The Diavik Waste Rock Project: Particle size distribution and sulfur characteristics of low-sulfide waste rock

Applied Geochemistry 36 (2013) 200–209

Contents lists available at SciVerse ScienceDirect

Applied Geochemistry journal homepage: www.elsevier.com/locate/apgeochem

The Diavik Waste Rock Project: Particle size distribution and sulfur characteristics of low-sulfide waste rock Lianna J.D Smith a,b, David W. Blowes a,⇑, John L. Jambor c,1, Leslie Smith d, David C. Sego e, Matthew Neuner d a

Department of Earth and Environmental Science, University of Waterloo, Waterloo, ON, Canada N2L 3G1 Rio Tinto (Diavik Diamond Mines Inc.), Yellowknife, NT, Canada X1A 2P8 Leslie Research and Consulting, Tsawwassen, BC, Canada V4M 3L9 d Department of Earth, Ocean and Atmospheric Sciences, University of British Columbia, Vancouver, BC, Canada V6T 1Z4 e Department of Civil and Environmental Engineering, University of Alberta, Edmonton, AB, Canada T6G 2W2 b c

a r t i c l e

i n f o

Article history: Available online 23 May 2013 Editorial handling by R. Fuge

a b s t r a c t Three large-scale instrumented waste rock piles were constructed at the Diavik Diamond Mine in the Northwest Territories, Canada. These experimental waste rock piles (test piles) are 15 m high and are part of an integrated field and laboratory research program to characterize and compare low-sulfide waste rock and drainage at various scales. During test pile construction, samples of the <50 mm fraction of waste rock were collected from two types of waste rock that are segregated during mining operations based on S content. The samples were analyzed for S content and particle size distribution. One test pile contained waste rock with an average of 0.035 wt.% S in the <50 mm fraction, within the operational S target of <0.04 wt.% S for the lower S waste rock type. The second test pile contained waste rock with an average of 0.053 wt.% S in the <50 mm fraction, lower than the operational S target of >0.08 wt.% S for the higher S waste rock type. The third test pile has a low permeability till layer and a low sulfide waste rock thermal layer covering a core of waste rock with average 0.082 wt.% S in the <50 mm fraction, which is within the operational S target of >0.08 wt.% S for the higher S waste rock. Particle size distributions for the lower and higher S waste rock are similar, but the higher S waste rock has a higher proportion of fine-grained particles. Sulfur determinations for discrete particle sizes of the <50 mm fraction illustrate higher S concentrations in smaller particles for both the lower S waste rock and the higher S waste rock. Similarly, S concentrations calculated for the >10 m scale, from composite blast hole cuttings, are lower than those calculated for the <50 mm scale. Acid–base accounting using standard methods and site-specific mineralogical information was used to calculate the ratio of neutralization potential to acid generating potential. A comparison of calculation approaches to pH and alkalinity data from humidity cell and test pile effluent suggest that ratios are very sensitive to the calculation method. The preferred calculation method was selected by comparing calculation results to pH and alkalinity data from humidity cell effluent collected over 95 weeks and test pile effluent collected over five field seasons. The preferred acid–base accounting values were obtained by calculating the average neutralization potential divided by the average acid potential of a sample set. This approach indicates that waste rock with >0.05 wt.% S is of uncertain acid-generating potential and effluent data indicate this waste rock generates acidic effluent; whereas lower S waste rock does not produce acidic effluent, consistent with the acid–base accounting predictions. Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction Large, heterogeneous, unsaturated stockpiles of mining waste can expose residual sulfide minerals to the atmosphere and result in acid mine drainage (AMD). AMD is characterized by drainage with elevated levels of acidity, SO2 and dissolved 4 ⇑ Corresponding author. Tel.: +1 519 888 4878; fax: +1 519 746 3882. 1

E-mail address: [email protected] (D.W. Blowes). Deceased.

0883-2927/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.apgeochem.2013.05.006

metals that are released by sulfide-mineral oxidation. AMD is controlled by coupled physical and biogeochemical processes that control mineral-weathering rates and hydrologic flow in mine-waste stockpiles (e.g. Strömberg and Banwart, 1999; Lefebvre et al., 2001b; Smith and Beckie, 2003). Waste rock piles are physically and geochemically heterogeneous with internal structures created during dump construction, particle sizes ranging from silt to boulders, and variable saturation, geochemical composition, and reactivity of the particles.

L.J.D Smith et al. / Applied Geochemistry 36 (2013) 200–209

Fundamental AMD-generation processes influenced by particle size include mineral weathering and fluid flow (water and air). Sulfide-mineral oxidation depends on the supply of water and O2 to the mineral site and the available reactive sulfide-mineral surface area in the pile. The reactive surface area is a function of the total available sulfide-mineral mass, the particle size distribution and the liberation or exposure of the sulfide mineral. Sulfide liberation is a function of the mode of sulfide mineral occurrence and the particle size distribution (Lapakko and Antonson, 2006; Lapakko et al., 2006). Particle size also affects the pile permeability, which is one control on the rate at which O2 and water move through the pile. Furthermore, particle size affects the development of reaction rims, which in turn affects the rate at which O2, and other reactants and products, diffuse through the particle to reach a reaction site. This transport-limited process can be described conceptually and mathematically by the shrinking core model (Levenspiel, 1972; Cathles, 1979; Davis and Ritchie, 1986, 1987; Davis et al., 1986; Wunderly et al., 1996; Lefebvre et al., 2001b; Mayer et al., 2002, 2003). In a series of batch reactors of discrete particle size fractions, Strömberg and Banwart (1999) determined that sulfide-mineral weathering rates in particles >0.25 mm differ from those in particles <0.25 mm. Sulfide-mineral availability to oxidation is a major control on environmental mass loading of acidity, SO2 4 and metals as well as AMD persistence. Reaction products and dissolved constituents are transported by infiltrating water. Matrix flow and flow channelization in unsaturated waste rock piles can result in spatial and temporal variability in drainage volumes and mass loadings, and may account for differences in weathering rates observed in field and laboratory studies (Velbel, 1993; Nichol et al., 2005; Stockwell et al., 2006). Textural and compositional waste rock heterogeneity, pile structure and pile geometry influence each physicochemical process and the coupling between these processes in unsaturated waste rock piles (Lamontagne et al., 1999; Lefebvre et al., 2001a; Nichol et al., 2005). This study examined particle size and S distributions, and acid– base accounting for field-scale experimental waste rock piles (test piles). This characterization study was part of the Diavik Waste Rock Project, which is an integrated field and laboratory study of the physicochemical evolution of unsaturated waste rock piles (Smith et al., 2013).

2. Site description Three 15 m test piles were constructed at the Diavik Diamond Mine (Northwest Territories, Canada; Fig. 1), where the average annual ambient temperature is 8.5 °C and temperatures fluctuate from an average maximum of 18 °C in July to an average minimum of 31 °C in January/February. The average annual precipitation is 280 mm (1999–2006) with about 60% occurring as snow (Environment Canada, 2008). The Diavik mine is an operating open pit and underground diamond mine. The kimberlite ore bodies are hosted in Archean granite and pegmatitic granite country rock that is massive and moderately to coarsely crystalline. The granite contains metasedimentary biotite schist xenoliths that comprise 4–6% of exposed pit walls and are variable in size and distribution (Fig. 2). The region is cut by a series of Proterozoic diabase dikes. Static tests conducted during the baseline geochemistry study indicate that the granite and pegmatitic granite country rock contain only trace sulfides (typically <0.02 wt.% S); they are considered non-acid generating with very low potential to leach metals during weathering. The biotite schist contains locally disseminated pyrrhotite [Fe1xS] and other minor sulfide minerals, and little carbonate. The sulfide-S content of the biotite schist varies from 0.01 to 0.56 wt.% S

201

with an average of 0.16 wt.% S. The biotite schist is considered potentially acid generating because of the low neutralization potential (NP). The silicate mineral assemblage was found to contribute little to the NP. The diabase dikes were also found to contain trace sulfides, but this rock type is considered geochemically insignificant on an operational scale because of its very low abundance. As part of the operational waste rock management program at the Diavik mine, waste rock is segregated into three types: Type I (target of <0.04 wt.% S), Type II (target of 0.04–0.08 wt.% S), and Type III (target of >0.08 wt.% S). Type I waste rock is comprised primarily of granite; Type II waste rock is comprised predominantly of granite with minimal biotite schist; and Type III waste rock is comprised of granite with a greater amount of biotite schist. Due to the biotite schist content, Type III waste rock is considered potentially acid generating. Each waste rock type is hauled to a designated stockpile or construction area. At the end of the mine life up to 120 Mt of waste rock, up to 75% of which could be Type III, will be stockpiled in a 60–80 m high permanent pile covering an area of up to 3.5 km2. Three test piles were constructed of run of mine waste rock. One test pile consisted of Type I waste rock, the second consisted of Type III waste rock, and the third consisted of Type III waste rock that was re-contoured and capped by a till layer and a layer of Type I waste rock (Covered test pile), based on the accepted closure design for the full-scale Type III waste rock pile (Smith et al., 2013). Instruments within the test piles were installed at the bases of each pile and on four angle of repose tip faces spaced 5 m apart. The instrumentation tip faces are referred to as Face 1, 2, 3 and 4. The outer slope face, which is not instrumented, is referred to as Face 5 (Smith et al., 2013). Laboratory kinetic experiments using waste rock collected in 2004 and 2005 were conducted in parallel with the field study. Kinetic testing was initiated in 2005. Leachate pH, electrical conductivity, and alkalinity are illustrated for the first 95 weeks of the kinetic tests. Effluent was also analyzed for Eh, anions, cations and nutrients (not reported here). 3. Characterization methods 3.1. Particle size sampling and analyses Samples for particle size determination ranged in mass from 5 to 10 kg. Particle size samples collected directly from the instrumentation face were excluded from the data set because they were not considered representative: instrumentation faces were characterized by large cobbles and boulders with few locations available for unbiased sampling of the <50 mm fraction (Fig. 4). Most samples collected from the face were composite samples, consisting of finer material scraped from larger boulders from several locations within about 2 m of the target sample location. Randomly selected particle size samples were analyzed according to ASTM method D422-63 (ASTM, 2002). Samples were split to 2 to 5 kg using a riffle splitter or the coning and quartering method. Samples were sieved using either a standard sieve set (1200 , 600 , 300 , 1½00 , 3=4 00 , ½00 , 3/800 , #4, #10, #20, #40, #60, #100, #200, and a pan for smaller size fractions) or a metric sieve set (40, 28, 20, 14, 10, 5, 2.5, 1.25, 0.625, 0.315, 0.160, 0.08 mm and a pan for smaller size fractions). To compare results between sieve sets the standard sieve set data were interpolated to correspond to the size fractions of the metric sieve sizes. 3.2. Sulfur and carbon sampling and analyses During open pit production blast holes were drilled 12 m deep with typical 7 m spacing between holes. The waste rock was clas-

202

L.J.D Smith et al. / Applied Geochemistry 36 (2013) 200–209

Fig. 1. Location of Diavik Diamond Mine.

Fig. 3. Typical haul load from which the <50 mm size fraction was sampled for particle size and S concentration. White square scale in photograph is 0.5  0.5 m.

Fig. 2. Typical distribution of bitotite schist xenoliths (dark lenses) in the open pit wall. Bench heights are 30 m.

sified by Diavik as part of the operational waste segregation procedure by determining the S content of samples of vertically integrated cuttings from each hole drilled for blasting. Drill cuttings were deposited by the drill in cone-shaped piles at each drill hole location. An aluminum trowel was used to bisect the cuttings pile and expose a vertical cross section of the tallest part of the cuttings cone. The trowel was used to sample the complete cross section, for a total sample volume of approximately 3 L. One duplicate sample was collected for every 25 samples. Sulfur content was determined for each drill hole in each blast pattern, and the S distributions were used to delineate mineable waste rock volumes (typically >10  10 m and 10 m deep) by waste rock type within each blast pattern. For times when the S analyzer was not function-

ing, the blast hole cuttings were classified visually based on lithology. For visual classification the lithological variations in the vertical cross-section of the blast hole cuttings cone were measured. Cuttings piles with <15% biotite schist were designated as Type I and those with >15% biotite schist were designated as Type III. Visual classification was typically more conservative than S assays of the cuttings (unpublished data). Sulfur results averaged for a mineable unit are referred to as blast pattern averages and >10 m scale. Samples of the <50 mm fraction were collected from most haul-truck loads (either 100 or 240 tonnes) delivered to the test piles during construction (Fig. 3). Samples were taken to measure the S content, C content and the particle size distribution of the matrix material. Haul trucks dumped the waste rock on the top of the test piles and samples were collected with an aluminum trowel before a track dozer pushed the waste rock over

L.J.D Smith et al. / Applied Geochemistry 36 (2013) 200–209

203

Fig. 4. Typical instrumentation face: (a) entire face; and (b) instrumentation face sample area. Square scale in (a) is 0.5  0.5 m and highlighted yellow with arrow for clarity; white square scale in (b) is 0.5  0.5 m. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

the side of the test pile. Samples were also collected directly from the instrumentation face as instruments were being installed. Samples of both the drill hole cuttings and test pile samples collected for S analysis weighed 2–5 kg. Samples were oven dried at 105 °C for 2–4 h. Cooled samples were crushed and homogenized. Subsamples of 0.26 ± 0.01 g of the pulverized material were analyzed using a LECO IR-432 Sulfur Determinator. This instrument was calibrated to seven reference standards, and the three closest to the average S content of the sample were used for calibration. A minimum of seven S blanks were analyzed per day. Sulfur standards were analyzed every tenth sample and sample duplicates were analyzed randomly. Particle size fractions (40, 28, 14, 10, 5, 2.5, 1.25, 0.625, 0.315, and 0.160 mm, and pan) of selected samples were retained and analyzed for S and C content. Each fraction was pulverized to <100 lm using a four-position Fritsch Pulverisette Analysette planetary ball mill and analyzed for S and C content using an Eltra CS2000 Carbon/Sulfur Determinator. Each sample was analyzed in duplicate, at minimum. Sulfur and C standards were analyzed every 10–15 samples with two to four standards analyzed each time; the two most similar results were used for calibration. Sulfur standards included 0.0021, 0.01, 0.011, 0.014, 0.034, 0.104, and 0.269 wt.% S. Carbon standards included 0.0054, 0.015, 0.019, 0.031, 0.468, 0.996, 1.02 wt.% C.

3.3. Kinetic tests Kinetic testing followed the standard ASTM5744-07, with the exception that sieved particles <25.4 mm (rather than <6.3 mm) were used for the cell charges. Each week included a 3-day dry period (0% relative humidity), a 3-day wet period (>95% relative humidity) and a 1-day flood leach with 500 mL of distilled water. Effluent from humidity cell experiments was sampled weekly and analyzed for a range of parameters. The pH was measured on unfiltered aliquots using an Orion Ross combination electrode (model 815600) calibrated with standard buffer solutions of pH 10, 7 and 4. Alkalinity was measured for aliquots filtered through 0.45 lm cellulose-acetate syringe filters. Alkalinity was measured using a Hach digital titrator and bromocresol green/methyl red indicator. Splits of the waste rock used in the humidity cells were analyzed for C and S content. The samples were pulverized and analyzed using the same method as described for the particle size sample C and S determinations. Type I and Type III waste rock used in the humidity cell experiments were collected in 2004 and 2005 and screened to

<25.4 mm. Three split samples from each waste rock type collected in each year were analyzed for S and C at an external laboratory. 3.4. Acid–base accounting calculations Ratios of NP to acid-generating potential (AP) are commonly used to infer the AMD development potential of waste rock. One common classification scheme recommends that NP:AP ratios less than 1:1 (i.e., NP:AP < 1) indicate potentially acid generating waste rock, NP:AP ratios between 3:1 and 1:1 (1 < NP:AP < 3) indicate uncertain acid generating potential, and NP:AP ratios greater than 3:1 (NP:AP > 3) indicate non-acid generating waste rock (INAC, 1993). The AP of each size fraction was calculated assuming the analytically-determined S concentration consisted entirely of sulfide minerals. This assumption is consistent with the feasibility-stage mineralogical studies of waste rock from the study site that showed negligible SO4 content. The AP calculations assumed the S will oxidize by the overall reaction:

ð9  3xÞ ð5  3xÞ O2ðaqÞ þ H2 OðaqÞ 4 2 2 þ ! ð1  xÞFeðOHÞ3ðsÞ þ SO4 þ 2H

Feð1xÞ SðsÞ þ

ð1Þ

Pyrrhotite with the composition Fe0.85S [Fe17S20] was determined to be the primary sulfide mineral in the country rock and would oxidize with O2(g) as the primary oxidant by the overall reaction: þ Fe0:85 SðsÞ þ 2:1375O2 þ 2:275H2 O ! 0:85FeðOHÞ3ðsÞ þ SO2 4 þ 2H

The NP for the particle size fractions was calculated assuming that the analytically-determined C content consisted entirely of CaCO3. This assumption is considered reasonable based on the feasibility-stage mineralogical information that showed negligible siderite [FeCO3] and other carbonate content in the waste rock, NP values for minerals provided in the literature (Jambor et al., 2007), and feasibility studies on the waste rock that suggest the silicate mineral assemblage contributes little to the NP. The calculated NP value for a given size fraction of a given sample was obtained using the arithmetic mean of the wt.% C from the replicates. Similarly, the calculated AP value for a size fraction was obtained using the arithmetic mean of the wt.% S from the replicates of that size fraction. Differences between replicates for both C and S analyses were typically <0.005 wt.%. The weighted average of sample C, S, and the sample NP and AP were calculated by multiplying the result for the replicates of a given size fraction by the proportion of the sample that size fraction accounted for, and then summing those products to give a weighted average for the sam-

204

L.J.D Smith et al. / Applied Geochemistry 36 (2013) 200–209

ple. The arithmetic mean of all the samples (based on the weighted averages of each sample) was then calculated to give an overall waste rock average for C, S, NP and AP. The overall average NP:AP of each waste rock type was calculated in three ways and compared. The first approach calculated the average NP:AP for each replicate of a given size fraction, then calculated the weighted average for each sample (w(NP:AP)) and then arithmetic mean of each of the two sample sets (l(NP:AP)). The second approach calculated the average NP and the average AP for each replicate of a given size fraction and then calculated the weighted average NP and AP based on the grain size distribution of the sample (wNP and wAP, respectively). The ratio of wNP and wAP was taken to give wNP:wAP for each sample to be able to compare w(NP:AP) to wNP:wAP. The arithmetic mean of the sample NP values was divided by the arithmetic mean of the sample AP values to give the overall NP:AP of each sample set (lNP:lAP). The third approach calculated the arithmetic mean of the sample wNP:wAP values (l(wNP:wAP). 4. Results and discussion 4.1. Particle size distribution The particle size distributions of the <50 mm fraction from the Type I test pile, Type III test pile, and Type III core of the Covered test pile were similar (Table 1; Fig. 5). The Type I test pile had the greatest variation in particle size among faces. The Type III test pile was slightly more fine-grained than the other two test piles. The d10 results (the diameter of particle that 10% of the material is finer than) illustrate that Type III waste rock has a greater proportion of fine particles compared to Type I waste rock. This distribution may be lithogically controlled. Biotite schist contained in Type III waste rock is more friable than granite; because Type III waste rock also contains a substantial portion of granite, the larger particle sizes are likely controlled by the granite fraction of the samples. Average test pile permeability to air is 1.4  109 m2, which is high in the range measured at other waste rock piles (Amos et al., 2009). Internal CO2(g) and O2(g) concentrations within the test piles remain at atmospheric levels (<0.01 vol.% CO2 and 20.9 vol.% O2, respectively; Amos et al., 2009), suggesting macroscale O2(g) supply does not limit sulfide-mineral oxidation. A single larger-scale (92 t) particle size measurement was conducted by Diavik for Type I waste rock for size fractions <900 mm (Fig. 6). Dawson and Morgenstern (1995) defined the boundary between waste rock piles that behave as soil-like and rock-like based on the fraction of material finer than 2 mm. Waste piles with more than 20% of waste rock passing the 2 mm sieve (i.e., d20 = 2 mm) are considered to behave in a soil-like manner, and coarser samples behave in a rock-like manner. The smallest sieve used for the 92 t sample was 2.5 mm, through which 14.3% of the sample passed, indicating that a Type I waste rock pile with similar characteristics would behave in a rock-like manner (i.e. less than 20% of the material is finer than 2 mm). Rock-like pile characteristics influence pile physicochemical processes including hydrology and, thus, solute flushing. Rock-like piles may have a larger contriTable 1 Average particle size for the <50 mm fraction and one standard deviation (r) for the Type I, Type III, and Covered test piles. The dn indicates the diameter of particle that n% of the material is finer than. d

Type I test pile (r)

Type III test pile (r)

Covered test pile (r)

d10 d30 d50 d60

0.49 mm (1.30 mm) 3.70 mm (4.73 mm) 11.24 mm (7.46 mm) 16.06 mm (7.99 mm)

0.20 mm (0.10 mm) 1.75 mm (1.76 mm) 6.93 mm (4.94 mm) 10.56 mm (4.60 mm)

0.25 mm (0.23 mm) 3.37 mm (3.63 mm) 11.16 mm (6.92 mm) 16.86 mm (8.95 mm)

bution of flow from macropore or channelized flow than from matrix flow. However, macropore flushing in the test piles seems to be limited to high intensity rainfall events (Neuner et al., 2013). By combining the <900 mm measurement with the average particle size curve for the <50 mm fraction provides an estimate of the overall fraction of smaller size fractions. For example, the <25.4 mm fraction, the same fraction used in the humidity cell experiments, accounts for 14.3% by mass of Type I waste rock <900 mm. 4.2. Sulfur distribution During open-pit operations, mineable units of each waste type were determined based on S concentrations of depth-integrated samples of blast-hole cuttings from each hole in a blast pattern (>10 m scale). Sulfur concentrations from blast patterns were averaged from the best available data from the assay results of blasthole cuttings, blast-pattern configurations, and the number of loads from each blast pattern delivered to each construction face. Waste rock used to build the test piles was classified as part of the standard waste rock management procedures and is considered representative of waste rock moved to the Type I and Type III production-scale areas of the waste rock dump during the same time periods as test pile construction. The average S concentration of the <50 mm fraction of the Type I test pile was 0.035 wt.% (n = 242, r = 0.019), towards the upper end of the Type I waste-rock designation (Fig. 7). Face 3 and Face 5 had the highest average sulfide contents, though the averages remain within the Type I target range. Individual Type I test pile samples ranged from 0.0028 to 0.26 wt.% S. Blast pattern averages for all faces of the Type I pile fell within one standard deviation of the average for the <50 mm samples. Blast pattern information for Face 4 was not available. The average S concentration of the <50 mm fraction of the Type III test pile was 0.053 wt.% (n = 270, r = 0.037), within the Type II waste rock designation. Individual samples from the Type III test pile ranged from 0.0085 to 0.27 wt.% S. Only the Face 1 average of the Type III pile was within the Type III waste rock designation; the average wt.% S of the base and Faces 3, 4, and 5 were within the Type II waste rock designation, and Face 2 was within the Type I waste rock designation. Blast pattern averages for all the Type III test pile faces, except Face 2, were within one standard deviation of the <50 mm sample average (Fig. 7). The average S concentration of the <50 mm fraction of the Covered test pile core (Type III waste rock) was 0.082 wt.% S (n = 183, r = 0.053), at the lower end of the Type III waste-rock designation. The Covered test pile core S concentration had the highest variability of all of the test piles (individual samples range from 0.006 to 0.38 wt.% S). The Covered test pile core was built after the Type III test pile with waste rock from different blast patterns within the open pit. Blast pattern averages for the Covered test pile core fell within one standard deviation of the <50 mm average for all faces except Face 3 (Fig. 7). Sulfur concentrations were more variable for Type III waste rock compared to Type I waste rock, emphasizing the variable nature of the waste rock S content due to the spatially discontinuous sulfide-bearing biotite schist xenoliths. 4.3. Relationship between particle size and sulfur content Blast pattern (>10 m scale) average S concentrations were typically lower than the <50 mm average S concentration but within one standard deviation (Fig. 6). This trend suggests a higher S concentration in smaller size fractions (represented by <50 mm samples) as opposed to the bulk rock (represented by blast hole samples). This relationship is consistent with a study by Lapakko (1994). Sulfur analyses for discrete particle size fractions <50 mm

L.J.D Smith et al. / Applied Geochemistry 36 (2013) 200–209

205

Fig. 5. Particle size distributions of the <50 mm size fraction for the (a) Type I test pile, (b) Type III test pile, (c) Covered test pile Type III core, and (d) till layer of the Covered test pile.

the relationship between particle size and reactivity. This relationship will help assess the potential for AMD and more accurately predict environmental mass loadings and persistence of AMD. 4.4. Acid–base accounting calculations

Fig. 6. Large scale (92 kg, <900 mm) particle size distribution with the averaged cumulative distribution from the <50 mm samples from the Type I test pile, Type III test pile, and Covered test pile Type III core for comparison.

showed greater S concentrations in Type III waste rock than Type I waste rock, and a general trend of increasing S concentration for particle sizes 61.25 mm for waste rock from both Type I and Type III test piles (Covered test pile waste rock particle size fractions were not analyzed). This relationship suggests the smaller size fractions would be more reactive. Individual particles of sulfide minerals were measured to be <200 lm in the feasibility-stage mineralogical study. Particles >1.25 mm have S concentrations that are more variable but lower than particles 60.315 mm (Fig. 8). The 61.25 mm size fractions also showed decreasing C content with increasing particle size for both waste rock types. Type I waste rock has similar C concentrations to Type III waste rock (Fig. 9). Comparison of S concentrations from the blast pattern averages, the <50 mm samples, and the discrete particle size samples indicates that a range of particle sizes must be evaluated to quantify

4.4.1. Calculation approach and comparison Calculated NP values are lower than values measured using the Sobek test (Sobek et al., 1978); however, measured and calculated values have a close relationship for a variety of rock types (Jambor et al., 2007). Furthermore, for samples in which all CO2 is present with Ca and Mg, calculating NP values from analytically determined CO2 avoids many of the perceived deficiencies of the traditional Sobek method, such as over- or under-acidification (Jambor et al., 2007; Lindsay et al., 2009), particle size controls (White et al., 1999), Fe(II) effects (Skousen et al., 1997; Jambor et al., 2003; Weber et al., 2004), and empirical fizz relationships (Weber et al., 2004). The w(NP:AP) calculation approach provided higher NP:AP compared to the wNP:wAP approach for both Type I and Type III sample sets, with few exceptions. The relative percentage differences (RPD) between individual sample w(NP:AP) and wNP:wAP for Type I samples ranged from 49% to 61% (average RPD of 54%, n = 5), with w(NP:AP) always greater than wNP:wAP. The RPD for Type III samples ranged from 9% to 135% (average of 56%, n = 31), with all but four samples having w(NP:AP) greater than wNP:wAP. There are similar discrepancies between NP:AP calculation approaches when calculating the overall NP:AP for the Type I and Type III sample sets. Type I l(NP:AP) was calculated to be 22.7; l(wNP:wAP) was calculated to be 13.0; and lNP:lAP was calculated to be 12.2. All NP:AP ratios suggest the sample would be non-acid generating, however values of 13.0 and 12.2 are more

206

L.J.D Smith et al. / Applied Geochemistry 36 (2013) 200–209

Fig. 7. Sulfur distribution for the (a) Type I test pile, (b) Type III test pile, and (c) Covered test pile Type III core for <50 mm sample S concentrations (closed gray symbols) and blast pattern S means (open symbols). Error bars on the mean S concentration for the <50 mm fraction (closed symbols) indicate one standard deviation. The dotted line indicates the pile average concentration (based on the <50 mm samples). The dashed lines indicate the cut-off values for Type I, Type II, and Type III designations.

(b) 0.35

0.30

0.30

0.25

0.25

0.20

wt. % S

wt. % S

(a) 0.35

0.15

0.20 0.15

0.10

0.10

0.05

0.05

0.00

0.00 0.1

1

10

0.1

Diameter (mm)

1

10

Diameter (mm)

Fig. 8. Sulfur distribution for discrete particle sizes for the (a) Type I test pile material and (b) Type III test pile material. Black symbols indicate average values and dotted line indicates test pile average for the <50 mm fraction.

(b) 0.35

0.30

0.30

0.25

0.25

0.20

wt. % C

wt. % C

(a) 0.35

0.15

0.20 0.15

0.10

0.10

0.05

0.05

0.00 0.1

1

10

Diameter (mm)

0.00

0.1

1

10

Diameter (mm)

Fig. 9. Carbon distribution for discrete particle sizes for (a) Type I test pile material and (b) Type III test pile material.

realistic when considering the low C and carbonate content of the sample waste rock. Type III l(NP:AP) was calculated to be 5.22; l(wNP:wAP) was calculated to be 2.7; and lNP:lAP was calculated to be 2.2. A NP:AP value of 5.5 suggests the waste rock would be non-acid generating whereas NP:AP of both 2.7 and 2.2 suggests the waste rock

would be of uncertain acid-generating potential for all accepted NP:AP classification schemes. The lNP:lAP value of 2.2 is most realistic when considering effluent data from humidity cells and the field experiments. Both Type III humidity cell and Type III test pile effluent have attained pH < 4.5 and alkalinity <0.5 mg L1 (as total CaCO3; Bailey et al., 2013; Smith et al., 2013). The lNP:lAP

207

L.J.D Smith et al. / Applied Geochemistry 36 (2013) 200–209

approach can be considered reasonable because particles in a waste dump would be in contact, thus one particle contributing a low NP:AP could be balanced by another particle contributing higher NP:AP. All NP:AP values reported subsequently are wNP:wAP or lNP:lAP but for clarity are referred to simply as NP:AP.

1:1

3:1

(b) 10 AP (kg tonne-1 CaCO3)

AP (kg tonne-1 CaCO3 )

(a) 10

4.4.2. Test pile waste rock The NP:AP ratios of the individual sample fractions of Type I waste rock ranged from 1.9 to 53.6 with a weighted average of 12.2 (n = 53; Fig. 10). Smaller particle fractions generally had lower mean NP:AP ratios (Fig. 11). All Type I mean NP:AP ratios were >5. These NP:AP predictions are consistent with observed effluent from the basal drain effluent of the Type I test pile that shows typ-

1

0.1

0.01

0

5

10

15

20

25

3:1

1

0.1

0.01

30

1:1

0

5

10

15

20

25

30

NP (kg tonne-1 CaCO3)

NP (kg tonne-1 CaCO3 )

Fig. 10. NP versus AP for all particle size fractions for (a) Type I test pile material and (b) Type III test pile material.

10

10

NP:AP

(b) 100

NP:AP

(a) 100

1

0.1

0.01

1

0.1

0.1

1

0.01

10

0.1

Diameter (mm)

1

10

Diameter (mm)

Fig. 11. NP:AP ratios for the <40 mm particle size fraction for the (a) Type I waste rock (b) Type III waste rock. Small gray symbols are individual samples NP:AP. Large open symbols represent size fraction average based on w(NP:AP) calculations. Large filled symbols represent size fraction average based on wNP:wAP calculation. Horizontal gray lines show the 1:1 and 3:1 lines. Dotted horizontal black line represents l(NP:AP) and dashed horizontal black line represents lNP:lAP).

60 50

4

30 20

2

10 0

70 8

60 50 40

6 4

30 20

2

10

-J an 01 -07 -J u 01 l-07 -J an 01 -08 -J u 01 l-0 -J 8 an 01 -09 -J u 01 l-0 -J 9 an 01 -10 -J u 01 l-1 -J 0 an 01 -11 -J u 01 l-11 -J an -1 2

01 -

Ja n 0 1 - 07 -J u 01 l-0 -J 7 an 01 - 0 8 -J u 01 l-0 -J 8 an 01 -09 -J u 01 l-0 -J 9 an 01 -10 -J u 01 l-10 -J an 01 - 1 1 -J u 01 l-11 -J an -1 2

0

01

40

6

10

pH

8

-1

70

(b) 80 Alkalinity (mg L as CaCO3)

10

pH

-1

Alkalinity (mg L as CaCO3)

(a) 80

Fig. 12. Trends of pH (black lines) and alkalinity (grey lines) from (a) the Type I test pile basal drain from 2008–2011 (no samples in 2007) and (b) the Type III test pile south basal drain from 2007–2011. After Bailey et al. (2013).

208

L.J.D Smith et al. / Applied Geochemistry 36 (2013) 200–209

(b) 80

8

60 50

30 20 10

6

-1

40

Alkalinity (mg L as CaCO3)

70

4 2

0

10

70 8

60 50 40 30 20

6

pH

10

pH

-1

Alkalinity (mg L as CaCO3)

(a) 80

4 2

10 0

0

20

40

60

80

100

0

20

weeks

40

60

80

100

weeks

Fig. 13. pH (black lines), and mean alkalinity (grey lines) for material used in the laboratory humidity cell experiments conducted at room temperature for (a) Type I material collected in 2004 with average 0.013 wt.% S and NP:AP of 9.0 (solid black and grey lines) and Type I material collected in 2005 with average 0.012 wt.% S and NP:AP of 5.1 (dotted black and grey lines); and (b) Type III material collected in 2004 with average 0.13 wt.% S and NP:AP of 0.6 (solid black and grey lines) and Type III material collected in 2005 with average 0.029 wt.% S and NP:AP of 7.6 (dotted black and grey lines).

ically circumneutral drainage with measurable alkalinity (Fig. 12; Bailey et al., 2013). Type III waste rock had lower NP:AP ratios than the Type I waste rock. Both waste rock types had similar C contents with weighted averages of 0.033 wt.% for Type I and 0.027 wt.% for Type III. Thus, the lower NP:AP in the Type III waste rock is caused by higher S content (weighted averages of 0.007 wt.% S for Type I and 0.033 wt.% S for Type III). Similar to Type I, Type III waste rock had a general trend of progressively lower NP:AP values for fractions <14 mm (Fig. 11). NP:AP values for individual sample fractions ranged from 0.09 to 115. The NP:AP of size fractions 65 mm were typically 62.2, the weighted average NP:AP of the individual samples (n = 297). A NP:AP of 2.2 suggests uncertain acid-generating potential. Effluent from the Type III test piles exhibits declining pH to <4.5 at the end of each field season, concomitant with depletion in alkalinity to <1.0 mg L1 (as total CaCO3; Fig. 12; Bailey et al., 2013). Comparing the calculated NP:AP to the effluent alkalinity and pH trends suggests this NP:AP calculation approach is reasonable but may still underestimate acid generating potential. 4.4.3. Humidity cell experiments Waste rock from 2004 and 2005 used in the Type I humidity cells had average S contents of 0.013 and 0.012 wt.%, respectively, and average C contents of 0.045 and 0.024 wt.%, respectively (average of three replicates). Both the 2004 and 2005 Type I waste rock were within the Type I waste rock designation (<0.04 wt.% S) and the NP:AP calculations suggest non-acid generating potential with NP:AP of 9.0 and 5.1 for the 2004 and 2005 waste rock, respectively. Laboratory results from the same humidity cells of Type I waste rock showed circumneutral pH levels (typically 5.5 < pH < 8.5) and low, stable alkalinity levels (2.8–17.1 mg L1 as CaCO3) for the 95 weeks of this data set. Type I cell effluent pH and alkalinity are consistent with the NP:AP ratio calculations that suggest the drainage will not be acidic (Fig. 13). The 2004 Type III rock in the humidity cells had an average S content of 0.13 wt.% S (average of three replicates), consistent with the Type III waste rock designation of >0.08 wt.% S. The 2004 Type III room temperature humidity cell effluent became acidic and then stabilized near pH 4. The NP values of less than 2.4 kg t1 CaCO3, reflect the paucity of calcite in the 2004 Type III waste rock sample. The calculated NP:AP of 0.6 predicted the sample would be acid generating. The humidity cell effluent contained little alkalinity initially and all of the available neutralization potential was consumed within the first 30 weeks of the humidity cell test. After the neutralization potential was depleted, alkalinity remained

<1.0 mg L1 (as total CaCO3) and the pH decreased and stabilized near pH 4 (Fig. 13). Type III waste rock collected in 2005 and used in the humidity cells had an average S content of 0.029 wt.%, within the Type I waste rock designations. This sulfide content is equivalent to an AP value of 0.91. The NP value of 6.9 and the NP:AP of 7.6 suggest this waste rock sample would be non-acid generating. In fact, the 2005 Type III waste rock used in the humidity cells had the highest NP of all the column waste rock for both the Type I and Type III waste rock, which emphasizes the variability in the Diavik waste rock and the effect of sample size. Humidity cell results for the 2005 Type III waste rock showed stable, slightly alkaline to circumneutral pH values and stable alkalinity levels. The humidity cell results from these 2005 Type III samples also exhibited the highest column effluent pH and alkalinity values, which persisted at relatively constant values for the 95 weeks of this study period. These results are consistent with the NP:AP ratio predictions (Fig. 13). The 2004 and 2005 Type I waste rock used in the humidity cells have similar NP:AP values and column effluent pH and alkalinity trends. The NP:AP values suggest non-acid generating waste rock. The Type I humidity cell effluent pH values were circumneutral and stable and the alkalinity values were low but stable. The 2004 Type III waste rock from the humidity cells had NP:AP values that suggested acid generating waste rock, which is consistent with the low pH and alkalinity values of the column effluent. The 2005 Type III column effluent had the highest pH and alkalinity levels of the columns, and was consistent with the NP:AP calculations that indicated this waste rock was non-acid generating. NP:AP ratios calculated for the waste rock used in the laboratory experiments, together with the column pH and alkalinity results from 95 weeks, suggest the lNP:lAP calculation approach is acceptable. 5. Summary and conclusions Particle size distributions of the waste rock from the Diavik mine site indicated the test piles have rock-like characteristics. Sulfur content measurements varied with the scale of the measurement: concentrations determined on <50 mm samples exhibited higher S concentrations than blast-pattern (>10 m scale) averages, and decreasing S concentrations with increasing particle size. A comparison of different NP:AP calculations suggests that NP:AP values are very sensitive to the calculation method: averaging NP:AP values from a given sample tends to provide a higher NP:AP compared to taking the average NP and dividing by the average AP of a sample set. A comparison of the field and laboratory re-

L.J.D Smith et al. / Applied Geochemistry 36 (2013) 200–209

sults suggested that ratios calculated using the latter approach based on measured S concentrations and interpreted using standard guidance ratios reasonably estimated the potential for acid generation for both the Type I and the Type III waste rock. This calculation approach was assessed for these samples by comparing NP:AP values to effluent pH and alkalinity trends from humidity cells and field-scale waste rock piles. Acknowledgements This research is part of the Diavik Waste Rock Research Program, a joint research program by the University of Waterloo, University of British Columbia and University of Alberta. The project is made possible through funding provided by the Natural Sciences and Engineering Research Council of Canada, the International Network for Acid Prevention, the Mine Environment Neutral Drainage program, the Canada Foundation for Innovation, and Rio Tinto (Diavik Diamond Mines Inc.). We thank G. Macdonald, B. Forsyth, S. Wytrychowski and J. Reinson for their support and commitment to the project. T. Deans, J. McKellar and E. McCulloch provided occasional field supervision and survey support. R. Amos provided field assistance and guidance. J. Bain, C. MacDonald, A. Mosher, K. Nicoll, J. Torbiak, A. McRae, S. Cousins, K. Nicol, B. Greenwood, O. Magbade, R. Pippy, M. McLeish, H. Ruiz, B. MacNeil and R. Simms provided field and laboratory assistance. M. King and D. Woodward of DDMI collected the blast-hole cuttings and performed the sulfur analyses at Diavik. S. Pinter of DDMI conducted the iSite survey. M. Moore conducted the kinetic tests. The authors also thank K. Lappako and an anonymous reviewer for their thorough reviews of the manuscript. References Amos, R.T., Blowes, D.W., Smith, L., Sego, D.C., 2009. Measurement of wind-induced pressure gradients in a waste rock pile. Vadose Zone J. 6, 953–962. ASTM D422-63 Standard Test Method, 2002. Standard test method for particle-size analysis of soils. In: Annual Book of ASTM Standards, American Society for Testing and Materials. Bailey, B.L., Smith, L.J.D., Blowes, D.W., Ptacek, C.J., Smith, L., Sego, D.C., 2013. The Diavik Waste Rock Project: persistence of contaminants from blasting agents in waste rock effluent. Appl. Geochem. 36, 256–270. Cathles, L.M., 1979. Predictive capabilities of a finite difference model of copper leaching in low grade industrial sulfide waste dumps. Math. Geol. 11, 175–191. Davis, G.B., Ritchie, A.I.M., 1986. A model of oxidation in pyretic mine wastes. 1. Equations and approximate solution. Appl. Math. Modell. 10, 314–322. Davis, G.B., Ritchie, A.I.M., 1987. A model of oxidation in pyretic mine wastes. 3. Import of the particle size distribution. Appl. Math. Modell. 10, 417–422. Davis, G.B., Doherty, G., Ritchie, A.I.M., 1986. A model of oxidation in pyretic mine wastes. 2. Comparison of numerical and approximate solution. Appl. Math. Modell. 19, 323–329. Dawson, R., Morgenstern, N., 1995. Liquifaction Flowslides in Rocky Mountain coal Waste Dumps, Phase 3, Final Report. Natural Resources Canada, Ottawa, Ontario. Contract Report 23440-3-9135/01. Environment Canada, 2008. Monthly Data Report for Ekati A, Northwest Territories, Bulk Data 1998–2007. . INAC [Indian and Northern Affairs, Canada], 1993. Guidelines for Acid Rock Drainage Prediction in the North. Catalogue Number R71-50/1E. Jambor, J.L., Dutrizac, J.E., Raudsepp, M., Groat, L.A., 2003. Effect of peroxide on neutralization-potential values of siderite and other carbonate minerals. J. Environ. Qual. 32, 2373–2378. Jambor, J.L., Dutrizac, J.E., Raudsepp, M., 2007. Measured and computed neutralization potentials from static tests of diverse rock types. Environ. Geol. 52, 1173–1185.

209

Lamontagne, A., Lefebvre, R., Poulin, R., Leroueil, S., 1999. Modelling of acid mine drainage physical processes in a waste rock pile with layered co-mingling. In: Proc. 52nd Canadian Geotechnical Conf., Regina, Saskatchewan, October 24–27, pp. 479–485. Lapakko, K.A., 1994. Comparison of Duluth Complex rock dissolution in the laboratory and field. In: Proc. 3rd Internat. Conf. Abatement of Acidic Drainage, vol. 1, April 25–29, 1994, Pittsburgh, PA, pp. 419–428. Lapakko, K.A., Antonson, D.A., 2006. Pyrite oxidation rates from humidity cell testing of greenstone rock. In: Proc. 2006, 7th ICARD, March 26–30, 2006, St. Louis, MO, Published by ASMR, 3134 Montavesta Rd., Lexington, KY 40502, pp. 1007–1025. Lapakko, K.A., Engstrom, J., Antonson, D.A., 2006. Effects of particle size on drainage quality from three lithologies. In: Proc. 2006, 7th ICARD, March 26–30, 2006, St. Louis, MO, Published by ASMR, 3134 Montavesta Rd., Lexington, KY 40502, pp. 1026–1050. Lefebvre, R., Hockley, D., Smolensky, J., Gelinas, P., 2001a. Multiphase transfer processes in waste rock piles producing acid mine drainage 1. Conceptual model and system characterization. J. Contam. Hydrol. 52, 137–164. Lefebvre, R., Hockley, D., Smolensky, J., Lamontagne, A., 2001b. Multiphase transfer processes in waste rock piles producing acid mine drainage 2. Applications of numerical simulation. J. Contam. Hydrol. 52, 165–186. Levenspiel, O., 1972. Chemical Reaction Engineering. J. Wiley, New York. Lindsay, M.B.J., Condon, P.D., Jambor, J.L., Lear, K.G., Blowes, D.W., Ptacek, C.J., 2009. Mineralogical, geochemical, and microbial investigation of a sulfide-rich tailings deposit characterized by neutral drainage. Appl. Geochem. 24, 2212– 2221. Mayer, K.U., Frind, E.O., Blowes, D.W., 2002. Multicomponent reactive transport modeling in variably saturated porous media using a generalized formulation for kinetically controlled reactions. Water Resour. Res. 38, 1174–1195. Mayer, K.U., Frind, E.O., Blowes, D.W., 2003. Advances in reactive-transport modelling. In: Jambor, J.L., Blowes, D.W., Ritchie, A.I.M. (Eds.), Environmental Aspects of Mine Wastes, Mineral. Assoc. Can. Short Course, vol. 31, pp. 383–302. Neuner, M., Smith, L., Blowes, D.W., Sego, D.C., Smith, L.J.D., Gupton, M., 2013. The Diavik Waste Rock Project: water flow though waste rock in a permafrost terrain. Appl. Geochem. 36, 222–233. Nichol, C., Smith, L., Beckie, R., 2005. Field-scale experiments of unsaturated flow and solute transport in a heterogeneous porous medium. Water Resour. Res. 41, W05018. Skousen, J., Renton, J., Brown, H., Evans, P., Leavitt, B., Brady, K., Cohen, L., Ziemkiewicz, P., 1997. Neutralization potential of overburden samples containing siderite. J. Environ. Qual. 26, 673–681. Smith, L., Beckie, R., 2003. Hydrologic and geochemical transport processes in mine waste rock. In: Jambor, J.L., Blowes, D.W., Ritchie, A.I.M. (Eds.), Environmental Aspects of Mine Wastes. Mineral. Assoc. Can. Short Course, vol. 31, pp. 51–72. Smith, L.J.D., Bailey, B.L., Blowes, D.W., Jambor, J.L., Smith, L., Sego, D.C., 2013. The Diavik Waste Rock Project: initial geochemical response from a low sulphide waste rock pile. Appl. Geochem. 36, 210–221. Smith, L.J.D., Moncur, M.C., Neuner, M., Gupton, M., Blowes, D.W., Smith, L., Sego, D.C., 2013. The Diavik Waste Rock Project: design, construction and instrumentation of field-scale experimental waste rock piles. Appl. Geochem. 36, 187–199. Sobek, A.A., Schuller, W.A., Freeman, J.R., Smith, R.M., 1978. Field and Laboratory Methods Applicable to Overburdens and Minesoils. EPA Report No. EPA-600/278-054. Stockwell, J., Smith, L., Jambor, J.L., Beckie, R., 2006. The relationship between fluid flow and mineral weathering in heterogeneous unsaturated porous media: a physical and geochemical characterization of a waste-rock pile. Appl. Geochem. 21, 1347–1361. Strömberg, B., Banwart, S., 1999. Experimental study of acidity-consuming processes in mining waste rock: some influences of mineralogy and particle size. Appl. Geochem. 14, 1–16. Velbel, M.A., 1993. Constancy of silicate-mineral weathering-rate ratios between natural and experimental weathering: implications for hydrologic control of differences in absolute rates. Chem. Geol. 105, 89–99. Weber, P.A., Thomas, J.W., Skinner, W.M., Smart, R.St.C., 2004. Improved acid neutralization capacity assessment of iron carbonates by titration and theoretical calculation. Appl. Geochem. 19, 687–694. White III, W.W., Lapakko, K.A., Cox, R.L., 1999. Static test methods most commonly used to predict acid-mine drainage: practical guidelines for use and interpretation. In: Plumlee, G.S., Logsdon, M.J. (Eds.), The Environmental Geochemistry of Mineral Deposits. Part A: Processes, Techniques and Health Issues. Rev. Econ. Geol. 6A, pp. 325–338. Wunderly, M.D., Blowes, D.W., Frind, E.O., Ptacek, C.J., 1996. Sulfide mineral oxidation and subsequent reactive transport of oxidation products in mine tailings impoundments: a numerical model. Water Resour. Res. 32, 3173–3187.