David G. Walsh, Foley & Lardner, Madison, Wis., for Wisconsin Utilities Ass'n, Inc., amicus curiae.

F. William Brownell, Norman W. Fichthorn, Hunton & Williams, Washington, D.C., for Alabama Power Co., amicus curiae.

Jeffrey S. Holik, Baker & Hostetler, Washington, D.C., for Aluminum Ass'n, Inc., amicus curiae.

Barton C. Green, Washington, D.C., for American Iron and Steel Institute, amicus curiae.

G. William Frick, Washington, D.C., for American Petroleum Institute, amicus curiae.

William E. Hynan, Washington, D.C., for National Coal Ass'n, amicus curiae.

Edward P. Giblin, Jr., Gaithersburg, Md., for Bechtel Power Corp., amicus curiae.

Paul J. Lambert, Bingham, Dana & Gould, Washington, D.C., for General Elec. Credit Corp., amicus curiae.

Carol L. Dalcanton, Pittsburgh, Pa., for Westinghouse Elec. Corp., amicus curiae.

Russell S. Frye, Chadbourne & Parke, Matthew B. VanHook, Washington, D.C., for American Paper Institute, Inc., amicus curiae.

Before CUDAHY and FLAUM, Circuit Judges, and GRANT, Senior District Judge.*

CUDAHY, Circuit Judge.

The Petitioner, Wisconsin Electric Power Company ("WEPCO"), challenges two final determinations issued by the Environmental Protection Agency (the "EPA"). In these determinations, the EPA concluded that WEPCO's proposed renovations to its Port Washington power plant would subject the plant to certain pollution control provisions of the Clean Air Act, as amended, 42 U.S.C. Secs. 7401 et seq. (1982). We affirm in part, vacate in part and remand to the EPA.

Some discussion of the Clean Air Act is required before turning to the merits of this case. In 1970, Congress enacted the Clean Air Act Amendments, Pub.L. No. 91-604, 84 Stat. 1676, to establish minimum air quality standards that would regulate the emission of certain pollutants into the atmosphere. To this end, Congress instructed the EPA to develop National Ambient Air Quality Standards ("NAAQS") that would specify the maximum permissible concentration of air pollutants in different areas across the country.

In section 111 of the 1970 Amendments, Congress required the EPA to promulgate New Source Performance Standards ("NSPS") in order to regulate the emission of air pollutants from new sources. These standards addressed hourly rates of emission and, in addition to new sources, applied to modifications of existing facilities that created new or increased pollution. Indeed, section 111(a)(2) of the Act stated that NSPS would apply to

any stationary source, the construction or modification of which is commenced after the publication of regulations (or, if earlier, proposed regulations) prescribing a standard of performance under this section which will be applicable to such source.

42 U.S.C. Sec. 7411(a)(2) (emphasis supplied). Congress then defined "modification" as

any physical change in, or change in the method of operation of, a stationary source which increases the amount of any air pollutant emitted by such source or which results in the emission of any air pollutant not previously emitted.

42 U.S.C. Sec. 7411(a)(4) (emphasis supplied).

Subsequently, faced with only varying degrees of success in controlling pollution in different parts of the country, Congress enacted the Clean Air Act Amendments of 1977, Pub.L. No. 95-95, 91 Stat. 685 (codified at 42 U.S.C. Secs. 7401-7642 (1982)). Congress revised the NSPS so that regulated sources of pollution would have to use "the best system of continuous emission reduction which (taking into consideration the costs of achieving such emission reduction, and any nonair quality health and environmental impact and energy requirements) the Administrator determines has been adequately demonstrated...." 42 U.S.C. Sec. 7411(a)(1)(C). In addition, Congress added a program for the Prevention of Significant Deterioration ("PSD"), concerned with increases in total annual emissions, to ensure that operators of regulated sources in relatively unpolluted areas would not allow a decline of air quality to the minimum level permitted by NAAQS. Air quality is preserved in this program by requiring sources to limit their emissions to a "baseline rate"; regulated owners or operators in areas that have attained NAAQS must obtain a permit before constructing or modifying facilities. 42 U.S.C. Sec. 7475(a)(1). Congress also essentially adopted its NSPS definition of "modification" for the PSD program. 42 U.S.C. Sec. 7479(2)(C).

From this statutory framework, the EPA promulgated regulations for both the NSPS and PSD programs. In this case, its regulations concerning modifications are central. The EPA defines "modification" in substantially the same terms used by Congress:

[A]ny physical or operational change to an existing facility which results in an increase in the emission rate to the atmosphere of any pollutant to which a standard applies shall be considered a modification within the meaning of section 111 [42 U.S.C. Sec. 7411] of the Act.

40 C.F.R. Sec. 60.14(a) (1988). To determine whether a physical change constitutes a modification for purposes of NSPS, the EPA must determine whether the change increases the facility's hourly rate of emission. 40 C.F.R. Sec. 60.14 (1988). For PSD purposes, current EPA regulations provide that an increase in the total amount of emissions activates the modification provisions of the regulations. 40 C.F.R. Sec. 52.21(b)(3) (1988).

Even at first blush, the potential reach of these modification provisions is apparent: the most trivial activities--the replacement of leaky pipes, for example--may trigger the modification provisions if the change results in an increase in the emissions of a facility. As a result, the EPA promulgated specific exceptions to the modification provisions:

The following shall not, by themselves, be considered modifications under this part:

(1) Maintenance, repair, and replacement which the Administrator determines to be routine for a source category ...

(2) An increase in production rate of an existing facility, if that increase can be accomplished without a capital expenditure on that facility.

(3) An increase in the hours of operation....

40 C.F.R. Sec. 60.14(e) (1988) (NSPS program); see 40 C.F.R. Sec. 52.21(b)(2)(iii) (1988) (PSD program). These regulations (and the statutes from which they derive) are the focal point of this case.

WEPCO's Port Washington electric power plant is located on Lake Michigan north of Milwaukee, Wisconsin. The plant consists of five coal-fired steam generating units that were placed in operation between 1935 and 1950. Each generating unit has a design capacity of 80 megawatts, but the recent performance of some of the units has declined due to age-related deterioration of the physical plant.

WEPCO and its consultant, Bechtel Eastern Power Corporation, conducted a Plant Availability Study in 1983 to examine and assess the condition of the power plant. As a result of the Study, WEPCO concluded "that extensive renovation of the five units and the plant common facilities is needed if operation of the plant is to be continued." Letter from Thomas J. Cassidy, Executive Vice President of WEPCO, to Jacqueline K. Reynolds, Secretary to the Public Service Commission of Wisconsin, at 2 (July 8, 1987) [Cassidy Letter] (emphasis supplied). The Study noted that the air heaters on the first four units had deteriorated severely, while the rear steam drums in units 2 through 5 had experienced serious cracking.1 Air heater deterioration prevented units 1 and 4 from operating at full capacity, while the potential for steam drum blowout required a reduction in pressure (and output) in units 2 and 3. The possibility of catastrophic failure (steam drum blowout) in unit 5 was so great that WEPCO shut down the unit completely.

As a result of this Study, WEPCO submitted a proposed replacement program (which it termed a "life extension" project) to the Wisconsin Public Service Commission for its approval, as required by state law. Wis.Stat. Sec. 196.49 (1987). WEPCO explained in its proposal that "[r]enovation is necessary to allow the Port Washington units to operate beyond their currently planned retirement dates of 1992 (units 1 and 2) and 1999 (units 3, 4 and 5) ... [and that renovation would render the plant] capable of generating at its designed capability until year 2010...." Cassidy Letter at 1-2. Among the renovations required were repair and replacement of the turbine-generators, boilers, mechanical and electrical auxiliaries and the common plant support facilities. Id. at 1. After preliminary review of the program, the Public Service Commission consulted the Wisconsin Department of Natural Resources (which then consulted EPA Region V) to determine whether WEPCO needed to obtain a PSD permit before commencing the repair and replacement program. David Kee, the Director of EPA Region V's Air and Radiation Division, then referred the matter to EPA Headquarters. See, e.g., 40 C.F.R. Sec. 60.5 (1988) (discussing the EPA's procedures regarding determinations of construction or modification).

EPA staff members conferred with WEPCO representatives between March and September 1988 to gain additional information regarding the proposed repair and replacement project. On September 9, 1988, EPA Acting Assistant Administrator Don R. Clay issued a memorandum in which he preliminarily concluded that the project would subject the plant to both NSPS and PSD requirements. Memorandum from Don R. Clay, Acting Assistant Administrator for Air and Radiation of the EPA, to David A. Kee, Director of Air and Radiation Division, Region V (Sept. 9, 1988) [Clay Memorandum]. The Clay Memorandum pointed out that the project would constitute a "physical change" resulting in an increase of production and emissions, which would therefore subject the plant to the relevant strictures of the Clean Air Act. Id. at 3-4. Further, the Clay Memorandum dismissed WEPCO's contention that the program was routine and was therefore exempt from the requirements of NSPS and PSD. This conclusion was adopted in toto by EPA Administrator Lee M. Thomas. Letter from Lee M. Thomas, Administrator of the EPA, to John Boston, Vice President of WEPCO (Oct. 14, 1988) [Thomas Letter].

Following the Thomas Letter, WEPCO continued to conduct capacity tests on the units. Based upon these tests, Assistant Administrator Clay issued a "revised final determination" that generally affirmed the EPA's earlier findings, but modified the baseline figures used by the EPA for units 2 and 3. Letter from Don R. Clay, Acting Assistant Administrator for Air and Radiation of the EPA, to John W. Boston, Vice President of WEPCO (Feb. 15, 1989) [Supplemental Determination].

Alleging that the EPA has misconstrued both the Clean Air Act and its own regulations, WEPCO appeals the EPA's final determination. We have jurisdiction to hear this appeal pursuant to 42 U.S.C. Sec. 7607(b) (1982).

Courts have generally accorded substantial deference to the EPA's interpretation of the Clean Air Act Amendments, reasoning that "considerable weight should be accorded to an executive department's construction of a statutory scheme it is entrusted to administer...." Chevron U.S.A. Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 844, 104 S. Ct. 2778, 2782, 81 L. Ed. 2d 694 (1984); see Union Elec. Co. v. EPA, 427 U.S. 246, 256, 96 S. Ct. 2518, 2525, 49 L. Ed. 2d 474 (1976); Train v. Natural Resources Defense Council, Inc., 421 U.S. 60, 75, 87, 95 S. Ct. 1470, 1479, 1485, 43 L. Ed. 2d 731 (1975); ASARCO Inc. v. EPA, 578 F.2d 319, 325 (D.C.Cir.1978). This deference with regard to the Clean Air Act follows logically from the highly technical provisions of the Amendments, Chevron, 467 U.S. at 848, 104 S.Ct. at 2784, and is consistent with the Administrative Procedure Act, which provides that agency actions are to be set aside only if they are "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law." 5 U.S.C. Sec. 706(2).

To be sure, this standard does not give the EPA unbridled discretion to construe the Clean Air Act Amendments free from judicial oversight. We must consider whether the EPA's construction comports with its statutory mandate and Congress's intent in enacting clean air legislation. But we cannot simply substitute our judgment for that of the EPA. Our role has been sharply defined and limited by the Supreme Court:

When a court reviews an agency's construction of the statute which it administers, it is confronted with two questions. First, always, is the question whether Congress has directly spoken to the precise question at issue. If the intent of Congress is clear, that is the end of the matter; for the court, as well as the agency, must give effect to the unambiguously expressed intent of Congress. If, however, the court determines Congress has not directly addressed the precise question at issue, the court does not simply impose its own construction on the statute, as would be necessary in the absence of an administrative interpretation. Rather, if the statute is silent or ambiguous with respect to the specific issue, the question for the court is whether the agency's answer is based on a permissible construction of the statute.

Chevron, 467 U.S. at 842-43, 104 S.Ct. at 2781-82 (footnotes omitted).

Further, we defer even more to an agency's construction of its own regulations. Lyng v. Payne, 476 U.S. 926, 939, 106 S. Ct. 2333, 2341, 90 L. Ed. 2d 921 (1986); see Wilkins v. Sullivan, 889 F.2d 135, 139 (7th Cir.1989); Homemakers North Shore, Inc. v. Bowen, 832 F.2d 408, 411 (7th Cir.1987) (agency construction of its regulations usually upheld). An agency's interpretation must be upheld " 'unless it is plainly erroneous or inconsistent with the regulation.' " Udall v. Tallman, 380 U.S. 1, 16-17, 85 S. Ct. 792, 801-802, 13 L. Ed. 2d 616 (1965) (quoting, in part, Bowles v. Seminole Rock & Sand Co., 325 U.S. 410, 413-14, 65 S. Ct. 1215, 1217, 89 L. Ed. 1700 (1945)). The principle of deference has particular force where, as is the case here, the subject being regulated is technical and complex. Aluminum Co. of Am. v. Central Lincoln Peoples' Util. Dist., 467 U.S. 380, 390, 104 S. Ct. 2472, 2479, 81 L. Ed. 2d 301 (1984); Wilkins v. Sullivan, 889 F.2d at 140; see also Skidmore v. Swift & Co., 323 U.S. 134, 140, 65 S. Ct. 161, 164, 89 L. Ed. 124 (1944) (rulings of agency constitute bodies of experience and informed judgment).

III. Like-Kind Replacement and Modification under the Act

With these principles in mind, we may address the merits. We must first consider whether WEPCO's Port Washington replacement program constitutes a modification under the terms of the controlling statute, 42 U.S.C. section 7411(a)(4). Cf. Blue Chip Stamps v. Manor Drug Stores, 421 U.S. 723, 756, 95 S. Ct. 1917, 1935, 44 L. Ed. 2d 539 (1975) (Powell, J., concurring) ("The starting point in every case involving construction of a statute is the language itself."). Section 7411(a)(4) defines modification as "any physical change ... which increases the amount of any air pollutant emitted...." 42 U.S.C. Sec. 7411(a)(4). Both parts of this definition--any physical change and an increase in emissions--must be satisfied before a replacement will be considered a "modification."

Certainly, under the plain terms of the Act, WEPCO's replacement program constitutes a "physical change." WEPCO proposes to replace rear steam drums on units 2, 3, 4 and 5; each of these steam drums measures 60 feet in length, 50.5 inches in diameter and 5.25 inches in thickness. Clay Memorandum at 4. In addition, WEPCO plans to replace another major component, the air heaters, in units 1-4. To implement this four-year program, WEPCO will need to make the replacements by taking the units successively out of service for nine-month periods. Id. These steps clearly amount to a "physical change" in the Port Washington plant. See Butler, New Source Netting in Nonattainment Areas under the Clean Air Act, 11 Ecology L.Q. 343, 349-50 (1984) ("[T]he new source review requirements are triggered not only when an operator builds a new plant, but also whenever the operator installs or alters a piece of equipment in an existing plant and thereby increases emissions.") (emphasis supplied).

WEPCO does not dispute that its steam drum and air heater replacements will result in an altered plant. But WEPCO does assert that Congress did not intend for simple equipment replacement to constitute a physical change for purposes of the Clean Air Act's modification provisions:

The plain meaning of "modify" is "to change or alter" [Webster's New World Dictionary] or "to make basic or fundamental changes in." [Webster's Ninth New Collegiate Dictionary] Reflecting the plain meaning of this term, Congress provided that a facility (1) must undergo a physical or operational "change" before it is evaluated under the modification provision.... Thus, under the plain meaning of the Act, a unit should not be deemed "modified" as a result of replacement of equipment with equipment similar to that replaced. As in the case of Port Washington, such like-kind replacement does not "change or alter" the design or nature of the facility. Rather, it merely allows the facility to operate again as it had before the specific equipment deteriorated.

Petitioner's Brief at 32-33.

Chevron instructs us to rely more on congressional direction and on agency construction (pursuant to congressional delegation) than on glosses found in the dictionary. What WEPCO calls "plain" is anything but plain and takes the definition far beyond the words enacted by Congress. Chevron, 467 U.S. at 843-45, 104 S.Ct. at 2781-83; see generally, R. Anthony, Report to the Administrative Conference of the United States: Which Agency Interpretations Should Bind the Courts and the Public? (1989) (explaining Chevron approach). Thus, whether the replacement of air heaters and steam drums is a "basic or fundamental change" in the Port Washington plant is irrelevant for our purposes, given Congress's directions on the subject: "The term 'modification' means any physical change...." 42 U.S.C. Sec. 7411(a)(4) (emphasis supplied). We follow Congress's definition of "modification"--not Webster's--when interpreting this term within the context of the Clean Air Act. Cf. Chevron, 467 U.S. at 861, 104 S.Ct. at 2790 ("[T]he meaning of a word must be ascertained in the context of achieving particular objectives....").

Nor can we find any support in the relevant case law for the narrow constructions of "modification" and "physical change" offered by WEPCO. The Supreme Court reported in Chevron that Senator Muskie, one of the principal supporters of the Clean Air Act, remarked: "A source ... is subject to all the nonattainment requirements as a modified source if it makes any physical change which increases the amount of any air pollutant...." 467 U.S. at 853, 104 S.Ct. at 2787 (quoting 123 Cong.Rec. 26847 (1977)) (emphasis supplied). And other courts considering the modification provisions of NSPS and PSD have assumed that "any physical change" means precisely that. See, e.g., National-Southwire Aluminum Co. v. EPA, 838 F.2d 835 (6th Cir.), cert. denied, --- U.S. ----, 109 S. Ct. 390, 102 L. Ed. 2d 379 (1988) (turning off pollution control equipment constitutes "physical change" and modification); Alabama Power Co. v. Costle, 636 F.2d 323, 400 (D.C.Cir.1979) ("[T]he term 'modification' is nowhere limited to physical changes exceeding a certain magnitude."); ASARCO Inc. v. EPA, 578 F.2d 319, 322 (D.C.Cir.1978) (NSPS applies to any stationary source that is "physically or operationally changed in such a way that its emission of any air pollutant increases.") (emphasis removed). Cf. United States v. Narragansett Improvement Co., 571 F. Supp. 688, 694-95 (D.R.I.1983) (replacement program not modification because, despite physical change, no increase in emissions).

Further, to adopt WEPCO's definition of "physical change" would open vistas of indefinite immunity from the provisions of NSPS and PSD. Were we to hold that the replacement of major generating station systems--including steam drums and air heaters--does not constitute a physical change (and is therefore not a modification), the application of NSPS and PSD to important facilities might be postponed into the indefinite future. There is no reason to believe that such a result was intended by Congress. The Clean Air Act Amendments were enacted to "speed up, expand, and intensify the war against air pollution in the United States with a view to assuring that the air we breathe throughout the Nation is wholesome once again." H.R.Rep. No. 91-1146, 91st Cong., 2d Sess. 1, 1, reprinted in 1970 U.S.Code Cong. & Admin.News 5356, 5356. In particular, the permit program established by the 1977 Amendments to the Clean Air Act represented a balance between "the economic interests in permitting capital improvements to continue and the environmental interest in improving air quality." Chevron, 467 U.S. at 851, 104 S.Ct. at 2786. The House echoed this theme in its Committee report: "[The compliance program is designed, in part,] to allow reasonable economic growth to continue in an area while making reasonable further progress to assure attainment of the [pollution-control] standards by a fixed date...." H.R.Rep. No. 294, 95th Cong., 1st Sess. 211, reprinted in 1977 U.S.Code Cong. & Admin.News 1077, 1290. A too restrictive interpretation of "modification" might upset the economic-environmental balance in unintended ways.

Consistent with its balanced approach, Congress chose not to subject existing plants to the requirements of NSPS and PSD. Members of the House recognized that "[b]uilding control technology into new plants at time of construction will plainly be less costly then [sic] requiring retrofit when pollution control ceilings are reached." H.R.Rep. No. 294, 95th Cong., 1st Sess. 185, reprinted in 1977 U.S.Code Cong. & Admin.News at 1264. But Congress did not permanently exempt existing plants from these requirements; section 7411(a)(2) provides that existing plants that have been modified are subject to the Clean Air Act programs at issue here. As Judge Boggs, dissenting in National-Southwire, reasoned: "The purpose of the 'modification' rule is to ensure that pollution control measures are undertaken when they can be most effective, at the time of new or modified construction. See 116 Cong.Rec. 32,918. (remarks of Sen. Cooper), reprinted in 1 Senate Committee on Public Works, A Legislative History of the Clean Air Act Amendments of 1970 (1974), at 260." National-Southwire Aluminum Co. v. EPA, 838 F.2d 835, 843 (6th Cir.) (Boggs, J., dissenting), cert. denied, --- U.S. ----, 109 S. Ct. 390, 102 L. Ed. 2d 379 (1988). Judge Boggs argued that the shutting down of pollution control equipment in an existing plant should not be considered a modification because it would not afford the utility an opportunity for "effective placement of new control technology." Id. Here the record is silent on this point (although the point is important). How easy or difficult would be "the effective placement of new control technology" in these renovated units is not clear, but we do know that the project already contemplates replacement of steam drums, air heaters and other components; each unit would, therefore, in any event be shut down for nine months.

Our reading of the phrase "any physical change" is also consistent with another of the basic goals of the 1977 Amendments: technology-forcing. The legislative history suggests and courts have recognized that in passing the Clean Air Act Amendments, Congress intended to stimulate the advancement of pollution control technology. See, e.g., S.Rep. No. 91-1196, 91st Cong., 2d Sess. 17 (1970) ("Standards of performance should provide an incentive for industries to work toward constant improvement in techniques for preventing and controlling emissions from stationary sources...."); Duquesne Light Co. v. EPA, 698 F.2d 456, 475 (D.C.Cir.1983); Alabama Power, 636 F.2d at 372; ASARCO, 578 F.2d at 327; United States v. SCM Corp., 667 F. Supp. 1110, 1126-27 (D.Md.1987). The development of emissions control systems is not furthered if operators could, without exposure to the standards of the 1977 Amendments, increase production (and pollution) through the extensive replacement of deteriorated generating systems.

The controversy involving WEPCO's alleged increase in emissions primarily concerns the regulations, not the statute: WEPCO argues that the EPA's regulatory method of measuring emissions is arbitrary and capricious. From a statutory standpoint, however, the modification provisions of the Clean Air Act Amendments are activated once a physical change is coupled with an "increase[ ] [in] the amount of any air pollutant emitted." 42 U.S.C. Sec. 7411(a)(4). See, e.g., United States v. Narragansett Improvement Co., 571 F. Supp. at 694. In the case before us, WEPCO does not dispute that its replacement program--intended to enable its deteriorated generators to operate at full capacity--will cause its emissions to increase from their current operating levels. The question for resolution, however, is whether the EPA properly construed its regulations by comparing actual emission rates with so-called "baseline" rates to determine the increase in emissions for NSPS and PSD purposes. We will discuss this subject later; but for purposes of the statutory requirement, we simply observe that the rejuvenated Port Washington plant will produce more emissions after the completion of the renovation project than the operating deteriorated plant produced shortly before the project was undertaken.

Although we have determined that WEPCO's repair and replacement program satisfies the modification provisions of the Clean Air Act Amendments, this is not the end of our inquiry. WEPCO's attack focuses primarily on EPA regulations, which in a number of respects are narrower than the statute. WEPCO argues that the EPA applied its regulations arbitrarily and capriciously to the Port Washington project.

1. Physical Change and the "Routine" Exception

EPA regulations define "modification" as "any physical or operational change to an existing facility which results in an increase in the emission rate to the atmosphere of any pollutant to which a standard applies." 40 C.F.R. Sec. 60.14(a) (1988). To a major degree, this definition parallels 42 U.S.C. section 7411(a)(2), and it is unnecessary to repeat the analysis already applied to the statute. See supra III(A)(1). However, the EPA has, in addition, used its regulations to exempt a number of activities from the broader definition. The exemption that may be relevant here is accomplished by the following language:

The following shall not, by themselves, be considered modifications under this part:

(1) Maintenance, repair, and replacement which the Administrator determines to be routine for a source category....

40 C.F.R. Sec. 60.14(e) (1988). See 40 C.F.R. Sec. 52.21(b)(2)(iii). WEPCO relies on this language to argue that, even if its repair and replacement program amounts to a physical change, it was specifically exempted by the regulations.

Again, we accord substantial deference to an agency's interpretation of its own regulations, especially with respect to technical and complex matters. Lyng v. Payne, 476 U.S. 926, 939, 106 S. Ct. 2333, 2341, 90 L. Ed. 2d 921 (1986); Aluminum Co. of Am. v. Central Lincoln Peoples' Util. Dist., 467 U.S. 380, 390, 104 S. Ct. 2472, 2479, 81 L. Ed. 2d 301 (1984). In this connection, to determine whether proposed work at a facility is routine, "EPA makes a case-by-case determination by weighing the nature, extent, purpose, frequency, and cost of the work, as well as other relevant factors, to arrive at a common-sense finding." Clay Memorandum at 3. The EPA considered all these factors in determining that the Port Washington project was not routine; first, the EPA observed that the nature and extent of the project was substantial: WEPCO proposed to replace sixty-foot steam drums (in units 2, 3, 4 and 5) and air heaters (in units 1, 2, 3 and 4) during successive nine-month outages at each unit. Id. at 4. Certainly, the magnitude of the project (as well as the down-time required to implement it) suggests that it is more than routine.

Further, the EPA points to WEPCO's admission in its application that "[work items] falling into the category of repetitive maintenance that are normally performed during scheduled equipment outages ... are not included in this application." Cassidy Letter at 1 (emphasis supplied). This admission suggests that WEPCO at first blush did not regard the repair and replacement project as ordinary or routine.

In addition, the EPA noted that far from being routine, the Port Washington project apparently was unprecedented: "WEPCO did not identify, and EPA did not find, even a single instance of renovation work at any electric utility generating station that approached the Port Washington life extension project in nature, scope or extent." Respondent's Brief at 44; see Clay Memorandum at 4 ("[T]his is a highly unusual, if not unprecedented, and costly project."). We surmise, although the record is silent, that the "case of first impression" character of the project may reflect historical practice in the electric utility industry of replacing old plants (at the expiration of their useful lives) with new plants, employing improved technologies and achieving improved efficiencies. This was the typical practice, rather than the mere extension of life of existing plants through massive like-kind replacements. Cf. Clay Memorandum at 4 ("[The Port Washington project's] purpose is to completely rehabilitate aging power generating units whose capacity has significantly deteriorated over a period of years, thereby restoring their original capacity and substantially extending the period of their utilization as an alternative to retiring them as they approach the end of their useful physical and economic life.").

WEPCO asks us to overlook the factors outlined in the Clay Memorandum and reverse the EPA primarily on the basis of earlier EPA decisions characterizing certain replacement programs as routine; WEPCO argues that the nature and extent of these "routine" projects parallel those of its Port Washington project. For example, WEPCO presented the EPA with a list of forty air heaters in other plants that had been replaced without triggering NSPS or PSD provisions. Letter from Mark P. Steinberg, Superintendent--Air Quality of WEPCO, to Dale Ziege, Wisconsin Department of Natural Resources (Jan. 11, 1989). But as WEPCO has acknowledged, the plate-type air heaters at issue in the Port Washington project must be replaced in whole; in contrast, the forty units where replacement was apparently considered routine contained a Ljungstrom basket or tubular type heater. The Ljungstrom basket type, at least, permits the replacement of the heat transfer surface without requiring the removal of the entire unit. Supplemental Determination at 6-7. Obviously, the precise nature of the physical change is a material factor in determining whether the change is routine, and for this purpose it is important that the subject of past EPA practice be closely comparable with the change under consideration here. See Thomas Letter at 3 ("PSD and NSPS applicability determinations are made on a case-by-case basis."). WEPCO has not demonstrated that the EPA's conclusion that the forty other air heater replacements were dissimilar is arbitrary and capricious.2

The purpose, frequency and cost of the work also support the EPA's decision here. WEPCO admits that the plans for extensive renovation "represent a life extension of the units from their planned retirement dates," Cassidy Letter at 2-3 (emphasis supplied), and it recognizes that "the renovation work items included in this application are those that would normally occur only once or twice during a unit's expected life cycle." Id. at 1. Indeed, WEPCO reported that it had never previously replaced a steam drum or "header" of comparable size at any of its coal-fired electrical generating facilities. Clay Memorandum at 5. Further, the Port Washington renovation project will cost at least $70.5 million. Letter from John W. Boston, Senior Vice President of WEPCO, to Gary D. McCutchen, Chief New Source Review Section of the EPA, at 4 (May 19, 1988). These factors suggest that the project is not routine.

WEPCO urges that the EPA's conclusions are supported by neither the evidence nor the provisions of the Clean Air Act Amendments. WEPCO reasons that because any replacement project will presumably extend the life of a facility, the EPA's reliance on life extension as a factor in denying the "routine" nature of a project is overbroad. Petitioners' Brief at 44. Although perhaps persuasive on its face, WEPCO's analysis is ultimately wide of the mark. While it is certainly true that the repair of deteriorated equipment will contribute to the useful life of any facility, it does not necessarily follow that the repairs in question would extend the life expectancy of the facility. The need for some repairs along the line is a given in determining in the first instance the life expectancy of a plant. WEPCO cannot seriously argue that its units' planned retirement dates of 1992 (units 1 and 2) and 1999 (units 3, 4 and 5) did not take into account at least minor equipment repairs and replacements.3 And WEPCO concedes that the Port Washington program will extend the life expectancy of the plant until 2010. The EPA concluded that the proposed project will increase the life expectancy of the Port Washington facility, and this conclusion was a factor in the finding that the work was not routine. These determinations were not arbitrary and capricious.

Still, WEPCO asserts that the cost, magnitude and nature of its Port Washington project are irrelevant for purposes of the "routine" exception to NSPS and PSD. WEPCO contends that the EPA has already addressed these factors--including the perpetuation of existing sources--through its so-called "reconstruction" rule:

(a) An existing facility, upon reconstruction, becomes an affected facility [subject to NSPS], irrespective of any change in emission rate.

(b) "Reconstruction" means the replacement of components of an existing facility to such an extent that:

(1) The fixed capital cost of the new components exceeds 50 percent of the fixed capital cost that would be required to construct a comparable entirely new facility....

40 C.F.R. Sec. 60.15 (1988).4 See United States Environmental Protection Agency, Electric Utility Steam Generating Units: Background Information for Proposed Particulate Matter Emission Standards 5-7 (1978). WEPCO believes that, because the air heater replacements will presumably cost less than six percent of a wholly new facility, the reconstruction provisions are not triggered. Therefore, WEPCO argues that the cost and scope of the project are relevant only to a "reconstruction" analysis and are not material for purposes of the routine exception to the modification provisions. Petitioner's Brief at 46.

WEPCO's analysis fails to note, however, the fundamental differences distinguishing the reconstruction and modification provisions. The reconstruction provision applies to any substantial replacement (more than 50% of the cost of a new facility) even if the replacement causes no subsequent increase in emissions. In sharp contrast, the modification provisions apply only when a physical change is accompanied by an increase in emissions. To argue, therefore, that the reconstruction provision is the exclusive determinant of whether the cost, nature and magnitude of a project will require the application of NSPS is to ignore the substantially different objectives of the reconstruction and modification provisions: The reconstruction provision is aimed principally at "discourag[ing] the perpetuation of a facility, instead of replacing it at the end of its useful life with a newly constructed affected facility," without regard to emissions, 39 Fed.Reg. 36946, 36948 (1974), while the modification provision applies to any physical change, without regard to cost, that causes an increase in emissions. See, e.g., ASARCO Inc. v. EPA, 578 F.2d 319 (D.C.Cir.1978); United States v. Narragansett Improvement Co., 571 F. Supp. 688, 695 (D.R.I.1983) ("a 'reconstruction' of an existing facility would occur 'irrespective of any change in emission rate' upon the replacement of a 'substantial portion of the existing facility's components.' "). Hence, we cannot agree that the EPA's consideration of the cost, magnitude and nature of the Port Washington project, for purposes of the modification provision of the regulations (and its "routine" exception), is somehow "preempted" by the reconstruction provisions of the regulations. The EPA's examination of these factors, therefore, was not arbitrary or capricious.

Thus far, we have not had to address the important differences between the PSD and NSPS programs. At this point, however, the differences become crucial, because each program measures emissions in a fundamentally distinct manner.

a. NSPS Measurements

As previously noted, the EPA's NSPS program is concerned primarily with increases in emission rates, expressed in kilograms per hour of discharged pollutants. 40 C.F.R. Sec. 60.14 (1988). The EPA compares the hourly emissions of the unit at its current maximum capacity to its potential emissions at maximum capacity after the change. Clay Memorandum at 9; see 40 C.F.R. Sec. 60 Ohio App. C (1988) (providing complex formulae for determining emission rate change). In this calculation, the agency disregards the unit's maximum design capacity; this factor often sheds little light on the unit's actual current capacity to produce emissions.5

The EPA applied these procedures in examining the generating units at Port Washington. The EPA asked WEPCO to submit figures for the actual operations and emissions of each unit at the Port Washington plant for the years 1978 to 1987; the EPA then relied upon the 1987 figures to calculate the emissions baseline against which post-replacement emissions could be compared. WEPCO, however, challenged the EPA's acceptance of these preliminary baseline figures, arguing that units 1, 2, 3 and 4 were capable of operating at higher rates of production than those calculated by the EPA based upon the 1987 figures. WEPCO conducted five ten-hour tests at each unit to determine its maximum capacity. Upon reviewing the test results, the EPA agreed that units 2 and 3 could be operated at their design capacities, and it revised the baseline levels for these units. The agency concluded that because there would be no increase in production or emissions, NSPS would not apply to these units following the renovation project. Nonetheless, the EPA refused to alter the baseline levels for units 1 and 4, noting that WEPCO's tests had not been conducted pursuant to the test protocol as required by the regulations and the Wisconsin State Implementation Plan (units 1 and 4 exceeded certain maximum allowable emission limits). Supplemental Determination at 8-9. Comparing these 1987 baseline levels to the maximum capacity of the plant after renovation, the EPA concluded that the renovation project would be subject to the provisions of NSPS.

WEPCO asks us to overturn the EPA's final ruling that the Port Washington project triggers NSPS. Specifically, WEPCO argues that, by using 1987 figures in determining the emissions baseline, the EPA failed to apply its own regulations: WEPCO asserts that these figures "reflected voluntary decisions by WEPC[O] regarding safety considerations (e.g., the 'zero' rate for Unit 5) and an electricity demand which did not require operation of the units at higher capacities." Petitioner's Brief at 15-16. WEPCO also posits that the EPA's refusal to compare representative pre-renovation emissions with actual post-renovation emissions is contrary to EPA regulations and amounts to an abuse of agency discretion.6

WEPCO's first assertion is easily dismissed. The EPA's choice of the 1987 figures was based entirely upon WEPCO's own data. And, when WEPCO complained that its own data did not reflect WEPCO's pre-renovation capabilities, the EPA permitted WEPCO to conduct new tests (pursuant to 40 C.F.R. Sec. 60 Ohio App. C (1988)) that eventually resulted in the revision of the baselines for units 2 and 3.

WEPCO's second charge is far more substantial. WEPCO argues that NSPS regulations require the EPA to use a "representative" year in determining a baseline rate of emissions. The EPA disputes this claim, arguing that "[a]s to NSPS, there is no 'representative emissions' concept.... Rather ... the baseline emission rates for units 1-5 are determined by hourly maximum capacity just prior to the renovations."7 Thomas Letter at 5.

WEPCO's interpretation of the regulations, at first blush, seems sensible: since the regulations require that the manual emission tests and continuous monitoring systems be based upon the "representative performance" of the facility, the emission factor test approach must also be based upon "representative performance." 40 C.F.R. Sec. 60.14 (1988); see 39 Fed.Reg. 36946, 36947 (1974) (explaining provision).8 Otherwise, the tests might reach inconsistent results, making the rate of emissions entirely dependent upon the type of test used by the facility. Hence, argues WEPCO, the EPA must examine the emission rates during a representative period, not 1987.

WEPCO's analysis, however, relies upon a flawed premise. WEPCO assumes that the phrase "representative performance of the facility" suggests that the EPA must choose a representative year. Read in context, however, the phrase refers generally to all the conditions of the test, not specifically to its timing:

Tests shall be conducted under such conditions as the Administrator shall specify to the owner or operator based on representative performance of the facility. At least three valid test runs must be conducted before and at least three after the physical or operational change. All operating parameters which may affect emissions must be held constant to the maximum feasible degree for all test runs.

40 C.F.R. Sec. 60.14(b)(2) (1988). Compare 40 C.F.R. Sec. 52.21(b)(21)(ii) (1988) (PSD program) ("The Administrator shall allow the use of a different time period upon a determination that it is more representative of normal source operation.") (emphasis supplied). Put simply, section 60.14 ensures that the operator will not doctor testing conditions to produce favorable emission results. The EPA's explanation of its regulations, which of course is given deference, supports this interpretation: "According to the proposed regulation, each set of emission tests (using manual tests or continuous monitors) conducted before and after a physical or operational change would consist of at least three runs, and would be conducted under representative operating conditions." 39 Fed.Reg. 36946, 36947 (1974) (emphasis supplied). WEPCO has not argued that it conducted its own tests under unrepresentative conditions, nor has it challenged any other part of the test protocol.9 And WEPCO does not claim that the tests were conducted during a period of operations that substantially differed from the normal operations of the deteriorated Port Washington plant. Further, the fact that the EPA permitted WEPCO to conduct additional emissions tests on the units (during which, presumably, WEPCO could maintain representative operating conditions) undermines WEPCO's assertion that the regulations were applied arbitrarily or capriciously.

b. PSD Measurements

Unlike NSPS, PSD is concerned with changes in total annual emissions, expressed in tons per year. The PSD regulations require preconstruction review of the construction or modification of major emitting facilities. These regulations define their key term--"major modification"--as "any physical change in or change in the method of operation of a major stationary source that would result in a significant10 net emissions increase of any pollutant subject to regulation under the Act." 40 C.F.R. Sec. 52.21(b)(2)(i) (1988) (footnote supplied).

Here the question is whether WEPCO's renovation project will result in "a significant net emissions increase" so as to trigger the "major modification" provision of the regulations and, as a result, PSD. To determine whether the project would result in an emissions increase, the EPA compared actual pre-renovation emissions with potential post-renovation emissions at the Port Washington plant. Specifically, the EPA first examined the two-year period of 1983 through 1984 as the pre-renovation baseline period, pursuant to 40 C.F.R. section 52.21(b)(21)(ii):

In general, actual emissions as of a particular date shall equal the average rate, in tons per year, at which the unit actually emitted the pollutant during a two-year period which precedes the particular date and which is representative of normal source operation. The Administrator shall allow the use of a different time period upon a determination that it is more representative of normal source operation. Actual emissions shall be calculated using the unit's actual operating hours, production rates, and types of materials processed, stored, or combusted during the selected time period.

40 C.F.R. Sec. 52.21(b)(21)(ii) (1988) (emphasis supplied). Because Administrator Thomas determined that the discovery of cracks in the rear steam drums led to a more recent "source curtailment," he relied upon the data from earlier years, 1983 and 1984, as the baseline to determine whether the renovation would cause an increase in emissions. Thomas Letter at 5. WEPCO does not challenge this component of the EPA's calculation.

Second, the EPA calculated the actual emissions of the plant following completion of the project. Generally, in order to apply PSD, the regulations require the EPA to find an "increase in actual emissions from a particular physical change or change in method of operation." 40 C.F.R. Sec. 52.21(b)(3)(i)(a ) (1988) (emphasis supplied). The EPA reasoned, however, that because the source "ha[d] not yet begun operations following the renovation, 'actual emissions' following the renovation [were] deemed to be the source's 'potential to emit.' " Clay Memorandum at 7. In support of its reliance on WEPCO's potential to emit, the EPA pointed to the regulations: "For any emissions unit which has not begun normal operations on the particular date, actual emissions shall equal the potential to emit of the unit on that date." 40 C.F.R. Sec. 52.21(b)(21)(iv) (1988).

WEPCO objects strenuously, and with good reason. In calculating the plant's post-renovation potential to emit, the EPA bases its figures on round-the-clock operations (24 hours per day, 365 days per year) because WEPCO could potentially operate its facility continuously, despite the fact that WEPCO has never done so in the past. And the EPA has admitted that it "assumed that emissions increases at Port Washington would come not from an increase in emission rate, but rather from increases in production rate or hours of operation." Supplemental Determination at 9; see Clay Memorandum at 7-8. The EPA responds that WEPCO can avoid these maximum calculations simply by consenting to federally enforceable restrictions on production rates and hours of operation, but WEPCO declines to do so. Clay Memorandum at 8; see 40 C.F.R. Secs. 52.21(b)(4), 52.21(b)(16) (1988). Thus, argues the EPA, it has no choice but to assume that the plant will be operated continuously.11

The first issue to be addressed is whether the EPA properly invoked the "potential to emit" concept in calculating the emissions increase. As explained above, the PSD regulations state that the EPA may rely upon a facility's potential to emit if the unit "has not begun normal operations on the particular date." 40 C.F.R. Sec. 52.21(b)(21)(iv) (1988) (emphasis supplied). WEPCO argues that this phrase should be interpreted to include only those units that have never been in operation, while the EPA urges that the phrase can be applied to both new and modified units.

The regulatory history of this phrase sheds little light on its proper interpretation. The EPA argues that it has always interpreted this phrase to include modified units; it asserts that its formulae for determining emissions increases have consistently assumed that "new or modified units " would be deemed to operate at maximum physical or federally enforceable levels. 45 Fed.Reg. 52676, 52718 (1980) (emphasis supplied). But the EPA's analysis here seems circular: in order to demonstrate that the Port Washington like-kind replacement project constitutes a modification, the EPA applies the potential to emit concept (to show an increase in emissions). And in order to apply the potential to emit concept to like-kind replacement, the EPA assumes that the plant is a "modified" unit. Although we accord great deference to an agency construing the statute it administers, Chevron, 467 U.S. at 844, 104 S.Ct. at 2782, and even more deference to an agency interpreting its own complex regulations, Aluminum Co. of Am. v. Central Lincoln Peoples' Util. Dist., 467 U.S. at 390, 104 S.Ct. at 2479, we cannot defer to agency interpretations that, as applied here, appear to assume what they seek to prove.12

We are also troubled by the EPA's assumption of continuous operations in calculating potential to emit at the Port Washington plant. Although we agree that the EPA cannot reasonably rely on a utility's own unenforceable estimates of its annual emissions,13 we find no support in the regulations for the EPA's decision wholly to disregard past operating conditions at the plant. Indeed, Alabama Power Co. v. Costle, 636 F.2d 323 (D.C.Cir.1979), which contributes importantly to the EPA's current PSD program, suggests otherwise. There, the D.C. Circuit held, in part, that the EPA must "take[ ] into account the anticipated functioning of the air pollution control equipment designed into the facility" when calculating the facility's potential to emit. Id. at 353. More important for our purposes, however, was the court's discussion of a unit's potential to emit:

If the source has no actual emissions because it has yet to commence operating, its hypothetical, projected emissions are included in the baseline. If, however, the source is an established operation, a more realistic assessment of its impact on ambient air quality levels is possible, and thus is directed.

Id. at 379 (emphasis supplied). The district court in United States v. Louisiana-Pacific Corp., 682 F. Supp. 1141 (D.Colo.1988), relying on Alabama Power, recently reached the same conclusion:

The broad holding of Alabama Power is that potential to emit does not refer to the maximum emissions that can be generated by a source hypothesizing the worst conceivable operation. Rather, the concept contemplates the maximum emissions that can be generated while operating the source as it is intended to be operated and as it is normally operated. Of course, it is possible that a source could be operated without the control equipment designed into it or that a Konus heater could be operated so badly that the fire would go out. Yet, Alabama Power stands for the proposition that hypothesizing the worst possible emissions from the worst possible operation is the wrong way to calculate potential to emit.

Id. at 1158.

In sum, we certainly do not suggest that the EPA may never subject replaced units to the potential to emit concept under its regulations. The EPA may, if it wishes, undertake notice and comment procedures to apply the potential to emit concept to like-kind replacement. See 42 U.S.C. Sec. 7607(d). But existing regulations do not seem to us to support such an application. We therefore believe that the EPA's reliance on an assumed continuous operation as a basis for finding an emissions increase is not properly supported. The EPA's determination that there has been a major modification for PSD purposes must be set aside.14

The final significant dispute in this case involves fuel switching. WEPCO proposed to the EPA that its "replacement project combined with an enforceable fuel switch would not 'result[ ] in an increase in the [sulfur dioxide and particulate matter] emission rate[s]' from those units." Petitioner's Brief at 50 (brackets in original); see 40 C.F.R. Sec. 60.14(a) (1988). Nonetheless, the EPA refused to permit WEPCO to utilize lower sulfur coal instead of implementing pollution control technologies to prevent an increase in emissions. The EPA explained that "the statute reflects a basic political decision that fossil fuel-fired sources not rely only on natural occurring less-polluting fuels to comply with the NSPS. Instead, Congress declared that compliance must depend in part upon the application of flue gas treatment or other pollution control technologies." Supplemental Determination at 10. Further, the EPA pointed to 40 C.F.R. section 60.14(b)(2) (1988), which requires that "operating parameters"--including fuel and raw materials--must be held constant in measuring emissions before and after renovations to determine whether the utility has undertaken a modification. WEPCO disputes the EPA's interpretation of the relevant provisions of the Clean Air Act Amendments.

Consistent with the Supreme Court's approach in Chevron, we first examine whether the statute evinces a clear congressional intent on the matter. We believe it does. Although the plain language of 42 U.S.C. section 7411 does not resolve the issue, the relevant legislative history provides ample support for the EPA's position. The House Conference Report, for example, states:

The agreement requires (1) that the standards of performance for fossil fuel-fired boilers be substantially upgraded to require the use of the best technological system of continuous emission reduction and to preclude use of untreated low sulfur coal alone as a means of compliance; ... (3) that for fossil fuel-fired sources, the new source performance standards must be comprised of both a standard of performance for emissions and an enforceable requirement for a percentage reduction in pollution from untreated fuel.

H.R.Rep. No. 564, 95th Cong., 1st Sess. 130, reprinted in 1977 U.S.Code Cong. & Admin.News 1077, 1510 (emphasis supplied). In addition, passages from the congressional debates reflect Congress's refusal to allow stationary sources to substitute low sulfur fuels to avoid a requirement of pollution control technology. See, e.g., III Senate Committee on Environment & Public Works, A Legislative History of the Clean Air Act Amendments of 1977, at 323, 353 (1978) (disapproving substitution of low sulfur coal for pollution control technology); IV Senate Committee on Environment & Public Works, supra, at 2653 (same). In these reports, Congress reasoned that the Administrator's previous standards--which had allowed fuel switching in lieu of pollution control technology--directly conflicted with the purposes of the NSPS program:

1. The standards give a competitive advantage to those States with cheaper low-sulfur coal and create a disadvantage for Midwestern and Eastern States where predominantly higher sulfur coals are available;

2. These standards do not provide for maximum practicable emission reduction using locally available fuels, and therefore do not maximize potential for long-term growth;

3. These standards do not help to expand the energy resources (that is, higher sulfur coal) that could be burned in compliance with emission limits as intended;

4. These standards aggravate compliance problems for existing coal-burning stationary sources which cannot retrofit and which must compete with larger, new sources for low-sulfur coal;

5. These standards increase the risk of early plant shutdowns by existing plants (for the reasons stated above), with greater risk of unemployment; and

6. These standards operate as a disincentive to the improvement of technology of new sources, since untreated fuels could be burned instead of using such new, more effective technology.

III Senate Committee on Environment & Public Works, supra, at 323. These purposes, reflecting technological and political choices, demonstrate that Congress rejected fuel switching as a method of avoiding the impact of NSPS. We believe Congress left us no choice on this issue.

In an era of increasing environmental concern, Congress enacted the Clean Air Act to "speed up, expand, and intensify the war against air pollution in the United States with a view to assuring that the air we breathe throughout the Nation is wholesome once again." H.R.Rep. No. 1146, 91st Cong., 2d Sess. 1, 1, 1970 U.S.Code Cong. & Admin.News 5356, 5356. The EPA is entitled to substantial deference in interpreting the technical provisions of the Act and its own regulations. We cannot grant deference, however, where the EPA has attempted to implement the Act's lofty goals in contravention of its own statutory regime. We therefore affirm in part and vacate in part, remanding the cause to the EPA for further proceedings not inconsistent with this opinion.