New Source Review reform adds complexity for US refiners

Fig. 1 shows the NSR program status in the US.

Although Pennsylvania's PSD program is SIP-approved, the state's rules use EPA-promulgated PSD rules at 40 CFR 52.21. The EPA and Pennsylvania are now enforcing the new reforms even though the program is SIP-approved.

The refining states of Texas, Louisiana, and Oklahoma will not see meaningful PSD reform until the respective SIPs are modified, sometime in 2006.

Although California and New Jersey enforce EPA's PSD rules, they—along with Pennsylvania and New England states—do not favor the reform program and have sued EPA to delay implementation. Refiners in those areas must realize that it may be necessary to obtain preconstruction permits from both the state and EPA region.

The New York State Department of Environmental Conservation, for example, has already publicly announced that it will no longer administer the PSD program; sources needing PSD permits are advised to contact the EPA office in Manhattan.

Promulgated changes

EPA promulgated five revisions to the PSD program in the Dec. 31, 2003, Federal Register:

• New procedures to determine the "baseline actual emissions" from which to measure whether a significant emissions increase has occurred.
• A new PSD applicability test that compares past actual emissions to future projected actual emissions (actual-to-future-projected actual [ATPA]) to determine if a project results in a significant emissions increase.
• New definitions of how companies can construct and implement plant-wide applicability limits (PALs) based on actual emissions.
• A new PSD applicability test for "clean units." This provides designated future projects at the clean units essentially a 10-year exemption from the need for a PSD permit, assuming emissions do not increase.
• A codified policy that excludes pollution control projects (PCPs) from needing a PSD permit.

Baseline actual emissions

EPA defined a new term, "baseline actual emissions," and a procedure for determining them in the final NSR rules. This term is not a replacement for the historical definition of actual emissions.

Baseline actual emissions are used to determine prechange emissions as part of the ATPA PSD applicability (or the traditional actual-to-potential emissions test), determine prechange emissions of a unit involved in a netting analysis, and set a PAL emissions cap. Historic actual emissions are still used for all air-quality impact analyses.

Under the new rules, a refiner at its discretion may choose any consecutive 24-month period in the previous 10 years (immediately before submittal of a complete permit application) to define baseline actual emissions.

The key points of this rule are that:

• All emissions units involved in a project must use the same consecutive 24-month period for each PSD-regulated pollutant.
• Sources may have different consecutive 24-month periods for different pollutants (including the determination of the netting baseline).
• Sources may use the last 2 years just as the old PSD rules required.
• The definition of projected actual emissions includes emissions due to start-ups, shutdowns, and malfunctions (SSM). It is therefore appropriate and necessary to include these emissions in the baseline where allowable.

Refiners must have adequate data to estimate baseline actual emissions accurately. Many existing plants have insufficient records older than about 5 years.

The refiner must adjust the baseline downward if it has accepted any legally enforceable limits that preclude operations at rates that were achieved during the baseline period. Also, the baseline must be adjusted downward if a source has agreed to any enforceable emissions limits in a permit.

Determining the optimum baseline period is not straightforward and adds at least 1 month to the application preparation for even a simple change. This time can be returned if PSD is not triggered but few refiners have 10 years of data.

Refiners must now establish new recordkeeping methods to take advantage of this provision in the future. Most companies have accepted legally enforceable limits in recent settlement negotiations.

ATPA applicability test

Revisions to the PSD rules supplement the existing actual-to-potential applicability test with an ATPA PSD applicability test for determining if a project triggers PSD. Any refiner can use either the ATPA or actual-to-potential applicability test.

The biggest disappointment is EPA's failure to adopt a potential-to-potential test. This approach bases emissions increases on increases in hourly potential emissions.

Projected actual emissions are based on the refiner's estimate of projected plant utilization and the emissions units' hourly emissions rate accounting for enforceable permit conditions. Additionally, the source may discount any increase in ATPA emissions above baseline actual emissions that could have been accommodated during the 24-month baseline period and is unrelated to the project.

A refiner can also exclude emissions resulting from "demand growth" that could have been accommodated before the change.

Key features of this rule are that:

• Companies are not obliged to make "projected actual emissions" an enforceable permit condition. Instead, companies must keep on site and readily available for inspection records that document the calculation of projected actual emissions.
• Records are required for: 5 years if the emissions unit's hourly potential-to-emit or design capacity does not increase, or 10 years if the hourly potential-to-emit or capacity does increase.
• Future projections should be based, in part, upon "information your company publishes for business-related purposes, such as a stockholder prospectus, or applications for business loans."

The company should keep significant records available for agency or public inspection. Typical records include continuous emission monitoring data, operational data, production levels, fuel-use data, or source test results.

The on site record of the ATPA applicability test should include a project description, identification of the emissions units with emissions changes affected by the project, and a description of the applicability test, including baseline actual emissions, projected actual emissions, and demand growth exclusions.

Most refiners are facing a 10-year recordkeeping requirement for turnaround projects; these can generally result in a minor increase in the potential-to-emit. The recordkeeping requirement is significant; therefore, facilities must consider establishing recordkeeping systems.

In exchange for additional recordkeeping, the company can use future projected actual emissions in the PSD applicability test in lieu of potential-to-emit. This could result in a project that is not significant under the new rules that was significant under the old rules.

Companies can avoid recordkeeping requirements by using the traditional actual-to-potential applicability test.

If a company's projection of future actual emissions is too low, a report to the permitting authority is required within 60 days after yearend. Although the rule does not discuss this type of event, it appears that PSD applies and enforcement would be at EPA's discretion.

Because the report is due at yearend, exceeding the projected actual emissions level has the potential for 365 daily violations of the PSD rule.

Plant-wide applicability limits

PAL is an optional rule that allows refiners to manage emissions without triggering major NSR. Although EPA has been issuing PAL permits since 1992, the reform rules establish clear procedures to determine the PAL baseline and renew or eliminate a PAL.

Sources with a PAL may make physical and operational changes without triggering NSR as long as emissions are below the caps specified in the PAL permit. If the cap is exceeded, a major NSR permit is then required.

Key features of this rule are that:

• Adding the significant level of the PAL pollutant to the baseline actual emissions of the PAL pollutant established the emissions cap.
• It must contain extensive monitoring, recordkeeping, and reporting requirements to ensure compliance.
• It must contain broad public participation requirements through either state rulemaking or permit-writing activities. The states retain considerable discretion to set the PAL lower than requested. States have the discretion, especially upon renewal, to lower the PAL based on air-quality concerns, regional growth margins, Class I areas, and other factors such as advances in control technology and cost-effectiveness.
• Emission controls or best-available control technology (BACT) are not required on new units, as long as the PAL is maintained.
• The PAL has a 10-year life.

EPA guidance on acceptable monitoring, recordkeeping, and reporting techniques in the preamble is conservative for refiners. Although mass-balance calculations are preferred, for example, EPA assumes that all of a PAL pollutant in fuel or raw material is emitted unless the refiner can otherwise accounted for it in the process.

In many cases, emissions factors alone are insufficient to establish a unit's actual emissions. Validation testing to develop site-specific emissions factors is needed if an emissions unit can emit a PAL pollutant at significant levels.

Emissions from SSM events can consume portions of the PAL emissions cap.

PALs can give refiners flexibility in exchange for additional monitoring, recordkeeping, and reporting requirements. As with baseline actual emissions, however, few refiners have sufficient data to construct the most advantageous PAL emissions cap.

More importantly, many rules do not allow for SSM emissions of criteria pollutants. Although exemptions for SSM emissions often appear in state regulations, they are not often part of the approved SIP.

Because the baseline actual emissions can only include legally allowable emissions, SSM emissions are not in the baseline, but they still consume PAL emissions allowances. Refiners need to be careful about this apparent inequity.

Provisions for recordkeeping also present many traps for a refiner. Combined with the recordkeeping requirements, there are many opportunities for unexpected enforcement problems. In particular, relying on emission factors adds a new level of uncertainty to EPA's ability to find a violated PAL based on different emissions measures.

Finally, refiners must realize the substantial procedural requirements, public participation requirements, and the substantial discretion of the state in revising a PAL.

Clean unit test

In the NSR revisions, EPA promulgated a new PSD applicability test for emissions units that are designated "clean units." A refiner, assuming emissions don't change, can make physical or operational changes at clean units without triggering PSD.

A unit that has undergone a formal BACT or lowest-achievable emission rate (LAER) determination is automatically a clean unit.

A unit that has been subject to equivalent controls can apply for clean unit status. The application process, however, requires an opportunity for a public hearing and an ambient air impact analysis.

Any refiner that has recently experienced a BACT or LAER analysis can take advantage of this new applicability test. There is no apparent downside.

Refiners that believe they have installed equivalent controls under a state permit should apply for clean-unit status. The process is similar to a permitting process and will likely require 9-12 months.

Even those refiners in states running a SIP-approved program should consider applying to EPA for clean unit status to be able to use the new applicability test immediately upon revision of the applicable SIP.

PSD exclusion

The revised PCP exclusion exempts (from the PSD program) physical changes at existing sources that are intended to reduce pollution or a particular pollutant.

There are essentially two ways a project can qualify as a PCP.

First, the rule includes a listing of projects that automatically qualify for the PCP. Typical air-pollution-control devices at refineries that automatically are considered PCPs include low-NOx burners, scrubbers, selective catalytic reduction, selective noncatalytic reduction, and floating roof storage tanks.

Second, any refiner may apply for the PCP exclusion through a case-by-case analysis wrapped into a permit. Projects that rely upon the PCP exclusion must be environmentally beneficial and may not cause a violation of a PSD increment, national ambient air quality standards, or air-quality related value.

The PCP exclusion can benefit refiners. Projects that are automatically considered PCPs only require a notice to the permitting authority before construction. The notice must address the rule's environmentally beneficial demonstration provisions; however, no permit is required, which clearly shortens the review cycle.

A minor permit is needed for projects requiring a case-by-case demonstration, which requires some lead-time.

Future NSR reforms

EPA has made it clear that it is still working on four important reforms to the NSR rules:

• In the Dec. 31, 2003 Federal Register, EPA proposed a definition of RMRR. (The final rule was issued after this article was written in August 2003. Implementation is currently stayed due to a challenge in the DC Circuit Court.)
• EPA has announced that it intends to propose additional rules governing when projects should be aggregated for a PSD applicability test, how to handle debottlenecked units, and the creation of PALs based on allowable emissions as opposed to actual emissions.

Three of these four reforms are critical to the refining industry. The RMRR proposal is critically important to the industry. The project aggregation and debottlenecking issues are also important to refiners.

In the enforcement initiative cases, EPA showed a propensity to aggregate projects over many years on the theory that, taken in total, they resulted in significant emissions increases because they increased refinery capacity. Of course, any project can be viewed as adding to overall capacity; a bright-line would therefore be beneficial.

Similarly, with respect to debottlenecked emissions units, EPA is likely to propose a rule eliminating debottlenecked units from the review. Until that time, the PSD permit application process will be complex because refiners must determine baseline actual and projected emissions for each debottlenecked unit, and maintain records for 5-10 years.

RMRR proposal

The proposed rule would add a definition of RMRR to the NSR rules. Uncertainty of what constitutes RMRR is the heart of many of EPA's enforcement actions.

The uncertainty in the current rule concerning whether a project is RMRR delays many projects; the only way to determine if an activity is RMRR is to negotiate with the local permitting authority. Most refiners consider turnarounds to be routine; but as many have learned, EPA usually disagrees.

The key points of this proposed rule are that:

• EPA will establish an annual maintenance, repair, and replacement allowance (AMRRA). The intent is to exclude certain activities related to the safe, reliable, and efficient operation of a facility from NSR review, provided that the project's total aggregated costs do not exceed the AMRRA. The AMRRA is not established yet; EPA is considering an approach similar to that used in the new-source performance standards.
• EPA will still maintain the case-by-case analysis.
• EPA may establish an equipment replacement provision that excludes NSR replacement activities if the costs are less than a specified but, as of yet, undefined percentage of a unit's replacement value. The replacement equipment would have to serve the same function as the original equipment and not alter a facility's basic design parameters. The stayed rule addresses only a new equipment replacement provision.
• As safeguards, EPA will not allow the annual allowance to cover certain activities. These include new process units, replacement of an entire process unit, or changes that increase hourly potential-to-emit.
• EPA is proposing a refinery-specific definition of process unit.

The current approach to RMRR (and project aggregation and debottlenecked emissions units) creates huge discontinuities in the NSR program.

Although EPA claims that RMRR is based on five factors—the nature, extent, purpose, frequency, and cost of a change—the Detroit Edison guidance document lays out 27 criteria that EPA used to determine if an activity is RMRR. No refiner is assured that the guidance is interpreted correctly. It is possible to reach any conclusion based upon the subjective criteria represented in Detroit Edison.

Until this issue is addressed in promulgation, refiners remain at enforcement risk for almost any activity associated with a plant turnaround.

Ambiguity in EPA's RMRR exclusion and pending enforcement actions creates no objective tie to determine if the RMRR exclusion applies.

Moreover, misuse of the future RMRR exclusion appears likely given the proposal's complexity.

Industry needs a bright-line test, but the complexity of the proposed approach may be a concern for many refiners.

Potential accounting rules will be important. Currently EPA proposes an annual approach but is open to longer periods for cyclical industries, such as refining, to deal with turnarounds.

Enforcement issues

Two issues must be addressed with respect to enforcement of the NSR program:

• What impact will the new NSR reform provisions have on EPA's existing enforcement initiative? Do some aspects of the proposal seem to help industry in the current NSR litigation? These include EPA's admission that pump replacements appear to be routine, and replacement of worn out equipment with updated and better versions is routine.
• Will the NSR reform provisions make it more likely that refineries will not unknowingly trigger NSR in the future?

The answers to these questions, unfortunately, reveal that the industry remains truly exposed to NSR and needs the substantial revisions promulgated before meaningful relief is available.

Existing enforcement initiative

EPA and the courts have made it clear that the promulgated and proposed NSR reforms do not apply retroactively.

This is an except from US' Response to Defendant's Notice of Supplemental Authority on Fair Notice and Routine Maintenance, filed on Jan. 14, 2003, in US vs. Southern Indiana Gas and Electric Company, Case No. IP99-1692-C-M/F, pending in the US District Court for the Southern District of Indiana:

"Counsel for the US have consulted with EPA management regarding SIGECO's 'Supplemental Notice' and have confirmed that the US' briefs currently before the Court accurately represent EPA's views, and that EPA's position is not affected by the recently published rulemaking proposal.

"The Dec. 31, 2002, proposal makes clear in at least seven places that the agency is merely soliciting public comments on possible changes to regulations which, if they are ultimately adopted, would be effective only, in the future [emphasis added]. Thus, the proposal makes no change in the governing regulations and represents no change in EPA's longstanding interpretation of them."

At a recent environmental conference in Keystone, Colo., John Cruder, Deputy Assistant Attorney General, Environmental and Lands Division, Department of Justice, announced that the DOJ would continue the refiners' enforcement initiative.

There are, however, several cases pending decision that will shed further light on the issue.

Additionally, EPA has recently announced plans to settle with one refiner and remains in active negotiation with several major refiners. Although the pace of settlement has slowed, EPA so far has not backed off.

Potential NSR enforcement issues

In its Enforcement Alert regarding the NSR program, EPA's Office of Enforcement and Compliance Assurance provided a discussion of the most common issues it noticed when reviewing NSR permit applications. These issues include:

• Improper use of exemptions. Essentially EPA implied that refiners relied improperly on the RMRR exclusion for many projects that should have been considered nonroutine and therefore subject to NSR.

This is the primary issue surrounding most of the enforcement initiative and aims squarely at turnaround issues. Replacing packing in a main fractionator column, for example, often changes the pressure balance in an FCCU. EPA, in enforcement proceedings, believes that this frees up air capacity in the FCCU and therefore increases emissions. Most refiners consider this activity routine maintenance.

Until a firm definition of RMRR is promulgated, refiners remain in the same precarious position they have been in for the last decade.

Nozzle replacement projects, wall replacements, cyclone replacement projects, and catalyst replacements all fall in this category.

• Improper emissions estimates. The NSR reform program makes this more difficult instead of less difficult; now affected facilities must determine baseline actual emissions. The 10-year look back and concerns about what represents adequate data are new challenges for refiners.

A huge source of potential error in many applications is the estimate of debottlenecked emissions.

As previously mentioned, the NSR reform program fails, so far, to address theses issues.

• Failure to recognize changes. Similar to the improper use of exemptions, this issue is a huge concern until EPA provides a firm approach to RMRR. And because debottlenecked units are often an issue associated with the failure to recognize changes, relief will not be available to the industry until EPA firmly defines an approach to debottlenecking.
• Improper netting. The NSR reform rules make netting more complex than in the past due to the need to calculate baseline projected emissions for each emissions unit involved in the netting analysis.

Because the units could each have a different definition of the time period representing baseline actual emissions, refiners must spend considerable effort defining the proper time period and then adjusting emissions levels to compensate for new requirements over the years.

Although the approach may be more fair, it is definitely more complex than the past procedure.

If EPA believed this was an underlying cause in NSR noncompliance, then the complexity can only lead to future concerns.

Recent developments

Recently, EPA has issued final approval for the NSR RMRR rule with respect only to the equipment replacement provisions, and opponents of the RMRR rule appealed the new provisions to the US Court of Appeals for the DC Circuit, which granted a stay. Also, two court decisions, based on two different interpretations of the original NSR rules (Duke and First Energy) have been issued.

The authors

Kenneth N. Weiss ([email protected]) is a principal with ERM, Exton, Pa., and is the national leader of the firm's air practice. He has more than 30 years' experience in air-quality issues management for government and industry. Weiss' areas of interest include Title V operating permits, NSR, and MACT. He has assessed the impact of the Clean Air Act on numerous major US corporations and facilities. Previously, he managed the environmental affairs department for Reynolds Metals Co. Before that, he was an environmental engineer for the Georgia Environmental Protection Division. Weiss holds a BS (1972) in chemical engineering from Georgia Tech and an MBA (1975) from Georgia State University.

Louis E. Tosi ([email protected]) is a partner in the environmental practice group for Shumaker, Loop & Kendrick LLP, Toledo. His principal areas of practice are environmental law, administrative law, and litigation. Tosi has experience in representing companies in matters before the US and Ohio Environmental Protection Agencies. He also regularly represents a number of industry groups in a variety of rulemaking settings and Comprehensive Environmental Response, Compensation, and Liability Act matters. He holds a BA and JD (1974) from Ohio State University. Tosi is a member of the Toledo, Ohio, and American bar associations.