------------------------------------------------------------------------------- On 07/09/91 [L-S document 396271, 56 FR 31176, 10766 lines] ENVIRONMENTAL PROTECTION AGENCY 40 CFR Part 80 [AMS-FRL-3965-9] Regulation of Fuels and Fuel Additives: Standards for Reformulated Gasoline AGENCY: Environmental Protection Agency. ACTION: Notice of proposed rulemaking. ----------------------------------------------------------------------------- SUMMARY: EPA proposes, in today's notice, two related programs implementing section 211(k) of the Clean Air Act (CAA or the Act) as amended by Public Law 101-549. The primary program under that section requires that gasoline sold in the nine worst ozone nonattainment areas be reformulated to reduce toxic and ozone-forming volatile organic compound (VOC) emissions. The second program prohibits gasoline sold in the rest of the United States from becoming more polluting. These regulations will take effect on January 1, 1995. DATES: Comments on this proposal will be accepted through August 15, 1991. EPA will hold a public hearing on July 15, 1991, from 9 a.m. until 5 p.m. and on July 16, 1991, from 8 a.m. until 3 p.m., regarding the contents of this proposal. If, after publication of this proposal but prior to the July 15-16, 1991, hearing, the Agency has issued a supplementary notice based on the results of a consensus that is reached through a continuing negotiated rulemaking process, the public hearing will also cover the contents of that notice. Additional information on the submission of comments and the public hearing can be found under "Public Participation" in the Supplementary Information section of today's notice. ADDRESSES: Interested parties may submit written comments (in duplicate if possible) to Air Docket Section (LE-131), U.S. Environmental Protection Agency, Attention: Docket No. A-91-02, 401 M Street, SW., Washington, DC 20460. The Agency requests that a separate copy also be sent to Carol Menninga, SDSB-12, EPA Motor Vehicle Emission Laboratory, 2565 Plymouth Road, Ann Arbor, MI 48105. The hearing will be held at the Westpark Hotel, 1900 North Fort Myer Drive, Arlington, Virginia. Materials relevant to this proposal have been placed in Docket No. A-91-02 by EPA. The docket is located at the above address in room M-1500, Waterside Mall (ground floor), and may be inspected from 8:30 a.m. to 12 p.m. and 1 to 3 p.m., Monday through Friday. A reasonable fee may be charged by EPA for copying docket materials. FOR FURTHER INFORMATION CONTACT: Carol Menninga, Standards Development and Support Branch, U.S. Environmental Protection Agency, 2565 Plymouth Road, Ann Arbor, MI 48105, telephone: (313) 668-4575. Richard A. Rykowski, Standards Development and Support Branch, U.S. Environmental Protection Agency, 2565 Plymouth Road, Ann Arbor, MI 48105, telephone: (313) 668-4339. To request copies of this notice contact: Marie Tolonen, Standards Development and Support Branch, U.S. Environmental Protection Agency, 2565 Plymouth Road, Ann Arbor, MI 48105, Telephone: (313) 668-4295. SUPPLEMENTARY INFORMATION: I. Use of Regulatory Negotiations EPA is developing the regulations proposed today through the process of regulatory negotiation. Over the past several years, the Agency has developed and employed this process as a means of developing rules that are acceptable to all the interests that will be significantly affected by the rules. Rules so developed are far less likely to be challenged in court. The process entails convening a negotiating committee that consists of representatives from EPA and all affected interests, generally including other government agencies, states, localities, industry, consumers and environmental groups. The Negotiated Rulemaking Act of 1990, Public Law 101-648, expressly authorizes use of the negotiated rulemaking process in appropriate circumstances and sets forth procedural requirements which the Agency has met for the negotiations being conducted on these regulations. A complete description of the negotiated rulemaking process and the Agency's decision to use that process for these regulations may be found in the notice of "Intent to Form an Advisory Committee to Negotiate Guidelines and Proposed Regulations Implementing Clean Fuels Provisions" (56 FR 5167, February 8, 1991) and in the notice of "Open Meeting of the Negotiated Rulemaking Advisory Committee; Clean Fuels Rules and Guidelines" (56 FR 8972, March 4, 1991). The Agency first convened the negotiating committee for these regulations on March 14, 1991. The committee includes representatives from EPA, the Department of Energy, the State and Territorial Air Pollution Program Administrators, the Association of Local Air Pollution Control Officials, the Northeast States for Coordinated Air Use Management, the California Air Resources Board, the American Petroleum Institute, the National Petroleum Refiners Association, the American Independent Refiners Association, the Rocky Mountain Small Refiners Association, the Clean Fuels Development Coalition, the Oxygenated Fuels Association, the Renewable Fuels Association, the American Methanol Institute, the National Council of Farmer Cooperatives, the National Corn Growers Association, the Petroleum Marketers Association of America, the Society of Independent Gasoline Marketers of America, the Independent Liquid Terminals Association, the Motor Vehicles Manufacturers Association, the Association of International Automobile Manufacturers, Citizen Action, the Sierra Club, the American Lung Association, and the Natural Resources Defense Council. The Agency believes that the negotiating committee has made significant progress towards arriving at a consensus on regulations implementing section 211(k). The Agency wants the regulatory negotiation process to continue because it believes that consensus can be reached. However, for EPA to meet the statute's November 15, 1991 deadline for promulgating final regulations implementing section 211(k), the Agency believes that it cannot wait for consensus to be reached before publishing a proposal. EPA has therefore decided to present in today's notice the positions being taken on different issues by one or more parties to the negotiation. The fact that an option has been included in this notice does not mean that every party to the negotiations, including the Agency, believes that it would be appropriate to adopt the option in the final rule. Indeed, a number of participants have expressed opposition to many of these options. Their inclusion in this notice does not constitute endorsement by all of the participants. EPA requests comments on the appropriateness of each of these options. If the negotiating committee reaches consensus, EPA will provide a supplemental notice describing the committee's recommendations and provide an opportunity for the public to comment on those recommendations. II. Statutory Provisions A. Overview Section 211(k) of the Act as amended prohibits the sale of gasoline that EPA has not certified as reformulated from being sold to consumers in the nine large ozone nonattainment areas that experienced the worst ozone pollution during the period of 1987 through 1989. Any other ozone nonattainment area may have the prohibition applied to gasoline sold within its borders at the request of the governor of the state in which it is located. Further, conventional gasoline sold elsewhere may not be more polluting than it was in 1990. The prohibitions take effect beginning on January 1, 1995, although a later effective date may be provided under certain circumstances in the case of areas opting into the reformulated gasoline program. Section 211(k) requires EPA to promulgate regulations establishing requirements for reformulated and conventional gasolines within one year of the amendments' enactment, i.e., November 15, 1991. Those regulations must include the specifications and performance standards that gasoline must meet to be considered reformulated; a process for certifying gasolines as reformulated; a program for granting tradeable credits to fuel producers that certify reformulated gasoline which is less polluting than required; and provisions implementing the prohibition against sale of conventional gasoline which is dirtier than it was in 1990. B. Compositional Specifications for Reformulated Gasoline Section 211(k)(2), provides that EPA's regulations shall require that reformulated gasoline comply with each of the following compositional requirements: An oxygen content not less than 2.0 percent by weight, a benzene content of not more than 1.0 percent by volume, and no heavy metals, including lead or manganese. That provision, however, permits the Administrator to waive the application of the oxygen content requirement and the heavy metal ban under certain circumstances. If the Administrator determines that compliance with the oxygen content requirement would interfere with the attainment of a national primary ambient air quality standard in a covered area, the application of that requirement may be waived, in whole or in part, for that area. In addition, if the Administrator determines that the addition of a heavy metal other than lead to reformulated gasoline will not increase toxic air pollutant emissions from motor vehicles on either an aggregate mass or a cancer-risk basis, the prohibition against the use of that metal in reformulated gasoline may be waived. C. Emission Standards for Reformulated Gasoline 1. NOx Emissions Section 211(k)(2)(A) also provides that emissions of oxides of nitrogen (NOx) from baseline vehicles may be no greater when using reformulated gasoline than when using the baseline gasoline which is defined in CAA section 211(k)(10) and described below in section D of this proposal. If the Administrator determines that compliance with this limitation is technically infeasible, considering the other requirements applicable under section 211(k), the Administrator may adjust or waive any of the other requirements, as appropriate, to ensure compliance with the NOx emission limit. 2. VOCs and Toxic Air Pollutants At the heart of section 211(k) are requirements that reformulated gasoline reduce ozone-forming and toxic air pollutant emissions. Section 211(k)(1) states that the EPA regulations implementing that section "shall require the greatest reduction in emissions of ozone forming volatile organic compounds (VOCs) (during the high ozone season) and emissions of toxic air pollutants (during the entire year) achievable through the reformulation of conventional gasoline, taking into consideration the cost of achieving such emission reductions, any non-air-quality and other air-quality related health and environmental impacts and energy requirements." Section 211(k)(10)(C) defines "toxic air pollutants" to mean the aggregate emissions of benzene, 1,3- butadiene, polycyclic organic matter (POM), acetaldehyde, and formaldehyde. Section 211(k)(3) requires that VOC and toxic emissions from baseline vehicles be reduced relative to the emissions attributable to the baseline gasoline. The reduction requirements for VOC emissions would apply during the high ozone season, and the requirements for toxic emissions would apply year round. The Act requires that both VOC and toxic emissions shall be assessed on a mass basis, rather than an ozone-forming or reactivity basis for VOCs, or a cancer-causing basis for toxic air pollutants. Under section 211(k)(3), the reductions in VOC and toxic emissions that reformulated gasoline must achieve are to be determined by comparing the emission reductions resulting from the use of a "formula" fuel (defined by section 211(k)(3)(A) and described below in section F of this proposal) with a specified 15 percent reduction in emissions from baseline vehicles (defined by section 211(k)(10)(A) as representative model year 1990 vehicles). The more stringent emission reduction becomes the minimum standard. VOC and toxic emission standards are to be determined separately, so that the "formula" fuel might set the standard for toxic emissions, while the 15 percent reduction standard might apply to VOCs. The determination of the proposed standards for both VOC and toxic emissions is discussed below in section III.D of this proposal. For the year 2000 and beyond, section 211(k)(3)(B) provides that VOC and toxics emission reduction requirements for reformulated gasoline are to be based on a comparison of the emission performance of the formula fuel to a 25 percent reduction in VOC and toxic emissions, relative to emissions from the baseline gasoline. Considering technological feasibility and cost, the Agency may adjust the "Phase II" emission standard to require more or less emission reductions, but in no case shall such an adjustment provide for less than a 20 percent reduction. Phase II reformulated gasoline standards will be developed through a later rulemaking. D. Composition of Baseline Gasoline Section 211(k)(10) defines "baseline gasoline" to mean one thing in the summer, which is the high ozone season, and another thing in the winter. It specifies the composition of summer gasoline in detail and leaves the composition of winter gasoline to be determined by the Administrator. 1. Summer Baseline Gasoline In the case of gasoline sold during the high ozone season, section 211(k)(10) defines 'baseline gasoline' as a gasoline which meets the following specifications: Baseline Gasoline Fuel Properties API Gravity 57.4 Sulfur, ppm 339 Benzene, volume percent 1.53 RVP, psi 8.7 Octane, R+M/2 87.3 IBP, degrees F 91 10%, degrees F 128 50%, degrees F 218 90%, degrees F 330 End Point, degrees F 415 Aromatics, volume % 32.0 Olefins, volume % 9.2 Saturates, volume % 58.8 There are, however, other fuel parameters for which the definition of baseline summer gasoline contains no specifications. Levels of oxygen, lead, and deposit-control additives are all defined for the 'formula' fuel as described below in section E, but are not specified for the baseline gasoline. EPA's proposals regarding the other parameters of summer baseline gasoline are discussed in section III.A.1 of this notice. 2. Winter Baseline Gasoline As noted earlier, the Act does not specify the composition of winter baseline gasoline. Instead, it requires that EPA establish specifications based on industry average gasolines sold in other than the high ozone season in l990. The high ozone season is defined in section II.H of this proposal. EPA's proposed specifications for winter fuel are set forth in section III.A.1 of this proposal. E. Formula Fuel Section 211(k)(3)(A) describes the following formula fuel as containing: --No more than 1.0 volume percent benzene, --No more than 25 volume percent aromatics, --At least 2.0 weight percent oxygen, --No lead, and --Additives to prevent the accumulation of deposits in engines or vehicle fuel supply systems. The Act is silent regarding many other compositional parameters of the 'formula' fuel, such as sulfur, Reid Vapor Pressure (RVP), octane (R+M/2), distillation points, API gravity, olefins, and saturates. EPA's proposal regarding these compositional elements of the formula fuel is set forth in section III.D of this notice. F. Affected Nonattainment Areas 1. Areas Covered by the Operation of the Law Section 211(k)(5) of the Act prohibits the sale of gasoline not certified as reformulated (i.e., conventional gasoline) in "any covered area." Section 211(k)(10) defines covered areas as the nine ozone nonattainment areas with the highest ozone design values during the years from 1987 through 1989 and with a 1980 population of over 250,000. Also defined as covered areas are ozone nonattainment areas reclassified as severe ozone nonattainment areas under section 181(b) of the Act, effective one year after reclassification. Title I of the Clean Air Act defines five levels of ozone nonattainment. In order of increasing severity, they are: Marginal, moderate, serious, severe, and extreme. Based on 1987 through 1989 ozone design values, a total of 96 areas of the country are at some level of ozone nonattainment. Of those, nine are classified under section 181(b) as severe or extreme, and fifteen as serious. To be reclassified as severe, an area must have an ozone design value of 0.16 ppm or higher. Based on United States census data and 1987 through 1989 ozone air quality design values, as published by EPA's Office of Air Quality Planning and Standards in August, 1990, the nine nonattainment areas that meet the criteria of the nine originally covered areas are the metropolitan statistical areas (MSAs) or consolidated metropolitan statistical areas (CMSAs) containing Baltimore, Maryland; Chicago, Illinois; Hartford, Connecticut; Houston, Texas; Los Angeles, California; Milwaukee, Wisconsin; New York, New York; Philadelphia, Pennsylvania; and San Diego, California. Of these nine areas, Los Angeles is designated as extreme, Hartford as serious, and the remaining seven are severe ozone nonattainment areas. A more detailed description of these areas and the counties they include can be found below in section VIII.C. 2. Opt-In Areas Under section 211(k)(6), the governor of a state may apply to have any ozone nonattainment area in the state included in the reformulated gasoline program. Upon receiving an application, the Agency is to publish it in the Federal Register and establish an appropriate effective date for including the area in the program, to take place not later than January 1, 1995, or one year after the application is received, whichever is later. The Administrator may extend the effective date by up to three years if he determines, in consultation with the Department of Energy, that there is an insufficient domestic capacity to produce the reformulated gasoline needed to supply opt- in areas. If the Administrator so finds, he is to extend the effective date for areas with lower ozone classifications before doing so for areas with higher classifications. According to EPA's data, there are 87 ozone nonattainment areas eligible to opt into the reformulated gasoline program. G. Averaging and Trading Section 211(k)(7) requires EPA to grant credits to persons who make or import and who certify gasoline that has more oxygen or less benzene or aromatics than is required to comply with sections 211(k)(2) and (3). Such credits may be used to demonstrate compliance with section 211(k) requirements, and they are tradeable. However, they must be used within the nonattainment area in which the credit-generating gasoline is sold. In addition, EPA may not grant or permit transfers of credits to the extent that the use of such credits would result, on average, in lower levels of oxygen or higher levels of benzene or aromatics in conventional gasoline in a nonattainment area than would occur in the absence of using such credits. H. Fuel Certification Section 211(k)(4) calls for fuels to be certified as reformulated if they comply with the compositional requirements and NOx emission limit established under section 211(k) and if they achieve equivalent or greater reductions than are achieved by a gasoline meeting the formula-fuel- or performance- standard-based requirements established under section 211(k)(3). EPA is to act on certification petitions within 180 days of receipt. If the Agency fails to act in time, the fuel is deemed certified until EPA does act. I. Prohibitions The linchpins of the reformulated gasoline provisions are the prohibitions in section 211(k)(5). That section provides that, beginning on January 1, 1995, the sale or dispensing by any person of conventional gasoline to ultimate consumers (e.g., car owners) in any covered area is prohibited. (Section 211(k)(10)(F) defines "conventional gasoline" as any gasoline that does not meet the specifications of a certification issued under section 211(k)). Section 211(k)(5) also prohibits any refiner, blender, importer, or marketer from selling or dispensing conventional gasoline for resale in a covered area (e.g., to a wholesaler) without segregating it from reformulated gasoline and clearly marking it as "conventional gasoline, not for resale to ultimate consumers in a covered area." Relatedly, it prohibits anyone who purchases properly segregated and labeled conventional gasoline from labeling, representing, or wholesaling it as reformulated gasoline. EPA is authorized to impose sampling, testing, and recordkeeping requirements to prevent violations of these prohibitions. J. Anti-Dumping Section 211(k) provides, not only for a program that cleans up gasoline sold in ozone nonattainment areas, but a program that maintains, on average, the current quality of gasoline sold in the rest of the country. The provisions of section 211(k)(8), which establishes the "anti-dumping" program, are described in section IX of this notice. III. Derivation of Emission Standards The first step in implementing the reformulated gasoline program is to determine the levels of the applicable emission standards. In the case of NOx, VOC, and toxic emissions, the Act specifies standards that are relative to emission levels from baseline vehicles using baseline fuel. In addition, it requires EPA to determine whether the specified formula fuel achieves reductions in toxic and VOC emissions greater than 15 percent of baseline emissions, as the specified performance standard otherwise requires. Finally, section 211(k)(1) provides that the VOC and toxic standards require the greatest achievable reductions, considering specified factors. In this section, EPA first lays the foundation for determining the emission standards by proposing a determination of baseline emissions. To do so, it defines or supplements the statutory definitions of baseline fuel, high ozone season, and baseline vehicle. The Agency then considers how different fuel parameters affect emissions of the regulated pollutants. With that background, it judges the emission reduction benefits of the formula fuel and the achievability of VOC and toxic emission reductions greater than those derived from a comparison of the formula fuel and the performance standard. A. Baseline Emissions 1. Baseline Fuel As described earlier, the statutory definition of baseline fuel specifies most, but not all, of the components of summer baseline fuel and delegates to EPA the task of defining winter baseline fuel based on 1990 industry averages. a. Summer Baseline Gasoline. As mentioned above in section II.D of this notice, there are some fuel parameters which are specified for the formula fuel, but for which the baseline summer gasoline, as defined in the Act, contains no specifications. For summer baseline gasoline, EPA proposes that the oxygen level be zero. Oxygenates were used in some fuels in 1990, with the most widely used fuel oxygenates being MTBE at an average of less than 2 percent and ethanol at less than 1 percent by volume. However, oxygenated fuels were often sold in specialized markets, such as under state-mandated programs to reduce wintertime emissions of carbon monoxide in CO nonattainment areas, in interim reformulated gasolines, or in some premium gasolines to increase their octane level. Because oxygenates were not more widely used, EPA believes it appropriate not to include oxygen in the summer baseline gasoline. The Agency welcomes any comments on the appropriate oxygen level of baseline summer gasoline. While lead had not been entirely eliminated from all gasoline sold in 1990, its use continues to decrease over time. To ensure a complete elimination of lead from gasoline, Congress included, in its 1990 amendments, subsection 211(n), which prohibits the use of leaded gasoline in motor vehicles beginning after December 31, 1995. Because gasoline sold for highway use will be virtually leadfree by the time the reformulated gasoline program takes effect, and additionally because the baseline 1990 model year vehicles are all designed to run on unleaded gasoline, the Agency believes that it is appropriate to define baseline gasoline as thus containing zero lead. Comments on this determination are encouraged. Deposit-control or detergent additives, while not specified for the baseline gasoline, were present to some degree in almost all gasoline sold in 1990. They are also present in gasolines used in vehicle emission testing programs, such as EPA's Emission Factor (EF) test program and the Auto/Oil Research Study, which is sponsored by both the automobile and petroleum industries, to determine the effects on emissions due to changes in certain fuel parameters. The inclusion of deposit-control additives in the 'formula' fuel, as described above in section II.E of this notice, is provided for in the Act. Section 211(1), of the Act, which, like the reformulated gasoline provisions, will take effect on January 1, 1995, requires the use of effective deposit-control additives in all of the nation's gasoline. In light of these factors, the Agency proposes that the baseline fuel likewise contain detergent additives. b. Winter Baseline Gasoline. As for baseline gasoline for wintertime comparisons, the Agency has based its proposed specifications on an analysis of surveys performed by the Motor Vehicle Manufacturers Associations and by Southwest Research Institute. i. Base Fuel Parameter Values. The data EPA used to determine the winter baseline gasoline specifications was culled from the Southwest Research Institute (SwRI) 1990 gasoline surveys (January-April and October-December) and the Motor Vehicle Manufacturer's Association (MVMA) 1990 winter gasoline survey (January). Fuel parameter values are presented in these surveys by city, grade and month. The number of samples per city, grade and month is also provided. These surveys, however, only sampled unleaded grades of gasoline. Nevertheless, since the vast majority of vehicles in 1990 operated on unleaded fuel, EPA believes that the 1990 baseline fuel can be appropriately based on unleaded gasoline specifications. ii. Sales Data. The best publicly available data on 1990 fuel sales is from Petroleum Marketing Monthly (PMM), a Department of Energy, Energy Information Administration publication. PMM reports fuel sales by grade, month, and states. The use of state data assumes uniform fuel consumption throughout a state. The sales data from PMM is listed under "Volumes of First Sales of Motor Gasoline by Grade." This data reflects first delivered sales of gasoline into the states where it is expected to be consumed. Comments concerning the appropriateness of these or other references for determination of an industry average fuel, and of specific data within these references, are requested. iii. Calculation Methodology. Initial survey parameter values were obtained by city, grade and month by combining SwRI and MVMA data. The following discussion explains the methodology used to combine the data of these two surveys to obtain a single initial parameter value by city, grade and month for the winter season. During the winter season, MVMA data is available for all parameters of interest only for the month of January. For all parameters except sulfur, aromatics, olefins and saturates, SwRI data is available year round on an every-other-month basis. For those four parameters, data is available only during January/February and July/August. MVMA samples in 23 continental U.S. cities while SwRI samples 53 continental U.S. cities. Twenty-two continental U.S. cities are sampled by both surveys. As mentioned above, for all but four fuel parameters, gasoline samples in the SwRI survey were taken on an every-other-month basis. Some cities were sampled on a January-March-May-July-September-November schedule, others on a February-April-June-August-October-December schedule. For averaging purposes, survey months were grouped into the following two-month pairs: January/ February, March/April, May/June, July/August, September/October, November/ December. Thus, data for SwRI cities sampled in January and SwRI cities sampled in February were assumed representative of gasoline sold in the city in the January/February time frame. For determination of a winter season baseline, the January/February, March/April, and November/December bimonthly data was used, along with the October data in those SwRI cities sampled in October. September data was not included because September is not a non-high ozone season month. For those 22 cities sampled both by MVMA and SwRI, the January data from MVMA and the Jan/Feb data from SwRI were averaged together on a sample- weighted basis (i.e., each sample from each survey was treated equally and independently) for each fuel parameter, gasoline grade and city. This step was unnecessary for the single MVMA city not part of the SwRI survey or for the 31 SwRI cities not part of the MVMA survey. iv. Nationwide Parameter Value Calculation. The calculation of a nationwide average for each fuel parameter of the winter baseline gasoline was accomplished step-wise as follows. The values for a given fuel parameter for each city, grade and bi-monthly (including October) period (either from the combined SwRI and MVMA data as described above or from either survey independently) were first weighted by grade fraction based on the gasoline grade sales data in PMM, resulting in one fuel parameter value per city per bi-monthly period. Next, using state fuel consumption data from PMM, the fuel consumption per city per bi-monthly period was determined by assuming uniform fuel consumption within a state and using the ratio of city to state population. The fraction of each city's fuel consumption per bi-monthly period to the total fuel consumption of all the survey cities in the period was then calculated using city population data. This resulted in a fractional fuel parameter value for each city in that bi- monthly period. The sum of these fractional fuel parameter values over all survey cities and bimonthly periods yielded the final, nationwide winter baseline fuel parameter. v. Oxygenate. The Agency proposes not to include oxygen content as a specification of the winter baseline gasoline for the same reasons it proposed not to include oxygen in the specification of summer baseline gasoline. Comments are requested concerning the inclusion or exclusion of oxygen content in the development of the industry average baseline gasoline. vi. Geography. The Agency proposes to determine the winter baseline gasoline on a nationwide basis (as per the summer baseline gasoline defined in the Act). Comments are requested on using this approach versus developing a regional, PADD or other geographically-limited baseline. vii. Final Winter Baseline Specification. Using the methodology described above, the following specifications for winter baseline gasoline were determined, and EPA proposes them today: Benzene, volume percent 1.64 Aromatics, volume percent 26.3 Olefins, volume percent 11.9 T90, degrees F 332 T50, degrees F 199 Sulfur, ppm 340 RVP, psi 12.3 2. High Ozone Season The term 'high ozone season' is used in section 211(k) to refer to the period of time during which VOC control is required. Sections 211(k) (3)(B)(i) and (10)(B)(i) provide that the Administrator is to define that period of time. The Agency is considering three options for defining the high ozone season. Under the first option, high ozone season would refer to the period from May 1 through September 15. This is the period during which most of the nation's ozone exceedances occur. (96 percent of all ozone exceedances from 1986-1988 occurred during this period, excluding those in California and Houston, which tend to experience high ozone concentration levels all year. Of the ten highest ozone levels experienced in each of the "nine cities" from 1986-1988, including California and Houston, 87 percent occurred from May 1 through September 15. This is also the period over which EPA volatility (RVP) standards for gasoline apply in all 48 contiguous states (40 CFR 80.27). A parallel schedule for reformulated gasoline and for volatility control would permit refiners of reformulated gasoline to meet requirements of both programs by producing only two separate product lines, one for summer and one for winter, rather than three separate product lines to be sold during three or four different enforcement periods. Moreover, defining high ozone season to be May 1 through September 15 would be in keeping with the basis for applying the volatility rule during the same period, since the primary purpose of that rule is also to reduce ozone-forming emissions. Under the second option, the high ozone season would be lengthened to include May 1 through September 30. EPA believes that this option may simplify record-keeping requirements, which are generally monthly, by being consistent with the monthly compliance periods being considered for reformulated gasoline in order to accommodate the emissions averaging program provided by section 211(k)(7) of the Act and described in this notice in section VIII. Under the third option, high ozone season would be defined uniquely for each ozone nonattainment area, based on the period during which ambient levels of ozone in that area exceed the ambient air quality standard. This period would be based on an analysis of past ozone data over a specified period (e.g., 1987-1989). The guiding principle might be, for example, to select the shortest period which encompasses 90 percent of all ozone violations occurring in each area within that area's high ozone season. VOC control would thus be focused on those periods during which each area is likely to suffer from high rates of ozone formation. The Agency welcomes information regarding the determination of area-specific high ozone seasons and encourages comments on all options for the definition of "high ozone season". 3. Representative 1990 Model Year Vehicles Under section 211(k) of the Act, reformulated gasoline must result in reduced emissions of VOCs and toxic compounds, relative to emissions from baseline or representative model year 1990 vehicles, when using the baseline gasoline. EPA proposes that representative model year l990 vehicles refers to all recent model year vehicles utilizing current vehicle technology. This could include 1986-1991 closed-loop vehicles with adaptive learning in order to take advantage of all available data on emissions from vehicles with technology comparable to that of actual model year 1990 vehicles. The use of all available data is critical to the development of a model for purposes of determining whether candidate fuels may be certified as reformulated. (See discussion of modeling below in section IV.) For certifying fuels using a test procedure rather than the model, EPA proposes that baseline vehicles be limited to vehicles from model years 1989 through 1991. Test results from each vehicle type will be weighted according to its 1990 model year sale fraction (described below in section V). 4. In-Use Basis When quantifying emissions from a vehicle or fleet of vehicles, it is important to realize that emissions will vary over a vehicle's lifetime, generally increasing with age due to factors including normal mechanical aging as well as possible malmaintenance or tampering. To ensure that the standards set for reformulated gasoline achieve the intended environmental benefits, the Agency proposes to assess emissions on an "in-use" basis, such that emissions from 1990 model year vehicles are the estimated average emissions from those vehicles over their lifetimes. Comments regarding the impacts of an in-use approach on determining representative vehicle emissions are requested. a. Use of EPA's Mobile4.1 Emissions Model. Traditionally, the Agency, along with state and local air pollution agencies, the auto industry, and other parties interested in estimating mobile source emissions have estimated in- use mobile source emissions using versions of EPA's Mobile emissions model. The model, which incorporates data from EPA's Emission Factor (EF) program involving testing of in-use vehicles, is capable of projecting motor vehicle emissions under a range of ambient conditions and a variety of regulatory schemes. The current version of the model, Mobile4.0 is currently being updated. This updated version, Mobile4.1, will be available for use in June. Because Mobile4.1 focuses on calendar year 1990 emissions and includes vehicle technology up to and including the 1990 model year, it will provide a more accurate estimate of in-use emissions from 1990 vehicles than previous versions of the model. Also, because Mobile4.1 will be used by states to estimate mobile source emissions for the purpose of developing their State Implementation Plans (SIPs), its use in determining the emissions baseline for reformulated gasoline will provide consistency between state and federal programs. While final Mobile4.1 emission estimates were not available in time to be included in this proposal, the baseline emission estimates described in this proposal are based on projections of a draft version of Mobile4.1. Projections from the final version will be available shortly and will be made public through a notice published in the Federal Register. EPA strongly believes that Mobile4.1 should be used as the source of baseline emission estimates for this rulemaking and requests comments on its use for this purpose. i. Temperature Conditions. EPA's Mobile emissions model has been developed to predict motor vehicle emissions on an area-specific basis. In order to use the Mobile model, it is necessary to specify a temperature range for that area in which motor vehicle emissions are being evaluated. Regarding the temperature conditions at which emissions from baseline, formula, and reformulated gasolines will be modeled, the Agency is proposing two options. Under the first option, EPA is considering modeling baseline emissions under different temperature conditions for gasoline sold in the cooler, more northerly areas classified as "Class C" areas under EPA's Phase II volatility regulations (40 CFR part 80, 55 FR 23659, June 11, 1990) than in warmer areas, classified as "Class B". This option is being proposed to account for the differences in these areas' ambient temperature conditions and the fact that levels of non-exhaust (evaporative, running loss, resting loss, and refueling) emissions and rates of ozone formation are associated with the temperature conditions in an area. Under the second option, EPA proposes modeling emissions for all areas under Class C conditions. This option is a simplification relative to the first option, but would generally be consistent with the Act's approach to defining the summer baseline gasoline, which represents a typical Class C area gasoline in that its Reid vapor pressure is 8.7 psi, although it is to be used as the baseline fuel for reformulated gasoline to be sold in all covered ozone nonattainment areas. Section III.D.3 of this notice lays out alternative proposals for the VOC emission standard applicable to reformulated gasoline sold in Class B covered areas. Under the first alternative, a more stringent standard would apply in Class B covered areas than would apply in Class C ozone covered areas. Under the second alternative, the same standard would apply to both Class B and Class C areas. The two options being presented here regarding temperature conditions are dependent, in a practical way, on those in section III.D.3. If a more stringent VOC emission standard is applied in Class B areas, then it would be appropriate to use different temperature conditions to model emissions in those areas. If the same VOC emission standard is applied in all areas, then the Phase II RVP standards will cause all fuels to meet the VOC standard, regardless of whether a Class B or Class C temperature range is used. Thus, for reasons of simplicity, the same temperature range could be used in both Class B and C areas. Comments are requested on the impacts of modeling baseline emissions under different temperature conditions for Class B and Class C nonattainment areas, as described above. ii. High Ozone Temperature Determination. For determining an appropriate range of temperatures at which to model high ozone period emissions for ozone nonattainment areas, two alternative proposals are presented here. Either option could be applied in conjunction with either of the temperature condition options described above. The first alternative would use the mean maximum and mean minimum daily temperatures at which the ten highest ozone exceedances occurred in each nonattainment area over the period from 1986 through 1989. These city- specific mean maximum and minimum temperatures would then be weighted by each areas' gasoline consumption level to derive a single temperature range for Class B and Class C ozone nonattainment areas (either one range for both Class B and Class C areas or else one range for each). These temperatures are being determined and will be placed in the docket for this rulemaking as soon as they are available. The second option would utilize two sets of diurnal temperature ranges which EPA has, in the past, used in regulatory analysis. For Class C and B areas, these diurnal temperature ranges are 72-96 and 74-106 degrees F, respectively and were based on the 90th percentile high temperatures from 1984 for all Class C cities combined and Class B cities combined. As such, they represent a somewhat more extreme set of high ozone conditions. The Agency encourages comments regarding appropriate temperature ranges for high ozone periods in Class B and Class C areas. b. Effects of Stage II Refueling Controls. EPA proposes that the conditions under which baseline vehicles emissions are modeled be representative of the conditions that will be encountered during the time when reformulated gasoline provisions will be in effect. During this period, all moderate, serious, severe, and extreme ozone nonattainment areas will be required to implement EPA's "Stage II" refueling controls as a means of limiting gasoline refueling emissions. Stage II controls require the use of emission control devices installed in service station pumps to recycle emissions that would otherwise be lost into the atmosphere. EPA's regulatory impact analysis supporting refueling emission regulations estimated the efficiency of Stage II equipment to be 86 percent in areas where the program is very strictly enforced and 62 percent where enforcement is minimal. The overall average efficiency for Stage II equipment in areas currently using Stage II controls has recently been estimated at 80 percent. EPA's Stage II implementation guidelines will allow exemptions for up to 25 percent of the nation's gasoline. (Exemptions are allowed for independent stations with throughput below 50,000 gallons per month and all stations with average throughput below 10,000 gallons per month.) The state of California, however, does not allow any exemptions from Stage II controls in their ozone nonattainment areas. Assuming that 25 percent of gasoline sold in nonattainment areas outside of California will be exempt, EPA projects that refueling controls will, on average, reduce refueling emissions by 66 percent in areas where the program is in effect. Comments are encouraged regarding this estimate of the in-use efficiency of Stage II refueling controls. Due to this program, refueling emissions will constitute a smaller fraction of total in-use VOC emissions at the time reformulated gasoline provisions go into effect than is currently the case. EPA thus believes that it would be appropriate to account for Stage II refueling impacts when assessing emissions attributable to reformulated gasolines. In order that emissions due to reformulated gasolines and to the baseline gasoline be assessed on a consistent basis, the Agency proposes also including the impacts of Stage II refueling control in estimates of baseline emissions. The Agency welcomes comments on the inclusion of Stage II refueling control effects in evaluations of emissions from the baseline gasoline and other gasolines. c. Effects of Enhanced Inspection/Maintenance Programs. A large portion of motor vehicle emissions are attributable to a small fraction of vehicles whose emission levels are extremely high due to tampering or malmaintenance. The enhanced inspection and maintenance (I/M) programs, which are mandated by the Act for all serious, severe, and extreme ozone nonattainment areas, will be addressing this category of emission sources by enforcing proper maintenance of exhaust and evaporative emission control equipment on motor vehicles. The Agency is in the process of developing the minimum criteria for enhanced I/M programs. The types of control measures being considered include anti-tampering inspections or canister purge and fuel tank pressure checks for evaporative emissions, and also an idle test or an "IM240" (a comprehensive emission test) for exhaust emissions. Mobile4.1 will be capable of modeling the emissions effects of enhanced I/M. Enhanced I/M, while not formally a part of Mobile4.1, has been discussed at all Mobile4.1 workshops. The criteria for the enhanced I/M program are expected to be defined and made available by July, 1991, at which time they will be placed in the public docket of this rulemaking. The Agency proposes to include the impacts of enhanced I/M programs on baseline emission projections since enhanced I/M programs will be in place when requirements for reformulated gasoline take effect. To bracket the potential impact of enhanced I/M, EPA has used two different I/M program scenarios in deriving estimates of toxic, VOC, and NOx emissions from baseline gasoline. These enhanced I/M scenarios represent two extremes with regard to the impact of enhanced I/M on the composition of reformulated gasolines. Under the first enhanced I/M scenario (referred to hereafter as the low evaporative or "low evap" case), the program is assumed to include canister purge and fuel tank pressure checks for evaporative and running loss emissions. For exhaust, it would require an IM240 test, with a stringent cut- point (the emission level above which a vehicle fails the test) for exhaust hydrocarbons. Under this scenario, 98 percent of the vehicle population would undergo inspections, and waivers would be granted for 2 percent of the inspected vehicles. Also, it is assumed that inspections would be performed on an annual basis. Under the second scenario, the "high evap" case, only an anti-tampering inspection would be required for evaporative and running loss emissions. Exhaust emissions would be assessed by an idle test, which is a simpler test than the IM240 test. Under this scenario, 95 percent of the vehicle population will undergo inspections, and waivers would be granted for 5 percent of inspected vehicles. As under the first scenario, inspections are assumed to be performed on an annual basis. The primary reason for including enhanced I/M benefits in the reformulated gasoline baseline is to focus the determination of reformulated gasoline's potential benefits on the sources of emissions that would remain after enhanced I/M programs are in place. The effect of including enhanced I/M would be to change the baseline ratio of exhaust to non-exhaust emissions, which could affect the types of fuel reformulations which are most cost effective. As will be seen in the next section, however, the definition of enhanced I/M is unlikely to significantly affect gasoline reformulation. This is true because the evaporative fraction is expected to continue to dominate overall emissions for both the high and low evap cases, and, consequently, volatility control will continue to be one of the major means of controlling total VOC emissions. Comments regarding the impacts of I/M programs on baseline emissions modeling are solicited. EPA also encourages interested parties to participate in the public process of developing enhanced I/M measures and to express their views regarding the impact of these programs on gasoline reformulations. 5. Projected Baseline VOC Emissions Assuming the above proposals for 1990 model year vehicles, in-use emissions and, Stage II refueling controls, and using Mobile4.0 and projected Mobile4.1 emission factors EPA estimates the Class C (with diurnal temperatures of 72- 96 degrees F) VOC baseline emissions as follows: Mobile4.1 Low High Model: Enhanced I/ evap evap M Scenario Mobile4.0 VOC VOC Exhaust (g/mi) 0.71 0.50 0.58 Hot Soak/Diurnal 0.25 0.25 0.81 Running Loss 0.53 0.22 0.53 Refueling 0.07 0.07 0.07 Total VOCs (g/mi) 1.56 1.04 1.99 For both baseline and standard-setting purposes, VOC emissions are proposed to include all oxygenated and non-oxygenated hydrocarbons and to exclude methane and ethane because of the low reactivity of these compounds. 6. Baseline Toxic Emissions As noted above, VOCs include all oxygenated and non-oxygenated hydrocarbons except for the simple compounds methane and ethane. All five of the toxic air pollutants whose emissions will be controlled through reformulated gasoline: Benzene, 1,3-butadiene, polycyclic organic matter (POM), formaldehyde, and acetaldehyde, also fall under the category of VOCs. Benzene, an aromatic compound, is a natural component of gasoline and, as such, is present in gasoline vapor emissions. Benzene is also formed from other aromatics during the combustion process and is emitted as exhaust. The four other toxic air pollutants subject to control by reformulated gasoline are solely products of combustion. Under high ozone conditions, all five toxics are present in exhaust emissions, and only benzene is present in evaporative, running loss and refueling emissions (non-exhaust emissions). EPA proposes to regulate aggregate toxics emissions based on exhaust and non-exhaust emissions during this period. Under non-high ozone or winter conditions, on the other hand, EPA believes that non-exhaust VOC emissions will be very small relative to exhaust VOC emissions, making non-exhaust benzene emissions very small as well. EFA therefore proposes to consider non-exhaust toxic emissions to be negligible outside of the high ozone season and to regulate aggregate toxics emissions based on total exhaust emissions during this period. EPA requests comments on this approach to toxics regulation. a. Proposed Exhaust Benzene. In the following section C, several correlations relating fuel parameters to emissions of toxic compounds are described. The first of these correlations, described further in section C.1, relates the benzene fraction of exhaust emissions to the levels of fuel benzene and aromatics. Under this correlation, the weight percent of benzene in exhaust non-methane/non-ethane emissions is equal to: 1.077 + 0.9441*(Bz) + 0.1133 x (Arom-Bz) where Bz is the volume percent of fuel benzene and Arom is the volume percent of fuel aromatics. EPA proposes that this correlation also be used to establish the level of benzene exhaust emissions from the baseline gasoline and welcomes comments. As the summer baseline gasoline contains 1.53 and 32 volume percent benzene and aromatics, respectively, the proposed benzene weight fraction of exhaust VOC emissions would be 0.0597. For the winter baseline gasoline described in section III.A.1 above, the analogous figure is 0.0542. b. Non-exhaust Benzene. i. Evaporative (Hot Soak and Diurnal) Benzene Emissions. Evaporative benzene emissions from a given vehicle are primarily a function of fuel benzene content, temperature, RVP and MTBE content. The MTBE effect is not a function of its oxygen content, but is a function of its chemical interaction with the other liquid compounds of the fuel. Data showing this effect is only available for MTBE at this time. Other ethers may show similar effects. Alcohols do not show this effect. Since hot soak emissions (evaporative emissions from a warm vehicle after it has been running) occur at higher temperatures that diurnal emissions (evaporative emissions from a sitting vehicle as the daily ambient temperatures rise and fall), the benzene fraction of hot-soak VOC emissions tends to be higher for a given fuel than that for diurnal VOC emissions. Evaporative benzene emissions also appear to be a strong function of the condition of the vehicle's evaporative emission control system, with the benzene fraction of evaporative VOC emissions being higher for those vehicles with properly operating systems and low VOC emissions (those likely to "pass" EPA's purge and pressure tests) and lower for those vehicles with inoperative systems and higher VOC emissions (those likely to "fail" EPA's purge and pressure tests). EPA assumes that the benzene fraction of both hot-soak and diurnal emissions for "pass" vehicles be based on the percentage of benzene in the fuel. This benzene fraction estimate is nearly equivalent to data submitted by ARCO (L.A. Rapp to R. Rykowski, May 15, 1991) and :somewhat lower than the Auto/Oil test results. The Auto/Oil test results show benzene to be 2.0 and 3.9 percent of hot-soak and diurnal VOC emissions for an industry average fuel which has the same benzene content and RVP as the summer CAA baseline fuel. For "fail" vehicles, EPA chooses to use the General Motors' tank vapor emissions model (which has been confirmed by both tank vapor data and similar models developed by CRC and EPA) for representative tank temperatures. For CAA baseline fuel, the benzene fractions of hot-soak and diurnal emissions are 1.143 and 1.033 percent, respectively. ii. Running Loss Benzene Emissions. Fractions of vaporous emissions of benzene from an operating vehicle (benzene running loss emissions) were also determined in the Auto/Oil program, although only two vehicles were tested for running loss emissions. These measurements varied widely, as may be expected for well-maintained vehicles like these with very low running losses. The great majority of in-use running loss emissions come from vehicles which fail EPA's purge or pressure tests. For the baseline levels of benzene running loss emissions, therefore, EPA proposes the use of General Motors' tank vapor emissions model described above for a representative tank temperature rise for both "pass" and "fail" vehicles. The result of using that model for running losses is a benzene fraction of running loss VOC emissions of 1.140. EPA requests comments and any additional data which may be available concerning the use of these figures for baseline fuel toxic emissions. iii. Refueling Benzene Emissions. Projections relating fuel benzene levels to the weight percent of benzene in refueling emissions have been estimated using EPA, American Petroleum Institute (API), and General Motors (GM) tank vapor emissions models. These three models project very similar values and have been shown to be consistent with available data. The GM model is the most sophisticated scientifically, so EPA proposes its use here. Using this model, EPA finds the benzene fraction of refueling VOC emissions for the baseline gasoline is 1.057 percent. EPA requests comments and any additional data which may be available concerning the use of this figure for baseline fuel toxic emissions. c. Formaldehyde, Acetaldehyde, and 1,3-Butadiene Emissions. The Auto/Oil test results on formaldehyde, acetaldehyde and 1,3-butadiene emissions due to the industry-average fuel can be applied directly to the reformulated gasoline baseline summer gasoline, since, as discussed above, the industry- average fuel fits the specifications of summer baseline gasoline. From that data EPA has calculated that the exhaust fractions of 1,3-butadiene, formaldehyde, and acetaldehyde are, respectively, 0.56, 1.25, and 0.89 weight percent of exhaust VOC emissions. EPA proposes the use of these fractions, applied to levels of summer baseline exhaust VOC emissions, to determine levels of summer baseline aldehydes and 1,3-butadiene. For the winter baseline fuel, EPA proposes to use the correlations proposed in section C below. These correlations, also based on Auto/Oil test results, quantify the effects of fuel aromatics, MTBE, olefins, and T90 on each toxic's fraction of VOC exhaust emissions. Given the specification of winter baseline fuel described above, the correlations project that 1,3-butadiene, formaldehyde and acetaldehyde represent 0.64, 1.39, and 0.98 weight percent of exhaust VOC emissions, respectively. Future Auto/Oil data, as well as other data, will be considered in estimating the effects of other oxygenates. d. Baseline Toxics Emission Projection. Based on the toxic pollutant fractions and correlations proposed above in section III.A.6, and using the VOC emission breakdowns derived above in section III.A.5, the following table lists EPA's estimated toxic emissions (mg per mile) from baseline vehicles when using summer baseline gasolines as they vary with the different options regarding use of EPA's Mobile model, low or high evap scenarios for enhanced I/M programs, and assuming Class C area temperatures. Mobile4.1 Model: Enhanced I/M Low High scenario Mobile4.0 evap evap Exhaust benzene (mg/mi) 42.40 29.87 34.65 Evaporative benzene 2.86 2.76 8.97 Running loss benzene 6.05 2.51 6.05 Refueling benzene 0.74 0.74 0.74 1,3-butadiene 2.77 2.78 3.22 Formaldehyde 5.75 6.28 7.28 Acetaldehyde 4.05 4.46 5.17 Polycyclic organics 1.40 1.40 1.40 Total toxics (mg/mi) 65.93 50.79 67.49 For winter baseline gasoline, toxic emissions are estimated to be as follows: Mobile4.1 Model: Enhanced I/M Low High scenario Mobile4.0 evap evap Exhaust benzene 51.06 35.96 41.73 1,3-butadiene 6.06 4.27 4.95 Formaldehyde 13.12 9.24 10.72 Acetaldehyde 9.24 6.51 7.56 Polycyclic organics 1.40 1.40 1.40 Total toxics (mg/mi) 81.48 57.38 66.36 B. Impacts of Fuel Parameters on NOx and VOC Emissions Having developed emission estimates for the baseline fuel, it is now necessary to determine how changes in the baseline fuel parameters will affect emissions, in order to evaluate the performance of the formula fuel, the performance of candidate reformulated gasolines, and the achievability of greater emissions. EPA presents two alternative proposals for predicting the emission impacts of changing fuel parameters by applying existing data to an emissions model. Under the first option, a simple model would be developed which includes only the effects of fuel benzene and aromatics levels on benzene emissions, RVP on non-exhaust VOC emissions, and oxygen and aromatics on exhaust VOC emissions, and oxygenates on emissions of aldehydes. A large amount of data is available on these fuel parameters and emission impacts, so that a model can be developed which predicts these effects of these parameters with a high degree of certainty. They are also the most likely fuel parameters to be changed in gasoline reformulations. Under the second option, a more comprehensive, more complex model would be developed that includes additional parameters, such as fuel sulfur and olefin levels and distillation points, whose emissions effects are less well known. The advantage of a more comprehensive model is that it would provide fuel producers with more ways of reformulating gasoline that could still be certified using a model instead of through more costly testing. On the other hand, adding fuel parameters could impose an added burden on some refiners, since any modeling of improvements in fuel characteristics to show reduced emissions would have to be accompanied by the modeling of increased emissions due to a worsening of those same fuel characteristics. Because roughly half of all gasolines have characteristics that are above an average value while the other half are below, for each fuel parameter roughly half of all refiners would need to make improvements simply to match the characteristics of the baseline gasoline. In addition, with the more comprehensive model, there is less certainty whether specific changes to the parameters actually produce the projected effect and less confidence that the emission reduction goals of the Act are being met. This option would also require a more complex and demanding enforcement mechanism. For use under either option, there is a variety of data available from different sources regarding the emission effects of changes in fuel parameters. Recently published results from the Auto/Oil research study constitute much of that data, and, in particular, provide the bulk of additional data that would be used in a more comprehensive model under the second option. The Auto/Oil tests used, as their baseline, a gasoline that meets the specifications of the CAA baseline summer gasoline, and because most of the Auto/Oil tests were performed on 1990 model year vehicles or vehicles with comparable relevant technology, the data resulting from these tests is relevant for potential inclusion in EPA's certification emissions model for baseline, formula, and reformulated gasolines. On the other hand, the Auto/ Oil test program used only ten vehicles, all low emitters, and thus would not fit the requirements of EPA's testing protocol described below in section V. Another important data source is EPA's Emission Factor (EF) database, which contains a substantial amount of data on oxygen and emissions. Other data is also available and will be described, along with Auto/Oil test results and EPA EF data, in the following paragraphs. While most test programs have been performed on well-maintained vehicles, the actual condition of vehicles on the road could have a strong influence on the in-use emission effects that will result from changes in fuel composition. Exhaust VOC emissions effects, particularly those which influence the effectiveness of the vehicle's catalyst, will likely vary between vehicles with low, high, and super high (high-high) exhaust emissions (low, high and super high emitting vehicles are described further in Section V of this notice). VOC evaporative and running losses, subject to the condition of the canister, are most likely to be accurately estimated by using different fuel factors for "pass" vehicles (those capable of passing evaporative emission tests) or "fail" vehicles, and weighing these factors according to the fleet percentage of these vehicle types. Because Mobile4.1 considers the above-mentioned vehicle emitter classes separately based on their emission rates, EPA proposes using, to the extent that they are available, separate exhaust, evaporative, and running loss emission effects for each emitter subclass. These effects will be weighted by vehicle type and incorporated into a single model projecting the effect of fuel parameters on total VOC emissions. The procedure for incorporating emission effects into a model will be consistent with the assumptions made concerning Stage II and enhanced I/M programs in estimating baseline emissions. 1. NOx and VOC Exhaust Emissions The Auto/Oil test results suggest both NOx and VOC emissions to be affected by a fuel's oxygen, sulfur, aromatics, and olefin levels and by its T90 percent distillation point (T90). NOx emissions may be reduced through higher aromatics, lower olefins, lower oxygen, lower sulfur, and higher T90, while VOC emissions may be reduced through lower aromatics, higher olefins, higher oxygen, lower sulfur, and lower T90. Some of these oxygen and aromatics effects are corroborated by other test data and could be applied to a simple model, while other correlations go beyond the RVP, oxygen, aromatics, and benzene effects proposed for a simple model, but could be included under the second option of a comprehensive emissions model as described above. Based on the Auto/Oil mean emission results, the correlations between these fuel parameters and VOC and NOx emissions can be characterized as follows: VOC exhaust (g/mi)=Baseline VOC exhaust+[1-(0.0060)x(32+Arom)]+[1- (0.010)+(Oxygen)]+[1-(0.0038)+(330-T90)]+[1-(0.00042)+(339-Sulfur)]; NOx exhaust (g/mi)=Baseline NOx emissions+[1+(0.00047)+(32+Arom)]+[1+(0.0088)+(Oxygen)]+[1-(0.0049)x(9.2- Olefins)]+[1+ (0.00047)x(330-T90)]+[1-(0.00030)+(339-Sulfur)]. Arom, Oxygen, Olefins, T90, and Sulfur refer, respectively, to the fuel volume percent aromatics, weight percent oxygen, volume percent olefins, T90 in degrees F, and sulfur content in parts per million. The above characterization will be explained further in the Regulatory Impact Analysis supporting this rulemaking which, as noted below, will be available at a later date. In addition to data from the Auto/Oil study, a substantial amount of data on VOC and NOx emissions at different oxygen levels is available as part of EPA's Emission Factor (EF) database. There is a great deal of data available which relates fuel oxygen to exhaust emissions. Still, not many fuel oxygenate levels have actually been tested. Most emission measurements were performed with MTBE at 2.7 weight percent oxygen or with ethanol at 3.5 weight percent oxygen, so some questions remain regarding whether emissions are a function of the oxygen level (weight percent) regardless of oxygenate type or of the oxygenate type and level (volume percent). There is also some speculation regarding the shape of the "curve" describing the relationship between either oxygen or oxygenate level and emissions. The Agency requests comments regarding the quantification of NOx and VOC emission effects due to fuel oxygen or oxygenate levels. The Agency requests comments on whether different oxygenates should be assessed differently for their effects on VOC and NOx emissions. Comments are also requested regarding the treatment, in a modeling approach, of oxygenates which were not included in EPA's EF or in Auto/Oil testing programs. Olefins, the familiar name for double-bonded hydrocarbon molecules, are among the most reactive compounds that are emitted by motor vehicles, combining very quickly with NOx in the presence of sunlight to form ozone. As projected by Auto/Oil data, decreasing a fuel's olefin content will result in both decreased olefin emissions and decreased NOx emissions, contributing thereby to reduced ozone formation. (While disagreements exist as to the precise ratio of olefin reactivity to typical VOC reactivity, all estimates show olefins to be substantially more reactive.) However, Auto/Oil data also indicates that the mass of exhaust VOC emissions is actually increased by decreasing levels of fuel olefins. Inclusion of this relationship in the VOC emission model would encourage high fuel olefin levels that could cause more, rather than less, ozone formation than the baseline fuel causes, or that would reduce ozone formation less than reformulated gasoline with low olefin levels does. EPA is considering including the emissions effect of olefins in the comprehensive option for a NOx emission model, but excluding it from the exhaust VOC emission model in order to avoid this inappropriate incentive. The Agency welcomes comments on this matter. Aside from such fuel parameters as olefins, whose inclusion in a certification model for VOC emissions would result in undesirable environmental impacts, EPA believes that, under the option of a comprehensive emissions model as described at the beginning of this section, all fuel parameters whose emissions effects can be reliably substantiated in time to be included in the reformulated gasoline rulemaking should be included in the derivation of the certification emissions model. In order to be reliably substantiated, EPA would have to be confident that the effect of a parameter was known for both low and high emitting vehicle types, or, if known for one type, that the results could be extrapolated for the other. EPA would also need to be confident that the emission effect was induced due to the specified parameter, and that the effect was independent of other fuel parameters or that its interaction with other parameters was known. The Agency would consider using current and future Auto/Oil data to model the effects of aromatics, olefins, oxygen, sulfur, and T90 on exhaust VOC and NOx emissions, and is also including EPA's EF data, along with Auto/Oil data, to model the effects of aromatics and oxygen. Under the first option of a simple model, the effects of olefins, sulfur, and T90 would be excluded, leaving only the effects of oxygen and aromatics. EPA is very interested in receiving comments on the impacts of fuel parameters on NOx and VOC exhaust emissions, and is especially interested in comments on the inclusion of these impacts in its certification emissions model. As the Agency finds that the fuel effects of additional parameters can be reliably substantiated, it will issue notices thereof. In order to provide some indication of the types of reformulated gasolines which could be certified under the modeling approach being proposed today the proposed regulations assume adoption of the second, more comprehensive option for the model (i.e., including all of the correlations (except for the effect of olefins on exhaust VOCs) that were developed from the Auto/Oil test results for low-emitting vehicles and applying them to all vehicles). 2. Evaporative, Running Loss and Refueling VOC Emissions The effects of fuel volatility (RVP) on evaporative, running loss, and refueling emissions are well characterized for Class C area summertime conditions within a volatility range of 7.8 to 11.5 psi and for Class B summer conditions between 7.0 and 10.5 psi. EPA proposes using its Mobile4.1 emissions model (which will address the effect of RVP under such conditions) to evaluate evaporative, running loss, and refueling emissions due to changes in gasoline RVP. As Mobile4.1 is not yet available, all projections made below utilize the effect of RVP on non-exhaust VOC emissions as projected by Mobile4.0. The Agency welcomes comments and information regarding the effects of volatility and other fuel parameters on nonexhaust emissions. C. Impacts of Fuel Parameters on Toxic Emissions As in the above section B on the Impacts of Fuel Parameters on NOx and VOC Emissions, the Agency is considering two options for predicting the impacts of fuel parameters on toxic emissions which are analogous to those for exhaust VOC emissions. The first is a simple model that includes the effects of fuel benzene, aromatics, oxygen, and RVP on toxic emissions. The second is of a more comprehensive model that also accounts for the impacts of other fuel parameters like sulfur content, olefins, and distillation points on toxics. 1. Exhaust Benzene Emissions Exhaust benzene emissions can be affected by fuel modifications in two basic ways. Some fuel effects will change the fraction of benzene in the exhaust, regardless of the total VOC mass that is emitted as exhaust. For instance, increasing or decreasing the levels of benzene in a fuel will lead to a direct increase or decrease in the benzene fraction of exhaust emissions. Moreover, changes in the level of benzene precursors (primarily nonbenzene aromatics) will affect the amount of benzene that is produced during combustion, also changing the benzene fraction of exhaust VOC emissions. Second, fuel modifications can affect the overall level of exhaust VOC emissions by affecting the efficiency of the engine in burning hydrocarbons or by affecting catalyst efficiency. In these cases, the benzene fraction of exhaust VOC emissions may stay relatively constant and benzene exhaust emissions will change proportionally with exhaust VOC emissions. Of course, some fuel modifications can produce a combination of these two effects. EPA proposes to analyze the effect of fuel modifications on exhaust toxic emissions by separating the two types of effects described above. This applies, not only to benzene, but to all five toxic air pollutants. With this approach, fuel modifications which change the level of VOC exhaust emissions can be considered to change the levels of toxic exhaust emissions proportionally. With respect to the effects of fuel modifications on the benzene fraction of exhaust VOC emissions, both fuel benzene and fuel aromatics appear to be the primary factors. A recent correlation developed by Chevron used the results of three studies (described in the Regulatory Impact Analysis) to relate fuel benzene and aromatics to exhaust benzene and characterizes the weight percent of benzene in exhaust VOC (nonmethane/nonethane) emissions as equal to: 1.077+0.9441x(Bz)+0.1133x(Arom-Bz) where Bz is the volume percent of fuel benzene and Arom is the volume percent of fuel aromatics. Because a sizeable amount of data went into these three studies, and because their results were all very similar, EPA proposes the use of this correlation for both summer and winter fuels and welcomes new information or suggestions regarding the effects of varying fuel parameters on exhaust benzene emissions. 2. Nonexhaust Benzene Emissions Benzene is the only toxic air pollutant that is emitted in measurable quantities from evaporative, running loss and refueling vapors. Reductions in fuel benzene may be expected to result in reductions in benzene emissions from all of these nonexhaust emission sources. The Agency proposes including this proportional effect of fuel benzene in nonexhaust benzene emissions in the emissions model. In addition to fuel benzene content, two other fuel parameters--RVP and MTBE, can also reduce nonexhaust benzene emissions. Reducing RVP reduces evaporative and running loss VOC emissions, since lower vapor pressure leads to lower emission levels of all pollutants, even if the fuel benzene level and the benzene vapor pressure remain constant. However, this effect is not proportional, meaning that benzene emissions will decrease less than one percent for every one percent decrease in VOC emissions. The effect of MTBE is more unusual. The presence of MTBE appears to depress benzene vapor pressure despite no change in fuel benzene content. This effect has been confirmed both by tank vapor data and the GM tank vapor model referred to above. EPA is considering alternative proposals for modeling the effects of RVP and MTBE on evaporative and running loss benzene emissions. Under the first option, EPA would utilize GM's tank vapor model to predict the effects of RVP and MTBE on tank benzene vapor emissions from both "pass" and "fail" vehicles. This model predicts that the benzene weight percent of hot soak VOC emissions for a fuel is described by the following relationship: Bzx(1.4448-0.080274(RVP) -(0.0342)x(OX)) where Bz is the volume percent benzene, RVP is in psi and Ox is the weight fraction oxygen in the form of MTBE. Similarly, the benzene weight percent of diurnal VOC emissions is: Bzx(1.3758-0.080274(RVP) -(0.0289)x(Ox)). Since additional data on the effects of RVP and MTBE on nonexhaust benzene emissions from both "pass" and "fail" vehicles will become available through future Auto/Oil testing, the second option would include the results of this future data to model nonexhaust benzene emissions in the certification model. The Agency welcomes comments and data on the relationships between fuel volatility and oxygenate on benzene emissions and between overall nonexhaust VOC emissions and benzene emissions. 3. Nonbenzene Toxic Emissions a. 1,3 Butadiene, Formaldehyde, and Acetaldehyde Emissions Effects. Available data vary regarding the effects of fuel parameters on 1,3- butadiene, formaldehyde, and acetaldehyde emissions. Under the first option of a simple, well-substantiated model, the Agency proposes to include the effects of fuel oxygen content on acetaldehyde and formaldehyde, both of which are oxygen-containing toxic air pollutants, since a substantial amount of collaborative data exists regarding these relationships, while the effects of other fuel parameters on formaldehyde and acetaldehyde emissions and the effects of oxygen on 1,3-butadiene display a range of varying and unpredictable results. Under the second comprehensive model option, the Agency proposes including the results of the Auto/Oil test program showing the effects of fuel aromatics, olefins, T90, and MTBE on these three toxics, as they have been included for other VOC and toxic emissions. Just as lower emissions of total exhaust VOCs will result in lower exhaust benzene emissions, so will they result in lower exhaust emissions of 1,3- butadiene, formaldehyde, and acetaldehyde. EPA proposes that other fuel modifications which reduce VOC exhaust emissions, such as higher oxygen levels or, under the second option, lower sulfur content, be assumed to reduce all toxic exhaust emissions proportionately. Additional Auto/Oil data will be available soon which quantifies the effects of ETBE and ethanol on 1,3-butadiene, formaldehyde, and acetaldehyde. EPA proposes to use this and all other available data to determine the effect of fuel modifications on these three toxic emissions, and would consider applying the results of this determination in the certification model, under the comprehensive model option, just as previously available Auto/Oil results have been proposed for that option. EPA welcomes comments on this approach and requests that any data showing the effect of fuel parameters that effect emissions of 1,3-butadiene, formaldehyde, or acetaldehyde be submitted. b. POM Emissions. Exhaust polycyclic organics (POMs) include a number of different toxic compounds, mostly high molecular weight aromatics. There is no data quantifying the impacts of gasoline reformulations on POM emissions. At the present time, there are no widely accepted test procedures for measuring POM in both the gaseous and particulate phases. In addition, they constitute a very small fraction of total toxic emissions (less than 2 percent). For these reasons, the Agency proposes that the emissions model contain no effects on POM emissions. Comments are welcome on this proposal. EPA also encourages comments on whether it would be more technically correct to treat POMs as a constant percentage of exhaust VOC emissions or as a constant value (in mg per mile). D. Proposed Emission Standards The final step in this analysis is to determine the appropriate levels of the VOC and toxic emissions standards that gasoline must meet in order to be certified as reformulated. As described earlier, section 211(k)(1) provides that gasoline is to be reformulated to yield the greatest achievable reductions in VOCs and toxic emissions, considering cost, energy, health and environmental impacts. Section 211(k)(3) provides that reformulated gasoline be required to comply with VOC and toxic emission standards determined by comparing the emissions performance of a specified formula fuel with specified performance standards. If the formula fuel achieves emissions reductions greater than the performance standards require, then the reductions achieved by the formula fuel become the standard; otherwise the performance standards apply. This determination is to be done separately for VOCs and toxics, so that the formula fuel may determine the standard for one of the pollutants and the performance standard may apply to the other. A fundamental issue of statutory construction is raised by the section 211(k) provisions regarding VOC and toxic emission standards. On the one hand, section 211(k)(1) calls for standards that require the greatest achievable reductions, considering specified factors. On the otherhand, section 211(k)(3) specifies VOC and toxic emission reductions that reformulated gasoline must achieve. Nothing in the language of section 211 or the Act addresses how these two provisions interrelate. The legislative history of section 211(k) does not definitively address the issue, either. One rule of statutory construction is that the specific governs the general. Application of that rule here would appear to mean that section 212(k)(3) determines the applicable reduction requirements. However, another rule of statutory construction is that every word of a statute is to be given effect; put another way, a statute should not be interpreted in a way that renders a word or provision superfluous. Applying this rule would seem to mean that section 211(k)(3) does not by itself determine the VOC and toxics standards, because such an interpretation would render the provision of section 211(k)(1), for standards requiring the greatest achievable reduction, meaningless. A corollary of this rule is that statutory provisions are to read together in a way that gives all of them effect. A possible reading of section 21l(k)(1) together with section 211(k)(3) is that the latter provision establishes minimum reduction requirements that reformulated gasoline must meet, while the former authorizes EPA to set more stringent standards if it finds such standards achievable in light of the specified factors. Another rule of construction that must be considered is that statutory provisions are to be interpreted in a manner consistent with Congress' purpose and policy in enacting the provisions. In the section below, EPA considers two alternative interpretations of the statutory provisions with regard to the VOC and toxic emissions standards: (1) Section 211(k)(3) establishes minimum standards that may be tightened pursuant to section 211(k)(1); or (2) section 211(k)(3) governs what standards apply. In any event, the first step in determining the required emission reductions is to determine the emission performance of the formula fuel relative to baseline emissions. For that determination to be made, the rest of the formula fuel's formula must be considered, since that statute only specified some of the fuel's parameters. 1. Definition of Formula Fuel As described above in section II.F, the formula fuel is defined in the Act as containing: --No more than 1.0 volume percent benzene, --No more than 25 volume percent aromatics, --At least 2.0 weight percent oxygen, --No lead, and --Additives to prevent the accumulation of deposits in engines or vehicle fuel supply systems. The Act is silent regarding other parameters that have been defined for baseline gasoline--sulfur, Reid Vapor Pressure (RVP), octane (R+M/2), distillation points, API gravity, olefins, and saturates. EPA proposes that, where a compositional characteristic is not specified for the formula fuel, its characteristic will be that specified for the baseline gasoline, in order that any emission reductions result from only the stated parameters, free of the many and varying influences that other gasoline properties may have on emissions. The Agency welcomes comments on specifications for the undefined components of the formula fuel. 2. VOC Emission Performance of Formula Fuel Relative to Baseline Gasoline: Summer Based on the impacts of fuel parameters on NOx, VOC, and toxic emissions discussed above in sections B and C, EPA estimates that exhaust VOC emissions are reduced between 2 and 11 percent from gasoline containing 2.0 weight percent oxygen in the form of MTBE (versus zero oxygen). The lower end of the range results from using Auto/Oil-determined MTBE effects and the upper end from using EPA-determined MTBE effects. Lower fuel benzene will only affect VOC exhaust emissions through its effect on total fuel aromatics, which is defined as 25 volume percent for the formula fuel. Lowering total fuel aromatics from 32 to 25 volume percent. will reduce exhaust VOC emissions slightly, based on Auto/oil test results. With regard to evaporative, running loss, and refueling VOC emissions, with constant fuel RVP, changes in benzene, oxygen, and aromatics content are not expected to have any effect. Thus, with a maximum reduction in exhaust VOC emissions of 2 to 11 percent and no change in the other VOC emissions, the reduction in total VOC emissions from the formula fuel relative to baseline gasoline is less than the 15 percent standard to which the Act defaults. Accordingly, the minimum VOC emissions performance standard for reformulated gasoline is a 15 percent reduction relative to baseline gasoline. 3. Class B VOC Emission Reduction Standard The volatility of gasoline, as measured by its Reid Vapor Pressure (RVP), is a major factor in the level of diurnal/hot soak, running loss, and refueling vapors emitted from vehicles. The warmer the temperature, the greater the contribution of RVP to high levels of these VOC emissions. EPA's Phase II volatility regulations (40 CFR part 80, 55 FR 23659, June 11, 1990) address the impact of gasoline volatility on ozone-forming emissions by limiting RVP to no greater than 9.0 psi in gasoline sold nationwide from May 1 through September 15. In addition, the RVP of gasoline sold in ozone nonattainment areas in those 23 southern states having the highest evaporative VOC emission levels (EPA Class A and B areas) may not exceed 7.8 psi from June 1 through September 15. These Phase II regulations will take effect May 1, 1992. Of the mandated nine areas in which only reformulated gasoline may be sold, Los Angeles, Houston, San Diego, and Baltimore are Class B areas and represent 45 percent of the total population in the nine areas. The extent to which reformulated gasoline will be an effective ozone control measure in these four covered areas will depend largely on the VOC emission reduction standard applicable to reformulated gasoline sold in these areas. As determined above in section D.2, the more stringent of the two alternative requirements laid out in section 211(k)(3) of the Act for VOC control is the minimum 15 percent reduction relative to emissions attributable to the baseline fuel. While the summertime baseline fuel has a specified RVP of 8.7 psi, gasoline sold in Class B ozone nonattainment areas during the summer will have a maximum RVP of 7.8 psi beginning in 1992. Reducing RVP from 8.7 to 7.8 is expected to result, with no other changes being made to the gasoline, in a 15 percent reduction in VOC emissions. Consequently, if reformulated gasoline achieves no more than a 15 percent VOC reduction relative to baseline fuel, it will achieve no more VOC reductions in Class B areas than the conventional gasoline that had already been sold in those areas for over two years. Today's notice lays out two alternative approaches to VOC standards, which incorporate four options. The first approach is premised on a reading of the statute as providing no more than a 15 percent emission reduction over the baseline gasoline. Under this approach, the Agency proposes an option that would require a 15 percent reduction in VOC emissions relative to baseline emissions during the high ozone season in all covered areas. The second approach is premised on the view that it would be inconsistent with the scope and intent of the reformulated gasoline program to apply a VOC standard that requires no more than a 15 percent reduction over baseline fuel. That the principal purpose of the reformulated gasoline program is to reduce ozone is clear from its scope. The program applies in the nine worst ozone nonattainment areas in the country and, at states' option, all other ozone nonattainment areas as well. It is not available to areas not in ozone nonattainment. According to this view section 211(k) authorizes a VOC reduction standard that achieves additional ozone reduction benefits for southern ozone nonattainment areas, including four of the nine worst ozone nonattainment areas. To ensure environmental benefits in all covered areas, the second approach would require that reformulated gasoline sold in Class C covered areas achieve VOC reductions of 15 percent over baseline fuel and that reformulated gasoline sold in Class B covered areas achieve more than 15 percent reduction in VOC emissions relative to baseline fuel. Specifically, the Agency proposes that, under one option, the VOC standard for gasoline sold in Class B covered areas would be a 30 percent reduction over baseline fuel, or equivalent to a 15 percent reduction beyond the VOC emissions benefits already achieved by Phase II volatility control. A second set of options would require Class B areas to use the Clean Air Act performance standards for the year 2000 for VOC emissions control. For 2000, the Act specifies a 25 percent reduction standard, which could be adjusted down to a minimum of 20 percent (section II.C), relative to the baseline emissions. EPA projects that an RVP reduction down to 7.1-7.2 psi, along with the formula fuel specifications, will achieve a 30 percent reduction in VOC emissions, relative to the baseline gasoline. EPA estimates that it is both technically feasible and cost effective to achieve a more stringent standard for VOC emission control in Class B areas and would have beneficial effects on health and the environment. (Cost and cost effectiveness estimates are still being developed. When these are completed, EPA will place them in the Docket, announce them in the Federal Register, and make them available, by mail, to parties interested in receiving them.) Because a 15 percent standard does achieve a substantial reduction in tons of VOCs in Class C covered areas, EPA would not consider a more stringent VOC standard to be appropriate for Class C areas. EPA requests comments on all options for VOC emission standards in Class B and Class C areas. Under the approach requiring more stringent VOC standards for Class B areas, there are three possible sub-options regarding the period over which the Class B VOC standard would apply. Under the first sub-option, and for ease of enforcement, a Class B VOC emission standard would be in effect over the entire high ozone season, as it is defined for reformulated gasoline. Under the second sub-option, and in order to be consistent with Phase II volatility control, the Class B VOC emission standard would apply from June 1 through September 30, when the Phase II 7.8 psi requirements apply, while a 15 percent standard would apply during the remainder of the high ozone season. While Phase II volatility regulations require the strictest levels of RVP control from June through September in all Class B areas, analyses supporting the volatility rule show that, in order to achieve emissions comparable to those of the average Phase I Class C areas using 9.0 RVP gasoline in July, nonattainment areas in many Class B states would actually require RVP controlled below 9.0 during only two or three months. Among the "nine cities", Baltimore is an example where sub-9 RVP control is only required in July. A third sub-option would be to apply the Class B VOC standard, on an area-specific basis, during only those months when RVP levels below 9.0 were found necessary in order to make emissions levels in those Class B nonattainment areas comparable with emission levels in Class C, 9 RVP, areas. A 15 percent VOC reduction would be required during all other months that fall within the high ozone season. The Agency encourages comments regarding the implications of these control season sub-options under the approach requiring a stringent Class B VOC emission standard. 4. Toxic Emission Reductions Due to the Formula Fuel, Relative to Baseline Gasoline: Summer As with ozone season VOCs, section 211(k)(3) sets, as the minimum standard for toxic emissions, a reduction of 15 percent over baseline fuel, or that which is achieved by the formula fuel, whichever is more stringent. The effect of the formula fuel on toxic emissions is more dramatic than its effect on VOCs. Its lower benzene and aromatic contents dramatically reduce both exhaust and nonexhaust benzene emissions. If MTBE is used as an oxygenate, the presence of 11 volume percent MTBE (the quantity needed to meet oxygen level requirements of 2.0 weight percent) has been shown to reduce nonexhaust benzene emissions, as described above in section 3. (For this and following emission estimates, the effects of oxygen have been modeled using MTBE, since Auto/Oil test data for other oxygenates has not yet been released) Based on the emissions effects proposed above, EPA estimates that a reduction in fuel benzene and aromatics to 1.0 and 25 volume percent, respectively, and the addition of 2.0 weight percent oxygen (via MTBE) to gasoline will reduce overall toxic emissions by 15.5-18.8% percent in the summer using a draft version of Mobile4.1 with Stage II refueling controls and a range of potential definitions of enhanced I/M (which have already been described). In the winter, the formula fuel is projected to reduce toxics emissions by 5.5 percent. While these estimates are being proposed as the toxic emission reductions achieved by the formula fuel, the values will likely change somewhat once Mobile4.1 has been finalized in June 1991 and enhanced I/M defined in June or July 1991. The projected effectiveness of the formula fuel in reducing toxic emissions may also be affected by new data received in response to this proposal. If new information warrants a change in the proposed toxic emission effects of the formula fuel, EPA will publish a Federal Register notice announcing the revision and will provide an opportunity for public comment. The Agency does anticipate, in any event, that the toxic emission reduction attributable to the formula fuel will exceed 15 percent, relative to the summer baseline gasoline, and that it will thus set the minimum standard for toxic emission reductions. The Agency has considered whether to apply a more stringent standard than the minimum standard for toxic emission control, as it has considered such an option for VOC control. Preliminary estimates indicate that the measures that would be required of refiners in order to reduce toxic emissions beyond the minimum requirements would be costly and not of great benefit. (The Regulatory Impact Analysis supporting this rulemaking will discuss in greater detail the environmental and economic impacts of reformulated gasoline.) EPA therefore proposes that, for toxics, the required reduction be based on the performance of the formula fuel, as described above. The Agency requests comments on the summer toxic emission standard. 5. NOx Emissions Attributable to the Formula Fuel Relative to the Baseline Summer Gasoline As explained earlier, section 211(k)(2) of the Act requires that NOx emissions from reformulated gasoline be no greater than the level of those emissions from baseline gasoline, but allows EPA to adjust or waive other requirements as needed to prevent the NOx cap from being exceeded. Since the formula fuel determines the toxics reduction requirement, it becomes relevant to determine whether the formula fuel would increase NOx emissions and, if the formula fuel does increase NOx emissions, whether adequate NOx control is technically and affordably achievable by adjusting other fuel parameters, without having to adjust any requirements of this program. An application of the correlation between various fuel parameters and NOx emissions proposed above in section IV.B.1 projects that the oxygen and aromatics content of the formula fuel will together result in an increase in NOx emissions of 2.1%, relative to NOx emissions from the baseline gasoline, assuming, as described above, MTBE as the oxygenate. The proposed correlation further projects that either a reduction in fuel olefins from 9.2 to 5.0 volume percent or a reduction in the sulfur level of gasoline from 339 to 270 will counteract the NOx emission increase due to the formula fuel without increasing either VOC or toxic emissions. Thus, the fact that the formula fuel itself increases NOx emissions does not necessitate that other requirements be changed to permit compliance with the NOx cap. 6. Formula Fuel Emission Projection Using the comprehensive option for a fuel certification emissions model, projections for Mobile4.1, Class C temperatures, and the emission effects proposed for that option in sections B and C above, the following table lists EPA's estimated VOC, NOx, and toxics emissions from baseline summer gasoline under the two enhanced I/M scenarios and from the formula fuel. (As described above in Section 3, oxygen effects have been modeled using MTBE as the oxygenate.) Baseline Formula Baseline Formula low evap w/MTBE hi evap w/MTBE Exhaust (g/m) .50 .47 .58 .54 Hot Soak/Diurnal .25 .25 .81 .81 Running Loss .22 .22 .53 .53 Refueling .07 .07 .07 .07 Total VOCs (g/m) 1.04 1.01 1.99 1.95 NOx (g/m) .930 .949 .930 .949 Exh Benzene (mg/m) 29.87 22.25 34.65 25.81 Evap Benzene 2.76 1.64 8.97 5.34 Running Loss Benz. 2.51 1.49 6.05 3.59 Refueling Benzene .74 .44 .74 .44 1,3-Butadiene 2.78 2.80 3.22 3.25 Formaldehyde 6.28 8.19 7.28 9.50 Acetaldehyde 4.46 4.71 5.17 5.47 POMs 1.40 1.40 1.40 1.40 Total TAPs (mg/m) 50.79 42.93 67.49 54.81 7. Toxic Emission Reductions Due to the Formula Fuel, Relative to Baseline Gasoline: Winter As explained above, the toxic emissions performance of the formula fuel relative to the summer baseline gasoline is higher than the mandated minimum 15 percent reduction. However, EPA estimates that the formula fuel achieves only a 5.5 percent toxics emission reduction relative to the winter baseline gasoline, well less than the minimum 15 percent toxics emission reduction that is required. Using the comprehensive option for a fuel certification emissions model, projections for Mobile4.1, and the emission effects proposed for that option in sections B and C above, the following table lists EPA's estimated VOC, NOx, and toxics emissions from baseline winter gasoline under the two enhanced I/M scenarios and from the formula fuel. (As described above in section 3, oxygen effects have been modeled using MTBE as the oxygenate.) High evap Low evap Enhanced I/M case fuel Baseline Formula Baseline Formula Exhaust VOCs (g/mi) .77 .71 .66 .64 Total VOCs /1/ (g/mi) .77 .71 .66 .64 NOx (g/mi) .930 .947 .930 .947 Exh Benzene (mg/mi) 41.73 35.49 35.96 30.59 1,3-Butadiene 4.95 4.88 4.27 4.20 Formaldehyde 10.72 13.27 9.24 11.43 Acetaldehyde 7.56 7.63 6.51 6.58 POMs 1.40 1.40 1.40 1.40 Total TAPs (mg/mi) 66.36 62.68 57.38 54.20 /1/ This assumes, as described above in section III.A.4, that wintertime evaporative emissions are negligible. There are two alternative proposals for setting the toxic emission standard. Under the first option, the winter toxics standard would default to the 15 percent reduction requirement with the required reductions being measured against the toxic emissions of winter baseline gasoline. The summer standard, on the other hand, would be determined by the performance of the formula fuel, that is, a projected reduction of 15.5-18.8 percent relative to the summer baseline gasoline emissions. Under the second option, the performance of the formula fuel would be averaged for the summer and winter baseline gasolines. The resulting annual average toxic emission reduction, if greater than 15 percent, would become the minimum standard for toxic emission control and would be applied on a year round basis. When averaging the summer and winter performance of the formula fuel based on the length of the season (assuming a high ozone season of five months) or based on the mass of toxic compounds which emitted during the season, the resulting annual average emission reduction of the formula fuel is, in either case, projected to be less than 12 percent. The standard would thus, under this option, default to a minimum 15 percent year round reduction in toxic emissions. Under either option, EPA proposes that the minimum standard apply for winter toxic emission control, as proposed for the summer. The Agency encourages comments regarding the impacts of setting a year round or a seasonal toxic emission standard for reformulated gasoline. IV. Fuel Certification by Modeling To certify a fuel as reformulated, the person refining, blending, or importing the fuel for ultimate sale in a covered area must declare the composition of the fuel and demonstrate to EPA that it meets, not only compositional requirements, but also emission performance standards for VOCs, toxic air pollutants, and NOx. (Section VIII on Reformulated Gasoline Compliance describes procedures by which refiner, blender, or importer would declare a fuel's composition and demonstrate its performance.) There are two possible ways that emissions performance might be demonstrated: By subjecting the fuels to vehicle emission testing or by applying the fuels' compositional specifications to a computer model that predicts vehicle emissions based on varying fuel characteristics. In either case, it will be necessary to declare all of the compositional characteristics of the candidate fuel that are defined for the baseline gasoline (API gravity, octane, distillation points, RVP, and sulfur, olefin, benzene, aromatics, and saturates levels) plus its oxygen level, oxygenate type, and heavy metal content. The Agency believes that demonstrating fuel effects on emissions through testing is likely to be very expensive and time consuming. At the same time, it believes that there is adequate data available on the emissions effects of some fuel parameters to construct a model that could reliably predict the emissions effects of the fuel reformulations involving changes in those parameters. Use of such a model would be vastly cheaper than fuel testing and would generally yield results as reliable as testing would. (See section V for more information regarding variability and other impacts on testing reliability.) For the reasons discussed above, EPA believes that certification by modeling should be an option for refiners, blenders, and importers of reformulated gasoline and encourages comments regarding the advisability of a modeling option for fuel certification. A. Contents of the Model EPA believes that any emissions model must be validated by substantial and reliable test data and proposes that a model option be available which includes the effects of certain fuel parameters on NOx, VOC, and toxics emissions effects. As described above in section III.B, the Agency suggests two options with regard to which fuel parameters would be included in such a model. Under the first option, the model would contain only the more established effects of fuel benzene and aromatics levels on benzene emissions, oxygenates on aldehyde emissions, oxygen and aromatics on exhaust VOC emissions, and fuel volatility on nonexhaust emissions. Under the second option, a more comprehensive model would also include the impacts of fuel sulfur levels and its T90 distillation point, and possible fuel olefin levels. Section III.B also contains a brief discussion of the implications of using one option over the other. Comments on the use of an emissions model for fuel certification and the contents of such a model are encouraged. B. Updating a Model Because of ongoing testing programs which are addressing fuel effects on emissions, and because further emissions effects may be determined based on fuel certification testing, it is likely that any model included in this final rulemaking would need to be updated to account for future findings. (If further findings become available in time to provide adequate public notice and opportunity for comment, EPA will include such findings in any model prior to publishing the final rule.) Afterwards, the model could be changed only through additional rulemakings. The Agency proposes two options for updating a certification model. Under the first option, any fuel certified by a model will be saleable through December 31, 1999 (up until Phase II reformulated gasoline requirements take effect). Once the model has been revised, fuel producers will have a choice of certifying fuels under either the original or the revised model. Under the second option, any fuel certified by a model will be saleable throuqh December 31, 1999, unless that fuel no longer meets the emission standards under an updated model, in which case the fuel would be certified for no more than two years after the updated model becomes effective. While the first option would provide greater flexibility for refiners, the second option would ensure that environmental benefits are met. Under either option, a new model will be developed for Phase II reformulated gasoline, which takes effect in 2000. Comments regarding updating a model and on the impacts of model updates on fuel certifiability are encouraged. V. Certification by Vehicle Testing As discussed above, data with which to develop an emission model is limited for many fuel parameters. As a result, the model described above is not able to incorporate all fuel parameters which may have an impact on emissions. Because of this, EPA is also proposing that a fuel producer be able to certify its candidate fuel through vehicle testing options when the candidate fuel either includes parameters not incorporated in the model or when parameter values fall outside of the modeled ranges. The following sections discuss proposals regarding the general requirements of the vehicle testing options, the emission reduction requirements, the test procedures, the test fleet requirements, the calculational and statistical compliance methodologies, and EPA's verification provisions. A. General Requirements 1. Appropriate Conditions for Testing EPA considers testing to be an important alternative means of fuel certification so that fuel producers have an opportunity to develop more cost effective formulations beyond those allowed by the emissions model. However, testing may not be an appropriate option in all cases. EPA proposes that the testing option be limited to only those situations where the characteristics of the candidate fuel clearly fall outside of the range of fuel parameters and/or their values covered by the model. Without such a constraint, it may, depending on statistical compliance criteria, be possible for a fuel producer to use the statistical variance associated with testing to certify a fuel through the testing option which would fail to be certified under the modeling approach. For example, a fuel that would fail to meet the VOC requirement through the model by a small margin could be tested and potentially pass due to the testing error associated with any vehicle testing program. The range of fuel properties covered by the models is discussed in section IV. In some cases, however, it may be appropriate to permit testing even though the effect of the pertinent fuel parameters are included in the model. This is especially true as it relates to the comprehensive modeling option discussed in section III.B where fuel parameter effects may be included based on limited testing. In this case the amount of testing involved in the fuel certification protocol may be more extensive and thorough than the information that went into developing the model. Thus, there may be comparable or greater confidence in the certification test results than in the model and testing would be an acceptable option. Results from such testing would be very useful to increase the confidence in the effects of a number of the parameters in the model. Comments on the criteria to be used in allowing use of the testing option are requested. 2. Testing Options EPA proposes that the testing option be coordinated with the modeling option such that a fuel producer could certify under the testing option by either testing for all emission types (exhaust, evaporative, running losses, and refueling) or, with the Administrator's approval, testing for only exhaust emissions and modeling the remaining emission types. As discussed above, if a candidate fuel's parameters or their values fall outside of the range covered by the model, then testing is required (unless test data would be more comprehensive than the data supporting the model); if not, then testing is not permitted. EPA proposes that this requirement apply to exhaust and non-exhaust (evaporative, running loss, and refueling) emissions in different ways, since fuel parameters that affect evaporative emissions are likely to have an exhaust emissions effect as well, while the opposite is not necessarily true. As a result, if testing is required for non-exhaust emissions, it is required for all emissions, but if testing is required for exhaust emissions, it may or may not be required for non-exhaust emissions. If the latter case is true and the fuel producer wishes to model non-exhaust emissions, the fuel producer must prove to the Administrator that the fuel's non-exhaust emissions can be accurately determined by the model by proving that the RVP of the fuel falls within the range of 7 to 11.7 psi, that the distillation curve of the fuel is normal up to the 10% point, and that the effect of any oxygenates on the benzene vapor pressure of the fuel is factored into the model. If the fuel producer wishes to test, these conditions must be proved to be false. By allowing non-exhaust emissions to be modeled under appropriate circumstances even though exhaust emission are measured, EPA believes that not only will the candidate fuel's emissions be more accurately determined, but also testing resources can be focussed on those emission effects which the models predict with the least degree of certainty. The results from testing that is performed can then be used to improve the models, as well as improve EPA's estimates of the air quality benefits of reformulated gasoline. To the extent testing is performed, EPA proposes that it be performed for all the pollutants included in the reformulated gasoline certification requirements, including toxics. Failure to have such a requirement could allow fuel producers to "game" the certification requirements by permitting them to utilize the modeling option for one pollutant when it would be advantageous and the test results for another pollutant when it would be advantageous. Certified reformulated gasolines may then not meet all of the applicable emission reduction requirements in-use. For example, testing may show that a fuel may meet the VOC requirement but fail the toxics requirement, while the model may suggest that the fuel may meet the toxics requirement. Allowing the fuel to use the model for toxics would ignore fuel impacts on toxics that may not be addressed by the model. Testing costs could be significantly reduced if only the pollutants other than toxics are measured by testing, and toxics are allowed to be modeled. However, since the testing option must be used when the candidate fuel's parameters fall outside of the range of the model, EPA does not believe situations will exist where only the pollutants other than toxics need be measured. As discussed earlier, if a fuel parameter is expected to impact toxics, and is not covered by the model, toxics must be measured. In addition, EPA believes that any fuel parameter expected to impact the pollutants other than toxics in an unknown fashion such that testing is required, will likely have an unknown impact on toxics emissions and thus warrant the measurement of toxics emissions, as well. Comments are encouraged on the above options. If additional options are identified that provide assurance that emission reduction requirements for all pollutants are obtained, then EPA will consider them in the development of the final rule. 3. Seasonal Limitation on Testing In order to be certified as reformulated, a gasoline must meet VOC, toxics, and NOx emission requirements in the high ozone season (summer) and toxics and NOx emission requirements outside of the high ozone season (winter). (See section II.) EPA believes that fuel producers would not likely utilize a testing option to certify non-high ozone season fuels. First, there is no VOC emission reduction requirement in the winter. Consequently, the candidate fuel formulations are more likely to be determined by the formula fuel specifications. Secondly, the cost of testing at winter temperatures would be significantly greater than testing at normal temperatures due to the need to use cold-room test facilities. As a result, EPA is proposing that testing be an option only for high ozone season fuel certification. EPA is, however, soliciting comment on an option that would allow any fuel producer wishing to certify a non-high ozone season fuel by testing to petition the Administrator for such testing. Under this option the petitioner would be required to provide appropriate rationale to support such a request. This rationale would simply need to present some evidence that the fuel parameter in question may have a different effect in winter than in summer. Under this option, EPA proposes that any testing that is allowed would be performed either under appropriate wintertime conditions or under the same conditions used for high-ozone season fuels. EPA requests comment on the appropriate conditions for such testing. 4. Fuels The fuels to be used for fuel certification include the candidate fuel and the baseline fuel since the candidate fuel's emission performance must be compared to the baseline fuel's to determine compliance with the VOC, NOx, and toxics emission requirements. Section II defines both the parameters and their corresponding values for the baseline fuel. For testing purposes, it is appropriate to allow the measured value of each of the parameters to be within some tolerance due to the difficulty of producing a test fuel to precise specifications and the statistical error associated with the measurement procedures. Since it is likely that the baseline fuel will be supplied by only one or two fuel suppliers, the variability due to fuel production will be minimized. Furthermore, since the fuel's properties can be tested multiple times, the statistical error due to the measurement procedures can also be minimized. As a result, EPA proposes that the baseline fuel's properties be within the tolerances defined in the table V.1. Due to the difficulty in accurately measuring the initial boiling point (IBP) and the fact that its value tends to be controlled by the RVP and the 10% distillation point, EPA proposes that no limitations be placed on IBP for fuel testing purposes. Table V-1.--Summer Baseline Fuel Properties Mean Tolerance API Gravity, deg.API 57.4 +/-0.3 Sulfur, ppm 339 +/-25 Benzene, wt% 1.82 +/-0.3 RVP, psi 8.7 +/-0.3 Octane, (R+M)/2 87.3 +/-0.3 IBP, deg.F 10%, deg.F 128 +/-5 50%, deg.F 218 +/-5 90%, deg.F 330 +/-5 End point, deg.F 415 +/-20 Aromatics, vol% 32.0 +/-2.7 Olefins, vol% 9.2 +/-2.5 Saturates, vol% 58.8 EPA also is proposing two alternatives to direct use of the baseline fuel as defined above which would reduce the error in the emission reduction estimates resulting from blending of the baseline fuel. In the first, the model would be used to determine the exhaust VOC emission impact of the actual measured values of the parameters in table V.1 relative to the required mean values. If the emissions impact of the actual measured values is more than two percent different from that of the required mean values, the fuel would not qualify as a baseline fuel. In the second option the model would be used to adjust the emission effects of the baseline fuel to account for fuel properties that differ from those listed in table V.1. EPA requests comment on both these options and the tolerances defined in table V.1. In order to effectively evaluate the emission performance of the candidate fuel and be able to monitor and enforce the quality of the candidate fuel once introduced into commerce, the characteristics of the fuel which are important for evaluating emission performance must be known and verifiable. Accordingly, EPA proposes that, at minimum, the fuel properties listed above for the baseline fuel also be specified for the candidate fuel. 5. Emission Reduction and Fuel Testing Requirements In order to be certified as a reformulated gasoline, a candidate fuel (high ozone season fuel) must result in the required VOC and toxics emissions reductions below the emissions resulting from the baseline gasoline and show no increase in NOx emissions. As discussed in section III, the required reduction in toxics emissions is based on the formula fuel's toxics emission performance. EPA currently projects that the formula fuel will result in a 18.6 to 22.1 percent reduction in air toxics emissions relative to the baseline fuel in the high ozone season. The Agency will issue a supplemental notice when it determines the actual percent reduction associated with the formula fuel once MOBILE4.1 is available and the definition and effectiveness of enhanced I/M are better known. (See section III for a more detailed discussion of this issue.) For VOC emissions, however, the formula fuel does not achieve the minimum 15 percent emission reduction required by section 211(k)(3) of the CAA. Instead EPA is proposing the alternative that the candidate reformulated gasoline be required to achieve a 15 percent reduction in VOCs over baseline gasoline for all areas, or a 15 percent reduction over baseline gasoline for ASTM Class C areas and a greater percent VOC reduction for Class B areas. In sum, the Agency proposes that for testing purposes the candidate reformulated gasoline be required to demonstrate no increase in NOx emissions, the required reduction in VOC emissions, and the reduction in toxics emissions determined by the formula fuel (currently estimated to be 18.6 to 22.1%). These reductions are all relative to the summer baseline fuel in the high ozone season. A fuel certified through testing would have to be segregated from other reformulated gasolines unless the fuel producer demonstrated that mixing this fuel with other fungible reformulated gasolines certified using the model did not deteriorate the emissions reductions of either fuel beyond that estimated by a linear averaging of the emissions effects of the relevant parameters of the two fuels. A second testing option proposed by EPA is to allow testing to determine the effectiveness of modifying a single fuel parameter (or a number of fuel parameters if changing a single fuel parameter naturally results in changes in others) at reducing emissions. If more than one new parameter is varied and they are not the natural consequence of each other, then the fuel EPA would consider the fuel as being tested under the first testing option above. Under this second option, the goal of the fuel producer would be to identify the emissions effects of a fuel parameter which is not yet part of EPA's certification model, so that the effect of this new parameter can be added to other fuel modifications which are addressed by the model to produce a certifiable reformulated gasoline. The key requirement of such a program is that the emissions effect of the new fuel parameter be additive to those already in the model and not duplicative. To best insure this, the reference fuel should be as close to the final, desired reformulated gasoline as possible, with the exception that the value of the new fuel parameter should be that of the baseline fuel. The candidate fuel containing the changed value of the new fuel parameter would then be the same as the reference fuel in every respect, except for the value of the one new fuel parameter. Again, a number of fuel parameters could differ between the candidate and reference fuels, if the differences all naturally resulted from changing a single fuel parameter. In such a case, the emission effect which is determined will be attributed to all the fuel modifications and not just one of them. All the fuel modifications would have to be present for the effect to be attributable to a given fuel. The emission reduction associated with this fuel parameter(s) will then be used to adjust the emission reductions granted to the candidate fuel by the model. While this option does not reduce the testing burden, it enables the fuel producer to evaluate and produce fuels beyond the applicable range of the model while still taking advantage of the model. It also enables test data to be used more readily to update and improve the model over time. In addition, it allows the fuel to be fungible if the effect of the new fuel parameter neither deteriorates when the fuel is mixed with other fungible reformulated gasolines nor deteriorates the emissions effects of other fuels. In general, EPA does not expect reformulated gasoline to be marketed as a number of segregated unique formulae. Rather, the Agency anticipates that reformulated gasoline will vary in composition to reflect different circumstances, but will retain the overall emission performance required relative to the baseline fuel. If a reformulated gasoline has a unique formula, meaning that its properties have not been shown to blend in a non- deteriorating manner, it is possible that blending with other reformulated gasolines might not result in the same in-use emission reductions, as would have been achieved by the individual fuels. Thus, as EPA proposed in section IV of this notice, for candidate fuels whose parameters have not been shown to blend in a non-deteriorating fashion, the fuels must be either segregated in the fuel distribution system or, if the nature of the deteriorating effect is known, modified such that blends with other reformulated gasolines still meet the required emission reduct