Document zQjkGxkrdx1wKqLnqDvkZ0xR7

EPA's sole reason for proposing the new MACI limits is to comply with its interpretation of LEAN v. LJ'A, 955 F.3d 1088 (D.C. Cir. 2020). While LEAN requires that EPA "address" all IAPs known to be emitted by a source category, it does not mandate that EPA set numerical MACT floors for every HAP, particularly those that are already controlled to an adequate margin of. safety. LEAN y. EPA, 955 F.3d 1088, 1097 (D.C. Cir. 2020) ("an emission standard includes as many limits as needed to control all the emitted air toxics of a particular source category") (emphasis added). The unachievable numerical limits proposed by EPA purportedly regulate previously "unregulated" but already adequately controlled HAPs. Cie. Section II (describing EPA's analysis of. HAP emissions in prior coke oven source category rulemakings). EPA should conclude, consistent with the CAA, and the LEAAT decision, that it is not "necessary" to amend the MACT standard to include these limits. B. EPA Should Have Included All Available Emissions Data When Setting the MACT Standards for Sources EPA ignored relevant data when establishing the MACT floor limits. When establishing "MACT floor," EPA must reflect the emission limits "achieved" by the "best performing" sources. e.g., S'ierra ('lub v. EPA, 353 F.3d 976. 980 (D.C. Cir. 2004). Where, as here, there arc fewer than 30 sources in a category or subcategory, the MACT floor for existing sources must be at least as stringent as "the average emission limitation achieved by the best performing 5 sources (for which the Administrator has or could reasonably obtain emissions information)." 42 USC 7412((i)(3)(B) (emphasis added).'' In determining the "average," EPA must consider variability and set the standard at a level that the source can meet "every day and under all operating conditions." Mossville Env( 'I Aciion Now v. EPA, 370 F.3d 1232, 1232 (D.C. Cir. 2004). For this rulcmaking, EPA calculated the MACT floor limits for the "previously unregulated HAP emissions" by ranking data and determining the top five (5) sources with emissions information, as per CAA sections 112(d)(2) and (3) for existing sources, then calculating the upper predictive limit ("UPL").2' 88 Fed. Reg. 55858, 55876 (Aug. 16, 2023). The LPL calculation relics upon estimating the true average and true variance. While the estimation of the average can be confidently done with a small number of samples, the estimation of the variance requires a substantially larger number of samples and in particular samples that cover the range of varying factors. EPA's methodology results in proposed MACT limits that do not approximate the limits "achieved" by the best-performing five sources, and therefore do not comply with the CAA. In fact, the five best performing sources cannot reliably meet EPA's proposed limits based on their own testing data, which EPA did not consider in its evaluation of the MACT floor. EPA should reevaluate these proposed MACT floor limits, ensuring that it considers the full range of historic For sources in a category or subcategory, the MACT standard may not be less stringent than the emission control that EPA determines Is achieved in practice by the best controlled similar source. 42 U.S.C. 7412(43). SunCoke's comments focus on EPA's establishment of MACT standards for c,xisiing sources, since all SlinCoke I INIZ coke plants constitute "existing sources." See supra Section VII. To the extent EPA believes any of SunCoke's I INIZ coke plants constitute a "new source" for purposes of the proposed amendments, SunCoke makes similar arguments with respect to EPA's development of the MACT standards, and notes that EPA is required to set limits based on worst foreseeable circumstances for best performing sources. Sierra nab I, El'1, 167 F.3d 658 (I) C Cir 1999) A UPI. is a statistical tool that EPA uses to account for the variability inherent to emissions and emissions testing. 27 Sierra Club FOIA 2025-EPA-04883 ED_018388_00000109-00028 SC_EVERSPLIT0005702