Environment Texas Citizen Lobby, Inc. v. ExxonMobil Corp.

824 F.3d 507, 46 Envtl. L. Rep. (Envtl. Law Inst.) 20100, 2016 U.S. App. LEXIS 9751, 2016 WL 3063302

• Court: Court of Appeals for the Fifth Circuit
• Decided: May 27, 2016
• Docket: 15-20030
• Status: Published

Opinion

FORTUNATO P. BENAVIDES, Circuit Judge:

This appeal concerns a Clean Air Act (“CAA”) citizen suit brought by Plaintiffs-Appellants Environment Texas Citizen Lobby Incorporated and Sierra Club (“Plaintiffs”) against ExxonMobil Corporation, ExxonMobil Chemical Company, and ExxonMobil Refining & Supply Company (collectively, “Exxon”). Exxon owns and operates an industrial complex (which includes a refinery and two petrochemical plants) in Baytown, Texas, and Plaintiffs allege that Exxon violated the federal permits governing operations at the complex thousands of times over a nearly eight year period. Specifically, and as relevant to this appeal, Plaintiffs allege that Exxon (1) repeatedly violated a permit condition “stating that emissions from ‘upset’ events are not authorized under any circumstances,” (2) repeatedly emitted pollutants at rates in excess of the hourly emission limits set forth in permit emission rate tables, (3) repeatedly emitted highly reactive volatile organic compounds (“HRVOCs”) at rates in excess of a 1,200 lbs./hr. emission limit, (4) repeatedly violated a prohibition on visible emissions from flares lasting more than five minutes during any two consecutive hours, and (5) repeatedly violated a number of other permit requirements, some emissions-related and some non-emissions-related, as reflected in “deviation reports” filed with the Texas Commission on Environmental Quality.

Plaintiffs sued Exxon for these and other alleged violations in the United States District Court for the Southern District of Texas. The district court conducted a thirteen-day bench trial and issued findings of fact and conclusions of law denying most of Plaintiffs’ claims and declining to order any relief. On appeal, Plaintiffs contend generally that (1) the district court erred in finding a total of only 94 actionable violations of Exxon’s permits, and (2) the district court abused its discretion in declining to impose any penalties, issue a declaratory judgment, or grant injunctive relief in remediation of the violations at issue. We now VACATE the district court’s judgment and REMAND for further proceedings.

I. BACKGROUND

Exxon’s Baytown industrial complex— the subject of the instant lawsuit — is comprised of a refinery, an olefins plant, and a chemical .plant. Overall, the complex is governed by five federal operating permits issued pursuant to Title V of the CAA. See 42 U.S.C. §§ 7661a-7661d. These federal permits (“Title V Permits”) incorporate various federal and state regulatory requirements and also incorporate by reference state permits issued pursuant to State Implementation Plan (“SIP”) - programs. Each permit at issue in this suit contains a Maximum Allowable Emission Rate Table (“MAERT”), which sets the maximum rates at which specific pollutants may be emitted from specific sources (or, in the case of “flexible” permits, groups of sources). It is also undisputed on appeal that (1) “[t]he permits for all three plants incorporate the Texas ‘HRVOC Rule,’ which limits facility-wide emissions of highly reactive volatile organic compounds to no more than 1,200 pounds per hour,” and (2) “[t]he permits for all three plants incorporate federal regulations prohibiting visible” plant flare emissions “for periods exceeding five minutes during any two-hour period.” Finally, each incorporated •permit involved in this case contains a series of additional “special conditions.” For example, and as relevant to the present appeal, a permit governing operations at the Baytown refinery provides under special conditions 38 and 39 that “[tjhis permit does not authorize upset emissions, emissions from maintenance activities that occur as a result of upsets, or any unscheduled/unplanned emissions associated with an upset. Upset emissions are not authorized, including situations where that upset is within the flexible permit emission cap or an individual emission limit.”

The state regulatory agency charged with enforcing these permit provisions in conjunction with the EPA is the Texas Commission on Environmental Quality (“TCEQ”). In order to facilitate TCEQ oversight and enforcement, state regulations require regulated entities to document “noncompliance and indications of noncompliance” with their permits in certain ways. Env’t Tex. Citizen Lobby, Inc. v. ExxonMobil Corp., 66 F.Supp.3d 875, 882 (S.D.Tex.2014). First, regulated entities must submit State of Texas Environmental Electronic Reporting System (“STEERS”) reports to the TCEQ documenting “emis-' sions events” 1 -that result in the release of pollutants at or above a threshold quantity. See 30 Tex. Admin. Code § 101.201(a); id. § 101.1(88)-(89). Second, regulated entities must maintain on-site records of “emissions events” that result in the release of pollutants below the relevant threshold quantity. Id. § 101.201(b). Third, regulated entities must submit semi-annual reports to the TCEQ documenting any “deviations” 2 from Title V permit requirements. Id. § 122.145(2). The TCEQ investigates each “reportable” event reflected in a STEERS report, reviews the on-site records of all “recordable” events, and has the authority to take enforcement action on any event should it deem such action necessary. In the present case, the record reflects that the TCEQ pursued enforcement and ultimately assessed over $1 million in penalties against Exxon based on a number of the “events” set out in its reports and records for the period relevant to this appeal. Furthermore, in 2012, the TCEQ and Exxon entered an “agreed enforcement order” which, among other things, requires Exxon to implement four “environmental improvement projects” in order to “reduce emissions at the Baytown Complex, including emissions from emissions events....”

As a supplement to the enforcement authority vested in the EPA and state regulatory agencies like the TCEQ, the CAA also authorizes “any person [to] commence a civil action on his own behalf’ against “any person ... who is alleged to have violated (if there is evidence that the alleged violation has been repeated) or to be in violation of ... an emission standard or limitation under [the CAA].” 42 U.S.C. § 7604(a)(1). The definition of “emission standard or limitation” includes any “standard,” “limitation,” “schedule,” “term,” or “condition” in a Title V permit. Id. § 7604(f)(4). Thus, any person may bring a so-called “citizen suit” under the CAA against a regulated entity that has violated a provision of its Title V permit, so long as the violation has been “repeated” or is “ongoing.” See id. § 7604(a)(1).

In December of 2010, Plaintiffs in the present case sued Exxon under the CAA’s citizen suit provision, alleging thousands of violations of Exxon’s permits over a period spanning from October of 2005 through the date of suit. 3 Plaintiffs raised seven counts in their complaint, five of which are at issue in this appeal. Specifically, Plaintiffs alleged (among other things) that Exxon (1) committed thousands of violations of the refinery permit condition providing that “upset emissions” are “not authorize[d]” (Count I); (2) committed thousands of violations of the MAERT emission limits for various pollutants in the complex’s permits (Count II); (3) committed 18 days of violations of the incorporated 1,200 pounds per hour permit limits on emissions of HRVOCs (Count III); (4) committed 44 days of violations of the incorporated permit prohibitions on visible emissions from flares for periods exceeding five minutes during any two-hour period (Count IV); and (5) committed over 4,000 days of additional violations of sundry regulatory requirements reflected in “deviation reports” that Exxon submitted to the TCEQ (Count VII). Plaintiffs sought the maximum statutory penalties for each of the violations, a declaratory judgment that Exxon violated its permits (and thus the CAA), a permanent injunction barring Exxon from further permit violations, attorneys’ fees and costs, and appointment of a “special master” to monitor implementation of relief.

As evidentiary support for the alleged violations, Plaintiffs relied exclusively on “Exxon’s STEERS reports of reportable emissions events, records of recordable emissions events, and Title V deviation reports covering the time period at issue.” Env’t Tex., 66 F.Supp.3d at 882. At the direction of the district court, the parties compiled the various reports and records into spreadsheets and “stipulated to [them] contents.” Id. The district court subsequently conducted a thirteen-day bench trial and issued findings of fact and conclusions of law in late 2014. Broadly, the district court concluded that only 94 of the thousands of alleged permit violations were “actionable,” and the court declined to order any of Plaintiffs’ requested relief. More specifically, (1) the district court treated Count I as alleging violations of MAERT hourly emission limits (essentially conflating Count I and Count II) and found no “actionable” Count I violations; (2) the district court found only 25 “actionable” Count II violations based on Plaintiffs’ ostensible failure to show that most of the violations were repeated violations of the same hourly MAERT limits; (3) the district court found only a handful of Count III and Count IV violations, as the rest of the alleged violations were not “corroborated”; (4) the district court found no additional Count VII violations, as Plaintiffs had failed to meet their burden of showing that the “indications” of noncompliance in the deviation reports were actual violations; (5) the district court declined to grant declaratory relief, because the court had “already made” findings on Exxon’s liability; (6) the district court declined to impose a penalty based, in part, on the finding that Exxon received no economic benefit from its failure to comply with its permits and the view that lengthy/serious violations could be offset by less lengthy/ less serious violations; and (7) the district court declined to grant injunctive relief, finding that the injury to the public from denial of an injunction would not outweigh the damage the injunction would cause Exxon. Plaintiffs now appeal.

II. DISCUSSION

As noted above, the district court in this case found that 94 of the thousands of alleged permit violations were “actionable” under the citizen suit provision of the CAA, and the.court declined to order any of Plaintiffs’ requested relief. Notably, the district court’s judgment on penalties went beyond merely concluding that no penalty was warranted for the violations it found actionable. Rather, the district court determined that even if every alleged violation were actionable, it would not impose a penalty. Env’t Tex., 66 F.Supp.3d at 904. We conclude that (1) the district court erred in finding 94 “actionable” pei’mit violations; (2) the district court abused its discretion when it weighed less lengthy/ less serious violations against more lengthy/more serious violations in its assessment of the CAA penalty factors; and (3) the district court erred in failing to consider certain evidence of Exxon’s economic benefit from noncompliance. We therefore VACATE the district court’s judgment and REMAND for assessment of penalties based on the correct number of actionable violations.

A. Liability

The liability claims at issue in this appeal largely hinge on the legal significance of undisputed facts. As noted above, the parties stipulated to the accuracy of Plaintiffs’ evidence, which consisted of spreadsheets detailing Exxon’s reports and records of “emissions events” and Title V “deviation reports.” However, the parties dispute nearly every legal conclusion to be drawn from the spreadsheets, including whether the spreadsheets — because they reflect Exxon’s legally required reports and records of “emissions events” — constitute admissions of permit violations. We will address each liability count in turn after briefly discussing the standard of review.

1. Standard of Review

“The standard of review for a bench trial is well established: findings of fact are reviewed for clear error and legal issues are reviewed de novo.” Preston Exploration Co., L.P. v. GSF, L.L.C., 669 F.3d 518, 522 (5th Cir.2012) (quoting Kona Tech. Corp. v. S. Pac. Transp. Co., 225 F.3d 595, 601 (5th Cir.2000)). However, “[t]he clearly erroneous standard of review does not apply to [those] factual findings made under an erroneous view of controlling legal principles.” Maritrend, Inc. v. Serac & Co. (Shipping) Ltd., 848 F.3d 469, 470 (5th Cir.2003) (quoting Lake Charles Stevedores, Inc. v. PROFESSOR VLADIMIR POPOV M/V, 199 F.3d 220, 223 (5th Cir.1999)).

“A finding is ‘clearly erroneous’ when there is no evidence to support it, or if the reviewing court, after assessing all of the evidence, is left with the definite and firm conviction that a mistake has been committed.” U.S. Bank Nat’l Ass’n v. Verizon Commc’ns, Inc., 761 F.3d 409, 431 (5th Cir.2014) (quoting Baldwin v. Taishan Gypsum Co., Ltd. (In re Chinese-Manufactured Drywall Prods. Liab. Litig.), 742 F.3d 576, 584 (5th Cir.2014)). When “the district court’s account of the evidence is plausible in light of the record viewed in its entirety,” this court “may not reverse it even though convinced that had it been sitting as the trier of fact, it would have weighed the evidence differently.” Id. (quoting Anderson v. City of Bessemer City, 470 U.S. 564, 573-74, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985)). Furthermore, if this court “determine^] that ‘there are two permissible views of the evidence,’ then [it] may not conclude that the [district] court’s choice between them was clearly erroneous.” Id.

2. Count I: The “No Upset Emissions” Condition

In Count I of their complaint, Plaintiffs alleged that Exxon violated incorporated provisions of the Title V Bay-town refinery permit providing “that upset emissions, emissions from maintenance activities that occur as a result of upsets, or any unscheduled/unplanned emissions associated with an upset, are not authorized in any amount.” Likewise, in the proposed findings of fact and conclusions of law that Plaintiffs submitted to the district court, they alleged under Count I that “Exxon violated the provisions of the Refinery’s Title V permit that prohibit upset emissions,” and Plaintiffs specifically cited special conditions 38 and 39 of incorporated refinery permit 18287 in support of this allegation. However, Plaintiffs also submitted a summary exhibit of Count I violations in which violations of “MAERT limits” were referenced. For this reason, the district court concluded that Plaintiffs’ allegations with respect to Count I had been “inconsistent,” and because the Count I summary chart listed violations “eontami-nant-by-contaminant,” the district court treated Count I as alleging violations “of conditions that apply to separate air contaminants,” i.e., MAERT emission limits. Env’t Tex., 66 F.Supp.3d at 895-96. In other words, the district court conflated Plaintiffs’ Count I allegations with their Count II allegations (addressed infra) and concluded there were no actionable violations under Count I for the same reason there were very few actionable violations under Count II. On appeal, Plaintiffs argue that the district court simply applied the wrong permit provisions to Count I; put another way, Plaintiffs claim that the court erred by applying the wrong law to the events set forth in Plaintiffs’ spreadsheets — a decision to which de novo review applies. See Maritrend, 348 F.3d at 470. We agree.

The aforementioned special conditions 38 and 39 state that “[t]his permit does not authorize upset emissions, emissions from maintenance activities that occur as a result of upsets, or any unscheduled/unplanned emissions associated with an upset. Upset emissions are not authorized, including situations where that upset is within the flexible permit emission cap or an individual emission limit.” An “upset event” (the emissions from which would be “upset emissions”) is defined under Texas law as an “unplanned and unavoidable breakdown or excursion of a process or operation that results in unauthorized emissions.” 30 Tex. Admin. Code § 101.1(110). Thus, because every “emissions event” recorded or reported by Exxon in this case also by definition involved “unauthorized emissions” as a result of “upset event[s]” or “unscheduled maintenance, startup, or shutdown activity,” id. § 101.1(28), Plaintiffs were clearly alleging under Count I that every emission of a pollutant during each recorded “emissions event” at the refinery was a violation of special conditions 38 and 39 (and, by extension, one of Exxon’s Title V permits).

The district court, however, believed it was “unclear exactly which standards or limitations Plaintiffs contend were violated under Count I,” largely because one of Plaintiffs’ summary exhibits setting forth Count I violations listed the violations under a heading for violations of “MAERT limits” rather than special conditions 38 and 39. 4 Env’t Tex., 66 F.Supp.3d at 896. The court thus suggested in a footnote that Plaintiffs were “combining a condition incorporated into a flexible permit that does not authorize upset emissions with conditions incorporated into the same flexible permit that limit separate air contaminants,” and because Plaintiffs’ Count I allegations were listed “contaminant-by-contaminant,” the court treated Count I as alleging solely violations of MAERT limits. Id. at 896 & n. 163.

On appeal, Exxon claims that the district court’s Count I analysis stemmed from its “rejection]” of “Plaintiffs’ theory that all upset emissions are actionable violations” and its “recognition] that these emissions were actionable, if at all, only if they exceeded the maximum hourly emission rates.” But this contention is inaccurate — the district court in fact expressly declined to “address whether the sole fact that there are allegedly multiple upset events makes those upset events actionable under the CAA or whether the condition referencing upset emissions constitutes a standard or limitation under the CAA.” Id. at 896. Indeed, in the district court’s view, Plaintiffs did “not contend every upset event is actionable because the condition that does not authorize upset emissions was repeatedly violated.” Id. As should be clear from the foregoing discussion, however, this is precisely what Plaintiffs contended, and we do not think the district court’s decision to ignore special conditions 38 and 39 follows from “Plaintiffs’ approach to proving repeated violations under Count I contaminant-by-contaminant.” Id. at 896 n. 167. Rather, because Plaintiffs alleged that each emission of each pollutant during refinery emissions events was a violation of the special conditions (regardless of MAERT limits), it is unsurprising under Plaintiffs’ actual Count I theory that they would list violations in such a manner.

Nevertheless, Exxon further argues that Plaintiffs’ allegations of permit violations in general are based on the fallacious theory that “unauthorized emissions” during emissions events violate state permits. Exxon claims, on the contrary, that emissions events are simply not governed by permits and are instead subject to other regulations. In support of this contention, Exxon cites a portion of its 2012 agreed enforcement order with the TCEQ, which provides that “[ejmissions events and [unplanned] MSS activities ... are not subject to permitting under 80 Tex. Admin. Code Chapters 106 or 116, and are regulated under 30 Tex. Admin. Code Chapter 101 and Tex. Health and Safety Code §§ 382.0215, 382.0216 and 382.085.” Exxon Mobil Corp., Docket No. 2011-2336-AIR-E, 2012 WL 780783, at *1 (Tex. Comm’n on Envtl. Quality Feb. 29, 2012). Based on this language, Exxon claims that “unauthorized” emissions from emissions events cannot violate a permit, because such emissions were never subject to permits in the first instance.

We see at least two problems with Exxon’s argument: first, “unauthorized emissions,” by definition, include “[ejmissions of any air contaminant ... that exceed any air emission limitation in a permit....” 30 Tex. Admin. Code § 101.1(108). Second, the language from the TCEQ agreed enforcement order, read in conjunction with the regulatory framework it references, appears to indicate simply that Exxon cannot be issued a permit by rule (under Chapter 106) or a permit for new construction or modification (under Chapter 116) that allows for emissions events. See id. §§ 106.4, 116.10-20. But this does nothing to suggest that emissions from such events are incapable of violating a permit, as evidenced by the fact that the TCEQ found violations of Exxon’s permits — including state-issued permit 18287 and the corresponding Title V permit — stemming from Exxon’s “fail[ure] to prevent unauthorized emissions” during several discrete emissions events at - the Baytown complex. Exxon Mobil Corp., 2012 WL 780783, at *4.

We accordingly conclude that the district court erred as a matter of law in treating Count I as alleging violations of MAERT limits rather than special conditions 38 and 39. Furthermore, we note (as did the district court) that the alleged “violations under Count I overlap to an extent with hourly emission limit violations under Count II,” but we do not agree that this is a reason to collapse the MAERT limits with special conditions 38 and 39. Rather, as Plaintiffs made clear in the court below, Count I sets forth the alternative theory that every “emissions event” at the refinery constitutes a violation of the “no upset emissions” provisions incorporated into the refinery’s Title V permit. As such, we believe that the district court’s judgment on Count I should be vacated and the case remanded for reconsideration of Count I together with Count II, which we will now address.

3. Count II: MAERT Limits

In Count II, Plaintiffs alleged that the “emissions events” set forth in Exxon’s reports and records encompassed over 13,-000 days of violations of the hourly numerical emission limits for specific pollutants contained in the Maximum Allowable Emission Rate Tables (MAERTs) of incorporated permits governing the Baytown refinery, olefins plant, and chemical plant. The district court found that Plaintiffs had not proven any “actionable” MAERT violations under the relevant permits for the refinery and olefins plants and had proven a total of only 25 actionable violations under the relevant chemical plant permits. The court premised its findings on the determination that because the CAA citizen suit provision authorizes suits for “repeated or ongoing” violations of “an emission standard or limitation ... in” a Title V permit, Plaintiffs had to prove repeated violations of the “same, specific” permit limitations. Env’t Tex., 66 F.Supp.3d at 895, 898. In the district court’s view, this meant that Plaintiffs had to show repeated violations of identical numerical emission limits from the MAERTs. And because Plaintiffs had categorized the violations in their spreadsheets by pollutant, with the applicable numerical limits in the spreadsheets often varying wildly for the same pollutants from one entry to the next, the district court concluded that Plaintiffs had not shown repeated violations of most MAERT limits. On appeal, Plaintiffs contend that the district court erred in viewing different numerical limits on emissions of the same pollutants from the same sources as distinct “permit limitations” for purposes of assessing whether MAERT limit violations were repeated or ongoing.

As noted previously, the CAA allows a person to bring a civil action “against any person ... who is alleged to have violated (if there is evidence that the alleged violation has been repeated) or to be in violation of ... an emission standard or limitation under [the Act].” 42 U.S.C. § 7604(a)(1). Based on this provision, the district court in this case concluded (and neither party disputes on appeal) that for a CAA violation 5 to be “actionable” in a citizen suit, “the plaintiff must prove by the preponderance of the evidence one of the following”: either “repeated violation of the same emission standard or limitation before the complaint was filed” or “violation of the same emission standard or limitation both before and after the complaint was filed.” 6 Env’t Tex., 66 F.Supp.3d at 894; see also Glazer v. American Ecology Envtl. Servs. Corp., 894 F.Supp. 1029, 1037-38 (E.D.Tex.1995). Accordingly, because an “emission standard or limitation” includes any “standard,” “limitation,” “schedule,” “term,” or “condition” in a Title V permit, 42 U.S.C. § 7604(f)(4), Plaintiffs concede that they had to prove by the preponderance of the evidence that violations of the same, specific conditions or limitations in Exxon’s permits were “repeated” in the past or occurred at least once before Plaintiffs filed suit and at least once after.

Despite this concession, Plaintiffs take issue with the district court’s finding that because violations were categorized in the spreadsheets and summaries by pollutant, with often-differing numerical limits listed, Plaintiffs failed to prove that most violations of specific MAERT limits were repeated or ongoing. Plaintiffs devote a significant portion of their brief on this point to the question of whether a MAERT limit for a particular pollutant from a particular source should be considered a single standard or limitation despite changes to the actual number of the limit over time. In short, Plaintiffs believe that multiple ex-ceedances of MAERT limits on specific pollutants from specific emission points (or groups of emission points) should be considered “repeated” even if the numbers of the limits vary due to intermittent permit amendments or renewals. Yet as Exxon points out, variations in the limits listed in Plaintiffs’ spreadsheets may, in at least some instances, be attributable to the presence of distinct numerical limits for ordinary conditions and maintenance, startup, and shutdown (“MSS”) activity within a single version of a permit. E.g., PeRmit No. 36476 (setting ordinary and MSS limits on chemical plant emissions). Nevertheless, the district court in this case did not distinguish between different emission limits in the same version of a permit and corresponding emission limits in different versions of a permit. Instead, the court simply determined that any time the listed “emission limit” in Plaintiffs’ tables varied numerically, a new permit “standard or limitation” was at issue. We conclude that this was error.

At least with respect to specific limits on particular pollutants from particular sources that change numerically due to amendments or renewals, we believe that such limits constitute the same “standard[s] or limitation!»” for purposes of determining whether violations are “repeated” or “ongoing” under the CAA citizen suit provision. 42 U.S.C. §§ 7604(a)(1) & (f)(4). This view is consistent with the approach taken by courts in assessing “ongoing” violations of Clean Water Act (“CWA”) permits. See Allen Cty. Citizens for the Env’t, Inc. v. BP Oil Co., 762 F.Supp. 733, 740-41 (N.D. Ohio 1991), aff'd, 966 F.2d 1451 (6th Cir.1992) (unpublished table decision). 7 In the CWA con text, courts have focused on pollutants and whether those pollutants have been discharged at higher rates than authorized by a permit, not simply on whether the same numerical thresholds are at issue. See id. at 740-41. We think this approach makes good sense given that, as the Fourth Circuit explained in a CWA case, “[t]he entire structure of [The Act] and regulations involves identifying specific pollutants and setting a permit limit for each pollutant of concern.” Chesapeake Bay Found. v. Gwaltney, 890 F.2d 690, 698 (4th Cir.1989). We accordingly hold that limits on emissions of specific pollutants from specific emission points (or groups of emission points in flexible permits) should constitute permit “emission standard[s] or limitation[s]” that 'may be violated repeatedly under the CAA citizen suit provision, regardless of whether the numerical values of the limits have been changed through amendments or renewals.

In light of our holding, we must vacate the district court’s judgment on Count II. On remand, the district court is instructed to determine the correct number of actionable Count II violations when treating corresponding limits on the same pollutants from different versions of the relevant permits as the same “standard[s] or limitation[s]” under the CAA. 8

4. Counts III and IV: HRVOCs and Smoking Flares

Under Counts III and IV, Plaintiffs alleged 13 violations (for a total of 18 days) of the incorporated “HRVOC rule” prohibiting emissions of highly reactive volatile organic compounds at a rate exceeding 1,200 lbs./hr. (Count III) and 42 violations (for a total of 44 days) of the incorporated “smoking flare rule” prohibiting visible emissions from flares for periods exeeed-ing five minutes during any two-hour period (Count IV). The district court counted 9 of the 13 HRVOC rule entries in Plaintiffs’ spreadsheets and 28 of the 42 smoking flare rule entries as “violations,” finding that the remaining entries were not “corroborated” as violations of the' rules because they either did not explicitly state limits had been exceeded or did not list opacity percentages and start/end times to allow for verification. Env’t Tex., 66 F.Supp.3d at 901-02. On appeal, Plaintiffs claim that the district court erred in requiring “corroboration” of these entries, as “Exxon conceded at trial that all of the alleged violations under Counts III and IV constituted ‘violations of an emission standard or limitation.’ ”

In an early portion of its order, the district court stated the following: “Exxon does not dispute that the alleged violations under Counts II, III, IV, and V of Plaintiffs’ complaint constitute violations of an emission standard or limitation.” Env’t Tex., 66 F.Supp.3d at 893 n. 153. Exxon argues on appeal that this statement referred only to “the actionability of various legal theories in general,” but we cannot agree. Far from merely acknowledging Exxon’s failure to dispute that Counts II, III, IV, and V involved “emission standards or limitations” that might hypothetically be violated, the district court expressly found it undisputed that “Exxon violated” standards and limitations under those counts. See id. at 893. In making this finding, the court was undoubtedly relying on an exchange at trial during which the court directly asked counsel for Exxon “which events from the stipulated tables” it claimed “do not constitute a violation.” In response, counsel pointed only to events under Counts I, VI, and VII. Indeed, even after the court clarified that it was “not talking about repeated or ongoing[,] [j]ust talking about the definition of violations,” counsel for Exxon replied that “those are the three areas I’ll point the court to.”

On the basis of this exchange, the district court clearly assumed each Count II event counted by Plaintiffs was undisputed as a violation, because it limited its focus in its findings of fact and conclusions of law to whether identical numerical permit limits were present in Plaintiffs’ tables such that repeated or ongoing violations of the same limits were “corroborated.” 9 See id. at 899-900. In other words, the district court appears to have treated the statements by counsel for Exxon as judicial admissions that “withfdrew]” the question of whether specific events were violations “from contention,” and it thus assumed the entries at issue under Count II were, factually, MAERT limit exceedances. See Martinez v. Bally’s La., Inc., 244 F.3d 474, 476-77 (5th Cir.2001). With respect to Counts III and IV, however, the district court concluded that a number of specific entries were not actionable because the entries themselves were not “corroborated” as violations. We find this differential treatment of Counts II, III, and IV irreconcilably inconsistent. If the district court believed there was a question as to whether particular events under the three counts constituted violations in the first instance, it should have analyzed each entry to determine whether, as a factual matter, the relevant limits were exceeded. If it believed the events under those counts were undisputed as violations, it should have analyzed simply whether the violations were “repeated” or “ongoing.” What the court did, however, was (1) analyze only whether violations, which it believed to be undisputed, were “repeated” or “ongoing” under Count II; and (2) analyze only whether specific entries corroborated violations of the relevant permit requirements under Counts III and IV. In light of the court’s analysis of Count II and its express finding that violations under Counts II, III, IV, and V were undisputed, we do not see how the district court’s treatment of entries counted by Plaintiffs as violations under Counts III and IV can be justified. We accordingly conclude that the district court erred in requiring “corroboration” for violations that, in a different portion of the same order, it explicitly found to be undisputed. On remand, the district court is instructed to include in its tally of Count III and Count IV violations the entries it rejected as “uncorroborated.”

5. Count VII: Additional Violations in Deviation Reports

Under Count VII, Plaintiffs alleged over 4,000 days of additional Title V permit violations based on “Title V deviation reports” that Exxon submitted to the TCEQ during the relevant time period. These reports contained entries reflecting the “emissions events” that were at issue in the other counts, as well as various non-emissions-related incidents (involving, for instance, reporting requirements). Plaintiffs asserted at trial that each incident contained in a deviation report constituted an actionable permit violation, and Exxon argued that none of the entries evinced “violations” at all under the definition of “deviation” set forth in the Texas Administrative Code. The district court agreed with Exxon, reasoning that (1) Texas law defines a deviation as merely “[a]n indication of noncompliance with a term or con dition of [a] permit,” 30 Tex. Admin. Code § 122.10(6); (2) Federal regulations confirm that “[a] deviation is not always a violation,” 40 C.F.R. § 71.6(a)(3)(iii)(C); and (3) given Plaintiffs’ decision to rely solely on the deviation reports themselves as evidence of underlying actionable violations, Plaintiffs had failed “to show how, in-light of [the aforementioned] provisions, the Deviations at issue ... [were] actual violations.” Env’t Tex., 66 F.Supp.3d at 903.

Plaintiffs contend on appeal that the district court misapplied the applicable “standard of proof’ in its ruling on Count VII, as the deviation reports contained “all of the prima facie evidence needed to establish that a permit violation occurred.” They accordingly argue that because Exxon failed to rebut the evidence of violations contained in the reports, the district court should have ruled that every incident in a deviation report was an actionable violation. More specifically, Plaintiffs note that federal regulations allow regulated entities to submit “other information” indicating that a “reported deviation was not a violation,” and Exxon in this case “submitted no such information as part of any of the Deviation Reports at issue, nor did it submit any such information at trial.”

We find Plaintiffs’ argument unpersuasive. While their briefing of Count VII is devoid of authority in support of their contentions regarding the “burden of proof,” Plaintiffs appear to be referring to their earlier reliance on cases reflecting that “a permittee’s own records of violations are sufficient to establish liability.” However, the cases Plaintiffs cite do not support the proposition that deviation reports specifically are sufficient to establish violations (as opposed to merely constituting “indications” of noncompliance). For example, the one CAA case Plaintiffs rely on involved Louisiana’s permitting and reporting system, which requires the filing of written reports “each time the refinery has an ‘unauthorized discharge.’ ” St. Bernard Citizens for Envtl. Quality, Inc. v. Chalmette Ref., L.L.C., 354 F.Supp.2d 697, 706 (E.D.La.2005). These reports would be akin to the STEERS reports mandated by the TCEQ, not the deviation reports at issue under Count VII. See also Concerned Citizens Around Murphy v. Murphy Oil USA, Inc., 686 F.Supp.2d 663, 680 (E.D.La.2010) (recognizing that “unauthorized discharge reports demonstrate that [the defendant] violated emission standards or limitations”).

Furthermore, to the extent Plaintiffs point to a lack of “other information” showing that deviations were not violations, testimony regarding the significance of stand-alone deviations and their relationship to compliance certification is precisely the type of evidence that could have aided the district court in resolving Count VII. See Compliance Assurance Monitoring, 62 Fed. Reg. 54,900, 54,937 (Oct. 22, 1997) (providing that a regulated entity “may include information in the certification to document that compliance was achieved”). In the absence of such evidence, however, we see no error in the district court’s conclusion that the Count VII deviation reports alone were insufficient to meet Plaintiffs’ ultimate burden of proving actionable violations. 10 See Carr v. Alta Verde Indust., Inc., 931 F.2d 1055, 1064 n. 7 (5th Cir.1991) (noting that “the burden [is] on the plaintiff to prove at trial his allegations of a continuing or intermittent violation”).

B. Remedies

Under the CAA, district courts have jurisdiction in citizen suits “to enforce” emission standards or limitations and “to apply any appropriate civil penalties.” 42 U.S.C. § 7604(a). For the thousands of days of permit violations alleged in this lawsuit, Plaintiffs sought (1) a declaratory judgment that Exxon had violated its permits '(and thus the CAA); (2) a statutory penalty of over $600 million (to be deposited in a special fund for use by the EPA pursuant to 42 U.S.C. § 7604(g)(1)); and (3) a permanent injunction prohibiting further permit violations. 11 The district court declined to order any of Plaintiffs’ requested relief. Plaintiffs now argue that (1) the district court abused its discretion in declining to issue a declaratory judgment, (2) the district court committed numerous errors (and thus abused its discretion) in declining to impose any penalties, and (3) the district court abused its discretion in declining to grant a permanent injunction. We will address each form of relief in turn.

1. Declaratory Judgment

The district court in this case refused to issue a declaratory judgment that Exxon had violated its permits and the CAA, because while it recognized that it was “undisputed Exxon violated some emission standards or limitations,” it viewed the more important issue as “whether any such violations are actionable under the CAA as a citizen suit” — and the court had “already made these findings.” Env’t Tex., 66 F.Supp.3d at 903. Plaintiffs now argue that the district court should have issued a declaratory judgment in order to “defin[e] and elarif[y] the nature of Exxon’s liability under the CAA.”

A determination of whether to grant declaratory relief is within the district court’s discretion, and a decision to deny declaratory relief is thus reviewed only for abuse of that discretion. United Teacher Assoc. Ins. Co. v. Union Labor Life Ins. Co., 414 F.3d 558, 569 (5th Cir. 2005). As we have previously explained, “[t]he two principal criteria guiding” the decision of whether to render a declaratory judgment “are .(1) when the judgment will serve a useful purpose in clarifying and settling the legal relations in issue, and (2) when it will terminate and afford relief from the uncertainty, insecurity, and controversy giving rise to the proceeding.” Concise Oil & Gas P’ship v. La. Intrastate Gas Corp., 986 F.2d 1463, 1471 (5th Cir. 1993) (quoting 10A Chaeles Alan Wright, Arthur R. Miller & Mary K. Kane, Federal Practice and Procedure § 2759 (2d ed. 1987)).

In this regard, we have recognized that a declaratory judgment may not serve a “useful purpose” when a fact-finder has already “settl[ed] the legal relations in issue.” Id.-, see also Am. Equip. Co., Inc. v. Turner Bros. Crane and Rigging, LLC, No. 4:13-CV-2011, 2014 WL 3543720, at *3 (S.D. Tex. July 14, 2014) (“Courts in the Fifth Circuit regularly reject declaratory judgment claims seeking the resolution of issues that will be resolved as part of the claims in the lawsuit.”). In Concise Oil, for instance, Plaintiffs brought an action for breach of contract and sought a declaratory judgment that the contract was valid; however, we declined to reverse the district court’s determination that such declaratory relief was unwarranted, because “the jury’s verdict and our affirmance ... on breach conclusively refute[d]” the contention that the contract had been terminated (as the defendants maintained). 986 F.2d at 1471. In the present case, the district court, sitting as fact-finder, found that Exxon had committed over 90 actionable violations of its permits (thus also violating the CAA). And while we vacate the district court’s judgment, our “affir-mance” of the broad conclusion that Exxon committed actionable violations of its permits diminishes any “useful purpose” that a declaratory judgment on that point might otherwise serve. As such, we find no abuse of discretion in the district court’s decision not to grant declaratory relief.

2. Penalties

As noted previously, the district court in this case went beyond merely concluding that no statutory penalties under the CAA were warranted for the few violations it found actionable — rather, the district court broadly concluded that “even if the Court had found every Event and Deviation in this case is actionable, the Court would still find. Exxon should not be penalized,” and it proceeded to analyze each penalty factor from that perspective. Env’t Tex., 66 F.Supp.3d at 904. Plaintiffs argue on appeal that the district court erred in its assessment of four of the statutory penalty factors, and thus its ultimate refusal to assess a penalty was an abuse of discretion. We will begin by discussing penalties under the CAA generally and will then address Plaintiffs’ arguments with respect to each penalty factor as the district court applied it in this case.

i. CAA Penalties Generally

The CAA provides that in a citizen suit, “[a] penalty may be assessed for each day of violation.” 42 U.S.C. § 7413(e)(2). The parties agree on appeal that imposition of a civil penalty is not mandatory under the CAA; rather, the decision whether to impose a penalty rests in the discretion of the court. 12 See, e.g., Pound v. Airosol Co., Inc., 498 F.3d 1089, 1094 (10th Cir.2007) (referring to “the penalty, if any, to be assessed for a violation of the Act”). In deciding whether to impose a penalty, however, a court must “take into consideration” seven statutory factors, “in addition to such other factors as justice may require.” See 42 U.S.C. § 7413(e)(1) (providing that the court “shall take [the factors] into consideration”); Pound, 498 F.3d at 1097-98 (recognizing that the statutory factors “must be considered in a CAA penalty analysis”). These factors are:

(1) the size of the business;

(2) the economic impact of the penalty on the business;

(3) the violator’s full compliance history and good faith efforts to comply;

(4) the duration of the violation as established by any credible evidence;

(5) payment by the violator of penalties previously assessed for the same violation;

(6) the economic benefit of noncompliance; and

(7) the seriousness of the violation.

42 U.S.C. § 7413(e)(1).

In the present case, the district court concluded that the “size of the busi ness” and “economic impact of the penalty” factors weighed in favor of assessing a penalty, as “Exxon [would] only be impacted by a large penalty and has the ability to pay the alleged maximum penalty.” Env’t Tex., 66 F.Supp.3d at 904. Plaintiffs do not contest this conclusion on appeal, for obvious reasons. The district court also concluded that because Exxon had previously paid $1,423,632 in penalties for some of the violations at issue (as a result of TCEQ enforcement actions), that amount should be “deducted from any penalty otherwise warranted.” Id. at 907. Plaintiffs likewise do not contest the district court’s resolution of this factor. With respect to the remaining factors, however, the district court concluded that each one either weighed against assessing a penalty or, at most, weighed neither for nor against assessing a penalty, and Plaintiffs vigorously contest the district court’s resolution and weighing of these remaining factors. We will accordingly address each penalty factor in roughly the same order as the district court, keeping in mind that (1) a district court’s analysis of the penalty factors is subject to review only for abuse of discretion, and (2) underlying factual findings are reviewed only for clear error. Sierra Club v. Cedar Point Oil Co. Inc., 73 F.3d 546, 573 (5th Cir.1996); see also United States ex rel. Adm’r of EPA v. CITGO Petroleum Corp., 723 F.3d 547, 551 (5th Cir.2013) (quoting Tull v. United States, 481 U.S. 412, 427, 107 S.Ct. 1831, 95 L.Ed.2d 365 (1987)) (acknowledging that “the process of weighing” similar CWA penalty factors is “highly discretionary,” and thus “[a] court’s determination of the amount of a penalty to be assessed is reviewed under the highly deferential abuse-of-discretion standard”).

ii. Compliance History and Good Faith Efforts to Comply

The district court in this case determined that Exxon’s compliance history and good faith efforts to comply with its permits weighed against assessment of a penalty. At the outset, the court noted that “the number of [events] at issue in this case [was] high” and might suggest that Exxon’s compliance history was “arguably inadequate.” Env’t Tex., 66 F.Supp.3d at 904-05. However, the court found that based on the extremely large size of the facility, it would not be “possible to” eliminate all emissions events, and thus “the number of Events and Deviations alone is not the best evidence of compliance history” or “good faith effort to comply.” Id. at 905. Rather, the district court concluded that Exxon “made substantial efforts to improve environmental performance and compliance” based on (1) the “significant reduction” in overall unauthorized emissions at the complex “over the years at issue in this case”; (2) Exxon’s agreement to undertake environmental improvement projects at the complex in an enforcement order with the TCEQ; (3) the lack of “credible evidence” that any of the alleged violations “resulted from a recurring pattern”; and (4) the “persuasive and credible” testimony from a chemical engineer and a former TCEQ official that Exxon’s “concerted effort[s] to comply” had contributed to the Baytown facility’s reputation as a “leader[] in maintenance and operation practices.” Id. at 905-06.

On appeal, Plaintiffs raise several challenges to the district court’s analysis of this factor. First, Plaintiffs contend that the district court essentially “invok[ed] presumed impossibility of compliance ‘as a reason to not impose penalties’ ” despite clear precedent and regulatory language indicating that “ ‘impossibility is not a defense’ ” to compliance with one’s permits. In this regard, Plaintiffs correctly identify that under Texas law, it is not “a defense in an enforcement action that it would have been necessary to halt or reduce the permitted activity in order to comply with the permit terms and conditions of the permit.” 30 Tex. Admin. Code § 122.143(4). In other words, if a regulated entity is incapable of operating in compliance with its permits, the “one simple and straightforward way ... to avoid paying civil penalties” is to “cease[ ] operations until it [is] able to” do so. Atl. States Legal Found., Inc. v. Tyson Foods, Inc., 897 F.2d 1128, 1141-42 (11th Cir.1990). However, as the district court in this case plainly noted, its invocation of Exxon’s size and the infeasibility of achieving full compliance was not an attempt to raise impossibility as a bar to imposing penalties; rather, the district court’s discussion on this point was simply a recognition that compliance history and good faith efforts to comply should be viewed in the context of the size and complexity of the emitting facility at issue, and thus “the number of [emissions events] does not alone mean Exxon did not make a good faith effort comply” with its operating permits for this particular complex. Env’t Tex., 66 F.Supp.3d at 905 n. 221. We accordingly reject Plaintiffs first argument.

Plaintiffs next claim that the district court clearly erred in finding that Exxon’s agreement to undertake four environmental improvement projects constituted a “substantial effort[]” at achieving compliance that demonstrated “good faith.” Plaintiffs argue that the agreement merely imposed toothless and overdue requirements in an attempt by Exxon and complied TCEQ officials to “undercut more stringent citizen enforcement,” as evidenced by the fact that the agreement “was negotiated, at Exxon’s instigation, only after” Plaintiffs gave notice of their intent to file suit. On this point, we think it is possible that the agreement between Exxon and the TCEQ is a sterling example of regulatory capture at its worst; however, it is also entirely possible that Exxon’s explanation for pursuing the agreement — that it wanted more “certainty” in enforcement— is valid and that the company did want to take good-faith steps towards reducing future compliance issues. Thus, given that there are “two permissible views of the evidence,” the district court was entitled to take the view it did, and we cannot second-guess that view now. U.S. Bank Nat’l Ass’n, 761 F.3d at 431.

Finally, Plaintiffs argue that the district court “did not consider several other factors relevant to Exxon’s compliance history,” including “the number of similar violations in the past” and “the prior enforcement lawsuit brought by the United States.” We were unable to discern, however, in what way Plaintiffs believe the district court’s failure to consider these “factors” was erroneous — the authority cited in Plaintiffs’ briefing, a CWA order from the Southern District of Mississippi applying that statute’s “history of violations” penalty factor, merely recognizes that “courts consider ... the duration and nature of’ past and present violations under the Act. United States v. Gulf Park Water Co., Inc., 14 F.Supp.2d 854, 864 (S.D.Miss.1998). Nowhere does the Gulf Park court state that specific consideration of the factors listed by Plaintiffs is required (or even appropriate) in all cases, and we have no reason to believe it would be. We accordingly conclude that the district court in this case did not abuse its discretion in determining that the “compliance history and good faith efforts to comply” penalty factor weighed against imposition of a penalty. 13

iii. Economic Benefit of Noncompliance

The “economic benefit of noncompliance” factor directs courts to “consider the financial benefit to the offender of delaying capital expenditures and maintenance costs on pollution-control equipment.” CITGO, 723 F.3d at 552. We have recognized in the Clean Water Act context that “a district court generally must make a ‘reasonable approximation’ of economic benefit when calculating a penalty under [the Act],” as a finding on the amount of economic benefit is “central to the ability .of a district court to assess the statutory factors and for an appellate court to review that assessment.” Id. at 552, 554 (quoting Cedar Point Oil, 73 F.3d at 576). We have also identified at least two general approaches to calculating economic benefit: “ ‘(1) the cost of capital, i.e., what it would cost the polluter to obtain the funds necessary to install the equipment necessary to correct the violation; and (2) the actual return on capital, i.e., what the polluter earned on the capital that it declined to divert for installation of the equipment.’ ” Id. at 552 (quoting United States v. Allegheny Ludlum Corp., 366 F.3d 164, 169 (3d Cir.2004)). 14 In the present case, the district court determined that “the most reasonable estimate of Exxon’s economic benefit of noncompliance is $0.” Env’t Tex., 66 F.Supp.3d at 908. In making this determination, the district court appears to have concluded that Plaintiffs failed to provide credible evidence of any “pollution-control equipment” that would have “corrected]” any of the alleged violations under either of the approaches to calculating economic benefit previously identified. Indeed, the district court rejected testimony from Plaintiffs’ expert on pollution-control technology, Mr. Bowers, as “vague and unde- tailed” and found that “neither Bowers nor any other evidence credibly demonstrated that any of Bowers’s suggested capital improvements would have prevented any of the [violations].” Id. at 907-08. Notably, the court also alluded to the four environmental improvement projects from the TCEQ agreed enforcement order 15 as “an effort to reduce emissions and unauthorized emissions events” and appeared to treat these projects as an indication that Exxon did not receive any economic benefit. Id. at 908. On appeal, Plaintiffs contend that the district court erred in ignoring evidence of economic benefit stemming from Exxon’s own admissions and the TCEQ’s agreed enforcement order. We agree that the district court erred in failing to enter findings on whether the four environmental improvement projects from the TCEQ order constitute evidence of economic benefit from noncompliance. 16

Caselaw makes clear that “economic benefit” in the penalty context can be calculated as the “benefit realized by a violator from delayed expenditures to comply with the [Act].” Allegheny Ludlum Corp., 366 F.3d at 178 (emphasis added) (analyzing the “economic benefit” penalty factor under the Clean Water Act). Courts applying this factor thus often “start[] with the ‘costs spent’ or that should have been spent to achieve compliance,” then “apply an interest rate to determine the present value. of the avoided or delayed costs.” Id. (quoting United States v. Smithfield Foods, Inc., 191 F.3d 516, 530 (4th Cir.1999)). In other words, the effect of spending money to achieve compliance is often not mitigation of economic benefit — rather, plaintiffs may point to such expenditures as evidence of the regulated entity’s economic benefit to the extent the delay in making those expenditures allowed the regulated entity to use the money it gaved productively. See United States v. Gulf Park Water Co., Inc., 14 F.Supp.2d 854, 863-64 (S.D. Miss.1998). For instance, the defendants in Gulf Park were alleged to have violated the CWA by discharging pollutants into United States waters without a required NPDES permit, and in an earlier proceeding, the district court ordered them to pay a deposit necessary to connect to a regional wastewater system. Id. at 857. The defendants complied and connected to the system in 1997 — subsequently, in assessing the penalty factors under the CWA, the district court concluded there was “no doubt that the defendants ... enjoyed an economic benefit in not having expended the funds necessary to connect to the Regional system in 1985, in 1989 or in 1991.” Id. at 864. The court also noted that while the plaintiffs had the burden of establishing economic benefit, “[t]he determination of economic benefit does not require an elaborate evidentiary showing,” as it is incumbent upon the court to “endeavor to reach” an estimate that “ ‘encompass[es] every benefit that defendants received from violation of the law1 ” regardless of the inherently “imprecise” or “somewhat speculative” nature of the inquiry. Id. at 863-64 (quoting United States v. Mac’s Muffler Shop, Inc., 25 ERC 1369, 1986 WL 15443, at *8 (N.D. Ga. Nov. 4, 1986)).

Turning to the present case, the district court rejected Mr. Bowers’ expert testimony regarding specific measures that could have been taken to achieve compliance as not “credible,” and because we are extremely deferential to a district court’s assessment of witness credibility, we conclude that this was not clearly erroneous. See Canal Barge Co., Inc. v. Torco Oil Co., 220 F.3d 370, 375 (5th Cir.2000) (“We cannot second guess the district court’s decision to ... discount a witness’ testimony.”). However, Plaintiffs also elicited detailed testimony from their economic benefit expert, Mr. Shefftz, about the benefit stemming specifically from the four environmental improvement projects contained in the TCEQ agreed enforcement order, and Mr. Shefftz made clear in his testimony that the calculation of benefit with respect to these projects did not rely on Mr. Bowers’ inputs at all 17 — rather, Mr. Shefftz took implementation dates from the order itself and cost estimates from one of Exxon’s environmental coordinators and used those figures to calculate the overall benefit from delaying implementation of the TCEQ projects between 2005 and 2012. Mr. Shefftz calculated that this amount was approximately 11.7 million dollars. Based on Mr. Shefftz’s testimony, Plaintiffs argued in their proposed findings of fact and conclusions of law that Exxon gained an economic benefit of at least this amount by not implementing the TCEQ projects earlier — which, according to testimony from Exxon’s own employees in the environmental department at the Baytown facility, it could have done. Nevertheless, while the district court expressly found that Mr. Shefftz’s methodology for calculating economic benefit was reliable, it did not treat the TCEQ projects as potentially indicative of Exxon’s economic benefit from noncompliance. Rather, it treated the projects as an indication that Exxon did not receive an economic benefit. In light of the evidence adduced by Plaintiffs, we believe the .court should have made findings on the critical question of whether Exxon received a benefit from failing to implement the TCEQ projects earlier. 18

We acknowledge that both of the approaches to calculating economic benefit identified in our caselaw require. some showing that delayed expenditures would be “necessary to correct” the violations at issue in the suit. See CITGO, 723 F.3d at 552. In the present case, the district court found no credible evidence to indicate that “any of Bowers’s suggested capital improvements would have prevented any of the Events or Deviations,” with the necessary implication being that none of those improvements would “correct” the violations. Env’t Tex., 66 F.Supp.3d at 908 (emphasis added). However, because the district court failed to address Plaintiffs’ evidence that Exxon received an economic benefit from delayed implementation of the TCEQ projects, it did not consider whether those projects were necessary to correct the violations. Looking to the order in which the projects are described, the TCEQ has specified that they “will reduce emissions at the Baytown Complex, including emissions from emissions events.” Exxon Mobil Corp., Docket No. 2011-2336-AIR-E, 2012 WL 780783, at *8 (Tex. Comm’n on Envtl. Quality Feb. 29, 2012). Furthermore, some of the projects appear to be correlated in at least a general way with at least some of the violations upon which Plaintiffs have sued. For example, one project aims to “more effectively monitor and troubleshoot” a refinery flare system in order to “improve the identification and characterization of flaring events” (Count IV), and the order estimates that the projects will specifically achieve reductions in HRVOC emissions (Count III). Id. at *7-*8. Finally, the district court itself recognized in its order that the projects reflect “an effort to reduce emissions and unauthorized emissions events” at the Baytown complex. Env’t Tex., 66 F.Supp.3d at 908. Thus, given that Plaintiffs adduced evidence regarding the benefit that Exxon received from foregoing earlier implementation of these projects, we conclude that the district court erred in failing to consider that evidence and enter specific findings as to whether the projects demonstrate that Exxon received an economic benefit from noncompliance. On remand, the district court is instructed to enter such findings, which will entail consideration of whether the projects are “necessary to correct” the violations at issue in this suit. 19

iv. Duration of the Violation

Under the CAA, courts must consider “the duration of the violation as established by any credible evidence” in determining whether and to what extent a penalty should be assessed. 42 U.S.C. § 7413(e)(1) (emphasis added). The district court in this case determined that the “duration of the violation” factor weighed neither for nor against imposition of a penalty, because “the duration of each of the [violations] differ[ed] tremendously,” and Plaintiffs sought maximum penalties for each emissions, event regardless of length. Env’t Tex., 66 F.Supp.3d at 906-07. Significantly, the court found that some of the violations were of “long” duration, but because other violations were “short,” the court concluded that the factor was neutral overall. On appeal, Plaintiffs argue that the district court abused its discretion in concluding that this factor did not weigh in favor of imposing a penalty, because the court essentially viewed shorter violations as “offsetting]” the violations of longer duration.

As Plaintiffs note, there is some authority in support of the proposition that, when multiple “intermittent” violations over a span of time are at issue, a court may consider the overall length of the period during which the violations occurred (rather than assessing each violation individually). See, e.g., United States v. Midwest Suspension and Brake, 824 F.Supp. 713, 736-37 (E.D.Mich.1993); United States v. A. A. Mactal Construction Co., Inc., No. 89-2372-V, 1992 WL 245690, at *3 (D. Kan. Apr. 10, 1992). In the present case, however, the district court appears to have read the statutory language as literally mandating that the duration of each “violation” within a series of violations over time be considered. Env’t Tex., 66 F.Supp.3d at 907 (discussing the durations of individual events). We need not resolve whether the “duration of the violation” factor requires scrutiny of the length of each individual violation or allows for assessment of an overall violation period, as even assuming the former approach is a proper one, the district court abused its discretion in this case by viewing violations as effectively offsetting each other.

An example serves to effectively illustrate the nature of the district court’s error: had Plaintiffs cherry-picked from Exxon’s reports only the violations that the district court found to be “long,” the court would have been unable to use “variability in duration” as a reason to conclude that this factor was neutral. Thus, given that Plaintiffs included with these long violations a host of other short violations, it would make little sense to say that the “duration of the violation” factor somehow applies differently to the lengthy violations in light of the inclusion of the short ones. Exxon claims that because it was actually the Plaintiffs who “fail[ed] to differentiate events based on duration ... at trial,” the district court “was free to reject Plaintiffs’ all-or-nothing approach.” However, the fact that Plaintiffs sought maximum penalties for each violation, regardless of length, has no bearing on whether the district court appropriately considered “the duration of the violation” under the CAA. See Pound, 498 F.3d at 1097-98 (noting that the court must consider each statutory factor in a CAA penalty analysis). Indeed, it is undisputed that Plaintiffs’ spreadsheets set forth the individual durations of the events at issue, and Plaintiffs’ repeatedly made clear the overall time period within which violations were alleged to have occurred. Thus, because the district court opted to consider the durations of violations individually, it should have considered whether any violation, standing alone, was sufficiently long to justify imposition of a penalty. Its failure to do so, and its decision to instead view the factor as neutral based on overall duration variability, was an abuse of discretion.

v. Seriousness of the Violation

In assessing the “seriousness of the violation” penalty factor, courts outside of this circuit have looked to the “risk or potential risk of environmental harm” posed by emissions/discharges and have found violations to be serious “even absent proof of actual deleterious effect.” Pound, 498 F.3d at 1099. Furthermore, courts have recognized that the overall number and quantitative severity of emissions or discharges may properly be relied upon as evidence of seriousness. See Pub. Interest Research Grp. of N.J., Inc. v. Powell Duffryn Terminals Inc., 913 F.2d 64, 79 (3d Cir.1990) (holding that the district court properly relied upon “the large number of gross exceedances in concluding that [the defendant’s] violations were serious”). In the present case, the district court began its assessment of this factor by noting that “each of the Events and Deviations differ tremendously,” with some of the events being “more serious because they emitted higher quantities of emissions” and “many more” of the events being “less serious.” Env’t Tex., 66 F.Supp.3d at 908-09. Thus, in “considering the amount of emissions,” the district court determined there were many more violations that were “not serious or less serious than were more serious.” Id. at 909. The court then went on to examine whether Exxon’s violations “adversely affect[ed] human health or the environment” and concluded there was no “credible evidence” that any of the events “even potentially” did so. Id. As with its analysis of the “duration of the violation” factor, we conclude that the district court abused its discretion in using “less serious” violations to essentially offset violations of a concededly “more serious” nature based on emission quantity alone.

The CAA instructs a court considering penalties to assess “the seriousness of the violation.” 42 U.S.C. § 7413(3)(1). When multiple violations are at issue, balancing “more serious” violations against “less serious” ones clearly runs contrary to this instruction, with the result being that serious violations become less so if accompanied by a sufficient number of insignificant violations. In other words, given the district court’s recognition that some of the emissions in this case were large enough to be considered “more serious,” we think it was an erroneous application of the “seriousness of the violation” factor to conclude that the existence of thousands of additional, smaller violations somehow tipped the scale against assessment of a penalty. If anything, the inclusion of many more violations with the most serious ones would only increase the overall degree of seriousness, rather than lessening it. See Powell Duffryn, 913 F.2d at 79. Thus, in light of the district court’s explicit recognition that some of the violations were more serious based on the amount of emissions alone, we conclude that it was an abuse of discretion to view this factor as weighing against imposition of a penalty due to the existence of thousands of additional, “less seri ous” violations. 20

3. Permanent Injunction

The final point of error asserted by Plaintiffs concerns the district court’s refusal to grant a permanent injunction prohibiting Exxon from committing further permit violations. “We review a district court’s grant or denial of injunc-tive relief for abuse of discretion.” Aransas Project v. Shaw, 775 F.3d 641, 663 (5th Cir.2014). “The party seeking a permanent injunction must ... establish (1) success on the merits; (2) that a failure to grant the injunction will result in irreparable injury; (3) that said injury outweighs any damage that the injunction will cause the opposing party; and (4) that the injunction will not disserve the public interest.” VRC LLC v. City of Dallas, 460 F.3d 607, 611 (5th Cir.2006) (citing Dresser-Rand, Co. v. Virtual Automation, Inc., 361 F.3d 831, 847-48 (5th Cir.2004)). The district court in this case denied the request for an injunction based on its finding that any injury to the public would not outweigh the damage an injunction would cause Exxon. Env’t Tex., 66 F.Supp.3d at 913. Regarding the injury to the public, the district court found that any future unauthorized emissions would not be any “more harmful to the public or the environment than past [emissions] were.” Id. With respect to the damage to Exxon, the district court found that granting a permanent injunction would be “excessively intrusive” because it would “entail continuing superintendence of the permit holder’s activities by a federal court.” Id. (quoting Friends of the Earth, Inc. v. Laidlaw Envtl. Servs. (TOC), Inc., 528 U.S. 167, 193, 120 S.Ct. 693, 145 L.Ed.2d 610 (2000)).

On appeal, and although their argument is rather cursory, Plaintiffs appear to contend that the district court abused its discretion by “rel[ying] on [the] clearly erroneous factual finding[]” that imposition of an injunction would significantly burden Exxon. See Aransas, 775 F.3d at 663 (noting that it is an abuse of discretion to “rel[y] on, clearly erroneous factual findings when deciding to grant or deny the permanent injunction”). Plaintiffs argue that Exxon “already tracks and reports compliance with its Title V permits” and “has immense financial resources which could be devoted to improving compliance,” and thus the district court erred in concluding that “even an injunction that did no more than require Exxon to prove it is complying with its permits” would be excessively burdensome. However, although it may be true that Exxon has the resources to comply with any injunction, we do not believe that fact alone establishes clear error in the district court’s finding that forcing Exxon to continuously prove its compliance with the CAA would be “excessively intrusive.” Thus, we conclude that the district court did not abuse its discretion in declining to grant a permanent injunction in this case.

III. CONCLUSION

In sum, we conclude that the district court erred in its analysis of Exxon’s liability under Counts I through IV and abused its discretion in assessing three of the CAA’s mandatory penalty factors. We accordingly VACATE the district court’s judgment and REMAND for assessment of penalties based on the violations that are properly considered “actionable” in light of this opinion. 21

1 . An "emissions event” is defined under Texas law as "[a]ny upset event or unscheduled maintenance, startup, or shutdown activity, from a common cause that results in unauthorized emissions of air contaminants from one or more emissions points at a regulated entity.” 30 Tex. Admin. Code § 101.1(28). “Unauthorized emissions” are in turn defined as "[e]missions of any air contaminant except water, nitrogen, ethane, noble gases, hydrogen, and oxygen that exceed any air emission limitation in a permit, rule, or order of the commission...Id. § 101.1(108).

2 . A "deviation" is defined under Texas law as "[a]ny indication of noncompliance with a term or condition of the permit as found using compliance method data from monitoring, recordkeeping, reporting, or testing required by the permit and any other credible evidence or information.” 30 Tex. Admin. Code § 122.10(6).

3 . Because Plaintiffs claimed many of the violations were "ongoing,” the period of alleged violations ultimately extended through September of 2013.

4 . While Plaintiffs do not offer an explanation for this variance, a simple review of the record provides an obvious one: human error. The summary tables for Plaintiffs’ Count II violations are identical to the summary tables for the Count I violations, with only the numbers and headings changed. Thus, when the Count II tables were used to make the Count I tables, it is likely that the "violations” heading was mistakenly left unaltered.

5 . Exxon "[did] not dispute” in the court below that "the alleged violations under Count[ ] II ... constitute violations of an emission standard or limitation.” Env't Tex., 66 F.Supp.3d at 893 n. 153. Exxon attempts to argue on appeal that it “never admitted” any entries under Count II were violations, "and the district court plainly understood that position since it did not find liability on all of the allegations in” that count. However, Exxon's argument — at least with respect to alleged violations under Count II — clearly runs contrary to its clarification at trial that "if Exxon Mobil files a reportable STEERS event, for example, in essence, it is making a report of releases that exceed, for example, an hourly limit with respect to a particular release.” Furthermore, the fact that STEERS reports and on-site records of "emissions events” reflect violations of emission standards or limitations does not mean that a defendant is per se liable under the citizen suit provision of the CAA. See 42 U.S.C. § 7604(a)(1) (providing that violations must be "repeated” or ongoing). Thus, the fact that the district court "did not find liability on all of the” events under Count II is not proof that the district court believed many of the events counted as violations by Plaintiffs were not, in fact, violations. Finally, with respect to Exxon's argument that specific entries in which the emission quantity — standing alone — would appear to fall below the applicable listed threshold were not shown to be violative of MAERT limits, we note that we were unable to locate in the record any point at which Exxon contested these entries before the district court. Rather, Exxon's proposed findings of fact and conclusions of law focused on whether Count II violations were "repeated” or "ongoing," and when asked directly by the district court "which events from the stipulated tables” it "claim[ed] [did] not constitute a violation,” Exxon did not mention Count II. Thus, to the extent Exxon asks us to conclude that most of the Count II violations, including a number of the violations that the district court found actionable, were not violations at all based on . an argument it never raised below, we decline to do so. See, e.g., Violette v. Smith & Nephew Dyonics, Inc., 62 F.3d 8, 11 (1st Cir.1995) (a defendant may not "evade the scrutiny of the district court” by raising a new defense on appeal "in order to create essentially a new trial”).

6 . The district court also noted a third alternative: namely, showing a “continuing likelihood of recurrence.” However, as the district court recognized, “Plaintiffs do not claim satisfaction of the third method of proving ac-tionability.” Env’t Tex., 66 F.Supp.3d at 894.

7 . We acknowledge, as the district court in this case did, that “[t]he 'to be in violation’ provision in the CAA is identical to the 'to be in violation' provision in the CWA,” and “[i]n-terpretations of the CWA provision are instructive when analyzing the CAA provision.” Env’t Tex., 66 F.Supp.3d at 894 n. 154; see also United States v. Anthony Dell’Aquilla, Enters. & Subsidiaries, 150 F.3d 329, 338 n. 9 (3d Cir.1998).

8 . We also note, once again, that on remand, the district court should consider the overlap between Plaintiffs' Count I and Count II claims with respect to refinery emissions.

9 . Indeed, it is for this very reason that we declined to address Exxon’s argument regarding whether specific emissions under Count II exceeded MAERT limits&emdash;Exxon never contested those emissions as violations below, and the district court rightly understood there to be no dispute on the point.

10 . We do not hold that deviation reports will always be insufficient to prove actual permit violations. Indeed, we note that some of the Count IV violations stem from information contained in deviation reports. We only conclude that, based on the record before us in this case, there was no error in the district court's refusal to find an actionable violation in every Count VII deviation. Because Plaintiffs chose to rely exclusively on the existence of the deviation reports, with little attempt to clarify their significance, as proof of hundreds of violations of different regulatory requirements, we see no basis (and Plaintiffs have not provided one on appeal) to disagree with the district court’s resolution of Count VII.

11 . Plaintiffs also sought attorneys’ fees and costs and appointment of a “special master” to monitor implementation of relief, but these are not at issue in the present appeal.

12 . Exxon expends considerable brief space rebutting what it views as "Plaintiffs' theory that civil penalties are mandatory.” However, it is fairly obvious from Plaintiffs’ briefing that they concede the point and do not argue that the district court was required to impose penalties as a matter of law.

13 . Plaintiffs raise two additional arguments in their briefing: first, they argue that Exxon's own calculations regarding the potential for implementation of additional emissions-re ducing technologies at the complex demonstrate that at least some of the violations were preventable, cutting against Exxon’s “good faith efforts to comply.” However, what Plaintiffs fail to mention is that Exxon's witness merely used calculations done by Plaintiffs as to the amount of emissions that could have been prevented by implementation of the relevant technology in order to show that even “giving every benefit of the doubt” to "Plaintiffs’ theory,” implementation of the technology would not have been economically reasonable. As such, Plaintiffs have not shown clear error on, this basis.

Second, Plaintiffs note the inconsistency between the district court’s conclusion that no "improvements could have been made to prevent recurrence” and the court’s finding that the four TCEQ-mandated environmental improvement projects "will reduce unlawful emissions.” For reasons we discuss in connection with the "economic benefit of noncompliance” factor, we do perceive a tension between the district court’s finding that the projects reflect "substantial efforts to improve ... compliance” and its finding that no improvements could have "prevented” any emissions events involved in this case. Env’t Tex., 66 F.Supp.3d at 905. Nevertheless, in light of the district court’s other reasons for weighing this factor as it did, we cannot conclude that this tension alone renders the court’s overall resolution of the "compliance history and good faith efforts to comply” factor an abuse of discretion. See CITGO, 723 F.3d at 551 (referring to our review of penalty factors as “highly deferential”).

14 . We have also recognized multiple approaches to the related question of how to "set the amount of the penalty” overall: "some courts use the ‘top down' approach in which the maximum penalty is set ... and reduced as appropriate considering the ... enumerated [penalty factors] as mitigating factors, while other courts employ the ‘bottom up’ approach, in which economic benefit is established, and the remaining ... elements ... are used to adjust the figure upward or downward.” Id. In the present case, the district court appears to have simply weighed the factors independently without adopting either approach — as it felt it had discretion to do — and the court noted that the result would be the same under any approach. Env’t Tex., 66 F.Supp.3d at 912 n. 267. Regardless, Plaintiffs do not challenge the district court’s overall approach to "set[ting] the amount of the penalty.”

15 . The four improvement projects are (1) plant automation venture — installing computer programs to monitor, identify, diagnose, and guide operations, which will help identify potential events so they can be addressed proactively; (2) fuels north flare system monitoring/minimization: additional monitoring probes and "on-line analyzers” that improve sensing and characterizing flaring events; (3) BOP/BOPX recovery unit simulators: developing and using "high-fidelity process training simulators” to improve operator training and reduce emissions events; (4) enhanced fugitive emissions monitoring: using infrared technology to locate leaks. Exxon Mobil Corp., 2012 WL 780783, at *8-*9.

16 . Plaintiffs’ argument about Exxon’s "admissions” regarding implementation of emissions-reducing technologies is the same argument that we addressed, and rejected, in footnote 13, supra.

17 . Mr. Shefftz did testify that the improvement projects should be considered a “subset of the total amounts” drawn from Mr. Bowers’ testimony, but this does not undercut the validity of the TCEQ projects as independent evidence of economic benefit. Mr. Shefftz’s point was simply that because Mr. Bowers’ suggested improvements would ostensibly encompass every measure needed to bring Exxon's facility to 0 permit violations, the $11.7 million benefit calculated by Mr. Shefftz with respect to the TCEQ projects should not be added on to the overall estimate of economic benefit. But this does nothing to suggest that the district court’s rejection of Mr. Bowers’ testimony also somehow precludes consideration of the TCEQ projects as evidence of economic benefit. Indeed, Mr. Shefftz explained that "[cjonceptually, it's a subset of [the overall estimate], but ... it’s also independent of [Bowers’] expert opinion.”

18 . Two arguments raised by Exxon warrant mention here: first, Exxon insisted at oral argument that the TCEQ order represents a regulatory decision to “forgo” maximum penalties in favor of other "corrective action,” and allowing a citizen suit to “capitalize” on the economic costs in the order by using them as economic benefits would be unfair — but this argument about compliance efforts "negating” economic benefit is precisely the argument that various courts have rejected under the economic benefit factor. As one district court recently stated, economic benefit is not “now negated by the fact that [the defendant] is working to remedy the issues ...,” because “the fact that [the defendant] must pay to bring his facility into compliance ... does not excuse his history of noncompliance.” Idaho Conservation League v. Magar, 2015 WL 632367, at *5 (D. Idaho 2015). Relatedly, in its response to Plaintiffs’ 28(j) letter, Exxon quoted the Supreme Court's opinion in Gwaltney for the proposition that allowing citizens to “file suit ... in order to seek the civil penalties that [a regulatory agency] chose to forgo” would "curtail[] considerably” the regulator’s "discretion to enforce the Act.” Gwaltney of Smithfield, Ltd. v. Chesapeake Bay Found., Inc., 484 U.S. 49, 61, 108 S.Ct. 376, 98 L.Ed.2d 306 (1987). Exxon's reference to Gwaltney is unhelpful, however, because the Supreme Court's discussion on this point was in the context of interpreting the Clean Water Act’s "to be in violation” language as barring citizen suits based on "wholly past violations of the Act.” Id. at 60, 108 S.Ct. 376. Yet as both parties acknowledge, the Clean Air Act was amended in 1990 to authorize citizen suits against any person "who is alleged to have violated” the Act. 42 U.S.C. § 7604(a)(1) (emphasis added). This amendment has been viewed as a direct response to Gwaltney, and indeed, neither party in this case disputes that the "to have violated” language authorizes citizen suits based on wholly past violations of the CAA. See Atl. States Legal Found., Inc. v. United Musical Instruments, U.S.A., Inc., 61 F.3d 473, 477 (6th Cir.1995) (recognizing that "after Gwaltney, Congress amended the Clean Air Act ... explicitly to allow citizen suits for purely historical violations....”). Thus, the proposition to which Exxon's Gwaltney quotation lends support appears to no longer apply in CAA citizen suits.

19 . In making its findings on remand, the court should be mindful that the economic benefit estimate must " 'encompass every benefit that defendants received from violation of the law’ " regardless of the inherently speculative nature of the inquiry. Gulf Park, 14 F.Supp.2d at 864 (quoting Mac's Muffler Shop, 1986 WL 15443 at *8). We thus believe that compliance expenditures or projects need not be tied specifically to prevention of each violation in order to establish that they are “necessary to correct” the violations overall, particularly in a case such as this where the violations are extensive and varied. Rather, we believe the inquiry should center on whether the projects will ameliorate the kinds of general problems that have resulted in at least some of the permit violations upon which Plaintiffs have sued.

20 . Because the district court acknowledged that for at least some of the emissions events, the sheer quantity of pollutants emitted made the violations "more serious,” we do not find it necessary to address the other asserted errors raised by Plaintiffs with respect to this factor.

21 . We note that the district court declined to address the applicability of any affirmative defenses, as it was unnecessary given the decision not to award penalties. Because we vacate that decision, however, we recognize that the district court may well be called upon to rule on Exxon’s claimed affirmative defenses on remand. See, e.g., 30 Tex. Admin Code § 101.222(b) (setting forth an affirmative defense for "[n]on-excessive upset events”).