United States v. Coffeyville Resources Refining

• Court: District Court, D. Kansas
• Decided: March 30, 2022
• Docket: 6:04-cv-01064
• Status: Unknown

Opinion

IN THE UNITED STATES DISTRICT COURT FOR THE DISTRICT OF KANSAS

UNITED STATES OF AMERICA and STATE OF KANSAS ex rel. KANSAS DEPARTMENT OF HEALTH AND ENVIRONMENT, Case No. 04-1064-JAR-KGG Plaintiffs,

v.

COFFEYVILLE RESOURCES REFINING & MARKETING, LLC,

Defendant.

MEMORANDUM AND ORDER Before the Court is Defendant Coffeyville Resources Refining & Marketing, LLC’s (“CRRM”) Petition for Judicial Review (Doc. 40) of the decision by Plaintiffs United States of America and the State of Kansas by and through the Kansas Department of Health and Environment (“KDHE”) to demand stipulated penalties under the parties’ 2012 Consent Decree. The Court has considered the parties’ original and supplemental briefs addressing both “threshold” and “merits-based” issues and is prepared to rule. As described more fully below, Defendant’s petition is denied to the extent it asks the Court to dismiss or reduce Plaintiffs’ stipulated penalty demand. I. Factual and Procedural Background

Defendant owns and operates a petroleum refinery located in Coffeyville, Kansas (“Refinery”). The Refinery processes crude oil into refined petroleum products, including propane, gasoline, distillates, and petroleum coke. Among numerous process units at the Refinery are the following three flares: the Coker flare (“Coker Flare”), cold water pond flare (“CWP Flare”), and an alky flare. Flares are open air combustion devices that destroy refinery waste gas, resulting in emissions of various air pollutants including sulfur dioxide (“SO2”). Defendant purchased the Refinery from the Farmland Industries’ bankruptcy estate in 2004. Immediately prior to the purchase, Plaintiffs entered into a Consent Decree with Defendant (“2004 CD”) that resolved some, but not all, Clean Air Act (“CAA”) violations at the

Refinery.1 In 2012, Plaintiffs and Defendant entered into the Second Consent Decree (“2012 CD”) under the Environmental Protection Agency’s (“EPA”) National Petroleum Refining Initiative (“NPRI”), which sought to reduce emissions and “level the playing field” across all American refineries.2 The 2012 CD contains a provision for stipulated civil penalties that Defendant must pay for “each failure to comply with the terms of this Consent Decree.”3 The 2012 CD requires, inter alia, that CRRM comply with Subparts J4 and Ja5 of the New Source Performance Standards (“NSPS”)—regulations promulgated by the EPA pursuant to Section 111 of the CAA. Paragraph 60 requires Defendant to comply with Subpart J; paragraph 61 requires Defendant to comply with Subpart Ja if, prior to termination, a flare becomes subject

to Subpart Ja. The parties do not dispute that by November 11, 2015, Subpart Ja applied to the CWP and Coker Flares and that Defendant was required to comply with it in lieu of Subpart J thereafter. Both subparts impose requirements for refinery flares in order to protect public health and the environment, including limiting hydrogen sulfide (“H2S”) concentration in the gas that is

1 Doc. 8. 2 Doc. 14. 3 Id. ¶ 180.a. 4 40 C.F.R. § 60.100, et seq. 5 40 C.F.R. § 60.100a, et seq. flared. When combusted, H2S forms SO2, compromising respiratory health, harming vegetation, and decreasing plant growth. The regulations also require refineries to monitor the H2S concentration in the gas being flared. Subpart Ja requires monitoring of other parameters including gas flow to each flare, performance tests and evaluations of monitoring equipment, adherence to monitoring equipment quality assurance and calibrations requirements, and

submission of flare management plans to EPA. On June 19, 2020, pursuant to paragraph 202 of the 2012 CD, Plaintiffs demanded stipulated penalties from Defendant under the 2012 CD for twenty-four different violations, eighteen of which were violations of Subparts J and Ja requirements. The parties engaged in informal dispute resolution as required under the 2012 CD. On January 8, 2021, after unsuccessful attempts to resolve the disputes informally, Plaintiffs sent Defendant a written notice ceasing informal dispute resolution in accordance with the 2012 CD. Plaintiffs then served their Statement of Position (“SOP”) on Defendant, setting forth their decision that Defendant is liable for $6,819,600 in stipulated penalties. Defendant ceased disputing one claim

and paid $2,600 in stipulated penalties, bringing the total stipulated penalty demand to $6,817,000. Claims 1–2 in the SOP allege that Defendant failed to comply with paragraph 60 of the 2012 CD and Subpart J by failing to install and operate a continuous emissions monitoring system at the Coker and CWP flares. Claims 3–18 allege that Defendant failed to comply with paragraph 61 of the 2012 CD and various Subpart Ja requirements at the Coker and CWP flares. In their supplemental response brief, Plaintiffs withdrew Claims 17–18. Before the informal dispute resolution had concluded as to the 2012 CD, Plaintiffs filed the First Supplemental Complaint on December 28, 2020.6 It alleged nine counts, including violations of the CAA, Kansas Air Quality Act (“KAQA”), and regulations “based on transactions, occurrences, and events that occurred after the filing of the original Complaint.”7 Counts 1 and 2 “are also violations of the 2012 Consent Decree,” based on exceedances of H2S

concentration limits at the Coker and CWP flares.8 On February 17, 2022, Plaintiffs filed a First Amended Supplemental Complaint (“FASC”), adding eight more claims.9 Defendant separately moves to dismiss the civil penalties sought by the State in all counts, Count 9 in its entirety, and to partially dismiss Count 17.10 That motion remains pending. The 2012 CD provides that Plaintiffs’ SOP is binding unless Defendant files a Petition for Review within sixty days of Plaintiffs’ SOP. On April 5, 2021, Defendant timely filed its petition requesting that the Court review the eighteen claims in the SOP for stipulated penalties based on violations of Subparts J and Ja.11 The petition asserts four “threshold” issues for the Court to consider and sought additional time for discovery and briefing on “merits-based” issues.

The Court denied Defendant’s motion for discovery and set a supplemental briefing schedule for the merits-based challenges. These briefs having all been filed, Defendant’s petition is ripe for consideration. The Court first considers Defendant’s challenges to Claims 1 and 2 based on

6 Doc. 32. 7 Id. ¶ 5. 8 Id. ¶ 7. 9 Doc. 90. 10 Doc. 91. 11 Doc. 40. The Honorable Monti L. Belot presided over this case when it was filed in 2004 and signed the consent decrees. This case was eventually reassigned to the undersigned on May 3, 2021, after the pending motions were filed. violations of Subpart J and then proceeds to Defendant’s challenges to the remaining claims under Subpart Ja. II. Standards Defendant’s Petition for Judicial Review asks this Court to resolve the parties’ disputes about sixteen alleged violations of the 2012 CD for which Plaintiffs demand stipulated penalties.

This Court has jurisdiction under Section XIII of the 2012 CD, which states that “[t]his Court shall retain jurisdiction of this matter for the purposes of implementing and enforcing the terms and conditions of the Consent Decree and for the purpose of adjudicating all disputes.”12 The 2012 CD requires the parties to comply with informal dispute resolution before bringing their dispute to the Court. The parties have engaged in this process and Claims 1–16 remain for this Court to address. The parties do not address the applicable standard of review outside the context of their previously-resolved discovery request.13 They agree that under Tenth Circuit law, “[a] consent decree is a negotiated agreement that is entered as a judgment of the court.”14 The Court

construes a consent decree for enforcement purposes as a contract; thus, “the terms of the decree and the respective obligations of the parties must be found within the four corners of the consent decree.”15 The Court applies Kansas law to interpretive issues relating to the consent decree.16

12 Doc. 14 ¶ 216. 13 In the initial round of briefing, the parties focused on whether the Court should be limited to an “administrative record” when considering the petition. The Court previously expressed its skepticism that the Administrative Procedures Act’s standard of review applies here given that it is not provided for in the 2012 CD, but declined to resolve the issue because it found that, even assuming the general standard set forth in Fed. R. Civ. P. 26(b) applied, discovery was not warranted. See Doc. 60 at 6. 14 Sinclair Oil Corp. v. Scherer, 7 F.3d 191, 193 (10th Cir. 1993) (citations omitted). 15 Id. (citing United States v. Armour & Co., 402 U.S. 673, 681–82 (1971)). 16 Id. (citing Air Line Stewards & Stewardesses Ass’n v. Trans World Airlines, Inc., 713 F.2d 319, 321 (7th Cir. 1983)); Johnson v. Lodge #93 of Fraternal Order of Police, 33 F.3d 1096, 1102 (10th Cir. 2004). Under Kansas law, the burden of proof in civil cases is generally proof by preponderance of the evidence.17 Because the 2012 CD does not provide for a different standard of review, this Court applies the preponderance standard to the parties’ factual disputes.18 Under this standard, a “‘preponderance of the evidence’ means that evidence which shows a fact is more probably true than not true.”19

Under Kansas law, if the language in a written contract “is clear and can be carried out as written, there is no room for rules of construction. To be ambiguous, a contract must contain provisions or language of doubtful or conflicting meaning, as gleaned from a natural and reasonable interpretation of its language.”20 “In considering a contract which is unambiguous and whose language is not doubtful or obscure, words used therein are to be given their plain, general and common meaning, and a contract of this character is to be enforced according to its terms.”21 “The cardinal rule of contract interpretation is that the court must ascertain the parties’ intention and give effect to that intention when legal principles so allow.”22

17 Ortega v. IBP, Inc., 874 P.2d 1188, 1192 (Kan. 1994). 18 See Sinclair Oil Corp., 7 F.3d at 193; United States v. Sanitary Dist. of Hammond, No. 2:93-CV-225- JTM-PRC, 2012 WL 6599919, at *6–7 (N.D. Ind. Dec. 18, 2012) (applying APA standard of judicial review based on the administrative record where consent decree explicitly provided for it); United States v. Minnkota Power Co- op, Inc., 831 F. Supp. 2d 1109, 1118–19 (D.N.D. 2011) (applying standard of review specified in consent decree: “The Court shall sustain the decision by NDDH unless the Party disputing the BACT Determination demonstrates that it is not supported by the state administrative record and not reasonable in light of applicable statutory and regulatory provisions.”); cf. United States v. Volvo Powertrain Corp., 758 F.3d 330, 147 (D.C. Cir. 2014) (finding no error in district court’s application of preponderance standard on petition for judicial review of a consent decree where evidentiary standard made little difference to end result, and the defendant waived it below). 19 In re B.D.-Y., 187 P.3d 594, 598 (Kan. 2008). 20 Gore v. Beren, 867 P.2d 330, 337 (Kan. 1994) (quoting Simon v. Nat’l Farmers Org., Inc., 829 P.2d 884, Syl. ¶ 2 (Kan. 1992)). 21 Wagnon v. Slawson Expl. Co., 874 P.2d 659, 666 (Kan. 1994) (quoting Barnett v. Oliver, 858 P.2d 1228, 1238 (Kan. Ct. App. 1993)). 22 Kay-Cee Enter., Inc. v. Amoco Oil Co., 45 F. Supp. 2d 840, 843 (D. Kan. 1999) (quoting Ryco Packaging Corp. v. Chapelle Int’l, Ltd., 926 P.2d 669, 674 (Kan. Ct. App. 1996)). Defendant is required to comply with certain federal regulations under the terms of the 2012 CD; thus, some of the parties’ disputes involve regulatory interpretation. In interpreting these regulations, the Court applies the same rules used to interpret statutes.23 The Tenth Circuit explains: We examine the plain language of the regulation and give each word its ordinary and customary meaning. Thus, in determining the plain meaning of a regulation, we do not consider the regulatory history or anything outside the text. If the language of the regulation is clear, we enforce the regulation in accordance with its plain meaning, giving no deference to a contrary interpretation by the Secretary.24

If the Court determines that a regulation is “genuinely ambiguous,” it applies Auer deference, “defer[ring] to the agency’s construction of its own regulation.”25 But a regulation is not genuinely ambiguous simply because it is difficult to read.26 Before concluding that a regulation is ambiguous, the Court must first exhaust its “legal toolkit” and find that “the interpretive question still has no single right answer.”27 This legal toolkit includes considering “the text, structure, history, and purpose of a regulation, in all the ways [the Court] would if it had no agency to fall back on.”28 If a genuine ambiguity remains after employing this toolkit, the Court then considers whether the agency’s reading is “reasonable.”29 And, if reasonable, the Court considers “whether the character and context of the agency interpretation entitles it to

23 Canyon Fuel Co. v. Sec’y of Lab., 894 F.3d 1279, 1287 (10th Cir. 2018) (quoting Mitchell v. Comm’r, 775 F.3d 1243, 1249 (10th Cir. 2015)). 24 Id. at 1287–88 (citations omitted). 25 Kisor v. Wilkie, 139 S. Ct. 2400, 2411 (2019) (discussing Auer v. Robbins, 519 U.S. 452 (1997)). 26 Id. 27 Id. 28 Id. 29 Id. controlling weight.”30 For example, deference to an agency’s interpretation may be inappropriate if it conflicts with a prior interpretation,31 or if it is a ‘“post hoc rationalizatio[n]’ advanced by an agency seeking to defend past agency action against attack.”32 With these standards in mind, the Court turns to the challenges raised by Defendant to Plaintiffs’ stipulated penalty demands.

III. Violations of Subpart J: SOP Claims 1 and 2 A. Background Paragraph 60 of the 2012 CD provides that the CWP and Coker flares are subject to NSPS Subpart J for Fuel Gas Combustion Devices, and “CRRM shall comply with those provisions.”33 The parties agree that the flares were subject to Subpart J from April 19, 2012, when the 2012 CD was entered, until November 11, 2015, when they became subject to Subpart Ja. Claims 1 and 2 relate to the location of the continuous emissions monitoring system (“CEMS”) used to monitor the H2S concentration of fuel gas combusted in the flares, which is required by 40 C.F.R. § 60.105(a)(3) and (4).

40 C.F.R. § 60.105(a)(3) and (4) provide in relevant part: (a) Continuous monitoring systems shall be installed, calibrated, maintained, and operated by the owner or operator subject to the provisions of this subpart as follows:

. . . .

(3) For fuel gas combustion devices subject to § 60.104(a)(1), either an instrument for continuously monitoring and recording the concentration by volume (dry basis, zero percent excess air) of SO2 emissions into the atmosphere or monitoring as provided in

30 Id. (citations omitted). 31 Christopher v. SmithKline Beecham Corp., 567 U.S. 142, 155 (2012) (citing Thomas Jefferson Univ. v. Shalala, 512 U.S. 504, 515 (1994)). 32 Id. (alteration in original) (quoting Auer v. Robbins, 519 U.S. 452, 462 (1997)). 33 Doc. 14 ¶ 60. paragraph (a)(4) of this section). The monitor shall include an oxygen monitor for correcting the data for excess.

(4) Instead of the SO2 monitor in paragraph (a)(3) of this section for fuel gas combustion devices subject to § 60.104(a)(1), an instrument for continuously monitoring and recording the concentration (dry basis) of H2S in fuel gases before being burned in any fuel gas combustion device.

(ii) Fuel gas combustion devices having a common source of fuel gas may be monitored at only one location, if monitoring at this location accurately represents the concentration of H2S in the fuel gas being burned.

Subpart J exempts certain gas streams from monitoring. Under subsection (a)(4)(iv), owners and operators are not required to monitor gases exempt under § 60.104(a)(1)—“process upset gases or fuel gas that is released to the flare as a result of relief valve leakage or other emergency malfunctions”—and gas streams combusted in a fuel gas combustion device that are inherently low in sulfur content. “Process upset gas” is defined as “any gas generated by a petroleum refinery process unit as a result of start-up, shut-down, upset or malfunction.”34 Under § 60.105(b), Defendant “may demonstrate that a fuel gas stream combusted in a fuel gas combustion device subject to § 60.104(a)(1) that is not specifically exempted in § 60.105(a)(4)(iv) is inherently low in sulfur.” To claim this exemption, Defendant must submit a written application for an exemption to the EPA Administrator.35 At the time Defendant acquired the Refinery, Farmland Industries used an H2S CEMS located immediately downstream of the Refinery’s fuel gas mix drum to monitor the H2S

34 40 C.F.R. § 60-101(e). 35 40 C.F.R. § 60-105(b)(1). concentration of the Refinery fuel gas to all combustion sources. Defendant contends that leading up to the 2012 CD, “CRRM had multiple discussions and negotiations with EPA and KDHE . . . . [and] NSPS Subpart J was a focus of several of those discussions.”36 But the only evidence of such discussion is a June 5, 2008 conference call between the parties that generated a “Task List” for Defendant to complete. One such task was that “CRRM will need to check with

operations department to identify any continuous streams which are routed to a flare (or other fuel gas combustion device) that are not monitored under NSPS J which would need an alternative monitoring plan (AMP) or need to be re-routed. – target date: July 31, 2008.”37 EPA chemical engineer Bill Peterson sent this task list to Defendant in order “to confirm Defendant’s claim that all continuous streams had been re-routed away from Defendant’s flares, including the Coker and CWP flares, and that only exempt streams pursuant to 40 C.F.R. § 60.104(a)(1) were allowed to release to the flares.”38 Before and after this conference call, CRRM representatives told Peterson that only exempt streams were allowed to release to the Coker and CWP flares, and EPA relied on this representation when the parties discussed Defendant’s past liability in the

course of negotiating the 2012 CD. In November 2015, Defendant installed new H2S, TRS, and flow monitors directly at its flares. After installation, Defendant had issues with the flow monitors communicating data to the Refinery’s data historian and data acquisition system, which store operating data and information monitored by the CEMSs. Defendant engaged the flow monitors’ manufacturer to address problems with the analog signal in 2015 and 2016. In the fall of 2015 and spring of 2016, Defendant undertook two substantial refinery-wide “turnarounds,” where the refinery units

36 Doc. 66, Ditmore Decl. ¶ 9. 37 Doc. 66-4 at 4. 38 Doc. 81, Peterson Decl. ¶ 6. and equipment were shut down in order to implement capital and expense projects and perform maintenance. EPA received the first monitoring data from the new monitors in mid-2016, which showed continuous high flow and high H2S concentrations well above the limit of 162 ppm, which Plaintiffs maintain proves that non-exempt streams were routed to the Coker and CWP

flares. According to Plaintiffs’ expert, if “the waste gas burned in a flare contains a different H2S concentration than [refinery fuel gas], the H2S monitor downstream of the fuel gas mix drum does not accurately represent the H2S in the fuel gas burned in the flare.”39 Claims 1 and 2 of the SOP allege that the H2S monitor on the fuel gas mix drum does not comply with NSPS Subpart J because it does not “accurately represent” the H2S concentration of fuel gas being burned in the Coker and CWP flares.40 Plaintiffs contend that because the monitor was located before the process units and therefore only monitored the H2S content of fuel gas entering the process units, it did not monitor the H2S content of fuel gas generated in the process units themselves. There are multiple connections between the process units and the flares after

the location of the monitor that were not re-routed back to the H2S monitor at the fuel gas mix drug; therefore, the monitor did not “accurately represent” the H2S concentration of fuel gas being burned at the flares. Plaintiffs seek $2.542 million in stipulated penalties per flare, per day from the date of the release from liability in the 2012 CD (April 19, 2012) until November 11, 2015, when CRRM was required to comply with NSPS Subpart Ja. CRRM disputes Claims 1 and 2 on the following grounds: (1) it did not violate Subpart J’s monitoring requirement as a factual matter; and (2) Plaintiffs’ knowledge of the alleged

39 Doc. 80, Sahu Decl. ¶ 20. 40 See 40 C.F.R. § 60.105(a)(4)(ii). violations since 2004 without compliance action prohibits it from now demanding the stipulated penalties in Claims 1 and 2, or in the alternative, permits the Court to reduce the penalty amount. The Court addresses these arguments in the next section. B. Discussion 1. Factual Disputes Regarding Subpart J Monitoring Violations

The Coker and CWP flares are fuel gas combustion devices subject to Subpart J.41 Because Defendant used an H2S CEMS located immediately downstream of the Refinery’s fuel gas mix drum to monitor the H2S concentration of the Refinery fuel gas to all combustion sources, it used “an instrument for continuously monitoring and recording the concentration of H2S in fuel gases before being burned in any fuel gas combustion device,” per § 60.105(a)(4). Therefore, Defendant was required to operate the monitor in a location that “accurately represents the concentration of H2S in the fuel gas being burned,” or demonstrate that an exemption applies. According to the SOP, there were multiple connections between the process units and the flares located after the monitor that were not re-routed back to the H2S monitor at

the fuel gas mix drum, and therefore were not monitored. Defendant argues that Plaintiffs fail to proffer sufficient evidence of a monitoring violation, and that any non-monitored streams were either accurately represented by the H2S analyzer on the fuel gas mix drum or exempt from monitoring under § 60.105(a)(4)(iv). Plaintiffs rely on the following evidence in support of their position that Defendant violated Subpart J: (1) the discrepancy between the H2S concentrations recorded at the flares and the monitor at the fuel gas mix drum based on data they began to receive in mid-2016; (2) Defendant’s own description of various streams to the CWP and Coker flares as being non-

41 40 C.F.R. § 60-101(g). exempt and unmonitored; and (3) that Defendant misconstrues the regulation’s definition of “relief valve leakage” in arguing that that exemption applies to some of the streams to the flares. Defendant argues, and Plaintiffs do not dispute, that Plaintiffs have the burden of proving a monitoring violation under Subpart J.42 The parties also appear to agree that Defendant bears the burden of demonstrating that an exemption to the monitoring requirement applies under

§ 60.105(a)(4)(iv). To be sure, the general rule is that while the plaintiff is required to prove a statutory violation, the party claiming the benefit of a statutory exemption to compliance bears the burden of proof on the exemption.43 With these burdens in mind, the Court first addresses whether Defendant meets the “accurate representation” provision of Subpart J before turning to the parties’ exemption arguments. a. Accurate Representation under § 60.105(a)(4)(ii) Defendant first argues that Plaintiffs fail to demonstrate that on each day between April 20, 2012 and November 11, 2015, Defendant’s monitor at the fuel gas mix drum did not “accurately represent” the H2S concentration combusted at the flares. First, Defendant claims

that Plaintiffs’ reliance on data from 2016 and 2017 to prove monitoring violations in 2012– 2015, is insufficient. It points to Plaintiffs’ own expert’s declaration that the composition of gases combusted at the flares can vary over time as evidence that Plaintiffs’ evidence of discrepancies in measurements over different years could not establish violations. Second, Defendant argues that other operational considerations render the data relied on by Plaintiffs unreliable. Namely, Defendant points to the 2015 and 2016 turnaround that involved extensive

42 Defendant incorrectly characterizes the accurate representation provision, § 60.105(a)(4)(ii), as an exemption. This is not an exemption; it is an alternative method of complying with the regulation’s monitoring requirement in subsection (a)(3) as explained above. 43 See United States v. First City Nat’l Bank of Houston, 386 U.S. 361, 366 (1967); Anderson v. Farmland Indus., Inc., 70 F. Supp. 2d 1218, 1226 (D. Kan. 1999). equipment upgrades, as well as other operational problems that rendered the new monitors’ data unreliable. Plaintiffs present strong circumstantial evidence that the fuel gas drum monitor did not accurately represent the H2S concentration at the flares on the specific dates in question. Once new monitors were installed at the flares in November 2015, they began recording concentration

levels far above the levels that were recorded at the fuel gas mix drum. According to Peterson’s Declaration, the first data he received from Defendant was in June 2016, showing high levels of H2S concentrations above the limit. Plaintiffs presented data in the SOP demonstrating that Defendant’s data “showed multiple exceedances of the 162 ppm H2S concentration limits at the flares after October 25, 2016.”44 Plaintiffs’ expert, Dr. Ranajit Sahu, discusses the specific data for the monitor at the fuel gas mix drum and the new H2S monitor at the CWP flare. According to Dr. Sahu, between October 25, 2016 and May 1, 2017, the monitor at the fuel gas mix drum reported H2S concentrations between 10 and 30 ppm for the vast majority of the hours excluding calibration.

During the same period, the new H2S monitors at the CWP flare consistently reported H2S concentrations up to 300 ppm. The H2S monitors at the flares were “pegged” at 300 ppm, meaning they could not measure concentrations above 300 ppm. Dr. Sahu contends that, based on the documents he reviewed, “the actual H2S concentration of gas measured by CWP flare monitor was likely multiples higher than 300 ppm.”45 The Court finds credible and persuasive Dr. Sahu’s independent expert opinion on this issue.

44 Doc. 54-3 at 27. 45 Doc. 80, Sahu Decl. ¶ 27. Defendant asserts that the two turnarounds at the Refinery explain the discrepancy in data relied on by Plaintiffs. The turnarounds occurred in the fall of 2015 and spring of 2016. The data cited by Plaintiffs is from later in 2016 and early 2017, after these turnarounds. Other than generic evidence that the refining process is “dynamic,” Defendant fails to explain how its two turnarounds in 2015 and early 2016 impacted the data relied on by Plaintiffs to establish the

monitoring discrepancy between the CWP flare and the fuel gas mix drum later in 2016 and early 2017. Defendant’s independent chemical engineering expert, David Wall, states that some of the connections to the flares release gases to manage process upsets, causing “small amounts of gas [to] continuously ‘leak’ to the flare during normal refinery operation.”46 And he contends that the turnarounds identified seventeen leaking valves in the various process units in the Refinery. But Defendant fails to explain how these leaks could explain the substantial data discrepancies identified by Plaintiffs’ evidence. The difference in recorded H2S concentration between the fuel gas mix drum monitors and the 2015 flare monitors are substantial—over 100

ppm different. The Court finds that Plaintiffs have presented circumstantial evidence that demonstrates by a preponderance of the evidence that Defendant violated Subpart J on each day between April 20, 2012 and November 11, 2015, because the monitor was not in a location that “accurately represents the concentration of H2S in the fuel gas being burned” under 40 C.F.R. § 60.105(a)(4)(ii). b. Exemptions from Monitoring under § 60.105(a)(4)(iv) As discussed above, subsection (a)(4)(iv) provides monitoring exemptions for gases under § 60.104(a)(1)—“process upset gases or fuel gas that is released to the flare as a result of

46 Doc. 65, Wall Decl. ¶ 31. relief valve leakage or other emergency malfunctions”—and gas streams combusted in a fuel gas combustion device that are inherently low in sulfur content. Defendant contends that these exemptions apply here. Defendant urges the Court to consider the 906 connections between the fuel gas mix drum and the flares and purports to account for all 906 connections and explain why any gas

released from these connections either would have been accurately represented by the monitor at the fuel gas mix drum, or was exempt from monitoring. Defendant relies almost exclusively on its independent expert Wall’s opinion to prove that exemptions apply. Wall’s opinion in turn is entirely based on his analysis of the various connections listed on Defendant’s Documentation, Minimization, and Analysis Tool (“DMAT”), a document it used to prepare the flare connections list for the Coker and CWP flares as of November 2015. He discusses the various categories of connections and opines whether they “would” be accurately represented by the fuel gas mix drum monitor, or were exempt. According to Plaintiffs, this itemized account of the 906 connections between the monitors was not presented by Defendant during the parties’ informal

negotiations, despite Plaintiffs’ request that Defendant provide the basis for its claimed exemptions. Nor were the DMAT tables that Wall relies on previously provided to Plaintiffs.47 According to Dr. Sahu, the DMAT tables differ from the Flare Management Plans (“FMPs”) previously provided to Plaintiffs in terms of the amount of information about each connection. As an initial matter, the Court is troubled by Defendant’s presentation of new evidence on this highly complicated, technical, and dense issue that apparently was not presented to Plaintiffs at the informal negotiation stage, at least not at this level of detail. Defendant admits in the reply brief that while it “has asserted all along that any non-monitored streams relieving to

47 See Doc. 80, Sahu Decl. ¶¶ 28–29. the Flares were exempt,” there is no “legal support” for the assertion that its newly specific arguments are too late.48 While it may be true that Defendant previously claimed that exemptions applied generally to the connections between the monitors’ locations during informal negotiations, it does not contest that its inventory of 906 connections based on the DMAT presented in its merits brief was not provided to Plaintiffs during informal negotiations.49

Indeed, Plaintiffs’ SOP makes clear that it asked Defendant to provide evidence to support its assertion that exemptions applied; there is no indication that the DMAT was put forward in response to those demands. Generally claiming that exemptions apply is a far cry from providing expert testimony and new documentation about hundreds of connections that Defendant asserts are explained away by exemptions. Of course, the purpose of the informal negotiation provision in the 2012 CD is to encourage the informal resolution of disputes before they reach the Court. Waiting to ask the Court to consider in the first instance the status of 906 different connections in the Refinery, based on dueling expert testimony, is not in keeping with the spirit of the informal negotiation requirement in the 2012 CD, to which the parties agreed to be bound.50 And the Court finds it

odd that if this was Defendant’s explanation for the discrepancy in the monitors’ readings, it would not provide this information to Plaintiffs during the lengthy process that led up to petitioning this Court for review. To assert that there is no “legal support” for Plaintiffs to

48 Doc. 84 at 9–10 n.4. 49 See Doc. 54-3 at 24–28 (addressing Defendant’s response to Plaintiffs’ stipulated penalty demands, which did not include the DMAT, including a reference to CRRM’s September 30, 2020 statement that it “has not found a way to identify each discrete instance in which process gas was released to a flare or relief valve was leaking to a flare for the period of time prior to November 2015.”). 50 Doc. 14 ¶ 219 (“Dispute resolution shall be commenced by a Defendant under the Consent Decree by giving written notice to another Party advising of a dispute pursuant to this Section XIII. The notice shall describe the nature of the dispute, and shall state the noticing Party’s position with regard to such dispute.”). complain about this strategy because this Court’s scheduling order permitted it to file a supplemental merits brief is disingenuous. Nothing in the Court’s Order permitting supplemental merits briefing signaled that the parties could include matters that were not fairly presented during the informal negotiation process. Indeed, had the Court understood the scope of Defendant’s intended belated argumentation, it may have reached a different result on

Defendant’s request for supplemental briefing. Despite Defendant’s eleventh hour attempt to bombard this Court with new evidence and highly technical explanations about hundreds of instances of claimed exemptions that were not fairly presented to Plaintiffs prior to petitioning this Court for judicial review, the Court will address Defendant’s assertion that these exemptions apply. The DMAT is the primary source Wall relied on in formulating his opinion that Defendant either complied with or was exempt from Subpart J’s monitoring requirement. According to Wall, the DMAT documented “[e]ach source and type of connection to the flare header system[s].”51 The connections were “highlighted by a group of third-party reviewers and

documented in the DMAT, who then had discussions with operators and/or engineers from each process unit to determine the frequency of contributions from each connection to the flares and the anticipated volume of such contributions.”52 Defendant attempts to account for all 906 connections listed in the DMAT between the refinery process units and the three refinery flares, 85 of which connect to the Alky flare that is not at issue here. As an initial matter, Plaintiffs’ expert calls into question the accuracy of the many classification judgments contained in the DMAT. Dr. Sahu points out that the DMAT does not

51 Doc. 65, Wall Decl. ¶ 13. 52 Id. make clear “the basis and methodology for classifications of high and low H2S concentrations.”53 The Court is persuaded by this evidence, particularly in light of Defendant’s inexplicable decision to produce the DMAT for the first time after the dispute reached the Court. The Court need not and will not engage in a connection-by-connection analysis to resolve whether Defendant has met its burden to show it was exempt from Subpart J’s monitoring

requirement. The Court only needs to find that Defendant’s itemized assertions of exemption do not fully account for the substantial discrepancy in H2S concentration measurements at the fuel gas mix drum and the flares, as demonstrated by the data presented by Plaintiffs. Plaintiffs address three categories of claimed exemptions and explain why Wall’s assertions of exemption are insufficient: (1) gas streams that Defendant claims are exempt because they are “inherently low in sulfur”; (2) two connections from the feed surge drums on the Vacuum No. 2 and No. 3 that Defendant claims are exempt as process upset gas, and (3) inadvertent leaks from the various connections that Defendant claims qualify for an exemption as “relief valve leakage.” The Court agrees with Plaintiffs that Defendant fails meet its burden of showing that the gas streams from

these connections are exempt from monitoring. Nor does Defendant’s evidence rebut Plaintiffs’ showing by a preponderance of the evidence that the monitor at the fuel gas mix drum did not accurately represent the concentration of H2S in the fuel gas being burned at the flares. i. Gas Streams “Inherently Low in Sulfur” First, Defendant contends that several gas streams are exempt because they are inherently low in sulfur. CRRM contends that 22 connections between the CWP flare and the Dehexanizer unit have a sulfur content so low that they would be accurately represented by the monitor at the fuel gas mix drum. But, as Plaintiffs point out and Defendant concedes, Defendant did not apply

53 Doc. 80, Sahu Decl. ¶ 13. for a monitoring exemption for these connections, as required by § 60.105(b), because they are “not specifically exempted under § 60.105(a)(4)(iv).”54 Accordingly, Defendant fails to establish by a preponderance of the evidence that these 22 connections are exempt under the regulation. Defendant asserts that these connections are also exempt as process fuel gas because they are used to relieve gas to the flares during a process upset. Plaintiffs respond that these

connections do not automatically qualify for an exemption as process fuel gas based on their designation as “maintenance” connections. Dr. Sahu explains that “[m]aintenance can occur at any time and can be conducted outside of start-up and shut-down time periods,” and thus, these connections “relieve to the CWP Flare during start-up, shut-down, malfunction, or upset events, but can also relieve during other periods of routine operation, such as ‘hot work’ maintenance.”55 Therefore, Defendant fails to show that these connections are exempt as process fuel gas. Similarly, Defendant fails to meet its burden of demonstrating that all 88 gas streams designated as “intermittent” or “continuous” are exempt from monitoring. Defendant asserts that the H2S concentration in these streams is accurately represented by the monitor at the fuel gas

mix drum because they would not have a measurable impact on the H2S concentration measured at the flares. These connections include control valves, manual bypass valves, manual vents, pump seals, sample vents, and sweep vents. As Dr. Sahu explains: Various process units are connected to the flares through these intermittent and continuous streams. The waste gas composition to each of these flares is different, and that composition varies over time. The single H2S monitor, located downstream of the single fuel gas mix drum, cannot logically represent the H2S concentration in the continuous and intermittent connections going to each of the two flares.56

54 See Doc. 64 n.2. 55 Doc. 80, Sahu Decl. ¶ 31. 56 Id. at ¶ 22. The Court agrees with Plaintiffs that Subpart J’s exemption provision does not allow Defendant to unilaterally decide not to monitor certain gas streams based on its belief that they would have a de minimus impact on H2S concentrations. And the Court is persuaded by Dr. Sahu’s declaration. As the Court previously determined, Plaintiffs’ data shows by a preponderance of the evidence that the fuel gas drum monitor did not accurately represent the H2S concentration at

the flares. These 88 gas streams that are designated by Defendant as “intermittent” or “continuous” are not exempt from monitoring based on Defendant’s unilateral determination that they would not have a measurable impact on the H2S concentration measured at the flares. ii. Connections from the Feed Surge Drums on Vacuum No. 2 and No. 3

Second, the parties discuss two connections from the feed surge drums on the Vacuum Unit No. 2 and No. 3, which impact the Coker flare only. According to Wall, these connections “open to the flare header when needed to maintain a safe and stable pressure of blanket gas in the vessel.”57 Wall estimates that “the additional contribution from the Vacuum Unit No. 2 and Vacuum Unit No. 3 feed surge drum vents could theoretically raise the total [Coker] flare gas H2S concentration to approximately 86.7 ppmv, which is well below the 162 ppmv H2S concentration limit.”58 Wall opines: [W]ith the exception of the Vacuum Unit No. 2 and Vacuum Unit No. 3 feed surge drum valves, gases combusted in the Coker flare from connections that were not monitored by the H2S analyzer at the fuel gas mix drum were either accurately represented by the H2S analyzer on the mix drum or only would have released gases to the flares that were generated from process units as a result of start-up, shutdown, upset, or malfunction or relief valve leakage. Based on historical data and engineering estimates, the gases from the Vacuum Unit No. 2 and Vacuum Unit No. 3 feed surge drum

57 Doc. 65, Wall Decl. ¶ 29.f.iv. 58 Id. valves would not have resulted in exceedances of the 162 ppmv H2S concentration standard for the Coker flare.59

Wall’s declaration demonstrates that these two connections released gas streams that were not in compliance with Subpart J’s monitoring requirement for the Coker flare. Defendant conflates two separate Subpart J requirements—monitoring and concentration limits. The fact that gas streams from these connections did not exceed the H2S regulatory limit does not mean that the fuel gas mix drum monitor accurately represented the measurement at the flares. Indeed, according to Wall, the H2S concentration from these connections could have been raised to 87 ppmv, well in excess of the average measurement at the fuel gas mix drum. That this amount did not exceed the concentration limit in Subpart J is irrelevant to the monitoring question, which asks if the single monitor used by CRRM between 2012 and 2015 “accurately represents the concentration of H2S in the fuel gas being burned” at the flares. Defendant’s position that “since these gases would not result in an exceedance of the 162 ppm H2S concentration standard, they were not required to be separately monitored, since the entire purpose of NSPS Subpart J’s monitoring requirements is to ensure compliance with that standard,” is unavailing because it is untethered to the plain language of the regulation.60 Thus, the evidence presented by Defendant about these two connections alone establishes that a connection-by-connection analysis fails to account for the disparate measurements between the old monitor and the new monitor at the Coker flare. iii. Relief Valve Leakage

Finally, Defendant contends that differences in H2S concentrations recorded from the monitor at the fuel gas mix drum and the flare headers could be explained by the gases leaking

59 Id. ¶ 36. 60 Doc. 84 at 13. from relief valves in various process units throughout the Refinery. Defendant claims such leaks are exempt as fuel gas that is released to the flares “as a result of relief valve leakage or other emergency malfunctions.”61 Plaintiffs argue that this exemption does not apply to relief valve leakage from non-emergency events. This exemption applies to “fuel gas streams that are exempt under § 60.104(a)(1).”62

Section 60.104(a)(1) in turn exempts “[t]he combustion in a flare of process upset gases or fuel gas that is released to the flare as a result of relief valve leakage or other emergency malfunctions.”63 The parties dispute whether “relief valve leakage” must occur as an emergency malfunction in order to be exempt. Defendant urges the Court to consider the definitions of “leak” and “leakage” and conclude that relief valve leakage contemplates “accidental or inadvertent release of gas or fluid under any number of circumstance (i.e., emergency or otherwise).”64 The term “leak” is defined as “to enter or escape through an opening usually by a fault or mistake”65 and “leakage” means “the act or process or an instance of leaking.”66 Based on these definitions, Defendant argues that the plain meaning of “relief valve leakage” is not

limited to emergency situations. Plaintiffs respond that “relief valve leakage” must be read in conjunction with “or other emergency malfunctions,” and when done so, it unambiguously applies to emergency situations only. Under the principle of noscitur a sociis, “a word is known by the company it keeps—to

61 40 C.F.R. §§ 60.105(a)(4)(iv), 60.104(a)(1). 62 Id. § 60.105(a)(4)(iv). 63 Id. § 60.104(a)(1). 64 Doc. 84 at 15. 65 Leak, Merriam-Webster.com, https://www.merriam-webster.com/dictionary/leak (last visited March 8, 2022). 66 Leakage, Merriam-Webster.com, https://www.merriam-webster.com/dictionary/leakage (last visited March 8, 2022). “avoid ascribing to one word a meaning so broad that it is inconsistent with its accompanying words, thus giving unintended breadth to the Acts of Congress.”67 Thus, Plaintiffs argue, relief valve leakage under the regulation occurs only when there is an emergency malfunction; it does not apply to “steady” relief valve leakage outside of an emergency situation. This reading of the exemption is consistent with EPA’s interpretive guidance.68

The Court finds that the regulation is unambiguous. The Court begins with the dictionary definitions of the words in question because the regulation does not define “relief valve leakage.”69 Defendant is correct that the ordinary meaning of “leak” does not require an emergency. But the Court agrees with Plaintiffs that as part of its plain language reading of the regulation, it must give effect to each word and clause, not just the meaning of a single word.70 In order to give effect to the words “or other emergency malfunctions,” relief valve leakage must occur as part of an emergency malfunction, not as steady streams of relief valve leakage that occur outside of an emergency, as urged by Defendant. Defendant’s interpretation of this language would render the words “or other” meaningless, instead allowing an exemption for

relief valve leakage and emergency malfunctions. Under the principle of noscitur a sociis and the requirement that this Court give meaning to each word in the regulation, Plaintiffs’ interpretation is the only one in keeping with the plain language of the regulation. Defendant suggests that the word “other” does not modify “relief valve leakage,” but instead distinguishes the phrase “emergency malfunction” from the word “malfunction” in the

67 Yates v. United States, 574 U.S. 528, 543 (2015). 68 See Applicability Determination from David Howekamp, U.S. EPA, to Armand S. Abay, Texaco Refining & Marketing, Inc. (May 14, 1998) (interpreting the exemption as applying only to emergency situations). 69 Canyon Fuel Co. v. Sec’y of Lab., 894 F.3d 1279, 1288 (10th Cir. 2018) (citation omitted). 70 See id. at 1289 (citing Bridger Coal Co./Pac. Minerals, Inc. v. Dir., Office of Workers’ Comp. Programs, U.S. Dep’t of Labor, 927 F.2d 1150, 1153 (10th Cir. 1991)). process upset gas definition. The Court disagrees. First, Defendant solely relies on regulatory history to support its interpretation and to argue that EPA only within the last ten years began to interpret the regulation as applying to fuel gas leakage in emergency situations. But as the Court has already stated, it is not to “consider regulatory history or anything outside the text” when examining the plain meaning of the language in the regulation.71

Second, Defendant’s interpretation is not supported by a plain reading of the regulatory language. The exemption applies to: “[t]he combustion in a flare of process upset gases or fuel gas that is released to the flare as a result of relief valve leakage or other emergency malfunctions.”72 Thus, the exemption applies to “the combustion in a flare of [1] process upset gases or [2] fuel gas that is released to the flare as a result of relief valve leakage or other emergency malfunctions.” The clause beginning with “that” is used here to define fuel gas, not process upset gases, and the term “or other emergency malfunctions” is a part of that defining clause. Moreover, the term “emergency malfunction” does not relate back to “process upset

gases,” which is separately defined from “fuel gas.” “Process upset gas” is defined as “any gas generated by a petroleum refinery process unit as a result of start-up, shut-down, upset or malfunction.”73 The exemption applies to process upset gas as so defined, “or fuel gas that is released to the flare either as a result of relief valve leakage or other emergency malfunctions.” “Fuel gas means any gas which is generated at a petroleum refinery and which is combusted.”74 The plain language of this phrase is that “other emergency malfunctions” references “relief valve

71 Id. at 1287. 72 40 C.F.R. § 60.104(a)(1). 73 Id. § 60.101(e). 74 Id. § 60.101(d). leakage,” not process upset gases. Accordingly, a plain language reading of this unambiguous regulation supports Plaintiffs’ interpretation. Defendant does not contest that the relief valve leakage it seeks to exempt did not occur as part of an emergency malfunction. Thus, to the extent unmonitored streams were due to steady relief valve leakage, they are not exempt from Subpart J.

In conclusion, the Court finds that Plaintiffs have met their burden of demonstrating a factual basis for the stipulated penalty demands in Claims 1 and 2 based on violations of Subpart J between April 10, 2012 and November 11, 2015. 2. Plaintiffs’ Delay in Challenging the CEMS Placement Defendant argues that Plaintiffs’ delay in challenging the placement of Defendant’s H2S monitor bars them from now seeking stipulated penalties for the period in question. Defendant provides several iterations of this argument—that Plaintiffs’ claims constitute a “post hoc reinterpretation” of the regulation’s monitoring requirements that the Court should reject because it constitutes unfair surprise, that the claims are time-barred under the doctrine of laches, and that

the stipulated penalties should at least be reduced under the Court’s equitable authority. The Court is not persuaded by any of these arguments. First, there is no evidence that Plaintiffs had knowledge of the violations at the time they entered into the consent decrees. Defendant fails to controvert Plaintiffs’ evidence that it relied on CRRM’s assurances before 2016 that only exempt streams were allowed to release to the Coker and CWP flares. It was not until the monitoring data was provided beginning in 2016 that Plaintiffs learned of the discrepancy in H2S concentration measurements. The absence of injunctive relief related to the fuel mix drum monitor in the 2012 CD does not mean that Plaintiffs made an affirmative determination that the monitor was compliant with Subpart J. According to Plaintiffs’ evidence, they did not have the data to make that determination before the 2012 CD. Plaintiffs did not affirm that Defendant’s monitor near the fuel gas mix drum was in compliance with Subpart J when it entered into the 2004 and 2012 Consent Decrees.75 Second, there is no evidence that the KDHE had knowledge of the violation when it issued construction and operating permits to Defendant in 2006 that required both flares to

comply with Subpart J’s monitoring requirements. Defendant contends that because the monitor was located at the fuel gas mix drum at the time, KDHE would have imposed a compliance plan and schedule in the permits if Defendant was out of compliance. But again, KDHE’s decision to issue these permits did not “affirm” or make a “determination” about the monitor’s compliance with Subpart J. There is no affirmative finding in those permits that the H2S monitor was in compliance with Subpart J. Indeed, the short permitting process and broad enforcement authority given to the EPA does not allow for an exhaustive procedure for regulators to investigate and resolve all areas of noncompliance.76 Defendant’s argument that Plaintiff’s interpretation of the regulation in the SOP

constitutes unfair surprise is also misplaced. The cases cited by Defendant discuss this issue in the context of Auer deference—once a court decides that a regulation is ambiguous and that the agency’s interpretation is reasonable, unfair surprise is one reason why a court may still decide not to defer to an agency’s interpretation of a regulation.77 But Defendant never argues that the regulation at issue here is ambiguous. Defendant argues that Plaintiffs misinterpreted the regulation and attempted to impose Subpart Ja prematurely to the Refinery’s fuel gas mix drum monitor. But the Court has determined that Plaintiffs did not misread the regulation in finding

75 See Doc. 8 ¶¶ 130, 138, 139; Doc. 14 ¶¶ 66.b, 231, 237. 76 See Citizens Against Ruining the Env’t v. E.P.A., 535 F.3d 670, 678 (7th Cir. 2008). 77 Kisor v. Wilkie, 139 S. Ct. 2400, 2417–18 (2019). that CRRM was in violation between 2012 and 2015. And Plaintiffs’ stipulated penalty demand cannot be considered unfair surprise given the lack of evidence it knew about the placement of CRRM’s monitor before 2016. Additionally, the defense of laches is not available to Defendant here. “Since at least 1940, ‘the general rule [has been] that the United States is not “subject to the defense of laches in

enforcing its rights.”’”78 Although there are some limited exceptions in specific cases, “the Tenth Circuit has generally declined to expand on those exceptions.”79 Defendant acknowledges this law, but asks the Court to apply an exception here because it is an egregious instance of laches.80 The Court disagrees. As already discussed, the evidence demonstrates that Plaintiffs first learned of the monitoring violation in 2016, informed Defendant of the potential enforcement action in 2018, and then the parties entered into tolling agreements. This was not an egregious delay. Finally, the Court declines to exercise its equitable authority to reduce the stipulated penalty demand under these circumstances. As discussed, there is no evidence that Plaintiffs

discovered the violation prior to 2016, nor that they had a duty to investigate prior to that time to determine whether there was a violation. The 2012 CD provides for stipulated penalties for Defendant’s failure to comply with Subpart J, and Plaintiffs have shown by a preponderance of the evidence that Defendant violated Subpart J on the dates alleged, triggering the stipulated penalties in paragraph 189. Under such circumstances, the Court does not have discretion to

78 FTC v. Superior Prods. Int’l II, Inc., No. 2:20-cv-02366-HLT-GEB, 2020 WL 7480390, at *2 (D. Kan. Dec. 18, 2020) (quoting FDIC v. Hulsey, 22 F.3d 1472, 1490 (10th Cir. 1994)) (citations omitted). 79 Id. (citing Hulsey, 22 F.3d at 1490). 80 See United States v. Admin. Enters., Inc., 46 F.3d 670, 673 (7th Cir. 1995). reduce the stipulated penalties that should be imposed.81 Therefore, Defendant’s request for an equitable reduction to the stipulated penalty demand is denied. In sum, the Court finds no undue delay in this case that would justify barring or reducing Plaintiffs’ stipulated penalty demands on Claims 1 and 2. IV. Violations of Subpart Ja: SOP Claims 3–16

Paragraph 61(a) of the 2012 CD provides: If, prior to the termination of this Consent Decree, a Flaring Device becomes subject to NSPS Subpart Ja for a regulated pollutant due to a “modification” (as that term is defined in the final Subpart Ja rule), the modified affected facility shall be subject to and comply with Subpart Ja, in lieu of NSPS Subpart J, for that regulated pollutant to which a standard applies as a result of the modification.82

It is undisputed that the CWP and Coker flares were modified and became subject to NSPS Subpart Ja on November 11, 2015. Claims 3–16 in the SOP assert violations of paragraph 61 of the 2012 CD at the Coker and CWP flares on various days between November 11, 2015 and June 30, 2017. Defendant challenges these stipulated penalty demands on several grounds. Defendant first challenges Plaintiffs’ stipulated penalty demands as not supported by the terms of the 2012 CD because: (1) the 2012 CD prohibits Plaintiffs from demanding both stipulated penalties under the consent decree and civil penalties in a separate complaint for Subpart Ja violations; and (2) Plaintiffs’ calculation of stipulated penalties does not comply with ¶ 189. Defendant also raises several factual challenges: (1) to Claims 3–4 on the basis that it complied with its obligation to operate and maintain flow monitors at the flares; (2) to Claims 5–6, 13–14, and 15–16 on the

81 See United States v. Volvo Powertrain Corp., 854 F. Supp. 2d 60, 71–72 (D.D.C. 2012) (fashioning equitable remedy where stipulated penalty provision did not apply to a consent decree violation). 82 Doc. 14 ¶ 61(a). basis that it timely complied with requirements for performance tests and evaluations; and (3) to Claims 11–12 on the basis that it complied with the span value requirements for the flares’ TRS analyzers. The Court addresses Defendant’s points of error in turn. A. Claim Splitting First, Defendant argues that Plaintiffs’ stipulated penalty demands are barred because

Plaintiffs are simultaneously seeking civil penalties for violations of Subpart Ja in the first two counts of their FASC. Plaintiffs maintain that these parallel remedies are permitted by the 2012 CD because they sought stipulated penalties before filing a new complaint, and because they seek to recover for different Subpart Ja violations in the FASC.83 The first two counts in the FASC are for exceedances of the H2S concentration limit at the Coker and CWP flares, in violation of Subpart Ja. The first count for relief applies to the Coker flare and alleges that Defendant exceeded the H2S concentration limit on at least 318 days since November 10, 2015, when Defendant became subject to Subpart Ja. In the alternative, Plaintiffs allege that the flare’s monitoring data was inaccurate, so Defendant failed to

“adequately operate, calibrate, and/or maintain the H2S CEMS on the Coker Flare on such days,” as required by Subpart Ja.84 The second count applies to the CWP flare and alleges exceedances on at least 486 days, or alternatively, that Defendant failed to adequately monitor or maintain the CEMS monitor on that flare, all in violation of Subpart Ja. On both counts, Plaintiffs seek “injunctive relief and the assessment of civil penalties to the United States of not more than the

83 After the briefing was complete on the petition, Plaintiffs filed the FASC. Doc. 90. Defendant filed a partial motion to dismiss the FASC on March 21, 2022. Doc. 91. The Court considers Defendant’s claim splitting arguments as applied to the FASC since the first two counts are the same. 84 Doc. 90 ¶ 140. per-day per-violation amounts set forth in Paragraph 82 above,” as well as injunctive relief and civil penalties to the State of Kansas.85 The issue of parallel remedies is explicitly addressed in paragraph 205 of the 2012 CD: 205. Subject to the provisions of Section XIV of this Consent Decree (Effect of Settlement/Reservation of Rights), the stipulated penalties provided for in this Consent Decree shall be in addition to any other rights, remedies, or sanctions available to the United States or State for CRRM’s violation of this Consent Decree or applicable law. Where a violation of this Consent Decree is also a violation of the Clean Air Act, CRRM shall be allowed a credit, for any stipulated penalties paid, against any statutory penalties imposed for such violation. The United States and State will not demand stipulated penalties for a Consent Decree violation if they have commenced litigation seeking penalties under the Clean Air Act for such violation. Notwithstanding the foregoing, the United States reserves all its rights to pursue, under the Consent Decree and/or outside of it, any other non-monetary remedies to which it is legally entitled, including but not limited to injunctive relief for violations of the Consent Decree.86

The parties have different interpretations of this provision, but the Court agrees with Plaintiffs that paragraph 205’s plain meaning is unambiguous and controls. Defendant repeatedly argues that paragraph 205 prohibits Plaintiffs from “simultaneously” demanding stipulated penalties under the 2012 CD and pursuing litigation for penalties under the CAA for the same violations. That is not what paragraph 205 says. As Plaintiffs correctly argue, they sought stipulated penalties before, not after, commencing litigation seeking penalties for Subpart Ja violations, which is explicitly permitted under paragraph 205. Plaintiffs filed their stipulated penalty demand on June 19, 2020; the Supplemental Complaint was filed on December 28, 2020. Moreover, paragraph 205 contemplates exactly this situation because it provides that: (1) the stipulated penalties under the CD are “in addition to any other . . . remedies . . . available to

85 Id. ¶¶ 143–44. 86 Doc. 14 ¶ 205. [Plaintiffs] for CRRM’s violation of this Consent Decree or applicable law”; and (2) where a violation of the CD is also a violation of the CAA, Defendant must be allowed a credit based on those stipulated penalties against any statutory penalties imposed under the FASC. Plaintiffs seek additional remedies available to them under the CAA in the FASC. To the extent Defendant becomes liable on those claims, they will be allowed a credit based on the stipulated

penalties already imposed for violating the 2012 CD on the same dates. B. Calculation of Stipulated Penalties Under Paragraph 189 Defendant challenges how Plaintiffs calculated their stipulated penalty demand under paragraph 189 of the 2012 CD, which governs the calculation of stipulated penalties for failure to comply with NSPS Subpart Ja. 189. Section V.J.: NSPS for Flaring Devices. For failure to comply with applicable NSPS Subparts A and J (or Ja if CRRM becomes subject to Ja during the term of this Consent Decree) requirements for flaring devices, including emission limits, per Flaring Device:

Period of Non-Compliance Penalty per day 1st through 30th day $500 31st through 60th day $1,500 Beyond the 60th day $2,000 or an amount equal to 1.2 times the economic benefit of delayed compliance, whichever is greater.

Specifically, Defendant objects that: (1) Plaintiffs’ “subsumation” approach—demanding a suspended stipulated penalty for concurrent Subpart Ja violations on a particular day at a particular flare—violates paragraph 189; (2) Plaintiffs incorrectly imposed graduated penalties; (3) Plaintiffs incorrectly demanded separate stipulated penalties for the Coker and CWP flares; and (4) Plaintiffs incorrectly demanded penalties on multiple days for discrete Subpart Ja violations. 1. Subsumed Violations When they calculated stipulated penalties for Subpart Ja violations, Plaintiffs assessed one stipulated penalty per flare, per day, even though they alleged that Defendant often violated multiple Subpart Ja provisions at the same flare on the same day. Plaintiffs refer to this as “subsumation” because one flaring violation on a particular day (the primary violation) subsumes

the stipulated penalties for any additional flaring violation that occurred at the same flare on the same day. Plaintiffs’ initial Demand explained: At times CRRM violated five or more NSPS Ja flaring requirements on a single day. However, Plaintiffs seek only one stipulated penalty per day per flare for these violations. Therefore, for all of CRRM’s Subpart Ja flaring violations with corresponding stipulated penalties that are subsumed in whole or part by other violations under this calculation, Plaintiffs reserve the right to seek stipulated penalties for those previously subsumed violations if any such violations are later withdrawn, or somehow found not to be violations.87

Plaintiffs referred to an Excel spreadsheet that they sent Defendant that identified all violations by type and date, including those they deemed subsumed. In the Demand, Plaintiffs stated an amount due for fully subsumed violations as “$0,*” and placed an asterisk next to certain other amounts, indicating that these amounts were “subsumed by other concurrent Subpart Ja violations.”88 On several violations, Plaintiffs stated that under paragraph 189 of the 2012 CD, “all of the stipulated penalties” for that violation “are subsumed by other concurrent Subpart Ja violations, as detailed in the previously emailed Excel spreadsheet.”89

87 Doc. 54-1 at 3. 88 Id. at 4–5 (Claims 3–4, 5–6, 7–8, 13–14). 89 Id. at 5–6 (Claims 9–10, 11–12, 15–16). Later, in a Supplemental Demand, Plaintiffs clarified that they were in fact demanding stipulated penalties for all violations that were labeled subsumed; however, because the Consent Decree limits arguably stipulated penalties to one stipulated penalty per flare, per day, for penalties that were described as “subsumed” in the demand letter, Plaintiffs are demanding such penalties but suspending CRRM’s obligation to pay such penalties at this time (and the escrowing of such funds if the claims are disputed) until Plaintiffs determine the amount of offset due to subsumation.90

This Supplemental Demand calculated the full amount of suspended stipulated penalties for the subsumed violations. Defendant challenges this approach, arguing that it deprived it of the 2012 CD’s dispute resolution provisions by allowing Plaintiffs to “suspend” penalties that exceed the maximum penalty per day provided under paragraph 189 and then “bring those secondary violations back to life if they fail to prove their ‘primary violation.’”91 By doing this, Defendant complains that it did not have the ability to choose to pay the stipulated penalty for those secondary violations rather than challenge them through the judicial review process. The Court finds that Plaintiffs’ approach, while not explicitly provided for in the 2012 CD, does not violate its plain and unambiguous terms. First, the parties agree that paragraph 189 sets the maximum stipulated penalty per day, per flare for violations of Subpart Ja. There is no dispute that Plaintiffs’ stipulated penalty demand cannot exceed these amounts. However, nothing in the 2012 CD prevents Plaintiffs from finding multiple violations on the same day at the same flare. Plaintiffs’ Demand and Supplemental Demand, along with the Excel spreadsheet detailing the stipulated penalties and whether or not they are subsumed, provided clear notice to

90 Doc. 54-2 at 3. 91 Doc. 56 at 15. Defendant about which provisions Plaintiffs claimed were violated on which days, and the total stipulated penalty that corresponded to the violation. This is in keeping with paragraph 202’s requirement that Plaintiffs’ demand include “the stipulated penalty amount the Plaintiffs have demanded for each violation (as can be best estimated), the calculation method underlying the demand, and the grounds upon which the demand is based.”92 Paragraph 202 does not require a

specific calculation method beyond what is provided for in paragraph 189; it only requires that Plaintiffs’ demand include the calculation method they used and the violations associated with its demand. Plaintiffs’ Demand, Supplemental Demand, and accompanying Excel spreadsheet complied with these requirements. Second, Plaintiffs’ approach complies with the plain terms of paragraph 189 when read in conjunction with paragraph 180, the introductory paragraph to the Stipulated Penalties section of the 2012 CD. That paragraph states that “CRRM shall pay stipulated penalties to the United States and State for each failure to comply with the terms of this Consent Decree provided herein.”93 The subsequent paragraphs in that section set forth the way stipulated penalties shall

be calculated, with paragraph 189 setting forth the calculation for Subpart Ja. The fact that paragraph 189 sets forth a maximum stipulated penalty per day simply operates as a cap to the daily stipulated penalty that can be collected. Third, Defendant has had a full and fair opportunity to seek judicial review of both “primary” and “secondary” violations that underlie the stipulated penalties demanded by Plaintiffs. Paragraph 202 contemplates that there may be more than one violation to which a stipulated penalty relates, as it provides that the “demand for the payment of stipulated penalties

92 Doc. 14 ¶ 202. 93 Id. ¶ 180 (emphasis added). will identify the particular violation(s) to which the stipulated penalty relates.” Plaintiffs set forth all applicable violations, some of which were concurrent, but only demanded that the maximum penalty be paid in escrow while the parties litigated their dispute pursuant to paragraph 203. Plaintiffs opted not to waive the concurrent violations, but instead, essentially hold the duplicative penalties in abeyance pending a determination about whether the primary

violations are either withdrawn, or determined by the Court not to constitute violations. Plaintiffs’ approach, although not explicitly provided for under the consent decree, allowed Defendant to seek review of all alleged violations at once, even though the actual penalties demanded are tied to dates of non-compliance with Subpart Ja, not the dates of each independent Subpart Ja violation. To the extent paragraphs 189 and/or 202 are ambiguous because they do not explicitly provide for a method of demanding stipulated penalties when there are concurrent violations, the Court must determine the intent of the parties “by considering all language employed, the circumstances existing when the agreement was made, the object sought to be attained, and other circumstances, if any, which tend to clarify the real intention of the parties.”94 The Court has

already found that the 2012 CD explicitly contemplated that there may be concurrent violations that justify a stipulated penalty demand. And the stated purpose of the 2012 CD is to “further the objectives of the Clean Air Act.”95 The 2012 CD also makes clear that “[t]he United States, after consultation with the State, may, in its unreviewable discretion, waive payment of any portion of stipulated penalties that may accrue under this Consent Decree.”96

94 Amoco Prod. Co. v. Wilson, Inc., 976 P.2d 941, 945 (Kan. 1999) (quoting Universal Motor Fuel, Inc. v. Johnston, 917 P.3d 877, 881 (Kan. 1996)). 95 Doc. 14 ¶ 13. 96 Id. ¶ 202. The Court cannot find that the parties intended to provide Plaintiffs with no recourse to demand stipulated penalties for concurrent violations in the event that the primary basis for the demand is withdrawn or dismissed on judicial review. Defendant suggests that Plaintiffs’ recourse was to issue a demand with multiple claims, but only “demand” stipulated penalties per flare, per day, so that the parties and the Court could litigate disputes over only those claims for

which Plaintiffs can actually demand stipulated penalties. But this would render the waiver provision meaningless—it would require Plaintiffs to waive payment of stipulated penalties that may accrue under the consent decree and remove their discretion from that determination. And it would require Plaintiffs to demand stipulated penalties one claim at a time per flare, which in this case could have potentially resulted in eight separate consecutive demands that could trigger judicial review. The Court cannot find that this procedure is in keeping with the parties’ intent as expressed in the consent decree when read as a whole. Plaintiffs’ approach is a reasonable method for demanding stipulated penalties for concurrent violations under the terms of the agreement.

Finally, it is unclear what relief Defendant seeks with this challenge. As discussed throughout this opinion, Defendant’s merits-based challenges—which they were given a full and fair opportunity to present for judicial review—are unavailing. Thus, the stipulated penalty demand for the “primary” violations have now been upheld and Defendant was provided an opportunity to challenge all grounds for which Plaintiffs could potentially demand stipulated penalties. Defendant is not and was not assessed an amount greater than what paragraph 189 allows. Even assuming Plaintiffs should have used a different method other than subsumation to demand stipulated penalties for concurrent violations, it would not lead to this Court’s finding that the amount actually demanded by Plaintiffs violates paragraph 189.97 2. Graduated Penalties for Violations on Non-Consecutive Days Defendant next challenges Plaintiffs’ calculation of stipulated penalties under paragraph 189 because they assessed graduated penalties of $1,500 or $2,000 regardless of whether the

specific violation continued consecutively for more than 30 or 60 days, respectively. But, as Plaintiffs point out, this argument depends on an incorrect reading of paragraph 189 as applying to a period of noncompliance for each type of Subpart Ja violation, rather than a period of noncompliance for any Subpart Ja violation. Plaintiffs calculated the total number of days of noncompliance with all Subpart Ja requirements and based its graduated penalties on this assessment, rather than a violation-by-violation assessment of noncompliance. For example, Claim 4 alleges a violation of Subpart Ja’s requirement to operate and maintain flow monitors at the Coker flare. If Plaintiffs determined that Defendant violated this provision for 40 days and ceased, but violated another Subpart Ja requirement on Days 41–45, they assessed penalties at

the graduated $1,500 per day amount rather than start over at $500 for the Day 41 violation. The Court agrees that this approach is supported by the plain language of paragraph 189, which sets forth tiers of stipulated penalties based on the “Period of Non-Compliance” and “[f]or failure to comply with [Subpart Ja] requirements for faring devices.” Nothing in this paragraph tethers the period of noncompliance to specific Subpart Ja violations. If the parties intended that

97 To the extent Defendant specifically challenges Claims 7–10 on the basis of subsumation, its challenge is denied for the reasons stated in this section. the penalties be tied to each separate Subpart Ja violation, they could have specified that in the consent decree, but they did not.98 3. “Per Flaring Device” Defendant argues that because the Coker and CWP flares are a “single affected facility” under NSPS Subpart Ja, Plaintiffs may only recover one stipulated penalty for any violation that

occurred at both flares on the same day. As Plaintiffs correctly note, Defendant yet again avoided the informal resolution procedures in the 2012 CD by failing to raise this argument during that process. Defendant also failed to identify this issue in its original petition for judicial review—neither the Court nor the parties were placed on notice of this “threshold” issue that does not touch on the substantive merits of the alleged violations.99 Nonetheless, the Court addresses this argument, as the plain language of the 2012 CD easily disposes of this claim. Paragraph 189 of the 2012 CD states that stipulated penalties “for flaring devices” are calculated “per Flaring Device.” The 2012 CD defines “Flaring Device” as “an Acid Gas Flaring Device and/or an HC Flaring Device.”100 The Coker flare is an Acid Gas Flaring Device.101 The CWP flare is an HC Flaring Device.102 Thus, under the plain terms of the 2012 CD, they are

separate flaring devices. The Court is not persuaded by Defendant’s argument that the term “flaring devices” in ¶ 189 is undefined. It is the plural of a defined term in the 2012 CD, agreed to by all parties, and it is used elsewhere as the same term of art defined by the consent decree. As paragraph 60 explains:

98 To the extent Defendant specifically challenges Claims 7–10 on the basis of its challenge to Plaintiffs’ imposition of graduated penalties, it is denied for the reasons stated in this section. 99 Doc. 14 ¶ 219. 100 Id. ¶ 14.p. 101 Id. ¶ 14.c. 102 Id. ¶ 14.x. CRRM currently operates the following Flaring Devices at the Refinery: (1) the Cold Pond Flare; (2) the Coker Flare; and (3) the Alky Flare. The Cold Pond Flare and the Coker Flare are “affected facilities” subject to the requirements of the NSPS, 40 C.F.R. Part 60, Subparts A and J for Fuel Gas Combustion Devices, and CRRM shall comply with those provisions. Within two (2) years of the Entry Date, the Alky Flare shall be an affected facility subject to, and CRRM shall comply with the requirements of 40 C.F.R. Part 60, Subparts A and J for Fuel Gas Combustion Devices.

The phrase “per Flaring Device” would be rendered meaningless if the Court adopted Defendant’s new contention that the stipulated penalties provision should apply to the Coker and CWP flares as one. Defendant’s argument that the possibility of violations at the Alky flare breathes meaning into its flawed interpretation is unavailing. All three flares are separate flaring devices for purposes of calculating penalties under ¶ 189 of the consent decree. 4. Continuing Violations, Claims 5–6 and 13–16 Claims 5–6 and 13–16 each allege untimely compliance with certain Subpart Ja requirements for H2S performance tests and evaluations at the flares. Plaintiffs assessed stipulated penalties for each day that each of these requirements was late until the test or evaluation was performed. Defendant challenges this approach, arguing that these claims are for discrete, one-time violations, so Plaintiffs may only recover for a single day of stipulated penalties for each claim—the day its performance was late. Plaintiffs respond that the plain language of the 2012 CD provides for daily stipulated penalties until performance is complete, that courts have found continuing violations where a requirement is an ongoing obligation, and that the regulations provide that violations continue until the requirement is met. The plain language of the 2012 CD resolves this dispute. Under paragraph 202, stipulated penalties “begin to accrue on the day after performance is due or on the [d]ay a violation occurs, whichever is applicable, and shall continue to accrue until performance is satisfactorily completed or until the violation ceases.”103 Similarly, paragraph 189 states that stipulated penalties for violations of Subpart Ja occur “per day.” The language in these two paragraphs is consistent with other provisions of the 2012 CD, which include graduated penalties for reporting violations where the report is due on a specific day.104 Here, Defendant was required to conduct certain tests and evaluations by a specific day. Plaintiffs demanded

stipulated penalties on the day after the test was due, and each day thereafter until the test or evaluation was completed. This complied with the terms of the consent decree. Defendant argues that the testing and evaluation requirements are discrete, not continuing violations, citing cases that discuss the continuing violation theory in the context of the statute of limitations.105 But the stipulated penalties are imposed under the 2012 CD, and its language controls how stipulated penalties are calculated. The Court finds that the 2012 CD unambiguously provides that stipulated penalties “shall continue to accrue until performance is satisfactorily completed or until the violation ceases.”106 Unlike the regulation at issue in Trident Seafoods Corp., the 2012 CD, to which the parties agreed to be bound, provides that the penalty

will continue to accrue until either performance is satisfactorily completed or until the violation ceases.107 Plaintiffs demanded stipulated penalties for each day that accrued after the violation

103 Id. ¶ 202. 104 See id. ¶¶ 190.b, 192.b, 192.c, 194, 196, 198.a. 105 See United States v. Reitmeyer, 356 F.3d 1313, 1321 (10th Cir. 2004) (considering whether a criminal offense is “continuing” for purposes of the statute of limitations); Toussie v. United States, 397 U.S. 112, 115 (1970) (same); United States v. Trident Seafoods Corp., 60 F.3d 556, 559 (9th Cir. 1995) (considering whether the defendant’s failure to give advance notice of the company's intent to remove asbestos was a continuing violation under the CAA and implementing regulations and finding that the language of the regulation failed to give notice that the violation would trigger a penalty “based on the length of time that the breach exists”); United States v. Midwest Generation, LLC, 720 F.3d 644, 647 (7th Cir. 2013) (considering whether failure to obtain a construction permit was a continuing violation for purposes of the CAA’s statute of limitations based on the language of the statute and implementing regulation). 106 Doc. 14 ¶ 202. 107 See Trident Seafoods Corp., 60 F.3d at 559. until performance was satisfactorily completed. This complied with the unambiguous terms of the consent decree. C. Merits-Based Challenges to Alleged Subpart Ja Violations 1. Flow Monitors, Claims 3–4 40 C.F.R. § 107a(f) required Defendant to “install, operate, calibrate and maintain, in

accordance with the specifications in paragraph (f)(1) of this section, a [continuous parameter monitoring system] to measure and record the flow rate of gas discharged to the flare.” The parties agree that Defendant installed flow monitors at the Coker and CWP flares on November 3, 2015, in order to comply with this provision, which became effective on November 11, 2015. Plaintiffs allege that Defendant failed to measure and record the flow of gas discharged at the flares from November 11, 2015 until October 21, 2016 at the Coker flare, and until April 14, 2017 at the CWP flare. Defendant contends that despite “operational issues” it experienced with the flow monitors on the dates at issue, it complied with 40 C.F.R. § 107a(f). According to Defendant’s Environmental Manager, John Ditmore, after the new flow

monitors were installed in November 2015, Defendant experienced issues with them communicating data to the Refinery’s data historian (Pi System) and data acquisition system (“DAS”), which store operating data and information monitored by various equipment, including CEMSs, installed throughout the Refinery. Ditmore asserts that the primary cause of these issues was the use of an analog signal that caused anomalies in the transmission of data between the DAS and the flow meters. Defendant subsequently switched this signal from analog to digital to help address these issues. And General Electric, the manufacturer of the Panametric flow meters, visited the Refinery on several occasions in November 2015, July 2016, August 2016, and October 2016 to install, calibrate, and/or perform maintenance on the flow meters. Plaintiffs counter that Defendant’s many reports submitted to Plaintiffs in 2016 and 2017 support its allegation that the flow monitors were not measuring and recording the flow rate of gas discharged to the flares. For example, in a December 2017 NSPS Subpart Ja Semiannual Report, Defendant admitted that “ongoing issues with the Cold Water Pond Flare flow meters” meant that “data is not available to accurately and/representatively report on the flow during that

discharge duration.”108 The reports attached to Peterson’s declaration demonstrate that Defendant’s issues with the flow meters went beyond mere communication problems and impacted the flow meters’ ability to measure and record the flow rate of gas discharged to the flare as required by the regulation. Moreover, to the extent Defendant argues that by seeking maintenance from General Electric and switching the flow meters’ analog signal to digital constitutes substantial compliance, it does not excuse CAA violations or Plaintiffs’ right to demand stipulated penalties under the 2012 CD. There is no provision in the regulation or the consent decree that excuses performance based on a finding of substantial compliance.109 Defendant’s challenge to

Plaintiffs’ stipulated penalty demands on Claims 3 and 4, based on a violation of 40 C.F.R. § 107a(f), is therefore denied.

108 Doc. 81-7 at 2; see also Doc. 81-3 at 4; Doc. 81-4 at 2; Doc. 81-5 at 2, 4. 109 See Pound v. Airosol Co., 498 F.3d 1089, 1097 (10th Cir. 2007) (stating that the CAA imposes strict liability for violations of the Act); United States v. B & W Inv. Props., 38 F.3d 362, 367 (7th Cir. 1994) (same). 2. Claims Based on Untimely Compliance with Tests and Evaluations, Claims 5–6, 13–14, and 15–16

Claims 5–6, 13–14, and 15–16 allege untimely compliance with certain performance tests and evaluations required under Subpart Ja for H2S and TRS monitors located at the Coker and CWP flares.110 a. Claims 5 and 6 First, under 40 C.F.R. § 60.104a(a): The owner or operator shall conduct a performance test for each . . . fuel gas combustion device to demonstrate initial compliance with each applicable emissions limit in § 60.102a and conduct a performance test for each flare to demonstrate initial compliance with the H2S concentration requirement in § 60.103a(h) according to the requirements of § 60.8.

Under § 60.8(a), “not later than 180 days after initial startup of such facility . . . the owner or operator of such facility shall conduct performance test(s) and furnish the Administrator a written report of the results of such performance test(s).” Claims 5 and 6 allege that Defendant failed to conduct a performance test within 180 days “after initial startup of such facility.” Plaintiffs maintain that initial startup at the CWP and Coker flares occurred when they were modified, which triggered Subpart Ja applicability. Because the flares became subject to Subpart Ja on November 11, 2015, at the latest, Plaintiffs maintain that Defendant was required to conduct these performance tests by May 9, 2016. Defendant did not conduct the tests until October 25, 2016 at the Coker flare and June 8, 2017 at the DWP flare.

110 The stipulated penalties Plaintiffs demand on these claim are suspended, as they are asserted to be subsumed claims in whole or in part. The Court nonetheless rules on these claims in the event Plaintiffs withdraw their primary claims on these dates and to ensure a complete record. Under Subpart Ja, “[s]tartup means the setting in operation of an affected facility for any purpose.”111 And “[a]ffected facility means, with reference to a stationary source, any apparatus to which a standard is applicable.”112 As Plaintiffs argue, the CWP and Coker flares did not become an “affected facility” for purposes of Subpart Ja until they were modified in 2014. EPA published in the preamble to its final rule on NSPS Subpart Ja its construction of how “startup”

applies to modified flares: For the purposes of this subpart, startup of the modified flare occurs when any of the activities in 40 CFR 60.100a(c)(1) or (2) is completed (e.g., when a new connection is made to a flare such that flow from a refinery process unit or ancillary equipment can flow to the flare via that new connection).113

Based on this guidance and the plain meaning of the regulation, Plaintiffs contend they used the latest possible date for Subpart Ja applicability as November 11, 2015, for purposes of calculating the performance test deadline. The Court agrees that EPA’s guidance is consistent with the plain meaning of the applicable regulations. Defendant argues that under § 60.8, because the flares are “existing” facilities, any “deadline for a performance test tied to their initial startup is clearly inapplicable.”114 The problem with Defendant’s interpretation of § 60.8 is that it reads out of the definition of “startup” the term, “affected facility,” and reads into the regulation the word “existing.” The flares were not an “affected facility” until they were modified, triggering Subpart Ja. The fact that they “existed” before this modification is not relevant under the plain terms of the regulations, which

111 40 C.F.R. § 60.2. 112 Id. 113 Standards of Performance for Petroleum Refineries; Standards of Performance for Petroleum Refineries for Which Construction, Reconstruction, or Modification Commenced After May 14, 2007, 77 Fed. Reg. 56422-01, 56,451 & n.11 (Sept. 12, 2012). 114 Doc. 64 at 25 (emphasis in original). references “initial startup of an affected facility.” To be sure, as Defendant points out, the word “initial” is not defined and the flares existed before they were modified in 2014. But the plain meaning of the word “initial,” is “placed at the beginning: FIRST.”115 Thus, under the terms of the regulation, most of which are defined, “initial startup of an affected facility” means the “the first setting in operation of any apparatus to which a standard is applicable for any purpose.”

Under this definition, the flares first became an “affected facility” when they were modified, which made Subpart Ja applicable for the first time. Under the plain and unambiguous terms of the regulation, Plaintiffs used a conservative estimate of November 11, 2015, for purposes of Subpart Ja applicability, and the Court therefore finds no error in tying the initial performance test deadline to this date. Plaintiffs’ interpretation of these performance test regulations is also in keeping with their purpose, which would be undermined by Defendant’s interpretation.116 The clear purpose of the performance test requirement is to demonstrate that the flares are in compliance with the H2S concentration limit in 40 U.S.C. § 60.103a(h).117 To find that there is no deadline to demonstrate

compliance undermines the stated purpose of the emission and testing requirements. Defendant argues that the regulation does not provide it with notice about the deadline that applied to conducting performance tests on its flares, and that EPA knows how to establish clear guidance about compliance deadlines but opted not to here. But this argument ignores the

115 Initial, Merriam-Webster, https://www.merriam-webster.com/dictionary/initial (last visited March 8, 2022). 116 Kisor v. Wilkie, 139 S. Ct. 2400, 2415 (2019) (explaining that the Court must “carefully consider[ ]” the text, structure, history, and purpose of a regulation, in all the ways it would if it had no agency to fall back on” (quoting Pauley v. BethEnergy Mines, Inc., 501 U.S. 680, 707 (1991) (Scalia, J., dissenting))). 117 See 40 C.F.R. § 60.104a(a) (“The owner or operator shall conduct a performance test for each . . . fuel gas combustion device to demonstrate initial compliance with each applicable emissions limit in § 60.102a and conduct a performance test for each flare to demonstrate initial compliance with the H2S concentration requirement in § 60.103a(h) according to the requirements of § 60.8.”). clear guidance provided in the definitions section of the regulation, as well as published EPA guidance, as described above. Defendant’s conclusory assertion that the EPA’s published guidance is merely “regulatory history” is unavailing. First, Defendant’s position is not merely tied to “regulatory history.” It is based on the text of the regulation and its definitions that squarely apply when construing the performance test standards at issue here. The Court has

found that the plain meaning of the regulation, when read in conjunction with the regulation’s definitions provision, is not ambiguous. Second, the EPA’s construction has been published in the Federal Register since 2012, so any argument that Defendant lacked notice of the applicable deadline, even assuming that it misinterpreted the regulation’s text, is unavailing. The Court, therefore, grants Plaintiffs’ stipulated penalty demand in Claims 5 and 6 to the extent they are not subsumed by other violations. b. Claims 13–14 Second, under 40 U.S.C. § 60.107a(a)(2)(ii), Defendant was required to “conduct performance evaluations for each H2S monitor according to the requirements of § 60.13(c) and

Performance Specification 7 of appendix B to part 60.” Section 60.13(c) in turn requires performance evaluations “during any performance test required under § 60.8 or within 30 days thereafter.” Therefore, Plaintiffs maintain in Claims 13 and 14 that Defendant was required to conduct a performance evaluation at least 30 days after the initial performance test, or by June 8, 2016. Instead, Defendant conducted its performance evaluations under this Subpart Ja provision on October 25, 2016. Defendant’s only challenge to the untimely evaluation claims in Claims 13 and 14 is based on the same deadline challenge to Claims 5 and 6—since the evaluation deadlines are tied to the deadlines for the performance tests, there is no deadline for these evaluations. For the same reasons stated above, the Court finds that the performance tests were due within 180 of November 11, 2015, and therefore the evaluations were due 30 days after the performance tests. There is no dispute that Defendant failed to complete the performance evaluations by this deadline, so Plaintiffs have demonstrated the violations in Claims 13 and 14 by a preponderance of the evidence. The Court, therefore, grants Plaintiffs’ stipulated penalty demand in Claims 13

and 14, to the extent they are not subsumed by other violations. c. Claims 15–16 Finally, under 40 U.S.C. § 60.107a(e)(1)(ii), Defendant was required to “conduct performance evaluations of each total reduced sulfur monitor according to the requirements in § 60.13(c) and Performance Specification 5 of appendix B to this part.” Like the performance evaluations for the H2S monitors, § 60.13(c) requires these performance evaluations “during any performance test required under § 60.8 or within 30 days thereafter.” Because Defendant was required to conduct a performance test for each flare to demonstrate initial compliance with the H2S concentration requirement in § 60.103a(h) according to the requirements of § 60.8 by May

9, 2016 at the latest, Plaintiffs maintain that performance evaluations on the TRS monitors at the flares were due by June 8, 2016. Defendant did not conduct these performance evaluations until October 25, 2016. First, Defendant challenges Claims 15 and 16 to the extent the deadline is tied to the H2S monitor performance tests alleged in Counts 5 and 6. To the extent Defendant argues that no deadline applies to these evaluations because no deadlines applied to the performance tests, the Court rejects this argument for the same reasons explained above on Claims 5 and 6. Defendant additionally argues that there is no requirement under the regulations that it conduct performance evaluations on the TRS monitors because there is no requirement for an initial test of those monitors like there is for the H2S monitors. But the Court finds that Plaintiffs’ interpretation of the regulations pertaining to the TRS performance evaluation requirement and deadline is consistent with a plain reading of the regulations, particularly when they are read together. The purpose of the TRS monitor evaluation is different than that for the H2S monitor evaluation, which was to determine compliance with emission limits. Under

§ 60.107a(e), sulfur monitoring is “for assessing root cause analysis threshold for affected flares.” Therefore, Defendant’s suggestion that a follow-up evaluation is not needed for the TRS monitors here because no initial test was conducted, is misplaced when considering the stated purpose of the regulation. Moreover, Defendant’s assertion that Plaintiffs’ interpretation of the regulation would require it to “conduct multiple, redundant performance evaluations on the flare TRS analyzers any time a performance test is conducted on any of these other process units at the refinery,”118 is not a fair reading of either the regulations or Plaintiffs’ position. Section § 60.107a(e)(1)(ii) requires a performance evaluation on the TRS monitor according to the requirements in

§ 60.13(c), which in turn requires performance evaluations “during any performance test required under § 60.8 or within 30 days thereafter.” Plaintiffs do not argue that any performance test under § 60.8 triggered Defendant’s obligation to conduct a TRS analyzer performance evaluation. Instead, they claim that Defendant was separately obligated to conduct performance tests on the H2S analyzers, and that those performance tests triggered the deadline for the performance evaluations on the TRS monitors at the flare headers, which are located in the same process unit at the Refinery as the H2S monitors. Defendant fails to identify any performance

118 Doc. 64 at 38. test required to be conducted at “other process units at the refinery” that would trigger the TRS analyzer performance evaluations at issue on these claims. Accordingly, the Court finds that the regulations unambiguously required Defendant to conduct these TRS analyzer performance evaluations by June 8, 2016. Because there is no question that Defendant failed to meet this deadline, the Court grants Plaintiffs’ stipulated

penalty demand in Claims 15 and 16 to the extent they are not subsumed by other violations. 3. Span Value for Flare TRS Analyzers, Claims 11-12 In Claims 11 and 12, Plaintiffs demand stipulated penalties for Defendant’s violation of 40 U.S.C. § 60.107a(e)(1)(i), which governs the appropriate span value for its TRS analyzers.119 The span value is “[t]he upper limit of a gas concentration measurement range that is specified for affected source categories in the . . . regulation.”120 Subpart Ja states that the span value on a TRS analyzer “should be determined based on the maximum sulfur content of gas that can be discharged to the flare (e.g., roughly 1.1 to 1.3 times the maximum anticipated sulfur concentration), but may be no less than 5,000 ppmv.”121

Between November 11, 2015 and April 30, 2016, Defendant used a span value of 200,000 ppm, or 20%.122 Plaintiffs maintain that this span value did not comply with § 60.107a(e)(1)(ii) because during that time period the maximum sulfur content that could be discharged to the flares was not less than 18.2%. Defendant responds that in 2016, the EPA

119 The stipulated penalties Plaintiffs demand on these claim are suspended, as they are all asserted to be subsumed claims. The Court nonetheless rules on these claims in the event Plaintiffs withdraw their primary claims that apply to these dates and to ensure a complete record. 120 40 C.F.R. Pt. 60, App. F, Procedure 1, § 2.3. 121 40 C.F.R. § 60.107a(e)(1)(i). 122 The parties vacillate between discussing the span value and sulfur concentration measurements in terms of a percentage and parts per million (“ppm”). A 1% sulfur concentration equates to 10,000 ppm. Thus, 20% sulfur concentration is the same as 200,000 ppm. Doc. 80, Sahu Decl. ¶ 35. The Court will generally reference the span value in terms of a percentage going forward. “demanded” that it set its span values at 100%, which was based on an incorrect reading of the regulation and of the analyzers’ historical data, which showed “virtually no instances” of the TRS analyzers on either flare recording sulfur concentrations above 20%. Defendant complains that in an effort to avoid an enforcement action, it voluntarily increased the span values, but maintains that based on sampling, its own engineering judgment, and knowledge of the Refinery

units and flow, its original setting of 20% was correct and in compliance with the regulation. Once again, the parties have differing interpretations of the regulation. Defendant argues that the maximum sulfur content should not be determined based on the mere possibility of a certain maximum sulfur concentration, particularly given that the TRS analyzers’ data shows that sulfur concentrations were less than 20% for more than 99% of all hours during the timeframe of November 11, 2015 and June 30, 2018. Defendant focuses on the word “anticipated,” which it contends means “expected or looked-forward to.”123 Given the data, Defendant maintains that the “maximum anticipated sulfur concentration” was “clearly” less than 20% and nowhere near the 100% value it set in 2016 in response to the EPA’s request for information.124

Plaintiffs respond that Defendant misreads the regulation by placing too much emphasis on the word “anticipated” and reading out the word “maximum.” The Court agrees, and finds that Defendant’s interpretation also reads out of the regulation the term “can be.” The regulation states that the span value is “determined based on the maximum sulfur content of gas that can be discharged to the flare.” The word “maximum” is defined as “the greatest quantity or value attainable or attained.”125 “Can” in this context is “used to indicate possibility . . . sometimes

123 Anticipated, Merriam-Webster, https://www.merriam-webster.com/dictionary/anticipated (last visited Mar. 9, 2022). 124 Doc. 84 at 21. 125 Maximum, Merriam-Webster, https://www.merriam-webster.com/dictionary/anticipated (last visited Mar. 9, 2022). used interchangeably with may” or to mean “logically or axiologically able to.”126 Therefore, the plain language of the regulation requires the span value to be determined based on the greatest value attainable or attained that may be discharged to the flare. Within the same sentence, immediately following this language, the regulation contains a parenthetical that states “e.g., roughly 1.1 to 1.3 times the maximum anticipated sulfur

concentration.” The parenthetical belongs to the sentence in which it is embedded, providing the span value based on specific multipliers. It again uses the term “maximum,” but this time adds “anticipated.” Thus, when read together, the span value is determined based on the greatest value attainable or attained that may be discharged to the flare, which should be specifically calculated as “roughly 1.1 to 1.3 times” the “greatest quantity attainable” of “expected” sulfur concentration. The parenthetical does not change the meaning of the sentence it belongs to, nor does it permit an estimated maximum sulfur concentration that disregards the maximum possible sulfur concentration that can be discharged to the flares. Moreover, Plaintiffs’ interpretation is in keeping with the purpose of the regulation, which is to “quantify[] the concentrations of high-

sulfur-containing streams as these would be the streams most likely to trigger a root-cause analysis.”127 Because the No. 3 Sulfur Recovery Unit (“SRU #3”) was reported by Defendant on August 29, 2016 to have had a maximum H2S concentration of 87%, Plaintiffs maintain this was the maximum sulfur content of gas that can be discharged to the flare and, therefore, span value should have been set at 1.1 to 1.3 times this amount.128

126 Can, Merriam-Webster, https://www.merriam-webster.com/dictionary/anticipated (last visited Mar. 9, 2022). 127 Standards of Performance for Petroleum Refineries for Which Construction, Reconstruction, or Modification Commenced after May 14, 2007, 77 Fed. Reg. 56422-01, 56,449 (Sept. 12, 2012). 128 Doc. 81-11 at 17. Defendant advances several arguments about why the span values should not be based on this 87% maximum concentration figure, and repeatedly argues that a 100% span value is not appropriate because the 87% value was either diluted by commingling with other gas streams, or occurred so rarely that it was not anticipated and therefore shouldn’t be used to determine the span value. Defendant claims its data shows that the maximum sulfur concentration recorded by

the analyzers was below 20% for more than 99% of the time during the period November 11, 2015 to June 30, 2018. Plaintiffs have demonstrated by a preponderance of the evidence that Defendant’s 20% span value on the dates in question was too low, and therefore did not comply with the regulation. First, Defendant argues that even if it is possible for some of the gas streams to have sulfur concentrations in excess of 20%, other process streams commingled with them to dilute the ultimate sulfur concentration of gases combusted at the flares. It contends that because only the comingled stream is discharged to the flares, high-sulfur-producing gas streams are not an appropriate maximum concentration measurement for purposes of span value. But Plaintiffs

factored this into their request for information that led to a finding that the span value measurements were too low. EPA responded to this exact argument by Defendant on April 22, 2016: Given the high level of sulfur in some of the process streams that can be routed to the flare headers, EPA maintains that the current TRS span value of 20% on both flares is likely too low. The variation of process stream sulfur concentration, flowrate, and baseline flowrates in each of the corresponding flare headers impact the span values; this information is needed to ensure the monitors are properly spanned and therefore, EPA is unwilling to withdraw the Information Request.129

129 Doc. 81-10 at 3. Dr. Sahu provided several calculations in his declaration based on actual data provided by Defendant, showing examples of how a high maximum sulfur concentration reported by Defendant, even when taking into account dilution from other gas streams with lower concentrations, necessitated a span value of more than 20%.130 Specifically, he calculated the span value based on the SRU #3 87% value and found that even accounting for the baseline flow

and H2S concentration in the flare after comingling, the span value should have been set at between 79% or 100%, depending on whether he used the baseline flow rate from 2015 or 2020. Either way, 20% was far too low. Defendant asserts in its briefing that “actual operating data shows that the TRS analyzers never recorded concentrations above 200,000 ppm,” so it could not have “anticipated” that gas streams in excess of that amount could be discharged to the flares.131 There are several glaring errors with this statement. First, as Defendant’s footnote immediately following this sentence admits, it is not true that the monitors “never” recorded concentrations above 200,000 ppm. Defendant admits that there were “hours in which the TRS analyzers did record sulfur

concentrations in excess of 200,000 ppm that occurred prior to the date the analyzers were certified, during analyzer calibrations, when the analyzers had failed a calibration, or during a cylinder gas audit.”132 Second, the regulation plainly does not require Defendant to determine what the “representative” amount of sulfur concentration discharged to the flares is in order to determine span value and disregard high sulfur concentration readings based on its own unilateral engineering judgment. And third, because the regulation requires a span value of 1.1 to 1.3 times the maximum sulfur concentration that can be discharged to the flares, Defendant’s

130 Doc. 80, Sahu Decl. ¶¶ 37–40. 131 Doc. 84 at 23 (emphasis in original). 132 Id. n.12. assertion that the TRS analyzers did not record concentrations above 200,000 ppm would still not demonstrate that a 20% span value was correct. Even if the maximum sulfur concentration was 20%, the span value would be 1.1 to 1.3 times that amount. Finally, Defendant’s assertion that there were “virtually no instances in which the TRS analyzers on either flare recorded TRS concentrations” that required a span value higher than

20% is not sufficient to show regulatory compliance. The regulation’s span value determination is not based on the “representative” amount of sulfur concentration, nor does it allow for the Refinery to discount actual data that it unilaterally determined was either not accurate or an errant, rarely captured measurement. The regulation requires the span value to be based on the maximum sulfur content of gas that can be discharged to the flare; not the amount that was actually discharged. The Court finds that Plaintiffs have shown by a preponderance of evidence that, even discounting the SRU #3 measurement, there were several days in 2016 where the TRS analyzers at the flares recorded sulfur concentrations that should have triggered a span value in excess of 20%.133 As such, Plaintiff’s petition to review claims 11 and 12 is denied and

Plaintiffs’ stipulated penalty demands are granted, to the extent they are not subsumed by other claims. V. Conclusion Defendant seeks judicial review of Plaintiffs’ stipulated penalty demand for $6,817,000 based on violations of paragraphs 60 and 61 of the 2012 CD. The Court has considered multiple rounds of briefing, along with the parties’ attached exhibits, on Defendant’s many legal and factual challenges to the alleged violations. For the reasons explained throughout this opinion, Defendant’s petition for review is denied to the extent it asks this Court to find that it did not

133 Doc. 80, Sahu Decl. ¶ 42. violate the 2012 CD. The petition is further denied to the extent it asks the Court to dismiss or reduce Plaintiffs’ stipulated penalty demand due as a result of those violations. IT IS THEREFORE ORDERED BY THE COURT that Defendant’s Petition for Judicial Review (Doc. 40) is denied. IT IS SO ORDERED.

Dated: March 30, 2022 S/ Julie A. Robinson JULIE A. ROBINSON UNITED STATES DISTRICT JUDGE