Technical Memorandum. 1.0 Deposition Algorithm

Transcription

1 Technical Memorandum To: Jim Sullivan, Minnesota Pollution Control Agency From: Pat Sheehy; Todd Fasking and Eric Edwalds, Barr Engineering Subject: NorthMet Plant Site Class II Protocol Addendum Date: July 6, 2016 Project: c: Kevin Pylka, PolyMet; Steve Sommer, Minnesota Pollution Control Agency The purpose of this memo is to provide and addendum to the Class II modeling protocol submitted for the NorthMet Project (Project) Plant Site submitted to the Minnesota Pollution Control Agency (MPCA) on April 29, This addendum addresses comments received from MPCA on the protocol and provided additional requested information. The following sections described the proposed procedure for addressing the effect of particulate deposition on modeled PM 10 concentration, the use of the Adjusted u* beta option in AERMOD to address low wind speed conditions and the proposed approach for maintaining the ambient air boundary for the Plant Site. These options are described in Section 1, 2, and 3 below, respectively. 1.0 Deposition Algorithm The default AERMOD deposition algorithm is proposed for the particulate matter less than 10 microns in aerodynamic diameter (PM 10) modeling to be submitted with the air emission permit application. Where AP-42 emission factors are used in the emissions calculations and different emission factors are used for PM 10 and particulate matter less than 2.5 microns in aerodynamic diameter (PM2.5), the fraction of particulate in each size category, based on the emission factors, will be used in the modeling. Where AP- 42 emission factors are calculated through the use of correlation equations, such as for unpaved roads, outdoor material handling sources, and tailings basin wind erosion (from AP-42 Sections , , and respectively) the ratio of the particle size multipliers for the fractions equivalent to PM 10 or smaller particles will be used to calculate the relative fraction of emissions in each size category. Note that the threshold friction velocity used for calculating wind erosion potential was developed separately for LTV Steel Mining Company (LTVSMC) tailings and NorthMet tailings based on overall size distribution following the procedures in AP-42 Section The particle size distributions were obtained from pilot study data for the NorthMet tailings and samples collected on-site for the LTVSMC tailings. The particle size multipliers in AP-42 Section were used to develop the Tailings Basin PM 10 and PM 2.5 wind erosion emission rates and therefore will be used for developing the appropriate particle size fractions for deposition modeling. In the case of sources for which all particulate emissions are assumed to be PM 2.5, all emissions will be assigned to the PM 2.5 size category. Barr Engineering Co MarketPointe Drive, Suite 200, Minneapolis, MN

2 To: Jim Sullivan, Minnesota Pollution Control Agency From: Pat Sheehy; Todd Fasking and Eric Edwalds, Barr Engineering Subject: NorthMet Plant Site Class II Protocol Addendum Date: July 6, 2016 Page: Ustar Adjustment Option The preferred air quality model AERMOD has a bias to overestimate concentrations for surface releases under low-wind stable conditions (Paine and Connors, 2013). A major reason for AERMOD s bias to overestimate atmospheric concentrations is that a default algorithm in AERMET underestimates the surface fraction velocity (u* or Ustar) for low-wind stable conditions. Recent versions of AERMET (version and later) provide an optional method for calculating u* for low-wind stable conditions (U.S. EPA, 2014). For the low-level volume sources and ground-level area sources found at PolyMet s Plant Site, the use of the ADJ_U* beta option in AERMET greatly reduces, but does not completely eliminate, the bias for the AERMOD/AERMET modeling system to overestimate atmospheric concentrations. Because the preferred air quality model (the default AERMOD/AERMET 3 modeling system) is less appropriate than using the ADJ_U* beta option in AERMET for modeling emissions from the Plant Site, the Plant Site Air Quality Modeling Protocol proposes the use of the ADJ_U* beta option in AERMET in the preparation of weather input files. See the separate technical memorandum submitted with this addendum NorthMet Class II Air Quality Modeling Protocols Ustar Adjustment Option for additional detail on the application of the u* Adjustment Option for the Plant Site air quality modeling. 3.0 Ambient Air Boundary The proposed ambient air boundary for the Plant Site is unchanged from the protocol previously submitted, with the exception of a small change on the northwest end of the site to move the ambient air boundary south and east of a snowmobile trail that runs east-west along the power line corridor and north-south along a trail. Please see attached technical memo NorthMet Ambient Air Boundary included as Attachment 2 for further explanation and figures showing the change. Access to the area excluded from ambient air will be restricted through a combination of locked or monitored gates, posting, patrols, security cameras, access agreements with other entities and the ability of terrain and vegetation to restrict access to the Plant Site. A detailed description of the how the ambient air boundary will be controlled will be provided in a separate technical memo NorthMet Ambient Air Boundary. Reference: Paine, B. and Connors J AERMOD low wind speed issues: Review of new model release. EPA Regional/State/Local Modelers Workshop, April 22-25, Dallas, TX. USEPA, Addendum. User s Guide for the AERMOD Meteorological Preprocessor (AERMET). U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards, Research Triangle Park, NC. EPA-454-B P:\Mpls\23 MN\69\ \WorkFiles\APA\Permitting\Air Permitting\Class II Modeling Plant Site\Protocol\With U star\plant Site Prot Addendum C2 memo.docx

3 POLYMET MINING April 28, 2016 Air Quality Permit Document Coordinator Minnesota Pollution Control Agency 520 Lafayette Road North St. Paul, MN DearSir/Madame, Please find attached two paper copies of the completed Air Quality Dispersion Modeling Protocol form (AQ.DM-01) submitted for the Poly Met Mining, Inc. Plant Site as part of the NorthMet Project. If you have any questions regarding this submittal, please contact me at Sincerely, Kevin Pylka Manager of Environmental Permitting and Compliance

4 Plant Site Class II Air Quality Dispersion Modeling Protocol Version 3 April 2016 NorthMet Project PolyMet Mining Incorporated Hoyt Lakes, MN This document provides the Class II dispersion modeling protocol for the Plant Site in the format requested by the Minnesota Pollution Control Agency (MPCA). This includes MPCA form AQDM-01, four figures and two attachments. The figures and attachments are listed below. Figures: Large Figure 1 Process Plant Model Source Layout Large Figure 2 Tailings Basin Model Source Layout Large Figure 3 Plant Site Ambient Air Boundary Receptor Grid - Close-Up Large Figure 4 Cumulative Impacts Assessment Boundary Receptor Grid - Close-Up Attachments (name references corresponding section of the AQDM-01 form): Attachment C Non-Default Modeling Options Attachment J Background Concentrations and Determination of Nearby Sources to be Modeled The AQMD-02 spreadsheet is incorporated in the Emission Inventory spreadsheet which was previously submitted to MPCA on March 26, Supporting files to accompany the protocol are posted to the project website.

5 AQDM-01 Air Quality Dispersion Modeling Protocol Protocol Form for Criteria Pollutant Modeling (Previously AQDMP-01) Doc Type: Air Dispersion Modeling Guidance Information on Page 14 Instructions: Permit applicants required to conduct air dispersion modeling should submit two paper copies of the completed Air Quality Dispersion Modeling Protocol form (AQDM-01), the Air Quality Dispersion Modeling Protocol Spreadsheet (AQDM-02), and all accompanying files to: Air Quality Permit Document Coordinator Minnesota Pollution Control Agency 520 Lafayette Road North St. Paul, MN Applicants may also submit an electronic version, in addition to the two required paper copies. Please note that all assumptions made in the air dispersion modeling analysis could result in air permit requirements. Electronic copies of the forms and accompanying files should be sent to: Only use the most current version of this form that is on the website versions prior to 01/24/2012 will not be accepted. Facility Information AQ facility/permit ID no.: Today s date (mm/dd/yyyy): 04/28/2016 Three-letter modeling facility ID (ex., XEK = Xcel Energy Allen S. King, MEC = Mankato Energy Center, etc.): Facility name: PolyMet Mining Inc. - Plant Site Facility street address: 6500 County Road 666 City: Hoyt Lakes County: St. Louis Zip code: State: MN Facility contact: Kevin Pylka Protocol prepared by: Eric Edwalds (Barr Engineering Co.) Facility contact phone: (218) Preparer phone: (952) Facility contact address: kpylka@polymetmining.co m Preparer address: eedwalds@barr.com *UTM coordinates of facility (NAD83, zone 15 extended ONLY): x = 564, m East, y = 5,271, m North *This should be the central location of the facility/source. PMP Files to accompany protocol Use the checkboxes to indicate that the following required files are included with the completed protocol form (unless otherwise noted). Please do not use spaces or special characters in the file names and pathways. 1. Sample AERMOD input files for each modeled criteria pollutant (*.inp, *.adi, *.ami) *Note: Input file should include receptor grid and building downwash (if applicable) 2. AQDM-02 form (spreadsheet) 3. BPIPPRM input file (*.bpi) (if applicable) 4. Elevation files for input into AERMAP (*.tif [NED files]) (if applicable) 5. Background data files with concentrations for each applicable pollutant (annual, seasonal, monthly, and daily, etc.) (if applicable) 6. Optional, but recommended, files and supporting documents Please list below: TTY or Available in alternative formats aq2-40 7/10/13 Page 1 of 16

6 Examples include: sample AERMOD output files (with CO RUNORNOT NOT ), hourly emissions file, images and figures, SIL analysis and/or paved roads fugitive dust modeling output files, etc. Large Figure 1 Process Plant Model Source Layout; Large Figure 2 - Tailings Basin Model Source Layout; Large Figure 3 Plant Site Ambient Air Boundary Receptor Grid Close-Up; Large Figure 4 Plant Site Cumulative Impacts Assessment Boundary Receptor Grid Close-Up Attachment C - Non-Default Model Options; Attachment J Background Concentrations and Determination of Nearby Sources to be Modeled; The AQMD-02 spreadsheet is incorporated in the Emission Inventory spreadsheet which was previously submitted to MPCA March 26, Modeling files (AERMOD input, receptor grids, hourly emission rate files) are included electronically. Section A. Purpose for Air Dispersion Modeling and Related Information 1. What is the purpose for conducting the ambient air dispersion modeling? (check all that apply) EAW EIS SIP PSD Permit requirement please specify permit requirement information (i.e., equivalent dispersion demonstration): Other please explain: Discretionary request from MPCA a. If EAW and/or EIS are selected, please specify the regulatory trigger for modeling (ex., air emissions increase of 250 TPY, 25 MW operating capacity or design, petition, voluntary, etc. see Minn. R. ch for more information) and the name of MPCA Environmental Review staff consulted with (N/A if not applicable or if no Environmental Review staff were consulted): MPCA EAW staff name(s): Date of consultation (mm/dd/yyyy): *Note: If EAW and/or EIS are to be performed and air modeling will be conducted, a cumulative effects analysis will need to be addressed, as per the CARD decision. Please then select NAAQS/MAAQS in question 8 (A.8). Contact the Environmental Review Unit Supervisor for any questions regarding applicability and requirements. 2. Were MPCA air dispersion modeling staff consulted while completing this form? Yes No a. If yes, please provide the following consultation information: MPCA modeling staff name(s): Jim Sullivan, Daniel Dix Date of consultation (mm/dd/yyyy): Multiple Topic of consultation: Expectations for protocol including meteorological data, pollutant receptor grids,and protocol elements. 3. What type of air emission permit does this facility currently hold? No current permit Federal (Title V/Part 70) State State Registration Capped General Other: 4. Will you be applying for a permit or a permit amendment for the project? Yes No 5. Please provide a project title and a description of the proposed project: a. Project title (10 words or less): NorthMet Project - Plant Site b. Project description (50 words or less): (Additional information can be provided in question 9 below) PolyMet plans to construct and operate a mine, to reactivate portions of the LTV Steel Mining Company facility and to build a hydrometallurgical concentrate processing facility at the former LTVSMC site. This protocol covers the Plant Site. More detail is available in the NorthMet Project Description Version 9 Submitted February 19, Is the proposed project subject to PSD? Yes No a. If yes, list pollutants: b. Is this facility considered a major source for PSD: Yes No 7. Has the PSD minor source baseline been set for: St. Louis County? Yes No a. If yes, for which pollutant(s) and the year(s) it was set (check all that apply)? NO PM PM2.5 SO PSD major source baseline (PM2.5 trigger date will be on Oct. 20, 2011, 1 year after F.R. publication date.) TTY or Available in alternative formats aq2-40 7/10/13 Page 2 of 16

7 NO PM PM SO What type of analysis will be conducted? (check all that apply) NAAQS/MAAQS PSD Class II Increments PSD Class I Increments SIL Analysis Screening Other: AERA 9. Additional information for this section that was not included above (if not applicable, place N/A in field): (Examples include: more information about the proposed project, whether the facility is in any SIP non-attainment or reclassified areas, etc.) The Project is a PSD synthetic minor and therefore the project does not trigger PSD modeling. The modeling approach in this protocol follows the approved methodologies used in the discretionary modeling completed for the Draft and Supplemental Draft EIS with updates that are limited to changes in guidance and software. Section B. EPA Pre-Processors and EPA Post-Processors 1. Will AERMAP be used? Yes No If no, please explain: a. What version of AERMAP is proposed to be used: AERMAP version If other, please explain: b. What type of elevation data will be used: NED 1/3 arc second NED 1 arc second Other - Please describe: MN DNR LiDAR aerial survey 2011 ftp://ftp.gisdata.mn.gov/pub/gdrs/data/pub/us_mn_state_mngeo/elev_lidar_arrowhead2011/metad ata/metadata.html All UTM coordinates must be in NAD83, Zone 15 Extended (not NAD27). 2. Will BPIP-PRIME version be used? Yes No If no, please explain: Tiering of buildings must follow MPCA s modeling guidance from the How To Model section of the April 2013 MPCA Air Dispersion Modeling Guidance For Minnesota Title V Modeling Requirements And Federal Prevention of Significant Deterioration (PSD) Requirements. (aq2-58) 3. Will MPCA pre-processed AERMET data be used? Yes No If yes, proceed to question 4. Note MPCA s pre-processed meteorological data with AERMET incorporates the following details: AERSURFACE version (or version 13016) is used to determine surface characteristics using 1992 LULC data. Yearly-averaged moisture conditions (wet, dry, or average) based on historical ranks are accounted for in AERSURFACE to aid in the determination of Bowen ratio values. Default 1.0 km radius for surface roughness and 10 km by 10 km domain for albedo and Bowen ratio used in AERSURFACE a. If no, will on-site meteorological data be processed and used? Yes No * If no, skip to question 4 and provide additional information in question 6. b. If yes to question a, please answer the following questions. i. Will AERSURFACE be used to determine surface characteristics around the meteorological tower? (Default is Yes ) Yes No ii. What version of AERSURFACE is proposed to be used: [Select from list] iii. What LULC data will be used? [Select from list] iv. If other, please explain: Will yearly-averaged moisture conditions (wet, dry, or average) based on historical ranks be accounted for in AERSURFACE (for the Bowen Ratio)? (Default is Yes ) Yes No v. Will the default 1.0 km radius for surface roughness, and 10 km by 10 km domain for albedo and Bowen ratio be used? (Default is Yes ) Yes No vi. If no, please explain: Please provide any additional processing details of the on-site meteorological data that is not provided above: 4. Are any EPA post-processors (such as LEADPOST) proposed to be used in the analysis? Yes No a. If yes, what post-processor(s) and version(s): TTY or Available in alternative formats aq2-40 7/10/13 Page 3 of 16

8 5. Are any user-developed pre-processors or post-processors proposed to be used in the analysis? Yes No a. If yes, what pre-processors or post-processors, and describe their functions: 6. Additional information for this section that was not included above (if not applicable, place N/A in field): The meteorological data set is Hibbing processed using AERMINUTE The MPCA reviewed and approved (3/25/2016 from Daniel Dix to Eric Edwalds; cc: Jim Sullivan) this data set for this modeling analysis. Section C. Model Selection and Options (Key CO Pathway Inputs) 1. Identify the air dispersion model and version proposed to be used in the analysis: Other a. If other, please list: AERMOD Version What criteria pollutants are required and will be modeled (check all that apply)? CO NO2 PM2.5 PM10 SO2 Pb Other: Please refer to Tables App.1 and App.2 in the Appendix for averaging times and form of standard for each criteria pollutant(s). Refer to the most recent version of the EPA s AERMOD User s Guide for correct pollutant IDs to use. Use EPA s most recent modeling guidance methods for PM What model options (CO pathway keywords) are proposed to be used in the analysis for the source under review (check all that apply)? (See Tables B-1 and B-2 of the AERMOD User s Guide Addendum ) Regulatory Default - list pollutants: SO2 Non-Default - List pollutants: NO2, PM10, PM2.5 Concentration Rural Urban ELEV FLAT BETA PVMRM OLM FLAGPOLE Other(s) not listed: a. If Urban, please specify population area, population, and surface roughness radius: Population area: Population: Surface roughness height: km b. If Non-Regulatory Default, please specify non-default options: OLM (1 hour NO2), deposition algorithm based on AERMOD Half-Life function (PM10), and Secondary Formation from Combustion Sources (PM2.5) see Attachment C for more detail c. If BETA is selected, please specify BETA options: (BETA options LOWWIND1 and LOWWIND2 are not to be used for regulatory purposes, and will not be accepted or review by the MPCA.) 4. Will alternative air dispersion models and/or methods, as specified by Appendix W, or applicable guidance, be applied (e.g., parallel version(s) of model, PVMRM, OLM, secondary formation coefficients, etc.)? Yes No Skip to question 5. The 1 hour NO2 SIL analysis assumes OLM and the options selected are included in question 5 below. Attachment C describes the proposed PM2.5 modeling methodology to account for If yes, please explain: secondary formation from the combustion sources at the Plant Site. a. If yes, will approval be required by MPCA and/or EPA Region V air modeling staff? Yes No OLM If yes, please select: MPCA (State-only action) EPA Region V (PSD/SIP action) If no, please explain: 5. If NO2 is required to be analyzed for the one-hour and annual NAAQS, what tier methodology(s) is proposed? N/A, NO2 not required (skip to question 6) Tier 1 (100% NOx to NO2 conversion, most conservative) Tier 2 (Default ambient ratio of 0.80, or an appropriate ratio value) Tier 3 (OLM, requires justification and approval by MPCA and/or EPA Region 5)* Tier 3 (PVMRM, requires justification and approval by MPCA and/or EPA Region 5)* *For Tier 3 options, please fill-out the NO2 questions in Section K: Pollutant-based Considerations. a. If Tier 2, please provide the ambient ratio proposed (default = 0.80): TTY or Available in alternative formats aq2-40 7/10/13 Page 4 of 16

9 b. If Tier 2, please provide justification for proposed ambient ratio: 6. If an alternative refined model is proposed for use (including non-default options like PVMRM or OLM), will the following criteria from Appendix W, section 3.2.2, paragraph (e), be met in this protocol? Yes No N/A If no, please explain: e. Finally, for condition (3) in paragraph (b) of this subsection an alternative refined model may be used provided that: i. The model has received a scientific peer review; ii. iii. iv. The model can be demonstrated to be applicable to the problem on a theoretical basis; The data bases which are necessary to perform the analysis are available and adequate; Appropriate performance evaluations of the model have shown that the model is not biased toward underestimates; and v. A protocol on methods and procedures to be followed has been established. 7. Additional information for this section that was not included above (including justification for non-default options, additional CO pathway keywords not mentioned above, values for options selected, etc.): Attachment C included with this protocol describes the Half-Life deposition method used for the PM10 modeling. Attachment C also describes the MPCA offset ratio method used to account for secondary formation of PM2.5 from combustion sources. The emission inventory previously submitted (3/26/2016) contains the offset ratio calculations. While the BETA option of adjusted ustar is available and has been approved for similar sources (EPA Region 10), PolyMet has chosen at this time not to pursue approval and application of this method. The limiting sources in the PM10 modeling (fugitive emissions generated near the ground) and the meteorological conditions (all of the maximum modeled impacts from the Plant Site sources occur for wind speeds < 1 m/s) are consistent with conditions for which the adjusted ustar BETA option was developed. While it is expected that even lower levels of impact would be modeled if the adjusted ustar BETA option were used, it is expected that compliance will be modeled successfully without using this methodology. The absence of the adjusted ustar BETA option applies an additional level of conservativism to modeled impacts associated with the described sources and conditions. Section D. Emission Source Characterizations and Parameters (Key SO Pathway Inputs) Include and list the facility s modeling parameters for all source types in the MPCA s Modeling Parameters Spreadsheet (Form AQDM-02). For background sources listed within SO Pathway, please see Section I. 1. Please indicate which of the following source characterizations are present at your facility and will be included for modeling analysis (check all that apply ): N/A = Source-type not present; Yes = Source-type present and will model; No = Source-type present but will not model. a. Point sources: N/A Yes No-please explain: i. Are any of the point sources capped and/or horizontal stacks (see guidance in section 6.1, AERMOD Implementation Guide (03/19/2009)) and, if yes, accounted for in the following? ii. No Yes exit velocity(s) = m/s Yes Non-Default POINTCAP and /or POINTHOR* *Please provide justification for use of non-default option in question ii, below. If using third-party software, please check with the vendor to determine if there are any issues with the use of BETA options POINTCAP and POINTHOR. Additional information for this subsection (if not applicable, place N/A in field): N/A b. Volume sources: N/A Yes No -please explain: Tip: Please refer to Figure App.1 in the appendix on calculating the lateral and vertical dimensions. i. Will there be any volume source(s) overlapping or within (2.15 * sigma Y) meters of any receptors? ii. No Yes * * Volume source should then be converted to an area source of commensurate size (per section 6.2 of EPA s AERMOD Implementation Guide (03/19/2009) or be further refined. Also see EPA s Haul Road Workgroup Guidance (03/02/2012). Additional information for this subsection (if not applicable, place N/A in field): TTY or Available in alternative formats aq2-40 7/10/13 Page 5 of 16

10 The Plant Site volume sources represent the fugitive dust from the vehicles traveling around the tailings basin and along the unpaved Dunka Road, tailings material handling, and building vents. The space heaters and other sources potentially venting into buildings are vented out of the roof of numerous plant site buildings that are primarily represented as volume sources. c. Area sources (includes AREACIRC and AREAPOLY): N/A Yes No-please explain: i. Additional information for this subsection (if not applicable, place N/A in field): The tailings basin erodible beach areas are represented by multiple area sources. The details are included in the previously submitted emission inventory spreadsheet. All beach areas are shown in Large Figure 2; a subset of these would be erodible at a given point in time. Either all beach areas will be modeled as erodible or the worst case erodible beach area location will be confirmed through modeling. d. Open pit sources: N/A Yes No-please explain: i. Additional information for this subsection (if not applicable, place N/A in field): N/A e. Line sources: N/A Yes No please explain: i. Additional information for this subsection (if not applicable, place N/A in field): N/A 2. Are fugitive emissions emitted from the source, and if yes, will they be accounted for in the modeling analysis? (Examples of fugitive emissions include but are not limited to: traffic on paved and/or unpaved roads, stockpiles of various materials, wind erosion, loadout, unloading, etc.) (N/A = no fugitive emissions are emitted; Yes = fugitive emissions are emitted AND will be included in modeling analysis; No = fugitive emissions are emitted and will NOT be included in modeling analysis) N/A Yes No-please explain*: * Please provide justification for excluding any fugitive emission sources within the modeling, including the use of the Paved Roads Fugitive Dust Policy. a. If yes, please list the facilities fugitive sources: material handling, unpaved road traffic, conveying, wind erosion Note: If modeling for paved road fugitive dust, please read and complete Section E. 3. Will all insignificant activities emitted from the source for PM10 (with emissions over 0.1 lb/hr), PM2.5 (with emissions over 0.02 lb/hr), or other pollutants be accounted for in the modeling? (N/A = no insignificant activities are emitted from the source; Yes = insignificant activities are emitted AND will be included in modeling analysis; No = insignificant activities are emitted and will NOT be included in modeling analysis) N/A Yes No a. If no, please provide justification for excluding any insignificant activities within the modeling: Refer to guidance under the General Modeling Information section of the April 2013 MPCA Air Dispersion Modeling Guidance For Minnesota Title V Modeling Requirements And Federal Prevention of Significant Deterioration (PSD) Requirements (aq2-58), under the Insignificant Activities subsection. 4. Are intermittent emissions sources (as described in the Intermittent Emissions subsection of the MPCA Air Dispersion Modeling Guidance) present at the facility? Yes No Skip to question 5. a. If yes, please provide detail on the types and operations of the intermittent emission source(s) for each applicable source (include the source IDs, any regular uses, testing frequencies, emergency uses, peaking vs. non-peaking units, days and hours of operation or testing, etc.): SV108/109/426 are backup generator stacks and SV304/305 are fire pump stacks. Operation beyond 1-hour per day would be indicative of an emergency situation where the other plant sources would not continue operating. b. If yes to question 4, will intermittent emission sources be included in the modeling analysis? Yes please list the source ID(s): TTY or Available in alternative formats aq2-40 7/10/13 Page 6 of 16

11 No please list the source ID(s): SV108, SV109, SV304, SV305, SV426 i. If no, please provide an explanation for not including the above sources in the modeling analysis: Best management practices will be used for the operation, maintenance and testing of these sources. When these sources are operating for more than testing and maintenance, that means other facility sources are not operating (i.e. a power interruption or other upset condition is occurring). 5. Does your facility have alternative operating scenarios? Yes No Skip to question 6. a. If yes, will multiple operating scenarios be modeled? Yes - # of scenarios to be modeled: No most conservative scenario will be modeled b. If yes to 5a above, please describe the operating scenarios and the differences between them: Note: If multiple scenarios will be modeled, please list the scenarios in the AQDM-02 spreadsheet. 6. Will emission factors/scalars (SO EMISFACT) be used to demonstrate compliance in the air dispersion modeling analysis? Yes No Skip to question 7. a. If yes, describe for which sources and the types of emission factors/scalars that will be applied: By Wind Speed: LCRSTA-K, PBLKTA-I ; By Month/Hour/Day: Tailings basin construction traffic. 7. Will hourly emissions file(s) be used for this analysis (HOUREMIS)? Yes* No *If yes, please provide the hourly emission file(s) with the submittal of this form and list file in question 6 under the Files To Accompany Protocol section. 8. Will all applicable PSD increment consuming and/or expanding sources be modeled for your source? Yes No-please explain (SIL-only, NAAQS-only, etc.): 9. Additional information for this section that was not included above, including discussion of options (if not applicable, place N/A in field): Wind speed scalars are used for the tailings wind erosion sources. The tailings wind erosion emissions are calculated using the wind erosion equation (which is non-linear with respect to wind speed). The month/hour/day scalars represent the construction truck fugitive dust during the tailings dam construction schedule. The hourly emissions files are used for the material handling sources which have wind speed as part of the emission factor. AQDM-02 spreadsheet is included with the emission inventory previously submitted (3/26/2016). Section E. Paved Roads Fugitive Dust Current MPCA policy regarding modeling of paved road fugitive dust emissions, in support of air quality permitting or environmental review, recommends that: New facilities or facilities undergoing physical expansions will not be required to model paved road fugitive dust emissions if a facilities predicted ambient impacts for PM10 and PM2.5 are less than a specified percent of the NAAQS and/or PSD Class II Increment. Physical expansions do not include increases in emission limits (emphasis added). This policy does not apply to modeling that supports permitting in maintenance areas or the development of State Implementation Plans. Exceptions to the policy can and will occur. Please see the MPCA air modeling webpage for policy. If paved road fugitive dust emissions are proposed to be included in your modeling analysis and you did not answer Neither to question 1b, MPCA guidance recommends that the source in question first model its facility (without including paved road fugitive dust emissions) plus nearby sources plus background (i.e., FAC w/o paved roads + nearby source impacts + background impacts). If your facility has multiple paved roads operating scenarios, results recorded in Table E-01 must reflect the most conservative scenario. Results can be recorded using Table E-01 below for PM10 and PM2.5. If not elected to include results with this protocol (using the methodology above), then please provide the results in the AQDM-06 (modeling report) form. 1. Does your facility have paved road fugitive dust emissions for PM10 and PM2.5? Yes No Please continue to the next section (F) a. Will your facility include paved road fugitive dust emissions in the modeling analysis? Yes No b. Is your facility either a new facility or an existing facility undergoing a physical expansion? New facility Existing w/ phys. expan. Neither * * If neither, policy does not apply. Please answer questions 2 4, and proceed to the next section (F) without filling in Table E-01. c. If yes to question 1 and no to question 1a, please provide justification for not including paved road fugitives: TTY or Available in alternative formats aq2-40 7/10/13 Page 7 of 16

12 2. How many vehicles per day drive on and off your facility s property? (Provide the maximum number from all scenarios.) Employee traffic and parking: 150 Third-party truck traffic: Does your facility have multiple operating scenarios for traffic on your property (i.e., seasonal traffic changes)? Yes No a. If yes, please provide additional details for the operating scenarios, such as changes in traffic counts, types of vehicles, silt loadings, cleaning frequencies, etc. 4. Additional information for this section that was not included above or below, including discussion of options and how paved road fugitive sources will characterized in the model (if not applicable, place N/A in field): Although the Project is being permitted as a new source, the Neither box was checked in item 1.b as this option seems most appropriate for this facility because the Project uses existing infrastructure (including paved roads) from previous industrial activity at the site. Paved roads were used at the previous facility at the time of the PM10 minor source baseline date, so paved road emissions do not consume increment.activities at the baseline date included higher staffing levels and production levels, so traffic levels were greater than would occur under NorthMet. Note: If completing Table E-01, modeling output files must then be submitted. Table E-01 (Use max modeled concentrations from all operating scenarios) Averaging Period NAAQS (μg/m 3 ) Modeled NAAQS Impact Concentrations w/ Background and Nearby Sources(μg/m 3 ) % of NAAQS/MAAQS PSD Class II Increments (μg/m 3 ) Modeled Class II Increment Impact Concentrations (μg/m 3 ) % of Class II Increments PM10 PM hour % % Annual % % 24-hour % % Annual % % Table E-02 indicates the resultant category(s) for your facility, based on the % of the standard(s) for PM 10 and PM 2.5 (see results in columns % of NAAQS and % of Class II Increments in Table E-01 above). This uses the highest % from all averaging periods for each pollutant and standard. The category descriptions are provided in Table E-03. Answers to question 2 above will help determine permit conditions if modeled concentrations result in a category 2 designation. Table E-02 NAAQS NAAQS/MAAQS Result(s) w/ Background and Nearby Sources (%) Cat. 1 Cat. 2 Cat. 3 PSD Class II Increments PSD Class II Result(s) (%) Cat. 1 Cat. 2 Cat. 3 PM % # < 60% 60% < # < 95% 95% < # 0.00% # < 35% 35% < # < 75% 75% < # PM % # < 80% 80% < # < 95% 95% < # 0.00% # < 40% 40% < # < 80% 80% < # Table E-03 Cat 1: Cat 2: Cat 3: Paved road fugitive emissions not required to be modeled, and no paved road fugitive dust permit conditions. Requirements in Minn. Rule apply. Paved road fugitive emissions not required to be modeled, with paved road fugitive dust permit conditions determined by levels of traffic at the facility. Paved road fugitive emissions are required to be modeled, with site-specific paved road fugitive dust permit conditions. Re-modeling and/or addition of paved road fugitive emissions source group required. Section F. Receptors (RE Pathway) Please refer to guidance from Tables 11, 12, and 13 of the April 2013 MPCA Air Dispersion Modeling Guidance for Minnesota Title V Modeling Requirements and Federal Prevention of Significant Deterioration (PSD) Requirements (aq2-58), as well as federal guidance TTY or Available in alternative formats aq2-40 7/10/13 Page 8 of 16

13 1. What type of receptor grid will be used? a. If other, or a combination, please describe: Discrete Cartesian b. Will grid be converted to discrete Cartesian? Yes No N/A already discrete Cartesian 2. What will be the grid dimensions? (Ex., radius of 10 km, 5 km by 5 km, etc.) 105 km X 110 km SIL grid, NAAQS cumulative impact assessment grids will extend to the SIL exceedance concentration and will be determined separately for each pollutant. 3. What is the proposed spacing of receptors for? a. Inside the property boundary(s): N/A meters b. On the fenceline(s): N/A meters c. On the property line(s): 25, 100, 250, 500 meters d. Beyond the property line(s): 50, 250, 500, 1000, 5000 meters 4. Will FLAGPOLE receptors be included in the receptor grid? (N/A = No high-rise structures w/ ambient air within 3 miles of source; No = High-rise structures w/ ambient air exist within 3 miles of source, but will not include in receptor grid.) Yes N/A No Please continue to the next question (6) 5. Additional information for this section that was not included above, including discussion of options (if not applicable, place N/A in field): There will be several receptor grids utilized as part of the permit application submittal. The SIL analysis results are discussed in detail in Attachment J. The use of the SIL results for developing the permit modeling receptor grids are discussed more generally in this section. Large Figure 3 shows a close-up of the receptor grid around the Plant Site ambient air boundary (AAB). The Plant Site AAB grid follows PolyMet s property boundary. PSD increment and NAAQS concentrations from NorthMet Plant Site sources are evaluated at this boundary. The Plant Site AAB grid will have 25 m spacing south of the main plant and tailings basin in areas of maximum modeled impacts. The remainder of the Plant Site AAB grid will have 100 m spacing around the boundary. From the boundary out 1 km the receptor spacing will be 250 m, except in the area of maximum impacts, where the receptor spacing will be 50 m. From 1 km out to 5 km from the boundary the receptors spacing will be 500 m. From 5 km to the maximum extent of the grid, receptor spacing will be 1 km. The Plant Site AAB receptor grid will be used to demonstrate that the project itself will not contribute to any increment or NAAQS excdeedances on the nearby facilities properties, which wouldn't have their own impacts included at receptors within their own ambient air boundary. Large Figure 4 shows a close-up of the cumulative impacts assessment boundary receptor grid. The former LTVSMC property boundary will be used for the NAAQS cumulative impacts analysis. The LTVSMC ambient boundary encompasses the extent of all of the former mining areas within the LTV footprint and includes the adjacent sources of Mesabi Nugget and the Cliffs Erie Pellet Yard. The cumulative impacts assessment boundary is appropriate for determining cumulative impacts from Plant Site and nearby sources as all of the area within the former LTV boundary is zoned mining industrial and general public access is not allowed. The three facilities share a common gated entrance at the Plant Site administration building. The cumulative impacts assessment boundary will have receptors at 100 m spacing along the boundary and at 1 km intervals out to the extent of the grid. Receptor density beyond the boundary will be increased in areas of elevated concentrations as necessary. As described in Attachment J, the extent of the cumulative impacts assessment grid will vary by pollutant, but all pollutants will be evaluted along the boundary. For this protocol submittal, two Significant Impact Level (SIL) grids have been used. For SO2, NO2, and PM10 the SIL analysis used the Plant Site ambient air boundary receptor grid which was used in previous modeling analyses. This grid has very dense receptor spacing immediately south of the Plant Site in areas of maximum modeled impacts. For the PM2.5 SIL analyses, a very large grid (105 km x 110 km) with less dense (500 m) receptor spacing near the Plant Site AAB was used out to 5 km from the Plant Site AAB to determine the extent of the refined modeling receptor grid. Section G. Meteorological Data (ME Pathway) Note: If modeling with more than one meteorological data set (i.e., portable facility), please list in question 8 the information requested in questions 1-4 for the additional data sets. 1. What meteorological surface station is proposed for use? Please indicate the station name, the state the surface station is located in, and the three letter call sign\identifier. (Ex.: Minneapolis/St. Paul Int l Arpt; MN; MSP) Or check the box to indicate that on-site surface meteorological data will be used instead of NWS surface meteorological data. Surface station name: Hibbing Airport State: MN Three-letter call sign/identifier: HIB TTY or Available in alternative formats aq2-40 7/10/13 Page 9 of 16

14 Onsite surface data 2. What meteorological upper air station is proposed for use? Please indicate the station name, the state the surface station is located in, and the three letter call sign\identifier. (Ex.: Chanhassen; MN; MPX) Upper air station name: International Falls State: MN Three-letter call sign/identifier: INL 3. What consecutive 5- year period will be used? 2009 to Were the proposed AERMET files pre-processed by MPCA staff? Yes No a. If yes, please provide the name of the met data zip file that was received or obtained from the MPCA and the date it was obtained: Name:.zip Date (mm/dd/yyyy): b. Please indicate what version of AERMET will be used: Other If other, please explain: AERMET; AERMINUTE 5. Will on-site meteorological data be used, instead of NWS meteorological data processed by MPCA staff? Yes No Note: If site-specific meteorological data will be collected and used, please follow the federal guidance (EPA s), as specified in section 8.3 and section (QA/QC) of 40 CFR Part 51 dated 11/09/2005 (Appendix W). a. If site-specific surface meteorological data will be collected and used, where will the location of the meteorological tower be set (city and state, coordinates, etc.)? N/A b. If site-specific meteorological data will be collected and used, what year of data is proposed to be used? N/A 6. What justification(s) applies for the proposed surface and upper air stations identified above? (Check all that apply) Similar surface characteristics as meteorological tower Similar wind patterns/characteristics Other Please describe: a. Please provide detail for your justifications (required): Similar land use characteristics Proximity to surface and/or upper air station(s) Hibbing has AERMINUTE data. Other nearby airports do not have AERMINUTE data. The Hibbing meteorological data set is the primary meteorological data set for dispersion modeling use on the Mesabi Iron Range. 7. Will wind speed categories and/or wind speed emission factors be used? No Skip to question 8. Yes for which pollutants: PM10, PM2.5 a. Please specify the wind speed categories (m/s) to be used for the ME WINDCATS keyword: User-specified ME WINDCATS: 4.70 m/s 7.80 m/s m/s m/s m/s Default ME WINDCATS: 1.54 m/s 3.09 m/s 5.14 m/s 8.23 m/s m/s b. Please list the user-specified wind speed emission factors for the SO EMISFACT WSPEED pathway. Include the source ID or range, pollutant, and six wind speed emission factors. This does not need to be an exhaustive list. PM10 and PM2.5: LCRST/PBLKT = 0 for windspeeds less than 4.7 m/s PM10: LCRST/PBLKT = 0/1.21E-05 PM2.5: LCRST/PBLKT= 0/1.81E-06 PM10: LCRST/PBLKT = 1.18E-05/4.60E-05 PM2.5: LCRST/PBLKT= 1.76E-06/6.90E-06 PM10: LCRST/PBLKT = 3.85E-05/8.33E-05 PM2.5: LCRST/PBLKT= 5.78E-06/1.25E-05 PM10: LCRST/PBLKT = 6.71E-05/1.19E-04 PM2.5: LCRST/PBLKT= 1.01E-05/1.79E-05 PM10: LCRST/PBLKT = 8.71E-05 /1.43E-04 PM2.5: LCRST/PBLKT= 1.31E-05 /2.15E Additional information for this section that was not included above (if not applicable, place N/A in field): LCRST sources A - K refer to LTV coarse tailings see: Wind Erosion_09-13_alternative_LTV_coarse_tailings PBLTK sources A - I refer to PolyMet bulk tailings see: Wind Erosion_09-13_alternative_PolyMet_bulk_tailings See Figure 4 for Tailings Basin wind erosion source configuration TTY or Available in alternative formats aq2-40 7/10/13 Page 10 of 16

15 Section H. SIL Analysis and Results 1. Will a SIL or impact area analysis be conducted in conjunction with this project, in order to determine if a cumulative analysis is required and to determine the extent of the modeling domain? No Will not model against SILs and instead proceed directly to conducting a cumulative analysis, with a modeling domain of 50km x 50km. Yes List for which pollutants: PM10, PM2.5, SO2, NO2 N/A Proceed to the next section (I) 2. If a preliminary SIL analysis has been conducted for this project, it is optional but highly recommended that results be provided (in the Table H-01 below), as well as including corresponding model output files. Note: Refer to the MPCA s Air Modeling Guidance, How To Model, Part II: Setting Up the Model, Step one Impact Area Evaluation for guidance. Table H-01, Class II Significant Impact Levels Modeling Results for: Pollutant SO2 PM10 PM2.5 NOx CO Averaging Time Modeled Impacts (H1H) (μg/m 3 ) SILs (μg/m 3 ) *As of 10/26/2010 % of SIL Exceed SIL? 1-hr % 3-hr % 24-hr % Annual % 24-hr % Annual % 24-hr % Annual % 1-hr % Annual % 1-hr % 8-hr % Yes - Refined Modeling Yes - Refined Modeling Yes - Refined Modeling Yes - Refined Modeling (blank) Radius of Impact (If exceeds SIL) 10 km 7.8, 5 km 24, 9.5 km 12.2 km km 3. Additional information for this section that was not included above (if not applicable, place N/A in field): SIL modeling was not conducted for SO2 3-hr, 24-hr, and annual averaging periods or for the NO2 annual averaging period as the 1-hour SILs for these pollutants have the largest radius of impact. The SIL results in Table H-01 reflect the maximum value on the SIL grid for each pollutant. However, the purpose of the SIL analysis (when results are above the SIL) is to determine the extent of the grid. The results listed as a percentage of SIL are therefore not particularly illuminating with respect to developing the full impact analysis grid. It should also be noted that exceedance of the SIL, by any amount, is not an indication of an air quality impact, but merely that additional analysis is warranted. Section I. Background Values Please refer to Section 2.4 and Table B-4 of the latest version of the AERMOD User s Guide, Addendum for guidance as well as the latest version of the MPCA s Air Dispersion Modeling Guidance for Title V and PSD. 1. Are background concentrations required for your analysis? *Tip: If NAAQS/MAAQS was selected in question A.8, the answer is Yes. Yes List pollutants required for: SO2, PM10, PM2.5, NO2 No Please explain (e.g., SIL analysis-only, etc.) and then proceed to question 2: Note: All background concentrations, including uniform and temporally varying concentrations, will need to be specified in the input file(s). a. If yes, please provide a description below of the background concentration(s) and any considerations given to the methodology: TTY or Available in alternative formats aq2-40 7/10/13 Page 11 of 16

16 Minnesota monitoring values from the most current three-year period were used. SO2 and NO2 are conservative as they located close to the Flint Hills Resources Pine Bend refinery - no adjustment was made to the background concentrations to account for the potential influence of the refinery on monitored values. PM10 and PM2.5 background concentrations were developed from Virginia, MN. The Virginia, MN monitor is a site representative of the Iron Range. The background concentrations will be developed from : PM hour 3-year average of 98 th %ile = ( )/3 = 18 µg/m 3 PM2.5 Annual 3-year average of annual conc = ( )/3 = 7.5 µg/m3 PM10 24-hour 3-year average of H2H conc = ( )/3 = 33 µg/m 3 PM10 Annual 3-year average of annual conc = To be determined Note PM10 annual is a MAAQS only standard. b. Background data sources (check all that apply): MPCA website EPA website Other explain: Annual AIr Monitoring Network Plan for Minnesota (data from in these reports) c. Examined monitors out to a distance of: km d. Monitors excluded from consideration and reasons why: Ely, MN has a single year of monitoring data for SO2 (2012) and NO2 (2012). Ely, MN also has PM2.5 monitoring data starting in 2013 and continuing to the present. Although the Ely site is located close to the Project site and has similar surrounding land use as the Project site as compared to the Twin Cities metropolitan area (where the proposed background values were obtained), the use of the Ely SO2 and NO2 monitoring data as background has not been approved for modeling because there is only a single year of data and due to the lack of industry or nearby population centers. The PM2.5 data were not used as background because the Virginia monitor better captures the contributions from nearby sources as described in more detail in Attachment J. The Ely PM2.5 data may be reconsidered for use as background depending on the modeling results. e. Facility characteristics: i. Facility land use (industrial, residential, agricultural, cropland, mixed, etc.): industrial with mixed forests ii. Facility setting: Urban Suburban Rural f. Ambient air monitor(s) characteristics: i. Monitor land use (industrial, residential, agricultural, cropland, mixed, etc.): ii. Monitor setting: Urban Suburban Rural g. Form of background value(s) for each applicable pollutant (maximum, 98 th percentile, etc.): SO2 (1-hour) - 3-year average of 99th percentile; FHR 443 NO2 (1-hour) - 3-year average of 98th percentile;fhr 423 PM10 (24-hour) - 3-year average of high-2nd-high; 327 First St. S, Virginia PM2.5 (24-hour) - 3-year average of 98th percentile; 327 First St. S, Virginia PM2.5 (annual) - 3-year average of weighted arithmetic mean; 327 First St. S, Virginia... PM - Small city; SO2 and NO2 Suburban / rural / industrial h. Please check user-generated background concentrations option(s) to be used for each applicable pollutant (check all that apply): SO BACKGRND BGflag i. File name(s): ii. iii. File creation date(s): BGflag(s) List all that apply and for which pollutant: SO BACKGRND HOURLY i. File name(s): (BGflag can be defined as ANNUAL, SEASON, MONTH, etc. See Table B-4, Appendix B, of the AERMOD User s Guide Addendum) 2. Additional information for this section that was not included above (if not applicable, place N/A in field): Seasonal background values for PM2.5 may be developed as a possible refinement to the analysis. The MPCA provided ambient monitoring data from Virginia, MN which may be used to develop seasonal background values. The PM10 24-hour increment is the most restrictive standard for the PM10 modeling, so further refinement to the PM10 monitored background concentrations will not be employed. Attachment J provides a detailed discussion of the sources influencing the background monitors and the combined use of background concentrations and explicitly modeled nearby sources TTY or Available in alternative formats aq2-40 7/10/13 Page 12 of 16

17 Section J. Nearby Sources *Contact MPCA air dispersion modeling staff for MPCA-generated products. Also, refer to the How to model, Part II section, Step four Compliance demonstration subsection in the MPCA s Air Dispersion Modeling Guidance. 1. Will any nearby sources be modeled explicitly for the analysis? N/A No* Yes List pollutants required for: PM10, PM2.5, SO2, NO2 *If no, please skip ahead to Section K: Pollutant-based Considerations. 2. Please select the methodology for selecting nearby sources (check all that apply): MNLookup Table ROI SQRM-D Tool* Other Please explain: *This is the MPCA-generated GIS Tool a. Please provide additional information on the selection process used to obtain the initial and final nearby source emission inventory, including any deviations from or additional work to MPCA approved approaches: *Nearby sources out to 50 km away from the subject source should be examined. *Please include facility ID s and/or names in the modeling input files of all nearby sources. Attachment J provides a detailed discussion of background concentrations and nearby sources to be explicitly modeled. b. Are there any nearby sources that are being omitted that were identified by MPCA approved methods? No Yes c. If yes to question b above, please provide the names of the sources omitted and the justification: Attachment J provides a detailed discussion of background concentrations and nearby sources to be explicitly modeled. d. What emission inventory year(s) is being used? N/A - using permitted emission rates 3. Describe the emission s type(s) that will be used in the nearby source modeling (check all that apply): Allowables (PTE s) Actuals User-collected Other a. Where does the emissions data come from (for each source)? *Example: Permit ID and action #, MPCA emissions inventory, stack tests, etc. Mesabi Nugget, modeling input files. {description of additional sources to be added after confirmation with MPCA} b. Please describe how the emissions are calculated (for each source): *Example: Process-based emissions, permitted value, merged-stack calculation, scaled emissions, etc. See Attachment J and NorthMet Plant Site emission inventory for specific emission calculation methods. 4. How will the nearby source(s) be characterized? (Check all that apply) Explicit Merged Parameters Other a. If other was selected for question 4 above, please provide further details: 5. Additional information for this section that was not included above (if not applicable, place N/A in field): Attachment J provides a detailed discussion of nearby sources to be modeled explicitly due to their likelihodd for combined impacts with NorthMet sources. Section K. Pollutant-based Considerations *Please refer to the April 2013 MPCA Air Dispersion Modeling Guidance for Minnesota Title V Modeling Requirements and Federal Prevention of Significant Deterioration (PSD) Requirements. (aq2-58) Secondary Formation of PM2.5 (PM2.5, NOx, and SO2): 1. Will this source have direct PM2.5 emissions? TTY or Available in alternative formats aq2-40 7/10/13 Page 13 of 16

18 Less than (<) 10 TPY SER (including no direct PM2.5 emissions) 2. Will this source have NOx emissions? Less than (<) 40 TPY SER (including no NOx emissions) 3. Will this source have SO2 emissions? Less than (<) 40 TPY SER (including no SO2 emissions) Greater than (>) 10 TPY SER Greater than (>) 40 TPY SER Greater than (>) 40 TPY SER 4. Please select the assessment case that applies to your facility for the secondary formation of PM2.5? (Refer to answers given above for questions 1 3. Also refer to the Secondary Formation of PM2.5 section of the April 2013 MPCA Air Dispersion Modeling Guidance.) Case 1 (No air quality analysis): Direct PM2.5 emissions < 10 TPY, NOx and/or SO2 emissions < 40 TPY Case 2 (Primary air quality impacts only): Direct PM2.5 emissions 10 TPY, NOx and/or SO2 emissions < 40 TPY Case 3 (Primary and secondary air quality impacts): Direct PM2.5 emissions 10 TPY, NOx and/or SO2 emissions 40 TPY Case 4 (Secondary air quality impacts only): Direct PM2.5 emissions < 10 TPY, NOx and/or SO2 emissions 40 TPY 5. Please describe the qualitative and/or quantitative approach to be used to address cases 3 or 4, selected in question 4 above (if applicable to your facility if not, then place N/A in field): The offset ratio method found on p. 40 of the MPCA's July 2014 Dispersion Modeling Guidance will be followed. The offset ratio calculations are included in the Emission Inventory spreadsheet previously submitted (3/26/2016).. NO2: 1. Is NO2 proposed to be modeled? Yes No* *If no, skip the rest of the NO2-related questions. 2. If Tier 3 is anticipated to demonstrate compliance in the modeling analysis (OLM or PVMRM), please provide the following details now to expedite MPCA s review: a. In-stack ratio of NO2/NOx (CO NO2STACK) (default=0.50): 0.50 b. Equilibrium ratio (CO NO2EQUIL) (default=0.90): If Tier 3 is anticipated, please provide details below regarding ozone values (Please check all option(s) proposed to be used): a. CO OZONEFIL (hourly): i. O3FileName: PMP_O3_5Y.dat ii. O3Units: UG/M3 PPM PPB iii. O3Format: Hourly b. CO O3VALUES O3flag: OZONUNIT: PPB PPM UG/M3 Note: O3flag can be defined as ANNUAL, SEASON, MONTH, etc. PPB is the AERMOD default for OZONUNIT. See Table B-2, Appendix B, of the AERMOD User s Guide, Addendum for Version i. Please input the O3values (i=1,n) for the O3flag specified above (C.5.d.ii): Example: For CO O3VALUES SEASON, n=4 values (winter #, spring #, summer #, fall #) c. CO OZONEVAL (monitored value): O3Units: UG/M3 PPB PPM Note: UG/M3 is the AERMOD default for OZONEVAL O3units. 4. Will separate NO2/NOx ratios (SO NO2RATIO SrcID) be used for specific sources while using the OLM or PVMRM options? Yes No 5. Will SO OLMGROUP ALL be used for the OLM option? N/A Yes No 6. Is EPA approval needed for the modeling protocol (e.g., Tier 3 NO2)? Tier 3 NO2 methodologies that require approval by Region 5 modeling staff need to have said approval before submission of the AQDM-06 report form and attach approval. Yes PSD/SIP permit action No State-only action 7. Additional information for this subsection that was not included above (if not applicable, place N/A in field): The hourly ozone file is included with the supplemental files TTY or Available in alternative formats aq2-40 7/10/13 Page 14 of 16

19 Appendix Figure App.1 Table App.1 Modeled Form of the NAAQS/MAAQS by Averaging Periods Pollutant 1-hour 3- hour 8- hour 24-hour Monthly Annual CO H2H - H2H NO 2 98 th percentile of the daily max. 1-hour values H1H Pb H1H - PM H6H of the multiyear values - H1H PM th percentile of daily max. 24-hour values - H1H SO 2 99 th percentile of the daily max. 1-hour values H2H - H2H - H1H Table App.2 Modeled Form of PSD Increment by Averaging Periods Pollutant 1-hour 3-hour 24-hour Annual NO H1H PM H2H H1H PM H2H* H1H SO 2 - H2H H2H H1H Criteria pollutants modeled for SIL analyses should be modeled as H1H s. Helpful Webpages, Documents/Guidance, and Modeling Tips Please consult the following webpages and documents for the most current modeling guidance and recommendations when filling out this form: U.S. EPA s Support Center for Regulatory Atmospheric Modeling: Please check the SCRAM webpage regularly for the most recent updates to guidance, models, and standards; especially, for modeling guidance related to: 24-hour PM2.5, 1-hour NO2, and 1-hour SO2 NAAQS. U.S. EPA s 40 CFR Part 51 Appendix W: U.S. EPA s AERMOD Implementation Guide: U.S. EPA s AirData: MPCA s Air Dispersion Modeling: MPCA s Ambient Air Monitoring Network Plan: MPCA s Environmental Review: USGS National Map Seamless Viewer (NED data): TTY or Available in alternative formats aq2-40 7/10/13 Page 15 of 16

20 For questions on this form, or data requests from MPCA air dispersion modeling staff, please send an to: Please be sure to include with your questions or requests: the form ID (ADQMP-01), facility name and permit #, and contact information. Questions can also be asked by calling one of the MPCA s air dispersion modeler s (phone numbers are listed on the MPCA s Air Dispersion Modeling webpage, link above). Tip: Please be sure to use UTM Coordinates, NAD83, Zone 15 Extended for all locational data. Zone 14 or 16 will not be acceptable. This is to keep consistency between coordinates of all Minnesota sources. Tip: For NAAQS modeling, please include the following source groups: SO SRCGROUP ALL BACKGROUND SO SRCGROUP BKG BACKGROUND SO SRCGROUP FAC [FAC sources) Tip: Plot files with combined averaging times for the same pollutant, and extensions such as *.grf, should be separated into individual *.plt files. For example, PM10 should have at least two individual plot files: one for the 24-hour averaging period for source group ALL and a second for the annual averaging period for source group ALL. Group ALL should include BACKGRND if conducting NAAQS modeling (e.g., SO SRCGROUP ALL BACKGRND ). Acronyms μg/m 3 AERMAP AERMET AERMINUTE AERMOD AERSURFACE AQ AQDMP-01 AQDMPS-01 BPIP-PRIME CO DEM EAW EIS EPA FAC FAR H1H H2H H6H H 2 S km LULC MAAQS MPCA NAAQS Micrograms per cubic meter AERMOD Terrain Preprocessor AERMOD Meteorological Preprocessor AERMOD 1-Minute ASOS Wind Data Processor AMS/EPA Regulatory Model AERMOD Surface Characteristic Tool Air Quality Air Quality Dispersion Modeling Protocol form Air Quality Dispersion Modeling Protocol Spreadsheet Building Profile Input Program for PRIME Carbon Monoxide Digital Elevation Model Environmental Assessment Worksheet Environmental Impact Statement U.S. Environmental Protection Agency 3-letter facility ID First-Approximation Representative High-first-high value High-second-high value High-sixth-high value Hydrogen Sulfide Kilometer Land Use Land Cover Minnesota State Ambient Air Quality Standard Minnesota Pollution Control Agency National Ambient Air Quality Standard NAD North American Datum of 1983 NED NO 2 NWS OLM Pb PM 10 PM 2.5 POINTCAP POINTHOR PPB PPM PRIME PSD PTE s PVMRM SCRAM SER SIL SO 2 SIP SMS National Elevation Dataset Nitrogen Dioxide National Weather Service Ozone Limiting Method Lead Particulate Matter less than 10 um in size Particulate Matter less than 2.5 um in size Capped-release point source (AERMOD beta, nondefault) Horizontal-release point source (AERMOD beta, non-default) Parts Per Billion Parts Per Million Plume Rise Model Enhancements Prevention of Significant Deterioration Program Potential to Emit Plume Volume Molar Ratio Method Support Center for Regulatory Atmospheric Modeling Significant Emission Rate Significant Impact Level Sulfur Dioxide State Implementation Plan Standardized Mobile Source UG/M3 Micrograms per cubic meter (μg/m 3 ) UTM Universal Transverse Mercator TTY or Available in alternative formats aq2-40 7/10/13 Page 16 of 16

21 Booster Pump House (#1) SV426 Barr Footer: ArcGIS 10.4, :37 File: I:\Client\PolyMet_Mining\Work_Orders\Permitting\Air_Permit_Application\Maps\Report\AQDMP_01_PlantSite_Protocol\Large Figure 1 Process Plant Model Source Layout.mxd User: arm2 Point Sources Sources Not Modeled Ambient Air Boundary Existing Buildings New Buildings Additional Buildings Note: Area 1 Building Vent and Administrative Building Stack are not shown. SV005 SV009 SV003 SV402 SV401 SV004 SV008 SV404 SV403 SV509 SV508 Concentrator SV507 (#6) SV506 Drive House 1 (#3) SV505 SV406 Flotation SV405 Water Treatment Plant SV504 Building SV503 SV304 SV221 SV502 SV305 SV408 SV016 SV407 SV014 SV421 Fine Crusher (#4) SV422 SV409 SV327B SV327C SV327A SV327D O2 Plant SV328 SV423 SV424 SV108 SV109 Limestone Preparation SV301 SV323 SV322 SV2532 SV333 SV331 Hydrometallurgical SV334 Plant SV332 Concentrate SV8003 Dewatering Concentrate Storage Reagents Coarse Crusher (#2) Warehouse 49 (#14) Warehouse Electrical (#15) General Shops (#12) Carpenter's Shop I ,000 Feet Waste Water Treatment Plant (WWTP) Rebuild Shop (#13) Description Proposed Model ID Permit SVID SV001 SV003 North Primary Crushing Stack SV002 SV004 South Primary Crushing Stack SV003 SV401 North 60 Distribution Box and 36" Crushers Stack SV005 SV005 North 36" Pan Feeders to Conveyor 1A Stack SV006 SV402 North 36" Pan Feeders to Conveyor 1B Stack SV004 SV403 South 60 Distribution Box and 36" Crushers Stack SV007 SV404 South 36" Pan Feeders to Conveyor 1A Stack SV008 SV008 South 36" Pan Feeders to Conveyor 1B and South Pan Feeder Stack SV009 SV009 North Pan Feeder Stack SV010 SV405 Drive House 1 East Transfer Stack SV011 SV406 Drive House 1 West Transfer Stack SV012 SV407 East Fine Crushing (1) Stack SV012 SV014 West Fine Crushing (1-2) and Transfer Points Stack SV016 SV408 North Transfer Point Stack SV017 SV016 South Transfer Point Stack SV019 SV502 Fine Ore Feeders North 1-2, SAG Transfers Stack SV020 SV503 Fine Ore Feeders North 3-5 Stack SV021 SV504 Fine Ore Feeders North 6-8 Stack SV022 SV505 Fine Ore Feeders North 9-11 Stack SV023 SV506 Fine Ore Feeders North Stack SV024 SV507 Fine Ore Feeders North Stack SV025 SV508 Fine Ore Feeders North Stack SV026 SV509 Fine Ore Feeders North Stack SV027 SV421 Fine Ore Feeders South 1-3 Stack SV028 SV422 Fine Ore Feeders South 4-6 Stack SV029 SV423 Fine Ore Feeders South 7-9 Stack SV030 SV424 Fine Ore Feeders South Stack SV013 SV409 West Coarse Ore Storage Stack SV063 SV108 Back up Generator #1 Stack SV064 SV109 Back up Generator #2 Stack SV035 SV2532 Autoclave Scrubber Stack SV037 SV8003 Plant Scrubber Stack SV043 SV221 Lime Slaker Stack SV049 SV322 Limestone Crusher Stack SV040 SV301 Autoclave Startup Boiler Stack SV044 SV323 Limestone Unloading Baghouse Stack SV065 SV304 Fire Pump #1 Stack SV066 SV305 Fire Pump #2 Stack SV031 SV331 Concentrate Storage Area #1 Vent SV032 SV332 Concentrate Storage Area #2 Vent SV033 SV333 Railcar Loading - Copper Concentrate Vent SV034 SV334 Railcar Loading - Nickel Concentrate Vent SV054 SV327A Oxygen Plant Cooling Tower Vent A SV055 SV327B Oxygen Plant Cooling Tower Vent B SV056 SV327C Oxygen Plant Cooling Tower Vent C SV057 SV327D Oxygen Plant Cooling Tower Vent D SV039 SV328 Oxygen Plant Adsorber Regeneration Heater Vent SV060 SV427 Boiler (New) Administration Building Stack SV067 SV426 Tailings Basin WWTP Backup Generator Stack SV080 A1BV Area 1 Bulding Vent DRAFT PROCESS PLANT MODEL SOURCE LAYOUT NorthMet Project Poly Met Mining Inc. Large Figure 1 NorthMet Plant Site Dispersion Modeling Protocol

22 Barr Footer: ArcGIS 10.4, :37 File: I:\Client\PolyMet_Mining\Work_Orders\Permitting\Air_Permit_Application\Maps\Report\AQDMP_01_PlantSite_Protocol\Large Figure 2 Tailings Basin Model Source Layout.mxd User: arm2 2EN JWN "6 JWU TBI S4 "6 AREA5 "6 SDW A5B JSD SD TAILINGS BASIN AREA "6 BCA HSD CA "6 SPD "6 HM4 CAS TWW "6 TBH JSH 1ES "6 WUE Ambient Air Boundary "6 Tailings Handling Source Locations LTV Coarse Tailings Wind Erosion Sources PMET Bulk Tailings Wind Erosion Sources Haul Road Sources (Source Count) A5B (12) JSD (39) BCA (67) JSH (38) CAS (74) JWN (43) TBH (55) TBI (124) WUE (116) I 0 1,000 2,000 4,000 DRAFT TAILINGS BASIN MODEL SOURCE LAYOUT NorthMet Project Poly Met Mining Inc. Coal Ash Landfill Unpaved Road Sources TWW (33) HSD (52) JWU (45) SDW (47) Feet Large Figure 2 NorthMet Plant Site Dispersion Modeling Protocol

23 Barr Footer: ArcGIS 10.4, :51 File: I:\Client\PolyMet_Mining\Work_Orders\Permitting\Air_Permit_Application\Maps\Report\AQDMP_01_PlantSite_Protocol\Large Figure 3 - Plant Site Ambient Air Boundary Receptor Grid - Close-Up.mxd User: arm2 Plant Site Ambient Air Boundary Receptor Grid Cliffs Erie Pellet Yard NorthMet Ambient Air Boundary (AAB) Approximate LTVSMC Ambient Air Boundary (AAB) Approximate Mesabi Nugget Ambient Air Boundary (AAB) Approximate Northshore Ambient Air Boundary (AAB) I Kilometers DRAFT PLANT SITE AMBIENT AIR BOUNDARY RECEPTOR GRID - CLOSE-UP NorthMet Project Poly Met Mining Inc. Large Figure 3 NorthMet Plant Site Dispersion Modeling Protocol

24 Barr Footer: ArcGIS 10.4, :45 File: I:\Client\PolyMet_Mining\Work_Orders\Permitting\Air_Permit_Application\Maps\Report\AQDMP_01_PlantSite_Protocol\Large Figure 4 - Cumulative Impacts Assessment Boundary Receptor Grid - Close-Up.mxd User: arm2 Minnesota Power - Laskin Energy Center Plant Site Cumulative Impacts Assessment Boundary Receptor Grid Approximate LTVSMC Ambient Air Boundary (AAB) NorthMet Ambient Air Boundary (AAB) Cliffs Erie Pellet Yard Approximate Mesabi Nugget Ambient Air Boundary (AAB) Approximate Northshore Ambient Air Boundary (AAB) I Kilometers DRAFT CUMULATIVE IMPACTS ASSESSMENT BOUNDARY RECEPTOR GRID - CLOSE-UP NorthMet Project Poly Met Mining Inc. Large Figure 4 NorthMet Plant Site Dispersion Modeling Protocol

25 Attachment C Non-Default Modeling Options This attachment provides additional information on the non-default modeling options described in Section C of the Plant Site MPCA AQMD-01 form. The non-default option is a deposition algorithm used for PM 10. This options is described in Section I below. Section II describes the proposed 24-hour PM 2.5 NAAQS modeling methodology to account for secondary formation of PM 2.5 due to the plant site combustion sources. An additional non-default option, OLM or Ozone Limiting Method, is also proposed for the Plant Site modeling (as was the case for previous modeling). However, the MPCA AQDM-01 form provides sufficient information to evaluate the use of OLM for the proposed modeling. I. Deposition Algorithm a. PM10 Particulate Deposition An alternative to the default AERMOD deposition algorithm (Reference (1)) was proposed and approved for the PM 10 modeling completed for the Plant Site for the Draft Environmental Impact Statement and the Supplemental Draft Environmental Impact Statement. The same approach is proposed for the modeling to be submitted with the air emission permit application. In this alternative approach, particulate deposition is represented by the Decay Term (D). As stated in the model user s guide (Reference (2)): The Decay Term in Equation (1-1) is a simple method of accounting for pollutant removal by physical or chemical processes. The decay term is calculated as a half-life as follows: D = exp [ - ψ x / u s ] Where: ψ = decay coefficient (s -1 ) x = downwind distance (m) u s= wind speed (corrected for release height) and ψ = / T 1/2 T 1/2 = pollutant half-life (s) Example : T 1/2 = 900 s x = 1000 m u s = 4 m/s P:\Mpls\23 MN\69\ \WorkFiles\APA\Permitting\Air Permitting\Class II Modeling Plant Site\Protocol\Without U star\attachment_c_-_ps_non-default_model_options_wo adjusted u star d3.docx 1

26 D = exp [ - (0.693/900) 1000 / 4 ] = In this case, 17.5% of the plume would be removed for a receptor at 1 km distance from the source. The model applies the Decay Term after conducting the other dispersion calculations. The pollutant half-life (T 1/2) is used to represent the various physical mechanisms which remove particulate mass from the plume. Deposition occurs from gravitational settling, removal by vegetation, particle agglomeration, and other mechanisms. The pollutant half-life was estimated using the gravitational settling velocity (Stoke s Law) term as described in the model user s guide (Reference (2)), as follows. v g = (ρ ρ air) 2 g d p c 2 / [ 18 µ ] * SCF (Equation 1-84) Where: v g = gravitational settling velocity (cm/s) ρ = particle density (g/cm 3 ) = 2.7 for haul roads (crushed ore) ρ air = air density (= 1.2 x 10-3 g/cm 3 ) d p = particle diameter (µm) = 7.4 for mass-mean particle size c 2 = units conversion constant (1 x 10-8 cm 2 /µm 2 ) µ = viscosity of air (1.81 x 10-4 g/cm/s) S CF = slip correction factor (see Equation 1-85; for a 7.4 µm particle S CF = 1.02) g = acceleration due to gravity (981 cm/s 2 ) For the particle characteristics given above (7.4 µm diameter, 2.7 g/cm3 density), the gravitational settling term v g = cm/s. The derivation of the 7.4 µm diameter is described in Exhibit 1. To calculate the pollutant half-life, it was assumed that the time required for a 7.4 µm particle to settle from an average release height of 5 m would approximate the time-dependent component of deposition. For a 5 m release height, the settling time is T = 5 m x 100 cm/m / cm/s = 1,100 s. The calculation of pollutant half-life and use of the decay coefficient to represent deposition was used in Midwest Research Institute s report prepared for the MPCA entitled: Iron Range Air Quality Analysis (Reference (3)). The MRI report also provided detailed emission inventories and modeling analysis (including model calibration) which were required for EPA approval of Minnesota s State Implementation Plan (SIP). During the late 1970s, the Iron Range was in nonattainment for TSP, and the 1979 modeling analysis was part of determining the actual attainment status and identifying mitigation measures and future attainment status. An important component of the modeling analysis was to adequately address particulate deposition so that the model results could be directly compared to monitoring data. To account for particle deposition, the 1979 modeling assumed a particle density of 3 g/cm 3, a particle diameter of µm, and a release height of 5 m, for a half-life of 370 s. The proposed pollutant half-life of 1,100 s is 3 times greater than was assumed in the MRI study, with the greatest difference in the half-life resulting from the smaller particle diameter used in this analysis. The 1979 analysis also assumed that particles less than 5 µm did not settle out. For this analysis, we propose

27 to model particles less than 2.5 µm as though they do not deposit. For the mining fugitive dust sources, PM 2.5 accounts for 10% of the total PM 10. Therefore, the modeled emission rate will be scaled by a factor of 1.1 to account for the PM 2.5 portion of the PM 10 that would not be predicted to settle out. b. PM10 Half-Life Modeling Analysis The AERMOD model is setup that when the half-life option is selected it is applied to every source in the input file and does not allow the user to specify which sources it can be applied to (similar to the urban source option). The purpose of modeling PM 10 with half-life is to account for the deposition of particulate matter from mechanically generated fugitive dust sources. There are numerous sources included in the Plant Site modeling analysis that are not considered mechanically generated. In order to model all of the PM 10 Plant Site sources together, the LAKES AERMOD-View Multi-Chem utility will be used to combine the post-files of the mechanically generated half-life sources and the non-half-life sources together by each hour and receptor to determine the combined 24 hour and annual PM 10 NAAQS and Increment concentration results. The LAKES Multi-Chem utility allows the user to model multiple emission rates for the same input file by splitting out a single AERMOD input file into individual source, unitized model runs. Each source is modeled individually at 1 g/s with an output plot file and post file for the maximum 1hour concentration. Multi-Chem then post-processes the individual source files, combining them together by hour and receptor and multiplying each source s individual concentration results by an emission rate specified in a text file input to the software. Using this utility creates a significant amount of data files due to a post file created for every individual source. Since the format of the post-files is the same whether for a single source or a combination of sources, PolyMet is proposing to use the Multi-Chem post-processing capabilities, but not for individual source, unitized post file runs. The PM 10 modeling for the Plant Site will be divided into two separate modeling files. The first file will model the non-half-life sources listed in the paragraph above in regulatory default mode using the hourly PM 10 emission rates calculated in the emission inventory. The input file will be setup to model the 1 hour concentration and output the maximum 1 hour high 1 st high concentration plot file and post file. The second file will model the mechanically generated fugitive dust sources using the half-life option using the hourly PM 10 emission rates calculated in the emission inventory multiplied by 1.1 to account for the 10% of the PM 10 that is made up of PM 2.5 and is not subject to deposition. The output will be the maximum 1 hour high 1 st high concentration plot file and post file similar to the first input file. Using the Multi-Chem post-processing utility, these post files will be combined on an hour by hour basis for each receptor. Since the actual modeled hourly PM 10 emission rates will be accounted for in the post files, the emission rate text file that is required for the Multi-Chem utility will list the emission rates for the two post-files as 1 g/s. In this way, the half-life and non-half-life sources will be combined in time and space and the 24 hour PM 10 high 2 nd high plot file can be produced for determining the Increment results and the high 1 st high through high 6 th high plot files can be produced for determining the NAAQS results. Using this method provides the option of modeling half-life for specific sources without the necessity of creating individual input files and an overwhelming number of post files.

28 II. PM2.5 Secondary Formation This section describes the proposed 24-hour PM 2.5 NAAQS modeling methodology to account for secondary formation of PM 2.5 due to the plant site combustion sources. The modeling analysis will use the off-set ratio method described in the National Association of Clean Air Agencies (NACAA) Report (Reference (4)). The method multiplies the direct PM 2.5 emissions of combustion sources with an offset ratio to reflect the PM 2.5 created from secondary formation. The offset ratio represents the amount of NO X or SO 2 that contributes to PM 2.5 concentrations through secondary formation. This value is called the total equivalent primary PM 2.5 emission rate. The equation (Primary PM 2.5 [TPY] + [SO 2 TPY]/10 + [NO X TPY]/100) was used to calculate the total equivalent PM 2.5 emission rates for each of the combustion sources. The combustion sources used the hourly SO 2 and NO X emission rates converted to TPY to calculate the total equivalent PM 2.5 emissions. The offset ratios for NO X and SO 2 used in this equation are from the NAACA document and were developed in that report for a source located in Northern Minnesota. There will be nine combustion sources at the Plant Site (five considered intermittent sources) which make up only 3.4% of the total direct PM 2.5 emissions. Therefore, secondary particulate formation is not expected to contribute significantly to the maximum modeled PM 2.5 concentration. References: 1. U.S. Environmental Protection Agency. AERMOD Deposition Algorithms Science Document (Revised Draft). March 19, User s Guide for the Industrial Source Complex (ISC3) Dispersion Models: Volume II Description of Model Algorithms (EPA-450/ b). March Maxwell, Christine M. and Hodgin, C. Reed. Iron Range Air Quality Analysis. s.l. : Midwest Research Institute, p National Association of Clean Air Agencies. PM2.5 Modeling Implementation for Projects Subject to National. January 7, 2011.

29 Exhibit 1: Determination of the mass mean particle diameter of 7.4 µm Information previously available from USEPA s Air Pollution Training Institute (APTI) website (under Basic Concepts in Environmental Sciences) was used to develop the mass mean particle diameter (Equation 1-84 of Reference (2)), and is appended at the end of this discussion. The EPA has reconfigured its online access to APTI resources such that the information relied upon to develop the mass mean particle diameter is no longer readily available through the links presented, however, the particulate matter information presented below is general in nature and relevant to this analysis. A size distribution curve representative of mechanically generated particulates is shown in Figure 2 of Appendix A Ambient Particulate Matter Size Distribution appended at the end of this discussion (previously accessed at the following website: Figure 2 of Appendix A shows the frequency % particle by mass for three types of atmospheric particulates (ultrafine, fine, and coarse - supercoarse). Table 1 was developed from Figure 2 of Appendix A and shows the mass percent of particulates in the 3 10 µm coarse supercoarse range which are associated with mechanically generated particulates 1 which are the subject of the decay coefficient calculation. Interpolation of the coarse - supercoarse curve for the coarse particle size range (between 2.5 and 10 microns which is representative of the fugitive PM 10 modeling) of Figure 2 of Appendix A leads to a mass mean diameter of 7.4 microns as shown below in Table 1 and Figure 1 and Figure 2: 1 Formerly found on the Particle Formation page ( physical attrition (mechanically generated) particles are shown to be primarily greater than PM10 and correspond to the coarse - supercoarse curve on Figure 2 of Appendix A.

30 Table 1 PM10 Mass Percent Particle Size Frequency % Particles by Mass (1) Normalized Mass % Cumulative Normalized Mass % % 2.7% % 8.0% % 16.0% % 28.0% % 43.1% % 60.0% % 77.8% % 100.0% (1) Interpolated from Figure 2 of Appendix A for the midpoint of the category (e.g., 10 micron frequency % is midpoint between 9 and 10 microns). Figure 1 shows the cumulative mass % (normalized to 100%) derived from Figure 2 of Appendix A. Note that Figure 1 did not use a mathematical representation of Figure 2 of Appendix A as one was not provided in USEPA s references. Instead, Figure 1 was obtained by selecting the particle size and reading the mass % off of the Y-axis in Figure 2 of Appendix A. The total mass % was normalized to 100% as shown in Table 1. Figure 2 shows the linear interpolation for the cumulative mass percent between the 7 and 8 µm particle sizes in Figure 1. As shown in Figure 2, the 50% cumulative mass percent intersects the 7.4 µm particle size. Figure 1 PM10 Cumulative % Mass

31 Figure 2 PM10 Particle Size Distribution 7-8 µm

32 Appendix A Excerpts from USEPA s Air Pollution Training Institute (APTI) Basic Concepts in Environmental Sciences Module 3: Characteristics of Particles - Particle Size Categories EPA Particle Size Terminology Since the range of particle sizes of concern for air emission evaluation is quite broad it is beneficial to divide this range into smaller categories. Defining different size categories is useful since particles of different sizes behave differently in the atmosphere and the respiratory system. The EPA has defined four terms for categorizing particles of different sizes. Table 1 below displays the EPA terminology along with the corresponding particle sizes. Figure 1 provides a visual comparison of the size of a fine particle (1 ), coarse particle (10 ), and a supercoarse particle (100 ). There is a substantial difference in size between these particles, all of which are considered moderate-to-large in air pollution control.

33 Regulated Particulate Matter Categories In addition to the terminology provided in Table 1 the EPA also categorizes particles as follows: Total Suspended Particulate Matter (TSP) PM 10 PM 2.5 Particles less than 0.1 Condensable Particulate Matter These particle categories are important because particulate matter is regulated and tested for under these categories. The National Ambient Air Quality Standard for PM 2.5 was remanded by a District of Columbia court in May of 1999 and is under litigation as of the writing of these modules (December 1999). Air quality standards are presented in these modules as they were promulgated by the EPA, with no presumptions made regarding the outcome of the pending litigation. Figure 2 displays a typical size distribution of atmospheric particulate matter that combines the two classification schemes discussed above. Total Suspended Particulate Matter Particles ranging in size from 0.1 micrometer to about 30 micrometer in diameter are referred to as total suspended particulate matter (TSP). TSP includes a broad range of particle sizes including fine, coarse, and supercoarse particles.

34 PM10 The U.S. EPA defines PM 10 as particulate matter with a diameter of 10 micrometers collected with 50% efficiency by a PM 10 sampling collection device. However, for convenience in these modules, the term PM 10 will be used to include all particles having an aerodynamic diameter of less than or equal to 10 micrometers. PM 10 is regulated as a specific type of "pollutant" because this size range is considered respirable. In other words, particles less than approximately 10 micrometers can penetrate into the lower respiratory tract. The particle size range between 0.1 and 10 micrometers is especially important in air pollution studies. A major fraction of the particulate matter generated in some industrial sources is in this size range. PM 10 is discussed in more detail in Module 6.

35 Attachment J Background Concentrations and Determination of Nearby Sources to be Modeled This attachment provides a detailed discussion of background concentrations and the nearby sources to be explicitly modeled for the NorthMet Mine Site Class II Area air dispersion modeling cumulative impact assessment. The approach presented herein follows directly from EPA s March 1, 2011 Memorandum Additional Clarification Regarding Application of Appendix W Modeling Guidance for the 1-hour NO 2 National Ambient Air Quality Standard. The Determining Background Concentrations section of the EPA memorandum (starting on p.12) provides guidance applicable to all pollutants and is not limited to the NO 2 NAAQS. Relevant excerpts from EPA s March 2011 Memorandum are reiterated (bold added) as they provide the foundation for the proposed approach: p. 12 goal of cumulative impact analysis: The goal of the cumulative impact assessment should be to demonstrate with an adequate degree of confidence in the result that the proposed new or modified emissions will not cause or contribute to violations of the NAAQS. In general, the more conservative the assumptions on which the cumulative analysis is based, the more confidence there will be that the goal has been achieved and the less controversial the review process will be from the perspective of the reviewing authority. p.12 nearby sources to include: As noted in the June 29, 2010 memo, Section of Appendix W emphasizes the importance of professional judgment by the reviewing authority in the identification of nearby and other sources to be included in the modeled emission inventory, and establishes a significant concentration gradient in the vicinity of the source under consideration as the main criterion for this selection. Appendix W also suggests that the number of such [nearby] sources is expected to be small except in unusual situations. p. 13 monitoring data: Section b of Appendix W states that [t]ypically, air quality data should be used to establish background concentrations in the vicinity of the sources(s) under consideration. Section c further states that [i]f the source is not isolated, it may be necessary to use a multi-source model to establish the impact of nearby sources. While many applications will be required to include both monitored and modeled contributions to adequately account for background concentrations in the cumulative analysis, we believe that these statements imply a preference for use of ambient air quality data to account for background concentrations where possible.

36 p. 13 double counting: Many of the challenges and more controversial issues related to cumulative impact assessments arise in the context of how best to combine a monitored and modeled contribution to account for background concentrations. Addressing these issues requires an assessment of the spatial and temporal representativeness of the background monitored concentrations for purposes of the cumulative impact assessment and the potential for double counting of impacts from modeled sources that may be contributing to monitored concentrations. p. 16 concentration gradients: Even accounting for some terrain influences on the location and gradients of maximum 1-hour concentrations, these considerations suggest that the emphasis on determining which nearby sources to include in the modeling analysis should focus on the area within about 10 kilometers of the project location in most cases. The routine inclusion of all sources within 50 kilometers of the project location, the nominal distance for which AERMOD is applicable, is likely to produce an overly conservative result in most cases. The March 2011 memo also provides a methodology for the analysis which includes: Identify nearby sources to be included in the modeled inventory o The MPCA s SQRMD tools provides the first step in this inventory by identifying all permitted sources within 50 km of the project site as shown in Large Figure J-1. The tool uses 2013 actual emissions to provide a first-cut on sources which may cause a significant concentration gradient in the vicinity of the source. Large Figure J-1 also depicts those sources which exceed the SQRMD thresholds and would potentially be considered as nearby sources to explicitly model. Determine the source characteristics and local meteorological and topographical factors that determine the spatial and temporal patterns of the source s ambient impacts. A three step process is identified which includes preparation of modeled concentration contour plots, identifying meteorological conditions associated with maximum modeled concentrations, and preparation of a wind rose to depict general flow characteristics. o o o Significant Impact Level (SIL) results figures, meteorological conditions and other factors contributing to maximum modeled concentrations are discussed by pollutant below. The primary topographical formation near the NorthMet Plant Site is the ridge along the Mesabi Iron Range formation. Also of topographical importance are the landscape changes due to former mining activities including open pits and waste rock stockpiles. The wind rose for Hibbing, Minnesota is shown in Large Figure J-2. Predominant winds are from the northwest and southeast. As discussed further below, these wind directions do not line up with sources west of the Mine Site.

37 Determine the location and magnitude of air quality data from ambient monitors located within the area. If there are no monitors located in the vicinity of the source, a regional site may be used to determine background. A regional site is one that is located away from the area of interest but is impacted by similar natural and distant man-made sources. o o o o The Virginia, Minnesota (pop. 8,700) ambient monitor (MPCA ID 1300) was established in 1968 to measure dust impacts from taconite mining and related activities. The site currently monitors Total Suspended Particulate (TSP), PM 10, and PM 2.5. The monitor is located approximately 32 km west-southwest of the Plant Site roughly equidistant to United Taconite Thunderbird Mine, U.S. Steel MinnTac, and ArcelorMittal Minorca Mine. Notably, the monitor is located downwind of MinnTac for the predominant northwest wind direction. The monitor also is influenced by Virginia Public Utilities as well as population-related sources. One year of pristine SO 2 and NO 2 ambient monitoring data were collected at Ely, Minnesota during 2012 and 2013 (located approximately 40 km north of the Plant Site). Continuous PM 2.5 monitoring started in Ely, MN in 2013 and 3 years of data are available for developing a pristine background concentration. The Virginia, MN monitor better charcaterizes PM 2.5 impacts from regional industrial sources. The ambient monitors for SO 2 and NO 2 which will be used to develop background are located within 2 km of Flint Hills Resources refinery in Rosemount, Minnesota. Ambient concentrations from population-based man-made sources (e.g., automobile traffic) at the Rosemount monitors would be expected to be conservative with respect to ambient concentrations at the Plant Site. Receptor Grids One of the project-specific considerations for developing background concentrations and the nearby source inventory for a cumulative impacts assessment is the use of two model receptor grids at the Plant Site the Plant Site ambient air boundary (AAB) receptor grid and the Plant Site cumulative impacts assessment boundary receptor grid. The Plant Site AAB grid follows PolyMet s projected property boundary whereas the cumulative impacts boundary is the extent of the former LTVSMC ambient boundary and reflects the extent of all of the former mining areas within the LTV footprint and includes the adjacent sources of Mesabi Nugget and the Cliffs Erie Pellet Yard. The cumulative impacts assessment boundary is appropriate for determining cumulative impacts from Plant Site and nearby sources as all of the area within the former LTV boundary is zoned mining industrial and general public access is not allowed. The three facilities share a common gated entrance at the Plant Site administration building. SO2 Nearby Sources Large Figure J-3 shows the 1-hour SO 2 SIL analysis model results. The 1-hour SO 2 SIL is the most restrictive of the SO 2 SILs, so SIL runs for the other averaging periods (3-hour, 24-hour, and annual) are not presented. As shown in Large Figure J-3, exceedances of the SIL can be seen to the east of the Plant

38 Site cumulative impacts assessment boundary up to approximately 10 km from Plant Site sources. These SIL exceedances occur for very light (< 1 m/s) winds from the east with very low mixing heights (7 to 17 m). The MPCA s SQRMD tool identified four SO 2 sources (US Steel MinnTac, United Taconite, Virginia Public Utilities, and Laskin Energy Center) within 50 km which exceed the SQRMD threshold value and could potentially be explicitly modeled. 1 Laskin Energy Center now only burns natural gas and therefore is no longer a significant SO 2 source. MinnTac, United Taconite, and Virginia Public Utilities are greater than 30 km from the Plant Site and would not have combined impacts with Plant Site sources at locations and for meteorological conditions where the Plant Site exceeds the SIL. Therefore, these distant sources will not be explicitly modeled for SO 2, but their potential impacts will be conservatively accounted for in the background concentration (see following section). Although Mesabi Nugget has low SO 2 emissions (< SQRMD threshold), Mesabi Nugget could potentially have combined impacts with Plant Site SO 2 sources. Therefore, Mesabi Nugget will be explicitly modeled for SO 2. Mesabi Nugget modeling parameters are included in the Nearby Source Model Inputs spreadsheet posted to the Project Website. SO2 Background Concentration MPCA Monitor 443 is located approximately 1.5 km southwest of Flint Hills Resources oil refinery (2013 annual reported emissions of 689 tons SO 2 ) in Rosemount, Minnesota. The year average 99th percentile maximum daily 1-hour concentration at Monitor 443 is 2 ppb (5.2 µg/m 3 ). Due to the close proximity of this monitor to a large SO 2 source, this background value conservatively represents potential SO 2 impacts from the SO 2 sources identified by the SQRMD tool which are well beyond EPA s recommended 10 km distance for inclusion of nearby sources. For comparison, the 2012 single year 99th percentile maximum daily 1-hour concentration at Ely, Minnesota was also 2 ppb (5.2 µg/m 3 ). Although this represents the same design value as for the Rosemount monitor, the annual concentrations at the Ely monitor were much lower. The 1-hour concentration are quite low at both sites relative to the NAAQS and the Rosemount site shows minimal impact from the nearby major source. NO2 Nearby Sources Large Figure J-4 shows the 1-hour NO 2 SIL analysis model results. The SIL analysis included the Tier 3 Ozone Limiting Method (OLM) screening methodology. As shown in Large Figure J-4, 1-hour NO 2 modeling results exceed the SIL from 1 to 5 km outside of the cumulative impacts boundary. These exceedances occur for very light winds (< 1 m/s) for all wind directions. Note that the OLM run also used the very conservative default value of 0.5 for in-stack NO 2 :NO X ratios The MPCA s SQRMD tool identified six NO X sources (ArcelorMittal Minorca Mine, Mesabi Nugget, US Steel MinnTac, United Taconite Fairlane Plant, Virginia Public Utilities, and Laskin Energy Center) within 50 km 1 The SQRMD tool requires a source to be existing to determine distance to other nearby sources. The SQRMD tool s coordinates for the Cliffs Erie Pellet Yard were used for the NorthMet Mine Site location.

39 which exceed the SQRMD threshold value and could potentially be explicitly modeled. ArcelorMittal, MinnTac, United Taconite, and Virginia Public Utilities are greater than 30 km from the Plant Site and would not have combined impacts with Plant Site sources at locations and for meteorological conditions where the Plant Site exceeds the SIL. Therefore, these distant sources will not be explicitly modeled for NO 2, but their potential impacts will be conservatively accounted for in the background concentration. There is a potential for combined NO 2 impacts with Plant Site sources from both Mesabi Nugget and Laskin Energy Center. Laskin Energy Center and Mesabi Nugget source inventories and source parameters for modeling are included in the Nearby Source Model Inputs spreadsheet posted to the Project Website. NO2 Background Concentration MPCA Monitor 423 is located approximately 1.5 km northwest of Flint Hills Resources oil refinery in Rosemount, Minnesota. The year average 98th percentile of the daily maximum 1-hour concentrations at Monitor 443 is 27.3 ppb (51.3 µg/m 3 ). Flint Hills Resources reported 1,266 tons of NO X emissions in 2013, making the MPCA Monitor 443 site subject to greater potential NO 2 concentrations than would be encountered at the NorthMet Plant Site. Due to the close proximity of this monitor to a large NO 2 source, this background value conservatively represents potential NO 2 impacts from the NO X sources identified by the SQRMD tool which are well beyond EPA s recommended 10 km distance for inclusion of nearby sources. Additionally, Rosemount has a much greater population density than Hoyt Lakes, so NO 2 concentrations from population-based sources (heating and automobile traffic) also make ambient monitored concentrations at Monitor 423 conservative with respect to NO 2 concentrations from population-based sources near the NorthMet Plant Site. For comparison, the 2013 single year 98th percentile of the daily maximum 1-hour concentrations concentration from Ely, Minnesota was 6 ppb (11.3 µg/m 3 ). PM10 Large Figure J-5 and Large Figure J-6 show the Plant Site PM Hour and annual SIL analysis results. The receptors which exceed the SIL are similar for both runs, with the 24-hour SIL having a slightly larger extent and receptors over the SIL just to the north and east of the Plant Site cumulative impacts boundary. Maximum modeled results at the Plant Site occur during low wind speeds and stable atmospheric conditions. The MPCA SQRMD Tool identified seven PM 10 sources (ArcelorMittal Minorca Mine, Mesabi Nugget, US Steel MinnTac, United Taconite Fairlane Plant, United Taconite Thunderbird Mine, Northshore Peter Mitchell Mine, and Laskin Energy Center) within 50 km which exceed the SQRMD threshold value and could potentially be explicitly modeled. 2 Laskin Energy Center now only burns natural gas and so is no longer a significant PM 10 source. ArcelorMittal Minorca Mine, US Steel MinnTac, United Taconite Fairlane Plant, United Taconite Thunderbird Mine are located close to the Virginia monitor and the Virginia 2 The SQRMD tool requires a source to be existing to determine distance to other nearby sources. The SQRMD tool s coordinates for the Cliffs Erie Pellet Yard were used for the NorthMet Plant Site location. Consequently, the SQRMD tool did not identify Cliffs Erie Pellet Yard as a nearby source.

40 monitored concentrations include the contributions from these sources (Large Figure J-1). Additionally, the maximum modeled concentrations from NorthMet sources occur during low wind speed conditions, and not meteorological conditions which would result in transport from the distant sources to result in potential combined impacts with NorthMet sources. Mesabi Nugget and the Cliffs Erie Pellet Yard are sources with a potential for combined PM 10 impacts with the Plant Site sources. Note that the receptors exceeding the PM 10 SIL on the east side of the Plant Site are on Northshore Peter Mitchell Mine property, so the Northshore sources would not be included in a cumulative impact at that location. Mesabi Nugget and Cliffs Erie Pellet Yard source inventories and source parameters for modeling are included in the Nearby Source Model Inputs spreadsheet posted to the Project Website. PM10 Background Concentrations The PM 10 design value (3-year average of the high-2 nd -high monitored concentration for 24-hour, and 3- year average concentration for annual) will be used as a conservative single value background in the cumulative PM 10 NAAQS and MAAQS model results. Because modeled attainment with the PM hour PSD increment is the limiting standard for the PM 10 modeling analysis, further refinement (e.g., seasonal values) to the background concentration is not proposed. PM2.5 Nearby Sources Large Figure J-7 and Large Figure J-8 show the Plant Site PM Hour and annual SIL analysis results. The 24-hour PM 2.5 SIL is exceeded in all directions from the Plant Site, with the furthest SIL exceedance 24 km to the southeast of the Plant Site. Maximum concentrations occur during very light winds (< 1 m/s) and stable conditions. The MPCA SQRMD Tool identified five PM 2.5 sources (Mesabi Nugget, US Steel MinnTac, United Taconite Fairlane Plant, Northshore Babbitt Mine, and Laskin Energy Center) within 50 km which exceed the SQRMD threshold value and could potentially be explicitly modeled. 3 Laskin Energy Center now only burns natural gas and so is no longer a significant source of primary PM 2.5 emissions. US Steel MinnTac and United Taconite Fairlane Plant are located close to the Virginia monitor and the Virginia monitored concentrations include the contributions from these sources (Large Figure J-1). Additionally, the maximum modeled concentrations from NorthMet sources occur during low wind speed conditions, and not meteorological conditions which would result in transport from the distant sources to result in potential combined impacts with NorthMet sources. Mesabi Nugget and the Cliffs Erie Pellet Yard are sources with a potential for combined PM 2.5 impacts with the Plant Site sources. Note that the receptors exceeding the PM 2.5 SIL on the east side of the Plant Site are on Northshore property, so the Northshore sources would not be included in a cumulative impact at that location. Northshore sources are not proposed to be explicitly modeled as they would only have a 3 The SQRMD tool requires a source to be existing to determine distance to other nearby sources. The SQRMD tool s coordinates for the Cliffs Erie Pellet Yard were used for the NorthMet Plant Site location. Consequently, the SQRMD tool did not identify Cliffs Erie Pellet Yard as a nearby source.

41 combined impact with Plant Site sources on the western portion of the cumulative impacts assessment receptor grid. Northshore s crushers and shop boilers (which are Northshore s largest PM 2.5 emission sources) are located approximately 25 km from the western portion of the cumulative impacts assessment receptor grid well beyond EPA s recommended 10 km distance for inclusion of nearby sources. Similarly, NorthMet Mine Site sources are not proposed to be explicitly modeled for the Plant Site PM 2.5 modeling as Mine site sources are approximately 17 km from the western portion of the cumulative impacts assessment grid. Both the Mine Site and the Northshore Peter Mitchell Mine are similar source types as those represented by the Virginia monitor, so the Virginia background concentration provides a reasonably conservative representation of potential PM 2.5 impacts from Northshore and the Mine Site without double counting their impacts. Mesabi Nugget and Cliffs Erie Pellet Yard source inventories and source parameters for modeling are included in the Nearby Source Model Inputs spreadsheet posted to the Project Website. PM2.5 Background Concentrations The Virginia monitor will be used to develop background concentrations. The PM 2.5 design value (3-year average of the 98th percentile monitored concentration for 24-hour, and 3-year average concentration for annual) may be used as a conservative single value to determine the total PM 2.5 NAAQS model results. A possible refinement will be to use the 98th percentile value of the seasonal concentrations. The Virginia PM 2.5 monitor reflects the contributions from nearby sources as well as distant sources. To avoid potential double counting of nearby source impacts, using the design value concentration from the Virginia monitor will provide a sufficiently conservative and representative account of combined impacts from the more distant sources identified using the MPCA s SQRMD tool. The Ely monitoring data may be further considered for background purposes depending on the modeling results and or the final inventory of sources agreed upon for inclusion in the cumulative modeling. The year average 98th percentile PM hour concentration at the Virginia monitor is 18 µg/m 3. The year average PM 2.5 annual concentration at the Virginia monitor is 7.53 µg/m 3.

42 K o o c h i c h i n g C o u n t y Louis Leustek & Sons Inc - Nonmetallic Barr Footer: ArcGIS 10.4, :11 File: I:\Client\PolyMet_Mining\Work_Orders\Permitting\Air_Permit_Application\Maps\Report\AQDMP_01_PlantSite_Protocol\Figure J-1 Nearby Source Locations MPCA SQRMD Tool.mxd User: arm2 I t a s c a C o u n t y NO x Sources That Exceed SQRMD PM 2.5 Sources That Exceed SQRMD PM 10 Sources That Exceed SQRMD SO 2 Sources That Exceed SQRMD *Note: SQRMD Tool requires source to exist. Cliffs Erie LLC was used for plant site location. US Steel Corp - Minntac Hoover Construction Co - Nonmetallic Virginia Department of Public Utilities United Taconite LLC - Thunderbird Mine Ambient Monitoring Site United Taconite LLC - Fairlane Plant 10 km Buffer around Mine Site and Plant Site AABs 50 km Buffer Around NorthMet Plant Site Ambient Air Boundary (AAB) MnDNR Mine Featues, 2015 ArcelorMittal Minorca Mine Inc Virginia PM Monitor S t. L o u i s C o u n t y Mesabi Nugget Delaware LLC Mesabi Bituminous Inc- Schley Mine A Cliffs Erie LLC - Hoyt Lakes Minnesota Power - Laskin Energy Center I Kilometers Northshore Mining Co - Babbitt ISD John F Kennedy High School L a k e C o u n t y L a k e S u p e r i o r DRAFT NEARBY SOURCE LOCATIONS MPCA SQRMD TOOL NorthMet Project Poly Met Mining Inc. Large Figure J-1 NorthMet Plant Site Dispersion Modeling Protocol

43 Large Figure J-2 Hibbing Wind Rose

44 Barr Footer: ArcGIS 10.4, :21 File: I:\Client\PolyMet_Mining\Work_Orders\Permitting\Air_Permit_Application\Maps\Report\AQDMP_01_PlantSite_Protocol\Figure J-3 1-HR SO2 Modeled Concentrations Plant Site SIL.mxd User: arm2 SO 2 1-Hour SIL Radius = 10 km SO 2 1-Hour SIL Concentrations (µg/m 3 ) NorthMet Ambient Air Boundary (AAB) Approximate Northshore Ambient Air Boundary (AAB) Approximate LTVSMC Ambient Air Boundary (AAB) Approximate Mesabi Nugget Ambient Air Boundary (AAB) MnDNR Mine Featues, 2015 I Kilometers DRAFT 1-HR SO 2 MODELED CONCENTRATIONS PLANT SITE - SIL ANALYSIS NorthMet Project Poly Met Mining Inc. Large Figure J-3 NorthMet Plant Site Dispersion Modeling Protocol

45 Barr Footer: ArcGIS 10.4, :16 File: I:\Client\PolyMet_Mining\Work_Orders\Permitting\Air_Permit_Application\Maps\Report\AQDMP_01_PlantSite_Protocol\Figure J-3 1-HR NO2 Modeled Concentrations Plant Site SIL.mxd User: arm2 NO 2 1-Hour SIL Radius = 12 km NO 2 1-Hour SIL Concentrations (µg/m 3 ) NorthMet Ambient Air Boundary (AAB) Approximate Northshore Ambient Air Boundary (AAB) Approximate LTVSMC Ambient Air Boundary (AAB) Approximate Mesabi Nugget Ambient Air Boundary (AAB) MnDNR Mine Featues, 2015 I Kilometers DRAFT 1-HR NO 2 MODELED CONCENTRATIONS PLANT SITE - SIL ANALYSIS NorthMet Project Poly Met Mining Inc. Large Figure J-4 NorthMet Plant Site Dispersion Modeling Protocol

46 Barr Footer: ArcGIS 10.4, :38 File: I:\Client\PolyMet_Mining\Work_Orders\Permitting\Air_Permit_Application\Maps\Report\AQDMP_01_PlantSite_Protocol\Figure J-4 24-HR PM10 Modeled Concentrations SIL.mxd User: arm2 PM Hour SIL Radius = 7.8 km PM Hour SIL Modeled Concentrations (µg/m 3 ) NorthMet Ambient Air Boundary (AAB) Approximate Northshore Ambient Air Boundary (AAB) Approximate LTVSMC Ambient Air Boundary (AAB) Approximate Mesabi Nugget Ambient Air Boundary (AAB) MnDNR Mine Featues, 2015 I Kilometers DRAFT 24-HR PM 10 MODELED CONCENTRATIONS PLANT SITE - SIL ANALYSIS NorthMet Project Poly Met Mining Inc. Large Figure J-5 NorthMet Plant Site Dispersion Modeling Protocol

47 Barr Footer: ArcGIS 10.4, :59 File: I:\Client\PolyMet_Mining\Work_Orders\Permitting\Air_Permit_Application\Maps\Report\AQDMP_01_PlantSite_Protocol\Figure J-6 Annual PM10 Modeled Concentrations SIL.mxd User: arm2 PM 10 Annual SIL Radius = 5.4 km PM Hour SIL Radius = 7.8 km PM 10 Annual SIL Modeled Concentrations (ug/m 3 ) NorthMet Ambient Air Boundary (AAB) Approximate Northshore Ambient Air Boundary (AAB) Approximate LTVSMC Ambient Air Boundary (AAB) Approximate Mesabi Nugget Ambient Air Boundary (AAB) MnDNR Mine Featues, 2015 I Kilometers DRAFT ANNUAL PM 10 MODELED CONCENTRATIONS PLANT SITE - SIL ANALYSIS NorthMet Project Poly Met Mining Inc. Large Figure J-6 NorthMet Plant Site Dispersion Modeling Protocol

48 Barr Footer: ArcGIS 10.4, :30 File: I:\Client\PolyMet_Mining\Work_Orders\Permitting\Air_Permit_Application\Maps\Report\AQDMP_01_PlantSite_Protocol\Figure J-7 24-HR PM2_5 Modeled Concentrations SIL.mxd User: arm2 S t. L o u i s C o u n t y PM Hour SIL Modeled Concentrations (µg/m 3) PM Hour SIL Radius = 24 km PM 2.5 Annual SIL Radius = 9 km NorthMet Ambient Air Boundary (AAB) Approximate Northshore Ambient Air Boundary (AAB) Approximate LTVSMC Ambient Air Boundary (AAB) Approximate Mesabi Nugget Ambient Air Boundary (AAB) MnDNR Mine Featues, 2015 I Kilometers DRAFT 24-HR PM 2.5 MODELED CONCENTRATIONS PLANT SITE - SIL ANALYSIS NorthMet Project Poly Met Mining Inc. Large Figure J-7 NorthMet Plant Site Dispersion Modeling Protocol

49 Barr Footer: ArcGIS 10.4, :33 File: I:\Client\PolyMet_Mining\Work_Orders\Permitting\Air_Permit_Application\Maps\Report\AQDMP_01_PlantSite_Protocol\Figure J-8 Annual PM2_5 Modeled Concentrations SIL.mxd User: arm2 S t. L o u i s C o u n t y PM 2.5 Annual SIL Modeled Concentrations (µg/m 3 ) PM Hour SIL Radius = 24 km PM 2.5 Annual SIL Radius = 9 km NorthMet Ambient Air Boundary (AAB) Approximate Northshore Ambient Air Boundary (AAB) Approximate LTVSMC Ambient Air Boundary (AAB) Approximate Mesabi Nugget Ambient Air Boundary (AAB) MnDNR Mine Featues, 2015 I Kilometers DRAFT ANNUAL PM 2.5 MODELED CONCENTRATIONS PLANT SITE - SIL ANALYSIS NorthMet Project Poly Met Mining Inc. Large Figure J-8 NorthMet Plant Site Dispersion Modeling Protocol