United States v. Murphy Oil USA, Inc. (155 F. Supp. 2d 1117, 2001)

OPINION AND ORDER

CRABB, Chief Judge.

Plaintiff United States of America is suing defendant Murphy Oil USA, Inc. for alleged violations of environmental laws. Primarily, plaintiff alleges that defendant made major modifications to the sulfur recovery unit at its Superior, Wisconsin oil refinery without obtaining permits required under the Clean Air Act, 42 U.S.C. §§ 7401-7671q, without complying with performance standards applicable to the work and without employing the best available control technology. Defendant denies any failure to comply. It maintains that all of the improvements it made to its sulfur recovery unit were motivated by the need to meet state and federal air quality standards and were undertaken in close cooperation with the Wisconsin Department of Natural Resources, the authority responsible for issuing Clean Air Act permits in Wisconsin. Plaintiff does not dispute defendant’s assertion that its projects were intended to improve its pollution control efficiency or that defendant followed the state’s directives. Rather, plaintiff contends that defendant avoided compliance by withholding critical information from regulators that would have indicated that its proposed modifications would trigger application of various regulations and permit requirements. Plaintiff is suing for the withholding of the information as well as for the penalties and injunctive relief available upon a showing that an operator modified a regulated pollution source without compliance with the provisions of the act. Besides the alleged Clean Air Act violations, plaintiff has alleged that certain acts and omissions of defendant violated various provisions of the Clean Water Act, 33 U.S.C. §§ 1251-1387, and the Resource Conservation and Recovery Act, 42 U.S.C. §§ 6021-6039e.

Before trial, the case was pared down slightly as a result of motions for partial summary judgment that led to findings in favor of plaintiff on some of its claims and in favor of defendant on one of its affirmative defenses, see Opin. and Order entered on May 18, 2001. It was trimmed further after plaintiff withdrew some of the 24 claims it had alleged originally, including all of the claims it had raised under the Emergency Planning and Community Right-to-Know Act, 42 U.S.C. §§ 6991-6991h. Trial to the court proceeded on the remaining claims and affirmative defenses June 7-16, 2001.

The bulk of trial time was devoted to plaintiffs four remaining Clean Air Act claims. The first is that defendant made major modifications to the sulfur recovery unit at its oil refinery in 1987-88 and in 1991-93 and that each project resulted in a net emissions increase of sulfur dioxide of more than 40 tons a year, obligating defendant to obtain a Prevention of Significant Deterioration permit. Plaintiff seeks an order prohibiting defendant from continuing to operate the sulfur recovery unit without obtaining a Prevention of Significant Deterioration permit and a civil penalty for making modifications to the sulfur recovery unit without the required permit.

In claim two, plaintiff seeks to enjoin defendant from operating its sulfur recovery unit without using best available control technology and to collect a civil penalty from defendant for making modifications to the unit without complying with this requirement of the Clean Air Act.

Plaintiffs third claim is that defendant did not provide all the relevant information plaintiff needed in order to determine whether the modifications defendant proposed to make in 1993 (routing its No. 2 distillate unifier into the sulfur recovery unit) would result in a significant increase in emissions from the sulfur recovery unit. Defendant admits that it withheld certain consultant reports but denies that anything in these reports was relevant, in the sense that it would have affected the permitting authority’s decision making. If defendant is correct, plaintiff will be prevented by operation of the applicable statute of limitations, 28 U.S.C. § 2462, from pursuing its demand for penalties against defendant on the first and second claim as they relate to the modifications that defendant made to the distillate unifier in 1992-93 because I have held that the statute of limitations may be tolled only if plaintiff proves that defendant made affirmative efforts to withhold relevant documents, that is, that defendant withheld documents knowingly and intentionally. (Plaintiff does not argue that defendant withheld relevant information in connection with the modifications it made before 1992; it concedes that the statute of limitations bars it from seeking penalties against defendant for making those modifications without obtaining a Prevention of Significant Deterioration permit.)

Plaintiffs third claim involves information provided by engineering consultants defendant hired at various times to suggest ways of making the sulfur recovery plant operate more efficiently. The information at issue includes the following: 1) reports prepared in 1987 by Sulfur Operations Support, a consulting firm, that include the consultants’ statement about the efficiency of the sulfur recovery unit before construction began in 1987, projected rates of efficiency after construction and references to the need for a larger combustion chamber; 2) reports from a firm known as Western Research that describe four stack tests and a material balance or performance test the firm conducted during 1989; and 3) reports from Becker, Losier & Associates that were prepared in 1989 and 1992 and discuss estimated recovery efficiency. Plaintiff contends that in addition to the consultants’ reports, defendant should have given the permitting authority information relating to a questionable stack test. The withholding of information claim can be reduced to two questions: whether the reports contain information that would have affected the permitting authority’s decision making and whether defendant would have had reason to know that it should have submitted the reports.

Plaintiffs fourth claim is based on its contention that since the completion of defendant’s modifications in the early 1990’s, the sulfur recovery unit has had the capacity to process more than 20 long tons a day of sulfur, that is, the unit’s “throughput” capacity has exceeded 20 long tons a day of sulfur. If plaintiff is correct, the unit has been subject to New Source Performance Standards, which would require defendant both to limit its sulfur dioxide emission concentration from the sulfur recovery unit to the standard set out in 40 C.F.R. § 60.104(a)(2) and to install an emissions control system to limit the sulfur dioxide concentration. Defendant denies that its sulfur recovery unit can process more than 20 long tons a day of sulfur and contends that even if the unit has done so on a few occasions, it was never designed to do so and cannot do so safely without risk of harm to its component parts. Additionally, defendant denies that the 20 long tons a day limit is measured by input (throughput) and not by output (production).

I conclude that defendant failed to submit to the Department of Natural Resources the information relevant to the department’s determination whether defendant’s proposed modifications of the No. 2 distillate unifier would require a Prevention of Significant Deterioration permit, compliance with New Source Performance Standards and use of best available control technology. If defendant had submitted the information, the department would have known that defendant did not qualify for a synthetic minor permit but needed a Prevention of Significant Deterioration permit and was required to comply with the New Source Performance Standards and best available technology. I conclude also that defendant knew that the withheld information would have been relevant to the department’s permitting decision and should have been disclosed. The failure to disclose this information tolls the running of the statute of the limitations, 28 U.S.C. § 2462. I conclude also that defendant’s sulfur recovery unit was designed to process at least 20 long tons a day of sulfur.

I conclude that plaintiff is entitled to judgment on its claims ten, eleven, twelve and thirteen, brought under the Clean Water Act, but that it has failed to prove its Resource Conservation and Recovery Act claim. A determination of the relief to which plaintiff is entitled will be made after the second phase of trial, which will be devoted to penalties and injunctive relief.

From the evidence adduced at trial, I make the following findings of fact.

FACTS

I. CLEAN AIR ACT

A. Background

Plaintiff United States of America brings this suit on behalf of the United States Environmental Protection Agency. (As in the earlier order, I will use “plaintiff’ to refer to the EPA.) At all relevant times, the United States has delegated to the state of Wisconsin the authority to issue Prevention of Significant Deterioration permits and to determine the applicability of New Source Performance Standards.

Defendant is a Delaware corporation that operates an oil refinery in Superior, Wisconsin. It takes crude oil from the Lakehead Pipeline and refines it into several petroleum derivatives, including at least six types of crude oil with varying sulfur content.

Although defendant’s refinery includes many different processes, the only one at issue for the purpose of the alleged Clean Air Act violations is the sulfur recovery unit, which is designed to treat the hydrogen sulfide that is a byproduct of the refinery process and arrives at the sulfur recovery unit in pipes from the various processing units throughout the refinery. Amine acid gas and sour water stripper gas are fed through separate piping into the sulfur recovery unit, which converts the hydrogen sulfide in the gases into elemental sulfur and sulfur dioxide through a process known as the Claus reaction. (Before defendant made the 1991-93 modifications to the sulfur recovery unit, it did not route the sour water stripper gas through the sulfur recovery unit but sent it directly to the incinerator without treatment.) Typically, two-stage Claus burners convert 90 to 95% of the hydrogen sulfur in waste gases into elemental sulfur. They do so by passing the gases over a catalyst at an elevated temperature and pressure in the presence of hydrogen. The resulting hydrogen sulfide is then sent to the sulfur recovery unit for conversion into elemental sulfur, with the unconverted hydrogen sulfide going to a tail gas incinerator that combusts the hydrogen sulfide, in the presence of oxygen to form sulfur dioxide. The elemental sulfur is routed to a sulfur pit as a product and sold commercially.

B. Defendant’s Sulfur Recovery Unit— 1987-88 Modifications

Sulfur recovery unit operators have a significant incentive to achieve highly efficient recoveries of elemental sulfur not only to increase the income from the sale of the product but to avoid Clean Air Act penalties for excessive pollution emission. Despite this incentive, defendant had difficulty operating its sulfur recovery unit from the time it was purchased in 1973 through the early 1990s. The unit is a small one built originally for a gas plant and not for an oil refinery. Up through the 1980s, it was often off line and had frequent mechanical failures. In 1986-87, defendant employed Sulfur Operations Support to analyze the unit’s problems and provide suggestions for design changes. The firm provided defendant with a number of reports in letter form. In a letter that appears to have been sent in the fall of 1986 (Exh. # 436), Sulfur Operations advised defendant that the sulfur recovery plant had an 85 to 90% recovery rate in the conditions under which it operated but defendant could expect to achieve a 91-93% rate if it could bring the plant’s conditions up to standard. In a January 20, 1987 letter (Exh. # 431 at 7), Sulfur Operations projected various recovery rates under different conditions, all of which would produce recovery rates of 89.5% or higher. In a February 13, 1987 letter (Exh. # 434 at 4), the firm noted that the existing combustion chamber was too small. In an April 10, 1987 letter, it calculated the rate of sulfur recovery from each of two design proposals it had made, showing the lower rate as 91.5% and the higher as 92.3%. In June 1987, defendant added a tail gas analyzer to the sulfur recovery unit.

In a February 1988 response to a notice of violation of the state’s Statewide Sulfur Dioxide Rule, defendant advised the Department of Natural Resources that the modifications it would be making to the sulfur recovery unit would enable it to operate the unit in July 1988 and to demonstrate full compliance with the proposed alternate emissions limits no later than September 1988. When the sulfur recovery unit and associated amine system were not operating, sour gases bypassed the sulfur recovery unit and were burned directly without pollution control,, causing an increase in sulfur dioxide emissions. In 1988, 1990 and 1991, there were times that defendant’s emissions exceeded the legal limits under the Wisconsin Administrative Code.

On April 8, 1988, defendant met with Department of Natural Resources staff to present a proposal and schedule to undertake various enhancements to the sulfur recovery unit to reduce emissions and bring the refinery into compliance with state sulfur dioxide rules. The department found the proposal acceptable. The changes to the sulfur recovery unit that defendant was planning to make included upgrading the primary burner, replacing the catalyst, installing a new and larger combustion chamber, adding an acid gas knock-out drum and a tail gas analyzer, a hydrocarbon coalescer in the upstream amine unit and a distributed control system. The changes it told the department about did not include increasing the size of the new combustion chamber, although it did say that the new chamber would provide inereaséd residence time. Defendant expected the changes to increase the reliability of the unit and improve its sulfur recovery efficiency. Before defendant made these modifications to the sulfur recovery unit, it never applied for a permit for them from the Department of Natural Resources, which administers the Clean Air Act in Wisconsin, or asked for an exemption from permitting requirements.

The capacity of a sulfur recovery unit to process gas streams and derive elemental sulfur is measured in long tons of sulfur per day. According to reports filed with the Bureau of Mines, defendant was producing an average of 142.5 long tons a month of sulfur during the two years preceding the start of the 1987-1988 modifications, that is, two years before the installation of the tail gas analyzer in June 1987.

In an October 17, 1986 letter to the Department of Natural Resources, James Gesick, manager of the Superior refinery, reported the results of two 1986 stack tests showing a 75-80% sulfur recovery efficiency. Exh. # 268. In September 1988, the department began enforcing a 393 pound an hour limit of sulfur emissions. Before June 1987, defendant’s air emissions inventory showed an average recovery rate of 76%.

Shortly before the start of trial in this case, Steven Dunn, an engineer with the Department of Natural Resources, made a nunc pro tunc calculation of defendant’s actual emissions during the two years preceding the installation of the tail gas analyzer using an 80% recovery rate and the reported monthly average sulfur production rate of 142.5 long tons a month. Dunn derived an actual emission rate of 957.6 tons of sulfur a year. Using the 76% figure shown by the air emissions inventory, he derived an actual annual emissions rate of 1,210. He calculated the sulfur recovery unit’s potential to emit after the planned changes were made as being 393 pounds an hour because of the state emissions limit that went into effect in 1988 and arrived at an annual rate of emissions of 1721, or an increase of 763.7 tons over the actual emissions calculated at an 80% rate and 511 tons over the 1,210 tons produced by using a 76% recovery rate. (In responses to requests for admission defendant filed on February 12, 2001, it agreed that no applicable federal emissions limitation was in effect in 1988 and that the state was enforcing a 393 pound an hour restriction.)

In a July 7, 1986 memorandum from plaintiffs Director of the Office of Air Quality Planning and Standards, the director stated that exemptions available under the New Source Performance Standards that applied to systems or devices with the primary function of the reduction of air pollution did not apply automatically under the Prevention of Significant Deterioration program. Exh. # 3039. The director concluded by saying that he was withdrawing an earlier memorandum on the topic (Exh. #3013) in which he had held that for Prevention of Significant Deterioration purposes, the term modification would include all the exemptions included in the New Source Performance Standards regulations.

In 1989, defendant retained the Western Research consulting firm to conduct a comprehensive full performance test on the sulfur recovery unit, which continued to have problems that required shut downs and produced excessive emissions of sulfur dioxide. In a letter dated September 1, 1989, the firm wrote Mark Miller, process engineer for the Superior refinery, confirming its agreement to consult and adding its' understanding that defendant’s goals were “a complete performance test of the sulfur recovery unit, an evaluation of current plant operating practices, and procedures and sulfur plant capability studies for possible future addition of distillate hydrotreating to the refinery.” Exh. #426. On September 19, 1989 and from October 24-26, 1989, Western Research conducted a full performance test of the sulfur recovery unit to evaluate its recovery performance under normal operating conditions and to help the refinery meet licensed emission guidelines during the compliance stack test scheduled for late October. Western Research reported a measured conversion efficiency from its October test of 96.04% of the total inlet sulfur, which it noted was consistent with the rated conversion efficiencies of 95.4% for the unit and which “displayed a large improvement over the September results [when the] estimated conversion efficiencies [were] between 93.3 percent and 95.1 percent.” Exh. # 427 at Bates No. MOII00495. The firm attributed the increased efficiency to operational changes carried out before the October test, id., but noted that during its October test, the unit was operating without the Air Demand Analyzer on line, which resulted in losses because of off-ratio conditions. Id. at Bates No. MOII00504. (The analyzer monitors the ratio of hydrogen sulfide to sulfur dioxide and adjusts the air flow automatically to keep the ratio at 2 to 1 or slightly higher for maximum efficiency of sulfur recovery. When the analyzer is not working, the operators must monitor the ratios and make the corresponding adjustments manually.) In addition, the final condenser coalescer was operating at a higher than desired temperature, leading to sulfur vapor in the tail gas.

Also, Western Research reported that during the September testing the sulfur recovery unit had demonstrated poor conversion efficiencies that Western Research attributed to high excess air values and possible diethylamine leakage into the condensers, possibly contaminating the converter beds.

The performance tests included a material balance (an analysis of the pertinent gas stream compositions, measurement of key process stream flows and the monitoring of the process temperatures and pressures). Id. at Bates No. MOII00499. A material balance can provide data acceptable for determining baseline emissions for permitting purposes. As part of the report, Western Research had a graph of net recovery efficiency (percent) and modified-Claus process practicable capability. Dotted lines and the words “Murphy Oil— Superior SRU” showed a net recovery efficiency for defendant’s sulfur recovery unit at above 96%. Id. at Bates No. MOII00509 (reproduced at Appendix A). This graph was submitted to the Department of Natural Resources as part of the No. 2 distillate unifier application, independently of the Western Research report and with the words “Murphy Oil — Superior SRU” deleted. Defendant did not supply the department with the graph in the form it was prepared by Western Research. It did not turn over any other portion of the Western Research report in 1992 or 1993 or even in 1999, when plaintiff was investigating this matter, although it turned over other previously withheld consultant reports in June 1999 and had the Western Research report in its possession before then. The report was discovered in December 1999. In addition, the Western Research report included a material balance of the sulfur recovery unit, summarizing the components going into and out of the unit.

Also in 1989, defendant retained Becker, Losier & Associates to propose possible improvements to the sulfur recovery unit. In a report dated September 28, 1989, Walter Losier noted the problems with the September 13, 1989 stack test that made the results unreliable and made recommendations to defendant for changes to solve the problems.

Plaintiffs policy is to give independent treatment to “major stationary sources” within another “major stationary source” for Prevention of Significant Deterioration purposes. Using a “nested facility” concept, it does not allow operators to hide a major stationary source within another one.

C. 1991-1993 Modifications

Defendant knew that it would have to increase the capacity of the sulfur recovery unit to produce low sulfur diesel fuel to conform with new environmental regulations. As of 1990, its unit’s design capacity was 15.1 long tons a day, Exh. #460; defendant wanted to be able to process an extra four long tons a day. Defendant retained Becker, Losier again in 1990, this time to study possible improvements in the unit to increase its processing capacity. In turn, Becker, Losier retained E & L Engineering to prepare a comparison study of the existing design of the sulfur recovery unit and a future design of revisions that would add a new 9.0 pound per square inch gauge blower and deeper sulfur seal legs, allowing an increase in acid gas flow equal to 16.0 long tons of sulfur a day. (Sulfur seal legs are devices used to balance the pressure of the process to the atmospheric pressure outside.) In April 1990, defendant asked E & L to add a new design study “to determine the unit capacity bottlenecks when additional acid gas is brought into the unit.” Exh. # 282.

On May 21, 1990, David Petty, an employee of defendant, wrote to the Department of Natural Resources, asking whether two proposed alternatives for modification of the existing sulfur recovery unit would trigger application of the New Source Performance Standards. He told the department that one of the options defendant was considering was simply “to replace the existing heat recovery unit and sulfur condenser with updated equipment of the same design but with different metallurgy.” Exh. # 284 at 2. Petty described the second option, which was to separate the amine and sulfur recovery systems, using the steam generated from the cooling water on a new heat recovery unit and sulfur condenser to heat the amine in a new, separate reboil-er. Id. Petty noted that the sulfur recovery unit had a nominal design capacity of 14 long tons a day and that ordinarily, the New Source Performance Standards do not address Claus units smaller than 20 long tons a day. He stated that neither of the options defendant was considering would “increase throughput capacity or emissions and that neither would constitute a reconstruction (because the anticipated cost of less than $1.5 million would not exceed 50% of the projected replacement price of $3.5 to 4 million).” Id. Petty did not tell the department that in fact, defendant was considering two designs that showed an increased gas flow rate of 15.7 long tons a day of sulfur in the feed gas with a new 8.0 pounds per square inch blower and an increased sulfur feed rate of 19.7 long tons a day with an 11 pounds per square inch blower, increased heat exchange surface area, 20 foot deep seals and larger gas pipes, as shown in Becker, Losier’s May 7, 1990 memorandum, Exh. # 460.

The department advised defendant that neither of the alternative plans for modification described by Petty would trigger New Source Performance Standards or Prevention of Significant Deterioration requirements or require a new source air pollution control permit. Exh. # 135. It asked defendant to submit a plan and schedule for “design, construction and operation of the unit” by July 30, 1990. Id.

Sometime before July 26, 1990, defendant asked E & L to base a design on two new blowers it had purchased: one of 11.5 pounds per square inch gauge and the other of 9.0 pounds. Defendant asked E & L to provide 17 to 18 foot sulfur seal legs to accommodate the larger blowers. On July 30, 1990, James Gesick, manager of the refinery, wrote to the Department of Natural Resources, saying that defendant had agreed to undertake the 1991 sulfur recovery unit improvements and that it had elected the second option, which involved separating the amine and sulfur recovery systems. (Gesick listed the two alternatives as 1) “Replace the existing heat recovery unit and sulfur condenser on a one-for-one basis with the same equipment”; and 2) “Separate the amine and sulfur recovery systems and provide a separate new reboiler.” Exh. # 3158.) Ges-ick provided a schedule for completion and noted that defendant was undertaking the work on the understanding that it would not trigger either New Source Performance Standards or Prevention of Significant Deterioration requirements. Id. Defendant never advised the Department of Natural Resources of the exact nature of the actual modifications it was making. It did not tell the department that it was installing new and more powerful blowers that would increase the unit’s capacity. It never submitted the reports from Becker, Losier & Associates with the reports from Becker, Losier’s subcontractor, E & L, showing the scope of the planned changes and the installation of the larger blowers, or the May 15, 1992 letter from the firm advising defendant that its sulfur recovery unit was operating at a 95 to 96% recovery rate and that this represented an improvement over the 90% to 92% efficiency rate of the unit before the hot gas by-pass was eliminated. Had defendant submitted these materials, they would have affected the decision making of the department.

During a September 21, 1990 conference on an unrelated matter, defendant told plaintiff that it was constructing a modification to its sulfur recovery unit under the guidance of the Department of Natural Resources and that the project was not subject to New Source Performance Standards. Defendant sent plaintiff copies of the 1990 correspondence between it and the department concerning the modifications to be made to the sulfur recovery unit. (As noted, this correspondence made no reference to defendant’s specific changes and its plan to add the more powerful blowers and accompanying sulfur seal legs.) Plaintiff gave preliminary approval to the Department of Natural Resources’ position that defendant’s construction of its second option was not subject to New Source Performance Standards and did not require construction or operation permits. Defendant began construction and provided both plaintiff and the Department of Natural Resources with periodic progress reports.

Defendant modified the sulfur recovery unit in 1991 by replacing the 9 foot sulfur seal legs with deeper 18 foot legs; installing a new and larger sulfur pit; installing two new air combustion blowers, one of 9.0 pounds per square inch and one of 11.5 pounds per square inch; replacing and increasing the heat exchange surface of the waste heat boiler; and enlarging the condenser surface area on the sulfur recovery unit.

As part of the 1991 project, defendant re-routed the sour gas generated by the sour water stripper to the sulfur recovery unit for pollution control purposes. Defendant never advised anyone in the permitting section of the Department of Natural Resources that it was re-routing the sour water stripper gas and thereby increasing the feed to the sulfur recovery unit. In a single-spaced, 3}4 page letter dated November 2, 1990, and written to Steven Dunn, a department engineer not working in the permitting section, Eder & Associates stated on defendant’s behalf that “[t]he waste gas incinerator stack number 15G-H1 does not comply with down wash minimization criteria. Emissions from this stack have not been tested but this source is being eliminated and emissions are being routed to the sulfur plant.” Exh. # 3178 at 1. Eder made no other mention of the re-routing in the November 1990 letter or in any other communication to the department. In April 1992, nearly two years after defendant had first sought approval of what was to be its 1991 modification project, Dan Rosenthal, the regional Department of Natural Resources compliance engineer, made an inspection visit to the refinery and was told about the rerouting.

In April 1992, defendant sought regulatory guidance from Rosenthal for a project involving improvements to defendant’s No. 2 distillate unifier that would enable defendant to produce diesel fuel that met the new sulfur limits of the Clean Air Act. Defendant wanted to increase the desul-furization capability of its unifier by installing a larger reactor. Rosenthal told defendant that the incremental feed to the sulfur recovery unit from the unifier would require an ah' permit application and that the project would be subject to Prevention of Significant Deterioration requirements, including air dispersion modeling and best available control technology analysis. Earlier, on March 19, 1992, Tom Graney, manager of engineering at the refinery, had sent a memorandum to refinery manager Ron Anderson asking about defendant’s obligations to the Department of Natural Resources in connection with the distillate unifier revamp. Graney wrote that the project would “increase plant S02 emissions due to the generation of additional feed to the sulfur recovery unit” of about 171 pounds an hour, which would produce an additional 816 pounds a day of sulfur dioxide discharged from the sulfur recovery unit incinerator stack, assuming a 90% recovery. Exh. # 358.

Defendant was eager to get the unifier project completed so that it would be able to start producing low sulfur diesel fuel by August 1, 1993. In a letter written to the department on July 6, 1992, Mark Miller, defendant’s process engineer, stated that the distillate unifier project would increase the hydrogen sulfide feed rate to the sulfur recovery unit and that there would be a concomitant increase in sulfur dioxide emissions. Exh. # 3291 at 1. He assured the department that the increased feed rate would be well within the design capacity of the sulfur recovery unit and illustrated the point with a table showing the then current daily maximum feed as 14 long tons a day and the estimated design maximum as 19 long tons. Id. at 1-2. He added that in his opinion the project was exempt as a “specified change in operation” because changes in sulfur removal at the No. 2 distillate unifier resulted in an increase in sulfur production that did not exceed the operating capacity of the sulfur recovery unit.

On July 14, 1992, defendant submitted a permit application seeking authorization to modify its No. 2 distillate unifier. It said nothing in the application about the fact it had increased the maximum sulfur input capacity of the sulfur recovery unit as part of the 1991 modifications it had made. In response to a request for guidance, plaintiff told the Department of Natural Resources that defendant would need a Prevention of Significant Deterioration permit unless it were to “net out” or restrict emissions through a “synthetic minor permit.” At the request of the Department of Natural Resources, defendant amended its application to obtain a “synthetic minor permit,” which is a means of avoiding the need for a Prevention of Significant Deterioration permit. To obtain such a permit, the applicant must meet certain criteria and agree to restrict emissions.

As part of the permit application, defendant had to show what its actual sulfur dioxide emissions were before construction began on the unifier and what the potential emissions would be after the project’s completion. Defendant chose to use a baseline period of August 1, 1989 to August 1, 1991, and a baseline recovery efficiency of 90%, taken from a stack test it had conducted on October 31,1989.

The Department of Natural Resources decided that the 1989-91 period should be changed to the 24-month period immediately preceding the No. 2 distillate unifier project. Consequently, defendant changed the baseline period to July 1990 to August 1992. This change resulted in lower pre-construction sulfur dioxide emission levels when defendant showed a 93% sulfur recovery efficiency for 10 of the 24 months. With lower pre-construction baseline emissions and the same post-construction emissions, the potential increase in emissions triggered Prevention of Significant Deterioration requirements.

On September 17, 1992, refinery manager Ron Anderson wrote the Department of Natural Resources, revising the No. 2 distillate unifier permit application and specifically revising the baseline recovery efficiency estimates by claiming a pre-project recovery efficiency of 86% for the 14-month time period before the 1991 sulfur recovery unit modification (July 1990 through August 1991), rather than the 90% recovery efficiency it had submitted for the entire time period of August 1, 1989 through August 1, 1991 in its original application for the distillate unifier work. Exh. # 313. To achieve the 86% efficiency rate, defendant averaged the results of two stack tests, the October 31, 1989 test that had been observed by the department and a September 13, 1989 stack test that had not been observed. According to defendant, the September 13, 1989 stack test showed an 82% efficiency recovery rate. Defendant did not tell the department that Western Research had conducted stack tests on September 19, 1989 and again on October 24-26, 1989, that the September 19 testing produced estimated overall conversion efficiencies ranging from 93.3% to 95.1%, that the October 24-26 tests showed a measured conversion efficiency of 96.04% or that during the unobserved September 13 test, the heat recovery unit tubes had been leaking diethylamine into the Claus side of the process. Had defendant given the Western Research report to the department, it would have affected the department’s decision on the permit application. Defendant urged the use of the September 13 test despite the fact that Mark Miller, defendant’s process engineer, had written to Tom Graney on May 29, 1992, saying that four performance tests had been run but that in his opinion, the sulfur recovery unit had been running properly only during the October 31, 1989 test. He had advised Graney that during the September 13 test, diethylamine had been leaking into the Claus side of the process and he had recommended bringing Western research back to run another test. Exh. # 361.

In the September 17 letter, Anderson listed the total sulfur shipped out of the plant each month for the two year period as the basis for setting baseline emissions. Exh. # 313. In a September 24, 1992 letter from Paul Yeung, permitting examiner, to Dan Rosenthal, the department’s compliance engineer in the Superior area, Exh. #204, Yeung wrote, “Both Murphy and the DNR have agreed that the basis for setting the baseline emission amount will be to use the total sulfur shipped out of the plant.”

Rosenthal opposed defendant’s efforts to use an 86% efficiency rate for any portion of the pre-construction period. In his opinion, the 82% figure was unreliable because it had been obtained from a stack test that the department had not observed and because his reading had convinced him that the normal range of recovery rates for sulfur recovery units is closer to the high 90s. Rosenthal argued his position vigorously with the department’s permitting authorities, but was unsuccessful in persuading them to use a higher recovery rate.

In November 1992, the Department of Natural Resources issued defendant Air Permit No. 92-POY-94, after determining that the modification involving the No. 2 distillate unifier would be exempt from Prevention of Significant Deterioration review because defendant had agreed to abide by synthetic minor permit limitations that restricted emissions from the refinery to levels below Prevention of Significant Deterioration applicability thresholds. The department concluded that the proposed change would be minor because the increase of sulfur dioxide emissions would be less than 40 tons a year.

Also in November 1992, Paul Yeung responded to a comment on defendant’s permit application from two residents of Superior. Among other things, he wrote that defendant’s “allowable sulfur dioxide emissions will be, for all practical purposes, the same as the sulfur dioxide emissions that [defendant] is currently emitting.” Exh. #154.

During the 30-day appeal period following issuance of the permit, plaintiffs inspectors toured defendant’s refinery to inspect aii’ pollution sources and determine compliance with all applicable air rules. Although this was not a full-fledged inspection but primarily an opportunity to become familiar with the air pollution sources in the region, plaintiffs staff discussed the sulfur recovery unit, the No. 2 distillate unifier project and the air permit. Plaintiff never filed an appeal of the permit or registered any objection to it.

During 1992, defendant constructed a new amine tower for separating hydrogen and hydrogen sulfide at the distillate unifier. Between 1992 and 1993, defendant increased the reactor size of the distillate unifier.

Lee Vail is Manager of Environmental Affairs for defendant. He has a Ph.D. in environmental engineering. In 1999, when plaintiff and the Department of Natural Resources were investigating this matter, Vail turned over the Sulfur Operations Support reports and two reports from Becker, Losier & Associates, without the reports by E & L. He did not turn over the Western Research report, although he had it in his possession from before June 1999, if not earlier. In correspondence with the department before the department discovered the Western Research report, Vail maintained that the department lacked any evidence to show that defendant had used an inaccurate stack test result as a basis for computing its pre-construction emissions.

D. Capacity of Sulfur Recovery Unit

In a letter dated September 11, 1990, Mark Miller wrote to McGill Environmental Systems, Inc., seeking information about the McGill reaction furnace installed in defendant’s sulfur recovery unit. Specifically, Miller wanted to know how the burner pressure drop would increase when the feed flow rate to the sulfur recovery unit was increased. Miller enclosed a table showing projected rates for its unit. Using the numbers in the table, Miller calculated the sulfur feed rate would be 22.7 long tons a day. Exh. 750. Miller believed that before 1991, the sulfur recovery unit had a maximum sulfur input rate of no more than 15 long tons a day; he projected a six or seven long ton a day increase with the changes to be made in 1991-1993.

In a letter dated December 22, 1995, Exh. #320, David Podratz, defendant’s manager of technical services, wrote to TP A, Inc., a consulting firm, about a new tail gas treating process. He informed TPA that

We have operated the existing [sulfur recovery] unit with sulfur production of up to 20 LTPD at recoveries of about 94%. I believe the ultimate capacity of the existing unit is about 22 LTPD. The new tail gas treating process should be designed so as not to limit our SRU capacity with recoveries of greater than 98%.

Defendant is required to provide to the United States Department of the Interior, United States Geological Survey information about monthly sulfur production, shipment and disposition. When sulfur production data for October 1996 are combined with defendant’s continuous emissions monitor data, the calculations show that the average sulfur input into the sulfur recovery unit for that month exceeded 20.27 long tons per day. Continuous emission monitoring data from 1994 through September 1998 indicate that there were 210 days with an input greater than 20 long days per ton of sulfur into the sulfur recovery unit, assuming a 95% recovery rate. (Assuming a 94% recovery rate, there were 14 days during the same time period when the input exceeded 20 long tons a day.)

Daily sulfur pit stick production data for 1991 through 1998 showed that the sulfur recovery unit produced over 20 long tons per day of sulfur on 148 days from 1991 through approximately September 1998. Sulfur pit stick measurements are prone to error; defendant measures the sulfur pit with nothing more sophisticated than a measurement tape with a bullet weight attached to a swivel at one end. The person doing the measurement cannot tell whether the weight is reaching the bottom of the pit and whether it is still upright or lying on its side because of the opacity of the sulfur. This uncertainty makes it easy to make a six inch mistake in measurement. However, the error can be either plus or minus actual sulfur levels. If the weight is lying on its side, the measurement will overstate the sulfur level; if, for some reason, the weight does not sink to the bottom of the pit, it will understate the sulfur level. Defendant relied on stick pit measurements in estimating its sulfur recovery efficiency rate for the stack tests performed on September 13, 1989 and October 31,1989. Exh. # 204.

With the changes made to the sulfur recovery unit in 1991-1993, the capacity of the unit to process sulfur increased to more than 20 long tons a day. The larger blowers and seal legs allowed the unit to be operated at higher pressure (up to about 11.5 psig), which means that the operator could “push” more sulfur through the sulfur recovery unit and increase the capacity. The lesser of the two blowers showed a capacity of 17.5 long tons a day; the larger showed a capacity of 21.7 long tons a day.

Brimstone Engineering Services did a performance evaluation of the sulfur recovery unit in July 1998. Included in its report was pressure drop information for the unit that both defendant’s expert, John Bourdon, and plaintiffs expert, Paul d’Haéne, relied upon in their calculations of maximum throughput capacity of the unit. Pressure drop information is necessary for knowing how much driving force is needed to push a certain amount of air through the piping system, given a known reaction furnace pressure. Maximum throughput in a sulfur recovery unit occurs at the point at which the pressure drop equals the available pressure on feed gases and on the air.

The Brimstone pressure drop information confirms that when the sour water gas feed is 0.6 long tons a day and the remainder is amine acid gas, the capacity of the sulfur recovery unit is up to 22.8 long tons a day if the control valves are wide open and not slightly pinched as they were in the Brimstone performance test. When the percentage of sour water gas feed increases to 1 to 1.3 long tons a day of sulfur, the sulfur recovery unit can process 21.7 long tons a day.

Defendant’s sulfur recovery unit has two control valves. The 6 inch valve does the bulk of the work and keeps the air coming in at the amount required by the assumed percentage of hydrogen sulfide in the feed. The smaher 2 inch valve acts as the trim controller, adjusting air flow in response to the air demand analyzer so as to maintain the optimum two parts hydrogen sulfide to one part sulfur dioxide.

When plaintiff published the new source performance standards of subpart J of 40 C.F.R. part 60 in 43 Fed.Reg. 10,868 (Mar. 15, 1978), it said that the standards of performance for new stationary sources referred to Claus sulfur recovery units that had sulfur production capacity in excess of 20 long tons a day. In the New Source Performance Standards subpart J, amended on October 25, 1979, plaintiff wrote that a sulfur recovery unit was subject to the standards if it had a processing capacity of more than 20 long tons a day. 44 Fed.Reg. 61,542 (Oct. 25,1979).

On October 22, 1992, John Rasnic, plaintiffs Director, Stationary Source Compliance Division, Office of Air Quality and Standards, issued a memorandum in which he wrote,

This memorandum amends the June clarification to make the 20 [long tons per day] exemption under [new source performance standards] Subpart J more practicable. The June 2, 1992 memorandum stated that the [long tons per day] applicability exemption refers to the amount of sulfur which a Claus plant is designed to produce. A definition of [long tons per day] based upon output would allow exemptions for inefficient, and thus low sulfur-producing facilities. Applicability based upon output would also apply differently to similar units depending on performance, and penalize efficient control devices with high sulfur recovery. A definition based upon feed rate and amount of sulfur input is more logical than a definition based upon the output.

Therefore, long tons per day means the design capacity of a Claus sulfur recovery plant based upon feed rate and content of hydrogen sulfide (expressed as sulfur) in the acid gas stream.

Exh. # 120.

In a memorandum dated July 7, 1986, plaintiffs Director of the Office of Air Quality Planning and Standards stated that exemptions available under the New Source Performance Standards to systems or devices with the primary function of air pollution did not apply automatically under the Prevention of Significant Deterioration program. Exh. # 3039. The director concluded by saying that the earlier memorandum on the topic (Exh. #3013) had been withdrawn.

II. CLEAN WATER ACT

A. Clean Water Act: Claims Ten, Eleven, Twelve and Thirteen

In May and June 1998, Daren Vanler-berghe, an environmental engineer with plaintiffs National Enforcement Investigations Center, inspected defendant’s refinery. At .the time, defendant gave Van-lerberghe a copy of its Spill Prevention Control and Countermeasure Plan that had been certified on April 4, 1996, with amendments dated November 1996 and June 1997.

1. Claim Ten

Slop oil tanks S-l and S-2 began operating in 1994 and 1995 respectively. When tanks S-l and S-2 began operation, they had no diked secondary containment area. Defendant had a 42-gallon spill at slop oil tank S-l on December 8, 1994, and a 50-gallon spill at slop oil tank S-2 on January 4, 1995. In May 1995, defendant added slop oil tanks S-l and S-2 to its Spill Prevention Control and Countermeasure Plan as potential spill sources. The plan listed the capacity of each slop oil tank as 572 barrels and the capacity of their diked area as zero. In March 1996, defendant amended its Spill Prevention Control and Countermeasure Plan to provide for installation of a sufficient diked secondary containment area for slop oil tanks S-l and S-2. Defendant’s amended plan stated, “Presently, the Slop Oil Tanks (tank numbers S-l and S-2) do not have diked secondary containment areas. Dikes will be installed in the future to provide adequate secondary containment.” This amendment was repeated in defendant’s plan of June 1997.

As early as January 1997, the Wisconsin Department of Natural Resources notified defendant of the need for remedial action at slop oil tanks S-l and S-2. Defendant did not propose the installation of a concrete containment area to the department until November 24, 1998. The department’s employee who is overseeing a spill site can recommend that the site be closed after he is satisfied with the cleanup of the site. In a letter dated May 29, 1998 to defendant’s process engineer, Mark Miller, departmental employee James Hosch wrote, “Once a completed site investigation report is submitted, Murphy can propose what levels of contaminants can remain at the site, what soil can be left in place until access is gained, site specific residual containment levels, and engineering and institutional controls.” In response to defendant’s request that the department close its investigation of the site surrounding slop oil tanks S-l and S-2, Hosch sent an e-mail to defendant’s Manager of Environmental and Process Safety, Liz Lundmark, to which he attached a memorandum in which he wrote, “Murphy proposes to [ ] install a concrete and a 60-mil HPDE membrane liner at the site to prevent future spills from infiltrating soil and to prevent precipitation from migrating into soil. Murphy needs approval before the end of August to install this system, if they are to go ahead this year.” (Emphasis in original.) The department did not ask or order defendant to postpone installation of the concrete containment area during the investigation. Defendant needed the department’s approval for closure of the site but not to install the concrete barrier.

At the time he conducted the National Enforcement Investigations Center inspection, Vanlerberghe thought that slop oil tanks S-l and S-2 were bulk storage tanks; however, he has changed his mind and now believes that they are not bulk storage tanks. In defendant’s June 1997 Spill Prevention Control and Countermeasure Plan, defendant states that slop oil tanks S-l and S-2 are in refinery area B, which is a bulk storage area. Defendant’s June 1997 plan does not mention that slop oil tanks S-l and S-2 drain into American Petroleum Institute Separator #1.

At the time of the May and June 1998 inspections, defendant had not received closure approval from the Department of Natural Resources in the S-l and S-2 area and had not installed a diked secondary containment area for slop oil tanks S-l and S-2. After the department gave verbal notification that the site would be closed in September 1999, defendant constructed secondary containment around tanks S-l and S-2.

2. Claim Eleven

Tanks 21, 22 and 23 began operating in 1961, 1961 and 1964 respectively. When the tanks began operation, they lacked a common diked area with a volume equal to the capacity of the largest tank, plus sufficient freeboard for precipitation, which is an additional 10% of the capacity of the largest tank, according to industry standard. In the May 1995 plan, defendant added tanks 21, 22 and 23 as potential spill sources. According to the May 1995, March 1996 and June 1997 versions of the plan, the largest of tanks 21, 22 and 23 is tank 23, holding 54,248 barrels with a containment area capable of holding 12,840 barrels. In March 1996, defendant amended its Spill Prevention Control and Countermeasure Plan to provide for increasing the capacity of the common diked area for tanks 21, 22 and 23 to a volüme equal to the capacity of the largest tank, plus sufficient freeboard for precipitation. Specifically, defendant’s amended Spill Prevention Control and Countermeasure Plan stated

[T]he volume of the diked area at the West Tank Farm is less than the capacity of tanks 21, 22 and 23. Repairs are planned to increase the capacity of the diked areas to a volume equal to the capacity of the largest tank, plus sufficient freeboard for precipitation. Until these repairs are completed, only one tank will remain in service and current operational procedures limit the capacity of this tank to be less than the diked area volume.

Before making this amendment to its plan, defendant had experienced difficulties at other tanks, such as overheating, that had prevented it from limiting tank contents to desired levels and had caused tanks to overflow. According to the March 1996 version of the plan, defendant had experienced spills in the past because of things such as “seal failure,” “valve left open” and “tank leaking.”

In the plan dated June 1997, defendant discussed secondary containment for Tanks 21, 22 and 23 under the section titled “Bulk Storage.” In 1997, defendant installed sufficient secondary containment for the diked areas of tanks 21, 22 and 23. On June 11, 2001, defendant measured the secondary containment area for tanks 21, 22 and 23 and found it had a capacity equal to more than 130% of the largest tank.

3. Claim Twelve

According to the May 1995 and March 1996 versions of defendant’s Spill Prevention Control and Countermeasure Plan, Tank 57 had a tank capacity of 89,000 barrels and a diked area capacity of 89,050 barrels. Defendant’s March 1996 Spill Prevention Control and Countermeasure Plan had been certified by a professional engineer. The version of its plan that defendant presented to Vanlerberghe during the 1998 inspection stated that Tank 57 had a tank capacity of 89,000 barrels and a diked area capacity of 89,050 barrels.

In February 2000, defendant recalculated the diked area capacity for Tank 57 as equal to more than 110% of the capacity of Tank 57. No changes had been made in the area of Tank 57 since 1995.

4. Claim Thirteen

Defendant’s Spill Prevention Control and Countermeasure Plan was certified by a professional engineer on April 4, 1996; however, the plan’s November 1996 and June 1997 amendments were not certified by a professional engineer.

During the three years following the April 1996 certification, defendant eliminated potential spill sources and updated the plan’s list of potential spill sources. Defendant did not get its Spill Prevention Control and Countermeasure Plan recerti-fied after it had made changes to its list of potential spill sources.

In November 1996, defendant added a new section to its plan entitled “Discharge Detection Systems,” describing alarms that would sound automatically in the event of a pipeline leak or rupture, as well as procedures for defendant’s personnel to follow when inspecting for equipment leaks, spills, splits and cracks. Also, defendant amended its plan to include visual inspection procedures and checklists. Specifically, defendant added Figures 14.7 and 14.8 (visual inspection procedures for storm water and non-storm water discharges) and Figure 14.9 (an annual facility compliance inspection checklist).

In June 1997, defendant amended its plan to add pumping procedures for an oil sump located in a truck loading area at the refinery. These pumping procedures had not been discussed in earlier versions of defendant’s plan. Also, defendant amended its plan to note the refinery’s redrawn internal boundaries, which regrouped tanks, drains and other structures at the refinery into six areas instead of nine. Because the plan had described oil storage and operation, sewer and surface drainage and secondary containment at the refinery on an area by area basis, redrawing the internal boundaries at the refinery meant that entire sections of defendant’s previous plan were no longer accurate and had to be amended. In June 1997, defendant amended its plan to state that area B of the refinery had a total of 66 tanks; previously its plan had stated that area B had 32 tanks.

III. RESOURCE CONSERVATION AND RECOVERY ACT

Defendant generates hazardous waste as a byproduct of its manufacturing processes at various locations throughout the refinery. From May 26 to June 4, 1998 and from June 15 to June 19, 1998, inspectors from plaintiffs National Enforcement Investigations Center conducted an environmental inspection of defendant’s refinery. Linda TeKrony toured the facility and reviewed records and documents.

At its wastewater treatment plant, defendant generates hazardous waste in the form of sludge, which is processed to remove recoverable oil, leaving thickener sludge containing hazardous solids. Defendant’s wastewater treatment plant collects these solids and pipes them to a thickener tank. Sludge that accumulates in the tank is pumped out into large containers and from those containers into 55-gallon drums that are disposed of off-site. On May 28, 1998, TeKrony observed four 55-gallon drums in defendant’s wash pad area that were being used to accumulate hazardous wastes, including contaminated gravel, oily pads and rags and wash pad sludges. All of the drums were labeled as hazardous waste and all had covers that formed a continuous barrier over each of the barrels so that there were no visible gaps. Each barrel cover contained a channel around its circumference that fit over the rim of the barrel; this ridge and groove design holds the cover in place in the closed position. None of the covers was secured by a barrel cover locking ring.

Defendant has never been cited for not having a locking ring on a barrel when the barrel is in storage status for fewer than 90 days.

OPINION

I. CLEAN AIR ACT

A. Background

Congress enacted the Clean Air Act “to protect and enhance the Nation’s air resources,” 42 U.S.C. § 7401(b)(1). In 1970, it charged plaintiff with establishing national ambient air quality standards that would specify the maximum permissible concentrations of certain air pollutants necessary to protect the public health. 42 U.S.C. § 7409. Once these standards had been set, each state could submit a plan providing for implementation of the standards within its borders. 42 U.S.C. § 7410. The law required that the plans be designed to prevent the significant deterioration of air quality in areas designated as “attainment” or “unclassifiable.” The states have primary responsibility for enforcing state implementation plans but the plans are enforceable by the federal government as well. 42 U.S.C. § 7413.

Although Congress wanted to speed the clean up of the nation’s air, it realized that many existing pollution sources would have difficulty complying with strict new requirements. It provided “grandfather” provisions for those facilities but anticipated that they would incorporate the newly required controls as they underwent modifications or replacement. Wisconsin Electric Power Co. v. Reilly, 893 F.2d 901, 909 (7th Cir.1990). Not all modifications came under the statute’s scope; the law covered only modifications that resulted in net emissions increases. See, e.g., 42 U.S.C. 7411(a)(4) (“ ‘modification’ means 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”).

B. The Need for Prevention of Significant Deterioration Permits for Modifications to Sulfur Recovery Unit

In 1987, when defendant began modifying its sulfur recovery unit, beginning with the installation of the tail gas analyzer, federal law incorporated into Wisconsin’s implementation plan prohibited the construction or modification of a “major emitting facility” without a Prevention of Significant Deterioration permit. 42 U.S.C. § 7479. In other words, if defendant planned to make any modifications that consisted of physical changes or changes in operation to a “major emitting source” (a stationary source that emits or has the potential to emit 100 tons a year or more of any air pollutant), such as a sulfur recovery plant, id., it would be required to obtain a permit if the modifications would result in a “significant net emissions increase” of sulfur dioxide. 40 C.F.R. § 52.21(b)(2)(i). A “significant” net emissions increase of sulfur dioxide means an increase in emissions of more than 40 tons a year. 40 C.F.R. § 52.21(b)(23)(i).

An increase in emissions is calculated by determining the excess of the source’s “potential to emit” over