United States v. Hoechst Celanese Corp.

964 F. Supp. 967, 27 Envtl. L. Rep. (Envtl. Law Inst.) 20487, 44 ERC (BNA) 1814, 1996 U.S. Dist. LEXIS 19946, 1996 WL 898377

• Court: District Court, D. South Carolina
• Decided: May 10, 1996
• Docket: Civil Action 0:92-1879-17
• Status: Published

Opinion

MEMORANDUM OPINION AND ORDER

JOSEPH F. ANDERSON, Jr., District Judge.

Introduction

The question presented in this case is whether Hoeehst Celanese Corporation’s Celriver Plant in Rock Hill, South Carolina, was required to comply with the National Emission Standard for Equipment Leaks (Fugitive Emission Sources) of Benzene (Benzene NESHAP), 40 C.F.R. Part 61, sub-part J, during the period 1984 to 1991. The Company implemented the NESHAP in 1991 and subsequently eliminated the use of benzene altogether from the Celriver Plant. Thus, the focus of this litigation is the 1984-1991 time period.

On behalf of the Environmental Protection Agency (EPA), the United States initiated this action against Hoeehst Celanese for allegedly violating the Clean Air Act, 42 U.S.C. § 7401 et seq., and the benzene NESHAP at Celriver. These regulations prescribe standards for equipment in certain stationary sources that produce, use, or otherwise have in service benzene, a hazardous air pollutant, by requiring subject plants to monitor equipment regularly for leaks, to repair leaks promptly, and to install (in some circumstances) equipment that prevents, captures, or destroys benzene emissions. The regulations also include reporting and record keeping requirements. In its complaint, the United States seeks civil penalties of up to $25,000 for each violation.

The pertinent regulation provides an exemption for “any equipment in benzene ser vice” that is “located at a plant site designed to produce or use less than 1,000 megagrams of benzene per year.” 40 C.F.R. § 61.110(c)(2). The EPA contends that the Celriver Plant was designed to use more than two million megagrams of benzene per year, and therefore was not exempt from the regulation. Hoechst Celanese contends that the Celriver Plant was designed to use (and, in fact, used) less than 1,000 megagrams of benzene per year, and therefore was exempt. The dramatic difference between the two calculations is caused by the different ways the parties count the benzene for purposes of the exemption.

The two approaches produce vastly different results because the Celriver Plant recycled benzene, continuously using the same benzene. The total quantity of benzene at the Celriver Plant site never exceeded 1,000 megagrams per year. Hoechst Celanese maintains that the benzene should be counted only once, and therefore the Celriver Plant was exempt under the regulation. EPA argues that the benzene should be counted each time it cycles through two separate points inside the system, and therefore the Celriver Plant was not exempt.

Hoechst Celanese further argues that, even if the exemption contemplated what it calls the “multiple count” 1 approach, the Company still should not be required to pay the civil penalty sought by the government because the Company did not receive the constitutionally required fair notice of the “multiple count” interpretation of the regulation.

The matter is now before the court on pivotal cross motions for summary judgment. The United States moved for partial summary judgment on the issue of liability for violations allegedly occurring between June 1986 and August 1992. The United States contends that Hoechst Celanese violated virtually all of the leak detection and repair requirements in the regulations, failed to install equipment designed to reduce fugitive benzene emissions, and failed to keep records and make reports to EPA. In total, the United States alleged thousands of violations of the Benzene Leaks NESHAP. Hoechst Celanese has cross moved for summary judgment on all claims. Oral argument was heard on both motions on February 2, 1996.

For the reasons set forth below, the court will grant the plaintiffs motion in part, by finding plaintiffs interpretation of the applicable regulation to be reasonable. The court will also grant, in part, defendant’s motion, concluding that because the Company was not afforded fair notice of the EPA’s interpretation of the regulation, no civil penalty should be awarded in this case.

Summary judgment may only be granted under Federal Rule of Civil Procedure 56 when there are no genuine issues of material fact and the moving party demonstrates that it is entitled to judgment as a matter of law. The court must inquire into the genuineness and materiality of all purported factual issues and view all facts and inferences in a light most favorable to the party opposing the motion. Drewitt v. Pratt, 999 F.2d 774, 778 (4th Cir.1993); Ross v. Communications Satellite Corporation, 759 F.2d 355, 364 (4th Cir.1985). When no genuine issue of material fact exists, the court has a duty to award summary judgment. Drewitt v. Pratt, 999 F.2d at 778-79.

I

The Celriver Plant was exempt under 40 C.F.R. § 61.1 10(c)(2) if it was designed to produce or use less than 1,000 megagrams of benzene annually. The facts regarding benzene at the Celriver Plant are essentially undisputed. Until March 1993, the Celriver Plant used benzene in, among other things, the production of acetic anhydride. 2 Hot ketene gases flowed into a quench chamber and were cooled with a continuous stream of benzene, with some of the ketene thereby being converted to acetic anhydride. The benzene used to cool ketene gases was called a “quench.” The substances in the quench chamber then moved to a main still. In the main still, benzene, water, and other substances rose to the top and exited, while acetic anhydride and acetic acid left from the bottom of the still. A stream of benzene flowed into the still to make this separation process more efficient. Benzene used to increase the efficiency of the main still was called a “reflux agent.”

Benzene entering the quench chamber and main still had a prescribed temperature and purity. After it entered the quench chamber and main still, the benzene became hotter and less pure. For this reason, some of the benzene leaving the main still went through a series of steps that removed impurities and cooled the benzene back to its original temperature, while other portions of the benzene were recirculated immediately back to the main still. When the purification process was complete, the benzene that had been purified also recirculated back to the quench chamber and main still.

II

EPA contends that the Celriver Plant was not exempt because it was designed to use more than 1,000 megagrams of benzene annually. The court has jurisdiction to review EPA’s interpretation, but “the scope of its review is limited to whether EPA’s interpretation is plainly erroneous.” Potomac Elec. Power v. EPA 650 F.2d 509, 513 (4th Cir. 1981), cert. denied 455 U.S. 1016, 102 S.Ct. 1709, 72 L.Ed.2d 133 (1982). EPA’s interpretation need not be the only reasonable one, or even the interpretation the court would have preferred if the question had first arisen in a judicial proceeding. Udall v. Tollman, 380 U.S. 1, 16-17, 85 S.Ct. 792, 801-02, 13 L.Ed.2d 616 (1965). Rather, the “agency’s interpretation of its own regulation is of ‘controlling weight unless it is plainly erroneous or inconsistent with the regulation.’ ” Talley v. Mathews, 550 F.2d 911, 919 (4th Cir.1977), quoting Bowles v. Seminole Bock & Sand Co., 325 U.S. 410, 414, 65 S.Ct. 1215, 1217, 89 L.Ed. 1700 (1945). See also Ford Motor Credit Co. v. Milhollin, 444 U.S. 555, 566, 100 S.Ct. 790, 792, 63 L.Ed.2d 22 (1980).

Courts defer to an agency’s interpretation of its own regulation for pragmatic and constitutional reasons. Deference to the agency’s view ensures that a nationally uniform interpretation prevails without the need for the Supreme Court to opine on the meaning of every regulation. Homemakers North Shore, Inc. v. Bowen, 832 F.2d 408, 412 (7th Cir.1987). An agency is also more suited than a court to make policy and scientific judgments within the agency’s area of expertise. Chevron U.S.A v. Natural Resources Def. Council, 467 U.S. 837, 844-45, 104 S.Ct. 2778, 2782-83, 81 L.Ed.2d 694 (1984) (deferring to EPA’s interpretation of the Clean Air Act); Citizens Against Refinery’s Effects v. U.S. EPA 643 F.2d 178, 181-83 (4th Cir. 1981). Furthermore, courts recognize that Congress often confers upon an agency, rather than the courts, the province of “filling the interstitial silences within a statute or a regulation.” Milhollin, 444 U.S. at 565, 100 S.Ct. at 796.

A court’s review of regulations under Section 112 of the Clean Air Act, such as the Benzene Leaks NESHAP, is particularly limited because of the broad discretion Congress gave EPA to control hazardous air pollutants. Under the Clean Air Act in effect when EPA promulgated the Benzene Leaks NESHAP,' Congress delegated authority to EPA to identity hazardous air pollutants and their sources, to determine methods of controlling emissions from these sources, and to determine the level of control necessary to protect public health. 3 Through its exercise of these powers, EPA has developed considerable scientific and policy expertise in the regulation of hazardous air pollutants. Natural Resources Defense Council v. U.S. EPA, 824 F.2d 1146, 1163 (D.C.Cir.l987)(era banc). This court treads lightly when construing the scope of an exemption that allows a source of hazardous air pollutants to go unregulated. The court will not substitute its own judgment on the scope of the exemption unless EPA’s interpretation is plainly erroneous or inconsistent with the regulation.

Ill

A. The Language of the Regulation

The exemption that Hoechst Celanese claims reads as follows:

§ 61.110(e) ... (1) If an owner or operator applies for one of the exemptions in this paragraph, then the owner or operator shall maintain records as required in § 61.246(1).

(2) Any equipment in benzene service that is located at a plant site designed to produce or use less than 1,000 megagrams of benzene per year is exempt from the requirements of § 61.112.

The government argues that Hoechst Celanese does not qualify for the exemption provided for in section 61.110(c)(2). Additionally, the government argues that even if Hoechst Celanese qualifies for the exemption, it nevertheless cannot claim the exemption in this ease because it never afBrmatively applied for the exemption. The court will address these issues seriatim.

1. The Interpretation of the Exemption

Interpretation of the exemption at issue in this litigation turns on the phrase “designed to ... use” in section 61.110(c)(2).

EPA claims to give the word “use” its broad, ordinary meaning: to put into service, to employ, to utilize, to consume. In EPA’s view, a plant is not exempt if, under its design, it uses benzene in one of these ways at a rate greater than or equal to 1,000 megagrams per year.

EPA maintains that the Celriver Plant was not exempt because it was not designed to use less than 1,000 megagrams of benzene annually. The Agency relies on the flow of benzene through two internal pieces of equipment in one process unit at the plant (namely, the main still and the quench chamber in the acetic anhydride manufacturing process) in reaching that conclusion. As EPA sees it, that equipment in the Celriver Plant was not “designed to consume” benzene. Rather, the equipment was designed to “utilize” benzene as a quench and as a reflux agent in the manufacture of acetic anhydride. The acetic anhydride manufacturing unit of the plant was designed to use benzene as a quench by putting it into service in the quench chamber, and to use benzene as a reflux agent by putting it into service in the main still. According to EPA, these two pieces of equipment in the acetic anhydride unit at the Celriver Plant were designed to put benzene into service at the rate of 1,517,700 mega-grams per year. Because these rates exceeded the exemption threshold of 1,000 megagrams, EPA concludes that the Celriver Plant as a whole was designed to use more than 1,000 megagrams per year and, hence, did not qualify for the exemption.

Hoechst Celanese reads the exemption far differently. The Company agrees that the exemption is phrased in terms of a rate. Beyond that, however, EPA’s and Hoechst Celanese’s analyses diverge significantly.

■ At the very outset, Hoechst Celanese alleges that the Government’s emphasis on the words “designed to use” is a litigation position embraced by the Government post hoc, and that the Government’s explanation of the exemption can be found only in the papers it has filed in this litigation. As such, the Company contends, the “designed to use” argument is entitled to no deference from this court. According to the Company, “designed to use” merely indicates that a plant site’s benzene usage is to be calculated on the basis of its maximum “design” or “name plate capacity,” rather than its “actual” capacity for any particular year. This is made clear, according to Hoechst Celanese, both from the context of the phrase as it is used in the exemption itself and from the rulemaking record.

Moreover, Hoechst Celanese argues, the Government’s multiple counting explanation of the exemption strays far from any common understanding of the word “use.” To illustrate, the Company gives the example of how much oil an automobile might be said to “use” in a year. An automobile that circulates several quarts of oil in its engine might plausibly be viewed as using either those several quarts of oil, or perhaps the amount of oil needed as “makeup” for that lost from the engine over the course of the year. Under multiple counting, however, the automobile would be said to “use” hundreds of thousands, if not millions, of quarts of oil per year.

Hoechst Celanese claims that, unlike the Government’s multiple counting explanation, its interpretation of the exemption-provision gives the word “use” its ordinary meaning. According to the Company, there are two dictionary definitions of the word that might apply in the context of the exemption: (1) to “expend or consume by putting to use,” and (2) to “put into action or service: have recourse to ...: to carry out a purpose or action by means of.” 4

At Celriver, Hoechst Celanese contends, the amount of benzene “used” in a year could thus mean either the amount of benzene the plant is designed to “use up” (ie., consume) over the course of a year, or the amount-of benzene that the plant is designed to'“utilizé” or “put into service” over the course of year (i.e., the amount used up during the year plus the amount remaining in equipment at year’s end). Either of these definitions is a “single counting” approach, the Company notes, contrasting them with the so-called “multiple counting” interpretation that the Government advances. The distinction is critically important, as it is not disputed that the Celriver Plant was exempt under either single counting interpretation.

2. The Administrative Record

Both sides assert that the administrative record supports their respective reading of the exemption language. Hoechst Celanese argues that the administrative record establishes that the exemption is based on the single counting concept of plant nameplate capacity or consumption, terms that are wholly . incompatible with EPA’s multiple counting approach.

For its part, EPA relies on the administrative record to support its claim that its interpretation is consistent with the purpose of the exemption. Typically, that portion of the administrative record formally adopted by the EPA Administrator (ie., the Preamble and rulemaking documents incorporated by reference in the Preamble) is to be read in harmony with the language of the regulatory provision so as to provide guidance as to the Agency’s true meaning. I turn, then, to an examination of the administrative record.

Deference to an agency’s interpretation is particularly warranted when the interpretation harmonizes with the purpose of the statutory scheme and the purpose of the regulations. Mourning v. Family Publications Service, Inc., 411 U.S. 356, 369-72, 93 S.Ct. 1652,1660-62, 36 L.Ed.2d 318 (1973); United States v. Unitank Terminal Service, 724 F.Supp. 1158, 1165-66 (E.D.Pa.1989) (interpreting the Benzene Leaks NESHAP).

One purpose of the Clean Air Act is “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). To fulfill this purpose, Congress required EPA to promulgate emission standards for hazardous air pollutants that protect the public with an ample margin of safety. 42 U.S.C. § 7412. EPA promulgated the Benzene Leaks NESHAP under this authority in order to protect the public from benzene that leaks from equipment, a type of pollution known as “fugitive emissions.” EPA believed that these controls would significantly reduce the incidence of leukemia for people living within twenty kilometers of leaking equipment. 49 Fed.Reg. 23,501 (June 6, 1984). As discussed below, the exemption was intended to reach only plants that were not cost-effective to regulate because they had few pieces of equipment and low emissions from leaking equipment.

EPA drafted the exemption in response to a comment that Benzene Leaks NESHAP proposed in 1981 was not cost-effective for some plants. EPA agreed that an exemption for such plants was appropriate. In a document supporting the final regulations, the agency wrote: “EPA believes it is reasonable to exempt plants from the standard when the cost of the standard is unreasonably high in comparison to the achieved emission reductions. Therefore, EPA decided to determine a cutoff for exempting these plants based on a cost and emission reduction analysis.” Benzene Fugitive Emissions — Background Information for Promulgated Standards, EPA 450/3-80-032b (“BID”), at 2-104.

EPA performed a cost and emission reduction analysis by analyzing actual data from plants that produced ethylene, cumene, and vinyl acetate. 5 Based on its review of five data points provided by the three studied production facilities, PES plotted lines purporting to make correlations between cost-effectiveness and (1) the nameplate capacity of plants producing so-called “synthetic organic chemical manufacturing industry” (“SOCMI”) chemicals, (2) emissions from such facilities, and (3) the number of pieces of equipment at the facilities.

According to EPA, the data show a rough correlation between a plant’s equipment count, emissions from leaking equipment, and plant production nameplate capacity, in that plants with higher production rates had more equipment and thus more fugitive emissions.

After assuming that the data it had derived from five data points was an accurate reflection of SOCMI production facilities generally, and by relating the data to benzene production plants, EPA’s consultant plotted a correlation between plants with nameplate capacity producing '1,000 mega-grams of benzene per year and benzene emissions of approximately 6 megagrams before regulation and emission reductions of about 4 megagrams after regulation. On the graph plotted by PES, emissions at these levels corresponded to plants having fewer than 87 pieces of equipment, such as valves, pumps, and pressure relief devices.

EPA determined that plants designed to use 1,000 megagrams of benzene per year would have approximately the same equipment count and emissions as plants that produced 1,000 megagrams of benzene annually. In the BID, EPA stated:

For plants in which the benzene emission reduction achieved by implementing. the standard is about 4 Mg of benzene per year, EPA has determined that the cost of achieving this emission reduction is unreasonable. In order to exclude plants on this basis, EPA selected a minimum cutoff of 1,000 Mg/year per plant site based on a benzene design usage rate or throughput.

BID at 2-105. The BID further explained that “[t]wo approaches for setting the [1,000 megagram per year] cutoff were considered: (1) number of sources at a plant site and (2) benzene usage per plant site.” BID at 2-104. EPA said it rejected a “cutoff based on the number of sources at a plant site” because such a cutoff “could not be applied readily” to small or intermittent users of benzene. Id. Rather, a cutoff “based on benzene design usage” was chosen “because it is easily understood and can be applied readily to a plant site.” Id. The BID also indicated that it would be reasonable to exempt from the NESHAP plants emitting benzene in “significant quantities” when the “cost of the standard is unreasonably high in comparison to achieved emission reductions.” BID at 2-104.

The Preamble to the Benzene Leaks NESHAP'similarly explains that the exemption was based indirectly upon an analysis of equipment count and emissions. The Preamble states:

EPA is exempting from the standard equipment at plant sites that are designed to produce or use 1,000 Mg/yr or less of benzene. This cutoff is based on the amount of equipment in a process unit and relates this amount to a design production rate. The 1,000 Mg/yr exemption would exclude most research facilities, pilot plants, and intermittent users of benzene from the standard.

And finally, even though pharmaceutical operations that are designed to use benzene in excess of 1,000 Mg/yr are subject to the standard, they have substantial equipment' inventories in benzene service, and, therefore, emit benzene in enough quantity to warrant coverage by the final standard.

49 Fed.Reg. 23,510 (June 6,1984).

EPA contends that its position that the Celriver Plant did not qualify for the exemption is consistent with the purpose of the regulation. Unlike plants that EPA intended to exempt, EPA says, the Celriver Plant had thousands of pieces of equipment and high fugitive benzene emissions. Between June 1984 and February 1991, for example, the Celriver Plant operated almost 17,000 pieces of equipment allegedly covered by the Benzene Leaks NESHAP. Between 1984 and 1993, the Celriver Plant’s fugitive benzene emissions ranged between 69 and 226 mega-grams annually. According to EPA’s Toxic Release Inventory, the Celriver Plant ranked in the top 5% of all plants reporting benzene fugitive emissions in every year between 1987 and 1993.

EPA raises other arguments to support its claim of consistency with guidance in the administrative record on applying the exemption. In the BID, EPA drew a parallel between benzene use and benzene “throughput.” According to EPA, the term throughput generally refers to the rate at which material is put through something, whether it be a system, a process, or a piece of equipment. Instead of looking at the “throughput” for the plant as a whole— which, as is explained below, Hoechst Celanese and its engineering experts contend is the only way for it to be consistent with the plant nameplate capacity element of the PES analysis and other aspects of the administrative record — the Government’s multiple count approach would base eligibility on the throughput for one or more pieces of equipment inside the plant process. To accomplish this, EPA equates “throughput” with the term “processing rate,” which, in turn, is more typically used when referring to the flow of material through a piece of equipment. This precise term, however, does not itself appear in the administrative record.

The BID states, “EPA selected a minimum cutoff of, 1,000 Mg/year per plant site based on a benzene design usage rate or throughput. Throughput is determined by a mass balance during the design stages of process operations, accumulating all benzene processed in 1 year.” BID at 2-15. This passage, EPA explains, describes both how throughput should be determined (by a mass balance), 6 and what benzene should be included (all benzene processed in one year). The Government argues that the “mass balance” referred to in the BID was not intended to be conducted on the plant manufactur ing process as a whole, but rather should really have been understood to contemplate one or more mass balances to measure the accumulated flow through pieces of equipment inside the plant manufacturing process selected by Agency staff after reviewing plant process diagrams.

Notwithstanding • the fact that the term “processing rate” does not appear in the rulemaking record, EPA argues that the BID reiterates the connection between “use” and processing rate or throughput in several other places as well. In a summary section under the heading “Benzene Usage Cutoff,” EPA stated:

An analysis by EPA indicated that the cost effectiveness of the standard for equipment at plant sites processing less than 1,000 Mg/yr is unreasonably high. Because this cost is unreasonably high in comparison to the minimal emissions reduction, achievable, EPA provided an exemption for plants designed for low benzene usage.

BID at 1-10. In response to a comment that research and development plants should be exempt, the agency stated:

a plant design usage rate or throughput rate of benzene equal to or less than 1,000 Mg/yr per plant has been selected as a cutoff. Based on the minimal amount of benzene processed at laboratory facilities and pilot plants, the 1,000 Mg/yr cutoff is likely to exempt most of these sources from the standard. Large-scale pilot plants or commercial tests might process more than 1,000 Mg/yr. If this happens, these plants would not have difficulty in achieving the standard since there are no major equipment requirements and the work practice requirements are reasonable, considering costs and emissions reductions available.

BID at 2-105. In response to a comment from a pharmaceutical company seeking an exemption, EPA stated:

And finally, even though pharmaceutical operations with a benzene throughput in excess of 1,000 Mg/yr are subject to the standard, they have substantial equipment inventories in benzene service and, therefore, emit benzene in enough quantity to warrant coverage by the final standard.

BID at 2-91.

EPA relies on these passages in support of its view that the Celriver Plant was not exempt. EPA contends that the Celriver Plant was not exempt based on the flow rate of benzene through two pieces of equipment inside' the manufacturing process — the quench chamber and main still. EPA obtained this flow from among the mass balances that Hoechst Celanese prepared, at EPA’s direction, on various benzene streams in Cehiver’s internal processes to show those streams’ “throughput,” which the Agency then defined as meaning “total cumulative flow through equipment in benzene service rather than net consumption.”' 7 .

Some parts of the administrative record provide support for the Defendant’s “consumption” interpretation. Chief among these is a paragraph from the Administrator’s Preamble which uses the word “consumed” or “consumption” four times, and appears to use the concept synonymously with the word “use”. This paragraph reads as follows:

The possibility that pharmaceutical operations could be adversely affected by the standard is very small. This is true for several reasons. First; most pharmaceutical plants use very little benzene. According to estimates contained in Market Input/Output Studies — Benzene Consumption as a Solvent ..., 1978 benzene consumption by pharmaceutical manufacturers was about 0.72 Gg [gigagrams]. No companies consumed more than 1,000 Mg/yr in 1978. The commenter states that they consumed about 325 Mg/yr during 1981. Thus, it is unlikely that phar maceutical operations would be affected by the standard because the final standard exempts equipment at plant sites that are designed to produce or use 1,000 Mg/year or less of benzene. Second, benzene consumption by the pharmaceutical industry is dechning rapidly.

49 Fed.Reg. 23,510 (June 6, 1984) (emphasis added).

In its brief the government explains that the placement of the words “consumed” and “consumption” in close proximity with the word “use” in both the Preamble and the BID is “accidental” and that EPA could have drafted these passages “more exactly”.

Hoeehst Celanese contends that the Preamble indicates that EPA intended to exempt small users of benzene, but EPA’s multiple counting interpretation would virtually eliminate the exemption in even the simplest and smallest of recycle operations. Hoeehst Celanese asserts, for example, that a research facility that intermittently recycles one gallon of benzene through an aquarium pump at a design flow rate of one gallon per minute would not be exempt.

Additionally, the precise term “processing rate,” (the. term relied upon by some EPA staff as the basis for a multiple count interpretation) is not mentioned anywhere in the record. Hoeehst Celanese argues that EPA adopted nameplate capacity and specifically rejected “processing rate,” a version of multiple counting, as unsuitable for this type of exemption. Moreover, it cites to a parallel EPA rulemaking to establish other standards for the synthetic organic chemical manufacturing industry where the use of that term was specifically rejected because it was “difficult to establish.” The government argues that although the words “processing” and “rate” do not appear together in the BID, the BID does describe the scope of the exemption in terms of the rate at which benzene is processed.

Hoeehst Celanese also points to the 1980 BID which listed the types of processes which were being considered for benzene regulation. The listed processes included those that produced benzene (refineries) and those that used benzene in the production of benzene derivatives. The 1980. BID had information on 133 sites, but did not list any of Hoeehst Celanese’s plants, some of which had been operating for decades.

These passages from the record, however, do not lead inevitably to the conclusion that EPA intended the word “use” in the exemption to mean only “consume.” As the Defendant concedes, a mass balance can be used to determine the flow of benzene through equipment. The Defendant prepared such a mass balance for EPA in order to show the flow of benzene through various parts of its plant, including the quench chamber and main still. Benzene throughput sometimes refers to a plant’s benzene consumption, but the words “throughput” and “consumption” are not synonymous. The word “throughput” can describe, for example, the flow of benzene through equipment like the quench chamber and main still. Furthermore, EPA maintains that the Defendant has not show how its consumption interpretation squares with the purpose of the Benzene Leaks NESHAP, which according to EPA was to control fugitive benzene emissions at all plants except those with few pieces of equipment and low emissions. 8 In sum, the passages cited by the Defendant from the administrative record do not establish that EPA’s interpretation is plainly erroneous.

The court concludes that EPA’s interpretation of the regulation is permissible and should be sustained by the court. Although Hoeehst Celanese’s interpretation of the regulation is also reasonable, the court must “defer to the reasonable judgment of the agency to which Congress has entrusted the development of rules and regulations to ensure [the protection of the environment].” General Electric Co. v. U.S. EPA 53 F.3d 1324, 1328 (D.C.Cir.1995). EPA’s interpretation of 40 C.F.R. § 61.110(c)(1) is not “plainly erroneous or inconsistent with the regulation.” Talley v. Mathews, 550 F.2d 911, 919 (4th Cir.1977).

3. The “Application” Argument

EPA’s other argument is that Hoeehst Celanese never applied for the ex- emption it now claims for itself. Hoechst Celanese contends, on the other hand, that the exemption is self-implementing because the plain language of Section 61.110(c)(2) provides that a plant designed to use less than 1,000 megagrams of benzene per year “is exem/pt from the requirements.” 40 C.F.R. § 60.110(c)(2) (emphasis added).

Courts look to the language and design of the whole law when construing the meaning of particular phrases, U.S. Nat. Bank of Oregon v. Independent Ins. Agents, 508 U.S. 439, 453-55, 113 S.Ct. 2173, 2182, 124 L.Ed.2d 402 (1993), and avoid constructions that render words superfluous. Ratzlaf v. United States, 510 U.S. 135, 140-42, 114 S.Ct. 655, 659, 126 L.Ed.2d 615 (1994). The paragraph containing the exemption begins, “[i]f an owner or operator applies for one of the exemptions in this paragraph____” 40 C.F.R. § 61.110(c)(1). Elsewhere, the NESHAP requires that applications made pursuant to Part 61 should be sent to the appropriate EPA regional offices. 40 C.F.R. § 61.04.

The criteria that must be met for the exemption is specified in 40 C.F.R. § 61.110(c)(2). The regulatory language does not give EPA discretion in determining who qualifies for the exemption; a plant qualifies for the exemption if it meets the criteria set forth in the regulation. EPA still maintains, however, that this lack of discretion does not make an application unnecessary.

The word “apply” in the regulation cannot be ignored. The EPA acknowledges, however, that the regulation does not specify the content of an application, nor does it give any real guidance as to an application procedure. Hoechst Celanese maintains that the regulation is self-executing and that the EPA has advanced the application argument only for the purposes of litigation and that in actual practice, the regulated community has not been applying for, nor have they been required to apply for this exemption. Hoechst Celanese also argues that key EPA staff have confirmed that there is no obligation to apply for the exemption.

EPA acknowledges that the ordinary meaning of the verb “apply” in 40 C.F.R. § 61.110(c)(1) means at least that plants seeking the exemption must “communicate” their request to EPA. 9 Because of this, an examination of the communications between EPA representatives and the Company after promulgation of the NESHAP in 1984 is appropriate. The relevant facts, derived from the undisputed evidence in the record, are as follows.

In 1984 Hoechst Celanese operated two Texas plants that recycled benzene: the Bishop and Pampa Plants. Following the June 1984 promulgation of the NESHAP, Hoechst Celanese undertook to determine the scope of the “use” exemption. The initial evaluation was made by Jaime Garza, a senior environmental engineer at Bishop. Mr. Garza considered a full range of regulatory interpretations — both plausible and implausible. He also asked for a review of the matter by in-house counsel, who evaluated the rulemaking record, including the regulatory Preamble and the BID. After these evaluations, Mr. Garza concluded that “use” meant “consumption” and that Bishop was exempt because it consumed less than 1,000 megagrams per year of benzene.

On August 31, 1984, Mr. Garza wrote a letter seeking confirmation of Bishop’s exempt status from the Texas Air Control Board (TACB), the agency to which EPA had delegated the authority to implement and enforce the benzene NESHAP in Texas.

Mr. Garza provided the TACB a detailed breakdown of annual benzene consumption and purchases for Bishop. His letter and attachments showed that Bishop recycled benzene and that its total consumption and purchases of benzene for each of the prior six years were less than 1,000 megagrams.

In response to Mr. Garza’s letter, another TACB representative (Mr. Rodgers) informed Mr. Garza about an August 20, 1984 letter sent from EPA Region VI (EPA’s re gional office for Texas) to a Texaco Chemical Company facility in that same EPA Region. The letter to Texaco said that the term “use” was “not meant to imply consumption,” but rather was “meant to reflect the overall quantity of benzene used in equipment at a facility,” and that the “total quantity in use at a facility needs to be considered” not just consumption.

After receiving a copy of that letter in September 1984, Mr. Garza called the TACB and discussed the matter with Mr. Palmer, the director of the TACB region responsible for the Bishop Plant. After reviewing the Texaco letter and checking with EPA, the TACB told Mr. Garza that the TACB wanted additional information regarding the amount of benzene “kept at the Bishop Plant site at any point in time.” In response, on October 22,1984, Mr. Garza sent another letter to the TACB, explaining that the maximum amount of benzene present at Bishop, both in equipment and storage, was less than 1,000 mega-grams a year.

Before responding to Mr. Gaza’s letters, Sabino Gomez, then Director of TACB’s Compliance Division, conferred with EPA Region VI about EPA’s interpretation of the word “use” in the exemption. From those conversations, Mr. Gomez understood that EPA interpreted the exemption to be based not on benzene consumption at a plant site, but rather on what a plant’s inventory of benzene was. Mr. Gomez also specifically asked EPA Region VI “whether ‘designed’ is the key word in this rule.” EPA Region VI staff, after consulting with EPA’s Office of Air Quality Planning and Standards in North Carolina, informed Mr. Gomez, “design capacity, irrespective- of actual use, is the operating factor.” 10

On the basis of those conversations with EPA and based on the information in Mr. Garza’s August and October 1984 letters, the TACB confirmed — by letter of December 7, 1984 — that Bishop was exempt from the benzene NESHAP. Significantly (for purposes of this litigation), as soon as Hoechst Celanese received the confirmation letter from the TACB, it circulated the information to company personnel whose responsibilities included the Celriver Plant.

TACB sent EPA a copy of the December 7,1984 letter confirming the exempt status of the Bishop Plant. After receiving the letter on December 21, 1984, EPA did not seek further information from either the TACB or Hoechst Celanese, nor did EPA. indicate that it disagreed with the TACB’s confirmation of the exempt status of a plant that recycled benzene as Bishop (and Celriver) did.

During this same period, Hoechst Celanese was also evaluating the terms of the exemption for its Pampa Plant. Like Celriver and Bishop, Pampa then recycled benzene continuously through a closed-loop system. Unlike Celriver and Bishop, however, in 1984 Pampa was consuming more than 1,000 megagrams of benzene per year. Consequently, in September 1984, the Pampa Plant applied to the TACB for a two-year waiver from compliance with the benzene NESHAP so that the plant could become exempt from the benzene NESHAP by reducing “the quantity of benzene consumed in the plant to less than 1,000 megagrams.” The TACB approved the waiver request for Pampa on April 9, 1985, stating that if benzene consumption were reduced to 1,000 megagrams per year by June 1986, the exemption would apply. Copies of both the waiver request and TACB approval were sent to EPA Region VI.

In June 1986, the Pampa Plant reported to the TACB that the facility had reduced its benzene consumption to less than 1,000 megagrams per year and was therefore entitled to the exemption. The Company provided EPA Region VI with a copy of the letter. On December 9, 1986, EPA Region VI inspector George Marusak conducted a compliance monitoring inspection at the Pampa Plant, during which he and the TACB evaluated whether Pampa was in compliance with the benzene NESHAP. Dr. Marusak’s Compliance Monitoring Report concluded that Pampa was exempt because it had reduced its “consumption of benzene to less than 1,000 Mg per year.” This report was re ceived and approved by Dr. Marusak’s supervisors at EPA Region VI. 11

As discussed later in this opinion, the Company did not communicate with EPA about an exemption specifically for the Celriver Plant until 1989. The facts reveal, however, that it was confirmed to the Company by EPA representatives that Hoechst Celanese’s other plants which used benzene in identical processes were exempt. Although the Company did not formally apply to the EPA for an exemption for Celriver, it did so indirectly through its communications with EPA regarding its Texas plants, which used essentially the same process as Celriver. Therefore, the court declines to accept the EPA’s assertion that a formal application was a necessary prerequisite to qualifying for the exemption and concludes that the communications with EPA were sufficient to qualify for an exemption for the Celriver Plant.

TV.

The second issue before the court is whether Hoechst Celanese received fair notice of the government’s interpretation of the benzene NESHAP exemption, as required under constitutionally based principles of due process. Hoechst Celanese is not liable and is not subject to penalties if the multiple counting interpretation described and deferred to above 12 is not “ascertainably certain” from the regulatory language and the Administrator’s statements in promulgating the rule. See General Electric, 53 F.3d at 1328-29, 1333-34. For the reasons discussed below, the court determines that Hoechst Celanese is not liable and is not subject to penalties.

A. The “Fair Notice” Standard 13

Where an agency has provided no pre-enforcement warning of its interpretation, the question facing a reviewing court is “whether the regulated party received, or should have received, notice of the agency’s interpretation in the most obvious way of all: by reading the regulations.” General Electric Co., 53 F.3d at 1329. Fair notice is not provided unless a regulated party acting in good faith is able to identify with “ascertainable certainty,” on the face of regulations and other public statements issued by the agency, the standard to which the regulating agency expects it to conform. Id. Without such fair notice, the regulated entity does not commit a wrong when it fails to meet the regulatory standard.

Unclear or ambiguous regulations do not provide the requisite notice. If a violation of a regulation subjects private parties to criminal or civil sanctions, a regulation cannot be construed to mean what an agency intended but did not adequately express. Gates & Fox Co. v. OSHRC, 790 F.2d 154, 156 (D.C.Cir.1986), quoting Diamond Roofing Co. v. OSHRC, 528 F.2d 645 (5th Cir. 1976). The test is “not what [the agency] might possibly have intended, but what [was] said.” United States v. Trident Seafoods Corp., 60 F.3d 556, 559 (9th Cir.1995) (brackets in original).

A regulated party may also be found to have been given fair notice of the agency’s interpretation in- those situations where, pri- or to the commencement of enforcement efforts, the regulated party was given actual notice of the interpretation. See General Electric Co., 53 F.3d at 1329. However, such notice must be an “authoritative interpretation of the regulation” from the regulating agency. Gates & Fox Co., 790 F.2d at 157.

Particularly compelling evidence that fair notice of a rule has not been given exists where the agency itself has not provided a definitive or consistent reading of the regulation. See, e.g., General Electric Co., 53 F.3d at 1332; Diamond Roofing Co., 528 F.2d at 649. When the “agency itself is uncertain of the meaning of its regulation” and “agency personnel give conflicting advice to private parties about how to comply with it,” it is “arbitrary to find the regulation ‘clear.’” Rollins Envtl. Services v. U.S. EPA, 937 F.2d 649, 653 (D.C.Cir.1991).

Due process concerns are clearly raised when unpublished “clarifying” interpretations cause the regulatory language and published guidance to be misleading. Misleading regulatory language, even if “clarified” with the best of intentions, is itself an ongoing cause of noncomplianee and confusion. Unlike the unpublished clarification, the official language is clearly authoritative and available to the public. Only the agency can amend the regulation and publish guidance when correcting an “accident” to ensure that they clearly communicate the meaning that will be enforced. Awarding penalties where the problem is the language chosen by the agency, rather than any unwillingness to comply would only “delay the day when ... regulations will be written in clear and concise language” so that regulated parties “will be better able to understand and observe them.” Diamond Roofing Co. v. OSHRC, 528 F.2d 645, 650 (5th Cir.1976).

B. Information in the Regulatory Record

Even though the court has deferred to the EPA’s interpretation of the NESHAP exemption as permitting a multiple counting of recycled benzene, the court believes that the regulatory record in this case 14 does not provide fair notice of that interpretation. Indeed, Hoechst Celanese has provided numerous examples of ways in which rulemaking record statements fail to give notice of a multiple count meaning. These examples and the court’s analysis thereof, have been set forth on pages 19-21 of this memorandum opinion and order.

Additionally, Hoechst Celanese provides an affidavit from Robert Ajax, Chief of the EPA unit that actually developed the benzene NESHAP. Mr. Ajax states that EPA chose to have the exemption on benzene “use,” based on single counting,' because it was an easy-to-apply surrogate for the cost-effectiveness policy that EPA wanted the exemption to implement. It is “unequivocally clear,” Mr. Ajax says, that “internal recycle” — i.e., multiple counting — was never “considered [n]or suggested as the basis for the NESHAP exemption.”

Similarly, the rulemaking record is inconsistent concerning EPA’s claim that a multiple counting approach would ensure that facilities with low fugitive emissions or few pieces of equipment in benzene service would be exempt while those with higher emissions and more equipment would be subject to the NESHAP. As noted in Attachment F of Hoeehst Celanese’s Initial Brief, under the government’s approach, some very large emitters of benzene would be exempt and some plants with very low emissions and very few pieces of equipment in benzene service would not be exempt.

In addition, the rulemaking record is somewhat inconsistent as to EPA’s claim that it intended to exempt only small sources. The Preamble shows that EPA also intended to exempt users of small amounts of benzene — a category that would include all those that recycled benzene as a solvent or quenching agent, as was done at Hoeehst Celanesé’s Celriver Plant and at the pharmaceutical facility described by — and declared to be exempt in — the NESHAP Preamble. 49 Fed. Reg. 23510 (June 6,1984). 15

In this regard, the court also finds compelling that in the months (and years) following promulgation of the NESHAP, because of the confusion precipitated by EPA staffers’ conflicting internal communications, various EPA offices understood the exemption in a variety of different ways. 16 In fact, the record in this case establishes that on various occasions from 1984 through at least 1990, EPA advised some facilities that they were exempt based on a single counting of recycled benzene and in other cases, EPA did not dispute state exemption decisions based on single counting. If the rulemaking record did not give EPA staffers notice of a multiple counting interpretation, it cannot reasonably be said that Hoeehst Celanese should have understood that to be what the Administrator meant.

The court concludes from all this that although the multiple counting approach is reasonable enough to be sustained when deference is given to the government’s present interpretation, the regulatory language and administrative record do not afford sufficient notice of the EPA interpretation.

C. Actual Notice from Events in 1984-1986

EPA argues that, apart from anything contained in the public record, Hoeehst Celanese had actual notice that EPA was interpreting the benzene NESHAP exemption as requiring a multiple counting of recycled benzene. According to EPA, this defeats Hoeehst Celanese’s reliance on a “fair notice” defense. EPA’s actual notice arguments are based largely on the previously described events that occurred in Texas after promulgation of the NESHAP,, in particular Hoeehst Celanese’s receipt of a copy of the Texaco letter, which stated that the word “use” was to mean the “total quantity in use at a facility.” 17 Although the Texaco letter cannot reasonably be said to have provided Hoeehst Celanese with actual notice that its benzene usage was to be determined by multiple counting, 18 EPA argues that the Texaco let ter did give Hoechst Celanese notice that the Agency did not intend to equate “use” with “consumption” and that, in turn, should have prompted Hoechst Celanese to ask EPA staff — rather than the TACB — whether its plants were exempt. Instead of seeking such confirmation, the government suggests that Hoechst Celanese chose to remain ignorant of EPA’s reading of the regulation.

The court disagrees. This argument ignores that Hoechst Celanese sought — and received — confirmation of the exempt status of its plants from both State regulators and EPA Region VI. In fact, Hoechst Celanese received a copy of the Texaco letter only because the Company went to the appropriate agency (the TACB 19 ) and asked it to confirm the exemption’s applicability. TACB responded to Hoechst Celanese’s inquiry only after first conferring with EPA about the matter.

Nor does the court find merit in EPA’s argument that the TACB’s reading of the regulation was irrational and thus should have prodded the Company to seek further guidance from EPA. The TACB’s single count application of the Texaco letter — although not based on just consumption — was still consistent with a single count “nameplate capacity” concept, which the NESHAP rulemaking record suggests may be the basis of the exemption. 20

Other government arguments that the TACB’s interpretation was invalid are not relevant here. The TACB was not saying that it advocated that interpretation. Rather, as explained by Mr. Gomez in his affidavit and by Mr. Garza in his deposition, the TACB was simply stating the interpretation that it thought was being advocated by EPA staff. The court finds unreasonable EPA’s suggestion that Hoechst Celanese had an affirmative obligation to question its exempt status further when its plants were exempt based on (1) a “use equals consumption” reading of the rule, (2) an alternative (“use equals inventory”) reading which both the TACB and Hoechst Celanese believed to be how EPA was reading the rule, and (3) any other reading of the exemption that called for a single counting of benzene at a plant site. The whole point of the “fair notice” line of cases is that, as a matter of due process, the regulatory obligations to which persons are subject flow from what the regulations themselves say, with ascertainable certainty. Companies are not obliged to seek “clarifications” from agency staff of every conceivable ambiguity that might be lurking in regulatory language. An “actual notice” argument can be predicated only upon notice that Hoechst Celanese actually received, not upon notice it might have received if it had continued to make unlimited additional inquiries.

The court also finds compelling that even if Hoechst Celanese had asked more individuals within EPA about the meaning of the exemption, it is far from clear that the Company would have been told about EPA’s multiple counting interpretation of the rule. As noted above, because of confusion generated in large part by internal memoranda from EPA staffers seeking to implement a multiple counting approach, EPA’s regional offices were inconsistent as to how the regulation should be applied and, on various occasions from 1984 through at least 1990, told inquiring companies that the exemption was to be based on a single counting of recycled benzene.

D. Evaluating Actual Notice After Commencement of Enforcement Against the Celriver Plant

Finally, the Government argues that Hoechst Celanese cannot rely on its “fair notice” defense for any violations that occurred after the summer of 1989, when EPA Region IV (EPA’s regional office in Atlanta) sent letters to the Celriver Plant concerning its compliance with the benzene NESHAP. The facts relevant to this argument, again derived from undisputed record evidence, are the following.

In June 1989, EPA Region IV sent a letter to Hoechst Celanese, stating that Celriver “may be subject” to the benzene NESHAP and asking for information on Celriver’s benzene “throughput on an annual basis.” The letter defined “throughput” as the “total cumulative flow through the process rather than net benzene consumption or usage.” Hoechst Celanese replied on June 26, 1989, reporting that Celriver was exempt because the plant’s consumption of benzene was far less than 1,000 megagrams annually.

On August 18, 1989, in reply to Hoechst Celanese’s June 26 letter, EPA Region TV sent a second letter stating, “it appears that you are unaware of EPA’s interpretation of benzene usage as the term is used to determine applicability----” Attempting, in its words, “to clarify what is meant by benzene usage,” EPA gave an example of how to calculate usage in a closed-recirculation system. In the example, benzene was considered “used” each time it was recycled. The letter indicated that this “interpretation of the term benzene use” was “based upon EPA Headquarters guidance provided shortly after [the benzene NESHAP] was promulgated” and that the basis for whatever was in that guidance (copies of which were not provided to Hoechst Celanese ) was in the Agency’s final BID.

In a September 1989 cover letter indicating that it did not then agree that multiple counting was a part of the benzene NESHAP exemption, Hoechst Celanese nevertheless provided the “throughput” calculations in the manner that Region TV had requested, by using flow rates at several points inside the process. This produced a Region TV-requested “throughput” calculation of over 2.5 million megagrams of benzene per year — a number that was orders of magnitude higher than the amount that the Celriver Plant ever had on site. While awaiting a response from EPA, Celriver also began work to determine what would need to be done to implement the NESHAP at the Celriver Plant if that was determined to be necessary.

On February 20, 1990, EPA Region IV sent the Celriver Plant a formal notice that EPA considered that plant to be in violation of the benzene NESHAP “based upon information provided by the [Celriver Plant] in response to EPA letters ... dated June 18, 1989 and August 18, 1989.” The Company set up a meeting with Region TV to discuss the exemption interpretation. At that April 1990 meeting, the Company affirmed that although it continued to believe Celriver was exempt, it would meet the terms of the benzene NESHAP at the Celriver Plant.

EPA Region IV then asked for — and within a few days received from Hoechst Celanese — the schedule that Hoechst Celanese intended to follow in meeting the terms of the NESHAP at Celriver. That schedule showed that while Hoechst Celanese could have implemented the NESHAP between 1990 and 1992 at a cost of between $2 and $3 million, the Company planned another much more expensive — approach, parts of which could be implemented more promptly. Specifically, Hoechst Celanese described a program to phase out the use of benzene entirely at one of its processes in less than one year, at an additional cost of $8 million. Hoechst Celanese estimated that it would be able to meet all remaining terms of the NESHAP no later than December 31, 1992.

The Celriver Plant then met the terms of the NESHAP by September 1991 — ie., within approximately eighteen months of the time that Hoechst Celanese gave its compliance schedule to EPA and within approximately two years of the time the Government contends the Company was given actual notice of EPA’s interpretation. 21 Later Celriver completely phased-out benzene usage at a cost of more than $40 million.

Based on these facts, the court concludes that it is inappropriate to find Hoechst Celanese liable or to impose penalties for actions it took between 1990 and 1992. The court bases this conclusion on several factors.

First, the court concludes that Hoechst Celanese did not have actual notice of EPA’s multiple counting interpretation in the summer of 1989. The interpretation in the letters from EPA Region TV were, after all, contrary to statements in the Preamble and contrary to the interpretations of the exemption that the Company had received from EPA Region VI, and state agencies in Texas, and Virginia. Thus, even if Hoechst Celanese may'have known from its correspondence that individuals in the Atlanta regional office were then subscribing to a multiple counting interpretation of the exemption, Hoechst Celanese had a legitimate basis for believing that EPA’s Region IV office was not speaking for the Administrator. In fact, the multiple counting reading of the exemption was not an authoritative reading even within Region TV since, during the same time period, other individuals within EPA Region IV were interpreting the exemption as calling for a single counting of benzene at a plant site. 22 Hoechst Celanese could not have been put “on notice” that EPA had adopted a multiple counting interpretation when, at the same time, the Agency itself was not consistently embracing such an interpretation.

Similarly, this court concludes that none of EPA Region TV’s communications with Hoechst Celanese — up to and including the issuance of a notice of violation (NOV) on February 20,1990 — can be deemed as having provided Hoechst Celanese an authoritative interpretation of the benzene exemption, when that interpretation is contradicted on its face by the language used by the Administrator in promulgating the final rule and was not communicated to Hoechst Celanese by the Agency as a whole. 23 Even though the court today concludes that the multiple counting interpretation is a “permissible” reading of the benzene exemption, that ruling does not reach back in time to establish that multiple counting was the Agency’s consistently held interpretation in 1989, 1990, or even in 1992, by which tune the Celriver Plant was meeting all the terms of the benzene NESHAP.

In any case, this court need not decide whether (and, if so, when) Hoechst Celanese was given authoritative actual notice of the Agency’s present interpretation. For even if Hoechst Celanese received actual notice, it is undisputed that Hoechst Celanese acted responsibly to meet — and far exceed — the terms of the NESHAP in advance of EPA-accepted deadlines. In fact, by spending an additional $8 million, it did so within the two-year compliance period that was available to qualifying sources under the waiver provisions in the NESHAP.

It is also undisputed that EPA never objected to the compliance schedule submitted by Hoechst Celanese in April 1990. This is significant because had EPA at any pot wished to compel Hoechst Celanese to meet the terms of the benzene NESHAP more This is significant because had EPA at any point expeditiously, it had options at its disposal that it never exercised. 24 In such cir cumstanees, where Hoechst Celanese never conceded that it was subject to the NESHAP but nonetheless agreed to implement the requirements of the rule and where EPA never objected to — or took steps to shorten — the compliance schedule that Hoechst Celanese submitted, fundamental considerations of fairness argue against a finding of liability for that time period.

Underscoring this conclusion is the fact that the Clean Air Act did not contemplate immediate compliance with the benzene NESHAP. Section 112(c) of the Act — as it read at the time of promulgation of the benzene NESHAP — specified that those subject to a hazardous pollutant emission standard had 90 days from the effective date of the rule to come into compliance. This 90-day period, though, could be extended for up to an additional 21 months (a total of two years) upon satisfaction of the standards for a waiver. EPA’s benzene NESHAP contains provisions mirroring the statute’s two-year waiver requirement. See 40 C.F.R. § 61.11.

In summary, had the benzene exemption provided fair notice that the Celriver Plant was not exempt when the rule was first promulgated, the Company would have been able to apply for and likely obtain a waiver for Celriver — as it did for its Pampa Plant— giving it up to two years in which to come into compliance. And, the facts show, the plant would have been able to come into compliance within a two-year period. It was not the Company’s fault that the opportunity to obtain a waiver may have passed by the time Hoechst Celanese was finally given actual notice by EPA that the Celriver Plant was not exempt. 25 The Company’s efforts to meet the terms of the NESHAP should be evaluated in light of the additional time for compliance to which it would likely have been entitled in 1984.

In light of the above facts, due process precludes a finding of liability in this ease. As one of EPA’s own administrative law judges aptly put it:

In order to continue to enjoy the confidence of the regulated community, the public, and the courts in the conduct of its enforcement activities, the government must “occasionally bear the consequences of unclear wording in the extensive and detailed implementing regulations for its statutory responsibilities. This should be regarded as a small price to pay for the maintenance of credibility and public trust, and for a reputation of fairness in dealing with the regulated community.”

In Re Phibro, 1994 CAA LEXIS 18 (Oct. 5, 1994) (Ex. 110), quoting In the Matter of K.O, Mfg., Inc., Docket No. EPCRA-VTI-89T-611 at 17 (Feb. 28,1998).

Conclusion

For all of the foregoing reasons, the court hereby determines that the EPA’s interpretation of 40 C.F.R. § 61.110(c)(2) is permissible and that interpretation is hereby sustained by the court. The government’s motion for summary judgment is granted to this extent only. The court also grants, in part, the motion for summary judgment by Hoechst Celanese Corporation, awarding no civil penalty to the plaintiff because Hoechst Celanese did not receive fair notice of the EPA interpretation during the time period it was using benzene at the Celriver Plant. Each side shall bear its own costs.

IT IS SO ORDERED.

1 . The government disagrees with the use of the words "multiple count” in describing its interpretation of the regulations. Whether the government's gloss on the regulation is denominated "multiple count,” "internal recycle,” or something else is not significant for purposes of this decision.

2 . The facts in this section describing the Celriver Plant’s design and use of benzene were derived from U.S. Mem. at 27-31, United States' Response to the Defendant's Motion for Summary Judgment and the Defendant’s Opposition to the United States’ Motion for Partial Summary Judgment ("U.S.Resp.”) at 3-6, and Hoechst Celanese's Initial Brief at 26. The court’s description of benzene use at Celriver focuses specifically on one of two process units at the plant, the acetic anhydride manufacturing process. A full description of this process is too technically complex to warrant its inclusion here, but the parties have provided a useful, if greatly simplified, depiction of certain of the most pertinent features of that process. The court notes that benzene was also used in another process unit at the Celriver Plant, the acid recovery unit. The Government does not discuss how benzene use would be calculated in this other process unit.

3 . Clean Air Act, sec. 4(a), §§ 111(b)(1)(A) and 112(b)(1), 84 Stat. 1676, 1684-85 (1970), sec. 110, § 112(e), 91 Stat. 685, 703 (1977), and Sec. 13(b), § 112(e)(5), 92 Stat. 3443, 3458 (1978) (current version at 42 U.S.C. §§ 7411-12 (Supp. V 1993)).

4 . Hoechst Celanese’s Initial Brief at 53 quoting Webster's Third New International Dictionary, p. 2523-24 (1976).

5 . EPA’s description of this cost and emission reduction analysis appears in U.S. Mem. at Appendix B. The court notes that, while the PES Memo was part of the rulemaking record, it was not itself published in the Federal Register, nor were its contents fully explicated in the BID. Rather, interested persons had to be able to identify the PES memo as being relevant to the benzene exemption and could then obtain a copy only by writing to EPA and requesting it.

6 . A "mass balance” is a technique based on the scientific principle that matter can never be destroyed. To apply a mass balance, one accounts for all mass or material entering, leaving, and remaining in a defined area. Typically, the accounting of a material is expressed in a mathematical equation. When all material is accounted for, the equation "balances." While both EPA and Hoechst Celanese agree with this general definition of the term "mass balance,” they disagree profoundly over the scope of the "defined area" on which this mass balance is to be performed under the exemption. At Celriver, EPA instructed Hoechst Celanese to provide benzene usage figures derived from mass balance calculations on individual process streams. The Company contends that both the plain language of the exemption and rulemaking record indicate that the mass balance is to be performed on the “plant site”' as a whole. Such a calculation, Hoechst Celanese maintains, is specifically intended to eliminate any multiple counting resulting from internal flow. For most facilities that use benzene in their manufacturing processes, it would generally result in the measurement of a plant’s consumption.

7 . Elsewhere in this same communication, the Agency observed that it "may be time-consuming to calculate actual annual benzene usage as interpreted by EPA,” and indicated that the Agency "would be willing to review any estimation methods [Hoechst] Celanese might use in formulating a response” to the Agency’s request. “Any response that relies that on estimation methods," EPA continued, "should contain enough backup information to allow us to evaluate the method used and the results obtained with it.”

8 . The Benzene Leaks NESHAP also exempts coke byproducts plants. 40 C.F.R. § 61.110(b).

9 . The definition of the verb "apply” includes, "To ask or seek aid, employment, or admission." Webster's II New Riverside University Dictionary.

10 . The government has not set forth any evidence which indicates that EPA told Mr. Gomez or anyone else at the TACB that "designed to use" was to be interpreted consistent with the multiple counting approach.

11 . In addition to these two Texas plants, Hoechst Celanese also operated the Celco Plant in Virginia. In 1984, Virginia’s State Air Pollution Control Board confirmed that the Celco Plant was exempt from the NESHAP based on a single counting of benzene at the plant site.

12 . During the past 12 years, the government has provided several different, inconsistent articulations and applications of a multiple counting approach. Hoechst Celanese argues that none of those is a cogent explanation of how to apply a multiple counting approach in a manner that will consistently provide the same result at the same plant. In fact, although the government has calculated Celriver's "use” of benzene based on the benzene flow through two specific internal pieces of equipment at that plant, it is not clear why the government chose those particular pieces over the many other flanges, valves, and other pieces of equipment in benzene service at Celriver.

13 . The government contends that Hoechst Celanese is in actuality asserting a defense of estoppel. It is well-settled that estoppel generally does not run against the United States. Office of Personnel Management v. Richmond, 496 U.S. 414, 110 S.Ct. 2465, 110 L.Ed.2d 387 (1990). This principle is grounded on the separation of powers doctrine. United State v. Marine Shale Processors, 81 F.3d 1329 (5th Cir.1996). ("Judicial adoption of estoppel based on agency misrepresentation would ... vest authority in [government] agents that Congress would be powerless to constrain.”) “Fair notice” on the other hand, is a principle rooted in the due process clause of the United States Constitution. See Mullane v. Central Hanover Bank & Trust Co., 339 U.S. 306, 314, 70 S.Ct. 652, 657, 94 L.Ed. 865 (1950) (The due process clause "prevents deference from validating the application of a regulation that fails to give fair warning of the conduct it prohibits or requires. ”). In determining that Hoechst Celanese should not be sanctioned in this case, the court relies on the fair notice doctrine, not principles of estoppel.

14 . The record includes the rule itself, the Preamble, and documents referred to and incorporated into the Preamble as part of the Administrator's further description and interpretation of the regulation. See 5 U.S.C. § 551(4) (definition of "rule” in the APA); 44 U.S.C. § 1507(4) (contents of the Federal Register “shall be judicially noticed”). Never having been superseded by a subsequent rulemaking, the Preamble is as much a current statement of what the regulation means today as it was when the benzene NESHAP was published on June 6, 1984. It provides evidence of the Administrator’s intent. See State of Vermont v. Thomas, 850 F.2d 99, 103 (2nd Cir. 1988).

15 . According to Hoeehst Celanese, when EPA was developing its benzene NESHAP, refineries and feedstock users of benzene produced or consumed millions of megagrams of benzene annually. In contrast, those that recycled benzene as a solvent or quenching agent — as Hoeehst Celanese's Celriver Plant did — collectively used or consumed a total of only 1,500 megagrams annually.

16 . Hoeehst Celanese points out that at least some of that confusion was generated by EPA staff's issuance of guidance documents that contradicted the Administrator’s clear language in the benzene NESHAP Preamble (calling for single counting) and tried to develop alternative approaches.

17 . The government attaches significant weight to a handwritten phrase at the top of Hoeehst Celanese’s copy of the Texaco letter which stated "Read it and Weep". The words were written by Mr. Garza and the memo was sent to Jim Mullins, Garza's counterpart at Hoeehst Celanese’s Pampa Plant. The government maintains that this comment is affirmative proof that Mr. Garza knew that EPA had adopted the multiple counting approach. After reviewing this letter as a whole, the court declines to make this bold presumption. Rather, the court finds that the most plausible explanation for this comment was that Mr. Garza read this letter to mean that the EPA might be reading the exemption as calling for an evaluation of the overall quantity of benzene used at a plant — that is, the plant's consumption of benzene plus the amount of benzene in service at the plant. Mr. Garza testified that he believed this approach would not affect his plant, but would affect Mr. Mullins' Pampa Plant and thus, would require his colleague to work harder.

18 . Hoeehst Celanese points out that according to the testimony of both the author of the EPA Region VI letter and the recipient of that letter, neither read it as saying that “use” is to be determined based on a multiple counting of benzene at a plant site.

19 . As noted above, EPA had delegated to the TACB the authority to address these issues. The TACB did so, continually informing EPA what it was doing. There is nothing in the record to indicate that EPA at any time advised the TACB — or Hoechst Celanese — that the TACB's actions were improper as to either the Bishop or Pampa Plants.

20 . TACB had before it information reflecting not just consumption, but the "overall quantity of benzene” physically available to be used in one year — the amount on site at the beginning of the year plus all benzene added during the year to make up for consumption.

21 . By August 1991, the Company believed that it had implemented all the terms of the NESHAP except for one pressure relief device, which Hoechst Celanese apparently did not begin operating in conformity with the NESHAP until the following month. Following that time, as explained by counsel for Hoechst Celanese during oral argument, Hoechst Celanese conducted a voluntary environmental audit at Celriver to determine if all plant equipment in benzene service was meeting the terms of the NESHAP. As a result of that self-audit, Hoechst Celanese discovered a single valve (out of a total of 174) that was not meeting the terms of the NESHAP; the Company replaced that valve by August 1992. The court concludes that Hoechst Celanese's treat ment of this single valve does not warrant a finding of liability. Whatever the explanation, even this task was completed before the December 1992 deadline specified in the compliance schedule which Hoechst Celanese submitted to EPA.

22 . On June 26, 1989, EPA Region IV concluded that the U.S. Department of Energy’s Savannah River facility (which recirculated benzene) would be exempt based on a its " ‘design’ benzene consumption rate.”

23 . Hoechst Celanese has pointed out that just one week after EPA Region IV issued its NOV for the Celriver Plant, EPA gave another company a conflicting single count interpretation of the exemption. A February 28, 1990 EPA Region VI inspection report for the Howell Hydrocarbons facility (which recycled benzene in a manner similar to that at Celriver) indicated that facility was exempt from the benzene NESHAP because less than 1,000 megagrams per year of benzene was "purchased as make-up material.”

24 .For instance, EPA could have issued an administrative order pursuant to § 113(a) of the Clean Air Act or the Agency could have pursued a civil enforcement action for an injunction under § 113(b).

25 . During oral argument, counsel for the government disputed whether the Celriver Plant would have qualified for a waiver, suggesting that waivers were not available for coming into compliance with leak detection and equipment tagging requirements. This ignores that Hoechst Celanese submitted to EPA, in April 1990, a comprehensive and coordinated plan for going beyond the terms of the benzene NESHAP and phasing out the use of benzene altogether in one of two process units at Celriver. Even if the court were to assume that extensions of time in order to comply solely with the leak detection and equipment tagging requirements were not generally contemplated under the waiver provision, the court is unwilling to conclude that the Agency would have been without discretion to grant Hoechst Celanese a waiver for the time necessary for the Company to implement its overall compliance plan, which included the additional environmental benefit of eliminating the plant’s use of benzene in one process. Cf. Monsanto Co. v. EPA, 19 F.3d 1201, 1207 (7th Cir.1994)(The Clean Air Act’s “waiver’ provision does not require the source to install the controls that will achieve compliance at the earliest possible date,” and if a company "has a choice between two control strategies, the EPA has authority to grant a waiver for a pollution prevention strategy even if that strategy would take slightly longer to implement than a less desirable strategy.").