

DATE:	July 24, 2015

SUBJECT:	Technical Support Document for the Secondary Aluminum Production Source Category Final Rule

FROM:	Mark Bahner, RTI International

TO:		Rochelle Boyd, U.S. Environmental Protection Agency (the EPA)

The purpose of this memorandum is to document the technical approach and rationale used to develop certain modifications to the National Emission Standards for Hazardous Air Pollutants (NESHAP) for Secondary Aluminum Production (40 CFR Part 63, Subpart RRR).
Capture Efficiency Assumption for Uncontrolled Furnaces
Test data related to the capture efficiency for uncontrolled furnaces were analyzed. Two tests were available. Both were conducted at Southwire Corporation in Hawesville, KY. Testing was performed for PM, HCl and PCDD/PCDF from April 30  -  May 02, 2007 at Furnace #5 in Mill 3, and reported on June 26, 2007. The test length was approximately 8 hours, which corresponds to the batch length for Melter/Holder 5. Testing was performed for PM, HCl and PCDD/PCDF from July 17  -  19, 2007 at a Group 1 furnace located in Rod Mill 1 and reported on September 5, 2007. The test length was approximately 3 hours, which corresponds to the Melter 1 batch length[1]. These tests were performed on a furnace stack and also on a temporary canopy hood, which was installed for the test, to capture emissions not directed through the furnace stack.
The results of the two tests are summarized in Table 1. From Table 1, for all pollutants measured via the furnace stack, it can be seen that the mean capture efficiency for the furnace stack was 71 percent, and the median capture efficiency was 78 percent. The EPA has considered various potential furnace stack capture efficiency assumptions and resultant stack test estimation factors for uncontrolled furnaces in light of the available data. For example, if the furnace capture efficiency of an uncontrolled furnace is assumed to be 50 percent, the furnace test results would need to be multiplied by a factor of 2 to calculate total emissions, and if the furnace capture efficiency was assumed to be 66.67 percent (approximately 67 percent), the furnace test results would need to be multiplied by a factor of 1.5 to calculate total emissions. Choosing a value of 80 percent, close to the median capture efficiency of 78 percent and the average capture efficiency for the HAP with the highest capture efficiency, would mean that test results would need to be multiplied by a factor of 1.25. The EPA has chosen 80 percent as providing the best estimate of the capture efficiency of uncontrolled furnaces for the pollutants being measured, based on the limited data. 
Additional tables related to this analysis are included in Appendix A to this memorandum. The tables can be summarized as follows:
      Table A1: This table presents data from the June 26, 2007 report. The data included pollutant emissions in pounds per hour and test sampling time. These are converted into total mass measured, so that the total emissions from the furnace stack can be compared to the total emissions in the canopy hood stack. For example, the during Run 3, hydrogen chloride (HCl) emissions in the furnace stack were 5.31 pounds (lb), and 0.84 lb in the canopy hood, for a total of 6.15 lb.
      Table A2: This table presents data from the June 26, 2007 report. The data are exclusively in pounds and the table includes a calculation of the percentage of pollutants emitted through the canopy hood stack versus the total mass of pollutants emitted through both the canopy hood stack and the furnace stack. For example, the 3-run average fraction for HCl in the canopy hood is 0.19, meaning that 19 percent of the HCl passes through the canopy hood.
      Table A3: This table presents data from the September 5, 2007 report. The data included pollutant emissions in pounds per hour and test sampling time. These are converted into total mass measured, so that the total emissions from the furnace stack can be compared to the total emissions in the canopy hood stack. For example, the during Run 3, hydrogen chloride (HCl) emissions in the furnace stack were 2.18 pounds (lb), and 0.84 lb in the canopy hood, for a total of 3.02 lb.
       Table A4: This table presents data from the September 5, 2007 report. The data are exclusively in pounds and the table includes a calculation of the percentage of pollutants emitted through the canopy hood stack versus the total mass of pollutants emitted through both the canopy hood stack and the furnace stack. For example, the 3-run average fraction for HCl in the canopy hood is 0.23, meaning that 23 percent of the HCl passes through the canopy hood.

Costs for Lowering the Dioxin/Furan (D/F) Emissions Limit for Group 1 Furnaces
We performed an analysis to evaluate the option of lowering the D/F emissions limit from 15 to 10 ug TEQ/Mg for group 1 furnaces processing other than clean charge at all facilities. Based on testing reported in information collection request responses,[2] and supplemental information received from facilities after the 2012 Proposal,[3] there are twelve Group 1 furnaces with emissions above 10 ug/Mg, as shown in Table 2. Eleven of the twelve furnaces are presently controlled with fabric filters, and it is assumed for the purposes of this analysis that activated carbon injection would be added to increase the level of control of D/F at these furnaces. One furnace is presently uncontrolled and it is assumed that a direct lime injection fabric filter would be added. In both cases, it was assumed that D/F emissions would be reduced by 85%. The total capital cost for adding activated carbon injection on each of the eleven furnaces already equipped with a fabric filter is calculated to be $170,578, and the total annualized cost for each furnace is estimated to be $112,188. For the one furnace without a fabric filter, the capital cost of adding a direct injection fabric filter is calculated to be $3,174,419 and the total annualized cost is estimated to be $926,856. These costs are based on costs for similar controls estimated for the Brick and Structural Clay Products (BSCP) rule.[4] Therefore, the total annualized cost for the additional controls on all twelve furnaces is estimated to be approximately $2,160,000 (see Table 3). The total annual D/F emission reductions if all furnaces above 10 ug/Mg reduced emissions by 85% would be 2.826 grams TEQ. Therefore, the calculated cost effectiveness is approximately $765,000 per gram TEQ of D/F (see Table 4). 
Compliance with Emission Limits during Startup/Shutdown

The EPA is promulgating standards for startup and shutdown for all secondary aluminum process units. The subpart RRR standards will apply at all times, including periods of startup and shutdown. Because the scrap processed at secondary aluminum production facilities is the source of emissions, emissions during startup and shutdown are expected to be no higher and probably much lower than emissions during normal operations since no scrap would be processed. 

For production processes in the secondary aluminum production source category where the standards are expressed in units of pounds per ton of feed or similar units (i.e. thermal chip dyers, scrap dryer/delacquering kiln/decoating kilns, dross-only furnaces, in-line fluxers using reactive flux and group 1 furnaces), the form of the standard would raise implementation issues if facilities were simply required to demonstrate compliance in the exact same way as they do during normal operations because the equipment is not receiving feed material during startup and shutdown periods. For example, the startup procedure for a furnace consists of gradually heating the furnace to normal operating temperature, without any addition of feed material. Since the feed rate is zero, the emission rate per ton of feed would be calculated to be infinite. 

One possibility would be to have concentration-based limits. However, these are probably not appropriate for the secondary aluminum industry, because hooding draws in large amounts of ambient air in order to increase hooding capture. Therefore, it is expected that emissions during startup and shutdown could be measured and determined under the applicable test method in pounds per hour (lb/hr). To demonstrate compliance during the startup and shutdown periods, for emissions standards that are in units of lbs per ton, the emissions in units of pounds per hour would be divided by the feed rate in tons per hour (ton/hr) from the most recent or current performance test and compared to the emission limit for the affected source. For example, the PM limit for a new or existing group 1 furnace is 0.4 lb/ton. If, during a startup or shutdown, the emission rate determined under the applicable test method is 2 lb/hr, and the furnace charge rate during the most recent performance test was 10 tons per hour, the calculated emission rate for that test during a period of startup or shutdown would be 0.2 lb/ton of feed (2 lb/hr of PM divided by 10 tons per hour of feed). This would be in compliance with the PM standard of 0.4 lb/ton of feed. If no performance test results were available, the furnace capacity (tons per hour of feed) could be used. Another possibility would be to demonstrate compliance by keeping records that show that during startup and shutdown no feed/charge or flux was added, and that only clean fuel was used or no fuel was used.

Representativeness for Nine-Company HAP Metals and HAP Organics Testing
The EPA used a nine-company ICR to gather testing data that would correlate emissions of THC surrogate to speciated organic HAPs and emissions of PM surrogate to speciated metal HAPs. Testing was limited to nine companies for efficiency and to avoid undue burden to the industry. The nine companies were chosen primarily because they were large companies operating many facilities, and because these companies are known to operate major source facilities that have thermal chip dryers or scrap dryer/delacquering/decoating kilns. Thermal chip dryers and scrap dryer/delacquering decoating kilns are the only types of equipment that have THC limits. Testing for these two pieces of equipment were limiting, in that few companies in the secondary aluminum source category currently operate those pieces of equipment. Therefore, the sampling of the pieces of equipment was quite representative, because the sampling represented a large fraction of the equipment operated by the industry. The sampling for HAP metals in fabric filter dust represented a much smaller fraction of the industry (less than 10 percent of all equipment). However, there were a total of 55 samples analyzed for the full suite of HAP metals (including arsenic, cadmium, total chromium and hexavalent chromium, manganese, mercury, nickel, etc.). Because 55 samples were taken from many facilities owned by nine companies, these results are believed to be representative of metal concentrations in fabric filter dust within the secondary aluminum industry and are suitable for use in developing speciated metal HAP emissions estimates from PM emissions for metal HAP that are mainly, or entirely, in particulate form (such as Pb, Ni, Cr, Mn, and Cd). More information regarding the data collected and the methodology to calculate emissions is available in the memorandum, " Development of the RTR Supplemental Proposal Risk Modeling Dataset for the Secondary Aluminum Production Source Category," which is available in the docket for this action.[3]
References
 EPA docket item # EPA-HQ-OAR-2010-0544-0204. Available at: http://www.regulations.gov/#!documentDetail;D=EPA-HQ-OAR-2010-0544-0204.
 Letter, Peter Tsirigotis (U.S. EPA) to Russell Mayfield (Mayfield Salvage Company) et al., "Requirement to provide information according to 42 U.S.C 7414," February 7, 2011. This was the cover letter for the all-company data collection ICR.
 Memorandum, Mark Bahner (RTI, International) to Rochelle Boyd (U.S. EPA), "Development of the RTR Supplemental Proposal Risk Modeling Dataset for the Secondary Aluminum Production Source Category" November 12, 2014. EPA docket item # EPA-HQ-OAR-2010-0544-0239. Available at: http://www.regulations.gov/#!documentDetail;D=EPA-HQ-OAR-2010-0544-0239. This memorandum also has a table that correlates RTI facility ID numbers with facility names and locations.
 Memorandum, Thomas Holloway (RTI, International) to Jeff Telander (U.S. EPA), "Methodology and Assumptions Used to Estimate the Model Costs and Impacts of BSCP Air Pollution Control Devices," November 6, 2014.


                                List of Tables


Table 1. Southwire Test Results Summary	6
Table 2. Group 1 Furnaces with Reported Emissions Greater than 10 ug TEQ/Mg Feed	6
Table 3. Group 1 Furnace D/F Reduction Options	7
Table 4. Cost-effectiveness Calculation for D/F Reduction Options	7

Table A1.  Southwire Testing Reported June 26, 2007: Complete Test Results	9
Table A2.  Southwire Testing Reported on June 26, 2007: Results in Pounds	10
Table A3.  Southwire Testing Reported September 5, 2007: Complete Test Results	11
Table A4.  Southwire Testing Reported on September 5, 2007: Results in Pounds	12


Table 1. 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
Mean for all
                                      All
                                     0.71
                                     0.29
Median for all
                                      All
                                     0.78
                                     0.22
   *PM = Particulate Matter; HCl = Hydrogen chloride gas; D/F = Dioxins/Furans


Table 2. Group 1 Furnaces with Reported Emissions Greater than 10 ug TEQ/Mg Feed


Table 3. Group 1 Furnace D/F Reduction Options 
Option
                                Furnace Status
                              Number of Furnaces
                         Capital Cost[1] Each (2011 $)
                          Total Annualized  Cost Each
                                   (2011 $)
                      Combined Total Capital Cost (2011$)
                    Combined Total Annualized Cost (2011$)
Limit of 10 ug/Mg TEQ for all
Controlled
                                      11
                                    170,578
                                    112,188
                                  $5,050,000
                                   2,160,000
Limit of 10 ug/Mg TEQ for all
Uncontrolled
                                       1
                                   3,174,419
                                    926,856
                                       
                                       
[1] Controlled furnaces have fabric filters, and are assumed to add activated carbon injection. Uncontrolled furnaces are assumed to add direct injection fabric filters (DIFF).

Table 4. Cost-effectiveness Calculation for D/F Reduction Options 
Option
                    Combined Total Annualized Cost (2011$)
                   Annual D/F Emission Reduction (grams TEQ)
                        D/F Removal Cost Effectiveness
                                 ($/gram TEQ)
Limit of 10 ug/Mg TEQ, assuming additional controls added that reduce D/F emissions by 85%
                                   2,160,000
                                     2.826
                                    765,000

                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                  Appendix A
                                       
                            Southwire Test Results
                                       
                                       


Table A1.  Southwire Testing Reported June 26, 2007: Complete Test Results 
* reported as 2378-TCDD equivalent



Table A2.  Southwire Testing Reported on June 26, 2007: Results in Pounds



Table A3.  Southwire Testing Reported September 5, 2007: Complete Test Results




Table A4.  Southwire Testing Reported on September 5, 2007: Results in Pounds

