BUKA ENVIRONMENTAL

941 9th Street Boulder, CO 80302 USA 303.324.6948/aamaest@gmail.com

Technical Memorandum
To: From: Date: Re: Kathryn Hoffman, Minnesota Center for Environmental Advocacy Ann Maest, PhD; Buka Environmental 12 March 2014 Comments on Polymets Supplemental Draft Environmental Impact Statement: Water quality and geochemical issues

Introduction
The comments contained herein are presented on behalf of the Minnesota Center for Environmental Advocacy. I appreciate the opportunity to comment on the NorthMet Supplemental Draft Environmental Impact Statement (SDEIS; MDNR et al., 2013). My comments focus on the geochemical and water quality aspects of the SDEIS and associated documents. I present a summary of my major findings and then discuss my non-technical and technical comments. The technical comments address baseline water quality and treatment targets, water management during mining, mine waste characterization, and model inputs, results, and uncertainties. A short list of recommendations and a conclusion are presented after the technical comments. The documents I relied on for my evaluation are listed in the References section of the memorandum.

Summary of Major Findings

o o The number of samples of NorthMet Project material in the waste characterization program is inadequate for the volume of wastes predicted to result from mining. Assumptions about the lack of or delay in acid generation and the lack of substantial metal/contaminant leaching are poorly substantiated and conflict with actual results. The SDEIS and associated documents declare that Category 1 wastes will never generate acid. Given operational realities and existing characterization results, this seems like a poor assumption and does not consider the potential to leach other contaminants, including those added by mining. Category 4 and some Category 2/3 wastes will likely go acidic more quickly than predicted in the model. If this is the case, submerging them in the pit may not be an effective management strategy. The SDEIS and associated documents do not discuss or even mention the need to dewater the area around the pits (using pumping wells) during mining. Dewatering wells will likely be required not just sumps in the pits themselves and no plans are in evidence for the treatment of pumped dewatering water during operation. At a minimum, the dewatering water is expected to have elevated concentrations of ammonia, nitrate, and sulfate and will require treatment. The treatment pilot tests assumed dewatering water would have nearbackground levels of nitrate and ammonia, so the effectiveness of the methods for removal of these constituents under mining conditions is not known.

Ann S. Maest, PhD

Geochemist

The Category 1 wastes are the only wastes that will remain on the surface in perpetuity. No liner is proposed for the facility, and the mine plan assumes that seepage collection and treatment will no longer be necessary at some unspecified time after mine closure. The leaching results for this waste category and the closely associated Category 2/3 wastes, and the likelihood that small differences in percent sulfur can readily be separated, suggest that releases from the pile could be more problematic than assumed in the SDEIS. Category 1 wastes will likely include wastes currently designated as Category 2/3 wastes, which will leach higher concentrations of contaminants. The modeling approach uses Monte Carlo simulations, which in theory cover a wide range of possible inputs and outcomes. However, because of limits placed on concentrations leached from the wastes and mined materials and the use of averages, the high end of potential contaminant concentrations has been excluded from consideration in the model. Monte Carlo simulations used in the water quality model do not simulate the timing or characteristics of natural processes or the potential environmental consequences of the interaction of these processes with mined materials. The model assumes that essentially all contaminants will be adsorbed or otherwise removed as mine-affected waters travel along prescribed flowpaths to receptors. The model does not simulate natural processes that could cause elevated concentrations of mine-related contaminants to reach groundwater or surface water resources, including flushes of leached contaminants during early snowmelt or other hydrologic events or movement along preferential pathways.

General Non-technical Comments

The amount of material to review for this proposed mine is overwhelming. Although its good to have a lot of data and information, some important material is missing, and data and information transparency is low especially in the most recent Waste Characterization Data Package (WCDP; Polymet Mining, 2013a). For example, the original concentrations for the humidity cell tests (HCTs) are not contained in the SDEIS or anywhere in the 808-page WCDP. Normally, an appendix will contain all measured concentrations. Instead, only summaries of the overall rates of release, as mg/kg/week, and only for a given statistical slice of the data (e.g., 95th percentile) are contained in the WCDP. After two document requests to the Minnesota Department of Natural Resources (MDNR), we received the original concentration data. However, the results for each sample were contained in individual Excel worksheets, and this made data evaluation and comparisons cumbersome. If more time had been provided after receiving the data, I would have done a more thorough evaluation of the humidity cell test results and compared rates used as inputs to the model against the original data. The WCDP is convoluted and difficult to follow; by comparison, the earlier waste characterization report (SRK Consulting, 2007) was substantially better written and more transparent in its information. One of the issues is numerous changes in approach that have occurred over time, including major changes in crucial cutoff values for waste categories. The primary purpose of waste rock characterization, according to SRK Consulting (2007b) was to divide the wastes into reactive and non-reactive waste as defined in Minnesota Rules. For example, initially, wastes were categorized according to sulfur content and copper:sulfur ratio. That approach was abandoned, and sulfur content alone was used to create waste categories. Wastes with a sulfur content greater than 0.05% were considered high-sulfur wastes (reactive), and low-sulfur wastes were those with values <0.05%S (non-reactive). SRK subsequently designated wastes into categories 1/2, 3, and 4; later the categories became 1, 2/3, and 4 (SRK Consulting, 2007; Polymet Mining, 2013a). The only graphs available for the HCT results displayed

Ann S. Maest, PhD

Geochemist

the results combining all samples with sulfur values less than and greater than 0.05% S. Because of the new cutoffs (e.g., 0.12 %S for Category 1 wastes), its difficult to see which categories are leaching which ranges of concentrations. As another example, the approach for characterizing leaching potential changed over time. Initially the input values were based on the HCT results later they made the inputs based on results from an aqua regia digestion1 for some but not all samples, and ultimately, values were based on a ratio of metal to sulfur in the SOLID samples, multiplied by concentrations of metals in the leachate (in solution) for some but not all samples. At some point, a short-term sequential leaching procedure (sequential meteoric water mobility procedure, or SMWMP) that used only six samples was also used to characterize the leaching potential of Category 1 waste rock. Some combination of these leachate values or rates based on the concentrations over time were used as inputs to the water quality model, and predicted concentrations in pit water, groundwater and surface water are based on this confusing array of inputs. Polymet should provide a concise summary of the changes in characterization approaches and model input values over time and state clearly which results were finally used to estimate water quality in the model. The original geochemical and water quality data (not only graphs) should be presented on the MDNR website in a user-friendly electronic format, with transparent information on the sources of all data.

Technical Comments
1. Baseline Water Quality and Water Treatment Targets
Baseline sulfate concentrations in surface water are close to the wild rice limit in many locations, and values exceed wild rice limits (10 mg/L sulfate) at certain locations at certain times of the year (SDEIS, Figure 4.2.2-3 and Table 4.2.2-14). Mercury concentrations in streams and lakes are close to or exceed allowable limits (1.3 ng/L because waters are subject to the Great Lake Initiative see SDEIS, p. 5-8) in a number of locations. As noted in the SDEIS (p. 4-24), all waters on the site are designated Outstanding International Resource Waters, and additions of bioaccumulative substances such as mercury are not allowed without an approved nondegradation demonstration. Sulfate and mercury are contaminants of concern for the NorthMet Project, and point and nonpoint discharges from mine operations will need to be severely limited. Model-predicted concentrations of mercury in the West Pit Lake and tailings seepage are very close to the standard, and the more conventional treatment approaches proposed during operations will not remove mercury to meet the standard; in fact, the lowest concentrations achievable were approximately three times the water quality standard (Barr Engineering Company, 2007a). Given the uncertainties in modeling, it is likely that mercury loads could increase to the Embarrass River (as stated in the SDEIS, p. 5-8) and the Partridge watershed. At a minimum, an independent review of potential mercury loadings and a nondegradation demonstration should be completed as part of the Supplemental Final Environmental Impact Statement (SFEIS). Treatment goals for effluent during operation, remediation, and closure are presented in Table 1 for a number of constituents. The lowest (i.e. most protective) relevant water quality evaluation criteria are also included in Table 1. The effluent targets highlighted in yellow exceed the most protective project evaluation criteria. Note that more targets meet the evaluation criteria during long-term closure. All treated effluent should at least meet the evaluation criteria for the NorthMet Project,
1

Aqua regia is a strongly acidic solution made up of hydrochloric and sulfuric acids. It will dissolve most oxide, carbonate, hydroxide, and sulfide minerals but will not dissolve silicate minerals, which can contain some of the contaminants of concern (Balcerzak, 2002).

Ann S. Maest, PhD

Geochemist

and there is no technical reason that they cant. Barr Engineering Company (2007a, p. viii) agrees that treatment targets should consider the most protective water quality standards: Because water will need to be moved between the Mine Site and the tailings basin, and because some of the water that will be stored in the tailings basin may eventually be returned to the West Pit or flow from the tailings basin to the natural groundwater, the process water quality targets established for treatment of the Mine Site process waters also considered the State of Minnesota surface water quality standards for Class 2 waters and the groundwater protection standards. These values establish more conservative water quality standards than those required in the Beneficiation Plant, but maintaining these concentrations throughout the operation significantly reduces the long term impact of the mining and processing operations on water quality after closure. The use of RO treatment during closure allows for more protective targets. It strikes me as odd that the better, more expensive, and more operationally difficult treatment methodology would be applied after the mine closes. Although the SDEIS and the mine plan state that treated water would not need to be discharged during the operational phase, as noted in Section 2, I believe that inflow volumes may exceed those predicted. In either case, a contingency should be included for discharge of treated water during operation that meets the most stringent water quality standards and criteria. Table 1. Wastewater treatment facility effluent targets for selected constituents during operation, reclamation, and long-term closure. WWTF Effluent WWTF Effluent WWTF Effluent Most Protective Targets Targets Targets Relevant Water Operation1 Reclamation1 Long Term1 Quality Evaluation Constituent (mg/L) (mg/L) (mg/L) Criteria2 Alkalinity (as CaCO3) <1 <1 <1 -Ammonia (as N) 0.040 --0.040b As 0.01 0.01 0.01 0.002c B Be Ca* Cd Cl Co Cua Hg Mg* Ni
a a

0.5 0.004 <1 0.004 230 0.005 0.03 1.3x10-6 <1 0.100 10

0.5 0.004 <1 0.004 230 0.005 0.016 -<1 0.090 --

0.5 0.004 <1 0.002 230 0.005 0.009 -<1 0.050 --

0.5 0.0002c -0.0025c 100d 0.0028b 0.0093c 1.3x10-6c -0.052c 10

Nitrate

Ann S. Maest, PhD

Geochemist

Table 1. Wastewater treatment facility effluent targets for selected constituents during operation, reclamation, and long-term closure. WWTF Effluent WWTF Effluent WWTF Effluent Most Protective Targets Targets Targets Relevant Water Operation1 Reclamation1 Long Term1 Quality Evaluation Constituent (mg/L) (mg/L) (mg/L) Criteria2 Pba Sb Se SO4 Zna Hardness (as CaCO3) 0.019 0.031 0.005 250 0.388 250 0.007 0.031 0.005 150 0.200 200 0.003 0.031 0.005 9 0.100 100 0.0032 0.0055b 0.005 10e 0.120c 250

Sources: 1 Adapted from 2012 Excel spreadsheet received from MDNR titled WWTF_Targets.xls, Table 1-34, Waste Water Treatment Facility Effluent Targets and Passive Treatment Parameters (Barr Engineering Company, 2007a); no target was included for Hg, nitrate, or ammonia, but process water quality targets were included in Table 4. 2 SDEIS, Tables 5.2.2-4 (surface water) and Table 5.2.2-2 (groundwater). WWTF=waste water treatment facility a water quality standard is hardness-dependent, evaluation criteria for surface water calculated at 100 mg/L hardness - if hardness is lower, standard will be lower; b for Class 2B waters (for aquatic life); c site-specific evaluation criterion for bedrock aquifer (surface water evaluation criterion is 0.004); c for Lake Superior Basin waters; d for Class 3B waters; e for Class 4A waters. * Calcium (Ca) and Magnesium (Mg) are the components of hardness. Although the hardness target ranges from 100 to 250 mg/L as CaCO3, the Ca and Mg targets are <1 mg/L.

2. Water Management during Mining

According to the SDEIS (Chapters 2 and 4) and the Water Management Plans (Polymet Mining, 2013b), groundwater and surface runoff flowing into the pits during mining will be collected in sumps and transported to the waste water treatment facility (WWTF) for treatment by chemical and filtration methods (Polymet Mining, 2013c). There is no mention of dewatering using wells around the pits. Without dewatering wells, the degree of hydrologic control outside the pit is uncertain. Even with dewatering wells, mine water capture is uncertain because of the fractured nature of the bedrock. Barr Engineering Company (2007a) states that there are four sources of mine site process water: groundwater entering the mine pit, direct precipitation, infiltration from stockpiles, and runoff from other mine site operations. The predicted water quality of leachate from waste rock stockpiles, leachate from the lean ore surge pile stockpile, and East Pit (I assume this is groundwater, although it does not say) is listed in tables in the document and summarized for the East Pit in Table 2. Predicted concentrations of sulfate, nitrate, and ammonia are similar to those in native groundwater; therefore, Polymet is assuming that pumped or collected groundwater near a blasted, potentially acid-generating pit will be changed from mining only in its metal content and then only slightly (Barr Engineering Company, 2007a, Table 2). Predicted concentrations of nitrate and ammonia in stockpile drainage were similar to values in Table 2, but predicted stockpile sulfate concentrations were much higher in the thousands of mg/L, reaching 9,600 mg/L by year 5, 10, or 15 of mining for all stockpiles. In contrast to these low predicted leachate and pit water concentrations, dewatering water from wells near the pits could have elevated concentrations of nitrate and ammonia from blasting agents,

Ann S. Maest, PhD

Geochemist

sulfate from oxidation of sulfides or unexpected infiltration of stockpile seepage, and potentially metals and other contaminants. Nitrate and ammonia will also be present in water collected in sumps in the open pits during operation, and treatment methods proposed for the operational phase will not effectively remove nitrate, ammonia, or sulfate, as reflected in the water quality targets shown in Table 1. At a minimum, removal of these components at higher concentrations has not been tested. The Pilot Testing Report (Barr Engineering Company, 2013) does not address removal of nitrate or ammonia and lists no treatment targets for these parameters. The water used for testing of the treatment process had very low nitrate and ammonia concentrations (generally at or below detection), and the testing program therefore did not adequately investigate removal of these constituents. Additional treatment pilot tests should be conducted to evaluate the removal of nitrate and ammonia at reasonable expected concentrations (at least 100 mg/L as N of each). Table 2. Predicted concentrations of ammonia, nitrate, and sulfate in East Pit water during mining. Parameter (mg/L) Year 1 Year 5 Year 10 Ammonia 0.085 0.093 0.092 Nitrate 0.085 0.093 0.092 Sulfate 37.4 25.4 26.3 Source: Barr Engineering Company, 2007a, Table 2. Concentrations of nitrate and sulfate in dewatering water at the Buckhorn Mine in far northern Washington State have had elevated concentrations of ammonia, nitrate, and sulfate over time, relative to baseline groundwater quality, as shown in Figure 1. To date, acid drainage has not been generated in the underground mine, and the pH of the dewatering wells is near neutral. Calculated site-wide baseline groundwater concentrations are 0.1 mg/L ammonia (as N), 1.33 mg/L nitrate+nitrite (as N), and 69.5 mg/L sulfate (Washington Department of Ecology, 2014, Table 7). As shown in Figure 1, nitrate and ammonia concentrations increased almost immediately, and sulfate concentrations took longer to increase. Nitrate and sulfate in mine sumps and waste rock seepage are also elevated; for example, nitrate and sulfate concentrations in waste rock seepage were 166 mg/L as N for nitrate and 1,100 mg/L for sulfate in June 2010. Polymets own documents show that nitrate and ammonia could be quite elevated in mine water during operations. Polymet Mining (2013a, Section 8.4.2) conducted a scoping-level analysis of possible nitrate and ammonia concentrations in waste rock seepage, and concluded that nitrate concentrations leaking from the Category 2/3 waste rock stockpile during operation would range from 100 to 300 mg/L (as N). I have not found a subsequent analysis of treated water that used a starting solution with nitrate values in this range. In the scoping-level analysis, ammonia concentrations in waste rock leachate were not reported. However, Polymet calculated that ammonia concentrations at the site boundary (deriving from escaped waste rock leachate) would likely not be higher than 1.5 mg/L (as N), and that with dilution, nitrate and ammonia concentrations in surface water would not exceed water quality standards. However, many rough assumptions go into these calculations, and they only considered waste rock seepage effects, not the potential effects of other uncontrolled nonpoint sources or treated discharge during mining. If excess mine water needs to be discharged during operations and it is not adequately treated, it could increase concentrations of these three constituents above levels that would be harmful to human health, aquatic biota, and wild rice. Polymet assumes that the water balance will work out perfectly during mining, and all collected mine-influenced water will be used in the mine processes or stored in pits or the tailings facility. However, as recently as February 2013 (Polymet, 2013b), the amount of water needed for the plant had not been determined for early and later mining years for any component of the plant site water

Ann S. Maest, PhD

Geochemist

balance(Figure 2). And the estimate given for Mine Year 10 is only an annual average value, with no indication of the possible range and uncertainty in the single number. If dewatering operations produce more water than can be used or stored and this is a reasonable possibility during a large storm or snowmelt event the excess dewatering water will require treatment before being discharged to streams. A graph from a mine site water balance study (Barr Engineering Company, 2007b) is included as Figure 3 in these comments. Polymet assumes that all process water (components shown in Figure 3) would be used in the mine processes for years 1 11 (Barr Engineering Company, 2007a (e.g., at the mill)). The largest component of process water is groundwater predicted to flow into the pit (Pit Groundwater in Figure 2). According to the work plan for the water balance study (Polymet Mining, 2007X, Appendix A, p. 2), flow rates of groundwater into the pits will be estimated using MODFLOW. The work plan also states: Fractured bedrock, such as is found at the mine site, is inherently difficult to simulate and predict even when extensive hydrogeologic data is available, which is not the case for the mine site. Even with this admission of large uncertainty, there is no contingency plan for pumping of groundwater surrounding the pit or for discharge to surface water if volumes exceed those needed or able to be stored. The mine water balance should be recalculated with the volume of dewatering water from pumped groundwater wells included. An estimate of dewatering water quality and a plan for treatment of dewatering waters should also be included in the SFEIS. Figures showing the volumes of water for each component of the water balance, under low stream flow/groundwater elevation and high stream flow/groundwater elevation conditions should be included in the main body of the SFEIS. It is important that the water balance issue, especially whether discharge of treated water will be needed during operation, be sorted out before the SFEIS is completed. If discharge is required during operation, at a minimum, a contingency NPDES permit will be required well before mining begins.

Ann S. Maest, PhD

Geochemist

Figure 1. Concentrations of (a) ammonia, (b) nitrate, and (c) sulfate in the original five dewatering wells at the Buckhorn Mine, northern Washington State, USA. Site-wide background groundwater concentrations are shown as a green dashed line (0.1 mg/L ammonia (as N), 1.33 mg/L nitrate+nitrite (as N), and 69.5 mg/L sulfate).

(a)

(b)

Ann S. Maest, PhD

Geochemist

(c)

Data source: Washington Department of Ecology, 2014.

Ann S. Maest, PhD

Geochemist

Figure 2. Estimates of water needed for ore processing at the plant site.

Source: Polymet, 2013b, Table 2-1. Figure 3. Process water generated at the mine site during mining. Note that with climatic variability, maximum process water volumes peak at ~3,700 ac-ft/yr (Barr Engineering Company, 2007b, Figure ES-C).

Ann S. Maest, PhD

Geochemist 10

Source: Barr Engineering Company, 2007b, Figure 4.

3. Mine Waste Characterization

3.1 Number of waste characterization samples
According to the most recent waste characterization report prepared by Polymet Mining (2013a), only 89 samples of waste rock from actual NorthMet project core have been characterized (Table 1)2. The number amounts to one sample for each 3.5 million tons of waste rock. I believe that the number of samples used for waste characterization of the NorthMet Project relative to the tonnage is lower than that for any other mine Environmental Impact Statement Ive reviewed3. Although no hard-and-fast requirements exist for the recommended number of samples, the GARD Guide (2014) includes the following suggestion, taken from Australian guidance: Static testing of several hundred representative samples of high and low grade ore, waste rock and tailings, the number dependent on the complexity of the deposit geology and its host rocks. Static testing refers to tests such as sulfur content, acid-base accounting, and whole rock analysis that do not predict changes in the environmental behavior of the mined materials over time. Each of the 89 samples was subjected to humidity cell testing, whole rock analysis (for metals), mineralogic analysis, physical analysis, and sulfur and carbonate content. No traditional acid-base accounting (ABA) analyses were conducted on the samples only sulfur, carbonate, and paste pH (SRK Consulting, 2007). The near total absence of carbonate minerals (SRK Consulting, 2007, p. i) strongly suggests that ABA test results would indicate that all samples are potentially acid generating. Instead, Polymet is relying on silicate weathering to counteract any acidity formed in the Category 1 wastes. MDNRs research results, described in detail in Polymet Mining (2007 and 2013a) show that over the long-term, the ABA predictions that categorize the waste as potentially acid generating might not be that far off. Table 1. Sulfur content, tonnage, numbers of waste characterization samples, and tons/sample for each waste category. Waste Sulfur Mean Sulfur Amount of # Waste One Rock Content Content Material Characterization Sample per Category (%S) (%S) Expected (tons) Samples* X Tons Category 1 Category 2/3 Category 4 0.12 0.12 0.6 > 0.6 0.06 0.21 1.90 217,000,000 83,000,000 8,600,000 42 26 21 5,170,000 3,190,000 410,000

Sources: Polymet Mining, 2013a, Section 4.3, p. 15; Polymet Mining, 2012, Table 2-1. *Humidity cell test, whole rock analysis, sulfur and carbonate content conducted on each sample. The number of humidity cell tests (HCTs; 89 in total, or possibly 92) is reasonable, but usually many more ABA and whole rock analyses are performed on materials that will become waste rock. I have reviewed information for mines in Nevada, for example, in which literally hundreds of ABA
2

In the Executive Summary of an earlier report (SRK Consulting, 2007), SRK states, PolyMets extensive test program and includes 92 ASTM humidity cells and 20 kinetic tests using MNDR methods. 3 I have reviewed approximately 200 mining Environmental Impact Statements and Assessments.

Ann S. Maest, PhD

Geochemist 11

measurements were made for each geochemical test unit for Environmental Impact Statements and each result is presented with an individual sample number rather than only as statistical compilations. The larger number of these less expensive and less time-consuming tests provides more information on possible environmental behavior of the waste materials, and sometimes relationships can be reasonably drawn between ABA or whole rock results, for example, and HCT results. Although a large number of total sulfur analyses were conducted for the block model, the results are only presented as statistical composites, e.g., mean or 95th percentile %S values for the three (previously four) waste categories or the seven geologic units (Figure 4). Figure 4. Generalized stratigraphic column showing units within the NorthMet deposit.

Source: Polymet Mining, 2013c.

Ann S. Maest, PhD

Geochemist 12

Polymet Mining repeatedly makes the argument that stratigraphic units within the deposit, and even within the Duluth Formation as a whole, are so similar in terms of their environmental behavior that distinctions should not be made (see, e.g., WCDP, p. 16). According to Polymet Mining (date unknown, p. 2): Stratigraphic unit is not a significant driver to the acid producing or metal leaching rates. (See Stephen Day memo of March 31, 2008, RS42 Agency Work Plan Request for Section 1.1 DRAFT). This may be true based on geology, but they do not have the waste characterization information to reliably back up these statements in terms of environmental behavior. Similarly, leaching results from samples from similar deposits is presented, but the extent to which these are similar enough to the NorthMet deposit is currently unknown. MDNR, most notably Kim Lapakko, has conducted a truly impressive long-term geochemical research program on the Duluth Deposit (see, e.g., MDNR, 2013), and they find that (MDNR, 2013, p. 1): Although there is some homogeneity in the bulk mineral content of rock throughout the Duluth Complex, the degree to which compositional variations within the complex might affect the water quality resulting from rock dissolution has not been determined. Assuming some relevance of all the testing on the Duluth Formation materials, and given the most striking results of Kim Lapakkos research that wastes with very low sulfur content (0.2 %S) can produce acid over the longer-term MDNR cannot rule out the possibility that all wastes produced from the NorthMet Project will produce acid or leach metals within the 20-year timeframe of proposed mining.

3.2 Ability to Separate Category 1 from Category 2/3 Wastes

If mining goes forward at the NorthMet Project, wastes will be separated in the field, based on initial estimates of waste category (e.g., Category 1 vs Category 2/3), and composite samples will possibly be taken to confirm the sulfur content of the piles before final placement. Because of this estimating and compositing, it is highly unlikely that the sulfur content of waste rock that ends up in the Category 1 Stockpile will only contain material with a %S of 0.12. As noted above, the categorization of the waste rock has undergone a number of modifications, and now all waste rock is considered reactive according to Minnesota Rules (determined to be, for this project, above a %S of 0.05). In previous versions of the waste characterization reports, Category 1 and Category 2 waste rock were combined. At that time the average %S content of the combined category was 0.08, a mere 0.02% S higher than the current average %S for Category 1 waste rock. At that time, as shown in Figure 3a, the category contained all seven stratigraphic units (1 through 7; see Figure 2). Although all units except for one had average %S values that met the current requirement for Category 1 wastes ( 0.12 %S), the maximum sulfur content for each unit was well into the sulfur content for current Category 2/3 (0.12 0.6 %S). The more recent waste characterization report (Polymet Mining, 2013a) proposes that all stratigraphic units in Category 1 have maximum %S values below 0.12 (Figure 3b), but the reality could be quite different.

Ann S. Maest, PhD

Geochemist 13

Figure 3. Predicted percent sulfur in (a) Category 1/2 waste rock in the block model and (b) Category 1 and 2/3 in the block model. (a)

Source: Modified from: Polymet Mining, date unknown, Table 1. (b)

Source: Modified from Polymet Mining, 2013a., Large Table 2. SRK Consulting (2007) also discusses the difficulty in separating rocks with similar sulfur contents and the implications for environmental behavior in the field: Predictions of drainage chemistry from Category 2 rock are susceptible to the assumption that the overall conditions within the waste rock will remain non-acidic and the composition will reflect rock classified as Category 2 in the block model. Under operational conditions, these assumptions may be affected by the accidental inclusion of small amounts of Category 3 and 4 rock that could become localized sources of acidic water and leaching metals. The effect of these inclusions could be to contribute to metal leaching and lowering of pH resulting in higher concentrations of metals in the drainage. Category 3 and 4 rock could

Ann S. Maest, PhD

Geochemist 14

become incorporated into Category 2 rock by a number of routes which could include waste heterogeneity (i.e. small-scale inclusions of Category 3 and 4 rock in Category 2 rock) and operational errors. The latter are factors such as mistakes at the operating face and dumping location. These errors will be minimized by management practices but some level of operational mishaps can be expected. Of course the same caveats apply to all waste rock categories. Based on the acknowledged difficulty in excluding material from higher-sulfur waste categories and the uncertainty associated with the leaching characteristics of any waste category, it is reasonable to assume that leachate concentrations from the Category 1 and other stockpiles could be higher than predicted. If this is the case, the environmental protection of groundwater, surface water, and aquatic biota relies on the optimal functioning of the engineered mitigation measures. After reviewing information on the predicted and operational environmental behavior of a number of large-scale metal mines in the United States (Kuipers and Maest, 2006, which found that water quality exceedences at 64% of case study mines were due to mitigation failures) and observing in great detail the environmental performance of the Buckhorn Mine (a six-year old underground gold mine in fractured bedrock in Washington State that has not been able to create or maintain a capture zone), my confidence in this assumption is fairly low - given current mining and mitigation techniques. The NorthMet Project is close to water resources and has a moderate to high acid generation and contaminant leaching potential. Case study mines in this category were more likely to have adverse effects to groundwater and surface water resources (Kuipers and Maest, 2006). Of the 25 case-study mines, only 60% and 68% exceeded standards in surface water and groundwater, respectively. With the inherent factors present at the NorthMet Project, 85% and 93% of the case study mines exceeded water quality standards in surface water and groundwater, respectively. Given the inherent characteristics of this project and the multiple operational uncertainties, and even though it would be more expensive, a more protective approach would be to backfill all waste rock, including Category 1 wastes. Polymet Mining (2013a, p. 2) notes that all waste rock will be managed as reactive under Minnesota Rules, part 6132.2200). If reactive wastes are not stored or modified to render than non-reactive, the operator much permanently prevent essentially all water from moving through the waste and collect and dispose of any draining residual waters. The Category 1 Stockpile would be permanent, but the leachate collection system would not be maintained in perpetuity. According to the SDEIS (Chapter 3), the Category 1 Stockpile containment system would be keyed into bedrock. However, the upper portion of the bedrock appears to have higher hydraulic conductivity than deeper bedrock, and the potential for movement of contaminated seepage along this preferential pathway was not evaluated in the SDEIS. Myers (2014) did evaluate this pathway and found that seepage could flow to surface water. At a minimum, lining of the Category 1 Stockpile should be considered as providing a net environmental benefit in the SFEIS.

3.3 Assumptions about Acid Drainage and Contaminant Leaching, and Inputs to Environmental Models
3.3.1 Assumptions about acid drainage and contaminant leaching

Ann S. Maest, PhD

Geochemist 15

The assumptions about the environmental behavior of NorthMet mined materials are most clearly stated in the earlier waste rock characterization report (SRK Consulting, 2007). I have found no evidence that these assumptions have changed markedly in the more recent documents. SRK Consulting (2007) evaluated the nearly two-decade-long MDNR studies and approximately one year (at the time) of Polymets leaching results and arrived at the following four conclusions for Duluth Complex rocks (p. i): o Acidic leachate has not developed after 18 years of testing of samples containing less than 0.4% total sulfur despite the presence of sulfide minerals and near total absence of carbonate minerals. o Sulfide oxidation rates are directly proportional to sulfur content with a secondary relationship related to sulfide mineralogy. Other variables (rock type, position in intrusive stratigraphy) do not appear to be important. o If acidic conditions develop, the transition to acidic pHs takes many years. Following the onset of low pH, and depletion of sulfide minerals, pHs recover. o Metal leaching is strongly linked to pH. Nickel in particular is sensitive to pH as it decreases below 7. Observed changes in pH and resulting enhanced metal leaching are linked to testwork conditions but provide useful information about metal leaching mechanisms. Simply stated, the overall conclusion is that mining of the NorthMet deposit would only create an environmental problem if the wastes become acidic, and that is unlikely to occur and if it does, it will self-correct over time. Although there are germs of truth in each of the four findings, there is much to refute them as well. Percent sulfur content and pH recovery Polymet states that Duluth Complex samples containing < 0.4%S have not developed acid over 18 years of testing. Stated another way, acid drainage could be a concern for all wastes that contain > 0.4% total sulfur. Using this definition, and given the %S cutoffs (0.12 for Category 1, 0.6 for Category 2/3, and no high-end cutoff for Category 4), all wastes except Category 1 are of concern in terms of acid drainage. Focusing only on acid drainage is insufficient lower-sulfur wastes have been shown to leach concentrations of other contaminants that are a concern for aquatic life and human health. As shown in Figure 4a, two samples with %S values < 0.4 have produced leachate with pH values below 6. The solubility of a number of metal compounds increases at pH values below 6. As noted in SRK Consulting (2007), no nickel solubility limits were exceeded (no nickel minerals precipitated) for hundreds of leachate samples when the pH was below 6. In addition, Polymet Mining (2013a) uses a pH of 6 as the low-end cutoff for non-acidic Category 2/3, 4, and ore in calculating the concentration caps (see Large Table 4). And the GARD Guide, an industry-based online guide to acid rock drainage issues, also considers pH 6 as the cutoff for designation of acidic drainage conditions (GARD Guide, 2014, Figure 1-1). While the cross-over from neutral to acidic pH values can be debated, it is clear that pH values are dropping in the two humidity cell test samples with %S values of 0.33 and 0.31 (HC11, which has hit values below pH 5, and HC18(D)). The pH of sample HC10 (0.21 %S) also decreased over time and dipped once into the <6 range. Therefore, Duluth Complex rocks with %S values < 0.4 have produced acid. The statement about the lack of importance of any other parameter (including geology, mineralogy, stratigraphic location) to environmental behavior of the wastes was addressed in Section 3.1 above.

Ann S. Maest, PhD

Geochemist 16

The statement about recovery of pH over time is not supported by the data. At least over the course of the nearly eight-year experiment conducted by MDNR, pH values are not recovering (see Figure 4a). And, as noted in MDNR (2013), these experiments are still ongoing. Figure 4b shows similar results for field monitoring of a mine waste dump in Canada. Results for Dunka pit blast hole samples (SRK Consulting, 2007; App. B.1) with higher %S values do show a drop and then a rise in pH, but pHs never recover to values above 6. Even though it would seem obvious that a complete depletion of sulfide minerals would cause the pH to recover, as stated by Polymet above, secondary minerals can still exist in the waste deposits, and some of them (especially hydrated iron sulfate salts) can store acid and metals that can be readily released during rain or snowmelt events; the secondary minerals can form again as interstitial waters in the waste pile evaporate under drier conditions (Nordstrom, 1982; Hammarstrom et al., 2005). Figure 4. (a) Results from 400-week long humidity cell testing of Duluth Complex rocks: pH. Results below the dashed line represent acidic conditions. Samples are from the Babbitt Prospect unless marked with a (D), which denotes samples from the Dunka Road Prospect. (b) Onset of acid drainage for a mine in Canada after eight years.

(a)

Source: MNDR, 2013, Figure 7.

Ann S. Maest, PhD

Geochemist 17

(b)

Source: Price, 2009. Relationship between metal concentrations and pH SRK Consulting (2007) states that metal leaching is strongly linked to pH, especially nickel at pH values below 7. While higher concentrations of nickel and a number of other metals are certainly more likely below pH 7, elevated concentrations of nickel and other metals in Duluth Complex leachate have been measured at neutral and more basic pH values. As shown in Figure 5, many test pile leachate samples (black xs) had nickel concentrations above 1 mg/L, reaching well into the tens of mg/L. Furthermore, there is so much scatter in the data that building an equation to express the relationship between pH and metal concentrations would produce unreliable results. Such an approach could be used for results from an individual leach test, for example, but would be too simplistic to come close to reflecting actual field conditions for a large pile or series of piles. Importantly, it would be easy to miss the higher concentration values using a single pH-metal relationship. In the field, the behavior of secondary minerals and hydrologic events (storms, snowmelt) can have a strong effect on leachate chemistry. SRK Consulting (2007, p. 12) discusses the seasonal variability in leachate pH and metal content for the six AMAX test piles from the Babbitt Prospect.

Ann S. Maest, PhD

Geochemist 18

Figure 5. Nickel concentrations in Babbitt Prospect test pile leachate and leach tests from Babbitt, Dunka, NorthMet, and Virginia Formation samples. The black xs are from the AMAX test piles, which contain Babbitt Prospect materials.

Key:

Source: SRK Consulting, 2007, Appendix H.2, Chart H.2.19.

Ann S. Maest, PhD

Geochemist 19

SRK Consulting (2007) acknowledges the lack of information regarding the relationship between pH and metals in waste rock leachate for the NorthMet Project (Section 7.4.1): Other components tended to occur at trace concentrations and have only been analyzed in leachates from PolyMets laboratory tests. As a result, the distribution with respect to pH is not well known. The lack of analyses of acidic leachates from Duluth Complex rocks limits the understanding of low pH solubility of elements such as antimony, arsenic, barium, beryllium, boron, chromium, mercury, molybdenum, selenium, silver and thallium. The data show however that many of these elements are more soluble under alkaline conditions than acidic conditions (Figure 7-1). Because several components occur as oxyanions this finding is consistent with expected chemical behavior. Problems with reliably predicting the relationship between pH and contaminants of concern affect modeling inputs and results, as discussed in the following section.

4. Modeling Inputs, Results, and Uncertainties

Polymet uses a confusing mix of sources for inputs to the GoldSim model, including: results from laboratory tests (HCTs, aqua regia digestions, short-term leach tests (SMWMPs), MDNR reactors, and microprobe analysis); leachate data from field test piles; concentrations from other mines; and thermodynamic (mineral solubility) information. The basic steps in their waste rock modeling process are (Polymet Mining, 2013a, p. 60): 1) Generate random non-acidic laboratory release rates 2) Generate random scale-up correction factors for water contact, particle size, temperature, and acidity 3) Calculate field release rates by multiplying non-acidic laboratory release rates by the particle size correction and temperature and acidity corrections (if applicable). 4) Partition the released mass between contact (flushed by runoff) and non-contact mass by multiplying by the contact correction. a. Contact mass is flushed by runoff and used to estimate drainage water quality. b. Non-contact mass remains with the waste rock and is flushed into water if the rock is submerged (i.e., in East Pit backfilling). This outline leaves out the step used by Polymet of applying concentration caps that limit the high end of the concentration data, which further reduces output concentrations in the model. Polymet Mining (2013a) states several times that MDNRs modeling assumptions are overly conservative, including MDNRs assumed distribution of bulk scale-up factors (p. 75), their estimates of concentration caps for Category 1 waste rock (p. 76), and their estimate of Category 1 stockpile pH (p. 78). I point out some issues in this section related to modeling that were not considered by either Polymet or MDNR.

4.1 Model Inputs and Concentration Caps

4.1.2 Use of Probabilistic or Monte Carlo Simulations in the Water Quality Model Monte Carlo simulations are all about random selection of inputs kind of like tossing a coin many, many times. The intent with this approach is to represent the range of possible values

Ann S. Maest, PhD

Geochemist 20

that could occur at the mine site and to characterize the uncertainty. But this approach does not honor real-world processes such as seasonal variability and correlations among constituents that could strongly affect water quality. US EPA (2013) noted in their comments on the GoldSim model that calcium and magnesium concentrations are not correlated in the GoldSim output, yet they are correlated in the natural stream chemistry, and this lack of correlation could incorrectly simulate the release of other constituents. In addition, if hardness is not simulated correctly, calculations of exceedences of hardness-based water quality criteria (e.g., copper, lead, zinc) would be off base. Monte Carlo simulations would also not capture the seasonal trends in pH and metal concentrations observed in AMAX field test pile seepage (SRK Consulting, 2007, p. 12) and Dunka stockpiles (Figure 6). When seasonal leachate concentration peaks coincide with low stream flows, pulses of contaminants from mining sources can have a larger impact on water quality and the health of aquatic biota. For example, Lin (1996) demonstrated that large pulses of metals are flushed seasonally from the 50- to 100-year old waste rock pile to groundwater. None of these potential outcomes are evaluated by either Polymet or MDNR. Seasonal flushing of contaminants from the waste rock piles and the pit walls should be evaluated in terms of concentrations generated in leachate and the effects on downgradient groundwater and surface water, using concentrations and flows that reflect observed and predicted seasonal flows and concentrations. Figure 6. Sulfate concentrations in leachate from Dunka stockpile 1811, 1999 to 2005.

Data source: Electronic data copy of Chemistry Leachate Data, CD attached to Polymet Mining, 2013a.

Ann S. Maest, PhD

Geochemist 21

4.1.2 Use of HCT data and (bizarre) calculations of release rates

Concentrations derived from laboratory leach tests, including HCTs, can underestimate field leachate concentrations (see, e.g., Schafer, 2008; Kuipers and Maest, 2006). A comparison of field and laboratory leachate results for this project shows a similar trend. As shown in Figure 5, measured field leachate concentrations for the AMAX/Babbitt test piles (black xs) are substantially higher than results from HCTs using similar material (dark blue filled diamonds). The same conclusions is expressed in the earlier WCDP (SRK Consulting, 2007, p. 77): Generally concentrations in leachates obtained from laboratory experiments were lower than indicated under field conditions. This indicates that the release of these elements is a function of pH but that the high liquid to solid ratio in the tests dilutes the reaction products. The liquid to solid ratio is raised even higher during the first week of HCT sampling, when it appears that Polymet used 750 mL instead of 500 mL of water to produce the first leachate sample (Excel files with HCT leachate data received from MDNR). This higher volume would further dilute any dissolved secondary salts, such as metal sulfate salts, that might have formed as weathering products on the test sample. Even with this extra dilution, many samples show higher sulfate concentrations early in the tests. Yet this information is completely ignored when generating release rates. Large Table 3 in the Polymet Mining (2013a) shows that most release rates for non-acidic conditions in waste rock were derived from HCT data. However, in the main body of this document (p. 54), it states that non-acidic release rates for Category 1 and 2/3 wastes and ore are instead based on %S values in solid samples. If this is the case, and it should be clarified in the FSEIS, it means that sulfate release rates for all Category 1 waste rock are based on an approach that ignores measured solution concentrations of sulfate in laboratory or field leach tests. Additional runs of the water quality model should be conducted for the FSEIS that capture the higher range of laboratory and field sulfate concentrations, especially for Category 1 waste rock simulations. As shown in Figure 7, there is a weak to moderately strong relationship between sulfate release rates and sulfur content for these two waste rock categories under non-acidic conditions. Using a linear regression, the R2 values for Category 1 and 2/3 waste rock are only 0.60 and 0.34, respectively. Polymet then applies a fixed-zero intercept, and the R2 values increase to 0.89 and 0.81, respectively. This approach yields an average release rate for all non-acidic Category 1 and 2/3 wastes and ignores higher sulfate release rates that would produce higher sulfate concentrations. This averaging approach also ignores initial flushes of sulfate, which occur over and over again under field conditions (see SRK Consulting, 2007, p. 12). US EPA (2013) also noticed in their review of the GoldSim model that mean groundwater and surface water concentrations were used instead of upper confidence limits. They noted that the use of means will eliminate those potential worstcase scenario conditionsthat could lead to environmental impacts. The poor correlation (without adjustment) and lack of inclusion of higher values suggest that the approach used to estimate sulfate release rates would underpredict maximum sulfate concentrations for Category 1 and 2/3 wastes and should not be used.

Ann S. Maest, PhD

Geochemist 22

Figure 7. Relationship between sulfate release rate and percent sulfur for non-acidic Category 1 and 2/3 waste rock at the NorthMet Project.

Source: Polymet Mining, 2013a, Figure 8-1. Polymet Mining (2013a, Section 8.1) goes on to discuss a complex set of approaches for calculating release rates for constituents other than sulfate. These approaches are summarized in Large Table 3 and include using the following: for copper - the Cu:S ratio from aqua regia results; for nickel in Category 1 wastes- the Ni:S ratio from Category 4 aqua regia results and the Ni:Mg ratio in olivine from microprobe results; and for zinc - the Zn:Mg ratio in aqua regia results. The multitude of approaches for generating model release rates in different waste categories seems enormously unnecessary, generates confusion, and lacks transparency. Some of the approaches also seem scientifically indefensible. For example, it is not clear how multiplying a value or a ratio in a solid (e.g., the Mg content of olivine) by a concentration in solution (mg of Mg in solution from the HCT tests) allows you to cross out units. Yet, the following equation (Polymet Mining, 2013a, Eqn. 8-8) is produced for the nickel release rate discussed above:

Even more importantly, I could find no convincing evidence was presented regarding why the Ni:S ratio in solid Category 4 waste rock is relevant to the Ni release into solution from Category 1 waste rock, whether Ni occurs primarily in olivine, or why ratios (e.g., Ni/S, Ni/Mg) are the appropriate measure to use to estimate nickel release. The inputs to the water quality model should be recalculated in a way that honors the available field and laboratory leachate information and avoids using ratios of constituents in solids that have not been shown to predict environmental behavior. Factors of safety could be included relatively easily to ensure that potential environmental effects are not being underestimated.

Ann S. Maest, PhD

Geochemist 23

4.1.3 Concentration caps and scale-up factors

Concentration caps The concentration cap concept was born sometime between the creation of the earlier waste characterization document (SRK Consulting, 2007) and the more recent one (Polymet Mining, 2013a). The basic outcome of applying concentration caps and correction factors is to reduce contaminant release rates used as inputs to the model and environmental concentrations produced as outputs by the model. A basic question about the concept of concentration caps is: why do you need to apply solubility-controlled concentration caps if you can put the results into a geochemical program that will calculate solubility limits for you? As noted in SRK Consulting (2007, p. 7): The first step in evaluation of the calculated scale-up concentrations is to examine the major components (sulfate, calcium, magnesium, etc) and some minor components (iron, aluminum). An appropriate approach is to enter the data into a thermodynamic equilibrium model (such as MINTEQA2, PHREEQE). These models can assist with identifying concentrations that are not supportable thermodynamically. According to Polymet Mining (2013a), the reason concentration caps were created was because the HCT leachate data indicated that they were needed: Previous interpretation of the humidity cell test results assumed that the rates of constituent leaching indicated by the humidity cells were free of concentration capping due to the high liquid-to-solid ratios used in laboratory kinetic tests. Data from the humidity cells demonstrate that this is not the case for NorthMet waste rock: nickel leaching, for example, can increase by orders of magnitude even as oxidation and weathering of the sources of nickel decrease. A new approach was needed to account for the influence of concentration caps on the humidity cell leaching rates. The problem is that this makes no sense, assuming that nickel leaching means nickel concentrations. If a concentration cap is in effect, you would see the same concentrations (of metals, for example) even as more and more material was dissolved. This observation would be a signal that no more metal could fit in solution and a solid was precipitating and controlling the dissolved concentrations. The pH-dependent concentration caps for acidic and non-acidic Duluth Category 2/3, 4, and ore samples mimic the 95th percentile and maximum values for the AMAX pile leachate and seem to be reasonably conservative (see Polymet Mining, 2013a, Large Tables 4, 13, and 14). However, the caps proposed for Category 1 waste rock are not conservative enough because they assume that pH values in the leachate will only go as low as pH 7 (see Polymet Mining, 2013a, Large Table 4 I believe this is MDNRs proposal; Polymets is less conservative). In fact, many pH values in Category 1 HCTs were below 7, and the majority of pH measurements for HCTs conducted by MDNR (2013) with %S values below 0.3 were below pH 7 (see Figure 4a above). Therefore, when the %S value is known to be in the range of Category 1 waste rock (<0.12%), leachate pH values are often below 7. In the real world of a NorthMet Mine, there will be less control over sulfur content in the Category 1 Stockpile, and it is more likely that rock with higher %S content will be included in the pile, as discussed in Section 3.2 of these comments. To better reflect the measured and likely pH conditions of Cateogry 1 stockpile leachate, pH-dependent concentration caps should use AMAX pile leachate concentrations for pH values between 6.0 or 6.5 and pH 7.5.

Ann S. Maest, PhD

Geochemist 24

Scale-up Factors Four subfactors were included in the bulk scale-up factor, and these subfactors were used to modify laboratory contaminant release rates to make them applicable to field conditions. The subfactors are water contact (kc), particle size (ks), temperature (kt), and pH (kpH). Laboratory constituent release rates are multiplied by the bulk scale-up factor to arrive at field-relevant rates. With one exception, the scale-up factors decrease laboratory release rates. The overall scale-up factors for the waste categories are: o Category 1 waste: 0.03 o Category 2/3, 4, and ore: 0.10 for non-acidic and 1.0 for acidic o Category 4, Virginia Formation: 0.10. Polymet Mining (2013a, p. 60) states that no acidity factor (KpH) is used for Category 4 Virginia Formation materials because the HCTs were already acidic. However, many of the acidic Category 2/3, 4 and ore HCT leachate samples were also acidic, and the bulk scale-up factor for this material is 1.0. Therefore, it seems likely that the scale-up factor for Category 4 Virginia Formation material is too low and should be re-evaluated. Figure 8-12 in Polymet Mining (2013a) shows the probability distribution of the time to onset of acidic conditions for reactive Duluth Complex Category 2/3 and 4 wastes and ore. The plot shows a 0% likelihood that Category 4 waste and ore would become acidic before about five years. However, results from the AMAX test pile for piles F1 F4, which had %S values of 0.64%, started going acidic after only one year, as shown in Figure 8. Sampling of the test piles started in spring or summer of 1978. The first acidic leachate samples (from FL4 pile) were measured one year later in spring 1979; samples from pile FL3 first became acidic after three years in summer 1981; and samples from piles FL2 and FL1 became acidic after 4.5 and 9.5 years, respectively. Based on these results, it appears that the time to onset for at least Category 4 wastes is overestimated and should start at one year rather than five years. A distribution could be obtained using data from the FL 1 4 AMAX test piles. If the wastes become acidic before they are backfilled, it is less likely that submerging the wastes will stop the acid production or metal leaching. Polymet Mining (2013a) proves that submerging wastes will stop the oxidation of sulfide minerals and the associated release of contaminants using equations (Polymet Mining, 2013a, p. 117). A better approach would be to conduct laboratory column tests that submerge already-acidic NorthMet or Duluth Complex rocks and evaluate the effect of subaqueous conditions on contaminant release. A similar study was conducted on acidgenerating samples from the Pebble Project in Alaska, and the results indicated that the wastes were still releasing acidity and metals, even after being submerged (Pebble Limited Partnership, 2011, Chapter 11; Graphs in Appendix 11I). If this is the case for NorthMet Project materials that produce acid quickly, such as Category 4 and some Category 2/3 wastes, submerging them in the pit may not be an effective management strategy. Submerged column tests should be conducted on Category 2/3 and 4 wastes that have gone acidic, and the results should be used to evaluate the effectiveness of submerging wastes.

Ann S. Maest, PhD

Geochemist 25

Figure 8. Leachate pH values for AMAX test piles FL1, 2, 3, and 4. The dashed line at pH 6 separates acidic and non-acidic results.

Data Source: Electronic data copy of Chemistry Leachate Data, CD attached to Polymet Mining, 2013a.

4.2 Assumptions about Processes along Flowpaths and Modeling Results

Although model-simulated concentrations are relatively high in the East Pit porewater, concentrations along groundwater flowpaths and in surface water are approximately at background values, and according to modeling results, mining a copper ore body will actually reduce copper concentrations in the Partridge River downgradient of the mine. As shown in Figure 9a, copper concentrations in the East Pit porewater after backfilling remain as high as 12.8 mg/L. Transport of pit water along the Category 2/3 surficial flowpath results in a concentration of only 3 g/L (Figure 9b), and by the groundwater arrives at the Partridge River, it is actually diluting pre-project copper concentrations. These surprising results are accomplished in part by applying adsorption factors in the model. Adsorption of metals and other mine-related constituents is strongly dependent upon a number of factors that can change markedly along groundwater flowpaths, including groundwater flow and water quality characteristics, the characteristics of the geologic materials, and the availability of adsorption sites. Given the lack of site-specific information on these factors that affect contaminant adsorption along potential groundwater flowpaths, a modeling run with no adsorption should be conducted and included in the modeling results presented in the FSEIS. If no adsorption is assumed, concentrations at a given location (e.g., Dunka Road) peak many years before scenarios with adsorption (Barr Engineering Company, 2008).

Ann S. Maest, PhD

Geochemist 26

Figure 9. Copper concentrations results from the GoldSim model, shown in purple: (a) in the East Pit porewater after backfilling; (b) in the East Pit Category 2/3 surficial flowpath at the property boundary; (c) and in the Partridge River.

Source: Excel file MineFeat_Perm_Concs_Timeseries_MineSite received from MDNR.

Source: Excel file GW_Cons_Timeseries_MineSite received from MDNR.

Ann S. Maest, PhD

Geochemist 27

Source: Excel file GW_Cons_Timeseries_MineSite received from MDNR. Groundwater flow is modeled as traveling in a homogeneous isotropic medium, even though the bedrock is fractured and the shallower glacial material is present in discontinuous layers. From Polymet Mining (2007X, Appendix A, p. 2): Fractured bedrock, such as is found at the mine site, is inherently difficult to simulate and predict However, by assuming that the aquifer acts as an equivalent porous media at the scale of the problem, it is possible to use conventional porous media modeling codes such as MODFLOW to predict the general direction and magnitude of groundwater flow. Myers (2014) used the observed higher hydrologic conductivity in the upper bedrock aquifer, which can more convincingly be simulated as a porous medium, to show in his model that contaminated water from the pits could make its way to surface water. Movement of mineinfluenced waters along bedrock fractures, for example, has caused increases in metal concentrations in streams affected by mining and complicated water quality improvement efforts (see, e.g., Caruso and Dawson, 2009). Adsorption of contaminants should not diminish concentrations markedly if mine-influenced water travels along a bedrock flow path or a preferential flow path in sedimentary or unconsolidated material. It is reasonable to assume no adsorption along such flow paths to conservatively assess the potential effects of contaminant transport from the mine. A more conservative scenario should be run in the model that does not include adsorption along flowpaths. As discussed in Section 4.1.2 of these comments, a scenario should also be run that includes seasonal flushing of contaminants from waste piles and pit walls to downgradient groundwater and surface water.

Ann S. Maest, PhD

Geochemist 28

Recommendations
o Information Transparency: Polymet should provide a concise summary of the changes in characterization approaches and model input values over time and state clearly which results were finally used to estimate water quality in the model. The original geochemical and water quality data should be presented on the MDNR website in a user-friendly electronic format, with transparent information on the sources of all data. Baseline Water Quality: An independent review of potential mercury loadings and a nondegradation demonstration should be completed as part of the Supplemental Final Environmental Impact Statement (SFEIS). Water Treatment: All treated effluent during all phases of mining should meet the most stringent applicable water quality evaluation criteria for the NorthMet Project. Water with project-relevant nitrate and ammonia concentrations should be tested for treatment. A plan for treatment and discharge of dewatering water during operation should be in place in the SFEIS. Mine Water Balance: The mine water balance should be clearly spelled out with volumes of water generated at the mine site and needed at the plant site under low streamflow/ groundwater elevation and high streamflow/groundwater elevation conditions at different points in the mine life. The estimates should include ranges that consider the need to pump groundwater around the pits and the effects of changing climate. Mine Plan: o Category 1 wastes should be lined or backfilled into the West Pit upon closure. o Submerged column tests should be conducted on Category 2/3 and 4 wastes that have gone acidic, and the results should be used to evaluate the effectiveness of submerging wastes in the pit. Model Inputs: o The manipulations that occur to leachate chemistry to get it into the model are opaque at best. A clearly written explanation should be provided that takes the reader through an example for one or two contaminants of concern (e.g., copper or nickel), for all waste rock/ore categories, under non-acidic and acidic conditions, that includes: (1) Initial concentrations, based on the release rates or method(s) used (some will be a function of pH); (2) concentrations after applying the four scale-up factors (water contact, particle size, temperature, and acidity); and (3) actual input values used in the model representing waste rock leachate concentrations. o Release rates should be based on field or laboratory HCT measurements rather than on complicated combinations of solid and solution values. Factors of safety could be included relatively easily to ensure that potential environmental effects are not being underestimated. o The pH-dependent concentration caps for Category 1 wastes should use AMAX pile leachate concentrations for pH values between 6.0 or 6.5 and pH 7.5. o Scale-up factors for Category 4 wastes should be re-evaluated. o Time to onset of acidic conditions for at least Category 4 wastes should be reevaluated. Model Assumptions: The model should be re-run with more protective assumptions, including an assumption of no adsorption along flowpaths and seasonal flushing of contaminants from waste rock and pit walls. A scenario should also be run that considers transport along a preferential flowpath to surface water. Model results should also include the pH in the pit, along flowpaths, and in the Partridge River.

Ann S. Maest, PhD

Geochemist 29

Conclusion
The release of contaminants from the proposed project is likely substantially underestimated, and the potential to affect downgradient and downstream aquatic resources is similarly underestimated. Major changes in the mine plan are needed to account for large unknowns in the water balance, especially the lack of plans for management and treatment of pumped dewatering water. The management of Category 1 wastes should be re-evaluated, and lining the stockpile or backfilling the wastes into the West Pit should be considered. The transparency of the modeling and waste characterization information needs to be increased substantially for the SFEIS.

References
Balcerzak, M. 2002. Sample digestion methods for the determination of traces of precious metals by spectrometric techniques. Anal.Sci. 18, 737-750. Available: http://www.quimlab.com.br/PDFLA/M%E9todo%20de%20digest%E3o%20de%20amostras%20para%20an%E1lises%20de%20metais %20preciosos.pdf Barr Engineering Company, 2007a. Addendum: Technical Design Evaluation Report RS29T Wastewater Treatment Technology. October. 369pp (included on SDEIS reference disk as Barr 2007e). Barr Engineering Company, 2007b. Mine Site Water Balance for the PolyMet NorthMet Mine Site. RS21, Draft-02. October. Produced for Polymet Mining. Barr Engineering Company, 2008. Mine Site Groundwater Impacts Predictions Evaluation at Dunka Road, Technical Memorandum. To: Stuart Arkley, DNR, From: Jer Mohr and Tina Pint, Barr Engineering Company. Project: 23/69-862. cc: Jim Scott, PolyMet Mining, Inc. October 24. Barr Engineering Company, 2013a. Final Pilot Testing Report, Plant Site Wastewater Treatment Plant Pilot Testing Program. Prepared for Poly Met Mining Inc. January (included in SDEIS Reference Disc as Barr 2013g). Barr Engineering Company, 2013b. Reverse Osmosis Pilot Test Report, SD026 Active Treatment Evaluation. Prepared for Cliffs Erie LLC and PolyMet Mining Inc. June. Caruso, B.S. and Dawson, H.E., 2009. Impacts of groundwater metal loads from bedrock fractures on water quality of a mountain stream. Environ. Monit. Assess. 153, 405-425. Available: http://link.springer.com/article/10.1007%2Fs10661-008-0367-6#page-1 GARD Guide, 2014. International Network for Acid Preventions GARD Guide, Chapter 4. Section 4.3.2.2. Number of Samples. Avaiable: http://www.gardguide.com/index.php/Chapter_4 and Chapter 1. The Global Acid Rock Drainage Guide. Available: http://www.gardguide.com/index.php/Chapter_1 Hammarstrom, J.M., Seal II, R.R., Meier, A.L., and Kornfeld, J.M. 2005. Secondary sulfate minerals associated with acid drainage in the eastern US: recycling of metals and acidity in surficial environments. Chemical Geology 215, 407 431. In: Geochemistry of Sulfate Minerals: A Tribute to Robert O. Rye. Paper 2. Available: http://digitalcommons.unl.edu/usgsrye/2. Lapakko, K. and Antonson, D., 2006. Laboratory Dissolution of Duluth Complex Rock from the Babbitt and Dunka Road Prospects, Status Report. [Minnesota Department of Natural Resources, Division of Lands and Minerals]. October. (referenced in Polymet Mining, 2013a)

Ann S. Maest, PhD

Geochemist 30

Lin, Xhixun, 1996. Leachate chemistry and precipitates mineralogy of Rudolfsgruvan Mine waste rock dump in central Sweden. Wat. Sci.Tech. 33 #6, 163-171. Abstract available: http://www.sciencedirect.com/science/article/pii/0273122396003046 Minnesota Department of Natural Resources (MDNR), United States Army Corps of Engineers, and United States Forest Service, 2013. NorthMet Mining Project and Land Exchange. Supplemental Draft Environmental Impact Assessment (SDEIS). November. Minnesota Department of Natural Resources (MDNR), 2013. Dissolution of Duluth Complex Rock from the Babbitt and Dunka Road Prospects: Eight-year laboratory experiment. June. Myers, T. 2014. Myers Groundwater Flow Transfer Model PolyMet FINAL. Prepared for the Minnesota Center for Environmental Advocacy. January. Nordstrom, D.K. 1982. Aqueous pyrite oxidation and the consequent formation of secondary iron minerals. In: Acid sulfate weathering, L.R. Hossner (ed.) Soil Sci. Soc. Amer. Publ., 10, 37-56. (Attached) PLP. 2011. Pebble Project Environmental Baseline Document 2004 through 2008 (with updates in 2010). Chapter 11. Geochemical Characterization, Bristol Bay Drainages. Prepared by SRK Consulting, Inc. for Pebble Limited Partnership. Available: http://www.pebbleresearch.com/ebd/bristol-bay-phys-chem-env/chapter-11/ Polymet Mining, Date Unknown. GC03 Waste Rock Water Chemistry Validation Sampling Program_2008.doc. Prepared in support of Appendix A, Waste Rock and Lean Ore Geochemical Characterization Plan, NorthMet Project, Minnesota, for RS53/RS42 Waste Rock Water Quality Waste Rock and Lean Ore NorthMet Project Draft 01, March 9, 2007. Polymet Mining, 2012. NorthMet Project Rock and Overburden Management Plan (Included on SDEIS Reference Disk as PolyMet 2012s). Polymet Mining, 2013a. NorthMet Project Waste Characterization Data Package, Version 10 Issue Date: March 7, 2013 (Included on SDEIS Reference Disk as PolyMet 2013l). Polymet Mining, 2013b. NorthMet Project, Water Management Plan Plant. Version 2. February 1. NorthMet Project, Water Management Plan - Mine. Version 2. Issue Date: January 9 (Included on SDEIS Reference Disk as PolyMet 2013e and PolyMet 2013f). Polymet Mining, 2013c. NorthMet Project, Project Description, Version 5. Issue Date: March 7, 2013 (Included on SDEIS Reference Disk as PolyMet 2013c). Price, W.A., 2009. Prediction Manual for Drainage Chemistry from Sulphidic geologic Materials MEND Report 1.20.1. Report prepared by: William A. Price, CANMET Mining and Mineral Sciences Laboratories, Smithers, British Columbia. V0J 2N0. Available: http://www.abandonedmines.org/pdfs/MENDPredictionManual-Jan05.pdf Schafer, 2008. Critical review of geochemical predictions at mines. Northwest Mining Association Meeting presentation, Reno, Nevada. December 3. (Attached)

Ann S. Maest, PhD

Geochemist 31

SRK Consulting, 2007. RS53/RS42 Waste Rock Characteristics/Waste Water Quality Modeling Waste Rock and Lean Ore NorthMet Project DRAFT. February/Draft 01, March 9, 2007 (Included on SDEIS Reference Disk as SRK 2007b). US EPA, 2002. National Recommended Water Quality Criteria: 2002. Office of Water, Office of Science and Technology. EPA-822-R-02-047. November. Available: http://water.epa.gov/scitech/swguidance/standards/upload/2008_04_29_criteria_wqctable_nrwqc -2002.pdf US EPA, 2009. National Recommended Water Quality Criteria for Priority Pollutants. Office of Water, Office of Science and Technology. Available: http://www.epa.gov/ost/criteria/wqctable/ US EPA, 2013. EPA Review of GoldSim 4-6-2013 (2).pdf. Washington Department of Ecology. National Pollutant Discharge Elimination System, Waste Discharge Permit No. WA0052434. Buckhorn Mine. Central Regional Office. March 2014. (Attached)

Ann S. Maest, PhD

Geochemist 32