


                                       
                                       
                                       
                                       
                                       
                                       
           National Emission Standards for Hazardous Air Pollutants:
                         Secondary Aluminum Production
                                       
                                       
Summary of Public Comments and Responses on Proposed Rule (77 FR 8576, February 14, 2012) and Supplemental Proposal (79 FR 72874, December 8, 2014)

                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                     U.S. Environmental Protection Agency
                 Office of Air Quality Planning and Standards
                     Sector Policies and Programs Division
                       Research Triangle Park, NC 27711
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                    (This page is intentionally left blank)
                                       
                                       
                                       
                                       
                                       


                               Table of Contents

Section	Page 
List of Acronyms and Abbreviations	4
Purpose and Background	5
Commenters	5
1. Risk Assessment	6
1.1 	Risk Assessment Inputs: Processing of ICR Data into Modeling File	6
1.2 	Risk Assessment Methods	8
1.2.1	Modeling of MACT Allowable Emissions	8
1.2.2	Modeling of Acute Exposure	9
1.2.3	Multipathway Screening	10
1.3 	Uncertainties in Risk Assessment	12
1.3.1	Individual Variability and Exposures	13
1.4	Risk Assessment Results	19
2. Rule Provision Changes	21
2.1 	Control Device Shutdown	21
2.2 	Change of Furnace Classification	22
2.3 	Hydrogen Fluoride Compliance	27
2.4	Hydrogen Chloride Compliance	27
2.5	Operating Requirements/Work Practices	29
2.6	Other Technical Corrections/Clarifications	32
3. Testing and Monitoring	38
3.1 	Testing Uncontrolled Group 1 Furnaces	43
3.2 	Worst Case Scenario Testing	44
3.3 	Reporting	48
3.5	Other	50
4. Compliance Dates	51
5. Startup, Shutdown, and Malfunction	52
5.1 	Removal of SSM Exemption	52
5.2 	Startup and Shutdown	64
5.3 	Malfunction and Affirmative Defense	74
6. Regulatory Impacts	89
6.1 	Cost and Economic Impacts	89
7. Miscellaneous Comments	92
7.1 	Commenter Endorsements	92
7.2	Other Comments	92

List of Acronyms and Abbreviations


ACGIH
American Conference of Governmental Industrial Hygienists
CAA
Clean Air Act
CBI
Confidential Business Information
CRWI
Coalition for Responsible Waste Incineration
D/F
Dioxins and Furans
EPA
U. S. Environmental Protection Agency
ERT
Electronic Reporting Tool
FAX
Fascimile
FR
Federal Register
HAP
HCl
Hazardous Air Pollutant
Hydrogen Chloride
HF
Hydrogen Fluoride
HON
Hazardous Organic NESHAP
MACT
Maximum Achievable Control Technology
MIR
Maximum Individual Risk
MOA
Mode of Action
NAAQS
National Ambient Air Quality Standards
NATA
National-Scale Air Toxics Assessment
NESHAP
National Emission Standards for Hazardous Air Pollutants
NFFS
Non-Ferrous Founders' Society
OM&M
Operation, Maintenance and Monitoring
PB-HAP
Persistent Bioaccumulative Hazardous Air Pollutant
PSD
Prevention of Significant Deterioration
RTR
Risk and Technology Review
SAPU
Secondary Aluminum Processing Unit
SIP
State Implementation Plan
SSM
Startup, Shutdown, and Malfunction
TTN
Technology Transfer Network
URE
Unit Risk Exposure










Purpose and Background 

On February 14, 2012, the EPA proposed revisions to the National Emission Standards for Hazardous Air Pollutants (NESHAP) for Secondary Aluminum Production to address the results of a residual risk and technology review and to correct and clarify rule requirements and provisions. The EPA received 12 comment letters on the 2012 proposal. On December 8, 2014, the EPA issued a supplemental proposal that addressed a limited set of issues. The EPA received 16 comment letters on the supplemental proposal. The purpose of this document is to summarize and respond to the comments received on the 2012 proposed rule and 2014 supplemental proposal that are not addressed in the final rule preamble; however, the final rule preamble also contains discussions of several of these topics. In the summarized comments below, we have indicated those comments that were submitted on the 2014 supplemental proposal. Unless otherwise noted, the remaining comments were received on the 2012 proposed rule. All comments are contained in Docket EPA-HQ-OAR-2010-0544.
Commenters	
The commenter, affiliation, and item number in Docket EPA-HQ-OAR-2010-0544 are listed below in Table 1.
 Table 1. Public Comments for Secondary Aluminum Production NESHAP, 2012 Proposed Rule and 2014 Supplemental Proposal, Docket EPA-HQ-OAR-2010-0544 
                        Docket No. EPA-HQ-OAR-2010-0544
                                Commenter Name
                                  Affiliation
                              2012 Proposed Rule
                                     0183
J. W. Vinzant
Kaiser Aluminum
                                     0189
Eileen A. Sottile
LKQ Corporation
                                     0190
Melvin Keener
Coalition for Responsible Waste Incineration (CRWI)
                                     0191
James Mallory
Non-Ferrous Founders' Society (NFFS)
                                     0192
J. W. Vinzant
Kaiser Aluminum
                                     0193
Charles D. Johnson
The Aluminum Association
                                     0194
Mike Palazzolo
Alcoa 
                                     0195
Russell Frye
SSM Coalition
                                     0196
Marcus Cooke
Cooke Companies International
                                     0197
Shannon Broome
Air Permitting Forum
                                     0198
J. W. Vinzant
Kaiser Aluminum (duplicate of 0192)
                                     0199
G. C. Barnett
Aleris International, Inc.
                          2014 Supplemental Proposal
                                       


                                     0286
Jerry Call
American Foundry Society
                                     0287
G. Vinson Hellwig, Robert H. Colby
National Association of Clear Air Agencies
                                     0288
G.C. Barnett
Aleris
                                     0289
Sean O. Alteri
Kentucky Department for Environmental Protection
                                     0290
J. Roger Crawford
Kaiser Aluminum
                                     0291
Garret Morris
Greater Baltimore Health Alliance 
                                     0292
Matt Owens
Century Aluminum
                                     0293
J. Roger Crawford
Kaiser Aluminum(duplicate of 0290)
                                     0294
James D. Jones
Alcoa
                                     0295
Charles D. Johnson
The Aluminum Association
                                     0296
James L. Mallory
Non-Ferrous Founders' Society
                                     0297
Shannon Broome
Air Permitting Forum 
                                     0298
Russell Frye
SSM Coalition
                                     0299
Ronald A. Amirikian
Delaware Division of Air Quality
                                     0300
Steven Klafka
Wingra Engineering
                                     0301
Emma Cheuse
Earthjustice
                                       



1. Risk Assessment
1.1 	Risk Assessment Inputs: Processing of ICR Data into Modeling File
Comment #1: In a comment on the supplemental proposal, commenter 0300 stated that when developing the original NESHAP for secondary aluminum, the EPA removed die casting operations from the NESHAP. The commenter stated that it would be appropriate for the EPA to revisit its decision to not develop NESHAP for die casting operations. The commenter stated that similar to the secondary aluminum production operations, furnaces at die casting facilities have emissions of dioxins/furans, chlorine, hydrogen chloride and particulate and the hazardous air pollutants released from the partial combustion of lubricants have not be identified, and likely contain products of incomplete combustion such as polycyclic aromatic hydrocarbons (PAHs).
Response: This comment was not made on the 2012 proposal and is outside the scope of the 2014 supplemental proposal, which solicited comment only on specific topics identified in the notice. See 79 FR at 72883-84. For informational purposes only, we have attempted to respond to the commenter's concerns.
The EPA disagrees with the commenter's statement that aluminum die casting operations were removed from the NESHAP and, therefore, are not subject to the NESHAP. Die casting operations that are substantially like operations at other secondary aluminum production facilities are subject to subpart RRR requirements. Specifically, a die casting facility that is a major source, is generally considered a secondary aluminum production facility and subject to subpart RRR if they operate a sweat furnace, a thermal chip dryer or a scrap dryer/delacquering kiln/decoating kiln. In addition, regardless of the types of operations conducted at a major source die casting facility, they are subject to the NESHAP if they process anything other than clean charge, internal scrap, or customer returns. See section 40 CFR 63.1503, definition of secondary aluminum production facility. Like a furnace at a major source die casting facility, a furnace at an area source die casting facility is subject to the subpart RRR emission standards for D/F with one exception. A furnace operated by an area source die casting facility that would be subject to subpart RRR only because the facility operates a thermal chip dryer, is not subject to the requirements of subpart RRR if the furnace melts only clean charge, internal scrap and customer returns, although the thermal chip dryer would be subject. See 40 CFR 63.1500(f). Although furnaces at die casters that are major sources of HAP would not be subject to subpart RRR if the furnaces handled only clean charge, internal scrap, or customer returns, and if the facility did not operate thermal chip dryers, delacquering/decoating kilns, or sweat furnaces, the EPA is not aware of any die casting facilities that are major sources of HAP that operate furnaces and are not subject to subpart RRR, and the commenter did not submit any contrary information.
Comment #2: In a comment on the supplemental proposal, commenter 0301 stated that the EPA's data collection found emissions of mercury and other pollutants for which the EPA has not proposed to set any emission standards; that is unlawful under section 7412(d) and (f), and National Lime Association. Commenter 0301 stated that as the D.C. Circuit has held, the EPA has a "clear statutory obligation to set emissions standards for each listed HAP." The commenter stated that the Act requires the EPA to set standards for all emitted hazardous air pollutants, and the EPA has no lawful basis for not doing so here; the EPA must set emission standards that assure the maximum achievable degree of emission reduction for mercury and other emitted HAPs that are currently unrestricted by the Secondary Aluminum standards.
Response: The subpart RRR MACT standards have regulated most HAP since promulgation in 2000. The standards include emissions limits for D/F, PM (as a surrogate metal HAP other than mercury), THC (as a surrogate for non-D/F organic HAP) and HCl (as a surrogate for acid gases including HF, chlorine and fluorine). 
We currently do not have sufficient data to establish emissions standards for mercury for this source category. To explain further, during the data collection efforts to support the RTR (including for developing emissions estimates and other inputs for the secondary aluminum RTR risk assessment), we gathered a substantial amount of emissions test data for PM, THC, HCl and D/F from many facilities through an information collection request (ICR). We also requested nine secondary aluminum companies to generate and submit compound-specific metallic and organic HAP data so that we could develop emissions estimates for all the HAP for input to the risk assessment model. The metallic HAP data collected were the measured concentration of the individual HAP metals in the baghouse dust. We did not collect actual emissions test data for any of these metal HAPs. To develop emissions estimates for each of these HAP metals for input to the risk model, a ratio of total PM to each HAP metal was developed based on the baghouse dust analyses. The ratios were then applied to the source specific PM emissions estimates to calculate the emissions estimates for each metal HAP for each source. This calculation approach is based on the assumption that the concentration of these metals in baghouse dust is the same as the concentration in the PM emitted from the process stack outlets. (See Development of the RTR Supplemental Proposal Risk Modeling Dataset for the Secondary Aluminum Production Source Category, EPA-HQ-OAR-2010-0544-0239). We believe this method resulted in reliable estimates of the emissions of the particulate metal HAPs (such as Pb, Be, Cd, As, Mn, Ni and Cr) since these metals are mostly, or entirely, in particulate form. However, mercury is different. Mercury is emitted in three main forms (i.e., gaseous elemental mercury, gaseous oxidized mercury, and particulate bound mercury). We have no data regarding what percent of the mercury are in particulate versus gaseous forms for this source category. However, based on our knowledge and experience of mercury emissions from other industries, we believe a significant portion would be in the gaseous forms. Therefore, the mercury data collected for this source category, which is entirely based on baghouse dust samples, are not measurements of mercury emissions since the mercury in the baghouse dust consists of the particulate bound mercury portion only. The gaseous mercury emissions were not measured from any units. Therefore, there are significant uncertainties regarding the mercury emissions estimates for this source category. 
We used these limited baghouse dust mercury data to generate rough estimates of mercury emissions for secondary aluminum sources for input to the risk modeling so that we could estimate the potential risks due to mercury emissions. Based on this approach, we estimated that the entire source category emits a total of less than one pound of mercury per year nationwide (as shown in table 3.1-1 of the risk assessment report). Nevertheless, these baghouse data and rough estimates of emissions are insufficient to establish emissions standards for mercury for this source category. Therefore, we are not promulgating emission standards for mercury as requested by the commenter at this time in this action. 
With regard to the other metal HAP and organic HAP, as mentioned above, the subpart RRR MACT standards have regulated those HAP since promulgation in 2000. The standards use PM as a surrogate to regulate the metal HAP (other than mercury) and THC as a surrogate to regulate non-D/F organic HAP. In addition, the 2000 rulemaking set separate MACT limits for D/F expressed as toxic equivalents and HCl as a surrogate for acid gases including HF, chlorine and fluorine, and in this rulemaking we are promulgating specific compliance provisions for HF, as discussed in the preambles to the proposed rules. Thus, the rule currently regulates all HAP known to be emitted from these facilities in measurable quantities.
1.2 	Risk Assessment Methods
1.2.1	Modeling of MACT Allowable Emissions
Comment #1: Commenter 0287 recommended that the EPA consider potential or allowable emissions, rather than actual emissions, as much as possible in evaluating residual risk. The commenter stated that because facility emissions could increase over time for a variety of reasons, and with them the associated impacts, the use of potential or allowable emissions is more appropriate; an analysis based on actual emissions from a single point in time could underestimate the risk. The commenter stated that the major source HAP thresholds are based on maximum potential-to-emit, as opposed to actual emissions, and air agencies issue permits based on potential emissions. The commenter stated that limiting the scope of a risk evaluation to actual emissions would be inconsistent with the applicability section of part 63 rules. The commenter stated that they were pleased that the EPA used allowable emissions in parts of the rulemaking but were concerned that the EPA continues to use actual emissions in other parts of its assessment. The commenter encouraged the agency to use allowable emissions in the future, including in assessing acute health risks.
Response: Consistent with previous risk assessments, the EPA considers both allowable and actual emissions in assessing chronic exposure and risk under section 112(f)(2). See, e.g., National Emission Standards for Coke Oven Batteries (70 FR 19998-19999, April 15, 2005); proposed and final National Emission Standards for Organic Hazardous Air Pollutants from the Synthetic Organic Chemical Manufacturing Industry (71 FR 34428, June 14, 2006, and 71 FR 76603, December 21, 2006). This approach is both reasonable and consistent with the flexibility inherent in the Benzene NESHAP framework for assessing ample margin of safety. As a general matter, modeling allowable emissions levels is inherently reasonable since this reflects the maximum level sources could emit and still comply with national emission standards. But it is also reasonable to consider actual emissions, where such data are available, in both steps of the risk analysis. See National Emission Standards for Coke Oven Batteries, 70 FR 19992, 19998 (April 15, 2005). The chronic inhalation risk assessment for the secondary aluminum source category was conducted using actual and allowable emissions, and all of the results were considered in determining risk acceptability and ample margin of safety.
The commenter also urges the Agency to rely on allowable emissions for the purpose of our acute screening assessment. The use of allowable emissions was not considered in the acute assessment due to the conservative assumptions used to gauge worst-case potential acute health effects. The conservative assumptions built into the acute health risk screening analysis include: (1) use of peak 1-hour emissions that are on average ten times the annual average 1-hour emission rates; (2) that all emission points experience peak emissions concurrently; (3) worst-case meteorology (from 1 year of local meteorology); and (4) that a person is located downwind at the point of maximum impact during this same 1­hour period. Thus, performing an acute screen based on allowable emissions would be overly conservative and at best, of questionable utility to decision makers.
1.2.2	Modeling of Acute Exposure
Comment #1: One commenter (0193) noted that the EPA concludes that the acute level health risk from the secondary aluminum source category is within acceptable bounds, even after using worst case values for dose response, dispersion modeling, hourly maximum exposure location, meteorology and stationary out-of-doors human activity, which the commenter stated exaggerates risk. The commenter emphasized that the acute risk estimates the EPA calculates are the result of combined worst case assumptions, are very conservative and have a low probability of occurring.

Response: The acute risk assessment is a screening-level assessment designed to identify facilities and pollutants that need further analysis and can provide information for our risk-based decisions. The screen provides us information such that for those facilities where the hazard quotients are less than 1, we do not expect acute effects, and for those situations where the hazard quotients are greater than 1, the potential for acute effects may exist. For those situations where the acute hazard quotients are greater than 1 based on our screening assessment, the fact that it is a screening-level assessment is taken into consideration. A finding of unacceptable risks is not based on any one risk metric (e.g., acute hazard quotients), but rather considering all of the health and other information available and the degree of uncertainty associated with that information.

Comment #2: In a comment on the supplemental proposal, commenter 0287 stated that they have expressed their concerns in the past with the EPA's use of Acute Exposure Guideline Levels (AEGLs) or Emergency Response Planning Guidelines (ERPGs) values to address acute exposures in the residual risk assessments and it appears the EPA is still using them for those purposes in this proposal. The commenter stated that in the December 2002 EPA document, "A Review of the Reference Dose and Reference Concentration Processes," the EPA stated that the primary purpose of the AEGL program is to develop guidelines for once-in-a-lifetime short-term exposures to airborne concentrations of acutely toxic chemicals; they are not meant to evaluate the acute impacts from routine emissions that occur over the life of a facility. The commenter stated that unlike the reference concentrations (RfCs) for chronic exposures, the AEGLs and ERPGs do not include adequate safety and uncertainty factors and cannot be relied upon to protect the public from the adverse effects of exposure to toxic air pollutants; the use of AEGLs or ERPGs in residual risk assessments is not appropriate and does not ensure that public health is adequately protected from the acute impacts of HAP exposure. The commenter stated that they were gratified to see that the EPA has increased its reliance on the California Reference Exposure Levels (RELs) to address acute exposures in the residual risk assessments and continue to urge the EPA to use the RELs for these assessments.

Response: The EPA does not rely exclusively upon AEGL or ERPG values for assessment of acute exposures. Rather, the EPA's approach is to consider various acute health effect reference values, including the California REL, in assessing the potential for risks from acute exposures. To better characterize the potential health risks associated with estimated acute exposures to HAP, and in response to a key recommendation from the SAB's peer review of the EPA's RTR risk assessment methodologies, we generally examine a wider range of available acute health metrics (e.g., RELs, AEGLs) than we do for our chronic risk assessments. This is in response to the SAB's acknowledgement that there are generally more data gaps and inconsistencies in acute reference values than there are in chronic reference values. In some cases, when Reference Value Arrays for HAP have been developed, we consider additional acute values (i.e., occupational and international values) to provide a more complete risk characterization. As discussed in the preamble to the proposed rule (79 FR 72874), the exposure guidelines the EPA considers depends on which exposure guidelines are available for the various hazardous air pollutants emitted. The EPA uses AEGL and ERPG values (when available) in conjunction with REL values (again, when available) to characterize potential acute health risks. However, it is often the case that HAP do not have all of these acute reference benchmark values. In these instances, the EPA describes the potential acute health risk in relation to the acute health values that are available. Importantly, when interpreting the results, we are careful to identify the benchmark being used and the health implications associated with any specific benchmark being exceeded.

1.2.3	Multipathway Screening
Comment #1: One commenter (0193) stated that the EPA's multi-pathway risk screening analysis combines multiple extreme assumptions and leads to an unrealistic estimate of risk. The commenter provided examples of the EPA's assumptions and data inputs which the commenter believes exaggerate potential risks:
 The site modeled is a farm adjacent to a secondary aluminum facility, including an aquatic pond. All foods are consumed daily for the entire duration by the farm occupants locally from on-site, including all fish from the pond.
 Downwind air dispersion of HAP's from the facility to the farm site and pond is assumed continuously using worst-case dispersion characteristics.
 Each person has an ingestion rate of food that exceeds expected body weight-normalized ingestion rates. In other words, all food stuffs are each consumed at a 90th percentile daily rate adding up to an enormous food intake rate per person. Fish consumption is all from the pond at the site, and the ingestion rate is assumed at the 99th percentile of daily population intake.
 The ingestion of fish also includes the assumption that upper-trophic level fish (due to an absence of large-bodied herbivore/detritivore fish to favor higher concentrations of bioaccumulative chemicals) results in higher concentrations in consumed fish. The linear food-chain assumption further maximizes concentration of bioaccumulative chemicals in higher-trophic level fish.
 The lifetime risk calculation includes continuous ingestion and dermal exposure at the site, including all food originating from the site, 365 days per year.
 Individuals were assumed to be exposed on a fraction of their bodies (i.e., their head, forearms, hands, lower legs, and feet) to contaminated soil from the TRIM.FaTE surface soil parcel with the highest concentration (N1) on a daily basis. For the water evaluation, individuals were assumed to be exposed to contaminated surface water with the same PBHAP concentration as the TRIM.FaTE screening scenario pond over their entire bodies on a daily basis.
 All livestock feed is from the local farm site (including grains and silage) and are all derived from most highly impacted locations.
 A nursing mother is assumed to be continuously exposed to contaminants by all ingestion pathways modeled and can pass the contaminants on to her infant through breast milk, including dioxins and mercury.
The commenter noted that it is remotely possible that a small number of these ultra-upper-bound assumptions might occur at some site at some point of time, it is unrealistic to model all of these extreme assumptions simultaneously and continuously for any realistic assessment of risk. Given the exaggerated levels estimated in the analysis, the commenter stated that the EPA has reached a reasonable conclusion with regard to the multi-pathway results. Beyond this conclusion, the commenter maintained, results of the multi-pathway analysis have no value and a more realistic and reasonable multi-pathway assessment would result in much lower risk estimates.
Response: The EPA originally developed the multipathway screening analysis to estimate worst-case non-inhalation exposures to PB-HAP. The approach uses numerous worst-case assumptions to ensure that exposure to PB-HAP is not underestimated. However since the time of proposal, the EPA has refined its emission screening thresholds for the multipathway analysis based on improved toxicity rating/scaling methods for POMs and dioxin congeners, as well as improved fate, transport, and uptake behavior through the aquatic food chain. We also expanded the multipathway analysis to include separate exposure scenarios, one for a subsistence fisher and one for a subsistence farmer. (See Appendix 4 of the risk assessment document for more details). Further, rather than using all worst-case meteorological and lake distance assumptions, we have incorporated more facility specific information into the multipathway screening scenarios. Given that it is highly unlikely that the most-exposed farmer is the same person as the most-exposed fisher, we consider these scenarios separately and do not consider it reasonable to add the risk results from these two exposure scenarios. While the screening analysis was not designed to produce a quantitative risk result, the factor by which the emissions exceed the threshold serves as a rough gauge of the "upper-limit" risks we would expect from a facility. Based on actual emissions from the secondary aluminum source category, facility-level emissions of POMs from all sources are below our revised multipathway screening threshold. With respect to dioxins, our revised multipathway screening threshold for cancer is exceeded by up to 70 times. None of the pollutants exceeded the chronic noncancer multipathway screening thresholds. Considering that these results are screening estimates that still incorporate a number of worst-case assumptions, it is likely that the actual multipathway cancer MIR and noncancer HQ values would be substantially less than 70-in-1-million and 1, respectively.
1.3 	Uncertainties in Risk Assessment
Comment #1: One commenter (0193) stated that the EPA's residual risk conclusions are reasonable; however the risk assessment calculations are likely to be substantial overestimations of the actual risk for secondary aluminum facilities. The commenter thought that each of these factors in and of themselves lead to an overestimation of risk, but when combined likely result in a synergistic impact leading to a much greater overestimation of risks. The commenter believes that the EPA's risk evaluation demonstrates more than an ample margin of safety for the existing Secondary Aluminum NESHAP.

Response: Uncertainty and the potential for bias are inherent in all risk assessments, including those performed for this rulemaking. Although uncertainty exists, our approach, which uses conservative tools and assumptions, ensures that our decisions are appropriately health protective and environmentally protective. A brief discussion of the uncertainties in the RTR emissions datasets, dispersion modeling, inhalation exposure estimates and dose-response relationships can be found in the supplemental proposal for this rulemaking (79 FR 72874). A more thorough discussion of these uncertainties is included in the Residual Risk Assessment for the Secondary Aluminum Source Category in Support of the 2015 Risk and Technology Review Final Rule, which is available in the docket for this action.

1.3.1	Individual Variability and Exposures
Comment #1: One commenter (0193) objected to the use of a safety factor of ten applied to the estimated hourly emissions maximum for short term acute health concerns such as for hydrogen chloride and hydrogen fluoride exposures. The commenter noted that the default value of ten is based on a study by the EPA of emissions in the Houston Texas area dominated by VOC emissions from petroleum and chemical operations that may not be representative of the emissions profile for hourly events in Secondary Aluminum. The commenter continued that a large uncertainty exists for evaluating acute risks, and the EPA has chosen a very conservative order-of-magnitude level safety margin in assessing acute exposure levels. The commenter stated that this level of conservatism will lead to a substantial overestimate of risk for acute exposures.
Response: This choice of a factor of ten for screening was originally based on engineering judgment. To develop a more robust peak-to-mean emissions factor, and in response to one of the key messages from the SAB consultation on our RTR Assessment Plan, we performed an analysis using a short-term emissions dataset from a number of sources located in Texas (originally reported by Allen et al. 2004). In that report, the Texas 	Environmental Research Consortium Project compared hourly and annual emissions data for volatile organic compounds for all facilities in a heavily-industrialized 4-county area (Harris, Galveston, Chambers, and Brazoria Counties, TX) over an eleven-month time period in 2001. We obtained the dataset and performed our own analysis, focusing on sources which reported emitting high quantities of HAP over short periods of time (see Appendix 3, "Analysis of data on short-term emission rates relative to long-term emission rates"). Most peak emission events were less than twice the annual average, the highest was a factor of 74 times the annual average, and the 99[th] percentile ratio of peak hourly emission rate to the annual hourly emission rate was 9. Based on these results, we chose the factor of ten for all initial screening; it is intended to cover all possible hourly 	peaks associated with routinely variable emissions. While there have been some documented emission excursions above this level, our analysis of the data from the Texas Environmental Research Consortium suggests that this factor should cover more than 99 percent of the short-term peak gaseous or volatile HAP emissions from typical industrial sources.
Thus, in the absence of source category-specific data or estimates, we use this default factor of 10 in the acute health screen. We further note that the commenter did not provide specific evidence that the acute factor of 10 is inappropriate for this source category, nor did the commenter suggest a different acute screening factor. However, even using this factor of 10, the potential for acute health risks from this source category was negligible.
Comment #2: One commenter (0193) asserted that the omission of a mobility/activity analysis is estimated by the EPA to overestimate risk by 25 percent for particulate HAPS (such as metals) based only on short-term mobility patterns, but the inclusion of long-term activity patterns, such as moving residences and job locations, would result in even lower risk estimates.
The commenter noted that another omission resulting in over estimation is consideration of indoor versus outdoor exposures in the assessment. The commenter argued that for particulate types of emissions the difference is significant, noting that the EPA estimates that indoor exposures for particulate types of emissions may be reduced by a value of 25 to 30 percent. The commenter concludes that the risks projected for metals and dioxin emissions are also additionally overestimated by a factor of 25 to 30 percent. 
The commenter maintained that combined with the mobility pattern issue impacts on risk, chronic individual and population risks from these factors are likely overestimated by an additional factor of two.
The commenter also argued that the EPA's omission of population mobility will bias the risk estimate for cancer incidence, as there is continued scientific interest in using a non-linear dose-response modeling approach for dioxin compounds.
Response: In general, we strive to strike a balance in our assessments, meaning that while some factors likely lead to overestimates of risk (e.g., 70-year exposure duration), others likely lead to underestimates of risk (e.g., building downwash not modeled). We conclude that the risk assessment for this source category is sufficient to support decisions on the acceptability of the risk and ample margin of safety. 
The commenter is correct that the EPA did not consider short-term or long-term mobility in the assessment. Mobility was not considered because we do not believe that assumption, in combination with other modeling assumptions, leads to a significant bias in the risk results in any one direction. For example, the EPA does not consider that homes and businesses may locate closer to facilities over time, which could lead to an underestimation of risk. Additionally, estimates of the maximum individual risk (MIR) contain uncertainty, because they are derived at census block centroid locations rather than actual residences. This uncertainty is known to create potential underestimates and overestimates of the actual MIR values for individual facilities, but overall, we believe it does not have a significant impact on the estimated MIR for a source category. Processes like deposition, plume depletion, and atmospheric degradation are not considered, which may introduce bias to the population exposure estimates. Finally, we did not factor in the possibility of a source closure occurring during the 70-year chronic exposure period, leading to a potential upward bias in both the MIR and population risk estimates; nor did we factor in the possibility of population growth during the 70-year chronic exposure period, leading to a potential downward bias in both the MIR and population risk estimates. These uncertainties are described in section 4.1 of the risk assessment document.
In assessing risk, we estimate the MIR or the maximum HI, which are defined as the risk (or hazard) associated with a lifetime of exposure at the highest concentration where we determine people may actually reside. For this assessment, we used the annual average ambient air concentration of each HAP at each census block centroid as a surrogate for the lifetime inhalation exposure concentration of all the people who reside in the census block. Basing exposure estimates at census block centroids may underestimate or overestimate actual exposure concentrations at some residences, as discussed in the response below. Further reducing exposure estimates for the most highly exposed residents by modeling their short-term behavior could add a systematic low bias to these results and is not expected to affect the MIR result. Thus, this risk analysis did not consider the short-term human activity patterns and their potential influence on exposure. Our experience with the National-Scale Air Toxics Assessments (NATA) (which modeled daily activity using the EPA's HAPEM model) suggests that, given our current understanding of microenvironment concentrations and daily activities, modeling short-term activity would, on average, reduce risk estimates about 25 percent for particulate HAP; it would also reduce risk estimates for gaseous HAP, but typically by much less. 
We acknowledge that our assumption that people may be present at their homes for 24 hours per day over a 70-year lifetime represents a scenario that likely overestimates the actual exposures received by people living near the facilities. Most people have daily activities that take them to areas where exposure concentrations are different and people may move to new residences periodically. Both of these behaviors will likely tend to lower lifetime exposures and, therefore, risk if they are no longer in the vicinity of sources in the source category. 
In our risk assessments, we estimate risk for the "maximum individual risk" (MIR) which is different than the "maximum exposed individual (MEI)." The MEI could be equal to or greater than the MIR as the MEI is the highest estimated risk to a hypothetical exposed individual, regardless of whether people are expected to occupy that area. The MIR, on the other hand, is the highest estimated risk to an exposed individual in areas people may actually occupy. We are using the MIR to be considered equivalent to the "individual most exposed" phrase from the CAA section 112(f)(2). This approach is consistent with the Commission on Risk Assessment and Risk Management report recommendation that exposure assessments should not be based on a hypothetical MEI.
Comment #3: In a comment on the supplemental proposal, commenter 0287 stated that in using long-term concentrations affecting the most highly exposed census block for each facility, the EPA's analysis dilutes the effect of sources' emissions by estimating the impact at the centroid of the census block instead of at the property line or wherever the maximum exposed individual is located. The commenter stated that census blocks can be large geographically, depending on the population density, so the maximum point of impact can be far from the centroid, including at or near the property line where people may live or work; the EPA itself alludes to this problem in the preamble. The commenter stated that even if the area near the property line is not developed, over time homes and businesses could locate closer to the facility. The commenter stated that while it is possible that population distribution is homogenous over a census block, this assumption is not necessarily accurate in considering the predicted impacts from the location of a source; using HEM-3, the EPA can identify the maximum individual risk at any point in a census block that is within a 50-kilometer radius from the center of the modeled facility. The commenter stated that based on HEM-3's power and ability, they suggest that the EPA abandon its use of the predicted chronic exposures at the census block centroid as surrogates for the exposure concentrations for all people living in that block; rather, they recommended that the EPA use the truly maximum individual risk, irrespective of its location in the census block.
Response: In a national-scale assessment of lifetime inhalation exposures and health risks from facilities in a source category, it is appropriate to identify exposure locations where it may be reasonably expected that an individual will spend a majority of his or her lifetime. In determining chronic risks, it is appropriate to use census block information on where people actually reside, rather than points at the property line, to locate the estimation of exposures and risks to individuals living near such facilities. Census blocks are the finest resolution available as part of the nationwide population data (as developed by the US Census Bureau); on average, a census block is comprised of approximately 40 people and about 10 households. In the EPA risk assessments, the geographic centroid of each census block containing at least one person is used to represent the location where all the people in that census block live. The census block centroid with the highest estimated exposure then becomes the location of maximum exposure, and the entire population of that census block is assumed to experience the maximum individual risk. In some cases, because actual residence locations may be closer to or farther from facility emission points, this may result in an overestimate or underestimate of the actual annual concentrations (although there is no systematic bias for average levels). Given the relatively small dimensions of census block in densely-populated areas, there is little uncertainty introduced by using the census block centroids in lieu of actual residence locations. There is the potential for more uncertainty when the census block are larger, although there is still no systematic bias. The EPA concludes that the most appropriate locations at which to estimate chronic exposures and risks are the census block centroids because: 1) census blocks are the finest resolution available in the national census data; 2) facility fence lines do not typically represent locations where chronic exposures are likely (i.e., people do not typically live at the fence line of facilities); and 3) there is no bias introduced into the estimate of the maximum individual risk (MIR) by using census block centroid locations. In its peer review of the methodologies used to estimate risks as part of the RTR rulemaking efforts, the EPA's Science Advisory Board (SAB) endorsed this approach. 
In addition to the approach described above, the EPA recognizes that where a census block centroid is located on industrial property or is large and the centroid is less likely to be representative of the block's residential locations, the block centroid may not be the appropriate surrogate. For this source category, as described in the risk assessment document available in the docket, in cases where a census block centroid was within 300 meters of any emission source (and therefore possible on facility property), we viewed aerial images of the facility to determine whether the block centroid was likely located on facility property. Likewise, we examined aerial image of all large census blocks within one kilometer of any emission source. If the block centroid did not represent the residential locations within that block, we relocated it to better represent them and/or we added receptors for residences nearer to the facility than the centroid. For this source category, we added several receptors for census blocks where the centroid location was not representative of the residential locations. Appendix 7 of the risk assessment, entitled Residual Risk Assessment for the Secondary Aluminum Production Source Category in Support of the 2015 Risk and Technology Review Final Rule provide additional information on these changes.
In assessing risk, we estimate the MIR, which is defined as the risk associated with a lifetime of exposure at the highest concentration where we determine people may actually reside. For this assessment, we used the annual average ambient air concentration of each HAP at each census block centroid as a surrogate for the lifetime inhalation exposure concentration of all the people who reside in the census block. Basing exposure estimates at census block centroids may underestimate or overestimate actual exposure concentration at some residences. Further reducing exposure estimates for the most highly exposed residents by modeling their short-term behavior could add a systematic low bias to these results and is not expected to affect the MIR results. Thus, this risk analysis did not consider the short-term human activity patterns and their potential influence on exposure. Our experience with the EPA's National-Scale Air Toxics Assessment (NATA) (which model daily activity using the EPA's HAPEM model) suggests that, given our current understanding of microenvironment concentrations and daily activities, modeling short-term activity would, on average, reduce risk estimates about 25 percent for particulate HAPs; it would reduce risk estimates for gaseous HAPs, but typically by much less.
Comment #4: In a comment on the supplemental proposal, commenter 0287 stated that in considering environmental justice issues, improvements are needed in the EPA's methods of evaluating environmental justice and encouraged the EPA to continue to consider these factors in developing the final rule and subsequent regulations. The commenter recommended that the EPA conduct the demographic analysis on individuals projected to experience a risk greater than 1-in-1-million and also on individuals living within five kilometers of the facility, regardless of projected risk, consistent with the approach used for the Hard and Decorative Chromium Electroplating and Chromium Anodizing Tanks source category. The commenter stated that the socio-economic analysis for this rule did not evaluate potential disparities within five kilometers for cancer risk at maximum allowable emission levels; this type of analysis is especially important in instances where a facility is located in a minority and low-income community. The commenter stated that the EPA only evaluated the risk to the population within a 50-kilometer radius, which could dilute the results by including populations not in the demographic groups most at risk. The commenter stated that this is especially the case if the source is located in or next to a minority or low-income population. The commenter recommended that an analysis at the five-kilometer distance be conducted to assess facility impacts to nearby environmental justice communities and that the rule writers work with the EPA Office of Environmental Justice to develop criteria and specific guidance on how to interpret and apply the outcome of these types of analyses in the rulemaking process.
Response: The commenter is correct that we performed the demographic analyses for the Secondary Aluminum source category differently than we did for the October 2010 proposed RTR for the chrome electroplating and anodizing source categories (75 FR 65067, October 21, 2010). We performed the demographic analyses for the chrome source categories using two approaches as examples of how such analyses might be developed, and invited public comment on the approaches used and the interpretations made from the results. In the first approach, we focused the analysis on the total populations residing within 5 km of each facility, regardless of their estimated risks, and examined the distributions across various demographic groups within those 5 km circles. That analysis was a "proximity" analysis in that it considered only the distance from the emission sources to surrounding populations, and not the estimated risks to those populations
In the second approach, we focused the analysis on the populations within 5 km of any facility estimated to have exposures to HAP which result in cancer risks of 1-in-1 million or greater or non-cancer hazard indices of 1 or greater. We examined the distributions of those risks across various demographic groups. In each approach, we compared the percentages of particular demographic groups to the total number of people in those demographic groups nationwide. We stated in the proposed rule for the chrome source categories that in future rules we planned to extend the analyses to cover the entire modeled domain for a facility (50 km radius) to capture all individuals with risks above 1-in-1 million or greater or non-cancer hazard indices of 1 or greater. We also stated that generally we have found that using a 5 km radius in the analysis will capture more than 90 percent of the individuals with cancer risks above 1-in-1 million.
In the February 2012 supplemental proposal for the RTR for the chrome source categories, our demographic analyses included populations within 50 km of each source (including those very near the sources) with risks of 1-in-1 million or greater or non-cancer hazard indices of 1 or greater (77 FR 6628, Feb. 8, 2012). We did not include analyses using a 5 km radius in that supplemental proposal. We believe that, where a risk assessment has been performed, it is more informative to consider the demographics of all populations (including those beyond 5 km) with elevated risks than to limit the demographics analysis to populations located within 5 km of a facility. Where a risk assessment has been performed, these populations are identified, and the source parameters such as low release height are taken into account. As discussed above, we have found that most exposure locations with the highest estimated risks are within 5 km of a facility, so extending the radius to 50 km has little impact on an analysis based on risks, but makes more sense because 50 km corresponds to the risk modeling radius and includes all populations with elevated risk estimates. We also note that we are working with the Office of Environmental Justice to implement the latest guidance, issued May 29, 2015, as appropriate, to ensure that environmental justice issues are being considered in the rulemaking process.
Comment #5: One commenter (0193) argued against the EPA's use of the upper bound estimates for dioxin related risk, and stated that given the wide-spread and common occurrence of dioxin and furans in nature, it is not likely that there are no defense mechanisms in the human body to protect against dioxin impacts, and it is unlikely that low dose exposures are as potent, molecule-for-molecule, as high dose exposures. The commenter concluded that the EPA dose-response model is likely to be a conservative overestimate of risks, especially at low exposure levels, but the quantification of the overestimate is not possible at this time.

The commenter cited several statements from the SAB Dioxin Review Panel concerning the need for nonlinear and linear methods for characterizing cancer risk which the commenter asserted indicate that the EPA's assumption of no-threshold linearity in the dioxin dose-response curve is likely to be most conservative and may overestimate risk substantially, especially at low dose levels.

Response: We disagree that the EPA should consider that the human body has natural defense mechanisms against health effects associated with dioxin exposure and note that the commenter provided no evidence in support of such assertions. In general the development of UREs and RFCs consider results from animal toxicological studies and/or human studies where an effect has been observed notwithstanding whatever natural defenses (if any) from HAP may be present in the species tested. Furthermore, the commenter provided no evidence for its conclusion that it is unlikely that low dose exposures are as potent as high dose exposures.
With respect to model use, the EPA's long standing science policy approach has been to use linear low-dose extrapolation as a default option for carcinogens for which the mode of action (MOA) has not been identified. We expect future EPA dose-response assessments may identify nonlinear MOAs and we intend to use those analyses, as appropriate, in our risk assessments. At this time, however, there is no available carcinogen dose-response assessment for inhalation exposure to dioxin that is based on a nonlinear MOA. 
Comment #6: One commenter (0193) disagreed with the EPA's assumption of a 70 year lifetime calculation for exposures to HAPs in the risk analysis, and the EPA's assumption that facilities will not experience partial or full curtailments of operations, asserting that this leads to additional overestimates of long term cumulative risk estimates. The commenter pointed out that over the last two decades, many Secondary Aluminum plants have been shut-down or experienced short and multi-year curtailments, leading to overestimation of cumulative risks. 
Response: The commenter is correct that the EPA assumes no curtailment of operations over the 70 year chronic exposure period. In assessing risk, we estimate the maximum individual risk (MIR) or the maximum hazard index (HI), which are defined as the risk (or hazard) associated with a lifetime of exposure at the highest concentration. For this assessment, we used the annual average ambient air concentration of each HAP at each census block centroid as a surrogate for the lifetime inhalation exposure concentration of all the people who reside in the census block. Basing exposure estimates at census block centroids may underestimate or overestimate actual exposure concentrations at some residences. The risk assessment also contains other assumptions that could lead to an underestimation or overestimation of risk; for example the assessment does not take into account population growth or human activity patterns over the 70 year period.
1.4	Risk Assessment Results
Comment #1: One commenter (0189) expressed support for the proposed amendments, stating that the emission standards proposed in this amendment will adequately protect the environment and communities close to Secondary Aluminum Production facilities, although the commenter also offered suggestions for the proposed rule (addressed elsewhere in this document or the preamble to the final rule).
Response: The EPA acknowledges the commenter's support.

Comment #2: In a comment on the supplemental proposal, commenter 0290 stated that they support The Aluminum Association's (TAA) comments asserting that the EPA's risk estimates are generally overestimated for hazardous air pollutant (HAP) emissions from secondary aluminum facilities because the EPA bases it calculations on unreasonable and overly conservative worst-case risk assumptions.

In a comment on the supplemental proposal, commenter 0295 stated that they support the EPA's finding of acceptable risks but believes that the EPA risk estimates are generally overestimated. The commenter stated that each of the factors assessed by the EPA in their risk assessment, including emissions, population location and exposures, multi-pathway exposures, and the dose response and health risk values of the emitted pollutants of concern, includes uncertainties that the EPA addresses in the preamble. The commenter stated that generally, the EPA addresses uncertainties by biasing the analysis high using a worst-case upper bound risk preference; the EPA also assesses ample margin of safety to provide risk protection as mandated under the CAA. The commenter stated that several key high-bias risk assumptions combine to make the estimates of risk high and well above actual risks likely to occur. The commenter concluded that the current MACT emission limits provide more than an ample margin of safety, including emissions from co-located facility-wide operations. The commenter stated that the factors that lead to this conclusion include:
 The determination of hourly emissions estimates based on annual emission rates,
 The decision not to address individual or population mobility, including long and short term activity patterns and indoor exposure differentials,
 The assumption of 70 year lifetime risk assuming no facility shutdowns, partial curtailments or other reductions in emissions that may occur,
 The conservative nature of the dose-response especially for D/F speciation that rely on unit risk estimates derived from the multi-stage single-hit no threshold models for calculating carcinogenic risk,
 For acute health concerns, the combined assumptions of worst case dose-response, worst case dispersion modeling, worst case hourly maximum exposure location, worst case meteorology and worst case stationary human activity,
 For the multi-pathway analysis, a combination of implausible assumptions for exposure/ingestion rates coupled with unlikely food source scenarios.

Response: Uncertainty and the potential for bias are inherent in all risk assessments, including those performed for this rulemaking. Although uncertainty exists, our approach, which used conservative tools and assumptions, ensures that our decisions are appropriately health protective and environmentally protective. We disagree with the commenter that the EPA's risk estimates are based on "unreasonable and overly conservative worst-case risk assumptions." A brief discussion of the uncertainties in the RTR emissions datasets, dispersion modeling, inhalation exposure estimates and dose-response relationships can be found in the proposal for this rulemaking (79 FR 72874). A more thorough discussion of these uncertainties is included in the Residual Risk Assessment for the Secondary Aluminum Production Source Category in Support of the August 2015 Risk and Technology Review Final Rule, which is available in the docket for this action.

Comment #3: In a comment on the supplemental proposal, commenters 0294 and 0296 stated that they believe that the EPA's conclusions regarding residual risk for the Secondary Aluminum source category are appropriate and they supported the Agency's proposed findings.

Response: The EPA acknowledges the commenter's support.



2. Rule Provision Changes
2.1 	Control Device Shutdown
Comment #1: Commenters (0192, 0193, 199) disagreed with the EPA's proposal to require a period of 24 hours before controls can be turned off on idled furnaces. 
Commenter 0192 stated that in batch-operated furnaces emissions stop once operations end, and that different furnaces have different emission characteristics so the shutdown sequence should be specified in the OM&M Plan.
Commenter 0193 argued that there is no reason to wait 24 hours to shut down APCD if a furnace is not operating. The commenter asserted that this is an unreasonable and unnecessary new restriction. The commenter proposed that the EPA allow controls to be turned off immediately after completion of the cycle, as there will be no emissions present and, further, that the EPA add the following new provision:
The owner or operator of a new or existing affected source that meets the following criteria may temporarily turn off associated fans, hoods, and APCD. During the idle period, the source will:
 Only burn clean fuel
 Not receive new charge, flux or alloying material 
 Before charging resumes, all associated fans, hoods and APCD must be turned on and operated continuously.
Facilities should also be allowed to turn off controls, without idling for 24 hours. The following reasons are examples:
 If a bag leak detection system (BLDS) alarm sounds and inspection requires entry into the baghouse to conduct a bag leak check.
 Additionally, if a source is going to do a routine internal baghouse inspection, the proposed change would require the furnace be idled 24 hours prior to this so the baghouse can be shut down.
 Finally, if a source is going to a complete bag change, the proposed change would add 24 hours of idle time before the baghouse can be shut off.
Commenter 0199 suggested the following rule language:
      "The owner or operator of a new or existing affected source that will be idle for at least 24 hours and will burn clean fuel only, and will not receive new charge, flux or alloying material, may immediately upon idling the source, temporarily turn off associated fans, hoods and any APCD. Before charging resumes, all associated fans, hoods and APCD must be turned on and operated continuously. "
      
Response: The commenters misinterpreted the proposed rule. The proposed rule did not require a furnace to be shut down for 24 hours before a control device can be turned off. The provision as finalized applies to "natural gas-fired, propane-fired or electrically heated group 1 furnaces that will be idled for at least 24 hours..." and allows associated capture and collection system fans and APCD to be temporarily turned off so long as the furnace will not receive new feed/charge, flux or alloying material. See 40 CFR 63.1506(m)(7) and Table 2 to Subpart RRR. Nor does the final rule require owners or operators to wait 24 hours before turning off a baghouse to respond to a bag leak detection system alarm, conduct a routine baghouse inspection, or conduct a change out of bags in a baghouse. The EPA's goal, as stated in the preamble to the proposed rule, is to allow owner/operators to turn off control devices for furnaces that will be shut down for extended periods, such as a weekend. The final regulation in 40 CFR 63.1506(m)(7) and Table 2 makes clear that if a furnace will be idled for at least 24 hours after the end of a cycle, then the associated capture and collection systems and APCD may be temporarily turned off. The rule also makes clear that before charging resumes, all associated capture and collection systems and APCD must be turned on and operated continuously.
Comment #2: One commenter (0189) noted that there are a variety of reasons, such as ordinary maintenance or market material issues, when there could be periods shorter than twenty-four (24) hours when the furnace could shut down or remain idle and not be charged with aluminum and/or fluxing. The commenter suggested that instead of allowing facilities to discontinue use of control devices only after remaining idle after a twenty-four hour period, the commenter proposed a thirty (30) minute time frame after the charge area is emptied and melting has completed. 
Response: As discussed in the preceding response, the regulation does not require a furnace to be shut down for 24 hours before a control device can be turned off. The purpose of this rule change is to allow facilities with furnaces that will be idle over a weekend, or other period exceeding 24 hours to discontinue use of control devices while the furnace is idle. The purpose of the rule change is not to allow control devices to be shut down if the furnace is idle for shorter periods of time. The EPA believes the added flexibility provided by the rule change will accommodate most situations identified by industry while balancing the need to practically and effectively enforce the emission control requirements of subpart RRR. In the EPA's experience, if control systems and devices for furnaces were allowed to be temporarily turned off for short time periods (i.e., less than 24 hours) based on the furnace being "idle" it would be difficult for a compliance inspector to retroactively confirm with certainty that capture and collection systems and APCD were being operated as required to achieve continuous compliance with emission standards and were being turned off only as permitted. The EPA has clarified in the final rule that use of control devices can be discontinued after the furnace cycle has been completed, if the furnace will remain idle for at least 24 hours. 
2.2 	Change of Furnace Classification
Comment #1: Commenters (0192, 0193) disagreed with requirements of §63.1514(c) for furnaces changing to from group 1 to group 2. The commenter stated that the EPA has provided no basis for reducing the limits or for requiring performance tests for group 2 furnaces, which, according to the commenter, have no potential for significant HAP emissions. 
Response: The original subpart RRR MACT rule subcategorized furnaces into group 1 and group 2 based on the difference in HAP emission potential from the materials melted (non-clean charge v. clean charge) and the type and amount of flux added (lack or presence of reactive flux). See 64 FR 6946, 6960 (February 11, 1999) (proposed rule). A group 2 furnace is defined as a furnace that melts, holds, or processes only clean charge and that performs no fluxing or performs fluxing using only nonreactive, non-HAP-containing/non-HAP-generating gases or agents. We defined and exempted group 2 furnaces from subpart RRR numerical MACT emission standards in part on the basis that they are clean and emit little or no HAP emissions; instead, we required work practice procedures under section 112(h). Id. at 6961. Although group 2 furnaces generally are not subject to numerical emission standards under subpart RRR, the EPA believes it is appropriate to require a furnace that seeks to change from group 1 to group 2, and therefore will no longer be subject to numerical group 1 emission standards, to demonstrate based on performance testing that it qualifies as a group 2 furnace. Unlike other group 2 furnaces, these furnaces were allowed to operate as group 1 furnaces without being restricted to clean charge and no reactive flux. The EPA is requiring these furnaces to demonstrate that they can operate within the emission profile of a group 2, rather than group 1, furnace. The EPA proposed specific emission levels for this purpose and received no comments that served to alter the proposed emission levels, and we are adopting them in the final rule as proposed. We note that the commenter did not submit any testing or other emissions information to document its assertion concerning the potential for HAP emissions. We also note that in the absence of a performance test requirement to confirm the low level of emissions of furnaces changing from group 1 to group 2, the EPA might consider whether allowing such changes is appropriate or whether its regulatory approach to group 2 furnaces should be reviewed.
Comment #2: Commenters (0192, 0199) objected to the requirements of §63.1514(d) for area sources to repeat testing every 5 years, as they are contrary to the requirements developed by the EPA that one-time testing was sufficient for area sources after a detailed evaluation for the 2000 NESHAP. The commenter asserted that since the original NESHAP included the ability to change operating modes there is nothing about changing modes that changes that situation. The commenter claimed that the EPA provides no basis in the proposed rule for imposing the burden of repeat testing on area sources.
Response: We did not intend for area sources to repeat testing every 5 years under the change of classification provisions, and we have made it clear in the final rule that area sources are not required to repeat testing every 5 years. 
Comment #3: In a comment on the supplemental proposal, commenter 0294 stated that the proposed limit at Section 63.1514(e) allowing four furnace operating mode changes in any 6-month period, with the ability of sources to apply to the permitting authority for additional furnace operating mode changes, is appropriate. 

In a comment on the supplemental proposal, commenter 0295 stated that while they objected in their comments on the 2012 proposal to the restrictions proposed for changing furnace classification, given the analysis provided in the supplemental proposal indicating that limited changes actually occur, and the EPA's proposal to allow furnace operating mode changes no more than 4 times in a 6-month period, or to obtain prior approval from the permitting authority if additional changes are needed, the requirements appear reasonable.

Response: The EPA acknowledges the commenters' support of the rule requirements on the frequency of switching furnace operating modes.
Comment #4: One commenter (0194) recommended that the EPA delete the parameter for "throughput" in §§63.1514(c)(1)(ii), (c)(3)(ii), (d)(1)(i) and (d)(2)(ii), and the requirement to establish a performance testing parameter. The commenter observed that this requirement is not necessary as emission limits under the rule are already normalized to process throughput (i.e., lb/ton).
Response: To ensure that any furnace that is changing its classification or operating mode is subject to the same testing requirements that apply to all furnaces at secondary aluminum production facilities, specific performance testing and compliance demonstration requirements, for example the requirement to establish process throughput or lime injection rates, have been removed from section 63.1514 in the final rule. Instead, section 63.1514 of the final rule requires that operating parameters be established in accordance with the performance testing and compliance demonstration requirements in sections 63.1511 and 63.1512.
Comment #5: One commenter (0194) stated that the requirements in Sections 63.1514(c)(1) and (d)(1) should be clarified to state that only uncontrolled group 1 furnaces must conduct performance testing for HF.
Response: The commenter is correct in that HF testing is only required for uncontrolled group 1 furnaces, and Section 63.1514 is revised accordingly in the final rule. 
Comment #6: One commenter (0194) stated that §§63.1514(a)(1)(ii), (a)(2)(iii), (a)(3)(ii), (a)(4)(iii), (a)(6)(iii), (b)(1)(ii), (b)(2)(iii) (b)(3)(ii), (b)(4)(iii), (b)(6)(iii), (c)(1)(ii), (c)(3)(ii) and (d)(1)(i) require the establishment of a "flux" rate parameter during performance testing, but it does not consider the fact that some group 1 furnaces do not flux. The commenter recommended that references to establishing a "flux rate" parameter during performance testing should be qualified to state "where applicable."
Response: In the final rule, as discussed above, the change of furnace classification provisions do not list monitoring and operating parameters that must be established; instead, the provision requires that the parameters be established in accordance with §63.1511(g).
Comment #7: One commenter (0194) noted that §§ 63.1514(a)(1)(ii), (a)(3)(ii), (a)(5)(iii), (b)(1)(ii), (b)(3)(iii), (b)(5)(iii), (c)(1)(ii), (c)(3)(ii) and (d)(1)(i) require the establishment of a "lime injection" parameter during furnace testing. The commenter observed that some baghouses used to control emissions from Subpart RRR regulated group 1 furnaces are not lime-injected, and the reference to establishing a "lime injection" parameter should be qualified to state "where applicable."
Response: In the final rule as discussed above, the change of furnace classification provisions do not list parameters that must be established; instead, the provisions require that operating parameters be established in accordance with §63.1511(g).
Comment #8: One commenter (0194) recommended that the performance test language throughout §63.1514 should be revised to match §63.1511(b)(1) of the current Subpart RRR rule:
      "The owner or operator must conduct each test while the affected source or emission unit is operating at the highest production level with charge materials representative of the range of materials processed by the unit and, if applicable, at the highest reactive fluxing rate."
One commenter (0199) asserted that in paragraph §63.1514(a)(l)(i), the EPA should clarify that highest throughput expected means throughput for the scrap chosen for the test, and that the scrap should be the worst-case material likely to be processed in the affected source.

One commenter (0199) stated that in paragraphs §63.1514(b)(l)(i), §63.1514(c)(l)(i), §63.1514(c)(3)(i), §63.1514(d)(1)(i), and §63.1514(d)(2)(i), regarding the rule text, "The performance test must be conducted with the scrap containing the highest level of contamination expected to be processed, at the highest throughput expected and using the highest rate of reactive flux injection expected ..." the EPA should clarify that these worst-case charge materials, and blends of these, have differing process rates and, therefore, the charge rate from the stack tests is not representative of the production rate that will be achieved during normal operations. 
The commenter (0199) disagreed with wording in paragraphs §63.1514(c)(1)(ii), §63.1514(c)(3)(ii) and §63.1514(d)(l)(i), "Parameters for throughput, capture, flux rate, and lime injection must be established during these tests," stating that throughput should not be included as a parameter. The commenter asserted that because the test requires "scrap containing the highest level of contamination expected to be processed ... " throughput should not be determined when running worst case materials.

Response: As discussed in the 2014 supplemental proposal (79 FR at 72902) and in section 3.2 of this document, we are clarifying the requirements for conditions under which performance tests for group 1 furnaces must be conducted. As provided in the final rule at 40 CFR 63.1511(b)(1), testing must be conducted "under representative conditions expected to produce the highest level of HAP emissions ... (considering the extent of scrap contamination, reactive flux addition rate and feed/charge rate)." If a single test condition is not expected to produce the highest level of emissions for all HAP, "testing under two or more sets of conditions (for example high contamination at low feed/charge rate, and low contamination at high feed/charge rate) may be required." Id. Regarding the establishment of operating parameter values used to monitor compliance, the final rule requires that those parameters be established as part of changing furnace classification in accordance with the section 63.1511(g) requirements for the establishment of monitoring and operating parameter values.

Comment #9: One commenter (0194) recommended that the EPA either delete the requirement to establish "parameters for capture" in paragraphs §§63.1514(a), (b) and (c) or provide detailed clarification on the specific parameters an owner or operator must establish for capture and then allow the aluminum industry an opportunity to comment on these specific requirements.
One commenter (0199) stated that in paragraphs §63.1514(a)(l)(ii), §63.1514 (a)(3 )(ii), §63.1514 (a)(4)(iii), §63.1514 (b)(1)(ii), §63.1514 (b)(2)(iii), §63.1514 (b)(3)(ii), §63.1514 (c)(1)(ii), §63.1514 (d)(l )(i), and §63.1514 (d)(2 )(iii), the EPA has not defined the term "capture" or what qualifies as a parameter suitable to meet this requirement. The commenter asked that the EPA define what is an acceptable capture parameter. The commenter stated that baghouse flow rate, baghouse pressure drop, or baghouse fan RPM are appropriate to be included as in such a definition of a capture parameter. The commenter also asked that the EPA make certain that it provide for alternative parameters to be used as appropriate.
Commenter (0197) recommended that the EPA should clarify the definition of a Capture and Collection System to read "systems including duct systems and fans and in some cases, hoods" to allow for a capture and collection system that does not have a hood. 
Response: To ensure that any furnace that is changing its classification or operating mode is subject to the same testing and compliance requirements that apply to the relevant classification of furnaces, specific performance testing and compliance demonstration requirements, including parameters for capture, have been removed from section 63.1514 in the final rule. Instead, section 63.1514 of the final rule requires that operating parameters be established in accordance with the performance testing and compliance demonstration requirements in sections 63.1511 and 63.1512. Specifically, in the final rule, furnaces that change their operating mode, like all furnaces subject to subpart RRR requirements, are subject to the applicable requirements for capture and collection systems in section 63.1506(c), the monitoring requirements for capture and collection systems in section 63.1510(d) as well as the testing requirements in section 63.1512, including the requirements for testing uncontrolled furnaces. We agree with the commenter's suggestion to modify the definition of "capture and collection system" to account for those systems that do not have a hood and the definition in the final rule reflects that change. 
Comment #10: One commenter (0199) suggested that §63.1514(a)(6)(i) and §63.1514(b)(6)(i) be changed to read, "Change label on the furnace to reflect uncontrolled operation." 
Response: The commenter correctly points out an inadvertent error in these two provisions. The final rule makes the appropriate corrections.
Comment #11: One commenter (0199) recommended that in paragraph §63.1514(c)(5)(iii), the text, "change the label on the furnace to reflect group 2 operation;" the label can be designed to reflect all planned operating modes and include compliant parameters for each mode.
Response: Because the operating mode of the furnace is not always obvious, the purpose of the label is to make clear to facility staff responsible for compliance, regulatory compliance inspectors, and others the current operating mode of the furnace. Therefore, it is necessary that the label affixed to the furnace communicate clearly the furnace's current operating mode, rather than all potential or planned operating modes, so the recommended change is not appropriate. 
2.3 	Hydrogen Fluoride Compliance
Comment #1: Commenters (0193, 0194) supported the EPA's rationale for adding an HF limit for uncontrolled group 1 furnaces that is the same as the current HCl emission limit. The commenter observed that both chloride- and fluoride-based salt fluxes are used to remove impurities or to dry dross; chlorine and fluorine are chemically similar; and the amount of flux required largely depends on the amount of impurities to be removed. The commenter further stated that establishing a flux usage limit for chloride- and fluoride-based fluxes will limit HCL and HF emissions from a furnace.
Response: The EPA acknowledges the commenters' support.

Comment #2: One commenter (0199) recommended that in §63.151l(c)(9), because the EPA has already approved Method 26 for HCl in an alternative request, it should list Method 26 along with Method 26A.
Response: The EPA has listed Method 26 along with Method 26A.
2.4	Hydrogen Chloride Compliance
Comment #1: Commenters (0192, 0193, 0194) supported the addition of the HCl estimation technique provided in §63.1510(t)((2)(ii), but noted that the definition of reactive flux does not specify which fluxes are considered to be reactive, except to state that argon and nitrogen are not reactive. The commenters suggested that the EPA include sodium chloride and potassium chloride in the list of non-reactive fluxes. Commenters 0192, 0193 included two attachments that demonstrate that sodium chloride and potassium chloride do not produce HCl when used as flux in aluminum refining processes. The commenters expressed concern that not recognizing the scientific evidence that sodium chloride and potassium chloride are non-reactive and do not produce HCl would lead to an unrealistic estimate of chloride emissions and unnecessary regulatory requirements with no environmental benefit. The commenters stated that this is particularly important for facilities that are area sources of HAPs and therefore do not have to test for HCl emissions. The commenters suggested that the rule language for §63.1510(t)((2)(ii) and §63.1512(e)(3) say, "chlorine contained in reactive flux" and "fluorine contained in reactive flux."
Response: In addition to the commenters' submissions, the EPA has obtained HCl emissions data from testing conducted at the Hydro Aluminum North America facility in Monett, Missouri. A test conducted on the Holding Furnace (EU0030) which receives molten metal from two melting furnaces and uses Metallpur(R)315 flux, which is primarily sodium chloride and potassium chloride, indicated low but detectable amounts of HCl (averaging 0.001 lb/ton for 3 runs; test results are in the docket for this rulemaking). Based on the limited information available to the EPA including the commenters' submissions and the Hydro Aluminum test results, the EPA concludes that use of flux containing sodium chloride and potassium chloride has the potential to produce HCl emissions. Therefore, we are not adding sodium chloride and potassium chloride to the list of non-reactive fluxes. We note that the Hydro Aluminum testing was conducted at a relatively low fluxing rate (25 pounds of flux per batch, 50 pounds of flux total per run of two batches) compared to furnaces such as rotary furnaces, which may use thousands of pounds of flux per furnace charge. For this reason, furnaces that use greater amounts of flux, if they use sodium chloride and potassium chloride fluxes, likely would have greater HCl emissions than the tested Hydro Aluminum furnace. However, we have revised the rule language for §63.1510(t)((2)(ii) and §63.1512(e)(3) to read, "chlorine contained in reactive flux" and "fluorine contained in reactive flux."
Comment #2: One commenter (0196) requested that the EPA correct the definition of cover flux. The commenter indicated that the definition is unclear and inaccurate. The commenter stated that cover flux is not reactive, even if the molten metal bed is agitated or moved. The commenter stated that the definition of cover flux proposed in the new Secondary Aluminum MACT is technically incorrect and is inconsistent with the definition that the EPA has used to define flux in the secondary aluminum industry for 38 years. The commenter continued by presenting the EPA standing definition of cover flux from the Air and Water Effluent Guideline for Aluminum Smelting (1974), given here:
      Fluxing. The addition of a covering flux to the molten aluminum melt forms a barrier for gas absorption and oxidation of the metal. The flux also reacts with nonmetallics, residues from burned coating, and dirt in the scrap, collects such impurities and allows physical separation from the molten aluminum. The exact composition flux cover used varies from smelter to smelter, but is generally some combination containing one or more of the following: sodium chloride, potassium chloride, calcium chloride, calcium fluoride, aluminum fluoride, and cryolite. A common flux mixture is 47.5 percent NaCl, 47.5 percent KCl, and 5 percent cryolite. At the melting point of aluminum, the fluxes usually range from a tacky semisolid to a liquid depending on the composition of the mixture and the technique used to remove it from the melt.
      The amount of flux used depends primarily on the material charged. Scrap containing a relatively large surface area, such as borings or turnings, creates large amounts of oxides and requires proportionally larger amounts of flux. The flux generally is added along with the aluminum scrap in amounts from less that 10 percent to 33 percent by weight of the material charged.
The commenter (0196) discussed why cover flux is used and described the processes in great detail. The commenter also stated that cover flux (NaCl, KCl, and cryolite) is inert and harmless. The commenter was concerned with the assertion that agitation of cover flux somehow causes reactions and increased emissions. The commenter indicated that calling cover flux reactive is inconsistent with the distinction between cover flux and reactive flux; and the physical constraints of processing secondary aluminum. The commenter recommended that the EPA should fix this inaccurate definition in the Final Rule, correcting terminology that defines cover flux as reactive flux if the metal bed is moved. 
Response: The commenter did not provide any stack test data that indicate that HCl is not formed when NaCl, and KCl are used as fluxes, or that HF is not formed when aluminum fluoride or cryolite is used in a flux. HF is known to be evolved from cryolite above 930F (500C) in the presence of water vapor (Alcoa MSDS for Synthetic Cryolite, available in docket EPA-HQ-OAR-2010-0544). However, we agree with the commenter that there are no test data indicating that agitation of the molten metal bath underneath a cover flux causes a cover flux to emit HAPs, and we have revised the definition of cover flux in the final rule as follows: Cover flux means salt added to the surface of molten aluminum in a group 1 or group 2 furnace, without surface agitation of the molten aluminum, for the purpose of preventing oxidation. Any flux added to a rotary furnace is not a cover flux. 
2.5	Operating Requirements/Work Practices
Comment #1: One commenter (0191) recommended that the agency not attempt to add specifications for cleaning processes such as those required for runaround scrap. The commenter did not believe that it is appropriate for the agency to delve into this topic and that the determination of the appropriateness of cleaning techniques be made in each affected facility. The commenter continued that if it was the agency's intention to continue to apply the requirements of the Subpart RRR rule to aluminum foundries that melt only clean charge and operate a thermal chip dryer, then the EPA should have stated that such facilities are covered only for the requirements covering the thermal chip dryer. The commenter also suggested that the definition of clean charge should be restated in its entirety, and not simply as a subtext in the discussion of appropriate scrap cleaning techniques. One commenter (0197) noted that the EPA solicited comment on whether it is necessary to include specifications for cleaning processes, principally thermal drying and centrifuging, to ensure that materials qualify as clean charge. The commenter asserted that any specifications for cleaning processes are not necessary for aluminum chips because there are sufficient financial, operational, and safety incentives to ensure chips are "clean." The commenter contended that such specifications are unnecessary for aluminum chips ("chips") in thermal chip dryers, given that facilities have sufficient operational incentives to ensure that chips are clean before processing them in their units. The commenter further explained the processes and incentives for cleaning chips. The commenter also noted that setting specifications is impractical as cleaning processes vary with the type and amount of contamination.
Response: In the proposal, we solicited comments on what methods could be used for cleaning scrap to ensure that scrap processed by certain methods qualified as clean charge. We did not propose scrap cleaning requirements but were interested in gathering information on appropriate specifications that could be applied to cleaning processes that would produce clean charge. However, we did not receive any information and are not making any rule changes to address specifications for cleaning processes.
Comment #2: Commenters (0192, 0193, 0194, 0199) opposed the proposed change to the §63.1510(b) rule language related to the revision of OM&M Plans. The commenters noted that the current language in §63.1510(b) related to submittal of revised OM&M Plans does not contain a 60-day waiting period and was part of an EPA and Aluminum Association settlement agreement under litigation of the 2000 Secondary Aluminum NESHAP ( 67 FR 16374, April 5, 2002). The commenters observed that the EPA provided no reason for changing this settlement agreement language, and thus did not provide the industry an opportunity to comment on the EPA's new reasoning. The commenters stated that the current approach for OM&M Plan revisions works well for State agencies and operating facilities, so the revised rule language would simply result in an unnecessary 60-day delay in operations' ability to respond to product or market needs. The commenters requested that the proposed changes to §63.1510(b) be deleted, otherwise, the EPA should describe why the change is needed and provide an opportunity for the aluminum industry to comment on the EPA's reasons. 
Commenter 0199 objected to the change because the OM&M Plan for a given facility commonly needs to be revised for any of a variety of reasons, such as the addition of a malfunction type, repair or replacement of equipment resulting in compliant operating parameter changes, new stack tests, or addition of new equipment, among others. The commenter observed that when these changes occur, it is necessary that the facility move forward and begin operating with the new procedures or parameters, without the need to wait for agency approval of the revised plan, or for a 60-day waiting period. The commenter noted that most states do not respond in any way to the submittal of an OM&M Plan. The commenter stated that, under the existing procedure, if the permitting authority determines any revisions are necessary, the owner or operator assumed the risk by beginning to operate under the revised OM&M Plan upon submittal.
In a comment on the 2012 proposal and the 2014 supplemental proposal, commenter 0197 and 0297 stated that regardless of the deadline set, the EPA should make clear that the source has achieved compliance by submittal of required documentation to the permitting authority in order to ensure that sources are not penalized for failure of regulatory authorities to act on submittals in a timely manner or situations where a regulatory authority requests plan revision and the pre-existing OM&M plan remains in effect after the compliance period. The commenters stated that alternatively, the EPA could revert to the prior rule language, under which a facility would operate under a newly-submitted plan unless the Agency indicated that revision was needed, and then the facility would be given a reasonable time to accomplish revisions and operate in compliance with that revised plan after submission.
Response: In response to the comments, we have removed from the final rule the proposed provision that an owner or operator must not begin operating under a revised plan until approval is received or until after 60 days. 
Comment #3: Commenters (0192, 0193, 0194) stated that anodized aluminum is clean charge material. Commenter 0193 argued that the EPA did not provide any data or information on what, if any, HAPs are emitted from melting anodized scrap, and that most dyes are inorganic and are used at very low application rates therefore this should not alter the scrap material. Commenter 0193 requested that the EPA maintain the current provisions.
Response: We proposed to clarify that anodized aluminum that contains dyes or sealants with organic compounds is not clean charge. Anodized aluminum containing organic dyes or sealants has the potential to result in formation of D/F when the aluminum is melted in secondary aluminum production processes, in the same way that D/F can be formed when aluminum coated with paints, other coatings, or lubricants is melted. The commenters submitted no test data or other documentation to support their assertion that anodized aluminum containing dyes or sealants with organic compounds is different from aluminum with paints, coatings, or lubricants that are already excluded from the definition of clean charge, such that melting such anodized aluminum in a group 1 furnace does not result in emissions of D/F or other organic HAPs. Therefore we are finalizing the provision as proposed.
Comment #4: The commenter (0197) indicated that the operation, maintenance, and monitoring provisions contained in section 63.1506(a)(5) of the proposed rule are redundant with the OM&M plan requirements and recommended that they should be deleted. The commenter stated that the OM&M plan requirements are sufficient to ensure compliance with the specific emission standards and work practices required under this Subpart. The commenter therefore proposed section 63.1506(a)(5) should not be included as a separate paragraph in this final rule.
Response: The OM&M plan requirements in section 63.1510(b) are consistent, and not redundant, with the provision in section 63.1506(a)(5) that an owner or operator must operate and maintain any affected source in a manner consistent with safety and good air pollution control practices for minimizing emissions. A determination of compliance with the provisions in section 63.1506(a)(5) include a review by the appropriate authority of information available including a review of operation and maintenance procedures, such as those contained in the OM&M plan. Section 63.1506(a)(5) is also consistent with the General Provisions in 40 CFR 63, Subpart A. We are adopting the revision to section 63.1506(a)(5) as proposed.
Comment #5: One commenter (0199) stated that the requirements of §63.1506(g)(5), (k)(3), and (m)(4) regarding continuous injection devices, may not be feasible in the case of repaired or replaced equipment, where a feeder setting may be lower on the repaired or replaced equipment than was established during the performance test while still providing the same lime flow rate (pounds/hour). According to the commenter, when this occurs, it should be sufficient to document the repair and show through a verification procedure that the lime flow rate is equal to or above the level established during the performance test at the new lime feeder setting.
The commenter suggested the following language be added:
      "For a continuous injection device, maintain free-flowing lime in the hopper to the feed device at all times and maintain the lime feeder setting at or above the level established during the performance test. In the event the equipment is repaired or replaced after the performance testing has been conducted, and the feeder setting required to maintain the lime flow rate is lower than that established during the performance testing, that lower the lime feeder setting should be documented in the OM&M Plan."
Response: In the final rule amendments, §63.1510(i)(4) has been revised as follows: "At least once per month, verify that the lime injection rate in pounds per hour (lb/hr) is no less than 90 percent of the lime injection rate used to demonstrate compliance during your most recent performance test. If the monthly check of the lime injection rate is below the 90 percent, the owner or operator must repair or adjust the lime injection system to restore normal operation within 45 days. The owner or operator may request from the permitting authority for major sources, or the Administrator for area sources, an extension of up to an additional 45 days to demonstrate that the lime injection rate is no less than 90 percent of the lime injection rate used to demonstrate compliance during the most recent performance test. In the event that a lime feeder is repaired or replaced, the feeder must be calibrated, and the feed rate must be restored to the lb/hr feed rate operating limit established during the most recent performance test within 45 days. The owner or operator may request from the permitting authority for major sources, or the Administrator for area sources, an extension of up to an additional 45 days to complete the repair or replacement and establishing a new setting. The repair or replacement, and the establishment of the new feeder setting(s) must be documented in accordance with the recordkeeping requirements of § 63.1517."

In addition, §63.1517(b)(4)(ii) has been revised as follows: 
"(ii) If lime feeder setting is monitored, records of daily inspections of feeder setting, including records of any deviation of the feeder setting from the setting used in the performance test, with a brief explanation of the cause of the deviation and the corrective action taken. If a lime feeder has been repaired or replaced, this action must be documented along with records of the new feeder calibration and the feed mechanism set points necessary to maintain the lb/hr feed rate operating limit. These records must be maintained on site and available upon request." 

Comment #6: One commenter (0199) noted that in regard to §63.1506(n)(l), "Maintain the total reactive chlorine flux injection rate and fluorine flux addition rate for each ...," until additional tests are conducted we do not have a fluorine injection rate to monitor or maintain. 
The commenter recommended that terminology should be consistent. The commenter observed that the proposal is currently written as chlorine flux injection rate and fluorine flux addition rate. The commenter suggested that the terms should be either flux injection rate or flux addition rate for both chlorine and fluorine.
Response: The EPA agrees with the commenter and this paragraph has been revised in the final rule to say "...chlorine flux injection rate and fluorine injection rate...." However, the EPA thinks both terms are equivalent and are used interchangeably throughout the rule.
Comment #7: In a comment on the supplemental proposal, commenter 0299 supports the additions and changes as they apply to sweat furnaces and believes they provide greater clarity and will improve understanding of the requirements for these facilities.
Response: The EPA acknowledges the commenter's support of the rule requirements for sweat furnaces.
2.6	Other Technical Corrections/Clarifications
Comment #1: One commenter (0192) requested that the EPA allow facilities to terminate the required scrap sampling and analysis if the facility has strict purchasing specifications and has generated a long-term database that shows very low oil and coatings content in scrap purchased in accordance with the purchasing specifications.
Response: Owners or operators are responsible for inspecting scrap to confirm that the contaminant level is no greater than that of the scrap charged during the performance test. Monitoring of scrap contamination is essential to demonstrate continuous compliance with the standard. Therefore, we are not changing the rule as recommended by the commenter.
Comment #2: One commenter (0192) agreed with the proposed change to the group 2 furnace definition to clarify that pots are not furnaces, but the commenter recommended that the words "to customers" should be deleted.
Response: In response to the comment, the last sentence of the definition of group 2 furnaces has been revised to read, "Unheated pots, to which no flux is added and that are used to transport metal, are not furnaces."
Comment #3: Commenters (0193, 0194) asked that the EPA clarify the definition of an aluminum scrap shredder. The commenters expressed concern that the addition of the term "shearing" to this definition would redefine the affected source and expand the applicability of scrap shredder requirements well beyond the types of shredders that the EPA considered when developing the 2000 Secondary Aluminum NESHAP. The commenters noted that under the 2000 NESHAP rule, the EPA established emission standards for aluminum scrap shredders that are stand-alone devices used to size a variety of aluminum scrap for further processing in a scrap dryer/delacquering kiln or furnace and have emission control devices to capture particulate emissions generated from crushing and grinding, as well as paints, coatings and other contaminants present on the scrap. The commenters' concern is that the addition of the term "shearing" could expand applicability of the rule to continuous "shearing" operations used to cut aluminum plate, sheet or foil products to final width. The commenters explained that this type of shearing operation is not a stand-alone operation for processing scrap, does not generate significant particulate emissions and does not have separate capture or add-on air pollution control. The commenters concluded that NESHAP requirements for aluminum scrap shredders (e.g., 0.01 gr/dscf limit, bag leak detector, etc.) could therefore not be reasonably applied to this type of "shearing" operation.
The commenters recommended that the EPA either delete the proposed addition of the word "shearing" to the aluminum scrap shredder definition or clarify in the definition that "shearing and cutting operations performed at rolling mills and aluminum finishing operations (such as slitters) are not aluminum scrap shredders."
Response: The definition has been revised by adding the sentence, "Shearing and cutting operations performed at rolling mills and aluminum finishing operations (such as slitters) are not aluminum scrap shredders."
Comment #4: One commenter (0192) supported removing the outdated guidance document for bag leak detection systems. The commenter stated that the use of the manufacturers operating and maintenance instructions is the most reliable method for assuring proper operation of the systems.
Response: The EPA acknowledges the commenter's support. 
Comment #5: One commenter (0194) supported using manufacturer's O&M instructions for bag leak detectors and deleting the reference to the EPA Triboflow leak detector guidance. The commenter stated that the leak detector guidance had caused confusion with respect to good practice for at least one of the commenter's locations.
Response: The EPA acknowledges the commenter's support. 
Comment #6: One commenter (0194) recommended that the added reference to "HF" in §63.1516(b)(2)(iii) should be deleted, because this provision relates to furnaces with add-on control devices and these furnaces are not subject to the new HF limit.
Response: We agree with the commenter and the final rule does not include the reference to HF in §63.1516(b)(2)(iii).
Comment #7: The commenter (0197) recommended that the EPA should clarify the definition of Clean Charge. The commenter noted that all materials listed in the first sentence are separated by semicolons with the exception of the final material, "runaround scrap". The commenter stated that "runaround scrap" is a distinct and defined item and a semicolon should be inserted after the "or higher; and before the "and runaround scrap" to delineate it from the previous term.
In a comment on the supplemental proposal, Commenter 0286 stated that the EPA needs to clarify the definition of "clean charge" to be consistent with its guidance and policy found in the Agency's Applicability Determination Index (ADI) Database. The commenter stated that on July 26, 2007, the EPA published in the FR recent postings to the ADI Database including determination M060039 (dated 03/01/2005), which provided guidance as to whether aluminum sows, ingots, and T-bars that have painted markings are considered clean charge under subpart RRR. According to the commenter, determination M060039 noted that there was a typographical error in the definition of clean charge and that the EPA intended these charge materials, "notwithstanding ink, grease or paint markings, should be treated as clean charge." The commenter stated that the EPA then published an amendment in the FR on October 3, 2005 correcting a punctuation error found in the definition of clean charge and clarified that the phrase "known by the owner or operator to be entirely free of paints, coatings, and lubricants" should only apply to aluminum scrap. The commenter recommends that the EPA should take the opportunity to clarify that the phrase, "known by the owner or operator to be entirely free of paints, coatings, and lubricants" applies only to aluminum scrap and does not apply to charge materials such as T-bar, sow, ingot, billet, pig and alloying elements. The commenter recommend that, to ensure that the definition of "clean charge" is consistent with the EPA's current policy and guidance on this matter as described in ADI M060039, the EPA add a sentence to the definition of clean charge (as proposed February 14, 2012) to read as follows: 
      "Clean charge means furnace charge materials, including molten aluminum; T-bar; sow; ingot; billet; pig; alloying elements; aluminum scrap known by the owner or operator to be entirely free of paints, coatings, and lubricants; uncoated/unpainted aluminum chips that have been thermally dried or treated by a centrifugal cleaner; aluminum scrap dried at 343 °C (650 °F) or higher; aluminum scrap delacquered/decoated at 482 °C (900 °F) or higher, and runaround scrap. Aluminum T-bars, sows, ingots, billets, pigs and alloying elements notwithstanding ink, grease or paint markings, are clean charge. Anodized aluminum that contains dyes or sealants with organic compounds is not clean charge."
Response: The comma following "or higher" has been replaced with a semi-colon in the final rule amendments to promote ease of reading; however, this is not a substantive change.
Regarding the comment received on the supplemental proposal concerning the definition of "clean charge" and certain aluminum sows, ingots, and T-bars, the scope of the 2014 supplemental proposal was limited to discrete topics that we expressly stated did not include "clean charge definition." 79 FR at 72884. Even in the 2012 proposal, our proposed revision to the definition of "clean charge" only addressed the topic of anodized aluminum. Accordingly, this comment is outside the scope of this rulemaking, and no action was taken in the final rule in response to the commenter's request.
Comment #8: One commenter (0199) suggested that in §63.1503 the definition of "tap" be revised to read: "Tap means the portion of an operating cycle when processed molten aluminum is poured from a furnace." The commenter expressed concern that as proposed, there may be a problem when conducting stack testing of multiple furnaces. The commenter noted that the commenter currently conducts stack testing of multiple furnaces with each one starting at a different point in the operating cycle and continuing through a complete operating cycle for each furnace, with each cycle containing a number of operating tasks including charging, melting, skimming, alloying, and tapping. The commenter stated that these stack tests have been accepted by both the EPA and the states. The commenter explained that for stack testing purposes, the beginning or end of the cycle can be at any point as long as all tasks are accomplished, meaning that tapping should not be defined as the end of the furnace cycle, rather tapping should only refer to the removal of molten aluminum from the furnace. The commenter noted that the commenter would not have the personnel required to simultaneously test multiple furnaces if the test runs all had to begin concurrently with charging and end with tapping.
Response: To address the commenter's concern, we have revised the definition of "tap" as follows, to address the commenter's concern: "Tap means the end of an operating cycle of any individual furnace when processed molten aluminum is poured from that furnace."
Comment #9: One commenter (0199) stated that the EPA has recognized through test plans and alternative requests that a facility can use both charge pounds and aluminum produced pounds within the same facility for different furnaces. The commenter provided as an example the commenter's Aleris Rolled Products, Inc. facility in Lewisport, Kentucky, which has melting furnaces that use pounds charged while the holders use calculated aluminum pounds produced. The commenter noted that this particular case is documented in the "Lewisport Uncontrolled Reverberatory Furnace Test Protocol", Appendix G in the commenter's Consent Decree (attached to comment letter). The commenter maintained that there is no justification for restricting individual furnaces from using pounds of aluminum produced or pounds charged when determining compliance.

Response: The proposed rules did not include changes to subpart RRR requirements regarding whether individual furnaces should be allowed to use pounds of aluminum produced or pounds charged when determining compliance and we do not believe a rule change is necessary based on the information available to the EPA.

Comment #10: One commenter (0193) supported the proposed addition of the new edition of the ACGIH guidelines manual for hooding requirements and allowing the older reference to remain. However, the commenter stated that due to the many different furnace types and configurations, the rule should be modified, with similar wording as is in the current Iron and Steel Foundry Area Source MACT, to state that each capture and collection system must be designed using acceptable engineering standards, such as those published by the American Conference of Governmental Industrial Hygienists. According to the commenter, this would allow hoods to be designed using the ACGIH guidelines as they were intended, as guidelines to assist in the design of capture and collection systems. The commenter recommended that the EPA clarify in the rule that the ACGIH-manual hooding specifications are intended to be used as guidelines, and that good engineering practices be included for site-specific instances where ACGIH hooding specifications are not practical and/or other engineering designs can be effective.
The commenter argued that the hooding guidelines promulgated by ACGIH were not intended to be regulation. The commenter stated that certain limitations exist within the guideline language that should be addressed by the EPA with engineering alternatives on a site specific basis. According to the commenter, where ACGIH hooding specifications are not practical due to furnace design, safety, and operating limitations/access, the EPA should consider work practice standards for furnace operations to minimize the potential for fugitive emission releases.
Commenter 0197 contended that the 23rd and 27th edition of Industrial Ventilation: A Manual of Recommended Practice, American Conference of Governmental Industrial Hygienists ("ACGIH Manuals"), do not contain specific requirements to guide facilities and should not be incorporated wholesale, by reference, into regulations. The commenter stated that EPA, in promulgating emission standards must explicitly state the requirements. The commenter recommended that the EPA should remove the reference to the ACGIH manuals, as well as the requirement in proposed section 63.1502 that capture and collection systems meet ACGIH Standards, the EPA should specify the elements that it believes are important and required so that the regulatory requirements are clear, and to the extent the manuals are referenced in the final rule, the EPA must specify which particular aspects of the manuals are being incorporated.
In a comment on the supplemental proposal, commenter 0297 stated that in the supplemental proposal, the EPA plans to provide an alternative to the ACGIH Manuals for design, installation, and operation of sweat furnaces only, and would not allow alternatives to the ACGIH Manuals for capture and collection systems installed on other sources subject to the rule. As they stated in their comments on the 2012 proposal, incorporating the ACGIH Manuals and requiring systems to be designed and installed to meet the ACGIH standards is inappropriate; the EPA must explicitly state the requirements as the ACGIH Manuals do not provide specific requirements for a facility to follow. The commenter stated that the 2012 proposal appears to actually exacerbate the problem in that it incorporates the entire manual rather than just sections 3 and 5. The commenter stated that due process and basic principles of rulemaking under the Administrative Procedure Act and Clean Air Act require the EPA to reasonably inform regulated entities of their obligations, but because the ACGIH Manuals are voluminous and it is unclear which specific aspects of the manual the EPA is intending to incorporate, finalizing the rule as proposed would contravene these fundamental legal requirements. The commenter stated that the manuals provide the general ventilation equations and simple diagrams for hoods and ductwork, also providing a framework to design simple conveyance systems; it does not and was not intended to address complex industrial exhaust systems with wide ranges of temperatures. The commenter stated that, thus, it is unclear what specific requirements from the manual the EPA intends to incorporate in proposed section 63.1506(c)(1) for the complex operations regulated here; under this proposal, sources will not have fair notice of the operating requirements for capture and collection systems.
Commenters (0192, 0193, 0199) requested that Secondary Aluminum Production facilities be allowed to use any ACGIH manual edition from the 23rd on rather than restricting the industry to the 23rd and 27th editions. 
Commenter 0192 supported the EPA's decision to allow use of the updated 27th edition of the ACIGH manual, and recommended that the EPA allow the use of any future editions that may be published. The commenter noted that the 27th edition has been renamed: "Industrial Ventilation: A Manual of Recommended Practice for Design." 
Response: We acknowledge commenter 0193's support for allowing the use of either the 23[rd] or 27[th] edition of the ACGIH Guidelines manual. In the preamble to the proposed rule, we solicited comment on whether there are alternatives to the ACGIH guidelines, such as ventilation rules or guidelines developed by other professional organizations, governmental agencies, or industry organizations that could be used in a more flexible rule provision. We also solicited comment on other ways to increase flexibility, such as using minimum face velocity, elimination of visible emissions, minimum pressure drop, or other suitable parameters to determine capture effectiveness. We did not receive any information from commenters or others on the potential parameters we listed but received the comments summarized above. We note that commenter 0193 requests that the rule provision be broadened to allow "acceptable engineering standards, such as those published by the [ACGIH]" and argues that the ACGIH manual be treated as guidelines as opposed to regulatory requirements, while commenter 0197 takes a different approach, asserting that the EPA should explicitly state the particular elements in the ACGIH manual that are required by the rule or preferably remove the regulatory reference to the ACGIH manual entirely. Based on our experience implementing the rule and in light of the lack of any comments containing recommended alternative ventilation rules or guidelines, as well as the lack of comments on potential capture efficiency parameter approaches listed by the EPA at proposal, we conclude there is no reason to adopt an approach different from the current one, i.e., incorporating by reference the ACGIH manual, with the revisions adopted in the final rule to allow use of the 27[th] edition in addition to the 23[rd] edition. We do not consider "acceptable engineering standards" to be sufficient or appropriate for capture and collection systems in this rulemaking, because this language does not provide sufficient guidance to the secondary aluminum industry and to permitting authorities as to what constitutes adequate capture and collection systems under the relevant provisions of subpart RRR. We also believe it is not necessary to revise the rule to specify particular elements of the ACGIH manual. We note that site-specific issues may continue to be addressed as facilities work with the EPA and delegated authorities during implementation of the rule. Accordingly, we are retaining the incorporation by reference of Chapters 3 and 5 of the 23[rd] edition and adding all of the 27[th] edition of the ACGIH Industrial Ventilation manual, as proposed.
The final rule incorporates by reference Chapters 3 and 5 of the 23[rd] edition, and the entire 27[th] edition, and the final rule accounts for the correct title of the 27[th] edition ("Industrial Ventilation: A Manual of Recommended Practice for Design"). However, it is not appropriate to incorporate by reference future editions of a document; in general, only editions already published may be incorporated by reference. 
In further response to the comment on the 2014 supplemental proposal concerning a proposed alternative to ACGIH Guidelines being limited to "sweat furnaces only", the commenter does not appear to oppose the proposed alternative for sweat furnaces but instead reiterates and supplements its comments on the 2012 proposal that ACGIH Guidelines should not be incorporated into Subpart RRR requirements. We have responded to the commenter's 2012 comments above, and to the extent the commenter provided new or additional comments in response to the 2014 supplemental proposal, they are outside the scope of that notice, which was limited to discrete topics that we expressly stated did not include "ACGIH Guidelines." 79 FR at 72884.
3. Testing and Monitoring

Comment #1: One commenter (0192) objected to the proposed language that "Performance tests be conducted under such conditions as the Administrator specifies based on representative performance of the affected source for the period being tested" under §1511(a). The commenter noted that this language seems to be extracted from subpart A section 63.7(e) but is incomplete and in a different context. The commenter questioned whether it has a different meaning in proposed subpart RRR. The commenter stated that if it does have a different meaning, it has not been explained or justified in the proposal; if it does not have a different meaning, it is unnecessary, and because it is incomplete and out of context, could lead to confusion. The commenter recommended that it should be deleted in either case.
Response: As discussed in the proposed and final rule documents, subsequent to a court decision we are eliminating the SSM exemption in subpart RRR, requiring that standards apply at all times, and making revisions to the applicability of the General Provisions. As in the proposed rule, the final rule removes from subpart RRR, Appendix A the applicability of §63.7(e)(1), which contains superseded references to periods of SSM, and revises §63.1511(a) to reflect pertinent portions of §63.7(e)(1). The commenter did not explain why it says revised §63.1511(a) is incomplete, out of context, and confusing, but making §63.7(e)(1) not applicable eliminates any conflict between the two provisions.
Comment #2: Commenters (0193, 0194) stated that monthly verification of lime injection rate is unnecessary and overly burdensome given that lime feed measurement is not likely to change significantly over short periods of time. The commenter recommended that the EPA establish a lime rate verification frequency of "at least once every 6 months" which would be consistent with all other monitoring devices used under the Secondary Aluminum NESHAP.
Commenter 0192 suggested that the EPA should exempt facilities with automated systems based on load cells to assure adsorbent flow from this requirement, as these systems create alarms or error messages when adsorbent flow is interrupted, making periodic verifications not necessary.
One commenter (0199) objected to the requirements of §63.1510(i)(4): "At least once per month, verify that the lime injection rate in pounds/hour (lb/hr) is no less than 90 percent of the lime injection rate used to demonstrate compliance during your performance test." The commenter asserted that this proposed measure contravenes the standard negotiated in a 2009 Consent Decree resolving a federal enforcement action against the company. The commenter agreed that lime flow rate is the important parameter and that it is necessary to periodically verify lime flow rates, but asserted that it is not necessary to verify lime flow rates monthly.
The commenter explained that under their Consent Decree, two time periods are used for the two lime flow rate methods: measuring volumetrically or using a scale. The commenter stated that if a scale is used, the Consent Decree provides that the commenter checks scale calibration semi-annually, and if the lime flow rate is determined volumetrically, it is appropriate to perform a routine verification. The commenter explained that where this has been necessary, the Consent Decree provides for bi-monthly verification if flow is monitored visually as required by §63.1510(i)(1)(i), and verification once every four (4) months if flow is monitored electronically as in §63.1510(i)(1)(ii). 
Response: The lime injection rate is a necessary parameter established during performance test to ensure continuous compliance with the emission standard, in the absence of monitoring emissions on a continuous basis and in light of performance testing that is required only every 5 years. Operation with potentially inadequate lime injection for a period of up to six months before the need for corrective action is determined could reduce the effectiveness of or damage the control device and thereby increase emissions above the emission standard. The commenters did not submit any supporting documentation for their assertions that monthly verification is "overly burdensome" or that lime feed measurement is "not likely to change significantly over short periods of time." Facilities with automated systems based on periodically calibrated scales or load cells should be able to verify lime injection rate easily by integrating the recorded hopper weight loss. Therefore, we are adopting the requirement as proposed.
Regarding the 2009 Consent Decree, the decree at paragraph 122 states clearly that each company is responsible for achieving and maintaining complete compliance with all applicable federal laws and regulations, and compliance with the Consent Decree does not necessarily mean compliance with the Clean Air Act or implementing regulations. Further, the Consent Decree does not limit the EPA's authority to revise subpart RRR.
Comment #3: Commenters (0192, 0193, 0194) stated that the new list provided in §1510(a) adds the phrase "including, but not limited to," which does not clarify which provisions of Subpart RRR actually apply to area sources. The commenters requested that the EPA delete the phrase from §1510(a). The commenter observed that if something is missing from §1510(a)(1) through (19), it should be added to provide clarity for area sources.
Response: Owners or operators of area sources are subject to monitoring requirements for those affected sources listed in §63.1500(c)(1)-(4). We are not changing the language in 63.1510(a) because the list is intended to be informational and a guide to applicable requirements to assist compliance by area sources, not to alter the applicability of promulgated Subpart RRR requirements. In addition to the requirements listed in § 63.1510(a), they must conduct whatever monitoring is necessary to ensure that these affected sources are operated in a manner consistent with safety and good air pollution control practices. 
Comment #4: Commenters (0192, 0193) requested that the requirement, "for the purpose of annual certifications of compliance required by 40 CFR part 70 or 71, the owner or operator of a major or area source subject to this subpart must certify continuing compliance" be removed, as this would result in an additional requirement for area sources and it is a Title V reporting burden that was specifically exempted for the secondary aluminum production.
Response: After review of the comments received, we agree this requirement would impose additional burden on area sources subject to Subpart RRR and provide limited benefit. We note the existing rule generally exempted secondary aluminum area sources from Title V permitting, and we did not propose to remove the exemption. The proposed compliance certification requirement for area sources is not included in the final rule.
Comment #5: Commenters (0193, 0194) stated that the proposed §63.1511(g)(5) states that "[i]f the owner or operator wants to conduct a new performance test and establish different operating parameter values, they must meet the requirements in paragraphs (g)(1) through (4) of this section and submit a revised site specific test plan and receive approval in accordance with paragraph (a) of this section." The commenters observed that paragraphs (g)(1) through (4) of this section are only relevant if an owner or operator wants to "use existing data in addition to the results of performance tests to establish operating parameter values....", but are not relevant if the facility "wants to conduct a new performance test and establish different operating parameter values." The commenters suggested that the new §63.1511(g)(5) should be revised to delete the reference to paragraphs (g)(1) through (4) to say: "[i]f the owner or operator wants to conduct a new performance test and establish different operating parameter values, they must submit a revised site specific test plan and receive approval in accordance with paragraph (a) of this section."
Response: The requirements of §63.1511(g)(1) through (4) apply to cases where an owner or operator wants to conduct a new performance test and establish operating parameter values, and in doing so wants to use existing data in addition to the results of the new performance test to establish operating parameter values for compliance monitoring. In the final amendments, paragraph (g) (5) has been revised to read, "If the owner or operator wants to conduct a new performance test and establish different operating parameter values, they must submit a revised site specific test plan and receive approval in accordance with paragraph (a) of this section. In addition, if an owner or operator wants to use existing data in addition to the results of the new performance test to establish operating parameter values, they must meet the requirements in paragraphs (g)(1) through (4) of this section."
Comment #6: Commenters (0192, 0194) supported the EPA's clarification that multiple new SAPUs are allowed under the rule. Commenter 0194 stated that creating a separate new SAPU greatly simplifies compliance and emissions sampling where multiple affected sources are controlled by a single lime-injected baghouse. Commenter 0192 agreed that grouping furnaces logically, based on their use, can make compliance calculations more straightforward. The commenter proposed that this concept should be extended to existing source SAPUs. 
Response: The EPA acknowledges the commenters' support for the revision providing for multiple new SAPUs, but we did not propose and are not finalizing a revision to allow multiple existing SAPUs, because it is only necessary to have one existing SAPU. 
Comment #7: In a comment on the supplemental proposal, commenter 0295 stated that they have concerns with the requirements found is sections 63.1510(d)(2)(ii)(D) and 63.1510(d)(2)(iii)(B) regarding the maintenance of readings `at or above 90 percent' of the pressure differential or fan RPMs in order to utilize the listed alternatives to annual flow rate measurements. The commenter stated that the 90 percent criteria are too restrictive to be of any practical benefit in the rule. This level of consistency is not routinely demonstrable for pressure differential and fan RPMs, especially for those involving an emission control device. The inherent repeatability of the measurement method itself already introduces variability of plus or minus ten percent when accounting for the repeatability of the test methods and equipment. In addition, hood static pressure can vary depending on the amount of dust loading on a fabric filter and the stage of cleaning associated with that filter, even when systems are operating normally. To account for this measurement and inherent system variability, the commenter believes that the regulatory requirement should be 80 percent of the indicator values.
In a comment on the supplemental proposal, commenter 0286 stated that, regarding the proposed alternatives (ventilation duct differential pressure or the fan revolutions per minute) to the use of Methods 1 and 2 for annual monitoring of capture/collection and closed vent systems (63.1510(d)(2)), the proposed requirement to maintain the measured parameter at or above 90 percent of the value measured during the most recent Method 2 test, is too restrictive to be of any practical benefit. The commenter stated that, presumably, if the alternative parameter is below 90 percent of the indicator value for any day of the year, the facility would be required to revert to measuring the air flow rate during that year to demonstrate compliance with the rule. The commenter stated that while this level of consistency may be representative for other types of parameters (e.g., oxygen content consistency in flue gases), it is not routinely demonstrable for pressure differential and fan RPMs, especially those involving an emission control device. The commenter stated that the inherent repeatability of the measurement method itself introduces variability of plus or minus 10 percent when accounting for the repeatability of the test methods and equipment; in addition, hood static pressure can vary depending on the amount of dust loading on a fabric filter and the stage of cleaning associated with that filter, even when systems are operating normally. The commenter stated that the allowable variability should be on the order of 80 percent of the indicator values to account for this measurement and inherent system variability.
Response: Where the EPA establishes a parametric operating limit to provide assurance of compliance associated with the performance of an operating parameter during a performance test, we rely on the capability of the facility to operate their equipment and emissions control devices, including those associated with parametric monitors, at the same level of operation as observed during the performance test. The parametric monitor must be a device capable of providing compliance assurance between emissions test events and capable of demonstrating when a distinct change in operation of the source or emissions controls has occurred. In the 2014 proposal, we proposed the daily monitoring of fan rpm as an alternative to the annual flow rate measurements. As a result of comments and the EPA's consideration of the comments, this rule provides alternatives to the requirement to conduct annual flow rate measurements of capture/collection and closed vent systems; all but one requires a parameter to be recorded daily and maintained at no less than 90 percent of the average value recorded during the most recent Method 1 and 2 performance test. Alternative monitoring parameters include differential pressure; fan RPM or fan motor amperage; static pressure measurements; duct centerline velocity using a hotwire anemometer, ultrasonic flow meter, cross-duct pressure differential sensor, venturi pressure differential monitoring or orifice plate equipped with an associated thermocouple. These monitoring alternatives rely on information from the parametric gauge reading as an indicator that a compliant volumetric flow rate is being achieved during the period between performance tests, which under these alternatives is extended to every five years. Variability in emissions test results is widely held to be +/- 15 percent, whereas parametric monitoring devices such as thermocouples can be calibrated to +/- 2 percent, pressure sensing devices to +/- 1%, amp meters to +/- 2 percent accuracy. Facilities should choose instrumentation that is capable of providing repeatable and accurate readings for their parametric monitoring records. Annual calibration of such devices is recommended, but not required by the rule. Furthermore, parametric monitors track trends that are generally not associated with process or other operational fluctuations, for instance an increase in production rate does not have a 1:1 relationship with an increase or decrease in fan motor amperage. So with parametric monitoring requirements, 90 percent (also described as a 10 percent deviation from the emissions test average) of the emissions test parametric reading average is selected as the trigger point for corrective action because it describes a control decision point; one where the monitoring data show that a process or control change has occurred. Where facilities are concerned that their parametric monitors may not provide this level monitoring assurance they would be well served to research the precision of the devices they currently have installed or plan to install and consider having their device(s) calibrated or choose one of the other compliance alternatives such as an annual direct flow rate measurement that does not rely on daily parametric monitoring measurement. With these devices calibrated or otherwise functioning normally the EPA believes that monitoring and maintaining a 90 percent operating requirement such as those in the monitoring alternatives provided in the rule is achievable for all facilities that are tracking their process rates regularly. In addition, neither commenter provided any data to support their arguments that a 90 percent level is inappropriate or that 80 percent is more appropriate.
Comment #8: In a comment on the supplemental proposal, commenter 0286 stated that the provisions for permanent total enclosures (63.1510(d)(2)(iv) acknowledges the potentially wide array of explanations for a measured parameter being outside an identified range by including a provision to investigate and restore systems when such a parameter is measured on occasion. The commenter stated that given the variability associated with ventilation systems and the requirement for daily parameter readings, a similar provision should be included for dealing with occasional static pressure and RPM measurements that are outside the applicable indicator value range and recommends that an provision such as follows be added to section 63.1510 (d)(2)(ii) and (iii):
      "When there are daily readings below 90 percent of the differential pressure (alternatively, RPM) measured during the most recent Method 2 performance test series, the facility may investigate and take steps to restore normal operation, in lieu of performing an annual flow rate measurement, provided that the reasons for the low reading are recorded."
Response: The EPA disagrees with the suggested revisions provided by the commenter. For parametric monitoring that is used by the source, instead of annual flow rate measurements, to assure continuous compliance with emission standards in the rule, corrective action must be taken to maintain monitoring parameters at or better than their performance testing baseline. Where a parameter deviates more than ten percent from this mark as an average value the EPA recognizes that the operation of the control responsible for the parameter being monitored has changed significantly and that the owner or operator must make the necessary repairs or adjustments to restore normal operations.
Comment #9: In a comment on the supplemental proposal, commenter 0294 stated that the EPA's proposal to incorporate permanent total enclosures (PTE) and the alternative inspection requirements for PTE are practical in the context of the affected sources within the secondary aluminum source category.
Response: The EPA acknowledges the commenter's support of the proposed rule language and the use of PTE in this respect.
Comment #10: The commenter (0197) stated that the EPA included thermal chip dryers in the lime injection monitoring requirements. The commenter believed that the inclusion of thermal chip dryers in proposed section 63.1510(a)(8) appears to be inadvertent because lime-injected fabric filters are not used as a control device for thermal chip dryers. If the addition of this requirement was not inadvertent, the commenter noted that it is inappropriate because the EPA did not provide notice, nor did it provide an explanation of why thermal chip dryer emissions should be controlled by a lime-injected fabric filter. The commenter recommended that thermal chip dryers should be removed from section 63.1510(a)(8).
Response: Lime injected fabric filters are not required for control of emissions from thermal chip dryers under the existing rule or the final amendments. However, if a lime injected fabric filter is used at the discretion of the owner or operator during the performance test, then it must continue to be used and the lime injection rate established during the performance test must be maintained.
3.1 	Testing Uncontrolled Group 1 Furnaces
Comment #1: In a comment on the supplemental proposal, commenter 0290 stated that as a part of the procedures to minimize untested fugitive emissions, the EPA should allow facilities to develop testing procedures on a case-by-case basis to demonstrate that the 67 percent capture efficiency assumption is not appropriate. The commenter stated that such a testing procedure might apply in particular to D/F emissions that are produced as a result of organic substances introduced with purchased scrap. According to the commenter, in a batch process those organic substances and their decomposition products have been emitted by the time the furnace charge has achieved a molten state and in some processes, the furnace remains completely closed up during the melting phase of the process and essentially 100 percent of emissions go up the stack. The commenter stated that a sequence of simultaneous stack tests with different durations could demonstrate that all D/F evolution occurs before the furnace is opened to perform necessary refining operations.
Response: It is possible that significant D/F evolution may occur when other-than-clean charge is initially placed in the furnace, while the furnace doors are open. If this is indeed the case, D/F capture might actually be lower than for pollutants like HCl. This possibility is supported by testing conducted at Southwire on September 5, 2007 (see table). In that testing, only 42 percent of the total D/F measured was in the furnace stack; this value was actually lower than the average of 71 percent for all pollutants and all tests (and on which the proposed 67 percent capture assumption was based). If D/F is emitted when the furnace doors are open, and escapes through open furnace doors, there would not necessarily be a relationship between the D/F capture efficiency and the amount of D/F generated during various times in the cycle. Therefore, it is not appropriate to measure the D/F emitted at various times during the cycle and attempt to estimate a capture efficiency based on those data. Therefore, the EPA believes that it is not appropriate to measure D/F in a sequence of simultaneous stack tests with different durations to infer a particular D/F capture efficiency, and such testing procedures are not allowed in the final rule.
Southwire Test Results Summary
Test Report Date
Pollutant
Three-run Average
Fraction in Furnace Stack
Three-run Average
Fraction in Canopy Hood Stack
June 26, 2007
PM
0.78
0.22
June 26, 2007
HCl
0.81
0.19
June 26, 2007
D/F
0.97
0.03
Sept. 5, 2007
PM
0.49
0.51
Sept. 5, 2007
HCl
0.77
0.23
Sept. 5, 2007
D/F
0.42
0.58
Average for all
All
0.71
0.29

However, it would be possible to estimate capture efficiency on a case-by-case basis by, for example, using the building as an enclosure, and measuring D/F emissions from powered roof vents. Such case-by-case demonstrations would need to be approved by the appropriate authority. Additionally, in light of comments, we have re-examined the Southwire testing data on which the proposed 67-percent capture efficiency assumption was based, and have revised the assumed capture efficiency to 80 percent.
3.2 	Worst Case Scenario Testing
Comment #1: In a comment on the 2012 proposal, one commenter (0199) asked that the EPA address and clarify in §63.1511 that stack testing is to be conducted when charging the worst-case material likely to be processed in the affected source. The commenter explained that this may mean multiple tests of separate charge material consisting of dross and separate charge material with high oil or other organic material, and for some sources, this will mean testing of a worst-case blend of these materials. The commenter suggested that the EPA provide clarification that the desired results from these tests are in fact "worst -case emissions expected under normal operations," and that this is not representative performance, as the new language implies.
The commenter also recommended that the EPA clarify that these worst case charge materials, and blends of these, have differing process rates and, therefore, the charge rate from the stack tests is not representative of the production rate that will be achieved during normal operations.
The commenter requested that until the EPA addresses these additional critical issues, the EPA acknowledge and confirm that the agreed-upon stack testing protocols from the Consent Decree are not impacted by this rule making and remain fully acceptable.
In comments on the 2012 proposal, commenters (0192, 0193, 0194) stated that in regard to changes proposed for the performance test language in paragraph §63.1511(b)(1), requirements to conduct a performance test at "the highest production rate" and "with the scrap containing the highest level of contamination" are mutually exclusive and such a requirement would not be workable. The commenters noted that the highest furnace production rates or throughputs are obtained when processing large uniform pieces of aluminum. The commenters observed that scrap with this sort of geometry (small surface area to weight ratio) contains less "contaminants" such as oil, paint or other materials per furnace charge than a scrap consisting of many small pieces, while production rates or throughput are lower when processing smaller, non-uniform scrap because of a lower density which takes up more space in the furnace. 
Commenter 0197 recommended that performance tests for thermal chip dryers should be conducted under representative operating conditions. The commenter recommended that the EPA should clarify this point in the final rule by revising Proposed section 63.1511(b)(1) to state that chips should contain representative levels of contamination expected to be seen under typical unit operating conditions.
Commenters 0193, 0194 expressed concern that the EPA's revision to §63.1511(b)(1) could lead to the need for two or more performance tests on each furnace to cover multiple rates of production, even though worst case emissions can clearly be represented by a single test condition under the current regulatory requirement in §63.1511(b)(1), and could double compliance testing costs for companies without good cause. The commenters requested that either the EPA delete the change to section §63.1511(b)(1) or describe why this is change is necessary and provide an opportunity for the aluminum industry to comment on the EPA's rationale.
In a comment on the supplemental proposal, commenter 0295 stated that while they concur with the changes that the EPA has made in the testing condition requirements as found in section 63.1511(b)(1) versus what was included in the 2012 proposal, it appears that the parenthetical inclusion of the word "normal" in the first sentence is un-necessary and potentially confusing. The commenter stated that it is not clear what "normal' means in the context of testing conditions expected to produce the highest level of HAP emissions. According to the commenter, historically, facilities have been known to establish an upper limit of what emissions might be by testing under conditions that are outside "normal" operations; for example, a facility that normally melts 30 to 40 percent purchased scrap in a group 1 uncontrolled furnace might test at 100 percent purchased scrap to set an upper bound of potential emissions. The commenter stated that a facility might accumulate purchased scrap of lower than average acceptable quality in order to make up a test with dirtier than usual charge materials. The commenter stated that test results from these activities are legitimate if the emission limits are met, even though the charge is outside the "normal" range. The commenter recommended that the EPA remove the parenthetical "normal" from the text in section 63.1511(b)(1).
In a comment on the supplemental proposal regarding the EPA's modification to section 63.1512 to include a requirement for performance tests to be conducted "under representative (normal) conditions," commenter 0290 stated that it is not clear what this means in the context of worst-case testing. The commenter stated that in the past, facilities have established an upper limit of what emissions might be by testing under conditions that are outside routine operations. The commenter stated that test results are valid if the emission limit is met, even though the charge is outside the "normal" range; facilities have routinely used these and other means to achieve "worst case" conditions for a test. The commenter recommended that the EPA should remove the parenthetical "(normal)" and clarify that testing outside of the routine range of operations is allowed as long as it is representative of worst case.
Response: In the 2012 proposal, we proposed amendments to clarify that performance tests under multiple scenarios may be required in order to reflect the emissions ranges for each pollutant regulated under Subpart RRR. We received comments on the 2012 proposal that the worst case charge materials, and blends of these, have differing process rates and, therefore, the charge rate from the stack tests is not representative of the charge rate that will be achieved during normal operations. Based on the comments received on the 2012 proposal and recognizing that it may be necessary to conduct performance tests under one or multiple scenarios to be representative of the range of normal operating conditions, in the 2014 supplemental proposal we proposed revised language in 40 CFR 63.1511(b)(1) to clarify the conditions under which subpart RRR performance tests must be conducted. The intention in the subpart RRR rule is to require testing under ``worst case'' conditions from the standpoint of emissions and to establish parameters based on such testing that ensure compliance under all operating conditions. For example, in a response to comments on the original proposed subpart RRR rule regarding the inlet temperature requirement for fabric filters, the EPA stated that testing under worst case conditions, such as higher than normal fabric filter inlet temperatures, could provide a larger temperature operating range, which would be used to monitor and ensure continuous compliance between periodic performance tests (65 FR 15699, March 23, 2000). In the EPA response-to-comments document (Summary of Public Comments and Responses on Secondary Aluminum NESHAP, December 14, 1999, Docket No. A - 92 - 61, item V - C - 1, comment 4.1.47), the EPA explained that requiring multiple tests over a range of different furnace operating conditions will show that the selected monitoring parameters are valid indicators of emissions and that it may not be possible for a single test to be representative of worst case conditions and that more than a single test may be required. It is not permissible, for example, to demonstrate compliance while processing relatively uncontaminated scrap, and then at a later time, when the supply of this scrap is constrained, process more heavily contaminated scrap, without demonstrating compliance under these changed conditions based on previous emissions testing or on new emissions testing if previous tests would not be representative of the emissions from the processing of the more heavily contaminated scrap.
To clarify the requirements for testing, we proposed in the 2014 supplemental proposal that performance tests be conducted under representative (normal) conditions expected to produce the highest level of HAP emissions expressed in the units of the emission standards for the HAP (considering the extent of scrap contamination, reactive flux addition rate and feed/charge rate). If a single test condition is not expected to produce the highest level of emissions for all HAP, testing under two or more sets of conditions may be required, e.g., high contamination (or high reactive flux rate) at low feed/charge rate and low contamination (or low reactive flux rate) at high feed/charge rate. Other operating conditions may also affect the worst case test for one or more regulated HAPs. Any subsequent performance tests for the purposes of establishing new or revised parametric limits are allowed upon pre-approval from the permitting authority for major sources or the Administrator for area sources. These new parametric settings are used to demonstrate compliance for the period being tested. We solicited comment on whether the proposed amendment adequately addresses and clarifies the requirement that multiple tests may be necessary to represent different operational conditions.
After considering the comments summarized above, we are adopting the rule amendment as proposed in the 2014 supplemental proposal, with the following exception. We agree with the commenter's contention that the parenthetical "normal" could lead to confusion and differing interpretations of the requirements for testing under conditions expected to produce the highest level of HAP (considering the extent of scrap contamination, reactive flux addition rate, and feed/charge rate), i.e., "worst case" emissions. The final rule does not contain the parenthetical "normal."
Comment #2: One commenter (0199) recommended that the EPA should clarify that because stack tests are designed to measure worst-case emissions, that when determining area source status, the emission factor determined during the stack test should be multiplied times the actual measured production rate achieved when the worst case material was processed during the stack test in the calculation of potential emissions. The commenter explained that if this is not done and the emission factor determined from a stack test performed when worst case materials are processed is combined with the highest production rate achieved processing the easiest material, the projected level of emissions will be much higher than is possible and area sources may unfairly be made major sources.
Response: Area source status is generally determined by the potential-to-emit of the source (see definitions of area source and major source in 40 CFR 63.2). If the permitted annual production cannot be achieved while processing the highest-emitting type of material, then this may be taken into account in conducting an area source determination. In some cases, the capacity to process a greater throughput of a lower-emitting (on a per-unit-production basis) material may result in a higher potential-to-emit. Absent an enforceable provision limiting emissions (for example, a limitation of the annual usage of reactive flux, or annual operating hours) or a physical limitation affecting potential to emit, potential-to emit determinations are based on the allowable emission rate.
Comment #3: One commenter (0192) on the 2012 proposal agreed with the new language the EPA is proposing for 40 CFR §63.1511(b)(1), stating that the language in the rule is appropriately broad so that testing of new flux materials, increasing the allowed furnace capacity, baghouse operating parameters and other operating parameters of the furnace could also be included. The commenter recommended that the EPA should clarify in the preamble to the final rule that other parameters can be changed during a performance test.
Response: Based on the comments received on the 2012 proposal and recognizing that it may be necessary to conduct performance tests under one or multiple scenarios to be representative of the range of normal operating conditions, we proposed revised language in 40 CFR 63.1511(b)(1) in the 2014 supplemental proposal to clarify the conditions under which subpart RRR performance tests must be conducted. As we explained in the 2014 supplemental proposal, the intention is to require testing under "worst case" conditions from the standpoint of emissions and to establish parameters based on such testing that ensure compliance under all operating conditions. The final rule clarifies that performance tests be conducted under representative conditions expected to produce the highest level of HAP emissions expressed in the units of the emission standards for the HAP (considering the extent of scrap contamination, reactive flux addition rate and feed/charge rate). The final rule also allows additional performance tests for the purposes of establishing new or revised parametric limits upon pre-approval from the appropriate authority.
Comment #4: The commenter (0197) indicated that the EPA does not define "highest level of contamination" for scrap used in an initial performance test, so it is unclear what is intended and the requirement should be deleted in any event. The commenter indicated that process equipment could be put at risk of fire, explosion, or equipment damage since it was not designed to handle wet or heavily oiled chips. 
Commenter 0193 asked that the word "scrap" in §63.1511(b)(1) should be replaced with the words "feed/charge material" in that all aluminum containing material to be processed in a furnace may not be scrap. 
Commenters 0192, 0193, 0194 suggested that the phase "while an air pollution control device is operating" should also be deleted because all furnaces do not operate with an add-on control device.
Response: In response to the concern that process equipment could be put at risk of fire, explosion, or equipment damage if required to conduct a performance test using scrap it was not designed to handle, such as wet or heavily oiled chips, the rule does not require testing while processing a material that, because of safety considerations, would never be processed in the unit.
In the final rule, we have replaced the word "scrap" with the words "feed/charge" and have deleted the phrase "while an air pollution control device is operating" as suggested.
Comment #5: In a comment on the supplemental proposal, commenter 0294 stated that the EPA's supplemental proposal addresses concerns previously articulated in their comments on the 2012 proposal that relate to operating conditions during performance testing. The commenter stated that they support the EPA's proposed revisions to the initial performance testing requirements at section 63.1511(b)(1). 
Response: The EPA acknowledges the commenter's support of the rule requirements for "worst-case" emissions scenario testing. As noted elsewhere in the response to comments, the only change to section 63.1511(b)(1) after the supplemental proposal is the removal of the parenthetical "normal" from the first sentence of the paragraph (b)(1).
3.3 	Reporting
Comment #1: One commenter (0193) objected to the proposed requirement to submit performance test reports electronically, because according to the commenter, it will impose a significant added financial burden on owners and operators of secondary aluminum smelters. The commenter asserted that the requirement constitutes double reporting and confidential business information may be jeopardized. The commenter requested that the EPA remove the electronic reporting tool (ERT) language. 
Response: The EPA believes that any costs of using the ERT are reasonable and not burdensome for several reasons. The electronic reporting requirement does not add any additional testing or change the timing of when test reports are due. The requirement only states that reports are to be submitted electronically. The ERT is fully functioning and has been used successfully by industry for several years. It was designed to capture all of the information which is submitted in a typical test report and was designed with input from States and the stack testing industry. 
The EPA worked with State and local agencies as well as stack testing companies (who typically prepare test reports) to develop the ERT. The ERT was designed to accept data and information that is typically collected during a performance test. Some State and local agencies have begun accepting the ERT as their reporting mechanism and with experience we believe acceptance by States will increase. In some instances state and local agencies may want a hard copy of a test report. The ERT can generate a printed test report or export the report to a word processor for reformatting. This report can be submitted to the State. The EPA believes that electronic reporting is a more efficient way to collect, store, and access test data and has set up a retrieval system such that States can access files that have been submitted using the ERT. As more States adopt electronic reporting, we believe that the need for paper reports will be diminished.
The commenter is incorrect that CBI would be jeopardized as a result of electronic reporting. Facilities can submit CBI according to the EPA's CBI procedures as always. The EPA will not make CBI publicly available, except in accordance with 40 CFR Part 2. In cases where some information in a test report is CBI, that information can be submitted separately as CBI.

Comment #2: The commenter (0197) noted that the Proposed Rule includes requirements for the submittal of reports in the records section (Proposed 40 C.F.R. § 63.1517(c)). The commenter also noted that the records section requires the source to "maintain" files and records, no reporting is required (40 C.F.R. § 63.1517(a)  -  (b)) and the first sentence of proposed section 63.1517(c) contains a reference to the requirements in paragraph (b). The commenter indicated that it is not clear if paragraph (b) is in the records or reporting section. The commenter recommended that the proposed section 63.1517(c) should be moved to section 63.1516 which contains requirements for reporting, and the reference to requirements in paragraph (b) should be clarified.
Response: In the final rule, the reporting provision that was incorrectly placed in section 63.1517 is now in the records section 63.1516, which also corrects the commenter's concern with the reference to paragraph (b).
Comment #3: In a comment on the supplemental proposal, commenter 0301 stated that, rather than requiring just recordkeeping of important compliance data as the EPA proposes, the EPA should instead require electronic reporting of all key compliance data points, so that the EPA, state and local governments, and concerned community members can provide an additional check to assure compliance. The commenter stated that if a facility is creating a record, there is very little additional time needed simply to submit the record electronically, along with other reports, and the public-interest and deterrent value of doing so is substantial.
Response: While we agree that it would be preferable for all key compliance data be recorded and reported electronically, the EPA has not yet developed electronic reporting systems for all compliance data currently required by the EPA regulations. The EPA continues to develop these electronic reporting systems and will deploy them as they are ready.
3.5	Other

Comment #1: One commenter (0194) agreed with the EPA's clarification for the daily SAPU emission limit calculation. The commenter noted that the commenter's locations have actually followed the proposed approach since the NESHAP became effective in 2003.
 
Response: The EPA acknowledges the commenter's support.
Comment #2: In a comment on the supplemental proposal, commenter 0301 stated that they support the EPA's proposal for annual inspection and improvements to compliance requirements.

Response: The EPA acknowledges the commenter's support of the rule requirements that clarify inspection and compliance requirements.
4. Compliance Dates

Comment #1: In a comment on the supplemental proposal, commenter 0296 stated that the compliance schedule in the supplemental proposal is much more reasonable than the compliance schedule in the 2012 proposal, allowing facilities an adequate window of opportunity for any new technology additions to be researched, selected, purchase, financed and installed; it is also consistent with the 2 year compliance schedule in subpart ZZZZZZ.
Response: The EPA acknowledges the commenter's support of the compliance dates.

5. Startup, Shutdown, and Malfunction
5.1 	Removal of SSM Exemption
Comment #1: One commenter (0195) stated that the EPA does not have the authority to amend existing MACT standards to make them more stringent, as is being done by proposing the new SSM standards. The commenter argued that the EPA's authority under CAA section 112(d)(6) is to "review and revise as necessary (taking into account developments in practices, processes, and control technologies), emissions standards promulgated under this section no less often than every 8 years." The commenter observed that the EPA did not invoke its section 112(d)(6) authority to support the Proposed SSM Provisions, but even if it had, section 112(d)(6) does not provide broad authority to reconsider aspects of previously issued MACT standards unrelated to "developments in practices, processes, and control technologies." The commenter maintains that the EPA cannot simply revisit and redo a MACT determination long after it has been issued, as the EPA attempts to do with the Proposed SSM Provisions. The commenter argued that the EPA cannot change its mind about what standards are required to comply with CAA sections 112(d)(2) and (3), nor can it recalculate a MACT floor based on subsequent performance.
The commenter asserted that reassessing existing NESHAP that were based on the MACT floor and imposing more-stringent requirements, would be inconsistent not only with the statute's careful provision of technology-review and residual-risk authority to follow establishment of MACT standards, but also with Congress' desire for finality evident in the judicial review of provisions of CAA section 307(b), which allows challenges to MACT standards within 60 days of promulgation. 
The commenter stated the EPA would have to justify why the decisions reflected in the current standards are wrong and why the new standards meet the required criteria that the EPA must satisfy in issuing MACT standards under CAA sections 112(d)(2) and (3). 
The commenter argued against the EPA's interpretation of Sierra Club v. EPA, stating that the Sierra Club decision interpreted the NESHAPs General Provisions, not what the EPA may or may not include in source-category-specific MACT standards. 
The commenter stated that although the D.C. Circuit and the EPA referred to the provision vacated in the Sierra Club decision as an "exemption" from hazardous air pollutant standards during SSM events, other portions of the NESHAP General Provisions impose various requirements that apply to sources both during SSM events (including the obligation to minimize excess emissions) and in anticipation of and following SSM events (including requirements to prepare a plan to address SSM events and to report SSM events).
The commenter pointed out that opinions where the court was looking at source-category specific MACT standards have emphasized the need for those standards to recognize and accommodate higher emission levels that occur at times other than normal operations. See, e.g., Sierra Club v. EPA, 167 F.3d 658, 665 (D.C. Cir. 1999).
The commenter asserted that the Sierra Club decision did not say that the same emission limitations that the EPA has derived for normal operations must also apply during SSM events. The commenter claimed that although an open-ended exemption from any standard under section 112 is inconsistent with the Sierra Club holding that, for section 112 MACT standards, "there must be continuous section 112-compliant standards", Sierra Club does not preclude the EPA from applying different standards during SSM events than apply during normal operations. The commenter argued that the opinion acknowledges that section 302(k)'s "inclusion of [the] broad phrase" "any requirement relating to the operation or maintenance of a source to assure continuous emission reduction" in the definition of "emission standard" suggests that the EPA can establish MACT standards consistent with CAA section 112 "without necessarily continuously applying a single standard." 
The commenter stated that there is ample precedent for the EPA applying a different standard during SSM events. The commenter noted that language the D.C. Circuit considered dispositive in interpreting the EPA's standard-setting authority under section 112  --  the statement in the CAA section 302 definition of "emission limitation" and "emission standard" that it "limits the quantity, rate, or concentration of emissions of air pollutants on a continuous basis"  --  has been in the statute since 1977. The commenter observed that throughout that time, the EPA has not required sources to meet NSPS emission limitations under CAA section 111 established for normal operations during SSM events. 
The commenter noted that Congress enacted the "continuous basis" language in section 302(k) knowing that the EPA's emissions standards under section 111 exempted SSM periods. The commenter argued that the "continuous basis" language inserted in 1977 is related to whether sources should be allowed to use temporary or intermittent pollution control technologies which entailed temporary reductions in emissions when weather conditions were poor, not what emission limitations apply during SSM periods, nor the EPA's established practice of exempting excess emissions during SSM events from the performance standards applicable to normal operations. The commenter noted that the legislative history indicates Congress was aware that alternative emission limitations might at times be necessary, even though the emission limitations were established based on the capability of "continuous controls" like scrubbers. See, e.g., S. Rep. No. 94-717 at 78 ("It is recognized that the source controls may not be available to achieve the full reduction required of a particular source under particular circumstances. In such case, supplementary programs can and should be developed. But this flexibility occurs only after imposition of the continuous emission limitation.").
The commenter asserted that the Sierra Club decision did not address whether the EPA could use a "design, equipment, work practice or operational standard," as authorized under CAA section 112(h) and included in the definition of "emission limitation" and "emission standard" in CAA section 302(k), in lieu of a numerical emission limitation during SSM events. The commenter stated that the EPA told the Court that the General Provisions SSM exemption struck down in Sierra Club was not an alternative standard based on the work practice standard authority. The commenter noted that the EPA argued in that case that section 112(h) was irrelevant to its authority to exempt excess emissions during SSM events. 
 The commenter maintained that the EPA cannot use the Sierra Club decision as a justification for ignoring an inability of even the "best performers" to achieve during SSM events the emission limitations the EPA has established for normal operations. The commenter asserted that the approach the EPA is proposing would not establish "continuous section 112-compliant standards" that the Sierra Club decision concluded are required, as MACT emission standards must be "achievable." The commenter stated that an emission limitation that applies during SSM events does not meet the requirement of CAA section 112(d)(2) that "emission standards" under that section be "achievable" if in fact the EPA has not demonstrated that the limitation is "achievable" with available technology. The commenter argued that an emission limitation that applies during SSM events has not been demonstrated to be "achieved" by the best-performing 12% of units in the category under CAA section 112(d)(3) unless the EPA can show that those best performers actually meet that emission limitation during SSM events. The commenter asserted that the EPA has not demonstrated that the emission limitations in the existing MACT standards, if they also applied during SSM events, would comply with section 112 when applied to SSM periods.
The commenter argued that the courts have long recognized that a "technology based standard discards its fundamental premise when it ignores the limits inherent in technology." NRDC v. EPA, 859 F.2d 156, 208 (D.C. Cir. 1988). The commenter provided examples. The D.C. Circuit recognized, in Portland Cement Ass'n v. Ruckelshaus, 486 F.2d 375, 398 (D.C. Cir. 1973), a decision reviewing standards under CAA section 111, that "`start-up' and `upset' conditions due to plant or emission device malfunction, is an inescapable aspect of industrial life and that allowance must be made for such factors in the standards that are promulgated." In Essex Chem. Corp. v. Ruckelshaus, 486 F.2d 427, 432 (D.C. Cir. 1973), another section 111 case, the court held that SSM provisions are "necessary to preserve the reasonableness of the standards as a whole." In National Lime Ass'n v. EPA, 627 F.2d 416 (D.C. Cir. 1980), another case reviewing emission standards promulgated under CAA section 111, the court held that the CAA requirement that NSPS be "achievable" means that the standards must be capable of being met "on a regular basis," including "under most adverse circumstances which can reasonably be expected to recur," including during periods of SSM. 
The commenter cited cases in support of the commenter's argument related to the Clean Water Act. 
The commenter stated that Weyerhaeuser Co. v. Costle, 590 F. 2d 1011 (D.C. Cir. 1978), does not support the EPA's position. The commenter asserted that in that case, the court was discussing a "technology forcing" standard, rather than one, like MACT, that is to be based on what is already being "achieved" or has been demonstrated to be achievable. The commenter claimed that the SSM events that the EPA acknowledges are expected to occur at sources subject to the Proposed Standards are a far cry from the unusual "'uncontrollable acts of third parties,' such as strikes, sabotage, operator intoxication, or insanity" that the court was considering in the passage quoted by the EPA. The commenter stated that industry is not requesting that the MACT standards provide relief from numerical emission limitations during those unusual types of events. The commenter emphasized that the Weyerhaeuser decision came long before NRDC v. EPA, 859 F.2d 156 (D.C. Cir. 1988) which, as noted above, affirmed the need for an upset provision to address circumstances where compliance with effluent limitations is impossible through no fault of the permittee, and which endorsed Marathon Oil.
The commenter noted that the EPA acknowledges this precedent in the rule preamble, but suggests that "intervening case law such as Sierra Club v. EPA and the CAA 1977 amendments undermine the relevance of these cases today." However, the commenter maintained that the EPA does not explain why it believes that to be the case. According to the commenter, Sierra Club v. EPA did not address this precedent, and the 1977 CAA amendments if anything support the conclusion that emission standards need to deal with the inability of a source to meet the normal emission limitations during particular circumstances. The commenter argued that the National Lime Ass'n decision, which relies in part on the cases the EPA referenced in the preamble, and which directly addresses the need for emission limitations that address reasonably anticipated adverse circumstances, post-dates the Clean Air Act Amendments of 1977 by three years.
The commenter pointed out that the Sierra Club decision rejected the EPA's asserted discretion to not to impose any emission standard covering SSM periods. The commenter asserted that the Sierra Club decision expressly recognized that different standards, including non-numerical standards, may (and in some cases must) apply during non-standard operating conditions, such as SSM events. The commenter added that the statement in the majority opinion that "Congress gave no indication that it intended the application of MACT standards to vary based on different time periods" is contradicted by other statements in the opinion that a MACT standard need not continuously apply a single emission limitation, is dicta because that was not the situation presented by the challenged regulations and argued by the EPA, ignores the extensive case law about technology-based limitations referenced above, and does not in any event say that the CAA precludes the EPA from adopting different emission limitations that apply during SSM events.
One commenter (0195) disputed EPA's approach to malfunctions is consistent with CAA section 112 and is a reasonable interpretation of the statute. The commenter observed that the EPA says it "has determined that CAA section 112 does not require that emissions that occur during periods of malfunction be factored into development of CAA section 112 standards," which the commenter maintains the EPA makes little effort to justify. The commenter disagreed with the EPA's statement that "[t]here is nothing in CAA section 112 that directs the agency to consider malfunctions in determining the level `achieved' by the best performing or best controlled sources when setting emission standards," claiming that there is nothing in CAA section 112 that allows EPA to ignore malfunctions and set MACT standards based on a level of emissions that even best-performing sources only achieve part of the time. The commenter asserted that EPA's rationale for not factoring in malfunctions directly contravenes congressional intent that MACT floor standards be based on what the best sources actually achieve.
The commenter recommended that the EPA to use authority under CAA section 112(h) to prescribe alternative design, equipment, work practice or operational standards where it is not feasible to set or enforce a numerical emission limit. The commenter disagreed with EPA's view that applying the concept of "best performing" is inconsistent with a source experiencing a malfunction. The commenter asserted that there are work practices  -  such as monitoring of operating parameters to identify a malfunction and stopping or cutting back the process accordingly  -  that represent the best practices for minimizing emissions during a malfunction, and these best practices represent the "the maximum degree of reduction in emissions of the hazardous air pollutants...achievable...through application of measures, processes, methods, systems or techniques" and reflect "the emission control that is achieved in practice by the best controlled similar source[s]".
One commenter (0195) asserted that the EPA's historical definition of "malfunction" is not in any way binding on how EPA defines the events that the Agency will refuse to take into consideration when establishing MACT emission standards. The commenter stated that by acting as if it were binding on EPA standard-setting, EPA has proposed a perverse approach to establishing emission standards in which the more common a situation is that results in higher emissions than normal operations, the less likely it is that EPA will write emissions standards to take that situation into account -- either in the standards themselves or in the affirmative defense.
One commenter (0195) stated that the elimination of SSM provisions in the existing MACT standards would put operators in a Catch-22 situation: the NESHAP would no longer accommodate the excess emissions that EPA knows will occur regularly during normal plant operations, due to process variability and control equipment operations and availability, yet those same events would appear to be ineligible for the "affirmative defense" that EPA proposes as a means of addressing excess emissions that occur despite a source's proper design and operation, because the "affirmative defense" is only available for excess emissions that result from "infrequent" events that cannot be "foreseen...and planned for." The commenter asserted that for the EPA to finalize these proposed actions would be unreasonable, arbitrary and capricious, and not in any way required by the Clean Air Act or judicial precedent.
One commenter (0195) stated that EPA ignores the undisputed existence of malfunctions even at best-performing sources, and the EPA claims falsely that the best-performing sources "achieve" emission levels that they undisputedly do not achieve part of the time. The commenter disagreed with the EPA's conclusion that malfunction events are not representative of best-performing sources. See, e.g., Cement Kiln Recycling Coalition v. EPA, 255 F.3d 855, 862-865 (D.C.Cir.2001) (recognizing that there is variability in the performance of control technologies, which needs to be accounted for in establishing emission limitations based on the MACT floor); Sierra Club v. EPA, 167 F.3d 658, 665 (D.C. Cir. 1999). The commenter observed that one goal of best performing sources is to operate in such a way as to avoid malfunctions, but the commenter believes that that is all the more reason for the EPA to acknowledge that those sources experience malfunction events. 
One commenter (0195) asserted in regard to malfunction events, that by proposing MACT standards that EPA recognizes even the best performing existing sources cannot achieve part of the time, EPA is going beyond the MACT floor, yet without making the demonstrations that the statute and case law require the Agency to make in order to impose beyond-the-floor MACT standards. The commenter claimed that the EPA has not justified as reasonable the conditions that the EPA proposes to impose on sources during malfunctions in order to be excused from civil penalties "taking into consideration the cost of achieving such emission reduction, and any non-air quality health and environmental impacts and energy requirements...." 
The commenter stated that it is indefensible for EPA to acknowledge that malfunctions are inevitable, even for the best performing sources, and yet refuse to include emissions data representing malfunctions (if such data exist) in calculating the MACT floor, and then require that those MACT floor limitations be met even during malfunctions. The commenter asserted that the EPA is obligated to consider representative data reflecting emissions during malfunctions in its MACT floor calculations for steady-state operating conditions, if that data can be gathered. 
The commenter recommended that to the extent EPA had access to continuous monitoring data for emission units covered by the NESHAP, EPA could have conducted analyses of emissions levels during malfunction events. The commenter noted that even if the continuous monitoring data are for parameters not regulated by the Proposed Standards, analysis of monitoring data for those other parameters during malfunction events might form a reasonable basis for EPA's assessment of what standards are achieved or achievable during malfunctions. The commenter suggested that many types of sources are required by many state agencies to submit deviation reports or malfunction reports when they experience a malfunction that causes an exceedance of an applicable limitation. The commenter observed that the EPA does not appear to have made any attempt to obtain and analyze such reports, in order to assess what type of requirement might reasonably apply to the subject emission units during malfunctions.
The commenter offered several options for setting MACT standards under CAA section 112 that would apply during malfunction events:
::	Establish an emission limitation that applies at all times, but that has an averaging time of sufficient duration that short, infrequent spikes in emissions due to malfunctions would not cause the source to exceed the emission limitation (while at the same time ensuring that the source does not operate in a way that causes frequent, lengthy excursions above the normal controlled emission rate);
::	Use the flexibility accorded by CAA section 302(k) (defining "emission limitation" and "emission standard" to include "any requirement relating to the operation or maintenance of a source to ensure continuous emission reduction, and any design, equipment, work practice or operational standard promulgated under" the CAA) to address emissions during malfunction events through operational requirements rather than by applying the same limits on pollutant concentrations in exhaust gases that apply during normal operations; and/or
::	Promulgate under CAA section 112(h) a design, equipment, work practice, or operational standard, or combination thereof, because it is not feasible to prescribe or enforce an emission standard.
The commenter stated that the EPA did not consider any of these types of options, and the lack of evaluation of these options renders the Proposed Rule arbitrary, and requires EPA to develop a new proposal.
One commenter (0193) disagreed with the EPA's proposed approach for regulating malfunctions, stating that it is unjust and creates a legal dilemma for owners/operators to be accused of exceeding unachievable emission standards due to events outside their control. One commenter (0194) expressed concern that the EPA, recognizing that malfunctions are not all reasonably preventable, creates a legal dilemma for owners/operators to be accused of exceeding unachievable emission standards due to events outside their control.
For technology-based standards, a commenter (0197) noted that EPA and courts have recognized that EPA must either (a) consider SSM in setting the standards or (b) provide an upset defense. The commenter indicated that EPA should revise its SSM provisions. 
Commenter 0298 on the 2014 supplemental proposal (same as commenter 0195 on 2012 proposal) stated when EPA ignores emissions that even best-performing sources experience during malfunctions, EPA breaches its duty to set standards on the basis of real-world performance; it fails to take into account how sources actually operate and unlawfully prohibits emissions that cannot be avoided. See, e.g., Sierra Club v. EPA, 167 F.3d 658, 665 (D.C. Cir. 1999) (a best-performing source should not violate a standard supposed to be based on what it "achieve[s] in practice"); see also 76 Fed. Reg. 15628 (rejecting a comment as "inappropriate" because it "would result in emission limits that even the best performing sources would be expected to exceed"). The commenter stated that every court that has addressed the issue, both under the CAA section 111 requirement that new source performance standards be based on the best demonstrated control technology and under the similar technology-based requirements of CAA section 112(d) and (h), has indicated that EPA must take the effect of malfunctions on emissions into account when establishing such technology-based emission standards. See Portland Cement Ass'n v. Ruckelshaus, 486 F.2d 375, 398-99 (D.C. Cir. 1973); Essex Chemical Corp. v. Ruckelshaus, 486 F.2d 427, 432 (D.C. Cir. 1973); Nat'l Lime Ass'n v. EPA, 627 F. 2d 416, 431 (D.C. Cir. 1980); Cement Kiln Recycling Coalition v. EPA, 255 F.3d 855, 872 (D.C. Cir. 2001); NACWA v. EPA, 734 F.3d 1115, 1158 (D.C. Cir. 2013) The commenter stated that while EPA may be implying in the supplemental notice that the opportunity that a source would have to argue to a court in an enforcement action that a malfunction justifies a reduced or no penalty, is a satisfactory means of accommodating the effect of malfunctions on what standards are achievable with available technology, the statute directs EPA to set standards based on what the available technology can do or, in the case of MACT floor standards, based on what the best performers are actually doing. The commenter stated that nothing in the statutory language or legislative history indicates that Congress intended EPA to establish MACT standards that apply at all times, but that even best performers cannot meet during unavoidable malfunctions, and then have penalties potentially mitigated in an enforcement action. According to the commenter, courts agree with this analysis, stating repeatedly that relying on enforcement discretion is no substitute for setting technology-based standards that reflect the performance of that technology during all reasonably anticipated conditions. See, e.g., Portland Cement, 486 F.2d at 398 n.91; Nat'l Lime, 627 F. 2d at 431 n.46; Marathon Oil Co. v. EPA, 564 F.2d 1253, 1273 (9th Cir. 1977). The commenter stated that, moreover, the enforcement discretion approach does not satisfy the purpose of CAA section 112 of reducing HAP emissions; a post-hoc case-by-case evaluation of emission circumstances, as EPA intends to handle all malfunction periods, will not prevent or reduce emissions of HAPs. The commenter stated that rather than the starting point being affirmative steps to prevent emissions as Congress intended (including steps that would have been encouraged by the conditions specified in the affirmative defense), the starting point for the approach in the supplemental notice is the post-emission presumption of wrongdoing by the source operator.
Commenter 0298 stated that EPA has tools for accounting for malfunctions in the standards once the affirmative defense is eliminated. The commenter stated that the supplemental notice repeats EPA's conclusory statements in the proposed rule that it would be too difficult for EPA to consider malfunctions when establishing emission standards and that doing so would result in standards that are too lenient. According to the commenter, the court decision that EPA cites, Sierra Club v. EPA, 551 F.3d 1019 (D.C. Cir. 2008), specifically acknowledged that EPA has authority under the CAA to set different standards that apply at different times, so long as some CAA section 112 compliant standard applies at all times and EPA has used that authority in numerous rulemakings to set alternative emission standards that apply during startup and shutdown periods. The commenter stated that there is nothing in EPA's mandate to set emission standards allows EPA to decline to even assess whether it has sufficient data, or can obtain sufficient data, to set numerical limitations that apply during malfunction events. The commenter stated that EPA also can use sections 112(h) and 302(k) to substitute a design, equipment, work practice, or operational standard for a numerical emission limitation; the use of work practices will in most cases be more practicable and more comprehensive than trying to establish numerical standards to address all malfunctions. The commenter stated that one approach that has proven effective in the past is to specify the elements and objectives of a plan that the source must develop, follow, and review to assure that the source is taking the appropriate steps before, during, and after a malfunction to minimize excess emissions; it is likely that most if not all or the sources identified as best performers have such plans and follow them to responsibly address malfunctions. The commenter stated that EPA has already shown that it can use work practices when a source is operating in other than normal steady-state operation, when EPA established work practice standards in lieu of numerical emission limitations for numerous source categories that apply during startup or shutdown. The commenter stated that before EPA promulgates revised emission standards for secondary aluminum plants that apply even during malfunctions but that contain no affirmative defense or other provision addressing malfunctions, EPA should determine an appropriate set of work practices, or a combination of work practices and alternative numerical emission limitations, that address higher emissions associated with malfunctions of properly designed and operated MACT technology, and EPA must offer its analysis for public comment before finalizing the deletion of the affirmative defense and the adoption of the alternative standards.
Response: The EPA disagrees in many ways with the commenters, first in that the EPA does not lack authority to make the changes proposed. We specifically disagree with the commenter's suggestion that Sierra Club v. EPA, 551 F.3d 1019 (D.C. Cir. 2008) is not relevant because it addressed the SSM exemption in the General Provisions rather than source category-specific MACT standards. The holding in Sierra Club that emissions limitations under section 112 must apply continuously and meet minimum stringency requirements, even during periods of startup, shutdown and malfunction, is clearly applicable to source category-specific MACT standards. As we explained in the 2012 and 2014 proposals, our SSM-related rule revisions are in response to the Sierra Club court's vacature of the SSM exemption in §§63.6(f)(1) and 63.6(h)(1). 77 FR at 8598-8600. When incorporated into section 112(d) regulations for specific source categories, these two provisions exempted sources from the requirement to comply with otherwise applicable MACT standards during periods of SSM. Subpart RRR, Appendix A incorporated several General Provisions related to SSM, specifically including §§63.6(f)(1) and 63.6(h)(1). The court's vacature rendered those provisions null and void prior to this rulemaking. The mandate implementing the court's decision was issued on October 16, 2009, at which time the vacated SSM provisions were clearly no longer in effect, and standards in subpart RRR applied at all times, including periods of SSM. The rule changes to remove the vacated SSM provisions from subpart RRR are ministerial actions by the EPA to reflect the vacature by the court. 
In addition, in Medical Waste Institute v. EPA, 645 F. 3d 420, 425-27 (D.C. Cir. 2011), the D.C. Circuit Court held that EPA may permissibly amend improper MACT determinations, including amendments to improperly promulgated floor determinations, using its authority under section 112(d)(2)-(3). In light of the court decisions in Sierra Club and National Lime Ass'n v. EPA, 233 F.3d 625, 633-34 (D.C. Cir. 2000), the exemption from standards in Subpart RRR for HAP emitted during SSM was not proper. Similarly, the Court's decision in Portland Cement Ass'n v. EPA, 665 F.3d 177, 189 (D.C. Cir. 2011) confirms that although 112(d)(6) requires EPA to reassess its standards every eight years, nothing prohibits it from doing so more often, including revising existing floors if need be. The commenter is thus incorrect that EPA lacks authority to amend promulgated MACT standards or that section 112(d)(6) provides exclusive authority to address standards that apply during SSM. 
We are not adopting different standards for periods of startup and shutdown, although we do not claim that the Sierra Club decision constrains our authority to set different standards, including work practice standards under section 112(h), for SSM periods. Instead we are applying the existing Subpart RRR MACT standards to periods of startup and shutdown because the scrap processed at secondary aluminum production facilities is the source of emissions, and we expect emissions during startup and shutdown to be no higher, and most likely significantly lower, than emissions during normal operations since no scrap is processed during startup and shutdown periods. We stated in the preamble to the 2012 proposed rule that we knew of no reason why the existing MACT standards should not apply at all times, and we have received no information to change that conclusion. For additional responses to comments concerning standards during periods of startup and shutdown and alternative methods for demonstrating compliance with production-based limits, see the discussion in section 5.2 below.
The EPA also disagrees with the commenter's suggestion that the existence of an SSM exemption in rules implementing section 111 in 1977 when Congress enacted the "continuous basis" language in the definition of "emission standard" is evidence that Congress approved of that regulatory SSM exemption. We do not believe the legislative history cited supports that Congress was aware or approved of that exemption. 
The commenters assert that the EPA must account for malfunctions when determining the emissions level achieved by the best performing sources, but this view is neither reasonable nor supported by the text of the CAA. A malfunction is defined as a "sudden, infrequent, and not reasonably preventable failure of air pollution control and monitoring equipment, process equipment or a process to operate in a normal or usual manner * * *" (40 CFR 63.2). The EPA interprets CAA section 112 as not requiring emissions that occur during periods of malfunction to be factored into development of CAA section 112 standards. Under CAA section 112, emissions standards for new sources must be no less stringent than the level "achieved" by the best controlled similar source and for existing sources generally must be no less stringent than the average emission limitation "achieved" by the best performing 12 percent of sources in the category. There is nothing in section 112 that directs the Agency to consider malfunctions in determining the level "achieved" by the best performing sources when setting emission standards. As the D.C. Circuit has recognized, the phrase "average emissions limitation achieved by the best performing 12 percent of" sources "says nothing about how the performance of the best units is to be calculated." NACWA v. EPA, 734 F.3d 1115, 1141 (D.C. Cir. 2013). In addition, accounting for malfunctions in setting emission standards would be difficult, if not impossible, and could lead to standards that are not reflective of levels achieved by a well-performing non-malfunctioning source. The EPA's approach to malfunctions is a reasonable interpretation of the statute. If a source fails to comply with the applicable CAA section 112 standards as a result of a malfunction event, as discussed above, the EPA would determine an appropriate response, and the source can raise any and all defenses in any enforcement action that may be filed.
The EPA's approach both accounts for variability associated with a reasonably foreseeable range of operating conditions and recognizes that enforcement mechanisms can address emission exceedances due to unpreventable equipment or process failures. While commenters may seek greater accommodation for malfunctions, such accommodation is not compelled by the Act.
Unlike start-up and shut-down, which are foreseeable operations, no one can predict the nature, scope, severity, timing, length, number or likely recurrences of malfunctions a source may  -  or may not  -  experience. Although the EPA bases emission standards on different manners of operations and circumstances, the EPA only takes into account conditions that are "foreseeable" and "which can reasonably be expected to recur." Sierra Club v. EPA, 167 F.3d 658, 665 (D.C. Cir. 1999); see also Nat'l Lime Ass'n v. EPA, 627 F.2d. 416, 431 n.46 (1980). Virtually no malfunction is foreseeable in terms of its nature, timing or effect on emission levels and no operator knows if or when a malfunction could recur.
The commenters identify no language in the statute that even implies that EPA must factor into MACT emission standards the emission levels that occur during malfunctions. Yet the commenters assert that CAA section 112 mandates that EPA account for malfunctions in establishing every MACT standard or set a separate MACT standard for malfunctions, so that they reflect higher emission levels that might occur during undefined, unidentified malfunction events. The commenters appear to rely on the silence of section 112, which does not speak directly to how malfunctions are to be treated to support their view, filling the void with the commenters' own judgments of how EPA must address such uncertain and undefined events.
Although EPA may have authority to address malfunctions through different mechanisms  -  albeit limited by the Court in Sierra Club v. EPA (striking down exemptions for malfunctions) and NRDC v. EPA (striking down an affirmative defense for malfunctions)  -  EPA is not required by section 112 to account for malfunction emissions by resetting MACT standards (e.g., with long averaging times) or by setting separate MACT standards (e.g., work practice standards) as suggested by the commenter.
A commenter asserts that relying on case-by-case enforcement discretion to address violations caused by malfunctions does not satisfy the purpose of CAA section 112 of reducing HAP emissions because it will not prevent or reduce emissions. We disagree. Applying the existing Subpart RRR MACT emission standards at all times meets the Act's purpose in section 112 and, as enforceable emission standards, limits HAP emissions in accordance with the Act. The commenter states that enforcement discretion is no substitute for accounting for malfunctions in the standards themselves and suggests that different standards or requirements could be applied to malfunction events. As discussed above, we are not required to account for malfunction emissions in section 112 standards, and based on the information available to EPA concerning compliance by sources with Subpart RRR standards, we find no reason to do so.
A commenter contends that section 112 mandates that EPA establish MACT standards so that they reflect higher emission levels that might occur during unidentified, unmeasured malfunction events. To support its argument, the commenter states that emission limitations must be "achieved in practice". The commenter asserts that the EPA has failed to develop standards that are "achieved in practice" if the best performing sources are expected to have malfunctions, and those malfunctions result in emissions that exceed the standard. But a source that suffers worker malfeasance or experiences a hurricane or a terrorist event may also exceed emission standards at certain times. These events, like malfunctions, do not transform achievable standards into unachievable standards. Indeed, under the commenter's reasoning, every MACT standard is potentially invalid because it does not reflect emission levels that may occur during the next malfunction, act of God, strike, malfeasance or other unpredictable event.
Hurricanes and malfeasance can occur at well-maintained and well-managed sources and can cause upset conditions that result in violations of emission standards, but this does not warrant factoring such unpredictable events into revised emission standards. Even if malfunctions were inevitable for all sources, including the best-performing sources, that does not make it possible to take them into account when establishing MACT emission standards, because they are still unknown in frequency, length, magnitude and, most importantly, effect on emission levels.
The commenter asserts that the EPA inappropriately relies on the CWA case Weyerhaeuser Co. v. Costle, 590 F.2d 1011, 1058 (D.C. Cir. 1978) because the Weyerhaeuser court based its holding on the fact that the standards at issue there were technology-forcing. While the Court did address this point, it went on to uphold EPA's case-by-case enforcement discretion approach to addressing malfunctions because interjecting factors dependent on "murky determinations concerning the sequence of events in the plant, whether those events would have been avoidable" and similar factors, is inconsistent with the intent of Congress to require numeric emission standards that are measurable and easily enforceable. 590 F.2d at 1057-58. The commenter also argues that the Weyerhaeuser decision only addressed effects from "uncontrollable acts of third parties" which are "a far cry" from events that would occur under Subpart RRR. In fact, the Court in Weyerhaeuser was addressing "unusual events such as plant start-up and shut-down, equipment failures, human mistakes, and natural disasters." 590 F.2d at 1056.
The commenter implies that Weyerhaeuser was overruled by another Clean Water Act case, NRDC v. EPA, 859 F.2d 156 (D.C. Cir. 1988), citing it for the proposition that a technology-based standard must accommodate defects in technology. That case was decided in 1988, long before the court decisions that found an exemption and an affirmative defense for malfunctions were contrary to the requirements of the Act. Moreover, the discussion of technology-based standards in the NRDC decision was dicta. The holding in NRDC was that water quality based standards do not need to account for malfunctions despite sources' use of technology for compliance. Although Congress revised the HAPs program in 1990 to have technology serve as the core for establishing emission standards, those standards are also not strictly control technology-based but rather are based on the emission levels actually achieved in practice, which the Court has explained goes beyond what can be achieved solely through available control technology. See Sierra Club v. EPA, 479 F.3d 875, 882-83 (D.C. Cir. 2007).
The commenter also offers the decision in Cement Kiln Recycling Coalition v. EPA to support its argument that malfunctions must be considered in developing emissions standards. However, nothing in Cement Kiln even implies that EPA has a duty to reflect in its emission standards emission levels that might occur during unpredictable malfunction events. To the contrary, Cement Kiln cuts to the heart of commenters' argument. The fundamental principle of the commenters' argument is that due to malfunctions, the MACT emission standards, which have been achieved by secondary aluminum production facilities, may be "unachievable" if specific sources experience a malfunction (citing Cement Kiln Recycling Coalition v. EPA, 255 F.3d 855, 872 (D.C. Cir. 2001); Sierra Club v. EPA, 167 F.3d 658, 665 (D.C. Cir. 1999); Portland Cement Ass'n v. Ruckelshaus, 486 F.2d 375, 398-99 (D.C. Cir. 1973); Essex Chemical Corp. v. Ruckelshaus, 486 F.2d 427, 432 (D.C. Cir. 1973); Nat'l Lime Ass'n v. EPA, 627 F. 2d 416, 431 (D.C. Cir. 1980); NACWA v. EPA, 734 F.3d 1115, 1158 (D.C. Cir. 2013)). The Court in Cement Kiln, however, rejected the notion that because a MACT standard may not be achievable by some sources, it is invalid. Cement Kiln, 255 F.3d at 861 ("EPA may not deviate from section 7412(d)(3)'s requirement that floors reflect what the best performers actually achieve by claiming that floors must be achievable by all sources using MACT technology."). See also Sierra Club v. EPA, 479 F.3d at 878; NACWA, 734 F.3d at 1150 (explaining that the argument that a MACT standard is invalid because it is not achievable by all sources at all times has been "roundly rejected" by the Court)."
Comment #2: One commenter (0195) argued that the EPA's proposal to replace §63.6(e)(1) with the somewhat different "general duty" language in proposed sections 63.1506(a)(5) and 63.1520(a)(8) is not being proposed under CAA section 112(d)(6) or 112(f), and the EPA therefore lacks authority to make this change to the existing RRR NESHAP. The commenter disagreed with the EPA's statement that the proposed sections 63.1506(a)(5) and 63.1520(a)(8) replace General Provisions requirements that reference vacated SSM provisions. The commenter stated that the D.C. Circuit did not vacate the provision the EPA proposes to replace, §63.6(e)(1), in Sierra Club v. EPA (which vacated the "exemption" for SSM events in sections 63.6(f)(1) and (h)(1), not the requirement for good air pollution control practices), and, contrary to the EPA's assertion in the preamble to the Proposed Rule, §63.6(e)(1) does not reference provisions that were vacated in Sierra Club v. EPA.
The commenter maintained that even if the EPA had authority to change the existing MACT standards in ways not required to address residual risk or new technology, the Agency would have to provide a cogent explanation of why the old rule was unacceptable and the new rule is necessary, which the commenter claimed the EPA has not done.
The commenter recommended that the EPA delete proposed section 63.1506(a)(5) from the final rule, but if the EPA does not delete the proposed section, the commenter recommended that the EPA re-propose the provision with some explanation of the basis and purpose for the provision, to allow the public an opportunity to provide meaningful comments, as required by CAA section 307(d)(3).
Response: The EPA disagrees with the commenter, in that the EPA does not lack authority to make the changes proposed (see response to comment #1 above). The EPA explained in the 2012 proposal that we were proposing to add language to this regulation to replace General Provision requirements that reference vacated SSM provisions. 77 FR at 8597, 8598-8600. We are revising the general duty language because some of the language in that section is no longer necessary or appropriate in light of the elimination of the SSM exemption. Specifically, the prior language in 63.1506(a)(5) and 63.1520(a)(8) characterized what the general duty entails during periods of SSM. With the elimination of the SSM exemption, there is no need to differentiate between normal operations, startup and shutdown, and malfunction events in describing the general duty. Sources will be subject to emission standards at all times, including during periods of startup and shutdown and malfunction, and other language in the final rule already makes clear that the general duty to minimize emissions applies at all times. Therefore, there is no need to specify that malfunctions need to be corrected as soon as possible or to further specify the application of the general duty requirement during periods of SSM. Our notice of proposal was sufficient on this topic and the rule revisions do not need re-proposal. 
Comment #3: One commenter (0195) maintained that even if the EPA could legally go back and change the existing MACT standards, it could not reasonably make those revised standards effective immediately. According to the commenter, sources subject to the Proposed Rule would be expected to comply with revised SSM requirements as soon as the rule becomes effective. The commenter argued that it would be arbitrary and capricious for the EPA to require compliance immediately, when CAA section 112(i) allows a compliance deadline of up to three years. The commenter observed that elimination of the SSM provisions in the existing rules could require plants to make significant changes to their production processes or their HAP collection and control systems, which would take significant time for design, engineering, acquisition, and installation. 
Response: We disagree that we lack justification for making rule changes to remove the vacated SSM provisions effective upon promulgation. As discussed above, the D.C. Circuit Court's vacature of SSM provisions in its Sierra Club decision rendered those provisions null and void prior to this rulemaking. EPA issued a letter dated July 22, 2009 from Adam Kushner, Director, Office of Civil Enforcement, to various parties that addressed the impact of the decision in Sierra Club v. EPA, 551 F.3d 1019 (D.C. Cir. 2008). The mandate implementing the court's decision was issued on October 16, 2009, at which time the vacated SSM provisions were clearly no longer in effect, and standards in subpart RRR applied at all times. The rule changes to remove the SSM provisions from subpart RRR are ministerial actions by EPA to reflect the vacature by the court. In addition, amendments to subpart RRR were proposed on February 14, 2012, at which time secondary aluminum plants were put on notice of the specific wording changes to subpart RRR. Thus, facilities have had ample notice of the SSM rule changes.
5.2 	Startup and Shutdown
Comment #1: Commenters (0193, 0194) suggested that the entire paragraph at §63.1516(b)(1)(v) in the current Subpart RRR rule needs to be deleted. The commenters noted that this paragraph refers to actions taken under a startup, shutdown and malfunction (SSM) plan, and the proposed rule removes SSM Plan requirements. 
Response: The commenter is correct and this paragraph has been deleted.
Comment #2: Commenters (0193, 0194, 0197) stated that the EPA has not evaluated or justified applicability of the existing MACT emission limits to startup, shutdown or malfunction periods. The commenters noted that the current MACT limits are based on performance tests conducted during normal process operation, not during SSM periods. Commenter 0192 noted that current work practices used by the industry during startup and shutdown of batch-operated furnace are adequate and regulatory requirements are not necessary.
Commenters (0192, 0199) asserted that the EPA's 2012 proposal to use the feed rate determined in the performance test to calculate production normalized emission rates during startup and shut down is inappropriate. The commenters explained that during startup and shutdown periods, little or no amount of aluminum is in a batch-operated conventional melting furnace, and thus no aluminum smelting-related emissions. The commenters observed that the zero production number during those startup and shutdown periods creates mathematical problems when calculating pounds per ton or micrograms per megagram production normalized emissions.
One commenter (0195) stated that the commenter's members do not have sufficient information to comment on whether in fact the emissions during startup and shutdown of the sources subject to the 2012 proposed rule will be no greater than emissions during normal operations, but expressed the opinion that the EPA has not adequately justified this conclusion. The commenter cited the possibilities that a control device may be less efficient until it reaches its design operating temperature, it may be less efficient when the pollutant concentrations in the gases to be treated are lower than during steady-state operation, or until a manufacturing process reaches steady-state operation, that process may generate substantially higher emissions, either on a total-mass basis or on a mass-per-unit-of-production basis.
The commenter suggested that the EPA conduct a thorough analysis and determine whether it is representative of the performance of best-performing sources (the MACT floor) to require facilities in these source categories to achieve the same emission limitation during startup and shutdown as during normal operations, and that the default assumption be that such special provisions during startup and shutdown are needed. The commenter asked that the EPA provide a factual analysis supporting the EPA's new conclusion that MACT standards can be achieved as well during all startup and shutdown periods.
The commenter expressed skepticism about whether the EPA conducted a sufficient analysis to demonstrate the achievability of the MACT emission limits established for normal operations during startup and shutdown events, because of likely limitations in the available data. The commenter stated that because of the difficulties in collecting reliable data during startup and shutdown periods, the EPA may have understated startup and shutdown emissions.
The commenter noted that it can be unclear whether a condition that leads to excess emissions should be characterized as a startup or shutdown event or a malfunction event, and the EPA may be analyzing data sets that exclude events that were treated as malfunctions but should be included in calculating average performance as startup or shutdown conditions.
Response: The commenters suggested that the EPA did not adequately analyze the ability of affected sources to meet Subpart RRR MACT emissions limits during startup and shutdown, but the EPA did provide an explanation of the basis for its approach in the 2012 proposal, including the EPA's conclusion that emissions during startup and shutdown will be low, and the commenters did not submit specific information to indicate that different standards are justified or to explain specifically why subpart RRR MACT standards are overly stringent during startup or shutdown. This failure to provide specific information to support their claims of unachievability is especially notable because the Subpart RRR MACT standards have already applied at all times, including periods of startup and shutdown, since the SSM exemptions incorporated into Subpart RRR were vacated in the Sierra Club decision. The EPA disagrees with the suggestion that there is a "default assumption" that "special provisions during startup and shutdown are needed". The commenters also do not identify or justify any specific different standards that should apply during startup and shutdown. 
For secondary aluminum production processes where the MACT standards are expressed in units of pounds per ton of feed or similar units (i.e., thermal chip dryers, scrap dryer/delacquering kiln/ decoating kilns, dross-only furnaces, inline fluxers using reactive flux and group 1 furnaces), the 2012 proposal included (and this final rule includes) an alternative method for demonstrating compliance with those limits based on emissions measured during startup and shutdown. In response to the commenters' concern that the absence of feed/charge during startup and shutdown will result in a zero production and create mathematical problems when calculating production-based emissions, the final rule specifies that the feed/charge rate used in the calculation come from the most recent performance test associated with a production rate greater than zero, or the rated capacity of the affected source if no prior performance test data are available.
In response to comments on the 2012 proposal and because conducting meaningful emissions testing during periods of startup and shutdown can be problematic, we proposed in the 2014 supplemental proposal (and are finalizing in this rule) an additional alternative method for demonstrating compliance with production-based subpart RRR emission limits during periods of startup and shutdown. As we explained in the 2014 supplemental proposal, these alternative compliance provisions reflect the MACT requirement for these periods. Specifically, compliance with production-based emission standards during startup and shutdown can be demonstrated by keeping records that show that the feed/charge rate was zero, the flux rate was zero and the affected source or emission unit either was heated with electricity, propane or natural gas as the sole sources of heat or was not heated. This alternative compliance method was based on the recognition that the source of HAP emissions is the processing of scrap and the use of fluxes during processing and that the heat for processing in this source category is generated exclusively by use of clean fuels -- natural gas, propane or electricity. The following records must be kept for sources that use this alternative method of compliance: the date and time of each startup and shutdown, the quantity of feed/charge and flux introduced during each startup and shutdown and the types of fuel used to heat the unit during startup and shutdown.
Section 112(h) allows the EPA to set work-practice standards when "the application of measurement methodology to a particular class of sources is not practicable due to technological and economic limitations." Section 112(h) "allows the EPA to substitute work practice standards for emission floors only if measuring emission levels is technologically or economically infeasible." Sierra Club v. EPA, 479 F. 2d 875, 880 (D.C. Cir. 2007). We specifically sought comment in the 2012 proposal on whether work practices should be applied for processes subject to production-based limits (i.e., standards expressed in units of pounds per ton of feed or similar units). We asked any commenters who supported the application of work practices to explain how the requirements of CAA section 112(h)(2) are met and to identify work practices that would be effective in limiting HAP emissions during periods of startup and shutdown for such processes. We did not receive and do not have information, from commenters or otherwise, concerning the impracticability of applying measurement methodology that would be necessary to prescribe work practice standards for periods of startup and shutdown in lieu of numerical emission standards contained in Subpart RRR.
We have reviewed the information available to us regarding startup and shutdown periods. Prior to the 2012 proposal we conducted an industry survey for emissions data including a request for information on emissions during startup and shutdown, but no such information on startup and shutdown emissions was received. A commenter notes the limited data available to the EPA, but neither this nor any other commenter submitted additional data on startup and shutdown emissions. Because scrap and reactive fluxing are the source of HAP emissions, we believe that startup and shutdown emissions will be very low and are not likely to cause a violation of the standards. We found no evidence that suggests that emissions are higher during startup or shutdown that might indicate a need for a different standard for these periods. The commenters have not provided any information to support a conclusion that sources cannot comply with the standards as proposed, and have not provided information to support development of different standards for startup and shutdown periods. Contrary to a commenter's argument, we believe that the Subpart RRR MACT standards provide "continuous section 112-compliant standards".
As we stated in the 2012 and 2014 proposals, the scrap processed and reactive fluxing used at secondary aluminum production facilities is the source of emissions. For that reason, we continue to expect that emissions during startup and shutdown will be no higher and probably much lower than emissions during normal operations since no scrap will be processed, even if, as the commenter suggests, control equipment is operating at less than optimal efficiency. We know of no reason why facilities should not be able to comply during periods of startup and shutdown. We recognize that there is no charging during startup and shutdown periods. For that reason we proposed and are finalizing two alternative methods for demonstrating compliance with production-based emission limits during periods of startup and shutdown.
Thus, we set standards based on available information as directed by section 112. The Subpart RRR standards that apply during normal operation will also apply during startup and shutdown.
In response to comments received, we proposed in the 2014 supplemental proposal and are finalizing, with some revisions, definitions of startup and shutdown. See Comment #3 below.
Finally, as in 40 CFR 63.7(e)(1), we are requiring in 40 CFR 63.1511(b)that performance tests conducted under this subpart not be conducted during malfunctions because conditions during malfunctions are often not representative of normal operating conditions. The EPA is adding language in 40 CFR 63.1517(b)(19) that requires the owner or operator to record the process information that is necessary to document operating conditions during the test and include in such record an explanation to support that such conditions are representative of startup and shutdown operations.
Comment #3: One commenter (0197) stated that startup and shutdown periods are not defined in the rule, but rather only in the preamble to the proposed rule. The commenter supported the EPA's definitions of startup and shutdown periods but, to reduce uncertainty, urged the EPA to include these definitions in the rule. 
In a comment on the supplemental proposal, commenter 0288 stated that the definition of "startup" and "shutdown" were problematic as applied to reverberatory furnaces. The commenter stated that the proposed definition of "shutdown" appears to require that a group 1 reverberatory furnace be tapped empty before shutdown can be said to have begun. The commenter stated that the implication is that emissions testing should include tapping the furnace until empty. According to the commenter, group 1 reverberatory sidewell furnaces cannot be tested under this definition, which would require the furnace cycle to include a complete drain and refilling. The commenter stated that filling a reverberatory furnace from an empty condition can take 12 hours or more depending on furnace size and design. According to the commenter, this is not a representative (normal) condition when operating a reverberatory sidewell furnaces, which operates with a heel of molten metal at all times during normal operation. The commenter stated that a common practice when starting a reverberatory sidewell furnace from an empty condition is to use clean charge and no reactive flux until the molten heel is raised above the arch between the sidewell and main hearth; once this arch is covered, normal charging and fluxing practices can begin. 
The commenter stated that a group 1 rotary furnace can be tested beginning empty and finishing empty. Although this does not represent every furnace cycle it does represent one of the normal operating modes for this furnace type. The commenter suggested that, to accommodate normal operating procedures for both reverberatory and rotary group 1 furnaces, the definition of "shutdown" be revised to the following: 
      "...that begins when the introduction of feed/charge is halted with the intention of removing all product and the intention of allowing the equipment to reach ambient temperature."
In a comment on the supplemental proposal on the definition of "startup," commenter 0288 referred the EPA to their comment on the definition of "shutdown," and added that a reverberatory sidewell furnace, from a cold start, must be heated to normal operating temperatures while feed/charge is added. The commenter stated that this feed/charge normally is made up of clean charge and no reactive flux and once the furnace heel is at normal operating levels (above the arch), normal charging can proceed. The commenter suggested the definition of "startup" be revised to the following:
      "... Startup ends at the point that feed/charge other than clean charge is introduced."
Commenter 0288 stated that the startup record required in paragraph 63.1517(b)(18) should include either identifying all materials charged to the furnace during startup or certification that only clean charge and no reactive flux was used until the arch between the sidewell and main hearth was covered. The commenter recommended that 63.1517(b)(18) be revised to the following:
      "For each period of startup or shutdown for which the owner or operator chooses to demonstrate compliance for an affected source based on a feed/charge rate, other than clean charge, of zero, a flux rate of zero, ..."
Commenter 0288 stated that for the same reasons, the type of charge along with quantity should be recorded. The commenter suggested that paragraph 63.1517(b)(18)(ii) be revised to the following:
The type and quantity of feed/charge and quantity of flux introduced during each startup and shutdown, and ..."
In a comment on the supplemental proposal with regard to the definition of "shutdown," commenter 0289 requested a clarification with how the definition relates to rotary tilt-type reverberatory furnaces, which operate on a batch-cycle. The commenter stated that, as proposed, the definition would classify the period between batches of this type of furnace, however brief and regularly scheduled, as a shutdown. The commenter stated that because this has implications for secondary aluminum processing units (SAPUs), and how emissions from members of a SAPU are counted, clarification as to whether the time between batches is a shutdown is appropriate.
In a comment on the supplemental proposal, commenter 0290 stated that they disagree with the proposed definition of shutdown in the supplemental proposal as it does not describe a furnace shutdown. The commenter stated that when a furnace is shutdown it is taken to a cold condition. The commenter stated that many melting and holding furnaces are operated batch-wise and between batches the furnace is still in operation and temperature is being maintained; the furnace is in a stand-by condition until the next batch is started. The commenter stated that the definition as proposed would apply between every batch and trigger burdensome record keeping requirements without any additional environmental benefit.
In a comment on the supplemental proposal, commenter 0294 stated that the typical operation of emission units within this source category are, for the most part, batch operations, i.e., feed/charge is introduced, feed/charge is halted, feed/charge previously fed into the emission unit is processed, and all of that feed/charge is subsequently removed from the emission unit; then another batch is begun and the cycle repeats itself. The commenter stated that as proposed, an affected facility would be shutting down numerous times daily as part of its normal batch operating mode; the commenter does not believe it was the EPA's intent to include in the definition of shutdown these periods of normal operations or to create unnecessary recordkeeping and reporting burdens for owners and operators. The commenter stated that they agree that the halting of feed/charge and the removal of product from the emission unit are aspects of a shutdown; however, there are other key aspects of a shutdown that must be incorporated into the definition; those aspects are the discontinuation of heat input and the reduction of the functional temperature of the emission unit. The commenter recommended that the EPA redefine "shutdown" to mean 
      "...the period of operation that begins when the introduction of feed/charge is intentionally halted, all product has been removed from the emission unit (e.g., by tapping a furnace), the source of heat is removed from the emissions unit, and a shutdown concludes when the emission unit has cooled to near ambient temperatures."
In a comment on the supplemental proposal, commenter 0295 stated that the proposed definition of a furnace shutdown in not accurate in that it would consider furnaces to be `shutdown' in situations when they are in stand-by mode in between processing batches of metal with the furnace still in operation, the temperature maintained at or near normal operating levels, and any applicable air pollution control equipment operating. The commenter recommended that the definition be revised to indicate that `shutdown' is a situation where the furnace temperature is brought to ambient after the metal charge has been removed to the greatest extent practicable as follows:
      "Shutdown means the period of operation ... that begins when the introduction of feed/charge is halted, the heat sources have been turned off, and all product has been removed from the emission unit to the greatest extent practicable (e.g., by tapping a furnace). Shutdown and ends when the emission unit is at or near ambient temperature."
In a comment on the supplemental proposal, commenter 0290 stated that the EPA correctly recognizes in the definition of startup that a shutdown furnace is in a cold condition by including the phase "warming from a cold start or a complete shutdown". The commenter recommended that the EPA add the phrase "the burners are turned off and the furnace is being taken to a cold state" to the proposed definition of "shutdown."
In a comment on the supplemental proposal, commenter 0289 stated that in the proposed definition of "startup," startup "begins with equipment warming up from a cold start or a complete shutdown." The commenter stated that there is no explanation for what constitutes a "complete" shutdown as opposed to the proposed definition of "shutdown" or how it differs from a "cold start." The commenter requested a clarification to forestall interpretation issues.
In a comment on the supplemental proposal, commenter 0295 stated that they have concerns with the definition of `startup' and recommended that the definition of startup be revised as follows:
      "Startup means the period of operation ... that begins with equipment warming from at or near ambient temperature. Startup ends at the point that flux or feed/charge is introduced."
Response: In response to comments on the 2012 proposal, the EPA included definitions of startup and shutdown in regulatory text for the 2014 supplemental proposal. We reviewed the comments on the supplemental proposal including those expressing concerns over the proposed definitions of "startup" and "shutdown," and the final rule addresses those concerns. In particular we revised the definitions to account for the fact that many furnaces are batch operations and are often in a standby condition that, under the proposed definitions, might have been considered to be shutdown. Also, we agree that the definition of "shutdown" should recognize that the beginning of shutdown includes halting of feed/charge, turning off the heat sources, and removal of product to the greatest extent practicable, and shutdown ends with cooling of the equipment to near ambient temperature. By modifying the definition of shutdown to begin after the removal of the heat source and to include cooling of the emission unit to near ambient temperature, the final rule distinguishes emission units that are shutting down from those that are in standby mode. The definition of startup also includes the introduction of flux as an end to startup. The final definition also recognizes that, after tapping, most furnaces (tilting furnaces are an exception) retain a molten metal heel and are not emptied completely. Regarding the definition of "startup," we disagree with commenter 0288's suggestion to revise the definition to state that startup ends with the addition of feed/charge "other than clean charge." The result of such a change would be that group 2 furnaces would never exit the startup mode. The commenter has not identified, nor do we envision any negative effects on sidewell furnaces if the definition of "startup" results in startup ending before the molten metal level reaches the top of the arch between the sidewell and the hearth. The final rule defines "shutdown" and "startup" for this source category as follows:
 Shutdown means the period of operation for thermal chip dryers, scrap dryers/delacquering kilns, decoating kilns, dross-only furnaces, group 1 furnaces, in-line fluxers, sweat furnaces and group 2 furnaces that begins when the introduction of feed/charge is intentionally halted, the source of heat to the emissions unit is turned off, and product has been removed from the emission unit to the greatest extent practicable (e.g., by tapping a furnace). Shutdown ends when the emission unit is near ambient temperature.
 Startup means the period of operation for thermal chip dryers, scrap dryers/delacquering kilns, decoating kilns, dross-only furnaces, group 1 furnaces, in-line fluxers, sweat furnaces and group 2 furnaces that begins with equipment warming from a shutdown, that is, the equipment is at or near ambient temperature. Startup ends at the point that flux or feed/charge is introduced.
Comment #4: One commenter (0197) indicated that the startup and shutdown emission calculation provision is confusing, may result in over reporting of emissions, and should be revised. The commenter indicated that the procedure is not a calculation of emissions, but rather an emission rate, and that according to the EPA's definition, there is no product (aluminum chips) in a unit during startup and shutdown periods and would therefore yield zero emissions. The commenter indicated that, with the exception of scrap shredders, emissions from sources covered by this subpart during these periods can be concluded to include only the byproducts of combustion. The commenter stated that these emissions, particularly in the case of dioxins and furans, should be negligible and may be accurately estimated using AP-42 emission factors or an engineering evaluation. The commenter recommended that the EPA should clarify the wording of the emission calculation equation and also allow facilities the option to demonstrate compliance with the standards through the use of AP-42 emission factors or an engineering evaluation. The commenter remarked that the EPA should not finalize a rule that inappropriately assumes emissions during startup are based on worst case emissions calculated during a performance test.
Response: We have revised the startup and shutdown emission calculation provisions in section 63.1513(f)(2) of the final rule to clarify how the owner or operator would be able to demonstrate compliance for those affected sources subject to an emissions limit expressed in units of pounds per ton of feed, or ug TEQ or ng TEQ per Mg of feed. As a further alternative to measuring emissions during startup and shutdown and dividing by the feed rate from the most recent performance test associated with a production rate greater than zero (or the rated capacity of the affected source if no prior performance test data is available), we are also allowing that records of feed or charge to the furnace may be used to demonstrate that the furnace is in compliance with the emissions limits during startup and shutdown. Because the source of HAP emissions is the feed or charge material and reactive flux material, emissions during startup and shutdown should be appreciably lower than the emissions limits. The EPA is, therefore, allowing sources to use recordkeeping to demonstrate compliance with the standards during startup and shutdown as an alternative to stack testing. The owner or operator would keep records showing that during startup and shutdown, the feed/charge rate was zero, the flux rate was zero and the affected source or emission unit was heated with electricity, propane or natural gas as the sole sources of heat or was not heated, although by definition, if feed or charge is introduced into the process, the affected source is not operating in a startup or shutdown mode and the requirements for normal operations would apply. 
Comment #5: In a comment on the supplemental proposal, Commenter 0294 stated that their standard safe work practices and standard operating procedures do not allow reactive flux or feed/charge materials to be added to furnaces, in-line flux boxes and other affected sources during a startup; for this reason, startup emissions are negligible and well below MACT emission limits. The commenter stated that documenting these work practices and procedures for every startup is an unnecessary recordkeeping burden and provides no environmental benefit. The commenter stated that all of their affected sources use either natural gas or electricity for heat and keeping a record of the types of fuel used to heat the unit for every startup also adds no value. The commenter recommended that the EPA should allow these facilities the option to submit an annual compliance certification stating that no flux and or feed/charge material was added and no heat source other than electricity, propane or natural gas was used during startup of the affected source. The commenter stated that the proposed records at §63.1517(b)(18) make no sense for shutdowns because, by definition, a shutdown begins when heat is removed and the introduction of feed/charge is halted; so, if there was no "lack of heating" or the flux rate and charge rate were greater than "zero", the source would not be in "shutdown" by definition. The commenter recommended that the proposed records at section 63.1517(b)(18) be deleted and replaced with a compliance provision in section 63.1516 (b)(2) that reads:
      "For each affected source choosing to demonstrate startup compliance based on §63.1513(f)(1): During each startup, no reactive flux and no feed/charge material was added to the emission unit and electricity, propane or natural gas was used as the sole source of heat." 
Response: We agree with the commenter that where a facility adheres to standard practices and procedures during startup and shutdown that exclude the addition of flux and feed/charge materials to emission units or the use of heat sources other than electricity, propane or natural gas, or is not heated, requiring the facility to document these practices for every startup and shutdown would be unnecessary and burdensome where there is an alternative method for documenting compliance during startup and shutdown. While we agree that documenting the work practices and procedures for startups/shutdowns under the conditions described can be reduced, we don't agree that annually is sufficient and are requiring documentation every 6 months. In the final rule, section 63.1517(b)(19) gives sources that choose to demonstrate compliance based on the startup and shutdown recordkeeping provisions of section 63.1513(f)(1), the option of submitting a semiannual report in accordance with section 63.1516(b)(2) of the final rule certifying that during startup and shutdown , no flux and no feed/charge were added to the emission unit, and electricity, propane or natural gas were used as the sole source of heat or the emission unit was not heated.
Comment #6: In a comment on the supplemental proposal, commenter 0298 stated that the proposed definitions that would delineate periods of startup and shutdown, during which alternative requirements would apply, as well as an alternative method for determining compliance are directionally consistent with portions of their comments on the 2012 proposal that urged the EPA to look more carefully at the Agency's assumption that facilities could meet limits based on normal operations during startup and shutdown just because less raw materials and fuel would be used during those periods.
Response: The EPA acknowledges the commenter's support of the rule requirements for demonstrating compliance during periods of startup and shutdown. See above for responses to specific comments on the definitions of startup and shutdown and alternative methods for demonstrating compliance with production-based emission limits during startup and shutdown.
Comment #7: One commenter (0195) stated that the EPA in the 2012 proposal has not justified adding proposed §63.1511(a) and eliminating reference to the NESHAPs General Provision on representative performance testing, §63.7(e)(1). The commenter asserted that the D.C. Circuit did not vacate the provision the EPA proposes to replace, §63.7(e)(1), in Sierra Club v. EPA, and according to the commenter, §63.7(e)(1) does not refer to or rely on provisions that were vacated in Sierra Club v. EPA. The commenter argued that the Sierra Club decision does not authorize the EPA to (1) change existing MACT standards so that the EPA, in turn, could (2) direct a source to conduct performance testing during abnormal operations, such as startups, shutdowns or malfunctions.
The commenter argued that requiring sources now to demonstrate compliance with numerical emission limitations in the existing MACT standards, using performance tests or continuous monitoring, even while the source is in startup or shutdown mode or is experiencing a malfunction, definitely has the effect of making those existing MACT standards more stringent. The commenter asserted that the EPA cannot simply "amend" existing MACT standards in a way that can make those standards more stringent without (a) providing an adequate justification for departing from the EPA's previous judgment about the appropriate level of the standards and the appropriate conditions for performance testing and (b) demonstrating that the revised standards would still meet the criteria of CAA section 112(d). The EPA's total failure to do so renders the Proposed Rule inadequate notice for public comment. If the EPA wishes to change the performance tests applicable to the existing MACT standards, making them more stringent, then it must at a minimum withdraw and re-propose the requirements, with adequate explanation and justification.
Response: As noted by the commenter, in addition to our ministerial action removing the vacated SSM exemption from emission standards (codified in General Provisions sections 63.6(f)(1) and (h)(1), formerly made applicable to this source category in Appendix A to Subpart RRR, and vacated by the Sierra Club decision), we also made inapplicable to this source category the SSM exemption from performance testing (see General Provisions section 63.7(e)(1)). The categorical exemption from performance testing for periods of SSM made sense when SSM periods were exempt from emission standards. Such an exemption is no longer appropriate because emission standards now apply at all times for this source category. Instead, the final rule provides in section 63.1511(a) that "[p]erformance tests shall be conducted under such conditions as the Administrator specifies to the owner or operator based on representative performance of the affected source for the period being tested." In addition, the final rule provides that performance tests should not be conducted during malfunctions. See section 63.1511(b). 
The removal of the exemption from performance testing for periods of startup and shutdown does not affect the stringency of the underlining emission standards nor does it alter the basic requirement that performance tests be conducted under conditions based on representative conditions. We note that subpart RRR requires initial performance tests in section 63.1511(b) and periodic performance tests every 5 years for major sources in section 63.1511(e), but these provisions do not necessarily require that performance tests be conducted during startup and shutdown periods, although the EPA maintains authority under Subpart RRR and the Clean Air Act to require such testing as appropriate. Sections 63.1511 and 63.1512 specify conditions and procedures for performance testing for this source category, some of which have been revised in this rulemaking as discussed in the preambles to the proposed and final rules, and elsewhere in the response to comments. In addition, we are finalizing two alternative methods for demonstrating compliance with production-based emission limits during periods of startup and shutdown, including one alternative method that relies on recordkeeping and not on performance testing. We note that the latter alternative method was proposed in the 2014 supplemental proposal, subsequent to Commenter 0195's comments on the 2012 proposal.
5.3 	Malfunction and Affirmative Defense
Comment #1: A commenter (0190) recommended that the EPA modify the affirmative defense provisions so that it is a "rebuttable presumption." The commenter was concerned that allowing a facility to interpose an affirmative defense for violations caused by malfunctions implies that the facility is guilty until proven innocent and improperly shifts the burden to the facility. Therefore, the commenter suggested that the EPA establish a rebuttable presumption (rather than affirmative defense) where it is presumed that any violation occurring during the malfunction was not the facility's fault unless the Agency proves certain facts that are enumerated in the rules. This will allow the Agency to challenge the alleged deviation without compromising the legal rights of either party.
While commenter 0190 would prefer the EPA use a rebuttable presumption, should the Agency keep the affirmative defense concept, the commenter suggested the following modifications to the language to make it more usable. The commenter suggested the EPA should drop the reference to "any" activity. There were also several references to "All" that would make it difficult to satisfy the requirements of an affirmative defense. In addition, the language in the provision was contradictory. In paragraph (a), the phrase "preponderance of evidence" was used while later in that paragraph (iii), the language referred to "any activity" meaning that more than preponderance of evidence was needed. This same trend occurred in paragraphs (5)  -  "All possible," (6) "All," (7) "All of the actions," and (8) "At all times." While "all" would include "preponderance," "preponderance" does not mean all of the time. The commenter suggested that the phrase "preponderance of evidence" is adequate and the references to "all" and "any" in the later paragraphs should be modified.
Commenter 0190 believed that a formal "root cause analysis" (as the term is understood by many engineers) should only be used when other reasonable methods fail to show what caused the malfunction or when the serious nature of an event might make such an analysis necessary. Moreover, other tools may be more appropriate (e.g., failure mode and effect, fault tree, etc.) or more powerful tools may be introduced in the future. The facility is the only one that can and should decide what tool to use to determine the cause of the malfunction. The commenter noted that to some companies and local regulators, the term "root cause analysis" implies a specific formal process. The commenter indicated that there are several techniques that may be called "root cause analysis," depending on the author and industry. If the EPA intended for the facility to investigate and fix the source of the malfunction so that it is less likely to recur, the commenter supported that concept but suggested that the Agency use an alternative term that does not carry a specific meaning. However, if the Agency envisioned a formal process for determining the root cause for every malfunction, no matter how simple, the commenter believed that this is unnecessary and would result in excess efforts with no environmental gains.
The commenter also noted that it is impossible to eliminate the causes for certain malfunctions (e.g., lightning strikes).
One commenter (0190) also noted that the EPA does not allow facilities to assert an affirmative defense for the exceedance of an emission limit during malfunctions if the EPA is seeking to enforce that emission limit through injunctive relief. The commenter stated that the Agency took that position based on a memorandum, State Implementation Plans: Policy Regarding Excessive Emissions During Malfunctions, Startup, and Shutdown at 2 (Sept. 20, 1999) (SIP SSM Memo). The commenter asserted that this policy is wrong and that the type of legal action or relief should have no bearing on the availability of this defense. The commenter concluded that the type of legal action the EPA uses to enforce a violation of its emission limits is simply irrelevant to whether the violation should be excused because of circumstances beyond the facilities control.
The commenter (0190) continued that not allowing an affirmative defense in an action for injunctive relief is arbitrary and capricious. As the D.C. Circuit Court stated in Essex Chem. Corp. v. Ruckelshaus, 486 F.2d 427. 433 (D.C. Cir. 1973) a case reviewing a § 111 rule, the court held that startup, shutdown, or malfunction ("SSM") provisions are "necessary to preserve the reasonableness of the standards as a whole." The D.C. Circuit Court of Appeals has also noted that "[a] technology-based standard discards its fundamental premise when it ignores the limits inherent in the technology." NRDC v. EPA, 859 F.2d 156, 208 (D.C. Cir. 1988). Therefore, the commenter indicated that the EPA should not apply a policy drafted to "ensure that SIPs provide for attainment and maintenance of the national ambient air quality standards ("NAAQS") and protection of prevention of significant deterioration (PSD) increments" and other risk-based programs, SIP SSM Memo at 2, to the CAA § 129 technology-based program.
Commenters (0190, 0193, 0194) suggested that the EPA clarify affirmative defense provisions by modifying the regulatory language in § 63.11501(e) to address the commenters' concerns and to make an affirmative defense a more useful tool (using strikeout to show text deleted and underline to show text added): 
      § 63.1520 Affirmative defense for violation of emission limit during malfunction.
      In response to an action to enforce the standards set forth in this subpart, you may assert an affirmative defense to a claim for civil penalties for violations of such standards that are caused by malfunction, as defined at § 63.2. Appropriate penalties may be assessed, however, if you fail to meet your burden of proving all of the requirements in the affirmative defense. The affirmative defense shall not be available for claims for injunctive relief.
      (a) To establish the affirmative defense in any action to enforce such a limit, you must timely meet the notification requirements in paragraph (b) of this section, and must prove by a preponderance of evidence that`:
      (1) The excess emissions:
      (i) Were caused by a sudden, infrequent and unavoidable not reasonably preventable failure of air pollution control and monitoring equipment, process equipment, or a process to operate in a normal or usual manner; and
      (ii) Could not have been reasonably prevented through careful planning, proper design or better operation and maintenance practices; and
      (iii) Did not stem from any activity or event that could have been reasonably foreseen and avoided, or planned for.
      (iv) Were not part of a recurring pattern indicative of inadequate design, operation, or maintenance; and
      (2) Repairs were made as expeditiously as possible when the applicable emission limitations were being exceeded. Off-shift and overtime labor were used, to the extent practicable to make these repairs; and
      (3) The frequency, amount and duration of the excess emissions (including any bypass) were minimized to the maximum extent practicable during periods of such emissions; and
      (4) If the excess emissions resulted from a bypass of control equipment or a process, then the bypass was unavoidable to prevent loss of life, personal injury, or severe property damage; and
      (5) All possible Reasonable steps were taken to minimize the impact of the excess emissions on ambient air quality, the environment and human health; and
      (6) All emissions monitoring and control systems were kept in operation if at all possible, consistent with safety and good air pollution control practices; and
      (7) All of the actions in response to the excess emissions were documented by properly signed, contemporaneous operating logs; and
      (8) At all times, the affected source was operated in a manner consistent with good practices for minimizing emissions; and
      (9) A written root cause analysis report has been prepared, the purpose of which is to determine, correct, and eliminate mitigate the primary causes of the malfunction and the excess emissions resulting from the malfunction event at issue. Facility personnel will determine the appropriate type of analysis required (may include but not limited to root cause analysis, failure mode and effect, fault tree, etc.) to identify the cause of the malfunction. The analysis report shall also specify, using best monitoring methods and engineering judgment, the amount of excess emissions that were the result of the malfunction.
      (b) Reports. The owner or operator seeking to assert an affirmative defense shall submit a written report to the Administrator within 45 days of the initial occurrence of the violation of the standards in this subpart, which may be the end of any applicable averaging period, to demonstrate, with all necessary supporting documentation, that it has met the requirements set forth in paragraph (a) of this section. The owner or operator may seek an extension of this deadline for up to 30 additional days by submitting a written request to the Administrator before the expiration of the 45 day period. Until a request for an extension has been approved by the Administrator, the owner or operator is subject to the requirement to submit such report within 45 days of the initial occurrence of the violation. A written report to demonstrate, with all necessary supporting documentation, that you have met the requirements set forth in paragraph (a) of this section. This report should be submitted as a part of the excess emissions and continuous monitoring system performance report and summary report as required in § 63.10(e)(3).
One commenter (0194) stated that the proposed affirmative defense language for periods of malfunction needs clarification and revision to make it usable. The commenter asserted that the proposed affirmative defense language is similarly made unusable by numerous limitations such as "could not have been prevented", "could have been foreseen and avoided", "all possible steps", "at all times" and "all actions ...documented". The commenter stated that this language so narrowly limits the affirmative defense that the provisions would have no practical application and provide no real legal defense for owners and operators, who are the intended users.
One commenter (0190) reported that the language the EPA is proposing in § 63.1520 is similar to the definition in § 63.2 with one major exception. In the proposed language, one of the conditions for an affirmative defense is that the excessive emissions were caused by a "sudden, infrequent, and unavoidable failure..." The General Provisions definition of malfunction uses the phrase "not reasonably preventable" instead of the word "unavoidable." The EPA obviously understands this difference in that they quote § 63.2 on page 8,599 of the preamble yet still makes a change in definition of malfunctions in § 63.1520. The commenter believed that it is inappropriate have two different definitions of malfunction and requested that the Agency revise the language to reflect the General Provisions definition of a malfunction.
One commenter (0191) commended the agency for adding an affirmative defense to civil penalties for violations of emission limits caused by malfunctions but recommended that the facilities asserting the affirmative defense be allowed sufficient latitude to demonstrate that the emission non-compliance was in fact caused by a sudden, infrequent, and unavoidable failure of air pollution control and monitoring equipment, that repairs were made as expeditiously as possible when the applicable emission limitations were being exceeded, and that effective steps were taken to minimize the impact of the excess emissions on ambient air quality, the environment and human health.
One commenter (0195) recommended that in proposed section 63.1520(a)(6), the phrase "if at all possible" should not be used because it is an extreme term that bears no relation to good air pollution control practices. The commenter suggested that this provision be qualified, as the EPA has qualified similar provisions in the NESHAP General Provisions in 40 C.F.R. §63.6 with caveats that the operation must be consistent with safety and good air pollution control practices, that the provision does not require the source to make further efforts to reduce emissions below what the standards require, and that it does not require regular operation of backup or standby pollution control equipment.
The commenter stated that the EPA cannot adopt the kind of absolute requirements implied by the Proposed Rule, without an explanation of why it is necessary and appropriate to do so. The commenter cited several recent NESHAP that inserted language clarifying that the general duty to minimize emissions "does not require the owner or operator to make any further efforts to reduce emissions if levels required by this standard have been achieved." See 76 Fed. Reg. 22,566, 22,583, col. 3 (April 21, 2011).
One commenter (0195) recommended that the language of proposed section 63.1520(a)(4) may be difficult or impossible for a source to demonstrate. The commenter stated that there could be substantial room for disagreement about what constitutes an appropriate exercise of air pollution control practice. The commenter maintained that there is nothing inherent to standards under CAA section 112 that requires a source to avoid bypassing control equipment to such a degree, nor has the EPA attempted to explain, why the CAA would disfavor bypassing "a process" in this way.
One commenter (0195) stated that the language in proposed section 63.1520(a)(5) regarding "all possible steps" is subjective, and presents a disconnect between the absolute and extreme requirement of the affirmative defense and the provisions of the CAA designed to attain national ambient air quality standards and protect human health and the environment. The commenter observed that the CAA does not require sources to take "all possible steps" to control emissions, even with respect to the residual risk of hazardous air pollutant emissions addressed under CAA section 112(f). The commenter claimed it is unclear how this criterion for qualifying for the affirmative defense differs from proposed section 63.1520(a)(3), which requires that the frequency, amount, and duration of excess emissions "were minimized to the maximum extent practicable." The commenter stated that unless the EPA explains what additional showing would be needed by section 63.1520(a)(5), that paragraph should be eliminated.
One commenter (0197) indicated that an affirmative defense is insufficient to absolve the EPA of its duty to set Section 112-compliant standards. The commenter stated that rather than attempt to promulgate section 112-compliant standards, the EPA proposes to establish an "affirmative defense" for malfunctions only. The commenter declared this "affirmative defense" to be insufficient to meet the EPA's obligations under section 112. The commenter also noted that the EPA has previously found a case-by-case basis consideration of potential enforcement to be a less desirable route to address malfunctions. 
One commenter (0195) stated that the proposed affirmative defense is not a substitute for addressing malfunction events in the emission standards as emission limitations achieved by best-performing existing sources under CAA section 112(d)(3), or as beyond-the-floor emission standards under CAA section 112(d)(2). 
The commenter questioned the EPA's legal authority in the CAA to shift the burden to the regulated community of proving (or disproving) essential elements of an alleged violation. The commenter asserted that if the EPA adopts an approach along the lines of the proposed affirmative defense, it should be stated instead in terms that, once a source has claimed that its excess emissions were related to a malfunction, it will not be considered to have violated the standards unless the enforcement authority demonstrates that the source is not entitled to claim the malfunction.
One commenter (0195) stated that the EPA should provide the same kind of affirmative defense that it proposes for malfunctions for startup and shutdown events as well.
One commenter (0197) asserted that, while the EPA has authority to establish an affirmative defense for SSM events, the criteria outlined in the proposed rule are arbitrary and capricious and should be revised. The commenter noted that the Proposed Rule would limit the affirmative defense to malfunctions only and not include startup and shutdown periods. The commenter discussed some legal precedent supporting the EPA's authority to provide an affirmative defense for malfunctions. However, the commenter believed that the proposed affirmative defense includes criteria that are unduly restrictive so as to not constitute a defense at all. 
One commenter (0195) stated that is not clear what the affirmative defense covers. The commenter asked if the term "civil penalties," which is not defined in the Proposed Rule, could possibly be intended to apply to a "civil administrative penalty" imposed by the EPA under CAA section 113(d), noting that the term "civil penalty" in other contexts means only penalties imposed by a court. The commenter asked if the affirmative defense applies to "noncompliance penalties" under CAA section 120 (which apply, inter alia, to noncompliance with a section 112 NESHAP). The commenter asserted that to meet the purported purpose of the affirmative defense, which is to provide relief from emission limitations that cannot be met at times even with equipment that is properly designed and maintained the affirmative defense would need to apply to civil and administrative penalties, including noncompliance penalties. The commenter noted that in the proposed rule's preamble, the EPA states that, if a source cannot prove its entitlement to the affirmative defense, appropriate penalties could be assessed in accordance with section 113 of the CAA. The commenter stated that this does not answer the questions presented in this paragraph.
The commenter stated that it is unclear as well how the affirmative defense would apply to enforcement actions by state and local governments, or to private citizen enforcement actions under CAA section 304. The commenter maintained that an affirmative defense should clearly state that it is applicable to enforcement actions by states or citizen-suit plaintiffs, as well.
The commenter observed that the EPA did not give any explanation for why the affirmative defense would not apply to injunctive relief. The commenter stated that if the excess emissions associated with the equipment or process failure really are not reasonably preventable, then there is no apparent reason why an affirmative defense to a claim for injunctive relief should not also be available. The commenter argued that in fact, as a matter of law, injunctive relief may not be available in cases where a civil penalty cannot be imposed. See Sierra Club v. Otter Tail Power Co., 615 F.3d 1008 (8th Cir. 2010) (under concurrent remedy doctrine, injunctive relief for a CAA violation is barred when civil penalty is barred by statute of limitations). 
The commenter asserted that maintaining liability for injunctive relief renders the affirmative defense particularly ineffective with respect to citizen suits. The commenter postulated the if the source is even potentially subject to injunctive relief and, therefore, could be required to pay the citizen-plaintiff's attorneys fees even if the source successfully demonstrated that it otherwise qualified for the affirmative defense, then the affirmative defense would not accomplish the EPA's stated objective of providing relief in situations where the emission limitations cannot be met despite proper design and operation of process and control equipment.
The commenter stated that the EPA has not addressed these and other apparent limitations and shortcomings of the affirmative defense, which make it an entirely inadequate substitute for setting MACT standards that include provisions for SSM events. 
The commenter stated that thus, at a minimum, the EPA should state that the affirmative defense applies to civil penalties, civil administrative penalties, noncompliance penalties, and injunctive relief, in an action brought by the EPA, a state, or a citizen-suit plaintiff. The commenter suggested that the EPA also should reword the affirmative defense, so that it states that a source "will not be deemed in violation of" the MACT standards for excess emissions or other deviations from the standards, associated with a startup, shutdown, or malfunction event, unless the event, and the source's response to the event, do not meet the criteria spelled out in the regulations. The commenter stated that configured in this way, the provision for malfunction events should be called something other than an "affirmative defense," such as an "alternative standard for SSM events."
One commenter (0195) disagrees with the EPA's tying the proposed affirmative defense to a continuously applicable section 112-compliant emission standard as discussed in Sierra Club v. EPA. The commenter argued that if the CAA required that the EPA be able to enforce a numerical limitation on a source's hazardous air pollutant emissions at all times, which the commenter disputes, then excluding injunctive relief from the affirmative defense, as the EPA proposes, would not meet that requirement. The commenter stated that injunctive relief is prospective only: it does nothing to prevent (or even penalize) a past excess emissions event, while under the approach the EPA proposes, a malfunction that causes emission limitations to be exceeded could trigger injunctive relief even where the affirmative defense prevented the EPA from imposing penalties, that injunctive relief would not address directly the past event (which, by definition, was unexpected), but would only require the source to do something different in the future. 
The commenter asked, if the Agency's providing for injunctive relief satisfies the CAA "requirement" for a continuously applicable section 112-compliant emission standard that the EPA posits, why is it necessary for the EPA to change the existing MACT standards to impose civil penalties for malfunctions (and then excuse a few of them with the affirmative defense), rather than simply providing that malfunction events, although they continue to be exempt from civil penalties, are not exempt from potential injunctive relief? The commenter stated that it would reduce if not eliminate the issues of reasonableness and Constitutional due process that the Proposed Rule presents when it imposes penalties for exceedances that the source could not reasonably have prevented.
One commenter (0195) stated that the affirmative defense as written is unreasonable and impracticable. The commenter asked that the affirmative defense must be substantially modified for it to provide any significant relief. The commenter suggested that the affirmative defense needs to state clearly that a source that qualifies for the affirmative defense shall not be deemed to have violated the applicable standards during that time. The commenter maintained that if the EPA does that, it may be unnecessary to state also that the affirmative defense relieves the source from liability for all types of penalties and injunctive relief (save criminal penalties), but that should be the clear effect of qualifying for the affirmative defense. The commenter recommended that the affirmative defense also should be available not only for malfunctions, but also for excess emissions during startup and shutdown.
The commenter discussed aspects of the affirmative defense that make it unavailable as a practical matter for many, if not most malfunctions. 
One commenter (0195) stated that several of the conditions for establishing an affirmative defense in proposed §63.1520(a) list numerous items which may be appropriate considerations in determining whether excess emissions during a malfunction should be considered a violation, but which ought not to be listed as mandatory criteria, all of which have to be met in order for a source even to raise an affirmative defense.
The commenter observed that a number of the conditions for establishing an affirmative defense use phrases that are subject to a wide range of interpretations, and that on their face do not recognize any need for reasonableness or cost effectiveness. The commenter suggested that in some cases, it may have been possible to prevent the malfunction, or to have further reduced the excess emissions, if the source had spent huge amounts of money or had imposed economically impracticable constraints on its operation. The commenter observed that affirmative defense, as proposed, leaves open the possibility that a source will be considered to be in violation because the enforcement authority decides subjectively that in one or more respects it would have been "proper" or "possible" for the source to take further steps to prevent or minimize the malfunction, allowing the EPA or the court to impose an extreme version of MACT during malfunction periods, without any application of the beyond-the-floor factors in CAA section 112(d)(2). The commenter noted that the descriptors in the criteria for demonstrating the affirmative defense (section 63.1520(a)(1)(ii), section 63.1520(a)(1)(iv),section 63.1520(a)(2), section 63.1520(a)(5), and section 63.1520(a)(6)) would inevitably lead to varying conclusions as to whether a violation has occurred, resulting in inconsistency from one jurisdiction to the next.
One commenter (0195) asserted that the requirement in proposed section 63.1520(a)(9) that the source "have prepared a written root cause analysis to determine, correct and eliminate the primary causes of the malfunction and the excess emissions resulting from the malfunction event at issue" does not implement the statutory criteria for standard-setting under section 112. The commenter maintained that if read literally, it would mean that a source could never take advantage of the affirmative defense if the source was unable to determine the primary cause of the malfunction or was unable to correct that cause. 
One commenter (0195) stated that 90 days is the minimum amount of time to provide the kind of extensive documentation required by the affirmative defense as currently written, including a completed root cause analysis, unless the EPA streamlines the criteria for affirmative defense. 
The commenter disagreed that the starting time for the proposed 45-day period for submitting a written report should be on the date of "the initial occurrence of the exceedance of the standards." The commenter observed that much of the required content of that written report could not be created until the malfunction event ceased, which could be many days after the malfunction commenced and a limit on the amount of time a source would have to prepare the report.
The commenter argued that the provision allowing an extension of the reporting deadline of up to 30 days would provide little practical relief as the source would have to submit the report within 45 days unless the EPA Administrator granted the extension request before the expiration of the initial 45-day period, which is not likely to happen until shortly before, if not after, the report would otherwise be due. The commenter recommended that the rule should provide that a request for extension of the reporting deadline on which the Agency has not acted within 10 days is considered granted.
The commenter requested that malfunction reporting be required on a semi-annual basis, for malfunctions that occurred during the preceding six months (the same frequency as similar reports pursuant to NESHAPs and Title V permit requirements). The commenter explained that this would enable the EPA to review the source's compliance history and the measures being taken to address malfunctions, without imposing unworkable reporting deadlines on sources or potentially encouraging sources to submit numerous "protective" reports to the Agency for malfunctions that may not ultimately result in the source exceeding a 30-day average emission limitation.
One commenter (0195) claimed that the requirement in proposed section 63.1520(b) to submit a written report to the Administrator within 45 days of the initial occurrence of the malfunction that demonstrates, "with all necessary supporting documentation," that the source met all of the multitude of criteria for the affirmative defense, is unreasonable and unnecessary. The commenter disagreed that a person could be determined to have acted unlawfully unless the person has submitted his or her entire defense before he or she is even notified of a potential enforcement action. The commenter asserted that the enforcement authority would be able to see, based on the kind of short malfunction or deviation report sources already submit under many air programs, that an exceedance of the Proposed Standards resulted from an unforeseen and unavoidable equipment failure or process upset. The commenter observed that it is extremely inefficient and burdensome, for both sources and regulators alike, to require a complete justification of the affirmative defense before the enforcement authority has indicated any need for further investigation.
The commenter further noted that affirmative defense notifications would swamp the EPA and state agencies with unnecessary paperwork, and for no clear benefit, because sources will need to submit notifications of the malfunction, with all the supporting documentation, whenever a malfunction defense could be claimed
One commenter (0197) supported the incorporation of the definition of "malfunction" in 40 C.F.R. §63.2 as a long-established definition that the regulated community is familiar with. The commenter noted, however, that the criteria outlined in the Proposed Rule to establish the affirmative defense are inconsistent with this definition and should be eliminated. The commenter declared that the criteria are substantially more restrictive than the requirements to show a malfunction under the definition, and these more restrictive criteria would result in only a subset of malfunctions being considered eligible for the affirmative defense, if any at all, undermining the purpose of the provision. The commenter noted that the criteria for establishing an affirmative defense may impose a greater burden on regulated entities than would otherwise be required by the standards; they are redundant of the general duty standard and may cause undue confusion as to the applicable requirements. The commenter stated that other criteria more stringent than, or at least redundant with, the general duty standard, should be eliminated. 
If the EPA subsequently determines to utilize an affirmative defense, then the commenter believed that the EPA should eliminate the criteria in the proposal, and require the regulated entity to establish that a malfunction, as defined in section 63.2 occurred. Alternatively, the commenter recommended that the EPA could use the Clean Water Act's upset defense as a model, rather than the 1983 and 1999 EPA memoranda on excess emissions that do not adequately address the treatment of SSM events in technology-based standards.
One commenter (0190) stated that, in the recently signed National Emission Standards for Hazardous Air Pollutants for Polyvinyl Chloride and Copolymers Production final rule (FR notice if available), the EPA incorporated the reporting requirements for an affirmative defense as a part of the excessive emissions reporting requirements of 63.10(e)(3). The commenter suggested that the Agency modify the reporting requirements in this rule to match the reporting requirements in 63.11895(b).
One commenter (0190) stated that as facilities and the EPA move toward electronic recordkeeping, it does not make sense to require keeping a "properly signed, contemporaneous operating logs" as a requirement for an affirmative defense. The commenter noted that there are a number of electronic methods for maintaining records currently available and suggested modifying this provision.
One commenter (0192) stated that EPA properly recognizes CAA section 112 does not require that emissions that occur during periods of malfunction be factored into development of CAA section 112 standards and objected to the proposed malfunction defense to the extent it includes a non-malfunction presumption. The commenter stated that creating a new program for malfunctions is not justified and conflicts with the straightforward definition under 40 CFR §63.2, and recommended that the proposed affirmative defense rule provision be dropped.
One commenter (0193) recommended the EPA establish a rebuttable presumption (rather than affirmative defense) where it is presumed that any violation occurring during the malfunction does not automatically result in civil liability and the imposition of civil monetary penalties unless the Agency proves certain facts that are enumerated in the rules.
One commenter (0195) asserted that being able to assert a defense is not the same as complying with emission limitations that are properly set in accordance with CAA section 112. The commenter expressed concern that despite the fact that a source believes it qualifies for the affirmative defense, it may be considered to have violated the standards -- and may have to report violations, certify noncompliance, etc. -- until there has been an enforcement proceeding and the source has successfully asserted the affirmative defense. The commenter stated that the affirmative defense places the source in the position of having to prove its innocence, rather than the EPA or another enforcement authority having to prove that the source violated the CAA.
One commenter (0195) maintained that the EPA's discussion of the regulatory burden of the affirmative defense indicates that the EPA does not believe the affirmative defense will mitigate to any significant extent the unreasonableness of imposing emission limitations that the Agency recognizes sources will not be able to meet when malfunctions occur. According to the commenter, the EPA states that in only a small number instances would source operators use the affirmative defense  -  presumably, the commenter stated, because the criteria the EPA proposes for qualifying for the affirmative defense will, as discussed above, be difficult if not impossible to meet. The commenter asserted that the EPA believes that it has so cabined the affirmative defense that it will effectively be unavailable, meaning that it will provide virtually no relief to address the legal necessity that the EPA not impose emission limitations that sources will be unable to achieve despite the proper design and operation of their facilities. The commenter stated that it is not credible for the EPA to claim that, despite the wide variety of circumstances at individual facilities and the wide variation in the number of facilities covered by particular NESHAP source categories, there will only be 0-3 circumstances in each of those source categories (or groups of source categories) qualifying for the affirmative defense over a period of three years.
One commenter (0195) disagreed with the EPA's estimate of the cost of the notification, reporting, and recordkeeping requirements associated with a source asserting the affirmative defense for a malfunction. The commenter believed that this estimate is extremely low, given the approximately one dozen different demonstrations the source is required to make under the criteria proposed for the affirmative defense, and given that this estimate also presumably includes the cost of preparing the written root cause analysis, which must not only "determine, correct, and eliminate the primary causes of the malfunction and the excess emissions resulting from the malfunction" but also must "specify, using best monitoring methods and engineering judgment, the amount of excess emissions that were the result of the malfunction."
The commenter stated that the burden estimate is arbitrary and irrational, noting a range of costs for the reporting and recordkeeping required to qualify for the affirmative defense for a single malfunction event for various recent NESHAP. 
One commenter (0195) stated that the reasons the EPA estimates so few incidents will qualify for the affirmative defense is that the Agency has defined "malfunctions" to be "infrequent" events that could not have been foreseen and has suggested that they cannot be "predictable and routine aspects of a source's operations." The commenter claimed that the EPA shows a bias that assumes sources currently operate with little concern for controlling excess emissions, because the sources can claim that those excess emissions are covered by the malfunction exemption in the current regulations. The commenter disagreed with this assumption. The commenter maintained that the SSM provisions in the existing Secondary Aluminum Production NESHAPs require the source to take measures to prevent or minimize excess emissions during SSM events to the extent practical and to follow a written SSM plan, and the source also is subject to a general duty under the NESHAPs General Provisions to operate and maintain the source "in a manner consistent with safety and good air pollution control practices for minimizing emissions." The commenter urged the EPA to not base its rulemaking on an unsupported assumption that sources subject to the Proposed Rule are not operating in accordance with existing regulations.
The following comments were received on the supplemental proposal regarding the proposed change to the affirmative defense provisions:
 Commenter 0298 stated that the supplemental notice implies that the affirmative defense was not a necessary part of the standards but was included in the standards "to provide a more formalized approach and more regulatory clarity." 79 Fed. Reg. 72,906. The commenter stated that this statement does not excuse the EPA's failure to reconsider entirely its statement of basis and purpose for the proposed rule with the affirmative defense for malfunctions removed. The commenter stated that, in the first place, the affirmative defense was a key element of the EPA's explanation for why it was lawful for the Agency to promulgate MACT standards that apply during malfunctions but that do not reflect what is achievable with available technology during malfunctions. See Proposed Rule, 77 Fed. Reg. at 8599-8600. The commenter stated that the Agency believed at the time that an affirmative defense provision was necessary to effectuate the congressional directive that MACT standards be based on what is achievable with available technology; otherwise, the EPA would have had no statutory authority to promulgate the affirmative defense. The commenter stated that they believe that the EPA staff almost inevitably had to have considered the effect of the affirmative defense on compliance obligations when they were deriving the proposed standards. The commenter provided the recent example of the MACT standard for boilers and process heaters in which the EPA revised its MACT floor calculation to factor in periods of malfunction that did not qualify for affirmative defense. The commenter stated that having taken the affirmative defense into account when deciding what emission standards to include in the proposed rule, the EPA cannot now eliminate the affirmative defense without reassessing the appropriateness of those emission standards and how they will affect secondary aluminum plants if the standards are no longer mitigated by the affirmative defense. The commenter stated that the EPA must then publish its analysis and seek public comment on it before promulgating a final rule that makes emission limitations based on data from normal operations applicable even during malfunctions but deletes any affirmative defense for malfunctions.
Commenter 0298 stated that the court decision cited as the reason for eliminating the affirmative defense provisions said nothing about (a) whether the EPA must consider the effect of malfunctions when it sets emission standards that apply at all times (except during some startup and shutdown periods); or (b) whether it is appropriate for the emission standards themselves to contain an alternative emission standard that applies during malfunction events that meet certain criteria. The commenter stated that while the court decision is grounds for the EPA eliminating the affirmative defense as it was written in the proposed rule, it in no way authorizes the EPA to remove the affirmative defense from the emission standards without substituting some other provision addressing malfunctions; nor does it allow the EPA to remove the affirmative defense without even considering whether/how the remainder of the emission standards must be revised in light of the removal of the affirmative defense.
Commenter 0290 stated that they object to the EPA eliminating the Affirmative Defense, to the extent those provisions may be interpreted to require compliance with emissions limits during a malfunction contrary to established law. The commenter stated that although it reasonable for a source to meet operational requirements during a malfunction that minimize emissions, during a malfunction a source cannot be expected meet the same emissions limitations designed for normal operations. 
Commenter 0295 stated that they are concerned that the removal of the affirmative defense provisions will lead to an expectation that emission limits must be complied with during malfunction periods, even with the clear language to the contrary in the proposed rule preamble. The commenter stated that the proposed approach of offering only enforcement discretion does not provide an adequate regulatory framework for facilities managing malfunction events. The commenter requests that the EPA reconsider its position and establish reasonable standards for malfunctions, rather than leaving a vague regulatory framework that could allow biased agency action or third party litigation for unavoidable situations. The commenter stated that due to the number and nature of potential malfunctions, establishing numerical emission limits is not technically feasible or practicable. The commenter recommended that the EPA establish malfunction work practice (such as a written malfunction response plan) and/or other alternative standard in place of the numerical MACT emission limits that were developed from normal operating conditions. 
Commenter 0296 urged the Agency to exercise its case-by-case enforcement discretion to provide flexibility as often as necessary in the event of violations caused by malfunctions.
Response: The EPA received numerous comments on the affirmative defense to civil penalties for violations caused by malfunctions that the EPA proposed to include in the rule. Several commenters supported the affirmative defense or supported the affirmative defense but made suggestions for revisions to its provisions. Other commenters opposed the affirmative defense or various components of the affirmative defense.
As discussed in the 2014 supplemental proposal, the D.C. Circuit Court vacated an affirmative defense in one of the EPA's Section 112(d) regulations. NRDC v. EPA, 749 F.3d 1055 (D.C. Cir. 2014) (vacating affirmative defense provisions in Section 112(d) rule establishing emission standards for Portland cement kilns). The court found that the EPA lacked authority to establish an affirmative defense for private civil suits and held that under the CAA, the authority to determine civil penalty amounts in such cases lies exclusively with the courts, not the EPA. Specifically, the Court found: "As the language of the statute makes clear, the courts determine, on a case-by-case basis, whether civil penalties are `appropriate.'" See NRDC, 749 F.3d at 1063. In light of NRDC, the EPA in the 2014 supplemental proposal withdrew the proposed affirmative defense and is not including a regulatory affirmative defense provision in the final rule. In the event that a source fails to comply with applicable CAA section 112(d) standards as a result of a malfunction, the EPA's ability to exercise its case-by-case-enforcement discretion to determine an appropriate response provides sufficient flexibility in such circumstances. Further, as the D.C. Circuit recognized, in an EPA or citizen enforcement action, the court has the discretion to consider any defense raised and determine whether penalties are appropriate. Cf. NRDC, 749 F.3d at 1064 (arguments that violation was caused by unavoidable technology failure can be made to the courts in future civil cases when the issue arises). The same is true for the presiding officer in EPA administrative enforcement actions. In light of the EPA's withdrawal of the proposed affirmative defense and the absence of an affirmative defense in the final rule, we are not responding to the comments opposing or suggesting revisions to the proposed provisions to establish an affirmative defense. With respect to comments that supported the affirmative defense or argued that further analysis or rule changes were necessary if the affirmative defense is not adopted, the preamble to the final rule, sections 5.1 and 5.2 above, and this section provide the EPA's reasons for not including an affirmative defense provision in the final rule as well as other responses.
Some commenters assert that EPA must reconsider its proposed decision to apply existing Subpart RRR standards to periods of malfunction in light of the withdrawal of the proposed affirmative defense, because the affirmative defense was necessary to ensure the MACT standards are achievable. A commenter cites an example of the EPA revising the MACT floor calculation for another source category to consider malfunctions. First, we disagree that the proposed affirmative defense was "necessary" to support the approach to malfunctions in the proposed rule (or in this final rule). As discussed above, we also disagree with the commenters' views regarding the term "achievable" as that term is used in section 112. Further, in reviewing the comments received on the 2012 proposal and 2014 supplemental proposal and developing the final rule, the EPA has considered all of the issues raised by commenters concerning the appropriateness of applying the existing Subpart RRR MACT standards to periods of malfunction and has concluded that the approach remains both consistent with the Act as interpreted in relevant court decisions and appropriate as a regulatory and compliance assurance policy matter. The fundamental analyses and considerations that support our approach to addressing violations caused by malfunctions was not affected by the withdrawal of the proposed affirmative defense. We again note that the malfunction exemption from Subpart RRR emission standards has not existed for several years, since it was vacated by the court, and we have received no indication from commenters or others of compliance problems or related enforcement responses or actions for this source category based on violations caused by malfunctions. Indeed, no commenter identified even a single actual case of a Subpart RRR compliance issue caused by a malfunction. Based on this record, we believe EPA's case-by-case enforcement discretion, as well as the discretion of a court in an enforcement action, to determine an appropriate response, including consideration of any defense raised concerning malfunctions and whether penalties are warranted in light of relevant facts, is a reasonable approach to malfunctions.

6. Regulatory Impacts

Comment #1: One commenter (0199) discussed the history of regulatory activities and agreements involving the commenter, the U.S. EPA, including Regional offices, state agencies, and the U.S. Department of Justice; cumulating with a Consent Decree (attached to the comment) and guidance regarding subpart RRR . The commenter stated that as a consequence of these negotiations, it was determined that the commenter would request formal alternative determinations from the EPA, which would be acted on and issued. The commenter has since operated all of its facilities using this agreed-upon guidance which has, to date, been fully accepted and utilized by the respective states and the EPA Regions in their role administering the Subpart RRR NESHAP rules.
The commenter expressed concern that the terms of certain provisions of the proposed changes to the Subpart RRR NESHAP rules because they seem to change or otherwise limit agreed-upon guidance which took extensive time and effort to develop and which has proven to be very helpful to both the commenter and the Government in the interpretation of the meaning and intent of these very complicated rules. 
Response: The requirements of these final amendments apply even if an individual facility is operating under a consent decree that previously established less stringent requirements.
6.1 	Cost and Economic Impacts
Comment #1: One commenter (0192) stated that because of safety concerns and the cost burden of installing capture systems for performance testing on uncontrolled group 1 furnaces, facilities will likely decide to not classify additional furnaces as group 1, and those furnaces will not recycle purchased scrap, reducing recycling rates, using more expensive and energy-intensive primary aluminum. The commenter noted that the EPA did not include this increased expense to facilities in its economic impact evaluation. 
Response: We do not agree that there will be any measurable decrease in secondary aluminum recycling rates related to the requirements for performance testing on uncontrolled group 1 furnaces.
Comment #2: One commenter (0192) noted that the EPA claimed a net benefit of $600,000 to "Allow Furnace Switching." The commenter maintained that the proposed rule does not freely allow furnaces to switch classifications, it severely restricts when a change can occur and requires redundant and unnecessary testing to do so. The commenter asserted that the EPA failed to include any cost to account for the downtime that facilities will experience when a baghouse malfunctions and a furnace classification change is prohibited or delayed by the proposed rule. The commenter estimated that this restriction could cost in excess of $1 million for a facility (the commenter provided the basis for this cost estimate in a separate Confidential Business Information submittal). The commenter claimed that the EPA did not include costs to account for unnecessary operation of baghouses when facilities are compelled by business conditions to process more expensive aluminum feedstock than is necessary if they had the flexibility to change furnace classifications to a group 2 Furnace or a group 1 Furnace without add-on control under the conditions set forth in the February 16, 2012 Applicability Determination. The commenter stated that the EPA did not include an estimate of the additional testing requirements included in section 63.1514, or the costly requirements and restrictions to operational flexibility.
The commenter stated that the EPA did not provide cost estimates for the requirement requiring the operation of control devices when furnaces are idled for less than 24 hours, which the commenter believes is unnecessary; the 60-day waiting period before a revised OM&M Plan can be implemented, which the commenter believes reduced facility's flexibility; or the regulatory burden on area sources to submit annual compliance certifications.
Response: As discussed above, in response to comments, the EPA will allow up to 4 furnace switches in any 6-month period. Therefore, we do not expect any cost associated with downtime when a baghouse malfunctions, because we do not anticipate that a furnace classification will be prohibited or delayed by the final rule. 
Prior to the promulgation of the 2015 amendments, the rule required that furnace control devices operate at all times. However, the 2015 revised rule allows control devices to be shut down when furnaces will be idled for more than 24 hours as a means to provide operational flexibility and cost savings when furnaces are idled over weekends or extended holidays. Therefore, this allows more flexibility and involves less cost, not any additional cost.
We have removed the 60-day waiting period before a revised OM&M Plan can be implemented as well as the requirement for annual compliance certification for area sources. 
Comment #3: One commenter (0191) cautioned the EPA to validate the actual size and scope of the regulated community subject to the subpart RRR rule. The commenter indicated concern over the initial ICR distribution list being derived incorrectly, erroneously identifying a number of facilities as Secondary Aluminum Production sources subject to the Subpart RRR rule. The commenter expressed concern that some of the 166 facilities that the EPA maintains are subject to the Subpart RRR rule might be facilities that either did not understand the applicability definition or had completed the ICR prior to being advised that it did not in fact apply to their facility. The commenter questioned the estimate of the number of facilities the EPA believes are actually subject to the rule may be overstated, thereby also calling into question all of the estimates on costs and benefits contained within the rule.
Response: As described in the memorandum, "Development of the RTR Risk Modeling Dataset for the Secondary Aluminum" (docket item EPA-HQ-OAR-2010-0544-0149) we believe there are approximately 161 facilities within the Secondary Aluminum Production source category. It is unlikely the number of facilities is overestimated because, as part of the risk modeling process, we identified facilities that had equipment regulated under Subpart RRR based on ICR responses . We also believe it is unlikely that a significant fraction of the facility owners who responded to the ICR erroneously stated to the EPA that their facility is subject to Subpart RRR. We note the commenter did not provide specific information about any individual facilities. We believe our estimate of the number of facilities subject to subpart RRR remains accurate although, in response to comments on the 2012 proposal and the 2014 supplemental proposal, we have revised certain estimates of costs and benefits. These are described in the preamble to this final rule.

7. Miscellaneous Comments
7.1 	Commenter Endorsements
Comment #1: One commenter (0193) supported the comments of Alcoa and Coalition for Responsible Waste Incineration (CRWI) (Commenters 0194, 0190).
Response: The EPA has responded to the comments of Alcoa and the Coalition for Responsible Waste Incineration (CWRI) (Commenters 0194, 0190). 
Comment #2: One commenter (0194) supported the comments of the Aluminum Association and the Coalition for Responsible Waste Incineration (0193, 0190).
Response: The EPA has responded to the comments of Alcoa and the Coalition for Responsible Waste Incineration (CWRI) (Commenters 0194, 0190).
Comment #3: One commenter (0192) supported and agreed with the Aluminum Association's (Commenter 0193) comments and evaluation concerning the overestimate of risk in the EPA's risk assessment calculations due to HAP emissions from secondary aluminum facilities. The commenter stated that to the extent not inconsistent with the comments provided herein, it incorporates by reference TAA's comments on the proposed rule.
Response: The EPA has responded to the comments of The Aluminum Association (Commenter 0193). 
Comment #4: One commenter (0199) supported the comments of the Aluminum Association (0193).
In a comment on the supplemental proposal, commenter 0290 stated that, to the extent consistent with their submitted comments, they incorporate by reference TAA's comments on the proposed rule.
Response: The EPA has responded to the comments of The Aluminum Association.
7.2	Other Comments
Comment #1: One commenter (0183) asked for an extension of the comment period.
Response: The comment period was extended from March 30, 2012 to April 13, 2012.
Comment #2: A commenter (0191) asked that the EPA incorporate by reference the full text of the Subpart ZZZZZZ rule into the docket for the current rulemaking, drawing specific notation to the language pertaining to the performance testing and monitoring requirements established for aluminum foundries subject to the rule.
Response: There is no need to incorporate Subpart ZZZZZZ into the docket for the current rulemaking, because it is already available in the Code of Federal Regulations. 
Comment#3: One commenter (0019) stated support for regulations that protect environmental safety.
Response: The EPA acknowledges the commenter's support.
Comment #4: In a comment on the supplemental proposal, commenter 0298 stated that they incorporated by reference their comments on the 2012 proposal.
Response: No action needed on this comment as their comments on the 2012 proposal are addressed elsewhere.
Comment, #5: In a comment on the supplemental proposal, commenter 0295 stated that in addition to their comments on the 2014 supplemental proposal, they note that many of the items it commented on in comments provided on April 13, 2012 regarding the proposed rule noticed on February 14, 2012 (77 FR 8576) will not be addressed until final rule promulgation expected in August 2015. The commenter stated that given the inter-relatedness of the 2012 comments with those being provided here, they stress the continued validity and applicability of its 2012 comments to this supplemental proposed rulemaking as well as reiterates their applicability to the original 2012 proposed rule.
Response: No action required; their comments on the 2012 proposal are addressed elsewhere.
Comment #6: In a comment on the supplemental proposal, commenter 0296 stated that the proposal defines the secondary aluminum industry as consisting of approximately 161 secondary aluminum production facilities, of which the EPA estimates 53 to be major sources of HAP. The commenter stated that it is a matter of continuing concern that the applicability definition for the current proposed revision to the subpart RRR rule still lists "Aluminum Foundry Facilities: NAICS 331524, and Aluminum Die-Casting Facilities: NAICS 331521" among the Industrial Source Categories affected by this proposed action. The commenter described the basis for their concerns regarding the potential erroneous classification of foundries and die casters as secondary aluminum production facilities. The commenter stated that there may still be reason to question the categorization of some of the facilities the EPA believes are actually subject to the rule, and therefore urge the agency to use caution in applying the applicability definition to the Industrial Source Categories it feels may be are affected by this proposed rulemaking.
Response: The list of industrial source categories, including aluminum foundry facilities and aluminum die-casting facilities, in Table 1 of the preamble to the supplemental proposal is a "guide for readers regarding the entities likely to be affected" and is not an "applicability definition" for the Subpart RRR rule. As stated plainly in the preamble, "[t]o determine whether your facility would be affected, you should examine the applicability criteria in the NESHAP." 79 FR at 72876. The applicability section of subpart RRR appears at 40 CFR 63.1500. The commenter did not suggest specific changes to the rule and we do believe that the comment warrants making any changes to the rule.
Comment #7: In a comment on the supplemental proposal, commenter 0296 stated that they concur with the Agency's determination that there have been no new developments in practices, processes or control technologies that are applicable to the secondary aluminum production source category.
Response: We acknowledge and appreciate the commenter's views.

