Pollutants for Area Sources: Industrial, Commercial, and Institutional Boilers (2011 Proposed Area Boilers Rule on Reconsideration), 76 Fed. Reg. 80,532, 80,535 (Dec. 23, 2011). The EPA created the exclusion because, in its view, temporary boilers are “insignificant sources! ] and were not included in the EPA’s analysis of the source category.”
Id.
The Agency eventually defined “temporary boiler” as “any gaseous or liquid fuel boiler that is designed to, and is capable of, being carried or moved from one location to another by means of, for example, wheels, skids, carrying handles, dollies, trailers, or platforms.”
See
2013 Area Boilers Rule,
b. Title V Permit Exemption for Synthetic Area Sources
As noted,
see supra
§ I.A.3, Title V of the CAA imposes a permit requirement on all owners and operators of major and area HAP sources.
See
42 U.S.C. § 7661a. The EPA, however, can exempt an area source subcategory if it finds “that compliance with such requirements is impracticable, infeasible, or unnecessarily burdensome on such categories.”
Id.
§ 7661a(a). When it proposed the Area Boilers Rule in 2010, the EPA considered exempting some area sources because, in its view, the existing restrictions on those sources made Title V duplicative.
See
2010 Proposed Area Boilers Rule,
But in the 2011 Area Boilers Rule, the EPA changed course and exempted synthetic area sources from the Title V permitting requirement.
See
3. The CISWI Rule
«In the CISWI Rule, the EPA created four CISWI subcategories: (1) incinerators (ie., “units designed to burn [solid] waste materials for the purpose of disposal”); (2) small, remote incinerators (“SRIs”) (ie., units that burn small waste batches); (3) energy recovery, units (“ERUs”) (ie., units that would be classified as boilers but for the fact they com-bust solid waste); and (4) waste-burning kilns (ie., units that would be classified as cement kilns if they did not burn solid waste). 2013 CISWI Rule,
The EPA then set numeric MACT limits for the section 7429(a)(4) pollutants.
12
See
2011 CISWI Rule,
The CISWI Rule shares several features with the Major Boilers Rule. In the CI-SWI Rule, for instance, the EPA also used the UPL, see id. at 15,722-27, as well as the pollutant-by-pollutant approach, see id. at 15,719-21, in setting MACT floors. Based in part on the differences between section 7412 and section 7429, the CISWI rule has four unique characteristics we briefly describe.
a. Startups, Shutdowns, and Malfunctions
As discussed,
see supra
§ ILB.l.c, the EPA imposed a work-practice standard for major and area source boilers during periods of startup and shutdown but declined to make any regulatory modification for malfunctions.
See
2011 Major Boilers Rule,
b. The Record-Keeping Requirement
Whether the EPA considers a combustion unit to be a boiler (and thus subject to section 7412) or a CISWI (and thus subject to section 7429) turns entirely on whether the unit combusts “solid waste.”
See id.
at 15,709. The term “solid waste” is defined in RCRA, 42 U.S.C. §§ 6901
et seq.,
and clarified by EPA regulation,
see
Identification of NonHazardous Secondary Materials that Are Solid Waste (NHSM Rule), 76 Fed. Reg. 15,456, 15,457 (Mar. 21, 2011).
See also
2011 CISWI Rule,
The source owner or operator initially decides whether the material its combustion unit burns meets the definition of solid waste.'
See id.
at 15,740. For this reason, the CISWI rule requires that the owner or operator of a combustion unit that burns materials “not clearly listed as traditional fuels” keep records explaining how the materials meet the regulatory definition of “non-solid waste.”
Id.; see also
40 C.F.R. § 60.2175(v). Failure to do so means, for the purposes of the EPA, that “the operating unit is a CISWI unit.” 40 C.F.R.
c. Emissions Averaging
• During the notice-and-eomment period, certain industry entities urged the EPA to allow a facility containing more than one CISWI unit to demonstrate compliance with the CISWI MACT standards by averaging the HAP emissions of all units in the facility. See Commercial and Industrial Solid Waste Incineration Units: Reconsideration and Proposed Amendments; NonHazardous Secondary Materials that Are Solid Waste (2011 Proposed CISWI Rule on Reconsideration), 76 Fed. Reg. 80,452, 80,463 (Dec. 23, 2011). Although it allowed facility-wide averaging in the Major Boilers Rule, the Agency declined to allow it for facilities with CISWI units. See id. The EPA explained, first, that “[t]he applicability of CISWI is such that each unit is an affected facility.” Id. In response to further comments, the EPA subsequently explained that it did “not believe [it had] the legal authority to allow emissions averaging in CISWI or under section [7429] generally because each individual unit is an affected facility.” Summary of Public Comments and Responses for Commercial and Industrial Solid Waste Incineration Units (CISWI Rule — Responses to Comments), EPA-HQ-OAR-2003-0119-2638-A2 (Dec. 2012), at 195.
d. Treatment of Units that Begin Combusting Solid Waste
Finally, in the preamble to the 2011 CISWI Rule, the EPA stated broadly that “[u]nits that begin combusting solid waste are considered
existing
sources under CI-SWI.”
II. STANDARD OF REVIEW
For each issue, the Petitioners argue that the EPA'either misinterpreted the CAA, acted arbitrarily and capriciously, or both. We review the EPA’s construction of the statute under the two-part framework established in
Chevron, U.S.A., Inc. v. Nat. Res. Def. Council, Inc.,
The CAA authorizes the Court to “reverse any [EPA] action found to be ... arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 42 U.S.C. § 7607(d)(9)(A). Our review is “narrow” and we will “not ... substitute [our] judgment for that of the agency.”
Motor Veh. Mfrs. Ass’n v. State
We review the EPA’s factual determinations for substantial evidence. 5 U.S.C. § 706(2)(E). We also “owe[ ] particular deference to EPA when its rulemakings rest upon matters of scientific and statistical judgment within [its] sphere of special competence and statutory jurisdiction.”
Am. Coke & Coal Chems. Inst. v. EPA,
III. INDUSTRY PETITIONERS’ CHALLENGES
A. STARTUPS, Shutdowns, and Malfunctions
Industry Petitioners raise two sets of challenges to startup, shutdown, and malfunction periods: (1) a challenge to the EPA’s failure to take malfunctions into account in the Major Boilers and Area Boilers Rules and (2) a challenge to EPA’s failure to take into account periods of startup, shutdown, and malfunction in the CISWI Rule. For the reasons that follow, we reject all of the Industry Petitioners’ claims related to startups, shutdowns, and malfunctions.
1. Periods of Malfunction in the Major Boilers and Area Boilers Rules
First, Industry Petitioners challenge the Major Boilers and Area Boilers Rules’ failure to take malfunctions into account in setting MACT floors.
See
2011 Major Boilers Rule,
Both sides agree that malfunctions are inevitable in the operation of area and major boilers. According to the EPA, “even equipment that is properly designed and maintained can sometimes fail and ... such failure can sometimes cause an excee-dance of the relevant emission standard.” 2011 Major Boilers Rule,
At first glance, the EPA’s chosen approach to malfunctions may seem counter-intuitive, as the Agency appears to have several reasonable alternatives: it could exempt periods of malfunction entirely from the application of the emission standards; or it could apply the standards to malfunctions while giving boiler owners the opportunity to defend against a penalty by demonstrating they were not at fault for the malfunction. But the EPA has previously been stymied in its attempts to implement either of these solutions, as this court has concluded neither approach is consistent with the Agency’s enabling statutes. For instance, in
Sierra Club III,
the EPA attempted to exempt major sources from complying with emission standards during start up, shut down, and malfunction.
See
Faced with an obvious dilemma, the EPA arrived at the approach it defends today. Malfunctions receive no special treatment and the EPA instead exercises “its enforcement discretion to address ex-ceedances of emission limits that may be caused by such uncertain, unpredictable events, on a case-by-case basis.” No. 11-1108 EPA Br. 38;
see also
No. 11-1141 EPA Br. 29. The EPA’s current treatment of malfunctions thus differs from its invalid affirmative defense provision because the Agency is exercising its
oum
regulatory enforcement power on an ad hoc basis outside the context of citizen suits. When an exceedance occurs during a malfunction, the EPA determines what enforcement action — if any — it should take by considering “the good faith efforts of the source to minimize emissions during malfunction periods, including preventative and corrective actions, as well as root cause analyses to ascertain and rectify excess emissions.” 2011 Major Boilers Rule,
For our purposes, we need not (indeed, must not) evaluate the policy implications of the EPA’s regulatory choice because our review is confined to determining whether the EPA’s regulation reflects a ' permissible reading of the applicable statute under
Chevron.
Here, we conclude that it does. The relevant statute requires only that the EPA set “achievable” standards, 42 U.S.C. § 7412(d)(2), and it defines ac-hievability to be no less “than the emission control that is achieved in practice by the best controlled similar source,” 42 U.S.C. § 7412(d)(3). The “best controlled similar source,” however, is unlikely to be a malfunctioning source, and the EPA is bound to enact a standard in keeping with emission limits achieved by that “best controlled similar source.” If anything, then, the statutory language on its face prevents the EPA from taking into account the effect of potential malfunctions when setting MACT emission standards. At the very least, the language permits the EPA to ignore malfunctions in its standard-setting and account for them instead through its regulatory discretion. Our
Sierra Club III
decision confirms this.
See
Nor do we agree with the Industry Petitioners’ secondary argument that the EPA acted arbitrarily and capriciously by failing to set a work-practice or a GACT management-practice standard for malfunction periods. First, the statute makes clear that these kinds of standards are to be set at the discretion of the EPA, so it would be difficult to interpret the statute consistently with its text while holding that the text’s
permissive
language in fact sets out a
requirement
that the Agency set work-practice or GACT management-practice standards. As to work-practice standards, “[t]he Administrator may, in lieu [of a numeric standard], promulgate a design, equipment, work practice, or operational standard, or combination thereof,” and any such standard set must “in the Administrator’s judgment [be] consistent with the provisions of subsection (d).” 42 U.S.C. § 7412(h)(1). As to GACT management-practice standards, “the Administrator may ... elect to promulgate” such standards with respect to certain “categories and subcategories 'of area sources.”
Id.
§ 7412(d)(5). It should go without saying that “may means may.”
McCreary v. Offner,
Second, the Petitioners have not demonstrated and the EPA does not concede that setting work-practice or GACT management-practice standards would even be feasible for periods of malfunction. As for work-practice standards, the EPA would have to conceive of a standard that could . apply equally to the wide range of possible boiler malfunctions, ranging from an explosion to minor mechanical defects. Any possible standard is likely to be hopelessly generic to govern such a wide array of circumstances. Similar problems exist for setting GACT management practices. These management practices would also need to apply to the wide range of possible malfunctions, and the EPA would need to determine that the standard would “reduce emissions of hazardous air pollutants,” an evidence-based standard that is difficult (perhaps impossible) to apply to the unpre- • dictable circumstances of malfunctions. 42 U.S.C. § 7412(d)(5). Thus, we reject the
In doing so, we are mindful that the EPA is not the only entity able to bring enforcement actions under the CAA, but that private citizens are also empowered to enforce emission standards by filing suit in district court. 42 U.S.C. § 7604(a). Assurances that the EPA will use its prosecutorial discretion to account for malfunctions would mean little if private citizens could seek strict enforcement of those same standards. But as we stated in
NRDC III,
“the Judiciary, not any executive agency, determines ‘the
scope’
— in
cluding the available remedies
— ‘of judicial power vested by’ statutes establishing private rights of action.”
2. Periods of Startup, Shutdown, and Malfunction in the CISWI Rule
In the CISWI Rule, the EPA made no modification for periods of start-up, shutdown, or malfunction. The Industry Petitioners argue that failing to account for these periods violated the EPA’s statutory instruction to set “achievable” standards. Additionally, the Industry Petitioners claim it was arbitrary and capricious for the EPA to set work-practice standards for startup and shutdown periods under the Major Boilers Rule but not under the CISWI Rule. Both arguments are without merit.
First, the EPA’s emission standards for small incinerators do take into account periods of shutdown and startup. The EPA based its standards for these machines on “short term stack tests' for pollutants,” in which incinerators are monitored during the course of normal operation, which includes daily startup and shutdown periods.
See
2011 CISWI Rule,
Second, as to periods of malfunctions, the same analysis applies to the CISWI Rule as applies to the Boilers Rules. The EPA adopted a reasonable interpretation of the CAA when it excluded periods of malfunction from its calculations of aehiev-ability given that malfunction periods are by their very nature unpredictable in terms of their effect on emissions. The EPA’s decision to account for malfunctions in its discretion is likewise a reasonable interpretation of 42 U.S.C. § 7412(d)(2) and (3).
B. The Pollutant-By-Pollutant Approach
The EPA must look to the performance of the best major boilers and CISWI incinerators when setting MACT floors for a pollutant. As described above, for new units, the EPA must set floors at the level achieved by the best similar unit in each subcategory. For existing units, the Agency must set floors at the level achieved by the best 12 per cent of similar units in each subcategory. 42 U.S.C. §§ 7412(d)(3)(A), 7429(a)(2). As a result, the EPA had to identify the best performing units in each subcategory when setting the MACT floors for the Major Boilers and CISWI Rules. But the EPA often could not identify a single unit or set of units that controlled all HAPs better than the other units in the subcategory. Instead, the EPA sometimes found that a unit might rank among the best in its subcategory at controlling emissions of one HAP, but among the worst at controlling emissions of a different HAP.
To address this problem, the EPA adopted a “pollutant-by-pollutant” approach in setting the MACT floors: instead of identifying the unit or units that best 'controlled
all
HAPs in the aggregate, the EPA used one unit or set of units to set the MACT floor for,
e.g.,
PM, and used a different unit or set of units to set the MACT floor for,
e.g.,
HC1.
See
2011 Major Boilers Rule,
The Industry Petitioners challenge the EPA’s use of the pollutant-by-pollutant approach. According to the Industry Petitioners, the CAA’s plain language requires the Agency to identify the best overall unit or set of units — not the best unit or set of units for a particular pollutant — in each subcategory when setting MACT floors. They further claim the EPA’s pollutant-by-pollutant approach was unreasonable with regard to SRIs because it resulted in a set of emission standards that no single unit in the subcategory had achieved in practice. We disagree, and conclude that the EPA’s pollutant-by-pollutant approach is a reasonable interpretation and application of the statute.
For the purposes of this challenge, the MACT floor provisions for major boilers and CISWI units are identical. Under both provisions, the EPA must set emission standards for new units based on “the emissions control that is achieved in practice by the best controlled similar unit, as determined by the Administrator.” 42 U.S.C. § 7429(a)(2) (CISWI); see also id. § 7412(d)(3) (major boilers). For existing units, the MACT floor is based on “the average emissions limitation achieved by the best performing 12 percent of units in the category.” Id. § 7429(a)(2) (CISWI); see also id. § 7412(d)(3)(A) (major boilers).
The Industry Petitioners claim this language unambiguously forecloses the EPA’s pollutant-by-pollutant approach. For new units, they assert, the statute requires the EPA to find the single unit that performs best overall and use this unit — and only this unit — to set standards for all regulated pollutants. For example, if Incinerator 3 were deemed the best overall performer in a subcategory, then the EPA would use
The Industry Petitioners read too much into the statutory language. It is true that the statute requires the EPA to base MACT standards on what is “achieved” by the best “unit” or “12 percent of units.” But, as the EPA argues, the statute says nothing about how the Agency should determine which units are the best.
Cf. Sierra Club v. EPA,
Here, the EPA’s choice is reasonable. The statute provides that emission standards shall reflect “the maximum degree of reduction in emissions of [regulated pollutants] that the Administrator ... determines is achievable for new or existing units in each category.” 42 U.S.C. § 7429(a)(2); see also id. § 7412(d)(2). It then provides that the “degree of reduction in emissions that is deemed achievable for new units in a category shall not be less stringent than the emissions control that is achieved in practice by the best controlled similar unit, as determined by the Administrator.” Id. § 7429(a)(2); see also id. § 7412(d)(3). Reading these provisions together, they support a pollutant-by-pollutant approach. The “best controlled similar unit” language does not exist in a vacuum; rather, it exists to measure the “degree of reduction in emissions that is deemed achievable.” Id. § 7429(a)(2); see also id. § 7412(d)(3). That “reduction in emissions” is the reduction in emissions of each pollutant listed in sections 7429(a)(4) and 7412(b)(1). The EPA’s approach to setting standards on a pollutant-by-pollutant basis thus comfortably fits within this statutory scheme.
Moreover, the Industry Petitioners have not explained how their preferred approach would better comport with the statute. Were the EPA required to determine which units perform best “overall,” we see at least two possibilities for how it could do so: First, the EPA could calculate a unit’s average emissions for each pollutant in consistent units of measurement, add these emissions together, and then choose the unit with the smallest overall sum in each subcategory. But this approach could produce arbitrary results, because the “best ■ performing” overall unit might emit unusually low quantities of some pollutants and unusually high quantities of others. This would mean the emission standards for some pollutants would be lenient while others would be stringent, with no principled reason for the difference. Alternatively,
The Industry Petitioners nevertheless argue that the CAA’s legislative history supports their preferred approach. In particular, they point to the floor comments of Senator Durenberger discussing the potential impact on MACT floors of mutually incompatible control technologies. 136 Cong. Rec. S17,238 (daily ed. Oct. 26,1990) (statement of Senator Durenberger). Mutually incompatible control technologies cannot be used at the same time and therefore present regulators with a dilemma. For example, say Technology 1 and Technology 2 cannot be used together. If Technology 1 is better at reducing PM than Technology 2, and Technology 2 is better at reducing CO than Technology 1, the EPA would have to choose which of the two technologies to factor into emission standards. In such situations, Senator Durenberger anticipated that the “EPA should judge MACT to be the technology which best benefits human health and the environment on the whole.” Id. The Industry Petitioners argue this statement demonstrates that Congress intended the EPA to make an overall determination of which units are the best performing “on the whole.”
Senator Durenberger’s statement does not support this broad principle. The statement merely explains that, where two technologies cannot be used together, the EPA should base MACT standards on the technology it considers best overall. Here, the Industry Petitioners do not identify any relevant control technologies that are mutually incompatible. Indeed, the EPA found in the CISWI Rule that “there is no technical reason why [the] air pollution control systems cannot be combined.” 2011 CISWI Rule,
Finally, the Industry Petitioners argue that even if the pollutant-by-pollutant approach is reasonable in some circumstances, it is arbitrary and capricious as applied to certain SRIs because it exacerbates certain problems posed by the “batch” nature of SRIs. As explained at
infra
§ III.E, SRIs burn waste in small batches. According to the Petitioners, this means that the SRIs that the EPA identified as best performing were, in reality, burning cleaner waste at the time emissions testing was done; they were not actually better than other units at removing or destroying waste. The pollutant-by-
This argument fails because the Industry Petitioners have not demonstrated that the Agency considered impermissible factors, failed “to consider important aspect[s] of the problem,” or offered an unreasonable explanation for its decision when setting the MACT floors for SRIs.
See State Farm,
C. The Energy-Assessment Requirement
The Major Boilers Rule and the Area Boilers Rule generally require sources with existing boilers to perform a one-time energy assessment. In the assessment, facilities must “identify energy conservation measures” — such as “process changes or other modifications to the facility” — “that can be implemented to reduce the facility energy demand,” thereby “reduc[ing] fuel use.” 2011 Area Boilers Rule,
The logic behind the assessment is straightforward. Boilers produce HAP emissions when fuel is combusted. Less combustion means fewer emissions. The EPA primarily justified the assessment as a beyond-the-floor MACT requirement under section 7412(d)(2).
See
2011 Area Boilers Rule,
Industry Petitioners raise three principal challenges to the energy-assessment requirement, none of which have purchase. The first challenge claims that the energy assessment regulates aspects of facilities that are off limits to the EPA— namely, the energy needs supplied by regulated boilers. Petitioners point to the language of the CAA, which requires the EPA to “list ... categories and subcategories of major sources and area sources” of enumerated air pollutants. 42 U.S.C. § 7412(c)(1). “For the categories and subcategories the Administrator lists, the Administrator” must set “emissions standards under” section 7412(d).
Id.
§ 7412(c)(2). As relevant here, the EPA defined the source categories to include “industrial boilers and commercial and institutional boilers.”
The Industry Petitioners misapprehend both the scope of the assessment and the CAA. The assessment requires facilities to evaluate energy systems “located on the site of the affected boiler,” including “[pjrocess heating[,] compressed air systems[,] ... facility heating, ventilation, and air conditioning systems,” and “[o]ther systems that use steam, hot water, process heat, or electricity, provided by the affected boiler.” 40 C.F.R. § 63.11237; see id. § 63.7575. Based on that evaluation, facilities must compile a “comprehensive report detailing the ways to improve efficiency, the cost of specific improvements, [anticipated] benefits, and the time frame for recouping those investments.” 40 C.F.R. pt. 63, subpt. JJJJJJ tbl.2; id. pt. 63, subpt. DDDDD tbl.3.
Contrary to the Industry Petitioners’ argument, the EPA has not “regulate[d] virtually every piece of equipment at all affected facilities.” No. 11-1141 Indus. Pet’rs’ Br. 19. Only “energy use systems” that “us[e] energy
clearly
produced by affected boilers” must be evaluated; facilities need not review the “total aggregation of all individual energy using segments of a facility.” 2013 Area Boilers Rule,
And that measured requirement falls within the EPA’s statutory authority. The CAA authorizes the EPA to regulate “major sources and area sources” of HAPs, and to subdivide those sources into categories and subcategories. 42 U.S.C. § 7412(c)(1), (c)(2). To Industry Petitioners, the authority to subdivide sources means the EPA may only regulate the narrowest applicable categorization — in this instance, commercial and industrial boilers. But the statute does not require so rigid a reading. While the EPA is permitted to subdivide sources, each subdivision remains a component of either a major or area “source.” Dividing sources into categories and subcategories does not make them any less of a “source” subject to the EPA’s regulation.
For that reason, the EPA explained that the Rules reach, respectively, “[a]ny area source
facility
using a boiler,” 2011 Area Boilers Rule,
The Congress’s definition of the terms major and area source supports this reading. At bottom, both terms refer to a “stationary source.” See 42 U.S.C. § 7412(a)(1), (a)(2). Stationary source, in turn, means “any building, structure, facility, or installation which emits or may emit any air pollutant.” Id. § 7411(a)(3). Against that backdrop, the Rules apply to any “building, structure, facility, or installation” that contains a boiler emitting the specified HAPs. The EPA’s regulatory authority reaches the relevant stationary source, of which the boiler is part.
That the EPA may regulate stationary sources does not mean it may regulate every nook and cranny of those sources. The CAA directs its authority to the establishment of emission standards; it does not provide some general power to superintend the business processes of plants and manufacturing facilities. In this case, however, we have no occasion to parse the precise parameters of the EPA’s authority to regulate aspects of area sources. It is enough to conclude that the challenged energy assessment — which applies only to systems that “us[e] energy clearly produced by affected boilers” — falls within the EPA’s authority under the CAA. 2013 Area Boilers Rule,
In the remaining two challenges, the Industry Petitioners take issue with the EPA’s justification of the energy assessment as a beyond-the-floor MACT standard and a GACT management-practice standard. We reject both challenges.
The assessment represents a valid beyond-the-floor MACT standard.
13
As discussed, after the Agency sets the MACT floor, it must determine “whether stricter standards are ‘achievable,’ ”
Nat'l Lime Ass’n,
(A) reduce the volume of, or eliminate emissions of, such pollutants through process changes, substitution of materials or other modifications, ...
(D) are design, equipment, work practice, or operational standards ... or
(E) are a combination of the above.
Id.
The EPA primarily justified the energy assessment as a beyond-the-floor measure designed to identify “process changes or other modifications to the facility” that would reduce fuel use and thereby reduce hazardous emissions. 2011 Area Boilers Rule,
The Industry Petitioners argue that the EPA skipped a step, imposing the energy assessment as a beyond-the-floor measure without first setting a relevant MACT floor. That is incorrect. The EPA first set a numeric MACT emissions limit for the categories and subcategories of sources subject to the energy assessment.
See
40 C.F.R. pt. 63, subpt. JJJJJ tbl.l;
id.
pt. 63, subpt. DDDDD tbl.2. The energy assessment represents a step beyond that — a
Before setting a beyond-the-floor measure, the EPA must consider whether it is “achievable” based on a number of factors, among them cost, “non-air quality health and environmental impacts and energy requirements.” 42 US.C. § 7412(d)(2). The EPA did so here. To begin, the EPA adequately considered costs. In the EPA’s estimation, “[t]he one-time cost of an energy assessment ranges from $2500 to $55,000 depending on the size of the facility.” 2010 Proposed Area Boilers Rule,
In addition to costs, the EPA considered non-air quality health and environmental impacts in general terms, concluding that “improving energy efficiency reduces negative impacts on the environment.” 2010 Proposed Area Boilers Rule,
We also find that the assessment is á valid GACT management practice. With respect to area sources, the EPA has discretion to require the use of “generally available control technologies or management practices ... to reduce emissions of hazardous air pollutants.” 42 U.S.C. § 7412(d)(5). The EPA justified the energy assessment as a GACT management practice for oil- and biomass-fired boilers.
See
2011 Area Boilers Rule,
The Industry Petitioners challenge that justification, claiming the energy assessment — which does not require implementation — cannot “reduce emissions of hazardous air pollutants.” 42 U.S.C. § 7412(d)(5). We disagree. The EPA did not need to make implementation mandatory to make the assessment lawful. Under the CAA, the EPA may sometimes act with a soft touch, rather than a firm hand. Here, the EPA selected a soft touch, requiring an assessment but not implementation. It was not unreasonable for the EPA to conclude, “after considering the structure of the requirement, the incentives it presents, and the likely behavior of sources, ... that sources will find it cost-effective to implement the conservation measures identified in the energy assessment.” 2011 Area Boilers Rule,
For those reasons, we reject the Industry Petitioners’ challenges to the energy-assessment requirement.
Section 7429 regulates combustion units that burn solid waste; units that do not burn solid waste will generally be regulated under section 7412. RCRA defines the term “solid waste” to mean (in part) “discarded material ... resulting from industrial [or] commercial ... operations.” 42 U.S.C. § 6903(27); see
id.
§ 7429(g)(6) (directing that “solid waste” carry “the meanings established by the Administrator pursuant to” RCRA). On the same day the EPA issued a rule setting emission standards for CISWI, it issued a separate rule fleshing out the meaning of solid waste in the context of combustion units.
See
NHSM Rule,
• The NHSM Rule generally provides that “non-hazardous secondary materials that are combusted are solid wastes,” 14 subject to several exceptions and exemptions. 40 C.F.R. § 241.3(a). Among the exceptions, non-hazardous secondary materials that meet certain “legitimacy” criteria do not qualify as solid waste. See id. § 241.3(b), (d). Source owners and operators may also seek a finding from the EPA that particular materials do not constitute solid waste when combusted by a third party. Id. § 241.3(c). And the rule exempts altogether a variety of materials from the definition of solid waste, including “traditional fuels.” Id. § 241.2.
The NHSM Rule is self-implementing: each source owner or operator must determine whether combusted materials meet the definition of solid waste.
See
2011 CI-SWI Rule,
Industry Petitioners challenge this last provision of the CISWI Rule. 15 They argue that the EPA cannot automatically treat units that fail to keep certain paperwork as CISWI units. Section 7429 permits regulation of “solid waste incineration units” — not units whose owners fail to file paperwork. As a result, the Industry Petitioners ask this court to invalidate the regulatory provision as exceeding the EPA’s statutory authority. 16
We decline the invitation. At
Chevron’s
first step, we find that “Congress did not speak directly, let alone clearly, to this issue.”
Am. Chem. Council v. EPA,
In this case, the EPA included within the revised CISWI Rule a presumption designed to enforce the Congress’s command. Section 7429 nowhere addresses whether the EPA may establish presumptions to ensure its regulations reach all sources burning solid waste. At the same time, the Congress plainly intended the EPA to regulate sources burning “any” solid waste, a goal presumably advanced by the recordkeeping presumption. See id. Against that backdrop, we cannot conclude that the presumption offends the text or purpose of section 7429.
Moving to
Chevron’s
second step, we conclude the recordkeeping presumption is reasonable. In
American Chemistry Council,
we upheld a regulation issued under RCRA defining hazardous waste to include any mixture or derivative of hazardous substances.
See
Similar reasoning applies here. The EPA crafted the presumption to reach sources likely to be burning solid waste, namely, those burning discarded materials other than traditional fuels.
See
42 U.S.C. § 6903(27) (defining “solid waste” to include, among other things, “discarded material”); 40 C.F.R. § 241.2 (exempting traditional fuels, defined as “materials that are produced as fuels ... that have not been discarded,” from the definition of solid waste). Such sources are subject to strict recordkeeping requirements.
See
40 C.F.R. § 60.2175(v). Within those confines, placing the burden on unit operators who have the mandatory obligation and the information to establish their non-regulable status is reasonable.
Cf. Am. Chem. Council,
There is, however, a difference between the presumption in this case and the one we upheld in American Chemistry Council. The CISWI recordkeeping presumption appears to turn on the failure to file paperwork, rather than the presence of a regulated substance. However broadly the Congress defined “solid waste incineration unit” in section 7429, the Congress did not allow for the regulation of now-waste burning sources — even when those sources fail to file paperwork. Indeed, had the EPA attempted to regulate sources based purely on a failure to file paperwork, we may well have reached a different conclusion.
But the CISWI presumption does not stretch so far. As explained, the presumption depends on factors beyond the mere failure to keep records. Sources subject to the presumption burn materials likely to qualify as solid waste, and must satisfy demanding recordkeeping requirements. The EPA acted reasonably when it presumed such sources were burning solid waste.
Despite the provision’s narrow reach, the Industry Petitioners fear it will sweep up sources not burning solid waste. To the extent that possibility exists, sources wrongfully regulated as CISWI have mul
We therefore reject the Industry Petitioners’ challenges to the recordkeeping presumption. 17
E. Waste-Stream Variance for SRI Units
The EPA regulated SRIs as a subcategory in the CISWI Rule. See Memorandum from Eastern Research Group, Inc., to Toni Jones, U.S. Environmental Protection Agency, CISWI Emission Limit Calculations for Existing and New Sources for the Reconsideration Final Rule (Jones Mem.) (Nov. 16, 2012) (No. 11-1125 J.A. 1159, 1162). There are 28 SRI units, all of which are located in Alaska, and the EPA had emissions data for nine of them. Id. As explained supra § I.B.3, the EPA used the pollutant-by-pollutant approach to establish MACT emission standards for these units. For new-unit standards, the EPA determined which of the nine units had the lowest emissions for a particular pollutant and set the MACT floor for that pollutant at the level achieved by the identified unit. See 42 U.S.C. § 7429(a)(2) (explaining that MACT floors for new units must be set at “the emissions control that is achieved in practice by the best controlled similar unit”). When setting MACT floors for existing units, the EPA had to calculate the average level of emissions achieved by the best performing 12 per cent of units. See id. It therefore determined which four sources had the lowest emissions for a given pollutant and set the emissions standard for that pollutant at the average level achieved by those four units.
The Industry Petitioners argue that the EPA’s approach was unlawful because it failed to account for the unique role that waste inputs play in emissions from SRIs. Unlike larger incinerators, SRIs burn small batches of waste at a time. Some batches include cleaner waste, such as wood and cardboard, while others include waste, such as sewage, that generates large quantities of S02 and other pollutants. Moreover, existing SRIs cannot use certain “end-of-stack” control technologies like wet scrubbers due to the Alaskan climate. The Industry Petitioners thus contend that emissions from SRIs are more closely tied to waste input than are emissions from other types of incinerators. This difference, they assert, required the EPA to take into account, when determining which SRI units were best performing for MACT floor purposes, the kind of waste an SRI unit was burning at the time of testing. Because the Agency did not do so, the Industry Petitioners contend the MACT standards for SRIs are arbitrary and capricious. We disagree.
To support their challenge, the Industry Petitioners advance two arguments, neither of which has merit. Petitioners first point to section 7429(a)(3), which directs the EPA to base emission standards on “methods and technologies for removal or destruction of pollutants before, during, or after combustion.” 42 U.S.C. § 7429(a)(3).
The EPA responds that the approach it adopted for SRIs complies with section 7429(a)(3) because “waste segregation”— that is, diverting dirtier waste to landfills and burning only cleaner waste — is a “method ... for removal ... of pollutants before ... combustion.” See 42 U.S.C. § 7429(a)(3). In fact, during notice and comment, the EPA estimated that many SRIs would choose to comply with the MACT standards by segregating their waste instead of by installing expensive control technologies. See Jones Mem. The Agency also determined that waste segregation was possible for SRIs because their waste often contained materials that could be recycled. Id. Finally, the Agency factored in any additional variance in emissions from these units by calculating the MACT floors according to the UPL formula described at supra §§ LB.l.a, IV.C. For these reasons, the Agency contends, it did not need to consider further any variation in emissions that might be caused by differences in waste inputs for SRIs.
The EPA has the better argument, based on both text and precedent. Textually, waste segregation plainly can be a “method[]” for “removal” of pollutants “before” combustion.
See
42 U.S.C. § 7429(a)(3). Accordingly, the EPA, when setting MACT floors, could not have looked solely to technologies used to reduce emissions during combustion.
Accord Sierra Club v. EPA (Sierra Club II),
Our holding in Sierra Club II confirms that our conclusion is correct. In that case, the EPA had acknowledged that kilns emitted lower levels of pollutants when burning cleaner clay but nevertheless based MACT standards only on the emission reductions achieved by control technology during the combustion process. Id. at 882. The Agency explained that clean clay existed only in certain areas and that transportation of the clay over long distances was impractical. Id. The EPA therefore considered only those emission reductions that were attributable to “deliberate steps kiln operators [took] to reduce emissions rather than to the ‘happenstance’ of being located near cleaner clay.” Id. at 883. But we rejected that approach, finding that “the Clean Air Act requires neither an intentional action nor a deliberate strategy to reduce emissions.” Id. Instead, where “non-technology factors” affect emission levels, we held the EPA must consider those effects when setting MACT floors. Id.
Applying that same reasoning, the EPA acted reasonably when it decided to corn-sider the emissions reduction that could be
The Industry Petitioners’ second argument also comes up short. According to Petitioners, the EPA selected the best performers for SRIs merely because those units happened to be burning batches of cleaner waste at the time of the emissions test. They claim this happenstance resulted in test data that did not reasonably estimate the
typical
performance of the units, and thus misidentified the best performers.
See Cement Kiln,
If the record supported this argument, it might well be persuasive; in
NACWA,
we accepted a similar contention that the EPA’s dataset for determining MACT floors must fairly represent a unit’s typical performance.
See
First, Petitioners cite evidence indicating that XTO Energy, which operates the incinerator that the EPA deemed the best performer for S02, was burning low-sulfur “waste wood, cardboard, and oily waste” during the relevant test runs. See Conoco-Phillips Co., Comment on EPA’s Proposed National Emission Standards for Hazardous Air Pollutants, EPA-HQ-OAR-2003-0119 (Feb. 12, 2012) (No. 11-1125 J.A. 1036). But the record does not show that the resulting test data were unrepresentative of XTO’s typical performance because the record says nothing about what XTO typically burns. Id.
Second, Petitioners note that Drift River, the unit the EPA deemed the” worst performer for S02, had emissions results similar to XTO Energy’s when burning low-sulfur waste, but results over 1,000 .times higher when burning high-sulfur waste. See id. (No. 11-1125 J.A. 1032-33). But again, the record does not say anything about the type of waste Drift River typically burns or its sulfur content; it merely demonstrates that the unit’s test results varied greatly from one run to the next. See id.
Third, Petitioners point to additional test data they provided for the Kuparuk unit, a source that met the EPA’s MACT standards for NOx. See id. (No. 11-1125 J.A. 1017, 1027-28). They claim this data shows that the Kuparuk unit “consistently” emits NOx levels exceeding that standard when burning sewage sludge. Id. This claim is both factually untrue — as the data reveals exceedances on only one day — and says nothing about whether the test' data that the EPA used was representative of Kuparuk’s typical performance. Id.
Instead, the record supports the EPA’s assertion that it gave Petitioners “multiple opportunities” to present data on the variability of waste streams for SRIs, but Petitioners never provided a reasonable empirical basis upon, which the Agency could adjust the MACT standards due to this variability. The Industry Petitioners have
In sum, no record evidence suggests that the current SRI emission standards are not achievable. The Industry Petitioners instead offer only general statements about the “small batch” nature of SRIs and the difficulty of using waste segregation or other controls in remote locations. These factors alone do not call into question the EPA’s assertion that controls such as waste segregation and technology upgrades are a feasible means of achieving compliance with the MACT floors that it established.
See
2011 CISWI Rule,
F. Carbon Monoxide as a Surrogate
In setting MACT standards for major boilers, the EPA used carbon monoxide (CO) as a surrogate for several of the HAPs that the Agency was required to regulate. A surrogate is another chemical that stands in as a proxy for the regulated HAP when the EPA sets numeric emission standards. The EPA regulates the surrogate in order to regulate the HAP, sometimes because the HAP itself is too difficult to measure.
We have previously approved the use of surrogates where the EPA’s choice of a surrogate for the HAP is “reasonable.”
See, e.g., Nat'l Lime Ass’n,
The Industry Petitioners base their first argument on a deficiency in the EPA’s dataset for coal-fired boilers’ emissions— i.e., the dataset contained numerous “non-detects” for organic HAP emissions. A test result is considered a “non-deteet” when emissions testing returns a value below that which the test methods are capable of detecting. According to the Industry Petitioners, multiple non-detects in a dataset demonstrate that it is “not feasible” to set a numeric emission standard for the affected HAP. As a result, they argue, the EPA should have set work-practice standards for these HAPs under section 7412(h)(2), which permits the EPA to set such standards when it is “not feasible” to set a numeric emission standard. See 42 U.S.C. § 7412(h)(2).
This argument fails because Petitioners have not explained how the non-detects here made setting numeric emissions “not feasible,” as that term is defined in the CAA. The CAA expresses a clear preference for MACT emission standards and limits the EPA’s ability to fashion more flexible work-practice standards.
Compare id.
§ 7412(d)(3) (providing that emission standards “shall not be less stringent” than the MACT floor),
with id.
§ 7412(h)(1) (permitting work-practice standards only if MACT standards are “not feasible”). To set a work-practice
This is a high bar and Petitioners have not demonstrated that the non-detects they have identified meet it. During notice and comment, the Agency reasonably explained that non-detects are present in many of its datasets because they are inherent to the imprecision associated with measuring boiler emissions.
See, e.g.,
2011 Major Boilers Rule,
We also reject the Industry Petitioners’ second argument that the EPA needed to explain why it established work-practice standards for other types of boilers in the unrelated “Utility MATS” rule. We take an “every tub on its own bottom” approach to the EPA’s setting of emission standards pursuant to the CAA.
Sierra Club I,
G. Health-Based Emissions Limitation for HCl
In the Major Boiler Rule, the EPA chose not to exercise its discretion to create more lenient emission standards for hydrogen chloride (HCl) based on health. The Industry Petitioners challenge this decision as arbitrary and capricious because, they claim, the Agency considered impermissible factors in reaching the decision and departed from its previous position without adequate justification. We disagree and hold the EPA reasonably chose not to establish a health-based emissions limitation for HCl.
The EPA generally must establish emission standards for all listed pollutants emitted from a source category based on what the best performing similar sources have achieved,
i.e.,
the MACT floor. The Agency, however, may consider adopting alternative health-based emission standards — which are more lenient — for pollutants with an established health threshold.
Using this authority, the EPA considered and adopted health-based emission standards for HC1 in an earlier rulemaking for major boilers.
See
2004 Boilers Rule,
The EPA again chose to consider a health-based standard for HC1 in the current rulemaking, but this time declined to set such a standard. 2010 Major Boilers Rule,
Applying this interpretation, the EPA suggested in its proposed rule that a health-based standard for HC1 might not be appropriate because these additional health and environmental considerations cautioned against a more lenient emission standard.
Id.
at 32,031. The Agency acknowledged, in particular, that its decision in the 2004 rule was based on data that considered
only
the chronic respiratory effects of HC1 exposure.
Id.
While affirming the validity of those findings, the EPA explained that those chronic impact studies did not consider the additional variables it had now identified, nor did it consider the potential acute or carcinogenic effects that might be caused by HC1 exposure.
Id.
And, because of these potential (though unproven) risks, the Agency resolved that it currently lacked sufficient information to establish an HC1 emission standard that would protect health with an ample margin of safety.
Id.
It thus requested additional data from stakeholders and the regulated community to help address its concerns,
After receiving numerous comments on the issue, the EPA declined to set a health-based standard in the final rule for two primary reasons: (1) the comments had not provided sufficient data on potential
cumulative
health and environmental effects caused by HC1 emissions from boilers and other nearby sources; and (2) the comments affirmed the potential
co-benefits
that limiting HC1 emissions might have in lowering emissions of other HAP and non-HAP pollutants. 2011 Major Boilers Rule,
The Industry Petitioners contend that the EPA’s consideration of the broad potential health and environmental impacts of HC1 rendered the Agency’s decision arbitrary and capricious. In particular, they argue that the Agency based its decision on two impermissible factors that were not supported by the record: (1) the potential cumulative effects of emissions from boilers and other nearby sources, and (2) the co-benefits of setting a more stringent MACT floor standard for HC1, We disagree on both counts.
The statutory text and purpose of section 7412(d)(4) amply support the Agency’s decision to consider potential cumulative risks associated with emissions from boilers and other nearby sources. Although other CAA provisions require the EPA to set emission standards based on the emissions from a particular source, section 7412(d)(4)’s plain language is not focused on emissions from any particular source. Compare 42 U.S.C. § 7412(d)(3) (instructing the EPA to set emission standards for sources at the level achieved in practice by the best controlled similar source), with id. § 7412(d)(4) (containing no mention of emissions from a particular source). The EPA’s consideration of the cumulative impacts from these emissions is also relevant to the Agency’s statutory mandate to ensure that a health threshold would protect health with an “ample margin of safety.” As such, the Agency had discretion to consider the potential risks associated with the cumulative emissions of boilers and other nearby sources under this provision.
The EPA was likewise free to consider potential co-benefits that might be achieved from enforcing the HC1 MACT floor. Section 7412(d)(4)’s text does not foreclose the Agency from considering co-benefits and doing so is consistent with the CAA’s purpose — to reduce the health and environmental impacts of hazardous air pollutants. The Agency was under no obligation to ignore the CAA’s purpose in making a final decision on whether to exercise a discretionary authority.
The Industry Petitioners attempt to refute this straightforward conclusion by pointing to “restrictions” in another provision, section 7412(d)(2). No. 11-1108 Indus. Pet’rs’ Br. 55-56. This provision requires the EPA to consider costs, non-air quality health and environmental impacts, and energy requirements in setting
maximum
achievable emission standards. Petitioners contend that these same “restrictions” must be read into section 7412(d)(4). But, even if we assume Petitioners are correct that these factors restrict the Agency’s ability to consider other factors under section 7412(d)(2), that provision furthers the statute’s command to set the strictest possible emission standards above what has already been achieved (i.e., the MACT floors). Section 7412(d)(4), by contrast, is a
Finally, the Industry Petitioners claim that the EPA’s decision was arbitrary because the Agency failed to support its reversal from the 2004 rule, in which it set health-based emission standards for HC1. Because the EPA changed its position, the Petitioners contend that the Agency had to present
factual
support for its decision to disregard the facts and circumstances that underlay its prior adoption of a health-based HC1 standard.
See FCC v. Fox Television Stations, Inc.,
At the outset, Petitioners misstate the EPA’s burden to justify its change in policy. Although an agency does not generally need to provide a more substantial explanation or reason for a policy change than for any other action, it must do so where “its new policy rests upon factual findings that contradict those which underlay its prior policy.”
Id.
at 515,
The EPA did so here by explaining that its prior decision focused too narrowly on the chronic respiratory effects of HC1 emissions without considering the broader implications of such emissions on health and environmental conditions. See 2010 Proposed Major Boilers Rule,
These enumerated concerns were sufficient to support the Agency’s decision not to adopt a health-based standard. Section 7412(d)(4) does not require that the EPA present affirmative factual data to reject a health-based standard. The provision requires just the opposite: in order to
impose
a health-based standard, the Agency must find that a health threshold can be set that provides an ample margin of safety. The EPA here determined that it could not do so, in part because it lacked relevant data like that discussed above. 2011
H. Emissions Averaging of Multiple CISWI Units in One Facility
Certain industry entities urged the EPA to allow facilities with more than one CISWI unit to demonstrate MACT compliance by showing that the average HAP emissions across all units at that location fell under the relevant cap. They pointed to the EPA’s allowance of emissions averaging in the Major Boilers Rule but the Agency defended its disparate treatment because, in its view, “[t]he applicability of CISWI is such that each unit is an affected facility.”
See
2011 Proposed CISWI Rule on Reconsideration,
Although the Industry Petitioners’ point is well taken — the plain terms of the CAA foreclose the EPA’s conflation of a CISWI “unit” and “affected facility,” see 42 U.S.C. § 7429(g)(1) (“facility” is comprised of “units”) — we agree that the-EPA has no statutory authority to allow emissions averaging under section 7429. 18 Section 7429 requires the EPA to regulate emissions from all “solid waste incineration units,” 42 U.S.C. § 7429(a)(2); see also id. § 7429(a)(4), and the CAA defines a “solid waste incinerator unit” as “a distinct operating unit” of a “facility,” id. § 7429(g)(1) (emphasis added). In other words, because the CAA mandates that the EPA regulate each “distinct” CISWI unit in a “facility,” the EPA cannot allow emissions averaging of all CISWI units in a facility. See id.
For this reason, the Industry Petitioners’
Chevron
challenge fails, notwithstanding the EPA’s minimal explanation set forth in its proposed CISWI Rule. It is axiomatic that an agency must “articulate[ ] an
adequate
explanation for its action,”
Int’l Fabricare Inst. v. EPA,
IV. ENVIRONMENTAL PETITIONERS’ CHALLENGES
A. Caebon Monoxide as a Surrogate
As explained at
supra
§§ I.B.1 and III.F, the EPA used carbon monoxide (CO) as a surrogate for several non-dioxin/furan organic HAPs when the Agency set the MACT floors for major boilers. In support of this approach, the EPA found that both CO and these HAPs were the products of “incomplete combustion.” 2010 Proposed Major Boilers Rule,
The Environmental Petitioners challenge this decision, arguing that the EPA has not adequately explained how setting emission standards for CO will accomplish what the statute plainly requires: that the EPA set emission standards for organic HAPs at the average level achieved by the best performers with regard to those HAPs. We agree and remand to the EPA to adequately explain how CO acts as a reasonable surrogate for non-dioxin/furan organic HAPs. We do not, however, vacate the current emission standards because we conclude that the Agency will likely be able to adequately explain its decision on remand and that vacatur would prove substantially disruptive.
The EPA may use a surrogate to regulate HAPs under section 7412 where “reasonable.”
See, e.g., Nat’l Lime Ass’n,
In the Major Boilers Rule, the EPA proposed using CO as a surrogate because, as relevant here: (1) the lowest possible CO emissions resulted in the lowest possible HAP emissions, and (2) the same combustion and oxidation control methods reduce both types of emissions.
See
2010 Proposed Major Boilers Rule,
Although we afford an agency’s scientific decision “an extreme degree of deference,”
see Kennecott Greens,
We recognize that there might be a context where a surrogate’s use is reasonable despite the presence of alternative control methods or technologies, but the Agency does not explain why it did not need to even
consider
whether such methods might further reduce HAPs here. For example, if the EPA used a surrogate that was closely correlated to the HAP and set surrogacy emission standards at a level that would eliminate HAP emissions altogether, the Agency might not need to account for alternative control technologies in its final rule. In that case, the use of the surrogate would not call into question whether the Agency had regulated the HAP as required by the statute because, after all, nothing is better than eliminating HAP emissions entirely. But the Agency offers us no analogous explanation or sup
We reject, however, the Environmental Petitioners’ other argument that combustion-related issues preclude the EPA from using CO as a surrogate for non-dioxin/fu-ran organic HAPs. The Petitioners contend that the EPA’s decision to use CO was arbitrary because record evidence demonstrated a breakdown in the correlation between CO and organic HAP emissions at CO emission levels below 130 parts per million (ppm). But the EPA explained that this apparent breakdown was most likely caused by the difficulty of measuring the regulated HAP at such extremely low emission levels, rather than by a flaw in the correlation between CO and organic HAPs. 2013 Major Boilers Rule,
Still, the EPA’s failure to address substantial record evidence on the potential availability of alternative control technologies or methods rendered the Agency's use of CO as a surrogate for certain organic HAPs arbitrary and capricious. We thus remand the portion of the Major Boilers Rule providing for CO’s use as a surrogate for non-dioxin/furan organic HAPs to the Agency for further consideration. We do not, however, vacate the current emission standards based on CO’s use as a surrogate. We may remand without vacatur where there is a likelihood of (1) cure on remand, and (2) a substantial disruptive effect that would result from vacatur.
See Heartland Reg’l Med. Ctr. v. Sebelius,
B. Exclusion op Certain Units prom MACT Analysis
In the Major Boilers Rule, the EPA created subcategories based primarily on the fuel combusted.
See
2013 Major Boilers Rule,
We cannot say the same about the EPA’s exclusion of certain high-performing units from its MACT-floor calculation. Although the EPA allowed sources that combust only 10 per cent of a subeategorydeflning fuel to join that subcategory, it declined to consider emissions from any source that burned less than 90 per cent of the subcategory-defining fuel when determining the average emissions level of the best performing sources in setting MACT floors for existing sources. And when it set ’ a subcategory’s MACT floors for new sources, the Agency declined to consider the emissions levels from any source that did not burn 100 per cent of the fuel. This disparate treatment makes a difference; several sources excluded from the MACTfloor determination were among the best performing sources (or, in some cases, the single best performing source) in that fuel-based subcategory.
The CAA, however, demands that source subcategories take the bitter with the sweet. Section 7412 mandates, without ambiguity, that the EPA set the MACT floor at the level achieved by the best performing source, or the average of the best performing sources, in a subcategory. See 42 U.S.C. § 7412(d)(3)(A), (B). It thus follows that if the EPA includes a source in a subcategory, it must take into account that source’s emissions levels in setting the. MACT floor.
The Agency, however, claims discretion to omit from MACT-floor computation sources it considers dissimilar. In support, it cites section 7412(d)(3), which provides that MACT standards must be no less stringent than “the best controlled
similar
source,
as determined by the [EPA].” Id.
§ 7412(d)(3) (emphases added). Our decision in
Sierra Club II,
The EPA tries to distinguish
Sierra Club II,
arguing that the issue in that case “was whether [the] EPA could exclude all units using the most-effective emission control technique because it might not be applicable to all existing units”; however, “[h]ere, [the] EPA is excluding a test result that is unrepresentative of typical operations of units in the subcategory, and thus is inappropriate to use in establishing the MACT floor.” No. 11-1108 EPA Br. 81. But it makes no difference whether the EPA exempts from consideration units with certain highly effective technology or units with impressive test results driven by the fuel combination it combusts. Either approach contravenes our holding in
Sierra Club II
that the EPA cannot ignore “the emission level actually achieved by the best performers (those with the lowest emission levels)” in the subcategory.
The EPA insists that if a source is “unrepresentative of typical operations of units in the subcategory,” it is “inappropriate to use [it] in establishing the MACT floor.” No. 11-1108 EPA Br. 81. Not so. “The idea is to set limits that, as an initial matter, require all sources in a category to at least clean up their emissions to the level that their best performing peers have shown can be achieved.”
Sierra Club I,
C. Upper Prediction Limit
Sections 7412 and 7429 create MACT-floor criteria that, for our purpose, are materially the same.
Compare
42 U.S.C. § 7412(d),
with id.
§ 7429(a)(2). In both provisions, the CAA mandates that MACT floors have maximum stringency but also be continuously achievable.
See id.
§ 7412(d)(2), (k);
id.
§ 7429(a)(2);
id.
§ 7602(k). Satisfying the statutory criteria is no easy task, especially because no source emits any HAPs at a constant level.
See
Page Mem. 6. Rather, emissions levels fluctuate over time and for many reasons.
See id.
at 3.
21
We have held,
see Mossville Envtl. Action Now v. EPA,
Further complicating the task is the way in which sources typically measure emissions. Virtually all of the data the EPA collects to set MACT floors come from the three-run stack test. Page Mem. 6. The three-run stack test, as the name suggests, involves three measurements of the source’s emissions taken over a short time period (ie., no more than a few days) with each of the three test “runs” lasting from one hour to four hours. Id. at 3. Because the tests provide three “snapshots” of a source’s emissions performance, they cannot accurately represent the source’s full range of emissions over all times and under all conditions. Id. at 3-4. Because stack testing typically involves “three separate runs,” however, it “will in most cases show some of a particular source’s variability over the short period of time during which testing was conducted.” Id. at 6 (emphasis added). 22
Based on the limitations inherent in stack testing, the EPA concluded that it could not set MACT floors based on that testing alone. It began using the UPL to account for the HAPs-emissions variety that stack-testing data do not reflect.
See NACWA,
Specifically, we struggled to pin down the EPA’s precise interpretation of the phrase “average emissions limitation achieved by the best performing 12 percent of units.” Id. at 1142-43 (quoting 42 U.S.C. § 7429(a)(2)). 23 As best we could tell, the EPA defended its use of the UPL as follows: “[bjecause the [UPL] represents the value which [the EPA] can expect the mean (i.e., average) of three future observations (3-run average) to fall below, based upon the results of the independent sample size from the same population, the [UPL] reflects average emissions.” Id. at 1142 (quoting Standards of Performance for New Stationary Sources and Emission Guidelines for Existing Sources: Sewage Sludge Incineration Units, 76 Fed. Reg. 15,372, 15,389 (Mar. 21, 2011)) (emphasis added) (some alteration in original). In our view, however, “the word ‘average’ ... seems to mean the average emissions limitation that the existing population of the best-performing 12 percent of incinerators has achieved.” Id. (emphases added).
Despite these doubts, we reasoned that the EPA could have “plausibl[y]” concluded that the UPL represents the “average emissions limitation achieved” by the best performing sources.
Id.
at 1143. That said, we were not willing to assume the EPA’s responsibility of “supplying] a reasoned basis” for its UPL use.
Id.
(quoting
Bowman Transp., Inc. v. Ark.-Best Freight Sys., Inc.,
Because the EPA also used the UPL in the Major Boilers Rule and the CISWI Rule, the Agency moved for a limited remand of the current petitions so that it could include its revised UPL explanation in the administrative records of these two regulations.
25
See
Page Mem. 2. On July
2. The Page Memorandum
The Page Memorandum recognized our “concern about the interpretation [we] believed [the] EPA was taking” of the word “average.” Page Mem. 3. It clarified that the Agency “does
not
interpret the term ‘average’ ” to mean “the average of a future 3-run compliance test.”
Id.
(emphasis added) (quoting
NACWA,
In the EPA’s judgment, “variability is a key factor in establishing” MACT standards because “[e]ach MACT standard is based on limited data from sources whose emissions are expected to vary over their long term performance.” Id. Specifically, “[t]he available emissions data are generally in the form of short term, three-run stack tests, with each test run lasting for between 1 and 4 hours.” Id. For this reason, the EPA concluded that it did not have information “encompassing] the emissions performance of a source over time.” Id. (emphasis added). And because the “EPA interprets ‘emissions performance’ ... to mean the emissions of a source over the long term, rather than just during a short-term stack test,” the EPA found it necessary to “appl[y] a methodology that predicts the actual emissions levels the source is achieving at times other than when stack testing was conducted.” Id. at 3-4 (emphases added).
The UPL is the methodology the EPA selected to account for' these limitations.
Id.
at 4. “[A] value derived from widely
The UPL, however, cannot demonstrate with absolute certainty the average emissions levels achieved by the best performing sources at all times (indeed, certainty is impossible without continuous monitoring). See id. Instead, the UPL equation produces a range of values that is expected, given the variance in the relevant stack-test data, to encompass the average emissions levels achieved by the best performing sources a specified percentage of the time. Id. at 4. To establish the MACT floor, the EPA calibrated the UPL equation to produce a range in which the average emissions levels of the best performing source or sources would be expected to fall 99 per cent of the time, which is referred to as a 99 per cent confidence interval. Id. Once the EPA had this range, it set the MACT floor at the top level of that range — hence, the “upper” in “upper prediction limit” — to arrive at a figure that, 99 out of 100 times, it expected the average emissions levels of the best performing sources to “achieve.” Id. Or, in the EPA’s words, “the 99 percent UPL is the level of emissions that” the EPA is “99 percent confident is achieved by the average source represented in a dataset over a long-term period based on its previous, measured performance history as reflected in short term stack-test data.” Id.
One of the equations the EPA used to calculate the UPL is as follows: 27
[[Image here]]
NACWA, 734 F.3d- at 1139. In this equation:
• “x” is the mean;
• “¿(0.99, n-1)” is a value called the “t-statistic,” the statistical tool used to set the confidence interval (here, 99 per cent);
• is the sample size;
• “m” is the number of stack tests that were run to calculate the mean (“x”); because most stack tests involve 3 “runs,” m usually equals 3;
• “s” represents the “standard deviation.”
See id.; see also Page Mem. 10-11.
3. Instant Challenges to UPL
After
the
EPA issued
the Page
Memorandum, the Environmental Petitioners renewed their argument that the UPL represents neither (1) the “average” emissions limit of the best performing source or sources in a subcategory, nor (2) the emissions levels “achieved” by the best performing sources in a subcategory. We believe that the EPA has carried its burden of demonstrating that the UPL “reflect[s] a reasonable estimate of the emissions achieved in practice by the best performing sources.”
Cement Kiln,
Our conclusion is driven, in large part, by the deference we owe the EPA when it determines how best to meet the technical challenges in its area of expertise. Indeed, the EPA “typically has wide latitude in determining the extent of data-gathering necessary to solve a problem” and, for that reason, we have “accorded
Chevron
deference to [its] interpretation of [the CAA] as allowing it to estimate MACT floors.”
NACWA,
The Agency has done so here. The Page Memorandum explains the limitations of stack-test
data
— i.e., the “snapshots” cannot reflect the best performing source’s or sources’ average emissions levels at all times and under all operating conditions. Page Mem. 6. The Page Memorandum also explains that the Agency chose the UPL as a tool “derived from widely accepted and commonly used statistical principles,”
id.
at 4, that “reasonably accounts] for
The Environmental Petitioners’ arguments to the contrary are unavailing. Their primary objection is that the UPL cannot reasonably estimate the “average” emissions level achieved by the best performing source or sources because the UPL represents “a level [the] EPA expects any future compliance test by any [source] in the top 12 percent to fall below.” No. 11-1108 Envtl. Pet’rs’ Br. 35 (emphases added) (internal quotation marks omitted). 28 But the Page Memorandum counters the Environmental Petitioners’ mistaken understanding of what the UPL represents. 29 According to the EPA, “the UPL does not represent the worst emissions performance of the best performing units at any time.” Page Mem. 4 (emphasis in original). 30 It is instead “the average level expected to have been achieved over time” by the best performing source or sources. Id. (emphasis in original). “In other words, the 99 percent UPL is the level of emissions that [the EPA is] 99 percent confident is achieved by the average source ... over a long-term period based on its previous, measured performance history as reflected in short term stack test data.” Id. (emphasis added).
Next, the Environmental Petitioners criticize the Page Memorandum’s explanation that the UPL represents the long-term average emissions levels achieved because “the first element of the UPL equation is the average of the short-term emissions test data from the best-performing sources.”
Id.
In their view, the UPL is no different from “saying that, over time, the average of 1, 2, and 3 = 2 + 500 because the first element in the equation (2) is the average of 1, 2, and 3.” No. 11-1108 Envtl. Pet’rs’ Br. 48. But the UPL does not simply tack an arbitrary increase on top of the stack-test average of the best performing sources. Rather, the UPL “allows [the] EPA to use emissions test data
and the data characteristics,”
which include “the distribution and sample size, along with the intrinsic variability
associated with those data,”
to estimate “an emissions limit based on a specified level of confidence such that an average best performing existing source would not be expected to exceed the limit a specified number of times.” Page Mem. 6 (emphases added). In other words, the UPL does not simply add
The Environmental Petitioners also attack the results produced by the UPL. They provide a series of charts that, in their view, demonstrate that the UPL sets MACT floors far too high to comport with the CAA’s mandate that floors represent “the maximum degree of reduction in emissions.” See 42 U.S.C. § 7412(d)(2). True, some of the charts show that the EPA has set a MACT floor above the highest emissions level recorded by the best performing sources’ stack testing. See No. 11-1108 Envtl. Pet’rs’ Br. 14-15; No. 11-1108 Envtl. Pet’rs’ Reply Br. 23. But this does not mean that the UPL is an arbitrary “average” proxy — for at least two reasons.
First, the charts selectively included are generated from data sets with considerable variance between the highest recorded stack test and the lowest. Unsurprisingly, if a handful of “snapshots” in a data set demonstrate that emissions levels experience high spikes and low plummets at discrete times, it is more likely that the average emissions level achieved by the best performing sources at
all
times might be high. This is because a data set with high variability will produce a higher UPL than a data set with low variability, even if the two sets share the same average. In other words, the UPL takes large variance into account and therefore naturally goes higher to arrive at the 99 per cent certainty the EPA thinks is appropriate.
32
Second, where the UPL suggested a MACT floor higher than the results of the stack tests, it often did so by insubstantial amounts. Indeed, for at least one chart, “the limit is a mere 4
millionths
of a pound per million Btu above the emissions test results of best performers, an una-larming amount given that the methodology is supposed to account for
variable
results.” No. 11-1108 Indus. Intervenors’ Br. 10 (emphases in original). For these reasons, the Environmental Petitioners have not convinced us that the EPA failed to satisfy the “minimal standard[] of rationality” that we require.
Ethyl Corp. v. EPA,
Finally, the Environmental Petitioners insist that “[t]he UPL predicts a level that hypothetical future tests will fall below, rather than estimating what boilers actually achieved,” in contravention of the requirement that MACT floors “reflect what
We believe that the UPL “reflect[s] a reasonable estimate of the emissions achieved in practice by the best-performing sources,”
Cement Kiln,
D. Beyond-the-Floor Standards for CISWI Units
The EPA declined to set beyond-the-floor standards for CISWI units. The Environmental Petitioners challenge that determination in three primary respects, each of which we reject. 34
Section 7429 of the CAA directs the EPA to set MACT standards in two steps. It first sets a floor level based on the best performing sources.
See
42 U.S.C. § 7429(a)(2). Next, it determines “whether a more restrictive standard is ‘achievable,’ ”
NRDC III,
When establishing MACT standards for CISWI, the EPA declined to establish beyond’-the-floor standards in the proposed rule,
see
2010 Proposed CISWI Rule,
The first challenge targets several instances in which the EPA refused to require sources to adopt, as a beyond-the-floor measure, controls that most sources would employ to meet the MACT floor standard. In each instance, the EPA determined that the relative costs outweighed the expected emissions gains. In the first such case, the EPA decided not to require liquid-fired energy recovery units to install dry sorbent injection and fabric filters as a beyond-the-floor measure, despite the fact that “four of the six” units would need to install those systems to meet the floor standard. 2011 CISWI Rule,
The Environmental Petitioners take issue with two other decisions along these lines. In the first, the EPA declined to set beyond-the-floor mercury control measures for kilns, citing a cost-effectiveness of roughly $351 million per ton. See Memorandum from Eastern Research Group, Inc., to Toni Jones, U.S. Environmental Protection Agency, Final Reconsideration Beyond-the-Floor Analysis for CISWI Units (Reconsideration Mem.) (Dec. 20, 2012), at ¶ 3.4.2 (No. 11-1125 J.A. 1232). In the second, a $26,000 per-ton implementation cost led the Agency not to establish stricter carbon monoxide control measures for calciner kilns. See. id. ¶ 3.4.3. Energy use — a factor mandated in section 7429(a)(2) — also entered the equation. With respect to calciner kilns, the technology used to reduce carbon monoxide would also increase energy requirements, and therefore increase energy costs. See id. In each of these decisions, the EPA reasonably applied the statutory factors. That Petitioners would have weighed the costs differently provides no grounds to displace the EPA’s otherwise reasonable determination.
In the second challenge to the decision not to set beyond-the-floor standards, the Environmental Petitioners contend the Agency arbitrarily failed to set emission levels
lower
than the MACT floor for categories likely to adopt technology capable of meeting those lower levels. Specifically, ac
That is incorrect. The Environmental Petitioners spin this yarn based on a line in the proposed rulemaking. There, the Agency speculated that kilns and energy recovery units would adopt fabric filters to comply with the MACT floor limit, and would “likely achieve a level of performance” below the floor. 2010 Proposed CI-SWI Rule,
In the third challenge, the Environmental Petitioners take issue with three determinations that rested on factors other than cost. First, the EPA declined to require coal-fired energy recovery units to adopt linkageless boiler management systems as a beyond-the-floor measure for carbon monoxide. See Reconsideration Mem., ¶ 2.3.1.1. While acknowledging that link-ageless systems were available at “fairly low-cost,” the EPA concluded it had insufficient data to determine the “actual reductions this control option would achieve” relative to an alternative control system. Id.
The EPA acted reasonably. The record suggests the EPA had scant evidence on the efficacy of linkageless control measures applied to coal-fired energy recovery units.
See id.
Had the Agency imposed a stricter standard based on controls for which it had precious little (if any) evidence, a reviewing court may well have concluded' the decision lacked “a rational connection between the facts found and the choice made.”
State Farm,
Second, the EPA rejected regenerative thermal oxidizers as a beyond-the-floor control for carbon monoxide in solid waste energy recovery units.
See
2011 CISWI Rule,
The Environmental Petitioners contend that the EPA failed to “suggest that these natural gas requirements are high in an absolute sense or relevant to achievability.” No. 11-1125 Envtl. Pet’rs’ Br. 51. We agree that the EPA’s analysis is less than
Despite these imperfections, we reject the challenge.
See Dist. Hosp. Partners, L.P. v. Burwell,
The Agency’s determination should be read in context. Elsewhere in the final rule, the EPA expanded on the energy and environmental impacts of thermal oxidizers, concluding that “[t]he combustion of fuel needed to generate additional electricity and to operate [thermal oxidizer] controls would yield slight increases in emissions, including NOx, CO, PM, and S02 and an increase in C02 emissions.” 2011 CISWI Rule,
Though courts are powerless to “supply a reasoned basis for the agency’s action that the agency itself has not given,” “[w]e will ... uphold a decision of less than ideal clarity if the agency’s path may reasonably be discerned.”
State Farm,
Third, and finally, the Environmental Petitioners challenge the rejection of dry sorbent injection and wet scrubbers as beyond-the-floor measures for waste-burning kilns. The EPA determined those measures would be cost-effective (at only $5,000 per ton) but declined to require them due to “uncertainty” surrounding “the appropriate control system that, some existing kilns would need to employ to meet” a stricter standard, “especially kilns that use ingredients with a high sulfur content.” See Reconsideration Mem., ¶ 3.4.5. Adding to that uncertainty, the EPA could not “account for potential costs at existing sources for additional scrubber water and spent sorbent.” Id. As before, the EPA reached a reasonable conclusion in the face of imperfect information. Had the EPA set a beyond-the-floor standard based on sor-bent injection and wet scrubbers, the Agency would have been flying blind. In avoiding that course, we conclude the EPA acted reasonably.
E. Regulation of Certain CISWI Units
The final CISWI Rule did not contain emission standards for burn-off ovens, cyclonic burn barrels, foundry sand reclamation units, soil treatment units, and space heaters. The Environmental Petitioners claim that the EPA unlawfully exempted these units from regulation by creating subeategories that capture only a subset of the units that the Agency is required to regulate as CISWI. The EPA, however, protests that it did not exempt these five types of units from regulation. Rather, the Agency determined that it lacked sufficient data to regulate the units at this time, and, with respect to some, it
We agree with the Environmental Petitioners that the Agency has violated its nondiscretionary statutory duty (1) to promulgate standards with respect to cyclonic burn barrels, and (2) to determine whether the remaining four types of units fall within the statutory definition of CISWI. The CAA requires the EPA to “establish performance standards ... for each category of solid waste incineration units” no later than November 15, 1994.
See
42 U.S.C. § 7429(a)(1)(A), (D). The statute then defines “solid waste incineration unit” as a “distinct operating unit of
any
facility which combusts
any
solid waste material from commercial or industrial establishments or the general public.”
Id.
§ 7429(g)(1) (emphasis added). That provision unambiguously requires the EPA to set emission standards for “any facility that combusts any commercial or industrial solid waste material at all,” subject only to the listed statutory exceptions.
NRDC I,
The Agency makes no effort to claim that cyclonic burn barrels fall outside the statutory definition for CISWI units. Nor could it — both the administrative record and the EPA’s brief make clear that cyclonic burn barrels “combust” solid waste.
See
2011 Proposed CISWI Rule on Reconsideration,
The EPA also had a duty to determine whether the other challenged sources— burn-off ovens (including foundry sand reclamation units), soil treatment units, and space heaters — were units that “com-bust” solid waste. Several commenters told the Agency that these units fell within the statutory definition of CISWI, and the EPA itself initially viewed some of these units as combusting waste.
See, e.g.,
CI-SWI Rule — Responses to Comments, at 74-76; 2010 Proposed CISWI Rule,
The CAA unambiguously requires that the Agency establish standards for all CI-SWI units. As a result, we grant the Environmental Petitioners’ petition for review on this issue and remand to the Agency to set emission standards for cyclonic burn barrels. The EPA must also determine whether the remaining four types of units are CISWI units and, if it finds that they are, it must set standards for them as well.
F. Delisting under 42 U.S.C. § 7412(c)
In contrast to major source subcategories (all of which the EPA must control), the CAA does not require the EPA to control emissions in every area source subcategory.
See
42 U.S.C. § 7412(c)(1), (3). The Act does, however, mandate that the EPA control area source emissions if the area source subcategory meets certain criteria. Section 7412(c)(1), for instance, requires the EPA to control any area source subcategory upon the Agency’s finding that emissions from the sources in the subcategory jeopardize either the environment or human health.
See id.
§ 7412(c)(3). If so, the EPA can establish either a MACT or a GACT standard.
See id.
§ 7412(d)(5). Similarly, if the EPA finds that capping emissions from an area source subcategory is necessary to achieve
In addition to prescribing requirements for inclusion of area source subcategories, the CAA provides a mechanism for removal of area source subcategories that, in the EPA’s view, no longer need to be controlled. Specifically, the EPA can “delete” any subcategory if it finds that no source or group of sources in it (1) emits cancer-causing HAPs at a volume sufficient to increase the lifetime risk of cancer in the population by more than one in one million and (2) emits non-cancer-causing HAPs at a level in excess of that which is adequate “to protect public health with an ample margin of safety” and to prevent against environmental harm. Id. § 7412(c)(9)(B). The section 7412(c)(9) process is known as “delisting.”
In 1998, the EPA identified several area source boiler subcategories — including oil-fired, industrial wood, commercial oil-fired and commercial wood-combustion boilers— as contributors to the “90 per centum of the aggregate emissions” of Hg and POM under section 7412(c)(6). See Source Category Listing for Section 112(d)(2) Rule-making Pursuant to Section 112(c)(6) Requirements, 63 Fed. Reg. 17,838, 17,839 (Apr. 10, 1998); When it decided to “list” these sources, however, the EPA included a caveat. It explained that it used the best emissions information it had at the time to conclude that these boiler subcategories produced enough Hg and POM emissions to justify section 7412(c)(6) control but it also admitted that it could not “assure that, this calculation of the 90 percent will remain constant.” Id. at 17,840.
The caveat proved prescient. When the EPA issued the 2010 Proposed Area Boilers Rule, it decided it needed to regulate only coal-fired boilers at the MACT level to control 90 per cent of Hg emissions.
See
For this reason, the EPA established GACT, rather than MACT, standards for the oil-fired and biomass-fired area source subcategories regarding these two pollutants.
See id.
It did not, however, make any of the “delisting” findings required by section 7412(c)(9) when it removed these area source subcategories from section 7412(c)(6)’s purview.
See
2011 Area Boilers Rule,
Because section 7412(c)(9) does not unambiguously apply to section 7412(c)(6) and because the EPA’s interpretation of section 7412(c)(9)’s delisting requirement is reasonable, we uphold the EPA’s decision as permissible under
Chevron
step 2. Section 7412(c)(9) provides that
In the Environmental Petitioners’ view, section 7412(c)(9) also applies to a second, subsidiary list — that contemplated by section 7412(c)(6), requiring imposition of the MACT standard. Granted, section 7412(c)(6) mandates that the EPA “shall ... list” source categories and subcategories if doing so is necessary to control 90 per cent of the aggregate emissions from seven enumerated pollutants.
Id.
§ 7412(c)(6) (emphasis added). But the use of the verb “list” in section 7412(c)(6) does not unambiguously establish that 7412(c)(9), titled “[deletions from the list,” applies. Because section 7412(c)(9) is ambiguous, we defer to the EPA so long as its interpretation is “based on a permissible construction.”
Chevron,
First, the EPA’s approach harmonizes sections 7412(c)(1), 7412(c)(3), and 7412(c)(9). Because the EPA must find that an area source “presents a threat of adverse effects to human health or the environment” before it regulates the source category at all, id. § 7412(c)(3), it makes sense to require the EPA to find that “no source in the category or subcategory ... exeeed[s] a level which is adequate to protect public health ... and no adverse environmental effect will result from emissions from any source” before it completely deregulates that category, id. § 7412(c)(9). It makes less sense to require the EPA to make the same findings before it opts for GACT instead of MACT standards, which occurs when the EPA removes a source from section 7412(c)(6)’s purview but continues to regulate it under section 7412(c)(1).
Second, the EPA’s approach is consistent with our decision in
New Jersey,
Finally, the Petitioners’ argument would freeze the EPA’s decision as to which sources need to be controlled to reach the requisite 90 per cent emissions reduction for the section 7412(c)(6) pollutants until it determines that “no source in the category or subcategory ... exceed[s] a level which is adequate to protect public health ... and no adverse environmental effect will result from emissions from any source.” 42 U.S.C. § 7412(c)(9). This, in turn, would hamper the EPA’s ability to respond to
G. Title Y Permit Exemption for Synthetic Boilers
The EPA has discretion to exempt one or more area source categories from Title V permitting requirements upon a finding “that compliance with such requirements is impracticable, infeasible, or unnecessarily burdensome on such categories.” 42 U.S.C. § 7661a(a). The EPA originally proposed exempting some area source categories because existing “testing, monitoring, notification, and recordkeeping requirements” rendered Title V permitting cumulative. 2010 Proposed Area Boilers Rule,
Environmental Petitioners argue the EPA’s decision to exclude synthetic boilers from Title V licensing requirements is arbitrary and capricious for two reasons. First, they say, the EPA arbitrarily concluded synthetic area sources would bear the same level of burden as other area sources in complying with Title V permitting requirements, rather than a lesser one.
See
No. 11-1141 Envtl. Pet’rs’ Br. 39-43. And second, they contend the EPA arbitrarily dismissed the additional compliance benefits of Title V licensing for these synthetic sources.
See id.
at 43-47. Under
State Farm,
“an agency rule [is] arbitrary and capricious if the agency ... offered an explanation of its decision that runs counter to the evidence before the agency.”
The EPA has authority under the CAA. to exempt sources from Title V permitting requirements if those requirements would be “impracticable, infeasible, or unnecessarily burdensome” on the area source. 42 U.S.C. § 7661a(a). The EPA previously developed a four-factor balancing test to determine whether Title V’s requirements are “unnecessarily burdensome.” See Exemption of Certain Area Sources from Title V Operating Permit Programs, 70 Fed. Reg. 75,320, 75,323 (Dec. 19, 2005). Under this test, the EPA considers whether: (1) Title V permitting would result in significant improvements in compliance with emission standards; (2) whether Title V permitting would impose significant burdens on the area source category; (3) whether the costs are justified, taking into account potential gains; and (4) whether there are existing enforcement programs in place sufficient to ensure compliance. See id. at 75,323-26. The EPA also must consider, consistent with the legislative history of the CAA, whether exemption would “adversely affect public health, welfare, or the environment.” Id. at 75,333-34. These factors are considered in combination and not every factor must point in favor of exemption for the EPA to choose that course. See id. at 75,323.
In its 2010 Proposed Area Boilers Rule, the EPA applied this balancing test and excluded almost all area source boilers
except
synthetic boilers that achieved “area” status via installation of a control technology (although it exempted those that achieved “area” status through operational
As to the second factor, the EPA noted that subjecting most area sources to Title V would “impose[] certain burdens and costs that do not exist outside of the [t]itle V program.” Id. at 31,912. One of the EPA’s major concerns was that “requiring permits for the large number of area sources could, at least in the first few years of implementation, potentially adversely affect public health, welfare, or the environment by shifting [s]tate agenciesf] resources away from assuring compliance for major sources with existing permits to issuing new permits for these area sources, potentially reducing overall air program effectiveness.” Id. at 31,913. For the third factor, the EPA concluded the costs of compliance would “impose a significant burden on many of the approximately 137,-000 facilities affected by this proposed rule” with only “low” potential gains in compliance. Id. at 31,912. Finally, for the fourth factor, the EPA determined that “[s]tate delegated programs are sufficient to assure compliance with this [rule],” and noted that the Agency retains authority to enforce this rule “anytime.” Id. The EPA therefore proposed exempting these area sources from the permitting requirements. See id. Environmental Petitioners are not currently challenging the exemption for non-synthetic area boilers.
However, in this 2010 rulemaking, the EPA also explained precisely why it declined to exempt synthetic area sources that installed air pollution controls from Title V requirements. First, the EPA noted these synthetic area sources “represent less than one percent of the total number of sources that will be subject to the final rule.” Id. at 31,913. The EPA also characterized these sources as “much more like the major sources” that are not exempt from Title V permitting requirements. Id. Further distinctions included that “many of these sources are located in cities, and often in close proximity to residential and commercial centers where large numbers of people live and work,” that they “have significantly higher emissions potential when uncontrolled” (even compared to synthetic boilers that adopted operational limits to attain area source status), and that many of these sources “are large facilities with comprehensive compliance programs in place” as opposed to small facilities, like schools or hospitals. Id. Given these distinctions, the EPA concluded additional public involvement and compliance oversight through Title V was “important to ensure that these sources are maintaining their emissions at the area source level.” Id.
But the EPA shifted its position in the 2011 Area Boilers Rule by deciding to
In its next iteration of the rule, the EPA endeavored to further explain its exemption of synthetic sources. The EPA again stated it “lacked sufficient information” to distinguish these synthetic sources from other area sources.
See
2011 Proposed Area Boilers Rule on Reconsideration,
Based on this record, the EPA’s reasoning has several fatal flaws that ren
Because its justifications for the final rule contradict earlier findings, the EPA must provide some reasoning to explain why its final decision “runs counter to the evidence before the agency.”
State Farm,
The EPA’s major oversight was its failure to explain why the rationale it used to exempt natural area sources from Title V could be identically applied to synthetic area sources. One of the Agency’s main justifications for exempting natural area sources was that their prolific numbers might overwhelm state and local regulatory agencies, diverting resources from other important environmental programs, thereby harming public health and welfare. The EPA never explained why requiring 48 synthetic area sources to comply with Title V would strain government resources to a comparable degree as would requiring the 137,000 natural area sources to comply. As discussed above, the EPA also did not explain how it suddenly determined these synthetic area sources were “similar in size and sophistication” to natural sources, when it had previously articulated several
The EPA relies on another problematic premise when it claims the potential benefits of subjecting synthetic area sources to Title V requirements are low. Both the EPA and Industry Intervenors argue that the added benefits of Title V would be minimal for these synthetic sources, relying solely on the rationale given for natural sources. But the EPA originally asserted “additional public involvement and compliance assurance requirements through title V [are]
important
to ensure that these sources are maintaining their emissions at the area source level.” 2010 Proposed Area Boilers Rule,
Similarly, for factor three’s balancing of costs and benefits, the EPA never justifies applying to natural sources — which tend to be small sites like schools, hospitals, and churches — the same rationale it applies to these larger synthetic sources, which tend to be located at refineries, chemical plants, and factories. Given these distinctions, it is at least possible this balancing would lead to a different outcome for synthetic sources. Taken as a whole, the EPA’s analysis fails to explain why several of the facts and characteristics it relied on for its initial assessment are no longer relevant— creating several glaring inconsistencies in the rulemaking record. The EPA offers no plausible reason for applying the results of the four-factor test for natural sources wholesale to these control-technology-dependent synthetic sources. We do not hold, however, that the EPA can never remove
With respect to remedy, there is a strong possibility that the Agency can properly explain its decision to exclude synthetic boilers from the Title V permitting requirement; moreover, vacating the decision would be unnecessarily disruptive for synthetic boiler operators who, in the interim, would not know whether they needed to begin the expensive, time-consuming process of obtaining a Title V permit.
See Allied-Signal, Inc. v. U.S. Nuclear Regulatory Comm’n,
H. GACT Standard Determinations
With few exceptions, the EPA has broad discretion to choose how to control area source emissions. For instance, the EPA has discretion to choose between GACT and MACT standards in the majority of cases. See 42 U.S.C. § 7412(d)(5). Even if the EPA chooses a MACT standard, it has discretion — although somewhat circumscribed — to set a work-practice standard instead of a numeric standard. Id. § 7412(h)(1). And the EPA has discretion when choosing among different GACT-standard options. See id. § 7412(d)(5).
Accordingly, we must uphold the EPA’s GACT-standard determinations so long as it “has considered the relevant factors and articulated a rational connection between the facts found and the choice made, and has not relied on [improper] factors.”
Nat’l Ass’n of Clean Air Agencies,
With these principles in mind, we address the Environmental Petitioners’ two challenges to the EPA’s discretionary decisions regarding the Area Boilers Rule.
1. EPA’s Selection of GACT Standards for Non-Hg Metals
The Environmental Petitioners argue that the EPA failed to support its decision-making when it established MACT standards for Hg and POM emissions from some coal-fired boilers but declined to regulate
non-'Rg
emissions under the MACT standard from the same boilers. We agree. Although the- EPA thoroughly explained why it chose to-impose one GACT standard instead of another,
2. EPA’s Selection of Certain GACT Standards
The Environmental Petitioners also challenge several of the EPA’s choices among different GACT standards. As noted,
see supra
§ I.A.l.c, the CAA provides virtually no instruction regarding GACT standards but the standards generally take the form of “methods, practices and techniques which are commercially available and appropriate for application by the sources in the category considering economic impacts and the technical capabilities of the 'firms to operate and maintain the emissions control systems.” S. Rep. No. 101-228, at 171 (1989). Because the EPA has ample discretion to choose the appropriate GACT standard, we will affirm its choices so long as we can discern reasoned decision-making from the record.
State Farm,
First, the Environmental Petitioners challenge the data set the EPA used to arrive at the numeric GACT standards for non-Hg-metal emissions from coal-fired boilers. Specifically, they contend that the EPA set the GACT limit based on boilers with no control technology, which resulted in a numeric standard of 0.42 lb/mmBtu. They insist that the EPA should instead have examined boilers outfitted with fabric filters, which would have resulted in a numeric standard of 0.03 lb/mmBtu. The EPA, however, thoroughly explained why it considered the uncontrolled boiler data set. Specifically, the controlled data set derives from the EPA’s “New Source Performance Standards” (NSPS) data, which, in the Agency’s view, could be used to set the non-Hg-metal GACT standard for boilers with a heat input capacity of 30 mmBtu/hr or greater but did not suffice for boilers with a lower heat input capacity. For this reason, the EPA examined its original data set, found that none of the coal-fired boilers in that set used control technology and, accordingly, set the GACT numeric standard at the emissions level achieved by the best performing uncontrolled source in that data set
(ie.,
0.42 lb/mmBtu). We are satisfied that the EPA exercised its discretion in a reasoned manner and, accordingly, we do not disturb it.
See Transactive Corp.,
Next, the Environmental Petitioners challenge the EPA’s decision to establish a tune-up requirement as a GACT management-practice standard for Hg and POM emissions from large biomass-fired and oil-fired boilers. In their view, other, more restrictive control technologies, including multiclones, 38 are “generally available” and their availability mandates that the EPA set numeric standards based on boilers that use those controls. But the EPA explained its approach:
A boiler tune-up requirement would potentially result in the same non-mercury metallic HAP reduction as a PM emission limit based on performance of mul-ticlones but would also reduce emissions of organic HAP. In addition the cost of a boiler tune-up appears minimal compared to the cost for testing and monitoring to demonstrate compliance with an emission limit.
See
2010 Proposed Area Boilers Rule,
Finally, the Petitioners challenge the EPA’s decision to set a tune-up requirement as a management-practice standard for small biomass-fired and oil-fired area boilers. The EPA adopted this approach because measuring PM emissions for smaller boilers is “not feasible.” 2010 Proposed Area Boilers Rule,
I. 30-Day Rolling Average
As discussed,
see supra
§ I.B.l.a, when the EPA sets a MACT floor, it begins by examining data generated by stack testing. Once the MACT standard is established, however, a source may (and in some cases, must) demonstrate compliance by implementing “continuous monitoring” instead of conducting additional stack tests.
39
See
2011 Proposed CISWI Rule on Reconsideration,
The calculation of a thirty-day rolling average is straightforward: the average of a source’s daily emissions for the immediately preceding thirty days. Each day produces a new rolling average and each “average is a separate compliance determination.” No. 11-1125 EPA Br. 88 n.17. In the EPA’s view, this “allow[s] operators sufficient flexibility for operational and control device adjustments should they be needed for short term fuel or waste characteristics variability.” 2011 Proposed CISWI Rule on Reconsideration,
The CAA vests the EPA with authority to “prescribe procedures and methods for determining compliance and for monitoring and analysis of pollutants.” 42 U.S.C. § 7661c(b). We have emphasized
First, they argue that the thirty-day rolling average fails at Chevron step 1 because it allows sources to emit HAPs continuously at the UPL-established MACT floor. Because they do not believe that the UPL represents the average emissions level achieved by the best performing sources, they argue that, ipso fac-to, allowing sources to continuously emit HAPs at the UPL level means that sources are permitted to emit at levels higher than the average levels achieved by the best performing sources. Because we have already concluded that the UPL is in fact a reasonable proxy for the average emissions level achieved by the best performing sources, see supra § IV.C, the Environmental Petitioners’ premise is inaccurate. And because the “total emissions from a unit complying with a rolling average must still be below the total emissions from a unit emitting continuously at the level of the standard,” No. 11-1125 EPA Br. 90, the Environmental Petitioners’ Chevron step 1 argument fails.
The Environmental Petitioners’
Chevron
step 2 argument fares no better. The EPA explained that (1) it expects to catch violations using a thirty-day rolling average “almost as much as for a shorter term average” and (2) it believes the longer average to be more effective in addressing “[cjoncerns of variability outside the oper-atorsf] control such as fuel content, seasonal factors, load cycling, and infrequent hours of needed operation.” 2011 Proposed CISWI Rule on Reconsideration,
Finally, we conclude that the EPA’s allowance of thirty-day rolling averaging does not reflect an arbitrary change in position. Although the Petitioners cite other rules that, in their view, manifest that the EPA once believed that longer averaging periods resulted in less stringent enforcement, most of the rules they cite have nothing to do with MACT-setting or MACT compliance
40
and none evidences an unexplained or unjustified deviation. Similarly, the Petitioners point to the EPA’s explanation in the 2011 CISWI Rule that “24-hour block averages ... would be inconsistent with the sampling time for the stack test data” to indicate an arbitrary change in position.
See
J. Fuel-Combustion-Based Subcategories
Section 7412 provides that the EPA may distinguish among “classes, types, and sizes” of sources when establishing emission standards. 42 U.S.C. § 7412(c), (d)(1). Under this authority, the EPA created subcategories of major source boilers based on the fuel the boiler was designed to burn. 2013 Major Boilers Rule,
These arguments fail. Section 7412(d) gives the EPA discretion to create subcategories based on boiler type, and nothing in the statute forecloses the Agency from doing so based on the type of fuel a boiler was designed to burn. Nor was the EPA’s decision to create such subcategories arbitrary and capricious. The Agency considered the relevant factors in coming to a reasoned decision that the type of fuel a boiler is designed to burn impacts the feasibility of emission standards. And, finally, the EPA based its technical judgment on sufficient record evidence. As a result, we deny the Environmental Petitioners’ challenge to the EPA’s subcate-gorization of major source boilers based on the type of fuel the boiler is designed to burn.
The Environmental Petitioners first claim that the text of the CAA forecloses the EPA from creating subcategories of “types” of boilers based on the fuel a boiler burns because a single boiler may use different fuels over the course of its lifetime. This may be true, but the Petitioners never .explain what it is about the word “type” that bars the EPA from regulating a boiler that burns “x” differently from a boiler that burns “y.” According to its ordinary meaning, “type” is easily broad enough to accommodate changes in boiler characteristics from year to year. See Oxford English Dictionary (2013) (defining “type” as a “general form, structure, or character distinguishing a particular kind, group, or class of beings or objects”). There is no textual reason then to assume that a boiler’s type must be written in stone.
Nor does our understanding of “type” write it out of the statute, as the Petitioners contend. The EPA has done what the term plainly encompasses: it has distinguished among boilers based on the kind' of fuel the boiler burned over the last year. It is thus not surprising that we have interpreted a similar provision to permit distinctions based on fuel inputs.
See Sierra Club v. Costle,
The Environmental Petitioners nevertheless claim that the EPA’s subcategories are arbitrary because a boiler is not of a different type when it can be a boiler “designed to burn coal” one year, and a boiler “designed to burn biomass” the next. But this argument fails for the same reasons as the Chevron argument we just rejected. The fact that boilers may switch from one type to another over time does not, alone, render a subcategorization arbitrary. With no discernable basis to find the EPA’s choice here questionable, much less arbitrary, we reject this argument too.
Finally, the Environmental Petitioners contend that the Agency failed to demonstrate with sufficient evidence, rather than mere assertions, that burning a different fuel makes the boiler a different class, type, or size. The Petitioners largely fail to develop this argument and, regardless, the EPA easily met its burden. The EPA based its decision on documented emissions data, several reports provided by the National Energy Technology Laboratory on boiler operations, and operating manuals provided by boiler manufacturers. See, e.g., Summary of Public Comments and Responses for National Emission Standards for Hazardous Air Pollutants for Major Sources: Industrial, Commercial, and Institutional Boilers and Process Heaters, EPA-HQ-OAR-2002-0058-3511A1 (Dec. 2012), at 558-63. These sources support the EPA’s decision to distinguish boilers based on the type of fuel they are designed to burn and the Agency’s conclusion that boilers designed for one fuel type are unlikely to use another fuel type. Id. The Petitioners present no contrary evidence, nor do they attack the validity or accuracy of the data that the EPA relied upon. We thus find no merit in the Petitioners’ various challenges to the EPA’s decision to subcategorize major boilers based on the fuel the boiler is designed to burn.
K. “Units that Begin Combusting Solid Waste” as “Existing” Sources
Section 7429(a)(2) distinguishes between “existing” and “new” CISWI units. The former must comply with floors set at the
The preamble to the 2011 CISWI Rule states, “[ujnits that begin combusting solid waste are considered
existing
sources.”
The Environmental Petitioners argue that the EPA’s broad statement in the 2011 CISWI Rule indicates that it imper-missibly changed its treatment of “modified sources” in contravention of the CAA. The EPA, however, agrees that any CI-SWI unit fitting the statutory criteria for a modified source must comply with new-unit MACT levels, not existing-unit MACT levels. See 42 U.S.C § 7429(g)(2). It also recognizes that its categorical statement in the 2011 CISWI Rule “may have been imprecise” and, in any event, it argues that the Environmental Petitioners have taken its statement out of context. See No. 11-1125 EPA Br. 73.
We agree with the Agency. The EPA’s later statement made clear that it intended to treat “sources that change fuels or materials” as “existing sources” unless “new source applicability,” as mandated by the Act, is “triggered.”
See
2011 Proposed CISWI Rule on Reconsideration,
L. Exclusion of “Temporary” Boilers from Area Boilers Rule
In the final 2013 Area Boilers Rule, the EPA excluded “temporary boilers” from regulation under section 7412.
See
(1) The equipment is attached to a foundation.
(2) The boiler or a replacement remains at a location within the facility and performs the same or similar function for more than 12 consecutive months, unless the regulatory agency approves an extension. An extension may be granted by the regulatory agency upon petition by the owner or operator of a unit specifying the basis for such a request. Any temporary boiler that replaces a temporary boiler at a location within the facility and performs the same or similar function will be included in calculating the consecutive time period unless there is a gap in operation of 12 months or more.
(3) The equipment is located at a seasonal facility and operates during the full annual operating period of the seasonal facility, remains at the facility for at least 2 years, and operates at that facility for at least 3 months of each year.
(4) The equipment is moved from one location to another within the facility but continues to perform the same or similar function and serve the same electricity, steam, and/or hot water system in an attempt to circumvent the residence time requirements of this definition.
2013 Area Boilers Rule,
Environmental Petitioners challenge this exclusion as a violation of the EPA’s obligations under the CAA to regulate all boilers listed under section 7412. By its own terms, the 2011 Area Boilers Rule “applies to all existing and new industrial boilers, institutional boilers, and commercial boilers located at area sources. Boiler means an enclosed combustion device having the primary purpose of recovering thermal energy in the form of steam or hot water.”
To the extent Environmental Petitioners challenge as unreasonable the EPA’s justifications for declining to set emission standards for temporary boilers, they cannot prevail. “Under arbitrary-and-capricious review, EPA’s determinations are presumptively valid provided [they] meet[] a minimum rationality standard.”
Nat’l Ass’n for Surface
Finishing,
Second, as both the EPA and Industry Intervenors note, the parallel rule for major source boilers has always explicitly excluded temporary boilers from its “industrial boiler” categorization. See 40 C.F.R. § 63.7491(j). The EPA thus considered commenters’ requests to add a similar clarification to the 2013 Area Boilers Rule and reasonably decided to do so. See, e.g., American Forest & Paper Association, Comments on Proposed Area Source Rule (AF & PA Comments), EPA-HQ-OARr 2006-0790-1939 (Aug. 23, 2010), at 58 (No. 11-1141 J.A. 389). EPA explained this choice in its proposed rule:
Owners and operators of regulated sources have pointed out that temporary boilers are small (less than 10 MMBtu/hr heat input) and are generally owned and operated by contractors, rather than the facility. As a result, they are not included in the facility’s operating permits because state and federal CAA operating permit programs have historically classified such units as insignificant sources. The owners and operators also noted that compliance with the •work practice requirements applicable to these small boilers would be complicated because they are typically located on site for less than a year, but would be subject to biennial management practice requirements. We agrée that the source category identified in subpart JJJJJJ should specifically exclude these temporary boilers because they have been considered insignificant sources, and were not included in the EPA’s analysis of the source category.
2011 Proposed Area Boilers Rule on Reconsideration,
The unique nature of temporary boilers favors their exclusion. These boilers tend to be rented for use on a temporary basis and come in “shop-fabricated package designs.” AF&PA Comments, at 58 (No. 11-1141 J.A. 389). Temporary boilers also “typically only fire gas or liquid fossil fuels (natural gas or distillate oil) which may be cleaner than the boiler(s) they are temporarily replacing. In addition, these units often do not have exhaust stacks that meet EPA Method 1 requirements for application of test methods.”
Id.
Regardless, during the rulemaking, Environmental Petitioners argued the EPA had “not explained why this is a distinction that justifies differential treatment, let alone an exemption.”
See
Area Boilers Rule — Responses to Comments, at 65. The EPA responded by explaining that rather than having “created a category or subcategory of ‘temporary boilers’ and then exempted them from the standards,” the Agency never “intend[ed] to regulate temporary boilers under the area source standards” in the first place.
See id.
The EPA further noted that, “[b]y their nature of being temporary, these boilers operate in place of another non-temporary boiler while that boiler is being constructed, replaced or repaired, in which case we counted the non-temporary boiler as the one being regulated.”
Id.
Finally, the Agency concluded regulation of temporary boilers was not necessary to meet its statutory emission requirements under sections 7412(c)(6) and 7412(c)(3) of the CAA.
The evidence before the Agency supported its decision to exclude temporary boilers. Indeed, the EPA “cogently explained]” why it exercised its discretion in this manner, such that this court concludes the Agency’s choice “was the product of reasoned decision making.”
U.S. Telecom Ass’n v. FCC,
M. Work-Practice Standards for Coal-Fired Boilers
When setting emission limits for area sources, the EPA enjoys greater discretion than when setting limits for major sources. With respect to major sources, the EPA has to promulgate MACT standards,
see
42 U.S.C. § 7412(d)(2), whereas for area-sources the EPA can generally promulgate more lenient GACT standards,
see id.
§ 7412(d)(5). The CAA, however, singles out seven particularly hazardous pollutants that require stricter regulatory standards, even for area sources.
41
Under section 7412(c)(6), the EPA must “list categories and subcategories of sources assuring that sources accounting for not less than 90 per centum of the aggregate emissions of each such pollutant” are regulated. The EPA listed a variety of area sources under section 7412(c)(6) in its 1998 rulemaking based on their Hg and POM emissions.
See
Proposed 2010 Area Boilers Rule,
Under section 7412(c)(6), the EPA was therefore required to set either a MACT limit under section 7412(d)(2), a health threshold under section 7412(d)(4), or a work-practice standard under section 7412(h) for all coal-fired boilers. The Agency chose to set MACT numerical emission limits for Hg and CO
42
at new and existing large coal-fired boilers. 2013 Area Boilers Rule,
Environmental Petitioners challenge the EPA’s decision to employ work-practice standards as a violation of 7412(d)(2)’s mandate to achieve the “maximum degree of reduction in emissions.” We examine Petitioner’s statutory argument step-by-step, as it hinges on the interplay between several statutory provisions. First, section 7412(c)(6) — which governs regulation of Hg and POM emissions — requires the Administrator to regulate sources of these pollutants under either section 7412(d)(2) or (d)(4). Section 7412(d)(4) allows the Administrator to establish health-based emission standards; it is not implicated here. Instead, the EPA decided to regulate coal-fired boilers under section 7412(d)(2). Section 7412(d)(2) instructs the Administrator to achieve “the maximum degree of reduction in emissions of the hazardous air pollutants ... that the Administrator, taking into consideration the cost of achieving such emissions reduction, and any non-air quality health and environmental impacts and energy requirements, determines is achievable for new or existing sources.” The Administrator is authorized to use several means to achieve this reduction including implementing “design, equipment, work practice, or operational standards ... as provided in [section 7412(h) ].” 42 U.S.C. § 7412(d)(2)(D). Section 7412(h)(1) states: “[I]f it is not feasible in the judgment of the Administrator to prescribe or enforce an emission standard for control of a [pollutant], the Administrator may, in lieu thereof, promulgate a ... work-practice standard ..., which in the Administrator’s judgment is consistent with the provisions of subsection (d) or (f) of this section.” Petitioners do not dispute the EPA’s ability to set work-practice standards here; they instead focus on section 7412(h)’s requirement that any such standards be “consistent with” subsection (d) — which requires the “maximum degree of reduction in emissions.” According to Petitioners, the EPA’s decision to set these particular work-practice standards fails at both Chevron steps.
With respect to Chevron step 1, Petitioners argue the “EPA does not claim the operational standards [for coal-fired boilers] are ‘consistent with the provisions of subsection (d) or (f)’ of § 7412.” No. 11-1141 Envtl. Pet’rs’ Br. 38. In other words, because these work-practice standards “do not even purport” to be consistent with section 7412(d)’s mandate to maximize reduction of emissions, “they are unlawful under Chevron step one.” Id. Petitioners point to the EPA’s specific findings to support this claim: for large coal-fired boilers, the EPA found that mercury emissions could be reduced by 75 to 82 per cent through the use of a fabric filter. Id. But, according to Petitioners, the “EPA admits the tuneup program [for small coal-fired boilers] will reduce emissions by only one percent.” Id. And, with respect to large coal-fired boilers undergoing startup or shutdown, Petitioners argue the “EPA does not claim that ‘following the manufacturer’s recommended procedures’ during startup and shutdown will reduce emissions at all.” Id.
In this case, Environmental Petitioners place too much emphasis on certain snippets of the statute without examining the larger context. For one, Petitioners seem to argue that the EPA must adopt work-practice standards that result in the maximum possible reduction of emissions, without taking into account any other considerations. But section 7412(d)(2) itself belies this claim: it says the EPA must promulgate standards that require “the maximum degree of reduction in emissions ... that the Administrator, taking into consideration the cost of achieving such emission reduction, and any non-air quality health and environmental impacts and energy requirements, determines is achievable.” 42 U.S.C. § 7412(d)(2) (emphasis added). This portion of the statute explicitly defers to the Administrator’s judgment regarding a standard’s “achievability,” even though it directs him to consider particular factors in making that assessment. Section 7412(h) similarly requires the Administrator to adopt a work-practice standard that in his judgment is consistent with section 7412(d)(2)’s mandate. We therefore cannot accept Petitioners’ suggestion that Congress unambiguously required the EPA to adopt standards that result in the maximum reduction of emissions that is technologically feasible.
Environmental Petitioners’ challenge to these work-practice standards as unreasonable under
Chevron
step 2 and arbitrary under
State Farm
presents a closer call. With respect to
Chevron
step 2, the court must “uphold an agency’s interpretation if it is reasonable.”
Ariz. Pub. Serv. Co. v. EPA,
Petitioners mount both a “facial” and a substantive challenge to the EPA’s rationale for adopting work-practice standards. First, Petitioners claim the EPA’s decision is arbitrary because it fails “to reconcile its approach with the statutory requirement [of section 7412(d)(2)].” No. 11-1141 Envtl. Pet’rs’ Br. 34. Specifically, Petitioners insist the EPA must explicitly state somewhere that these particular work-practice standards are “consistent” with section 7412(d)(2). See id. at 33-34. Otherwise, Petitioners contend, the Court must simply “assume that the Agency heeded § 7412(h)’s ‘consistent with’ requirement, notwithstanding the EPA’s failure to acknowledge and apply that requirement in the record.” No. 11-1141 Envtl. Pet’rs’ Reply Br. 10.
The Agency responds that, “by identifying the tune-up and startup/shutdown requirements as ‘work practices,’ [it] is stating that those standards are issued under section 7412(h) and are consistent with the requirements of. section 7412(d) (ie., MACT).” No. 11-1141, EPA Br. 71. The EPA did acknowledge it has an obligation to maximize emission reductions under
But the lack of an explicit statement does not automatically condemn this portion of the rule.
See Bowman,
1. Small Coal-Fired Boilers
First, with respect to small coal-fired boilers, the ERA determined that a biennial tune-up requirement would best comply with section 7412(h)’s requirements. As a starting point, the EPA surveyed a sample of state regulatiohs mandating various work-practice standards for small coal-fired boilers; the Agency found that ten states required tuneups, two required periodic inspections, one required operator training, and one required operation in accordance with manufacturer specifications. 2011 Area Boilers Rule,
The EPA elected to implement a work-practice standard because the typical method used to measure emissions of Hg and CO could not be used to sample emissions from stacks with small diameters (less than 12 inches).
See
2011 Area Boilers Rule,
Environmental Petitioners counter that while tune-ups may minimally reduce HAP emissions, they do not maximize this reduction per section 7412(d)(2)’s mandate.
45
But Petitioners view section 7412(d) too myopically; under that section, the Administrator is empowered to adopt standards that “tak[e] into consideration the cost of achieving such emission reduction.”' 42 U.S.C. § 7412(d)(2). The EPA here examined the costs of imposing a numeric emission standard on small coal-fired boilers and found that option “not feasible” due to high costs and monitoring difficulties, considerations equally permissible under section 7412(d)(2). Petitioners argue that'requiring the use of a fabric filter would have resulted in greater reductions, but they are unable to point to any small coal-fired boiler that currently uses such a filter.
See
2010 Proposed Area Boilers Rule,
Petitioners’ most compelling argument involves the EPA’s lack of data on small coal-fired boilers. As they point out, the EPA’s summary of its 2008 combustion survey makes no mention of any small coal-fired boilers.
See
MACT/GACT Mem., App. D-3, tbl.l (No. 11-1141 J.A. 523). And the EPA never directly addressed whether control technologies, such as fabric filters, were useable by small boilers; “[t]he only claim EPA made in the record is that tuneups are the most effective option that [small] coal-fired boilers ... are
currently
using, not that tune-ups yield the maximum reduction ‘achievable.’ ” No. 11-1141 Envtl. Pet’rs’ Reply Br. 11. In
Sierra Club II,
this court agreed with Sierra Club’s challenge to the EPA’s use of a work-practice standard instead of an emission floor because the “EPA never determined that measuring emissions from ceramics kilns was impracticable; it determined only that it lacked emissions data from ceramics kilns. EPA thus had no basis under section 7412(h) for using work practice standards.”
Ultimately, though, the high level of deference afforded the EPA counsels in favor of upholding this work-practice standard. Although the EPA did not explicitly state that tuneups were the best option to reduce emissions while still “considering costs,” this finding can be inferred from the record as a whole. For instance, the EPA found that “[n]one of the States for which we have an inventory have an applicable emissions limit” for small coal-fired boilers, except New Jersey, which actually has tune-ups as its work-practice standard. 2010 Proposed Area Boilers Rule,
2. Large Coal-Fired Boilers Undergoing Startup or Shutdown
The record for large coal-fired boilers undergoing startup or shutdown is less extensive but again the EPA’s determination is reasonable. While large coal-fired boilers are required to meet numeric emission standards during “normal” operations, the EPA adopted a work-practice standard for the temporary periods of startup and shutdown.
See
40 C.F.R. § 63.11214 (“[Mjinimize the boiler’s start-up and shutdown periods following the manufacturer’s recommended procedures, if available. If manufacturer’s recommended procedures are not available, you must follow recommended procedures for a unit of similar design for which manufacturer’s recommended procedures are available.”). Environmental Petitioners contend that the EPA never stated this practice would reduce emissions at all, and therefore it has not met its burden under section 7412(d)(2). But we have already explained that no
express
finding of consistency with section 7412(d)(2) need be made. Here, the record suggests that the work-practice standard the Agency chose would reduce emissions, and we therefore can “reasonably [] discern[]” the Agency’s path.
Bowman,
V. CONCLUSION
For the foregoing reasons, we grant the petitions in part and deny them in part. Specifically, we vacate the MACT standards for all major boiler subcategories that would have been affected had the EPA considered all sources included in the subcategories, as explained at supra § IV.B. We also remand, without vacatur, to the EPA to: (1) adequately explain how CO acts as a reasonable surrogate for non-dioxin/furan organic HAPs; (2) set emission standards for cyclonic burn barrels; (3) determine whether burn-off ovens, soil treatment units, and space heaters are CI-SWI units and, if so, to set standards for those types of units; (4) adequately explain the exclusion of synthetic boilers from Title V’s permitting requirements; and (5) adequately explain the choice of GACT standards over MACT standards for non-Hg metals.
So ordered.
Notes
. National Emission Standards for Hazardous Air Pollutants for Major Sources: Industrial, Commercial, and Institutional Boilers and Process Heaters (2011 Major Boilers Rule), 76 -Fed. Reg. 15,608 (Mar. 21, 2011), as amended, National Emission Standards for Hazardous Air Pollutants for Major Sources: Industrial, Commercial, and Institutional Boilers and Process Heaters (2013 Major Boilers Rule), 78 Fed. Reg. 7,138 (Jan. 31, 2013).
. National Emission Standards for Hazardous Air Pollutants for Area Sources: Industrial, Commercial, and Institutional Boilers (2011 Area Boilers Rule), 76 Fed. Reg. 15,554 (Mar. 21, 2011), as amended, National Emission Standards for Hazardous Air Pollutants for Area Sources: Industrial, Commercial, and ' Institutional Boilers (2013 Area Boilers Rule), 78 Fed. Reg. 7,488 (Feb. 1, 2013).
.Standards of Performance for New Stationary Sources and Emission Guidelines for Existing Sources: Commercial and Industrial Solid Waste Incinerator Units (2011 CISWI Rule), 76 Fed. Reg. 15,704 (Mar. 21, 2011), as amended, Commercial and Industrial Solid Waste Incineration Units: Reconsideration and Final Amendments; Non-Hazardous Secondary Materials that Are Solid Waste (2013 CISWI Rule), 78 Fed. Reg. 9,112 (Feb. 7, 2013).
. See, e.g., 40 C.F.R. pt. 60, subpts. Cd, Ce, Eb, AAAA, BBBB, EEEE, FFFF, LLLL, MMMM.
. The EPA must keep the HAPs list current. See 42 U.S.C. § 7412(b)(2), (3).
. The CAA defines “stationary source’’ as “any building, structure, facilily, or installation which emits or may emit any air pollutant.” 42 U.S.C. § 7411(a)(3).
. The four major boiler subcategories for which the EPA established work-practice standards include "[n]ew and existing units that have a designed heat input capacity of less than 10 MMBtu/hr, and new and existing units in the Gas 1 (natural gas/refinery gas) subcategory and in the metal process furnaces subcategoiy.” 2011 Major Boilers Rule,
. As discussed below, however, the EPA does allow sources to demonstrate MACT compli-anee by use of "continuous monitors.” See infra § IV.I.
. Specifically, the startup and shutdown work-practice standard requires a source to follow "the manufacturer's recommended procedures for minimizing periods of startup and shutdown.” 2011 Major Boilers Rule,
. As noted above, see supra § I.A.l.a, the EPA has some discretion in promulgating emissions limits for area HAP sources. Exercising its discretion, the EPA had previously determined that natural gas-fired area boilers did not emit HAPs at a level necessitating regulation. See National Emission Standards for Hazardous Air Pollutants for Area Sources: Industrial, Commercial, and Institutional Boilers (2010 Proposed Area Boilers Rule), 75 Fed. Reg. 31,896, 31,900 (June 4, 2010).
. As used in the Area Boilers Rule, the difference between “large” and "small” units depends on the heat-input capacity of the unit.
See
2013 Area Boilers Rule,
. These pollutants are (1) PM, (2) sulfur dioxide (S02), (3) HC1, (4) nitrogen oxide (NOx), (5) CO, (6) lead (Pb), (7) cadmium (Cd), (8) Hg, (9) dioxins and dibenzofurans, and (10) opacity (where appropriate). 42 U.S.C. § 7429(a)(4).
. In addition to challenging the assessment as a beyond-the-floor measure, the Industry Petitioners claim the assessment represents an invalid work-practice standard. But ''[t]he energy assessment is not ... a work practice standard, and EPA makes no claim that it is.” No. 11-1141 EPA Br. 47 n.9. Therefore, we decline to address that contention.
. The NHSM Rule defines non-hazardous secondary material to "mean[] a secondary material that, when discarded, would not be identified as a hazardous waste.” 40 C.F.R. § 241.2.
. In their reply brief, the Industry Petitioners clarify that they do not challenge the EPA’s authority to require sources to keep records.
.The Industry Petitioners also argue the EPA arbitrarily failed to provide sufficient notice of the recordkeeping presumption. We disagree. The Industry Petitioners had sufficient notice of the CISWI Rule, which was promulgated after notice and comment and "give[s] fair warning of the conduct it prohibits.”
Gen. Elec. Co. v. EPA,
. The Industry Petitioners also contend that the CISWI Rule functions as a form of injunc-tive relief in violation of 42 U.S.C. § 7413(a). That is incorrect. The provision is neither styled nor operated as a form of injunctive relief. Cf. 42 U.S.C. § 7413(a) (permitting the Administrator to issue, among other forms of relief, “an administrative penalty order” or "an order requiring [a person in violation of EPA regulations] to comply with such requirement or prohibition”).
. The EPA does have statutory authority under section 7412 to allow facility-wide emissions averaging in the Major Boilers Rule. See 42 U.S.C. § 7412(a)(1) ("major source[s]” defined as "any stationary source or group of stationary sources located within a contiguous area and under common control” (emphasis added)); see also id. § 7411(a)(3) (“stationary source” defined as "any building, structure, facility, or installation which emits or may emit any air pollutant”).
. The EPA concedes that it once allowed, in a different rule, emissions averaging for units subject to section 7429 but has since concluded that it does not have the statutory authority to do so. Although the Industry Petitioners argue that the Agency arbitrarily changed its position, the fact that the EPA may have acted outside its authority in a rule is not at issue here. “[P]revious statutory violations," of course, “cannot excuse” new ones.
New Jersey
v.
EPA,
., In its brief, the EPA argued that the Environmental Petitioners' challenge was moot either because the challenged MACT standards had been remanded for other reasons or because inclusion of the allegedly dissimilar sources would not have affected the MACT standard. During oral argument, however, it conceded that it misunderstood the scope of the Petitioners' argument, which argument challenges unremanded MACT standards that have in fact been affected by the EPA’s decision to omit certain high-performing sources from its MACT-floor analysis. See Oral Arg. Recording pt. B at 48:28-49:22. We believe that the Environmental Petitioners’ challenge is not moot and has not been waived.
. See also Page Mem. 2-3 ("This variability occurs due to a number of factors, including measurement variability (both sampling and analysis) and short term fluctuations in the emission levels that result from short-term changes in fuels, processes, combustion conditions, and controls.").
. See also Page Mem. 5 ("[E]ven single three run tests, which are performed over a short period of time, typically show different emissions levels during each individual test run.”).
.
See also NACWA,
.
See also NACWA,
.In
NACWA,
we had other problems with the EPA’s use of the UPL. Specifically, the EPA had explained that "a smaller dataset may have greater variability, and thus a higher [UPL].”
NACWA,
. The Environmental Petitioners urge us to ignore the Page Memorandum, insisting that it "provide[s] a series of new interpretations and assertions that, rather than ‘explaining’ the prior record, instead contradict and revise the agency’s earlier position,” in contravention of
NACWA
and the scope of the remand the Agency requested regarding the Major Boilers Rule and the CISWI Rule. No. 11-1108 Envtl. Pet'rs' Br. 41. But our
NACWA
decision did not, as the Petitioners would have it, require the EPA to adopt our belief that the Agency construed “average” to mean "the average of a future 3-run compliance test.”
See NACWA,
. The EPA used "one of several equations” to calculate the UPL depending on “certain characteristics of [the] dataset,” including the distribution of data within the dataset. Page Mem. 4. Here, we set out the equation the EPA used for a dataset with a "normal distribution.” Id. at 10. For our review, we need not recount the other, somewhat more complicated equations the EPA used in determining the UPL for datasets with, e.g., a "lognor-mal distribution.” See id. ("Even though they differ due to separate mathematical properties associated with each distribution, the UPL equations share a common format....”); see generally id. at 11 (describing lognormal distribution equation).
. See also No. 11-1108 Envtl. Pet’rs’ Reply Br. 15 ("It is, as its name indicates, an upper limit — the emissions limitation that every member of the best-performing 12 percent will fall below....” (emphasis in original) (quotation marks omitted)).
. The Environmental Petitioners’ argument rests, at least in part, on their contention that we should not consider the Page Memorandum at all. We decline their invitation to ignore the explanation we ordered the EPA to provide.
.See also Page Mem. 5 (It is “generally ... reasonable to establish a [MACT floor] standard that all the best performing 12 percent of existing sources can meet without any modification because the statute requires the Agency to establish the standard at the average level of performance of the best 12 percent of sources.” (emphasis in original)); id. at 14 (”[T]he MACT floor represents the average emission level achieved by the best performing sources, not the worst emission level achieved by those sources.” (emphases in original)).
. See also Page Mem. 6-7 (''[T]he UPL equation that is used to account for variability and [to] calculate the MACT floor standard depends on the distribution of the data.”); id. at 11 ("The UPL ... is directly related to the confidence level and to the variance, meaning that as either of these values go up or down, so does the UPL value.”).
. The EPA "selected the 99 percent level in order to provide reasonable assurance that the limit can be met at all times by a source with emissions at the average level achieved by the best performing source or sources.” Page Mem. 10. The Environmental Petitioners have not challenged the EPA’s choice of a 99 per cent confidence level, as opposed to a lower level of certainty, and we express no opinion on that choice. And we reiterate that the more specific concerns we had with the UPL when we decided
NACWA
— in particular, the UPL's accuracy “given a small dataset”— are not before us.
. See also Page Mem. 4 ("[T]he 99 percent UPL is the emissions level that the source would be predicted to be below 99 out of 100 performance tests, including emissions tests conducted in the past, present, and future.”); id. at 10 (“The confidence level, in this case 99 percent, is the percentage of measurements (past, present, and future) that are predicted to fall at or below the UPL value.”).
. Although the EPA does not argue that the court lacks jurisdiction to consider this argument, Environmental Petitioners raise the issue defensively, contending that they satisfied the CAA’s administrative exhaustion provision. We agree. During the rulemaking process, the Petitioners comprehensively critiqued the EPA’s proposed rationale for rejecting beyond-the-floor standards.
See, e.g.,
Natural Resources Defense Council, Comments on Standards of Performance for New Stationary Sources and Emission Guidelines for Existing Sources, EPA-HQ-OAR-2003-0119 (Aug. 23, 2010), at 11-16 (No. 11-1125 J.A. 668-73). Many of those comments challenged the EPA’s consideration of costs and other factors — the same types of issues Petitioners now ask the Panel to resolve. Because the Environmental Petitioners raised the relevant issues “with reasonable specificity” during the period for public comment, our jurisdiction is not in question. 42 U.S.C. § 7607(d)(7)(B);
see Portland Cement Ass’n v. EPA,
.EPA interprets the statutory factor of “cost” to permit consideration of cost-effectiveness,
NRDC III,
. This argument suffers from an additional flaw: the Environmental Petitioners appear to treat as interchangeable proposed emissions rules for new units with the final rules applicable to existing ones. That apples-and-oranges comparison underscores the weakness of the argument.
. The EPA asserts that it has not made a final decision with regard to the regulation of the five units at issue here — a claim that calls into question our jurisdiction, which under the CAA is limited to "final” actions.
See Portland Cement,
. A multiclone is a PM "mechanical separator!].”
See
2010 Proposed Area Boilers Rule,
. As the name suggests, a continuous monitoring system measures the source’s emissions at all times and generally takes one of two forms: (1) a continuous parameter monitor, which measures, e.g., a source's temperature, pressure or oxygen content; or (2) a continuous emissions monitor, which measures the pollutant concentration in the source’s emissions.
. The only exception is the EPA's 1996 Medical-Waste Incinerators Rule, which provides that '‘[t]he period of time over which emissions are measured and then averaged to determine compliance with the regulation ... must correspond to the period of time over which emission levels were measured and averaged in determining the emission limits in-eluded in the regulation.” Standards of Performance for New Stationary Sources and Emission Guidelines for Existing Sources: Medical Waste Incinerators, 61 Fed. Reg. 31,-736, 31,748 (June 20, 1996). This twenty-year-old statement, however, does not detract from the EPA’s well-reasoned defense of the thirty-day rolling average in the CISWI Rule.
. These seven pollutants are: (i) alkylated lead compounds, (ii) polycyclic organic matter (POM), (iii) hexachlorobenzene, (iv) mercury (Hg), (v) polychlorinated biphenyls, (vi) 2.3.7.8-tetrachlorodibenzofurans, and (vii) 2.3.7.8-tetrachlorodibenzo-p-dioxin. See 42 U.S.C. § 7412(c)(6).
. Because the EPA chose to regulate POM emissions indirectly — by using CO emissions as a surrogate — the standards it set under section 7412(c)(6) are for CO rather than POM.
See
2013 Area Boilers Rule,
.In their brief, Environmental Petitioners alternate between the terms "operational standards” and "work-practice standards,” both of which fall under section
. The EPA's discussion of boiler tune-up advantages occurs mainly in the context of its decision to select a GACT standard rather than any numeric emission standards for certain boilers. Environmental Petitioners challenge this decision on similar grounds, see supra § IV.H. But the benefits of periodic tune-ups also apply to the coal-fired boilers at issue here.
. Petitioners also point to a comment they made in the record arguing that a tune-up standard "would not achieve emission reductions that are consistent with the definition of MACT,” and urge that the EPA never addressed these concerns. See National Association of Clean Air Agencies, Comments on EPA Proposals for Regulation of Hazardous Air Pollutants (HAPS), EPA-HQ-OAR-2006-0790, EPA-HQ-OAR-2002-0058, EPA-HQ-OAR-2003-0119 (Aug. 23, 2010), at 21-22 (No. 11-1141 J.A. 417-18). But this comment was specifically addressed to gas-fired, boilers, and it is inapposite to the EPA’s consideration of standards for coal-fired boilers.