OPINION OF THE COURT
BECKER, Circuit Judge.The American Lung Association and other clean air advocacy groups brought this “citizens” suit in the district court for the District of New Jersey to compel New Jersey to promulgate and implement a system of ground-level ozone emission regulations. Under the Clean Air Act, 42 U.S.C. §§ 7401-7642 (1982), the United States Environmental Protection Agency (“EPA”) is given the authority to promulgate maximum levels for air-borne pollutants. Each state is required to adopt a State Implementation Plan (“SIP”) that details what measures it will take to ensure that the air in its region does not contain more than the federally determined acceptable levels of pollutants. See generally Concerned Citizens of Bridesburg v. EPA, 836 F.2d 777, 780-81 (3d Cir.1987) (describing structure created by Clean Air Act). American Lung Association and the other plaintiffs argued in the district court that New Jersey has obligated itself in its SIP to adopt and enforce several types of regulations to limit emissions of ozone, but has failed to do so.
In the first phase of a bifurcated proceeding, the district court granted plaintiffs’ motion for summary judgment against the state defendants (New Jersey Governor Thomas Kean, the New Jersey Department of Environmental Protection (“NJDEP”), and Richard T. Dewling, the Department’s Commissioner). 670 F.Supp. 1285. EPA was also named as a defendant, but it sided with plaintiffs against the state defendants. The district court rejected New Jersey’s contention that it had obligated itself only to investigate the regulatory programs outlined in the SIP, and declared instead that New Jersey was required to embark on these schemes of regulation. The court denied motions by several petroleum industry trade associations to intervene in this portion of the trial, although that industry would be heavily affected by the regulations.
In the second phase of the trial, which was dedicated to setting a schedule for promulgation and implementation of the regulations, the trade associations were allowed to intervene by consent of the parties.1 The district court adopted the compliance schedule suggested by New Jersey, which had been approved by the plaintiffs and by EPA. It rejected a longer timetable proposed by the trade associations. The trade associations have brought this appeal seeking to challenge both the finding of liability against New Jersey and the validity of the compliance schedule adopted by *322the district court. They raise three contentions.
First, the trade associations challenge the jurisdiction of the district court to entertain a citizens’ suit against the state in its regulatory capacity. We conclude, however, that the plain language of the Clean Air Act permits it. Second, one of the trade associations, the American Petroleum Institute (“API”) seeks to challenge the district court’s holding that New Jersey was required to adopt and enforce the regulatory programs. We conclude, however, that since the trade associations were denied intervention in the liability phase of the trial and did not appeal that determination, they may not challenge the determination of liability via this appeal.
Third, the trade associations seek to invalidate the scheduling order entered by the district court. They assert that the schedule was set with undue haste, and they challenge the substance of the schedule. We conclude, however, that under the circumstances, the district court acted properly in allowing slightly more than a month for the submission of proposed timetables and that it acted within its discretion in adopting the timetable proposed by NJDEP for the promulgation and implementation of ozone emission regulations. We will therefore affirm.
I. BACKGROUND
A.
The Clean Air Act of 1970, 42 U.S.C. §§ 7401-7642, was enacted in response to the growing threat that air pollution poses to human health. See id. § 7401. The statute creates a program of cooperative federalism for achieving cleaner air. The EPA is given the responsibility for setting National Ambient Air Quality Standards (“NAAQS”), which set maximum permissible levels for certain air-borne toxins. Id. § 7409. It is then up to each state to produce an implementation plan to reduce emissions from pollution sources within the state so that it complies with the NAAQS. Id. § 7410. These SIPs are promulgated by state agencies after notice and comment and must be approved by the EPA after it conducts its own notice and comment proceedings. Id. The SIPs are not merely advisory; once EPA approves a SIP the state is obligated to comply with it. Id. § 7413(a)(2). The Clean Air Act of 1970 provided that all states must be in compliance with the NAAQS within three years after the adoption of their SIPs or, in other words, by the mid-1970s. Id. § 7410(a)(2)(A).
By 1977, however, many states contained regions that were still not in compliance with the NAAQS. Congress amended the Clean Air Act to adjust to this fact. The 1977 amendments provided that states not yet in attainment must attain compliance with most NAAQS “as expeditiously as practicable, but ... not later than December 31, 1982.” 42 U.S.C. § 7502(a)(1). However, the amendments created a conditional exception to this deadline for two regulated toxins. States were given until December 31, 1987, to come into compliance with the ozone and carbon monoxide NAAQS, but only if the states met several conditions. In relevant part, the states first had to be able to establish that they could not have come into compliance earlier despite their adoption of all reasonably available measures. Id. § 7502(a)(2). Second, the states were required to “identify other measures necessary to provide for attainment of the applicable national ambient air quality standard” beyond the “reasonably available” measures and adopt them. Id. § 7502(b)(ll)(C). Third, the SIPs had to contain “enforceable measures to assure attainment” of the NAAQS “as expeditiously as practicable but not later than December 31, 1987.” Id. §§ 7502(c), 7502(a)(2).
B.
This case involves the problem of ozone pollution in New Jersey. Ground-level ozone is a primary element of smog. Prolonged exposure to ozone in levels exceeding the NAAQS can cause breathing difficulties, chest tightness or pain, coughing, headaches or nausea in otherwise healthy people. And ozone emissions are particu*323larly problematic for people with asthma and other respiratory ailments. See Affidavit of Fred M. Jacobs, April 22, 1987.
At issue is the revised SIP adopted by New Jersey and approved by EPA in 1980, in which New Jersey obligated itself to a program that would bring it into compliance with the ozone NAAQS by the end of 1987. Plaintiffs’ suit, brought in January 1987, contended that New Jersey had failed to comply with its SIP by failing to promulgate and implement several types of pollutant regulations according to the timetable contained in the SIP. This appeal involves regulations concerning the installation of devices to limit the escape of ozone during automobile refueling (“Stage II controls”)2 and during the loading of barges carrying gasoline (“barge loading controls”).
C.
As noted above, the district court bifurcated the trial. The first phase was to determine whether the New Jersey SIP required the promulgation of regulations mandating the controls in question. The second phase was to set a schedule for compliance if such regulations were required. In April 1987, two groups of trade associations — the American Petroleum Institute and the National Association of Convenience Stores (“NACS”) and similar associations — moved to intervene as defendants for both phases. On July 24, 1987, their motions were denied, but API was allowed to file a brief amicus curiae. In the first phase of the proceedings, the district court granted plaintiffs’ motion for summary judgment against the state defendants, holding that the SIP required NJDEP to promulgate regulations to implement Stage II controls and barge loading controls. The court rejected New Jersey’s argument, and the argument of API as amicus, that the language of the SIP had obligated New Jersey to do no more than investigate these areas of regulation.
Prior to the second phase of the trial and pursuant to consent of all parties, the court permitted limited intervention by the trade associations “for the sole purpose of addressing the ... remedies required as a result of the Court’s [liability] opinion.” Consent Order Granting API Leave to Intervene as Party Defendant at 2 (Oct. 5, 1987); Consent Order Granting NACS, et al., Leave to Intervene as Party Defendants at 2 (Oct. 29, 1987). The parties and the intervenors were given slightly more than a month from the time of the judgment of liability to submit proposed schedules for implementation of the required regulations. See Dist.Ct.Op. at 2 (Nov. 19, 1987) [hereinafter “Scheduling Opinion”] [1987 WL 31764],
NJDEP proposed a schedule which was endorsed by the plaintiffs and by EPA. The trade associations proposed a much longer timetable. The court adopted the schedule suggested by NJDEP. See Scheduling Opinion at 16-17. NJDEP has already promulgated the first stage II regulations, and the state defendants do not appeal from the district court judgment for liability or from the scheduling order. The trade associations, however, have brought this appeal seeking to overturn the district court’s judgments with respect to both the conclusion that New Jersey had obligated itself to adopt Stage II and barge loading controls and the eourt-adopted timetable for promulgation and implementation of the regulations.3 The trade associations have not appealed the denial of their motion to intervene in the first portion of the trial.
II. JURISDICTION
API asserts that the district court had no jurisdiction over a suit brought *324against a regulator, as opposed to a polluter. Plaintiffs respond that jurisdiction was proper in the district court under section 304 of the Clean Air Act (codified as 42 U.S.C. § 7604), the citizens’ suit provision. Section 304 provides in relevant part that
any person may commence a civil action on his own behalf—
(1) against any person (including ... (ii) any ... governmental instrumentality or agency to the extent permitted by the Eleventh Amendment to the Constitution) who is alleged to be in violation of (A) an emission standard or limitation under this chapter....
42 U.S.C. § 7604(a). The Clean Air Act defines “emission standard or limitation” in relevant part as follows:
any condition or requirement under an applicable implementation plan relating to ... vapor recovery requirements....
42 U.S.C. § 7604(f)(3). API does not dispute that the Stage II and barge loading controls at issue in this case are requirements under a SIP that relate to vapor recovery. And the statute makes it clear that state agencies can be defendants in a citizens suit. Since NJDEP is a government agency in violation of an emission standard, the district court concluded that the language of section 304 “clearly provides this court with jurisdiction over this case.” 670 F.Supp. 1285,1292 (1987) [hereinafter “Liability Opinion”].
API contends that, notwithstanding the language of section 304, the Clean Air Act does not give the district court jurisdiction to entertain suits by citizens against the state in its capacity as regulator, rather than its capacity as polluter. We must consider this argument regardless of whether, as discussed below, API may challenge the judgment of liability, because a federal court has “a special obligation to ‘satisfy itself not only of its own jurisdiction, but also that of the lower courts in a cause under review.’ ” See Bender v. Williamsport Area School District, 475 U.S. 534, 541, 106 S.Ct. 1326, 1331, 89 L.Ed. 2d 501 (1986) (quoting Mitchell v. Maurer, 293 U.S. 237, 244, 55 S.Ct. 162, 165, 79 L.Ed. 338 (1934)).
API bases its argument on Citizens Association of Georgetown v. Washington, 535 F.2d 1318 (D.C.Cir.1976) (per curiam). In that case, the D.C. Circuit found itself to be without power to adjudicate plaintiff’s demand that the District of Columbia revoke construction permits for certain projects. That court stated that “the legislative history ... quite clearly indicates that section 304(a)(1) confers federal jurisdiction only over suits against polluters, and, under certain conditions, the Administrator of the EPA.” 535 F.2d at 1321. The legislative history that the Georgetown court refers to is congressional debate on the 1970 version of the Act, which included statements by several Congressmen that under section 304 citizens would be able to sue polluters, but does not include any mention of suits against regulators. See id.
The fundamental problem with this argument is that, even if we accept the D.C. Circuit’s conclusion, Georgetown is a pre-1977 case, construing the pre-1977 statute through its legislative history, whereas the definition of “emission standard or limitation” which the plaintiffs and the district court rely on here was adopted as part of the 1977 amendments to the Clean Air Act. The D.C. case thus does not speak to the jurisdictional issue in this case.
Turning to the legislative history of the 1977 amendments to section 304, we find that, like the earlier legislative history, it contains no reference to suits by citizens to compel regulation. For example, a proposed amendment to the section explicitly stated that citizens would be allowed to sue “sources” but did not mention state regulators. See H.ContRep. No. 564, 95th Cong., 1st Sess. 173, reprinted in 1977 U.S.Code Cong. & Admin.News 1077,1553-54. Furthermore, the amendment to the statute that plaintiffs point to is a change in the definition of the term “emission standard,” rather than a change in the first portion of section 304 which sets out who can be sued.
Despite these arguments, we conclude that we do have jurisdiction under section 304 to adjudicate citizens’ suits against the *325state in its regulatory capacity. “Where the language of the statute is clear, only ‘the most extraordinary showing of contrary intentions’ justifies] altering the plain meaning of a statute.” Malloy v. Eichler, 860 F.2d 1179, 1183 (3d Cir.1988) (quoting Garcia v. United States, 469 U.S. 70, 75, 105 S.Ct. 479, 482, 83 L.Ed.2d 472 (1984)). The language of a statute tells us what the majority of Congress agreed upon; bits of legislative history tell us less than that and generally do not trump the facial meaning of the statute.
As noted above, the explicit language of the Clean Air Act permits this suit, since the regulations at issue are requirements relating to vapor recovery.4 And there is no contrary indication in the legislative history to the 1977 amendments that could overcome this facially valid reading of the statute. The revisions to section 304 were not extensively debated, and the discussions that do exist relate to earlier versions of the amendment. At most there is an omission of the discussion as to the issue of suits against regulators. In some circumstances such an omission might indicate that Congress did not wish to alter the prior meaning of the section in this respect. This argument fails here, however, because there was not a universally accepted interpretation of section 304 prior to 1977. For example, in the same year in which Georgetown was decided, the Second Circuit allowed a suit against state regulators requiring them to enforce provisions of a SIP, albeit without discussing the jurisdictional issue. See Friends of the Earth v. Carey, 535 F.2d 165 (2d Cir.1976).5 The omission of discussion on the issue can thus just as easily be seen as ratifying the Second Circuit practice of allowing such suits, as the D.C. Circuit practice of disallowing them.
Additionally, we note that the 1977 amendments represented a trade-off with the states in response to the fact that many states had either been dilatory in meeting, or were unable to meet, their responsibilities under the Clean Air Act. States were given more time to comply with the NAAQS, but the amendments added teeth to ensure state compliance. See Currie, Relaxation of Implementation Plans Under the 1977 Clean Air Act Amendments, 78 Mich.L.Rev. 155, 187 (1979) (describing the 1977 amendments as federal “blackmail” of the states to ensure compliance with the requirements of the Clean Air Act). For example, the 1977 amendments, for the first time, conditioned some types of federal funding on state compliance with the Clean Air Act. See 42 U.S.C. §§ 7506(a), 7616(b)(2). Permitting citizens to sue states to make sure they enact the regulatory schemes to which they have committed themselves furthers this Congressional goal.
We therefore conclude that the district court did have jurisdiction to decide this case. We next turn to the question whether API may properly challenge in this appeal the district court’s finding of liability against the state defendants.
III. THE RIGHT OF API TO CHALLENGE THE LIABILITY DECISION6
As noted above, API moved to intervene as a party defendant for purposes of both the liability and scheduling phases of the trial. See Notice of Motion of API for Leave to Intervene (April 23, 1987). This *326motion was denied. See Order (July 24, 1987). By consent of all the .parties, API was subsequently allowed to intervene “for the sole purpose of addressing the remedies required as a result of the [liability] opinion.” Consent Order Granting API Leave to Intervene as Party Defendant at 2 (Oct. 5, 1987). Although API could have, as part of this appeal, appealed the denial of intervention with respect to the liability phase of the trial, see Stringfellow v. Concerned Neighbors in Action, 480 U.S. 370, 107 S.Ct. 1177, 1182, 94 L.Ed.2d 389 (1987), it did not do so. See API Notice of Appeal (Dec. 18, 1987).
This Court has held that “one who is properly denied intervention cannot appeal the merits of the case.” Harris v. Pernsley, 820 F.2d 592, 603 (3d Cir.), cert. denied sub nom. Castille v. Harris, — U.S. -, 108 S.Ct. 336, 98 L.Ed.2d 363 (1987). There is a good reason for this rule. When a district judge denies a motion for intervention, the judge is making a determination that the party trying to intervene does not have a sufficient interest in the matter being adjudicated to require joinder and that interests in convenience do not mandate joinder. See Fed.R.Civ.P. 19, 20. See also 7 C. Wright, A. Miller & M. Kane, Federal Practice and Procedure § 1652 at 372 (1986). There is no reason to allow a party to challenge on appeal issues which it could not litigate at the trial level because its interest was deemed insufficient. The district court denied API’s motion to intervene, either of right or permissively, in the liability phase of the trial. As this decision was never appealed, we must presume that it is correct. Thus, API may not challenge the liability decision of the district court.
In opposition to this, API argues that this Court’s decision in Cerro Metal Products v. Marshall, 620 F.2d 964, 969 (3d Cir.1980), holds that limited intervenors are allowed to appeal the full final judgment of the court. We do not believe that that case stands for so broad a proposition. In Cerro, we allowed an intervening union to appeal a motion that had been decided before it intervened, despite the fact that in allowing the intervention the district court had said that the union would not have the right to require the redetermination of issues already decided in the case. However, that case is distinguishable from the case at bar in three ways.
First, the Cerro court rested its decision on a factual determination that the district court’s limitation did not “purport[ ] to foreclose the Union’s right to appeal the prior ... order.” 620 F.2d at 969 (emphasis added). By its terms, it purported to foreclose the union only from district court redetermination of determined issues. See id. In the instant case, on the other hand, the district court did purport to foreclose API from appealing the merits by a district court order that API never appealed. The language of the consent order that allowed API to intervene solely in the scheduling phase of the trial does not repudiate the court’s prior denial of its motion to intervene in the liability phase, and that denial certainly purports to foreclose API from participating as a party in the liability phase both in the district court and on appeal.
Second, the Cerro court was concerned that the district court might have been attempting to condition an intervention as of right, an action which might be beyond its authority. See id. at 969 & nn. 6 & 7. In contrast, API’s motion to intervene as of right was denied; thus API was only a permissive intervenor in the scheduling phase of the trial. The Cerro court’s concern about conditioning mandatory intervention is therefore inapplicable to the case at bar.
Finally, the Cerro court concluded that the general rule is “that an intervenor may appeal from any order adversely affecting the interest that served as a basis for intervention.” Id. at 969. As API was an intervenor only with respect to schedule-setting, its sole basis interest was in the timing of the implementation of the regulations. Thus application of the “general rule” articulated in Cerro forecloses API’s attempt to appeal the grant of summary judgment with respect to liability. The fact that appeal was allowed despite a conditional grant of intervention in Cerro is thus inapposite to the case at bar.
Alternatively, API contends that it should be allowed to appeal as a non-party *327whose interests are affected by the decision. Other circuits have allowed such appeals, but only in extraordinary cases, where required by the equities. See, e.g., Martin-Trigona v. Shiff, 702 F.2d 380, 385-86 (2d Cir.1983); cf. Citibank International v. Collier-Traino, 809 F.2d 1438, 1441 (9th Cir.1987) (denying appeal by a non-party because the case was not sufficiently "exceptional”)- However, none of these cases involved a non-party whose pri- or motion to intervene had beén denied. To allow API to appeal the merits despite the fact that it did not appeal the denial of intervention would constitute a sub silentio reversal of the district court’s judgment that API did not have sufficient interest to be included as a party in the adjudication of the merits of the case. We will not permit API to make this end-run around the procedures for appellate review that were readily available to it.
Furthermore, even if we were to adopt an “exceptional circumstances” test, this case does not present the kind of exceptional situation in which justice will be hindered if a non-party is not allowed to appeal. API can protect its interests by contesting in state court the validity of the NJDEP regulations as they are promulgated, or by lobbying New Jersey to amend its SIP.
We conclude that since API has not appealed the denial of its motion to intervene in the liability phase of the trial, it may not challenge the district court’s determination of liability. Thus the only claim remaining for us to adjudicate is the validity of the district court’s scheduling order.
IV. THE VALIDITY OF THE SCHEDULING ORDER
The citizens’ suit provision of the Clean Air Act gives the district court the authority to “enforce ... emission standard[s] or limitation[s].” 42 U.S.C. § 7604(a). As discussed above, see supra at 323, the emission standard at issue in this case was the SIP requirement that New Jersey promulgate and implement Stage II and barge loading controls. However, in this case it was impossible for the court simply to enforce the schedules set in the SIP for implementation and promulgation because most of the dates set in the SIP had long since passed. See New Jersey SIP for Ozone and Carbon Monoxide at 54-55 (Sept. 1983). The district court was thus presented with the task of revising the schedule, consonant with the purpose of the Clean Air Act, to ensure that New Jersey came into compliance with the ozone NAAQS “as expeditiously as practicable.” 42 U.S.C. § 7502(a).
The scheduling order entered by the district court is an equitable order, made within the ambit of the district court’s discretion to fashion appropriate remedies. We will set aside such an order only if it constitutes an abuse of discretion. See Kingstate Oil v. M/V Green Star, 815 F.2d 918, 922 (3d Cir.1987). Appellants challenge both the way in which the scheduling order was set and the substance of the order.
A.
All parties were given thirty-six days to submit proposed scheduling orders to the court. NACS argues that the district court abused its discretion in setting so brief a period for development of a schedule, because it did not afford NJDEP time to hold hearings and thoroughly investigate the feasibility of a schedule for implementing Stage II controls.7 As NACS put it, “there was no compelling deadline *328which dictated the District Court’s immediate action,” NACS Brief at 13, and thus the district court should have given the litigants at least several more weeks so that the agency could gather and deliberate about all relevant information.
We disagree with NACS’s implication that speed was not of the essence in setting the timetable for implementation of the regulation. In adopting the 1977 amendments to the Clean Air Act, which provided that states must be in compliance with the ozone NAAQS “as expeditiously as practicable but not later than December 31, 1987,” 42 U.S.C. § 7502(a), Congress made it clear that the states were to embark on the task of improving air quality as quickly as they could. We further note that the violation of law for which the judge was granting relief in the scheduling phase of the trial was that New Jersey had dragged its feet in implementing the regulations it had obligated itself to adopt in its SIP. Finally, we note that NJDEP has not joined in this appeal, and thus has not contested the amount of time it was given to propose a schedule; apparently it is satisfied that it was given enough time to come up with a reasonable proposal.
In view of the facts that: (1) the Clean Air Act mandates speedy action; (2) in this suit the American Lung Association and other groups were specifically seeking relief from the state’s delays in implementing required regulations; and (3) NJDEP itself apparently thinks the time it was given was adequate, we cannot say that the district court abused its discretion in granting the parties thirty-six days to develop recommended timetables.
B.
NACS and API also argue that the district court abused its discretion in adopting the schedule proposed by NJDEP for implementing Stage II control regulations, contending that the evidence suggests that the schedule will be impracticable. The schedule requires large gasoline stations to be in compliance with the new regulation by December 30,1988, and requires smaller stations to achieve full compliance by December 29, 1989. See Scheduling Order, app. A at A-l. The district court based its decision on evidence introduced by NJDEP to the effect that “it would take 25 weeks to install Stage II controls at the 2,000 largest stations and 27 V2 weeks to install controls at the remaining regulated stations.” Scheduling Opinion at 12. The district court found that NJDEP’s proposed schedule which provided 27 weeks for large-station installation and 43 weeks for the remaining stations was “certainly practicable.” Id. Appellants contend that the timetable adopted by the court is neither physically nor economically practicable.
As support for the feasibility of its proposed implementation schedule, NJDEP submitted notes documenting a telephone survey in which members of NJDEP contacted various contractors in New Jersey and asked them how long it would take them to install Stage II controls, and how many crews they had available, and then recorded and tabulated this information. As appellants point out, this survey is incomplete — for example, it did not take account of the fact that some crews might be busy installing Stage II controls in other states. Nonetheless, we do not believe that the district court abused its discretion in setting the schedule because appellants did not offer persuasive evidence to contradict the district court’s conclusion as to practicability.
Appellants submitted a report they had commissioned earlier for submission to EPA after EPA had asked for public comment on Stage II controls. The report discussed the feasibility of implementing Stage II controls, using Philadelphia and Dallas as representative cities. However, the timetables in the report are based on assumptions not necessarily relevant for New Jersey. For example, the Philadelphia estimate is based on the time it took to *329implement Stage I controls8 in Philadelphia, the implementation of which was not a procedure rushed by court order, and which was constrained by the number of contractors available in the Philadelphia area, which may be different from the number available in New Jersey.
Most of the other evidence that appellants offered as to the availability of contractors in New Jersey and the time it would take them to install stage II equipment comes from affidavits from representatives of oil companies and gas station associations — entities that have an economic interest in having the implementation of Stage II control regulations proceed slowly. Furthermore, these estimates are vague and are based on general past experience rather than specific evidence. There is no reason to believe that these estimates are better than the estimates made by NJDEP.9 The district court was therefore justified in adopting NJDEP’s estimates.
Moreover, the fact that the plan adopted by the court was in fact proposed by NJDEP adds to its credibility. NJDEP has no incentive to unduly accelerate the regulatory process. Additionally, NJDEP not only supervised notice and comment on the feasibility of implementing Stage II controls at the time the SIP was drafted, but the agency also subsequently monitored notice and comment on two proposals for Stage II implementation. Thus, NJDEP drew not only on the phone survey but also on its expertise on the subject of Stage II regulation, garnered in the regulatory process. EPA also endorsed the NJDEP plan, while dismissing the intervenors’ proposed scheduled as “protracted and misleading.” Letter from EPA (Nov. 7, 1987). EPA has expertise in the area of Stage II regulation, so it was right for the district court to give weight to EPA’s opinion. All of these factors support the district court’s decision.
Finally, the district court has retained jurisdiction over this matter. See Scheduling Order at 6. To the extent that it turns out to be physically impracticable for the gas stations to comply with the district court’s order, they can introduce new evidence, and the court can modify its prior order.
C.
Appellants also argue that the timetable set by the district court for implementation of Stage II controls is economically impracticable. They argue that the cost of installing Stage II controls is being made needlessly expensive because the timetable does not coordinate Stage II control installation with other gas station excavations and thus might require gas stations to engage in two separate excavations.
We believe that the district court was correct in dismissing appellants’ economic practicability arguments. As the court pointed out, it is not for a federal court to revise a SIP merely because it thinks it is better able to create economic efficiency. See Scheduling Opinion at 10-11. The SIP does not call for coordination of excavation at gas stations; it calls for implementation of Stage II controls by the end of 1987. If parties think that changes in circumstances make the requirements of the SIP unduly harsh, the proper course in general is to comply with administrative procedures for revising the SIP. See 42 U.S.C. § 7410(a)(3)(A) (setting out standard by which the EPA Administrator approves SIP revisions).
Appellants point out that the statute mandates that ozone regulation should proceed “as expeditiously as practicable,” 42 *330U.S.C. § 7502(a) (emphasis added), and argue that, despite the fact that the statute contemplates revision by agencies rather than the courts, a court attempting to effectuate the statutory purpose should take into account whether the requirements of a SIP are economically practicable. We need not reach the issue of whether it would ever be appropriate for a court to take into account arguments relating to the cost of compliance because in this case, although appellants offered evidence of duplication of costs, they have never offered evidence that would prove that the increase in cost by the failure of the schedule to coordinate the timing of excavations would be so significant as to make the timetable that the court adopted economically impracticable.
D.
API also argues that the district court abused its discretion in adopting NJDEP’s proposed timetable for implementation of barge loading controls. API points out that the U.S. Coast Guard is currently considering adopting nation-wide barge loading control regulations, and that the United States Department of Transportation has requested that states refrain from adopting such regulations so that a uniform national policy can be adopted and barges moving in interstate commerce are spared from inconsistent barge loading control regulations. See Letter from Secretary of Transportation Dole to Governor Deukmajian of California, (Feb. 25, 1986). The schedule adopted by the court provides for final implementation of barge loading regulations by February 28, 1990. See Scheduling Order, app. A at A-2. The Coast Guard, on the other hand, contemplates receipt of an advisory committee report on the subject of barge loading controls by February 1989 and expects to promulgate final rules by February 1990. See Scheduling Opinion at 15.
The district court acknowledged that the New Jersey regulations might clash with the eventual federal regulations. The court thus directed New Jersey to consult with the Coast Guard, so that it could benefit from the Coast Guard’s expertise, and so that the regulations New Jersey adopts can anticipate, insofar as possible, the forthcoming federal regulations. See Scheduling Opinion at 16. Given that the district court has done this much to minimize potential conflicts, and given that the schedule set in the adopted timetable already gives barge owners a more than two-year reprieve from the timetable set in the SIP, we cannot say that the district court abused its discretion in setting the schedule for barge loading controls.10
API also argues that the timetable set by the court will engender unnecessary expenses because the barge owners and dock owners may have to pay to implement controls that conform to the New Jersey regulations, and pay again to modify the controls if the federal regulations are adopted and are not identical, and because the schedule is not tied to the schedule on which barges are taken out of service for routine maintenance. We believe that the district court correctly rejected these arguments for the same reasons, discussed above, that it correctly rejected the arguments that the schedule for Stage II controls would create duplicate expenses. See supra at 329-30.
V. CONCLUSION
In light of the foregoing, the order of the district court will be affirmed.
. The trade associations are the American Petroleum Institute, the National Association of Convenience Stores, the Petroleum Marketers Association of America, and the Society of Independent Gasoline Marketers of America.
. Stage I controls, in contrast, are devices that limit the escape of ozone when the gas pumps are filled during the delivery of gasoline to the service station.
. Although plaintiffs’ claims against EPA remain before the district court, appellate jurisdiction is proper over the judgment against the state defendants because the court found, consonant with Fed.RXiv.P. 54(b), that there is no just reason for delay of review of this judgment, and consequently entered a Rule 54(b) order. See Dist. Ct. Scheduling Order at 1 (Nov. 19, 1987) [hereinafter "Scheduling Order”].
. In this regard, we also note that in Georgetown, the D.C. Circuit relied not simply on the legislative history of the act, but also on the narrowness of the definition of "emission standard” in the 1970 version of the statute. See 535 F.2d at 1322-23.
. This Court has also allowed a citizens’ suit against the state in its regulatory capacity, but without discussion. The suit in Delaware Valley Citizens’ Council v. Pennsylvania 533 F.Supp. 869 (E.D.Pa.) aff'd 678 F.2d 470 (3d Cir.), cert. denied, 459 U.S. 969, 103 S.Ct. 298, 74 L.Ed.2d 280 (1982) was brought by a citizens’ group against the state on the ground that the state had refused to create a program mandated by its SIP. See also League to Save Lake Tahoe, Inc. v. Trounday, 598 F.2d 1164, 1173 (9th Cir.1979) (finding jurisdiction over a suit by a citizens group to invalidate a state agency’s grant of building permits).
.The other intervenors, NACS, et al., do not seek to challenge the liability decision. They attack only the scheduling order.
. NACS contends that under the doctrine of "primary jurisdiction” the district court was required to defer to NJDEP, and allow NJDEP to set the schedule because NJDEP has greater expertise in the area. See MCI Communications Corp. v. AT & T, 496 F.2d 214, 220 (3d Cir.1974) (stating that under the doctrine of primary jurisdiction, relevant expert agencies “ 'should not be passed over’ ’’ in cases that raise " ‘issues of fact not within the conventional experience of judges' ”) (quoting Far East Conference v. United States, 342 U.S. 570, 574, 72 S.Ct. 492, 494, 96 L.Ed. 576 (1952)). We find this principle of deference to expert agencies to be inapposite to the instant case for two reasons. First, the schedule that was adopted was the one proposed by NJDEP. The agency thus did bring its expertise to bear on the problem. Second, in this case the agency was the defendant in the suit. To require the district court to defer to the *328judgment of the agency in this case would effectively defeat the provisions of the Clean Air Act that allow for judicial oversight of agency action (or inaction).
. As noted above, these are controls that prevent the escape of ozone when the gasoline is moved from trucks into the gas station’s underground storage tanks.
. Appellants do offer one stronger piece of evidence, an affidavit submitted by the head of an association of petroleum contractors, the industry that would be installing the Stage II controls. See Affidavit of Mark Towey Concerning Stage II Vapor Recovery. The number of contractors he identifies is similar to the number identified by the NJDEP survey. Towey states that these contractors are booked months in advance, and that they will thus be unable to finish installations on the court's schedule. However, it was within the discretion of the district court to disbelieve this affidavit which is inconsistent with the evidence put forth by NJDEP. See Scheduling Opinion at 13.
. It is possible that if the Coast Guard does embark on a nationwide regulatory scheme, state regulations will be preempted. However, as there are as yet no federal regulations, there is no current preemption problem. If one arises, the district court can address it, as it has retained jurisdiction.