On June 10, 2002, MSCC petitioned the United States Court of Appeals for the Ninth Circuit for review of EPA's May 2, 2002, final SIP action. Subsequently, MSCC and EPA agreed to a stay of the litigation pending EPA's final action on this FIP. The case is captioned Montana Sulphur & Chemical Company v. United States Environmental Protection Agency, No. 0271657. No petitions for judicial review were filed regarding EPA's May 22, 2003, SIP action.

On July 12, 2006 (71 FR 39259), EPA proposed Federal Implementation Plan (FIP) provisions for the Billings/Laurel, Montana area because of our disapproval of portions of Montana's Billings/Laurel SO2 SIP. In our proposal, we indicated that our FIP would not replace the SIP entirely, but instead would only replace elements of, or fill gaps in, the SIP.

In promulgating today's rules, EPA is fulfilling its mandatory duty under section 110(c) of the Act. Under section 110(c), whenever we disapprove a SIP, in whole or in part, we are required to promulgate a FIP. Specifically, section 110(c) provides:
``(1) The Administrator shall promulgate a Federal
implementation plan at any time within 2 years after the
Administrator
(A) Finds that a State has failed to make a required submission or finds that the plan or plan revision submitted by the State does not satisfy the minimum criteria established under [section 110(k)(1)(A)],\1\ or
\1\ Section 110(k)(1) requires the Administrator to promulgate minimum criteria that any plan submission must meet before EPA is required to act on the submission. These completeness criteria are set forth at 40 CFR 51, Appendix V.
(B) Disapproves a State implementation plan submission in whole or in part, unless the State corrects the deficiency, and the Administrator approves the plan or plan revision, before the Administrator promulgates such Federal implementation plan.''

Thus, because we disapproved portions of the Billings/Laurel SO2 SIP, and the attainment demonstration, we are required to promulgate a FIP.

Section 302(y) defines the term ``Federal implementation plan'' in pertinent part, as:
``[A] plan (or portion thereof) promulgated by the Administrator to fill all or a portion of a gap or otherwise correct all or a portion of an inadequacy in a State implementation plan, and which includes enforceable emission limitations or other control measures, means or techniques (including economic incentives, such as marketable permits or auctions or emissions allowances) * * *.''

More simply, a FIP is ``a set of enforceable federal regulations that stand in the place of deficient portions of a SIP.'' McCarthy v. Thomas, 27 F.3d 1363, 1365 (9th Cir. 1994). As the Court of Appeals for the D.C. Circuit noted in a 1995 case, FIPs are powerful tools to remedy deficient state action:

The FIP provides an additional incentive for state compliance because it rescinds state authority to make the many sensitive technical and political choices that a pollution control regime demands. The FIP provision also ensures that progress toward NAAQS attainment will proceed notwithstanding inadequate action at the state level.

Natural Resources Defense Council, Inc. v. Browner, 57 F.3d 1122, 1124 (D.C. Cir. 1995).

When EPA promulgates a FIP, courts have not required EPA to demonstrate explicit authority for specific measures: ``We are inclined to construe Congress' broad grant of power to the EPA as including all enforcement devices reasonably necessary to the achievement and maintenance of the goals established by the legislation.'' South Terminal Corp. v. EPA, 504 F.2d 646, 669 (1st Cir. 1974). As the Ninth Circuit stated in a case involving a FIP with farreaching consequences in Los Angeles: ``The authority to regulate pollution carries with it the power to do so in a manner reasonably calculated to reach that end.'' City of Santa Rosa v. EPA, 534 F.2d 150, 155 (9th Cir. 1976), vacated and remanded on other grounds sub nom. Pacific Legal Foundation v. EPA, 429 U.S. 990 (1976).

In addition to giving EPA remedial authority, section 110(c) enables EPA to assume the powers that the state would have to protect air quality, when the state fails to adequately discharge its planning responsibility. As the Ninth Circuit held, when EPA acts to fill in the gaps in an inadequate state plan under section 110(c), EPA `` `stands in the shoes of the defaulting State, and all of the rights and duties that would otherwise fall to the State accrue instead to EPA.' '' Central Arizona Water Conservation District v. EPA, 990 F.2d 1531, 1541 (9th Cir. 1993). As the First Circuit held in an early case: ``[T]he Administrator must promulgate promptly regulations setting forth `an implementation plan for a State' should the state itself fail to propose a satisfactory one * * * The statutory scheme would be unworkable were it read as giving to EPA, when promulgating an implementation plan for a state, less than those necessary measures allowed by Congress to a state to accomplish federal clean air goals. We do not adopt any such crippling interpretation.''

South Terminal Corp. v. EPA, supra, at 668 (citing previous version of section 110(c)).

The Billings/Laurel SO2 FIP establishes emission limits and compliance determining methods for four sources located in Billings/Laurel, Montana, to replace/fill gaps in portions of the SIP we disapproved, and to
[[Page 21420]]
support our attainment demonstration. Three of the sources are petroleum refineries: CHS Inc., ConocoPhillips (including the Jupiter Sulfur facility), and ExxonMobil. The fourth source is Montana Sulphur & Chemical Company, which provides sulfur recovery for the ExxonMobil refinery.

The following is a summary of the major components of our FIP rule: (1) The FIP establishes flare emission limits at all four sources (150 lbs SO2/3hour period at all but the Jupiter Sulfur flare, 75 lbs SO2/3hour period shared limit for the Jupiter Sulfur flare and the Jupiter Sulfur SRU/ATS stack) and monitoring methods to determine compliance with those limits. The FIP includes an affirmative defense to penalties for violations of the flare limits that occur during malfunction, startup, and shutdown periods. To determine flare emissions, the FIP requires concentration monitoring (which can consist of continuous monitoring, grab sampling, or integrated sampling) and continuous flow monitoring.
(2) The FIP prohibits the burning of sour water stripper overheads in CHS Inc.'s main crude heater and requires CHS Inc. to keep the valve between the old sour water stripper and the main crude heater closed, chained, and locked.
(3) The FIP provides that emission limits for identified ExxonMobil refinery fuel gas combustion units are contained in the SIP, and establishes compliance determining methods for instances in which the H2S concentration in the refinery fuel gas stream exceeds 1200 ppmv. These methods involve the use of lengthofstain detector tubes on a onceperhour frequency.
(4) The FIP provides that emission limits for ExxonMobil's Coker CO Boiler stack, when ExxonMobil's Coker unit is operating and Coker unit flue gases are burned in the Coker CO Boiler, are contained in the SIP. The FIP establishes compliance determining methods for these emission limits that require measurement of the SO2 concentration and flow rate in the Coker CO Boiler stack using CEMS.
(5) The FIP establishes emission limits on MSCC's SRU 100meter stack, based on good engineering practice (GEP) stack height credit of 65 meters, with compliance with these limits to be determined using methods already approved in the SIP. The FIP does not provide variable emission limits for this stack.
(6) The FIP establishes emission limits and compliance determining methods for MSCC's auxiliary vent stacks and SRU 30meter stack. In addition to mass limits, the FIP establishes concentration limits on fuel burned in the units that vent to the auxiliary vent stacks and SRU 30meter stack. These concentration limits are 160 ppm H2S per 3hour period and 100 ppm H2S per calendar day. When trigger events specified in the rule occur, MSCC must measure the H2S concentration in the fuel using lengthofstain detector tubes on a onceper3hour period.
(7) The FIP establishes various recordkeeping and reporting requirements.

It is important to note that, in cases where the provisions of the FIP address emissions activities differently or establish different requirements than provisions of the SIP, the provisions of the FIP take precedence. We also caution that if any of the four sources are subject to requirements under other provisions of the Act (e.g., section 111 or 112, part C of title I, or SIPapproved permit programs under part A of title I), our promulgation of the FIP does not excuse any of the sources from meeting such requirements. Finally, our promulgation of the FIP does not imply any sort of applicability determination under other provisions of the Act (e.g., section 111 or 112, part C of title I, or SIPapproved permit programs under part A of title I). II. Issues Raised by Commenters and EPA's Response
A. FIP Not Necessary
1. Ambient Data and Historical Modeling Show Attainment
(a) Comment (CHS Inc., COPC, ExxonMobil, NPRA, MPA, MDEQ, MSCC, WETA): The FIP is not necessary for attainment of the NAAQS because ambient data show that the Billings/Laurel area has been for many years and continues to be in attainment with both the Federal and State SO2 ambient air quality standards for all averaging periods.

Response: EPA does not agree that a FIP is not necessary because ambient data show attainment of the SO2 NAAQS. Ambient monitoring is limited in time and in space. Ambient monitoring can measure pollutant concentrations only as they occur; it cannot predict future concentrations when emission levels and meteorological conditions may differ from present conditions.

EPA has long held that ambient monitoring data alone generally are not adequate for SO2 attainment demonstrations. Additionally, a small number of ambient SO2 monitors usually are not representative of the air quality for an area. (See reference document GGGGG, April 21, 1983, memorandum from Sheldon Meyers, Director, Office of Air Quality Planning and Standards (OAQPS), to Regional Air and Waste Division Directors, titled ``Section 107 Designation Policy Summary,'' and reference document HHHHH, September 4, 1992, memorandum from John Calcagni, Director, Air Quality Management Division, OAQPS, to Regional Air Division Directors, titled ``Procedures for Processing Requests to Redesignate Areas to Attainment.'')

Typically, modeling estimates of maximum ambient concentrations are based on a fairly infrequent combination of meteorological and source operating conditions. To capture such results on an ambient monitor would normally require a prohibitively large and expensive network. Therefore, dispersion modeling is generally necessary to
comprehensively evaluate sources' impacts and to determine the areas of expected high concentrations. (Id.) Air quality modeling results would be especially important if sources were not emitting at their maximum level during the monitoring period or if the monitoring period did not coincide with potentially worstcase meteorological conditions. Further, ambient monitoring data are not adequate if sources are using stacks with actual heights greater than good engineering practice stack height (which indeed is the case with MSCC and ConocoPhillips) or other dispersion techniques for which SIP/FIP modeling credit is not allowed. (See also our discussion of related issues in our final action on the Billings/Laurel SO2 SIP (67 FR 22168, 2218522187, May 2, 2002.))

Ambient monitoring data and air quality modeling data for a particular area can sometimes appear to conflict. This is primarily due to the fact that modeling results may predict maximum SO2 concentration at receptors where no monitors are located.

Moreover, our SIP Call for the Billings/Laurel area was based on modeled violations of the SO2 NAAQS, not monitored violations. (See reference documents Y and Z.) We took final action on the SIP Call in our May 2, 2002, action on the Billings/Laurel SIP (67 FR 22168, 22173), and we are not revisiting it in this FIP action. It would be inconsistent and inappropriate to now rely solely on monitoring to determine necessary measures and demonstrate attainment.

It is especially important to recognize that, as a result of our partial and limited disapproval of the Billings/
[[Page 21421]]
Laurel SO2 SIP, we are legally obligated to promulgate a FIP for the area. See section 110(c)(1) of the CAA, 42 U.S.C. 7410(c)(1). However, the SIP deficiencies that triggered our partial and limited disapproval were varied and were not necessarily associated with problems that could be measured at an ambient monitor. For example, one basis for disapproval of the SIP was the State's use of improper (too tall) stack height credit for MSCC in modeling attainment of the NAAQS. In the real world, emissions at the actual (100 meter) height of the stack create less impact on monitored ambient concentrations in the Billings/Laurel area than if the emissions were emitted from a lower stack. Nonetheless, we had to partially disapprove the SIP due to the State's inappropriate grant of stack height credit, and section 110(c) of the CAA requires that we correct the deficiency. Since the State did not model attainment at the proper stack height credit for MSCC's stack, it was necessary that we do so and set emission limits for the stack consistent with our attainment demonstration. We believe MSCC has consistently been meeting the emission limits we are adopting, so there may be no reduction in actual emissions from the stack, but that does not mean the CAA allows us to forego this aspect of the FIP.

Likewise, CAA sections 110(a)(2)(A) and (C) require that SIP control measures be enforceable. We disapproved several source monitoring methods because they were not adequate to determine compliance under all operating conditions. It may be impossible to measure the impact these SIP deficiencies may have on ambient SO2 concentrations in the area, but the CAA still requires that we correct the deficiencies. Regarding the emission limits and compliance determining methods for the flares, the Stateonly flare limits, which the State relied on to demonstrate attainment, may have positively impacted flare emissions in the past few years. However, the State did not include the Stateonly flare limits or adequate compliance determining methods in the SIP. Thus, the SIP remains deficient. We now have the responsibility to ensure that emission limits relied on to demonstrate attainment are included in the SIP and are practically enforceable, consistent with the requirements of section 110 of the Act.
(b) Comment (MSCC, MDEQ): The State's SIP modeling, along with appropriate emission limits, show attainment of the NAAQS.

Response: EPA addressed this issue in its actions on Montana's SIP submissions. As explained in those actions, EPA does not agree that the State's SIP modeling, along with appropriate emission limits, show attainment of the NAAQS. EPA's formal determinations regarding the attainment demonstration and emission limits were made in final actions on May 2, 2002 (67 FR 22168) and May 22, 2003 (68 FR 27908). The FIP fills the gaps for the provisions we disapproved.

We note that we have not reopened our SIP actions as part of this action. Thus, to the extent the commenters are expressing their disagreement with EPA's actions on the SIP, their comments are not relevant to this action, and EPA is not reconsidering them here. (c) Comment (ExxonMobil): EPA's proposed FIP ignores the substantial improvement in air quality in the Billings/Laurel area and instead predicts exceedances of NAAQS based upon modeling performed as long as 15 years ago. EPA's FIP proposal must be further examined in light of subsequent developments, including correct modeling and consideration of currently available information indicating compliance.

Response: See response to comment II.A.1.(a), above, regarding ambient data and response to comments in section II.E., below, regarding modeling.
2. Existing Controls Sufficient
(a) Comment (MDEQ, MSCC, COPC, ExxonMobil, MPA, NPRA, WETA): The FIP offers questionable improvements because the existing control plan provisions submitted by the state are adequate and contain sufficient SO2 emission controls and strategies and provide for the implementation, maintenance, and enforcement of the SO2 NAAQS.

Response: EPA addressed the adequacy of Montana's SIP submissions in its final actions on the SIP. As explained in those actions, EPA does not agree that the State's SIP control plan provisions are adequate and contain sufficient SO2 emission controls to show attainment of the NAAQS. EPA's formal determinations regarding the attainment demonstration and emission control plan were made in final actions on May 2, 2002 (67 FR 22168) and May 22, 2003 (68 FR 27908). In our May 2002 and May 2003 actions we disapproved various control plan provisions. The FIP fills the gaps for the provisions we disapproved. The FIP offers necessary improvements to the SIP by imposing new emission limits and reliable compliance determining methods to ensure attainment of the SO2 NAAQS.

We note that we have not reopened our SIP actions as part of this action. Thus, to the extent the commenters are expressing their disagreement with EPA's actions on the SIP, their comments are not relevant to this action, and EPA is not reconsidering them here. (b) Comment (CHS Inc., WETA, COPC, MDEQ, ExxonMobil, NPRA): In addition to the SIP, SO2 emissions in the Billings/Laurel area have decreased as a result of Consent Decrees and Montana Air Quality Permit changes. These limits are all federally enforceable because there are Title V operating permit conditions (CHS Inc.). EPA did not consider these emission reductions in making its determination that the FIP was necessary. The FIP proposal does not otherwise acknowledge the practical effects of the recent consent decrees between the primary refinery parties subject to regulation as well as other permitting actions that have occurred over the past eight years (MSCC, COPC).

Response: EPA did not consider the emission reductions that resulted, or will result, from the consent decrees and/or State permit revisions to determine that the FIP was necessary or include the emission reductions in our modeling for several reasons.

First, the FIP is required because we disapproved the SIP, and the State has not made revisions to the SIP to address the SIP's flaws. As noted in other responses, because we disapproved the SIP, we have a legal obligation to promulgate a FIP. See CAA section 110(c), 42 U.S.C. 7410(c).

Second, even though permits and consent decrees are federally enforceable, some permits can be revised without EPA approval and consent decrees have a limited lifespan.\2\ To protect the integrity of the attainment demonstration, and our statutory role in assessing SIP/ FIP adequacy, we believe that stationary source emission limits necessary to demonstrate attainment must be included in the FIP (or approved SIP). See, e.g., CAA sections 110(a)(2)(A), 110(i), 110(k)(3) (6), and 110(l), 42 U.S.C. 7410(a)(2)(A), (i), (k)(3)(6), and (l). This ensures that changes to those limits will only be made with EPA's approval as a SIP or FIP revision,
[[Page 21422]]
following notice and comment rulemaking.
\2\ The State can revise construction permits without EPA approval, and, while EPA has authority to object to Title V permits, that authority is only available to ensure that underlying applicable requirements are included in the Title V permits. Thus, if those underlying requirements change, EPA may have no recourse at the Title V stage.

Third, the consent decrees and permitting actions, for some emission points, do not contain SO2 emission limits that are consistent with the averaging times of the SO2 NAAQS, specifically, the 3hour and calendar day averaging periods. For example, the SIP establishes 3hour, calendar day, and calendar year emission limits for CHS Inc.'s FCC regenerator/CO boiler stack. The January 17, 2007, final State construction permit (reference document IIIII) and the consent decree (reference document JJJJJ) indicate that the FCC regenerator stack SO2 emissions shall not exceed 50 ppm by volume (corrected to 0% O2) for a 7day rolling average [or a fresh feed of 0.3 percent by weight] and 25 ppm by volume (corrected to 0% O2) for a 365day rolling average. None of the commenters has suggested these limits be converted to FIP mass limits that would apply over a 3hour averaging period, and the State has not submitted a SIP revision with such limits.

It should be noted that EPA did solicit comment on whether we should limit the main flares to 500 pounds of SO2 per calendar day. This value is consistent with the trigger point for certain analyses contained in settlements (i.e., consent decrees) between the United States and CHS Inc., ConocoPhillips, and ExxonMobil. We received limited comments on this proposal and have decided to keep the limit at 150 pounds of SO2 per 3hour period to maintain consistency with the State's Stateonly limit.
B. EPA Exceeded Its Authority in Proposing a FIP
1. State's Responsibility
(a) Comment (WETA, MPA, ExxonMobil): EPA's role is limited to determining whether or not a SIP is attaining and maintaining the NAAQS. Selecting the source mix and various control measures to achieve these ends has been determined by courts to be the sole responsibility of the state. EPA's proposed action intrudes on the primary responsibility of the state and local governments to implement the Clean Air Act (MSCC).

Response: The commenters' characterization of EPA's role regarding SIPs is not accurate. We lack authority to question a state's choices of emissions limitations if they are part of a plan that satisfies the standards of the Clean Air Act. Train v. Natural Resources Defense Council, 95 S.Ct. 1470, 14811482 (1975). In our 2002 and 2003 actions, we found that Montana's SO2 SIP for Billings/Laurel did not fully satisfy CAA requirements. See 67 FR 22168, May 2, 2002 and 68 FR 27908, May 22, 2003. Thus, pursuant to section 110(c) of the CAA, 42 U.S.C. 7410(c), we are required to promulgate a FIP. In doing so, we stand in the state's shoes and have authority to determine emissions limitations and other measures for specific sources to fill gaps in the SIP. Central Arizona Water Conservation District v. EPA, 990 F.2d 1531, 1541 (9th Cir. 1993); South Terminal Corp. v. EPA, 504 F.2d 646, 668 (1st Cir. 1974) (citing previous version of CAA section 110(c)).

We note that we have not reopened our SIP actions as part of this action. Thus, to the extent the commenters are expressing their disagreement with EPA's actions on the SIP, their comments are not relevant to this action, and EPA is not reconsidering them here. (b) Comment (WETA): Since the State of Montana has already taken appropriate actions to reduce sulfur dioxide emissions, EPA does not have the authority under the CAA to adopt the proposed FIP.

Response: See response to comment II.B.1.(a), above. The adequacy of the State of Montana's actions has already been considered by EPA in other rulemaking actions that addressed the State's SIP submission. Those actions are not the subject of EPA's present rulemaking, which promulgates the necessary measures to remedy the deficiencies EPA identified in its prior SIP reviews.
(c) Comment (MSCC): States have primacy, and because EPA did not choose to exercise its rights in the comprehensive and competent state decision process, EPA may not default and then act.

Response: Under section 110(c) of the Act, EPA is not required to participate in a state's administrative process before promulgating a FIP.
(d) Comment (MSCC, MDEQ, ExxonMobil): EPA has no authority to question the wisdom of a state's choices of emission limitations if they are part of a plan that satisfies the standards of Sec. 110(a)(2) of the Act. As long as the ultimate effect of a state's choice of emission limitations is compliance with the NAAQS, the state is at liberty to adopt whatever mix of emission limitations it deems best suited to its particular situation. There is no evidence provided by EPA that Montana reached its material conclusions or choices in the SIP unreasonably. Additionally, EPA has not shown that additional controls beyond the SIP measures adopted by Montana are necessary to meet or assure SO2 NAAQS compliance.

Response: See our responses to comments II.A.1.(a) and II.B.1.(a), above. Much of this comment pertains to our actions on Montana's SIP. We are not revisiting or reopening comment on those actions here. Our basis for finding that the SIP was not adequate to ensure attainment and meet other CAA requirements is described in our actions on the SIP. Once we disapprove part or all of a required SIP, section 110(c) of the Act requires that we issue a FIP. Our obligation in this action is to correct the SIP deficiencies we previously identified. Thus, the findings that triggered our responsibility to promulgate a FIP were established in the prior rulemaking actions reviewing Montana's SIP. EPA is not required to repeat those findings in the FIP rulemaking itself.
(e) Comment (ExxonMobil): EPA cannot propose a FIP to replace a SIP, unless the SIP is substantially inadequate to ensure compliance with the CAA.

Response: The commenter misstates the standard for promulgation of a FIP. Section 110(c) of the CAA is straightforwarda FIP is required if (1) EPA finds that a state has failed to make a required submission; (2) EPA finds that a plan submission does not satisfy the completeness criteria established under section 110(k)(1)(A) of the CAA; or (3) EPA disapproves a SIP in whole or in part. EPA partially disapproved the Billings/Laurel SO2 SIP; thus, a FIP is required. Contrary to the commenter's assertion, the obligation to promulgate a FIP is not contingent on an EPA finding of substantial inadequacy. As explained above, the findings triggering our responsibility to promulgate a FIP were made in the prior actions reviewing Montana's SIP.
(f) Comment (MSCC): The commenter claims EPA's action violates the Tenth Amendment to the Constitution. The commenter also claims EPA's FIP is dictating the required controls in contravention of the holdings in Commonwealth of Virginia v. EPA, 108 F.3d 1397 (D.C. Cir. 1997) and Bethlehem Steel v. Gorsuch, 742 F.2d 1028 (7th Cir. 1984).

Response: Our FIP compels no action on the part of the State and is not coercive vis[agrave]vis the State. Our FIP contains requirements applicable to four private companies. The Tenth Amendment is not implicated. Nor do our actions contravene Commonwealth of Virginia or Bethlehem Steel. The former case held that EPA cannot, in a SIP Call, dictate that a state adopt a particular control measure to
[[Page 21423]]
demonstrate attainment of the NAAQS. EPA had issued a SIP Call finding that the SIPs of 12 states were inadequate to meet the ozone NAAQS and in its SIP Call rule, specified that the states needed to submit SIPs that included the California Low Emission Vehicle Program. In this matter, we are promulgating a FIP, not issuing a SIP Call. We are not directing any action by the State. Thus, the Commonwealth of Virginia case is not relevant to our FIP. Bethlehem Steel is also not relevant to our FIP action. In that case, the 7th Circuit held that it was improper for EPA to partially approve an Indiana SIP revision so as to render it more stringent than the State intended. We are promulgating a FIP in this action, not acting on a SIP; thus, Bethlehem Steel does not apply. As we note elsewhere, once we disapprove a SIP, we are required to promulgate a FIP, and in promulgating the FIP, we stand in the state's shoes. See section 110(c) of the CAA, 42 U.S.C. 7410(c); Central Arizona Water Conservation District v. EPA, 990 F.2d 1531, 1541 (9th Cir. 1993).
(g) Comment (MSCC): The commenter argues that the cases EPA cited in the preamble to the proposed Billings/Laurel FIP, regarding its FIP authority, do not speak to the central question``When and on what authority may the EPA undertake the draconian act of displacing a state's implementation plan?'' The commenter argues that the question is particularly sensitive in this case because the State and the sources spent years negotiating the SIP.

Response: As noted in response to comment II.B.1.(e), the CAA requires that we promulgate a FIP whenever we disapprove a SIP, in whole or in part. While we are sensitive to the fact that the State and sources spent years negotiating the SIP, that does not change our obligation under the CAA.
2. No Adequate Basis for FIP
(a) Comment (MSCC, ExxonMobil): Because EPA must find substantive noncompliance with some provision of the Clean Air Act, specifically, failure to attain NAAQS, and because that finding of substantial inadequacy must be clearly stated, the present FIP decision must fall. It is inadequate on both counts. EPA has not provided any evidence that the State plan is not working.

Response: See our response to comment II.B.1.(e), above. The evidence supporting EPA's determinations regarding the adequacy of Montana's SIP is contained in the record for those rulemaking actions, and need not be repeated here. EPA's disapproval of the SIP triggered the obligation for a FIP. No separate showing that the State plan is not working or does not meet CAA requirements is needed as part of this action. Commenters' comments regarding EPA's SIP actions are not relevant for this rulemaking.
(b) Comment (ExxonMobil): Even when the EPA has statutory authority for a particular rule, its technical decisions about the level of pollutant reduction needed to comply with the CAA and the control strategies necessary to meet the level of pollutant reduction must be rational. Courts ``confronted with important and seemingly plausible objections going to the heart of a key technical determination * * * '' will not presume that EPA would never behave irrationally. South Terminal Corporation v. Environmental Protection Agency, 504 F.2d 646, 665 (1st Cir. 1974). In South Terminal Corporation, various interested parties challenged EPA's FIP on technical grounds. Id. at 66266. The court held that EPA failed to adequately support its decision to promulgate the rules contained in the FIP and remanded the case to EPA to develop the record. Id. at 666. The court questioned EPA's position in light of contradictory modeling and data, concluding that ``it is not clear whether or not the ambient air at Logan meets, or will without controls by mid1975 will meet, the national primary standard.'' Id. 664. Similarly, in the present FIP proposal, EPA has neither determined appropriate current modeling nor used currently available information.

Response: The standards for judicial review of this rulemaking action are contained in section 307(d)(9) of the CAA, 42 U.S.C. 7607(d)(9). We believe the emission limitations and other requirements in this FIP are reasonable and that the situation in the cited case is not analogous.\3\ The commenter has not identified any modeling that contradicts our attainment demonstration, which forms the basis for the FIP's emission limitations; nor has the commenter shown that a different model would result in substantially different emission limitations. Our responses pertaining to model selection and input data are contained in section II.E., below. Further, we note that it does not appear the commenter is suggesting that the entire SIP should be redone based on more current modeling and more uptodate information. On the contrary, the commenter seems satisfied with the EPAapproved emission limitations in the SIP,\4\ which were based on the very modeling that the commenter now claims is unreliable.
\3\ In South Terminal Corporation, EPA had determined emissions reductions needed to achieve the ozone and carbon dioxide NAAQS based on monitored values that the Court found highly questionable (petitioners claimed the ozone monitor was defective). South Terminal Corporation, 504 F.2d 646, 662 (1974). The commenter seems to suggest that the Court rejected EPA's modeling approach, but in fact, the Court was satisfied with the rollback modeling that EPA used. Id.
\4\ Among other things, the commenter asserts that the state SIP requirements are adequate to protect the NAAQS. See reference document YYYY, page 27.
(c) Comment (ExxonMobil): Citing Hall v. United States Environmental Protection Agency, 273 F.3d 1146, 1159 (9th Cir. 2001), the commenter states that in acting on a SIP, the test EPA applies is to ``measure the existing level of pollution, compare it with the national standards, and determine the effect on this comparison of specified emission modifications.'' The commenter argues that in the FIP proposal, EPA did not correctly identify the existing level of pollution and ignored the substantial evidence of permanently reduced SO2 emissions and levels in the Billings/Laurel area. The commenter also argues that EPA's authority is limited by its mandate under the CAA to ensure attainment and maintenance of the NAAQS as well as the CAA's other general requirements.

Response: See responses to comments II.A.1.(a), II.A.2(b), and II.E.1.(e) and (g). Also, the Hall case involved a challenge to EPA's approval of a SIP revision for Clark County, Nevada, and EPA's interpretation of section 110(l) of the CAA, which provides that EPA may not approve a SIP revision if it would interfere with attainment or other applicable requirements of the CAA. EPA asserted that its approval of the Clark County SIP revision was consistent with section 110(l) because the revision did not relax the existing SIP. The Court disagreed, holding that 110(l) requires morea determination that the specific revision, when considered in the context of the SIP elements already in place, can meet the Act's attainment requirements. Hall at 1152, 1159. It was in these circumstances that the Court expected EPA to determine the extent of pollution reductions required and evaluate whether the reductions resulting from the revision would be sufficient to attain the NAAQS.

In its reference to Hall, the commenter appears to be conflating two disparate concepts. The Hall Court was addressing EPA's action on a SIP revision and indicating that EPA was not adequately evaluating whether Clark County's rule change would interfere
[[Page 21424]]
with attainment and other CAA requirements. The Court was not establishing a standard for a FIP or indicating that EPA was requiring more than necessary for the area, which seems to be what the commenter is suggesting in the case of the Billings/Laurel FIP. As we explain in greater depth elsewhere in this notice, we are not starting from scratch with our FIP. Instead, we are working within the framework of the existing Billings/Laurel SIP to fill the gaps resulting from our partial and limited disapproval of discrete SIP elements. In this unique circumstance, where only discrete elements of the SIP were deficient, the CAA does not require us to reevaluate or replace the entire SIP or the basic modeling approach upon which it was based. Nothing in the CAA requires EPA to reject an entire SIP when only certain elements within it are not approvable, and doing so, where that is not necessary to address a discrete deficiency, would be inconsistent with the basic scheme of cooperative federalism embodied in the CAA.

Nor are we required as part of this FIP to revisit our SIP Call or the bases for our SIP disapproval. Our task is to fix the portions of the SIP that were deficient. It is reasonable to continue to treat as valid the factors we found adequate to support the portions of the SIP we approved, and augment and/or replace those factors that we found inadequate. In fact, based on the holding in Train v. NRDC, 421 U.S. 57 (1975), recited by this commenter and others, it would be inappropriate for EPA to now reject or replace the portions of the SIP that we approved as meeting the CAA's requirements, because to do so would be to intrude on the State's authority under the CAA to establish the mix of controls for the area.\5\ The State, of course, remains free to submit a SIP revision that reflects a different mix of controls across all the sources. This would be the mechanism, for example, whereby the State could adopt SIP limits that correlate to refinery consent decree limits.\6\ If the State were to submit such a revision, we would evaluate the revision according to the Act, our regulations, and the relevant cases.
\5\ To the extent the commenter is arguing that we may do no more in this FIP than appears minimally necessary to attain the NAAQS, we reject that notion as well. See, e.g., Central Arizona Water Conservation District v. EPA, 990 F.2d 1531, 1541 (9th Cir. 1993) (EPA ``stands in the shoes of the defaulting State, and all of the rights and duties that would otherwise fall to the State accrue instead to EPA.'') Under the CAA, states are not restricted to barely meeting the NAAQS. In fact, the opposite is truestates may exceed minimum requirements. See CAA section 116, 42 U.S.C. 7416. In any event, our modeled attainment demonstration resulted in projected values just at the 24hour SO2 NAAQS (365 [mu]g/m\3\) and just below the 3hour SO2 NAAQS (1291.5 [mu]g/m\3\). However, we think we had discretion to adopt limits (to replace those we disapproved) consistent with modeled ambient concentrations further below the NAAQS, if we had felt a larger margin of safety was justified to ensure attainment and maintenance. \6\ As we allude to in sections II.A.2.(b), II.D.4., and II.E.1.(e), the consent decree limits would need to be translated into limits that support an attainment demonstration for the SO2 NAAQS. In sections II.A.2.(b) and II.D.4., we identify some of our concerns with the consent decree limits. (d) Comment (ExxonMobil): EPA's proposal imposes costly technology requirements not rationally designed to achieving their stated objectives. While EPA has authority to impose an emission limitation, the emission limitation must be necessary to attain NAAQS. City of Santa Rosa v. EPA, 534 F.2d 150, 155 (9th Cir. 1976), vacated on other grounds, 429 U.S. 990 (1976). The EPA derived its authority in City of Santa Rosa from its statutory mandate to ensure compliance with NAAQS and the fact that no alternative to its proposal was adequate to ensure compliance with NAAQS. It is clear that Montana's existing SIP, supplemented as it is by further state and federally enforceable consent decrees are a more than adequate alternative.

Response: The cited case actually stands for the proposition that EPA's authority to adopt measures to meet the NAAQS is expansive. EPA adopted a FIP provision that would have required a substantial reduction (up to 100%) in the supply of gasoline to major metropolitan areas in California, including Los Angeles. Even the EPA acknowledged that the rule would cause severe social and economic disruption, and the EPA Administrator at the time publicly advocated amendments to the CAA to provide relief from EPA's own FIP rule. Nonetheless, the Court held that economic and social disruption are not cognizable if (1) a measure is necessary to attain the NAAQS; (2) there is no statutory limitation on EPA's authority to adopt the measure; and (3) there are no equally effective, less burdensome alternatives. City of Santa Rosa at 151154.

The measures EPA is promulgating in this FIP are in no way comparable to the reduction in gasoline supply at issue in the City of Santa Rosa case. Our FIP is narrowly tailored to fill the gaps in the Billings/Laurel SIP. Section 110(c) requires us to promulgate the FIP. There is no statutory limitation on our authority to adopt the measures we are adopting. On the contrary, section 110(a)(2)(A) of the Act requires enforceable emission limitations as necessary or appropriate to meet the applicable requirements of the Act, which include attainment and maintenance of the SO2 NAAQS. Using ISC, the same model the State used to set the commenter's emission limits in the SIP, we have determined emission levels consistent with attainment and established corresponding emission limits on the flares, MSCC's main stack, and other emission units, whose emission limits we disapproved in our SIP action. While the authority to require monitoring, recordkeeping, and reporting requirements can be inferred from CAA sections 110(a)(2)(A) and (C), section 110(a)(2)(F) of the Act specifically indicates that the EPA Administrator may prescribe the installation, maintenance, and replacement of monitoring equipment by stationary sources, as well as reporting requirements. Our requirement for the refineries and MSCC to install monitoring equipment to measure flare gas flow and concentrations is consistent with this authority and is rationally related to the goals of the FIP, i.e., to ensure attainment and maintenance of the SO2 NAAQS. We do not believe estimating flare emissions or emissions from other units is a sufficient substitute for realtime monitoring for purposes of this FIP; estimation is not an equally effective technique.

The commenter argues that the existing SIP and the State and federally enforceable consent decrees are a more than adequate alternative to our FIP requirements. This comment ignores the fact that we disapproved portions of the SIP as not meeting the CAA's requirements. Elsewhere we explain that the consent decree provisions are not sufficient to meet the CAA's requirements under section 110 related to attainment and maintenance of the NAAQS. See, e.g., sections II.A.2.(b), II.D.4., and II.E.1.(e).
(e) Comment (MSCC): EPA's failure to issue the FIP within the CAA's twoyear deadline is important in this case. As a result of EPA's delay, EPA should have to consider the cleanup of emissions that has occurred and significant changes in modeling technology.

Response: We regret that it has taken this long to issue the FIP. We disagree that missing the twoyear deadline obviates our duty or the need for the FIP. The State has not submitted a SIP revision correcting the portions of the SIP that we disapproved, despite the passage of time. Regarding the argument that we should have considered the reduction in emissions since we disapproved the SIP, see our responses to comments in section II.A. In section II.E, we respond to comments arguing
[[Page 21425]]
that we should have used newer modeling technology.
C. Flare Monitoring
1. Flare Flow Monitoring
(a) Comment (MSCC): The core flowmeter technology application for flare systems seems to be an established technology, with thousands of installations completed around the world on other types of gas and liquid streams. However, none was identified that is following the precise specifications of the FIP proposal. Installation and operation of a flow meter in flare gas service at MSCC are probably achievable today, but not at the flow range below 1 fps, and not with conventional QA/QC procedures. Flow monitors have a difficult time measuring or reliably detecting low flow velocities (under approximately 1.0 fps) without false positives or false negatives. EPA should revise the proposed rule that currently indicates:
``[t]he minimum detectable velocity of the flow monitoring device(s) shall be 0.1 feet per second (fps). The flow monitoring device(s) shall continuously measure the range of flow rates corresponding to velocities from 0.5 to 275 fps and have a manufacturer's specified accuracy of 5% over the range of 1 to 275 fps.

The revised rule should read ``[t]he minimum resolution of the flow monitoring device(s) shall be 0.1 feet per second (fps) when measuring flow rates above 1.0 fps. The device(s) shall continuously measure the range of flow rates corresponding to velocities from 1.0 to 275 fps and have a manufacturer's specified accuracy of 5% over the range of that range.''

The rule should also clarify if ``accuracy'' is intended to be 5% of the fullscale range of the instrument (13.7 fps is 5% of 275 fps), or if this is intended to be 5% of the measured flow, which would be 0.05 fps at a flow of 1 fps, and would clearly be nonachievable with a resolution of 0.1 fps.

Response: EPA proposed the volumetric flow monitoring specifications based on what we saw was achievable in vendor literature (see reference documents NN and OO) and what was being required by regulation in the Bay Area Air Quality Management District (BAAQMD) (see reference document LL) and South Coast Air Quality Management District (SCAQMD) (see reference document CCC).

The commenter asserts that installation and operation of a flow meter at the flow range below 1 fps are not achievable. However, various sources indicate that ultrasonic flow meters can measure in the range of 0.1 to 1 fps. For example, in ``Flare Gas Ultrasonic Flow Meter,'' J.W. Smalling, L.D. Brawsell, L.C. Lynnwoth and D. Russel Wallace, Proceedings ThirtyNinth Annual Symposium on Instrumentation for the Process Industries, 1984, the authors reported ``initially, a modest objective was established to develop an ultrasonic flow switch capable of detecting leaks in flare lines corresponding to flow velocity on the orders of 0.3 ms/ (1 ft/s). As testing continued, however, it became apparent that the equipment could measure flows below 0.03 m/s (0.1ft/s) and up to at least 6 m/s (20 ft/s) in flare stacks * * *'' (see reference document KKKKK). See also reference document OO, ``the DigitalFlowGF868 meter achieves rangeability of 2750 to 1. It measures velocities from 0.1 to 275 ft/s (0.03 to 85 m/s) in both directions, in steady or rapidly changing flow, in pipes from 3 in. to 120 in. (76 mm to 3 m) in diameter.''

Additionally, the BAAQMD (see reference document LL) and SCAQMD (see reference document CCC) require flow meters on flares. BAAQMD requires that the minimum detectable velocity shall be 0.1 fps and the SCAQMD requires monitors with a velocity range of 0.1 to 250 fps. Based on conversations with the BAAQMD, it appears that the refineries in the Bay Area have installed flow meters meeting the requirements of the rule (see reference document OOOOO).

Based on the above, we conclude that flow meters are available that can measure in the velocity range below 1.0 fps, and other regulatory authorities are requiring such flow meters with success.

The commenter also claims that installation and operation of a flow meter are probably not achievable with conventional QA/QC procedures. The QA/QC procedures are discussed below in response to comment II.C.1.(d).

The commenter argues that flow monitors have a difficult time measuring or reliably detecting low flow velocities (under approximately 1.0 fps) without false positives or false negatives. As indicated in the response to comment II.C.1.(b) below, there are approaches available for improving measurement accuracy in the 0.1 to 1.0 fps range. In addition, as the response to comment II.C.1.(b) indicates, in the final FIP we are specifying a separate accuracy range for the velocity range of 0.1 to 1 fps. Finally, we describe how we are addressing the false positive and false negative flows in response to comment II.C.1.(c).

The commenter asked that the rule clarify if ``accuracy'' of the instrument is intended to be 5% of the fullscale range of the instrument or 5% of the measured flow. In the rule, we have clarified that ``accuracy'' of the instrument is the accuracy of the measured flow and not the ``fullscale range'' of the instrument.

The commenter also suggests some changes to the rule. Apart from adding a separate accuracy range for the velocity range of 0.1 to 1 fps and clarifying that accuracy is based on the measured flow, we are not making any additional changes to this aspect of the rule. We explain our reasoning in the response to this comment II.C.1.(a) and in the responses to comments II.C.1.(b)(d), below.
(b) Comment (ExxonMobil, WSPA): Manufacturers of flow monitoring instrumentation publish impressive performance specifications regarding velocity measurement range and accuracy, but often manufacturers' claims are not actually achieved in practice over the long term. To achieve a high level of measurement performance in the field requires adequate lengths of straight flare header pipe upstream and downstream of the monitor, the absence of flow disturbances, etc. Where these criteria cannot be met, the advertised or predicted performance of the flow monitoring system may not be fully realized in practice. MSCC claimed that significant piping modifications and possible flare relocation would be required to provide such runs at accessible locations. CHS Inc. asserted that it is likely that the CHS refinery flare header will not have adequate distances of undisturbed piping for ideal installation. In this case, either major, costly piping modification will be required or the accuracy criteria will not be achievable.

Response: The commenters are correct that piping modifications may be appropriate to optimize the measurements. Each flare system will have unique flow measurement location issues and will have to be addressed on a casebycase basis. Sources may need to work with the flow monitor manufacturer and flow testers to assure that the monitors meet the FIP's specifications for accuracy and representativeness and manufacturer's requirements for assuring ongoing equipment performance.

In addition to making piping modifications (e.g. flow straighteners), other approaches are available to improve the measurement accuracy in the 0.1 to 1.0 fps range. Among the approaches are the use of additional monitoring paths, monitoring paths of longer length, and unconventional monitor configurations and path locations. Another approach involves
[[Page 21426]]
the use of Computer Fluid Dynamics (CFD) for the existing piping. CFD analysis has been used to provide correction factors for a series of velocities across the range of flow velocities. For example, these factors have been used to correct flow measurement data for disturbances caused by upstream pipe irregularities. These approaches are discussed in ``A Total Approach to Flare Gas Flow Measurement for Environmental Compliance,'' Gordon Mackie, Jed Matson and Mike Scelzo, Institute of Measurement and ControlEnvironmental Conference 2006. (See reference document LLLLL.) (See also Note to Billings/Laurel SO2 FIP File regarding conversations with GE Sensing (reference document MMMMM)).

Finally, to address concerns regarding the measurement accuracy in the 0.1 to 1.0 fps range, we are revising the rule to indicate that the flow monitor must have a manufacturer's specified accuracy of 20% over the range 0.1 to 1 fps. Based on conversations with a vendor, we believe this is achievable. The vendor indicated that they have provided methodologies for sources to meet the SCAQMD rule, which also requires 20% accuracy in the 0.1 to 1.0 fps range. Methodologies include a second interrogation path or straightening of pipe. (See reference document MMMMM.)
(c) Comment (ExxonMobil, WSPA, NPRA, MSCC): Consistently achieving low flow detection limits can be very difficult. Spurious signal, resulting in ``eddy'' currents and backandforth flows in the flare header, can easily limit the detection and accuracy of low flow readings. Furthermore, sometimes a flow monitor will show an indication of flow even though water seals ahead of the flare stack remain intact (i.e., there is not flow to the flares). Other regulations in other jurisdictions allow the sources other means to positively determine when the flare is not operating (e.g., flare on/off monitoring device, pressure of water seal). ExxonMobil recommends that similar language be considered by the stakeholder process for inclusion in the EPA's proposed FIP, and thereby remove the uncertainty of low flow reading. MSCC claimed that the EPA proposed FIP language should be revised to allow flare operations to be monitored by other means, and to disregard low flow readings when the flare is not operating to eliminate falsely reported SO2 emissions, when in fact there are none.

Response: We agree that it is appropriate to include in the regulation the ability to use other secondary means to determine whether flow is reaching the flare when the flow monitor indicates low flow. If the secondary device indicates that no flow is going to the flare, yet the continuous flow monitor is indicating flow, the presumption will be that no flow is going to the flare. We have revised the final rule to allow the use of flare water seal monitoring devices to determine whether there is flow going to the flare, in addition to the continuous flow monitoring device. See response to comment II.F.1.(a) regarding the comment seeking a stakeholder process. (d) Comment (ExxonMobil, WSPA): A limitation of flare gas monitoring systems is the inability to provide for an independent ``in situ'' verification of accuracy. For example, there is no practical way to vary the flare gas flow that the monitor sees, and no practical way to utilize a reference method. Consequently, the calibration of a monitor is performed electronically, and the demonstration of accuracy is based on that calibration method. MSCC asserted that the proposed FIP does not provide adequate guidance to allow development of an acceptable QA/QC system for routine calibration or daily checks of the system. Without clear guidance, it is not possible to specify a system for a systems integrator (DAS/reporting) or an enduser to design or build a system to accomplish these checks.

Response: Since refinery flares contain highly variable flows and highly combustible material, in situ verification of flow measurement accuracy is difficult. For that reason, the performance specifications in the FIP rely in large part on procedures developed by the ultrasonic flow monitor manufacturers \7\ for commissioning monitors to assure the monitors will meet performance specifications on an ongoing basis. Manufacturers have established procedures for conducting annual or more frequent verifications of the performance of installed flow monitors as well as for the initial installation and performance verification (see reference document NNNNN). Based on manufacturer established procedures (Id.), we expect that the annual verification procedures will address elements such as:
\7\ Ultrasonic flow monitors will most likely be the monitors installed to meet the FIP's flow monitoring performance

specifications.

1. Verification of the Flowmeter with Reference Transducersthe purpose is to evaluate all flowmeter subsystems with factory certified ultrasonic transducers;

2. Mechanical Inspection of Flowmeter Transducersthe purpose is to visually verify the integrity of the flare gas flowmeter transducers and to clean any accumulated debris from the transducer faces;

3. Zero Flow Verificationthe purpose is to evaluate the operation of the transducer pair in the flare gas process (the integrity of the original process transducers is tested in a controlled environment);

4. Input/Output Verificationthe purpose is to verify the calibration of the analog I/O of the flare gas flowmeter;

5. Electronic Flow Simulationthe purpose is to demonstrate the operation of the flare gas flowmeter over the full measurement range of the instrument; and

6. Flowmeter System Reinstallation and Testthe purpose is to verify that all mechanical systems were properly aligned.

It should also be noted that since ultrasonic flow monitors do not contain any moving parts, their performance is not expected to deteriorate over time. One ultrasonic flow monitoring vendor provided information on the reliability and availability of the transducers (sensors in the flare that transmit and receive the ultrasound) they have installed. The information indicates that the 3,998 transducers installed between first quarter 2005 and first quarter 2007 had a reliability percentage of 94.32% and an availability percentage of 99.96%. (See reference documents MMMMM and XXXXXX.) (See also reference document LLLLL, ``A Total Approach to Flare Gas Flow Measurement for Environmental Compliance,'' Gordon Mackie, Jed Matson and Mike Scelzo, GE Sensing, Institute of Measurement and Control, Environmental Conference 2006, and reference document NNNNN, April 5, 2007, email from Jed Matson, GE Sensing, to Laurie Ostrand, EPA, containing flare gas flow meter procedures.
(e) Comment (COPC): ConocoPhillips asserts it would need to replace a GE Panametrics flare flow monitor that is wellsuited to the variable flow conditions it experiences, but does not conform precisely to the proposed specifications. It is difficult to quantify what additional benefit this change would provide although the cost is significant and quantifiable. The benefit evaluation is further clouded because of the rela

FOR FURTHER INFORMATION CONTACT Laurie Ostrand, Air and Radiation Program, Environmental Protection Agency (EPA), Region 8, 1595 Wynkoop Street, Denver, Colorado 802021129, (303) 3126437,
ostrand.laurie@epa.gov.