572178
research-article2015
WMR0010.1177/0734242X15572178Waste Management & ResearchPereira de Almeida et al.
Original Article
Reduction of acid rock drainage using steel slag in cover systems over sulfide rock waste piles
Waste Management & Research 2015, Vol. 33(4) 353–362 © The Author(s) 2015 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0734242X15572178 wmr.sagepub.com
Rodrigo Pereira de Almeida, Adilson do Lago Leite and Anderson Borghetti Soares
Abstract The extraction of gold, coal, nickel, uranium, copper and other earth-moving activities almost always leads to environmental damage. In metal and coal extraction, exposure of sulfide minerals to the atmosphere leads to generation of acid rock drainage (ARD) and in underground mining to acid mine drainage (AMD) due to contamination of infiltrating groundwater. This study proposes to develop a reactive cover system that inhibits infiltration of oxygen and also releases alkalinity to increase the pH of generated ARD and attenuate metal contaminants at the same time. The reactive cover system is constructed using steel slag, a waste product generated from steel industries. This study shows that this type of cover system has the potential to reduce some of the adverse effects of sulfide mine waste disposal on land. Geochemical and geotechnical characterization tests were carried out. Different proportions of sulfide mine waste and steel slag were studied in leachate extraction tests. The best proportion was 33% of steel slag in dry weight. Other tests were conducted as follows: soil consolidation, saturated permeability and soil water characteristic curve. The cover system was numerically modeled through unsaturated flux analysis using Vadose/w. The solution proposed is an oxygen transport barrier that allows rain water percolation to treat the ARD in the waste rock pile. The results showed that the waste pile slope is an important factor and the cover system must have 5 m thickness to achieve an acceptable effectiveness. Keywords Mining waste, steel slag, acid rock drainage, geochemical tests, geotechnical tests, cover system.
Introduction Sulfide minerals may be exposed to the atmosphere during the waste disposal process in mining or in large civil constructions. These processes move large volumes of material or alter the hydrogeological factors in an area. In the presence of water, oxygen and bacteria (Thiobacillus ferrooxidans), the sulfide minerals produce an acidic solution and Fe3+ (Stumm and Morgan, 1981; US Environmental Protection Agency, 1994; Pastore and Mioto, 2000; Borma and Soares, 2002; Boscov, 2008; MEND Program, 2009). Thus, oxidation of sulfides, such as pyrite, generates acid rock drainage (ARD) that can be represented by
FeS2(s) + 3.5 O 2 + H 2O → 2SO 4 2− + 2H + + Fe 2+ (1)
Fe2+ + 0.25 O 2 + H + → Fe3+ + 0.5 H 2O (2)
The reaction (1) releases sulfuric acid. In the groundwater, it may cause solubilization of the rocks and release heavy metals. According Shinobe and Sracek (1997), an acid plume containing high concentrations of dissolved ions is formed. Furthermore, if it reaches the groundwater, rivers and lakes, it pollutes the
environment through bioaccumulation and biomagnification within the food chain (Murta, 2006). The neutralization of this plume may occur naturally in the presence of limestone or artificially by adding lime (CaO) or calcite (CaCO3) to the sulfide material. According to Boscov (2008), the following reactions neutralize the ARD: SO 4 2− + CaCO3(S) + 2H + + H 2O → CaSO 4 .2H 2O(s) + CO 2
(3)
Fe3+ + 2H 2O → FeO ( OH )(s ) + 3H + (4)
Fe3+ + 3H 2O → Fe ( OH )3(s ) + 3H + (5)
Universidade Federal de Ouro Preto, Ouro Preto, Brazil Corresponding author: Rodrigo Pereira de Almeida, Universidade Federal de Ouro Preto, Ouro Preto, R. Dr. Noé Ferreira, 111, Bela Vista, Patos de Minas 38703234, Brazil. Email: [email protected]
Downloaded from wmr.sagepub.com at UNIV OF PITTSBURGH on November 14, 2015
354
Waste Management & Research 33(4)
In equation (3), the pH is increased, and in equation (4), the FeO (OH) precipitates. Finally, as shown in reaction (5), Fe(OH)3 is released which is known as “Yellow Boy”. Steel slag, a waste obtained in the steel refining process, has high concentrations of lime. The mineralogical composition is β-dicalcium silicate (larnita), calcium ferrite, wustita (FeO), lime (CaO), periclase (MgO), portlandite (Ca(OH)2), calcite (CaCO3), and other lesser minerals (Fernandes, 2010; Sousa, 2007; Machado, 2000). Therefore, calcite is present in steel slag as has been proven in laboratory and field tests developed by Ziemkiewicz and Skousen (1998). Machado and Schneider (2008) conducted leachate extraction tests with mixtures of steel slag and coal sulfide tailings. As result of these tests, the optimal ratio of each material was established. Salviano et al. (2010) evaluated steel slag as a replacement for lime in ARD treatment systems. Ziemkiewicz and Skousen (1998) have shown that steel slag has the advantage of providing higher levels of alkalinity to the environment for a long period of time. Experiments by Simmons et al. (2002) show that the steel slag has high alkalinity and releases that alkalinity for at least 10 years. In addition, the steel industry produces 100–170 kg of steel slag per ton of steel produced (Machado, 2000; Sousa, 2007). Annual production is around 5 million tons; 2.24 million tons are disposed of on land (Machado, 2000). Around 40% of steel slag is currently used as road metal. The remaining 56% is stockpiled (IBIS, 2007). Thus, much of the production of this waste is not recycled. In the southern region of Brazil, 310 million tons of sulfide waste was produced since 1925 (Santos et al., 2013). The major part of the waste material was accommodated without technological intervention. Currently, overburden piles and tailing dams are constructed for retention and minimal treatment. Borghetti Soares and Souza (2008) developed pilot tests for a cover system for the Mina do Verdinho coal mine located at Santa Catariana, Brazil. Amorim (2008) and Ribeiro (2011) developed a cover system for a tailing dam in Paracatu, Minas Gerais, Brazil. The Mine Morro do Ouro in Paracatu produces 15 tons of gold per year. The metal concentration is around 0.4 g of gold per ton of excavated material. The amount of sulfide waste produced is approximately 3.75 million tons per year. Furthermore, Assis (2006) studied plants to revegetate this area during the mine closure process. This study evaluates the use of steel slag to neutralize the ARD generated by a material sampled in an abandoned mine, the Mina de Pirita, located in Ouro Preto, Brazil. Subsequently, a reactive cover system was designed using steel slag to minimize acid generation. Although this cover system allows an infiltrating water flux, the oxygen flux required to activate reactions (1) and (2) described above is blocked. Accordingly, Herrmann et al. (2010) suggest that steel slag has a promising potential as a suitable cover layer material under the right conditions.
Materials Samples of steel slag (ES1), a by-product from the electric arc furnace at Usiminas steel plant in Ipatinga, Brasil (Fernandes, 2010), were obtained. This material had been stored for a period of 6 months and then was collected and transported to Universidade Federal de Ouro Preto (UFOP). The sampling process was carried out according to NBR 5564 (ABNT, 2011). About 200 kg of this material was crushed using a jaw crusher (type Blake, 130 mm × 100 mm, 1 camshaft) and a roller crusher (120 mm × 80 mm). In addition, a sample of the sulfide waste was collected from the abandoned mine, Mina de Pirita located between the towns of Ouro Preto and Mariana in Minas Gerais, Brazil. Pyrite (FeS2) for sulfur and sulfuric acid production had been extracted from this mine between 1930 and 1960 (Moraes, 2010). The samples from the Mina de Pirita (AMP) were collected and transported to the geotechnical laboratory at UFOP. The amount of this material stored was also around 200 kg.
Methods The ingress of oxygen into the layers of the cover system was evaluated numerically in this work using the software Vadose/w (Geo-Slope International Ltd, 2008). The cover system designed may be classified as an inhibiting reaction and as oxygen transport barrier (O’Kane et al., 2002). The method of Van Genuchten (1980) was chosen to obtain the hydraulic conductivities in the unsaturated zone.
Geochemical tests The samples were geochemically characterized by the following methods: X-ray diffraction, chemical composition by total acid digestion, paste pH and conductivity analyses (O’Kane et al., 2002; Camargo et al., 2009), cation exchange capacity (CEC) determination and specific surface (Pejon, 1992) and leach extraction tests. In the leach extraction tests, mixtures with different proportions of ES1 and AMP were carried out to find the optimal amount of steel slag that is required to neutralize the ARD. The leaching columns were constructed using PVC columns with 9.3 cm diameter and 22 cm height. The samples were prepared using the dry weight of each material. The experiment consisted of six columns with different mixtures, as shown in Figure 1 and described in Table 1. The volume of water per each column was the amount corresponding to two rainy months totaling 700 mm m-2. Therefore, the total water volume of each column was 4755.04 ml. Water was gradually added to the columns during the test period of 100 days. This corresponds to 47.5 ml of water per day. The leachate was retained in beakers and measurements of pH, Eh, electrical conductivity (EC) and temperature were performed periodically.
Downloaded from wmr.sagepub.com at UNIV OF PITTSBURGH on November 14, 2015
355
Pereira de Almeida et al.
Figure 1. Diagram of columns with different proportions of ES1 and AMP.
Geotechnical tests The input data to the numerical modeling were as follows: saturated hydraulic conductivity, water retention curve, and coefficient of volume compressibility (mv) obtained by consolidation test. The geotechnical properties were determined according to the procedures below: - particle size distribution (PSD) (ABNT, 1984d); - Atterberg limits (ABNT, 1984a, 1984c); - specific gravity (ABNT, 1984b); - consolidation-saturated test (ABNT, 1990); - saturated permeability (ABNT, 1995, 2000); - soil water characteristic or moisture retention curve (Gardner, 1937; Rodrigues, 2007).
Cover system design The Vadose/W (Geo-Slope International Ltd, 2008) software uses finite elements to analyze flow in the non-saturated zone. This program addresses the interaction of the soil with the atmosphere. Thus, temperature and rainfall data are inputs. The evaluation of the volume of infiltration water and evapotranspiration are outputs. A key purpose of this tool is to design and evaluate the covers systems performance. The cover system was designed over a waste pile as shown in Figure 2. The waste pile was 20 m high and 40 m long. The analyses were “transient coupled”. The climate type in Ouro Preto is seasonally humid. The size of the standard element of the finite element mesh was 1.2 m. The cover system was designed in three layers:
Layer 1: material collected from the Mina de Pirita (AMP). Layer 2: mixture M33 cited in Table 1, its thickness ranged. Layer 3: composed of steel slag (ES1), 20 cm thick. The mixture M33 was used as layer 2’s material, because according to the leach extraction tests, it demonstrated the ability to neutralize the ARD. In addition, the clay materials in its composition provide high capillary values. Therefore, the steel slag present in layers 3 and 2 contributes to the reaction role of the system. The functions of layer 3 are also to improve slope stability, protect against erosion (Motz and Geiseler, 2001) and prevent excessive evaporation during dry periods. This layer maintains a high saturation degree that can reduce the oxygen ingress (O’Kane et al., 2002). Yanful (1993) suggests that the cover systems must address the reduction in the oxygen ingress into the sulfide materials. If the degree of saturation remains above 85%, it reportedly effectively reduces the oxygen flux. Layer 2 was similarly analyzed in a range from 0.8 to 9.0 m thick. The cumulative gas flux, cumulative liquid flux and minimum degree of saturation were determined for each thickness and climate condition. The upper boundary conditions simulated the climate of Ouro Preto, Brasil. This climate is typically humid. This boundary condition was symbolized in Figure 1 by double arrows. There were two sequences of simulations: one using the maximum precipitation, which corresponds to 2200 mm/m², and then about a 50% reduction corresponding to 1021 mm/m² per year. These precipitation values are extreme (Oliveira, 2008) and correspond to the years 1985 and 1990. Another boundary condition represented by triangles on the right-hand side of Figure 1 is drainage (Q = 0 m³/day). Finally,
Downloaded from wmr.sagepub.com at UNIV OF PITTSBURGH on November 14, 2015
356
Waste Management & Research 33(4)
Figure 2. Cover system analyzed as a reaction inhibiting and oxygen transport barrier. Table 1. Table containing the proportions of material in each leaching test. Mixture
ES1 (g)
AMP (g)
Total (g)
ES1 (%)
ES1 M33 M25 M20 M14 AMP
300 150 100 75 50 0
0 300 300 300 300 300
300 450 400 375 350 300
100.0 33.3 25.0 20.0 14.3 0.0
the dashed line in Figure 1 is the initial water table that was automatically reviewed during the interactions performed.
Results and discussion Geochemical results The minerals identified in ES1 by X-ray diffraction methods are: calcium oxide, CaO; wustita, FeO; quartz, SiO2; and one other mineral of lesser concentrations.The values obtained by total acid digestion chemical analysis are described in Table 2. The results are similar to the results obtained by Pena (2004), Sousa (2007), Fernandes (2010) and Herrmann et al. (2010). A steel slag sample (ME1) classified by Sousa (2007) as class II (non-inert and non-hazardous) exhibited solubilized concentrations of aluminum higher than the limit regulated (ABNT, 2004). Moreover, the difference in the percentage of lime (CaO), which is higher than 10%, may be caused by a 1-year atmosphere exposure. The paste pH values and electrical conductivity of ES1 were respectively 12.5 and 218.3 mS/cm. The pH is highly alkaline and the electrical conductivity shows a polluted leachate; it is in agreement with Salviano et al. (2010). In addition, the CEC value
is 0.29 meq/100 g and the specific surface area is 2.1 m2/g. This demonstrates low reactivity and a small specific surface area. In sequence, the X-ray diffraction test identified the following minerals for AMP: biotite, kaolinite, quartz, goethite, erlichmanite, iron oxides and other lesser minerals. Erlichmanite (OsS2) is a sulfide and indicates an acid-generating potential. The presence of biotite, quartz and kaolin were also detected by Moraes (2010) in samples collected from the same mine. Chemical analysis showed that the most significant part of the material (45.5%) is represented by quartz (SiO2). The iron oxides are secondary (26.87%), possibly indicating precipitation of ARD leachate (Dold, 1999). The presence of aluminum is also significant with 15.12%. Sulfur confirms that it is a sulfide material with almost 2% of total weight. In addition, arsenic was detected; leach extraction tests performed by Moraes (2010) also demonstrated the presence of arsenic. The measured values of pH and EC of AMP were respectively 2.04 and 17513.6 mS/cm. These data confirm the occurrence of the ARD at high rates (MEND Program, 2009). The CEC values and specific surface area of AMP was 11 meq/100 g and 136.4 m2/g, respectively. This confirms the presence of clay minerals (kaolinite) (Pejon, 1992). The pH measurements graph of the leach extraction tests, shown in Figure 3(a), indicates consistent results compared with tests by Machado and Schneider (2008) who studied sulfide coal tailings from a region in southern Brazil mixed with steel slag. The ES1 sample has maintained an alkaline pH between 10 and 12 during all experiments. Furthermore, this sample exhibited a white leachate, suggesting that the material is highly rich in calcium. The pH of the column that contained only AMP remained acidic at about 2.3. Values below pH 3.5 indicated the presence
Downloaded from wmr.sagepub.com at UNIV OF PITTSBURGH on November 14, 2015
357
Pereira de Almeida et al. Table 2. Results of total acid digestion chemical analysis for steel slag. Constituent
Al2O3
CaO
Fe2O3
MgO
MnO
P2O5
TiO2
SiO2
S
ES1 ME1
5.41 2.36
32.17 43.80
35.83 36.86
11.38 1.68
2.73 4.95
1.06 1.43
0.48 –
9.32 8.03
0.09 0.09
(%) (%)
of acidophilic bacteria (MEND Program, 2009). To this extent, Martins et al. (2004) studied samples from the Mina de Pirita and identified the T. ferrooxidans bacterium. These mixtures had the pH gradually increased throughout the test period. There was an exception for M33 that had fluctuated for the first 20 days and, afterwards stabilized at about pH 6.5. A pH in the range of 6–8 can be considered within a neutral “range” according to the Canadian MEND Program (2009). The M25 pH was neutral after 100 days at a proportion of 1:3 that was suggested by Machado and Schneider (2008) to neutralize sulfide coal tailings. The reduction potential (Eh) of AMP showed high values, above 500 mV, throughout the monitored period as demonstrated by Figure 3(b). These values are in agreement with Martins et al. (2004) and Machado and Schneider (2008). In addition, these data indicate a strong reduction potential and a presence of acidophilic bacteria. The M14 and M20 mixtures attenuated the oxidation process (Figure 3(c)). However, they remained in the range 400–500 mV. The M25 test reduced the values of Eh, and after 90 days this value was reduced to about 300 mV. M33 has reached similar values. The M33 mixture has showed Eh values between 200 and 300 mV since the beginning. These results demonstrate an immediate neutralization. In the first 10 days, the EC dropped in all columns containing AMP. Initially, these values were high, which showed elevated ion concentrations. After 10 days, the values were reduced and were subsequently stabilized in the range of 2000 mS/cm, indicating a polluted leachate. Other works showed stabilization around 2000 and 2500 mS/cm (Machado and Schneider, 2008; Moraes, 2010). The EC reduction may be explained by the absorption process proposed by Ekolu and Azene (2012) as a possible mechanism in the reduction of ion concentrations in the leachate. The highly alkaline mixture from the calcium oxides is negatively charged and the cations (mainly metals) in water are attracted and chemically absorbed onto the calcium oxides. The ES1 column maintained about 2000 mS cm-1 for almost all studied periods. This is due to calcium release. Finally, temperature measurements (Figure 3(d)) have remained close to the environmental temperature, even though the ARD reaction is exothermic.
Geotechnical results The geotechnical tests provided the results cited in Table 3. According to the Unified Soil Classification System (USCS) (American Society for Testing and Materials, 1985), the materials were classified as follows.
- ES1: well-graded sand (SW). - AMP: well-graded sand with clay (SW-SC). - M33: well-graded sand with clay (SW-SC). The soil–water characteristic curve is shown in Figure 4(a), in which the ES1 exhibits a distinct behavior compared with M33 and AMP. It is typically due to the difference in PSD. The ES1 is coarsegrained soil, while M33 and AMP present fines in their composition. Materials that present high moisture retention can be used as a capillary or oxygen inhibiting barrier (O’Kane et al., 2002). The graph of the unsaturated hydraulic conductivity versus matrix capillarity can also be seen in Figure 4(b). The ES1 shows high values of conductivity for small suction values and decreased values for higher suction values, this being typical for coarsegrained soils (O’Kane et al., 2002). In contrast, the AMP conductivity is smaller than ES1 for small values of matrix suction and increases for high levels of suction. This means that this material has clay in its PSD. The M33 unsaturated hydraulic conductivity has shown the smallest values in the beginning of the matrix suction. This may be explained by a clogging process caused by ferric hydroxide (“Yellow Boy”) precipitation.
Numerical modeling As M33 has efficiently neutralized ARD during the leach extraction tests, it was proposed to be the material for layer 2. Thus, to control its efficiency as a cover system, the water flow and oxygen ingress have been addressed at the bottom of this layer. The thickness of layer 2 and the climate boundary were varied for each interaction performed as observed in Figure 5(a). The climate boundary is based upon the precipitation related to the 1021 and 2200 mm m-2 per year (Oliveira, 2008). The results for other precipitation data in Ouro Preto are intermediate values. By analyzing the graph of maximum cumulative gas flux in Figure 5(a), the reduction in the gas flux (up to 1 m thick layer) is small; gas passes with little interference between the pores. In addition, the cumulative gas flux decreases significantly for the thickness between 1 and 3 m. For 5 m thick and more, the gas flux is very low. The maximum surface cumulative liquid flux is constant up to 1.60 m thick layer (Figure 5(b)). This is likely due to the two well-defined seasons. During the wet season, infiltration dominates; during the dry season, evaporation is predominant. This impacts the cumulative flux at the base of this layer. With thicknesses above 1.6 m, the cumulative liquid flux increases linearly. This layer contains more material; thus, it is impacted less by evaporation. During the wet season, this layer allows water infiltration, but, in the dry season, it maintains a high saturation degree due to moisture storage.
Downloaded from wmr.sagepub.com at UNIV OF PITTSBURGH on November 14, 2015
358
Waste Management & Research 33(4)
Figure 3. Leach extraction test results: (a) pH; (b) Eh (mV); (c) EC (electrical conductivity, µs/cm); (d) temperature (°C).
Downloaded from wmr.sagepub.com at UNIV OF PITTSBURGH on November 14, 2015
359
Pereira de Almeida et al. Table 3. Results obtained from geotechnical tests. Geotechnical results Material PSD Clay (φ< 0.002 mm): Silt (0.002
research-article2015
WMR0010.1177/0734242X15572178Waste Management & ResearchPereira de Almeida et al.
Original Article
Reduction of acid rock drainage using steel slag in cover systems over sulfide rock waste piles
Waste Management & Research 2015, Vol. 33(4) 353–362 © The Author(s) 2015 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0734242X15572178 wmr.sagepub.com
Rodrigo Pereira de Almeida, Adilson do Lago Leite and Anderson Borghetti Soares
Abstract The extraction of gold, coal, nickel, uranium, copper and other earth-moving activities almost always leads to environmental damage. In metal and coal extraction, exposure of sulfide minerals to the atmosphere leads to generation of acid rock drainage (ARD) and in underground mining to acid mine drainage (AMD) due to contamination of infiltrating groundwater. This study proposes to develop a reactive cover system that inhibits infiltration of oxygen and also releases alkalinity to increase the pH of generated ARD and attenuate metal contaminants at the same time. The reactive cover system is constructed using steel slag, a waste product generated from steel industries. This study shows that this type of cover system has the potential to reduce some of the adverse effects of sulfide mine waste disposal on land. Geochemical and geotechnical characterization tests were carried out. Different proportions of sulfide mine waste and steel slag were studied in leachate extraction tests. The best proportion was 33% of steel slag in dry weight. Other tests were conducted as follows: soil consolidation, saturated permeability and soil water characteristic curve. The cover system was numerically modeled through unsaturated flux analysis using Vadose/w. The solution proposed is an oxygen transport barrier that allows rain water percolation to treat the ARD in the waste rock pile. The results showed that the waste pile slope is an important factor and the cover system must have 5 m thickness to achieve an acceptable effectiveness. Keywords Mining waste, steel slag, acid rock drainage, geochemical tests, geotechnical tests, cover system.
Introduction Sulfide minerals may be exposed to the atmosphere during the waste disposal process in mining or in large civil constructions. These processes move large volumes of material or alter the hydrogeological factors in an area. In the presence of water, oxygen and bacteria (Thiobacillus ferrooxidans), the sulfide minerals produce an acidic solution and Fe3+ (Stumm and Morgan, 1981; US Environmental Protection Agency, 1994; Pastore and Mioto, 2000; Borma and Soares, 2002; Boscov, 2008; MEND Program, 2009). Thus, oxidation of sulfides, such as pyrite, generates acid rock drainage (ARD) that can be represented by
FeS2(s) + 3.5 O 2 + H 2O → 2SO 4 2− + 2H + + Fe 2+ (1)
Fe2+ + 0.25 O 2 + H + → Fe3+ + 0.5 H 2O (2)
The reaction (1) releases sulfuric acid. In the groundwater, it may cause solubilization of the rocks and release heavy metals. According Shinobe and Sracek (1997), an acid plume containing high concentrations of dissolved ions is formed. Furthermore, if it reaches the groundwater, rivers and lakes, it pollutes the
environment through bioaccumulation and biomagnification within the food chain (Murta, 2006). The neutralization of this plume may occur naturally in the presence of limestone or artificially by adding lime (CaO) or calcite (CaCO3) to the sulfide material. According to Boscov (2008), the following reactions neutralize the ARD: SO 4 2− + CaCO3(S) + 2H + + H 2O → CaSO 4 .2H 2O(s) + CO 2
(3)
Fe3+ + 2H 2O → FeO ( OH )(s ) + 3H + (4)
Fe3+ + 3H 2O → Fe ( OH )3(s ) + 3H + (5)
Universidade Federal de Ouro Preto, Ouro Preto, Brazil Corresponding author: Rodrigo Pereira de Almeida, Universidade Federal de Ouro Preto, Ouro Preto, R. Dr. Noé Ferreira, 111, Bela Vista, Patos de Minas 38703234, Brazil. Email: [email protected]
Downloaded from wmr.sagepub.com at UNIV OF PITTSBURGH on November 14, 2015
354
Waste Management & Research 33(4)
In equation (3), the pH is increased, and in equation (4), the FeO (OH) precipitates. Finally, as shown in reaction (5), Fe(OH)3 is released which is known as “Yellow Boy”. Steel slag, a waste obtained in the steel refining process, has high concentrations of lime. The mineralogical composition is β-dicalcium silicate (larnita), calcium ferrite, wustita (FeO), lime (CaO), periclase (MgO), portlandite (Ca(OH)2), calcite (CaCO3), and other lesser minerals (Fernandes, 2010; Sousa, 2007; Machado, 2000). Therefore, calcite is present in steel slag as has been proven in laboratory and field tests developed by Ziemkiewicz and Skousen (1998). Machado and Schneider (2008) conducted leachate extraction tests with mixtures of steel slag and coal sulfide tailings. As result of these tests, the optimal ratio of each material was established. Salviano et al. (2010) evaluated steel slag as a replacement for lime in ARD treatment systems. Ziemkiewicz and Skousen (1998) have shown that steel slag has the advantage of providing higher levels of alkalinity to the environment for a long period of time. Experiments by Simmons et al. (2002) show that the steel slag has high alkalinity and releases that alkalinity for at least 10 years. In addition, the steel industry produces 100–170 kg of steel slag per ton of steel produced (Machado, 2000; Sousa, 2007). Annual production is around 5 million tons; 2.24 million tons are disposed of on land (Machado, 2000). Around 40% of steel slag is currently used as road metal. The remaining 56% is stockpiled (IBIS, 2007). Thus, much of the production of this waste is not recycled. In the southern region of Brazil, 310 million tons of sulfide waste was produced since 1925 (Santos et al., 2013). The major part of the waste material was accommodated without technological intervention. Currently, overburden piles and tailing dams are constructed for retention and minimal treatment. Borghetti Soares and Souza (2008) developed pilot tests for a cover system for the Mina do Verdinho coal mine located at Santa Catariana, Brazil. Amorim (2008) and Ribeiro (2011) developed a cover system for a tailing dam in Paracatu, Minas Gerais, Brazil. The Mine Morro do Ouro in Paracatu produces 15 tons of gold per year. The metal concentration is around 0.4 g of gold per ton of excavated material. The amount of sulfide waste produced is approximately 3.75 million tons per year. Furthermore, Assis (2006) studied plants to revegetate this area during the mine closure process. This study evaluates the use of steel slag to neutralize the ARD generated by a material sampled in an abandoned mine, the Mina de Pirita, located in Ouro Preto, Brazil. Subsequently, a reactive cover system was designed using steel slag to minimize acid generation. Although this cover system allows an infiltrating water flux, the oxygen flux required to activate reactions (1) and (2) described above is blocked. Accordingly, Herrmann et al. (2010) suggest that steel slag has a promising potential as a suitable cover layer material under the right conditions.
Materials Samples of steel slag (ES1), a by-product from the electric arc furnace at Usiminas steel plant in Ipatinga, Brasil (Fernandes, 2010), were obtained. This material had been stored for a period of 6 months and then was collected and transported to Universidade Federal de Ouro Preto (UFOP). The sampling process was carried out according to NBR 5564 (ABNT, 2011). About 200 kg of this material was crushed using a jaw crusher (type Blake, 130 mm × 100 mm, 1 camshaft) and a roller crusher (120 mm × 80 mm). In addition, a sample of the sulfide waste was collected from the abandoned mine, Mina de Pirita located between the towns of Ouro Preto and Mariana in Minas Gerais, Brazil. Pyrite (FeS2) for sulfur and sulfuric acid production had been extracted from this mine between 1930 and 1960 (Moraes, 2010). The samples from the Mina de Pirita (AMP) were collected and transported to the geotechnical laboratory at UFOP. The amount of this material stored was also around 200 kg.
Methods The ingress of oxygen into the layers of the cover system was evaluated numerically in this work using the software Vadose/w (Geo-Slope International Ltd, 2008). The cover system designed may be classified as an inhibiting reaction and as oxygen transport barrier (O’Kane et al., 2002). The method of Van Genuchten (1980) was chosen to obtain the hydraulic conductivities in the unsaturated zone.
Geochemical tests The samples were geochemically characterized by the following methods: X-ray diffraction, chemical composition by total acid digestion, paste pH and conductivity analyses (O’Kane et al., 2002; Camargo et al., 2009), cation exchange capacity (CEC) determination and specific surface (Pejon, 1992) and leach extraction tests. In the leach extraction tests, mixtures with different proportions of ES1 and AMP were carried out to find the optimal amount of steel slag that is required to neutralize the ARD. The leaching columns were constructed using PVC columns with 9.3 cm diameter and 22 cm height. The samples were prepared using the dry weight of each material. The experiment consisted of six columns with different mixtures, as shown in Figure 1 and described in Table 1. The volume of water per each column was the amount corresponding to two rainy months totaling 700 mm m-2. Therefore, the total water volume of each column was 4755.04 ml. Water was gradually added to the columns during the test period of 100 days. This corresponds to 47.5 ml of water per day. The leachate was retained in beakers and measurements of pH, Eh, electrical conductivity (EC) and temperature were performed periodically.
Downloaded from wmr.sagepub.com at UNIV OF PITTSBURGH on November 14, 2015
355
Pereira de Almeida et al.
Figure 1. Diagram of columns with different proportions of ES1 and AMP.
Geotechnical tests The input data to the numerical modeling were as follows: saturated hydraulic conductivity, water retention curve, and coefficient of volume compressibility (mv) obtained by consolidation test. The geotechnical properties were determined according to the procedures below: - particle size distribution (PSD) (ABNT, 1984d); - Atterberg limits (ABNT, 1984a, 1984c); - specific gravity (ABNT, 1984b); - consolidation-saturated test (ABNT, 1990); - saturated permeability (ABNT, 1995, 2000); - soil water characteristic or moisture retention curve (Gardner, 1937; Rodrigues, 2007).
Cover system design The Vadose/W (Geo-Slope International Ltd, 2008) software uses finite elements to analyze flow in the non-saturated zone. This program addresses the interaction of the soil with the atmosphere. Thus, temperature and rainfall data are inputs. The evaluation of the volume of infiltration water and evapotranspiration are outputs. A key purpose of this tool is to design and evaluate the covers systems performance. The cover system was designed over a waste pile as shown in Figure 2. The waste pile was 20 m high and 40 m long. The analyses were “transient coupled”. The climate type in Ouro Preto is seasonally humid. The size of the standard element of the finite element mesh was 1.2 m. The cover system was designed in three layers:
Layer 1: material collected from the Mina de Pirita (AMP). Layer 2: mixture M33 cited in Table 1, its thickness ranged. Layer 3: composed of steel slag (ES1), 20 cm thick. The mixture M33 was used as layer 2’s material, because according to the leach extraction tests, it demonstrated the ability to neutralize the ARD. In addition, the clay materials in its composition provide high capillary values. Therefore, the steel slag present in layers 3 and 2 contributes to the reaction role of the system. The functions of layer 3 are also to improve slope stability, protect against erosion (Motz and Geiseler, 2001) and prevent excessive evaporation during dry periods. This layer maintains a high saturation degree that can reduce the oxygen ingress (O’Kane et al., 2002). Yanful (1993) suggests that the cover systems must address the reduction in the oxygen ingress into the sulfide materials. If the degree of saturation remains above 85%, it reportedly effectively reduces the oxygen flux. Layer 2 was similarly analyzed in a range from 0.8 to 9.0 m thick. The cumulative gas flux, cumulative liquid flux and minimum degree of saturation were determined for each thickness and climate condition. The upper boundary conditions simulated the climate of Ouro Preto, Brasil. This climate is typically humid. This boundary condition was symbolized in Figure 1 by double arrows. There were two sequences of simulations: one using the maximum precipitation, which corresponds to 2200 mm/m², and then about a 50% reduction corresponding to 1021 mm/m² per year. These precipitation values are extreme (Oliveira, 2008) and correspond to the years 1985 and 1990. Another boundary condition represented by triangles on the right-hand side of Figure 1 is drainage (Q = 0 m³/day). Finally,
Downloaded from wmr.sagepub.com at UNIV OF PITTSBURGH on November 14, 2015
356
Waste Management & Research 33(4)
Figure 2. Cover system analyzed as a reaction inhibiting and oxygen transport barrier. Table 1. Table containing the proportions of material in each leaching test. Mixture
ES1 (g)
AMP (g)
Total (g)
ES1 (%)
ES1 M33 M25 M20 M14 AMP
300 150 100 75 50 0
0 300 300 300 300 300
300 450 400 375 350 300
100.0 33.3 25.0 20.0 14.3 0.0
the dashed line in Figure 1 is the initial water table that was automatically reviewed during the interactions performed.
Results and discussion Geochemical results The minerals identified in ES1 by X-ray diffraction methods are: calcium oxide, CaO; wustita, FeO; quartz, SiO2; and one other mineral of lesser concentrations.The values obtained by total acid digestion chemical analysis are described in Table 2. The results are similar to the results obtained by Pena (2004), Sousa (2007), Fernandes (2010) and Herrmann et al. (2010). A steel slag sample (ME1) classified by Sousa (2007) as class II (non-inert and non-hazardous) exhibited solubilized concentrations of aluminum higher than the limit regulated (ABNT, 2004). Moreover, the difference in the percentage of lime (CaO), which is higher than 10%, may be caused by a 1-year atmosphere exposure. The paste pH values and electrical conductivity of ES1 were respectively 12.5 and 218.3 mS/cm. The pH is highly alkaline and the electrical conductivity shows a polluted leachate; it is in agreement with Salviano et al. (2010). In addition, the CEC value
is 0.29 meq/100 g and the specific surface area is 2.1 m2/g. This demonstrates low reactivity and a small specific surface area. In sequence, the X-ray diffraction test identified the following minerals for AMP: biotite, kaolinite, quartz, goethite, erlichmanite, iron oxides and other lesser minerals. Erlichmanite (OsS2) is a sulfide and indicates an acid-generating potential. The presence of biotite, quartz and kaolin were also detected by Moraes (2010) in samples collected from the same mine. Chemical analysis showed that the most significant part of the material (45.5%) is represented by quartz (SiO2). The iron oxides are secondary (26.87%), possibly indicating precipitation of ARD leachate (Dold, 1999). The presence of aluminum is also significant with 15.12%. Sulfur confirms that it is a sulfide material with almost 2% of total weight. In addition, arsenic was detected; leach extraction tests performed by Moraes (2010) also demonstrated the presence of arsenic. The measured values of pH and EC of AMP were respectively 2.04 and 17513.6 mS/cm. These data confirm the occurrence of the ARD at high rates (MEND Program, 2009). The CEC values and specific surface area of AMP was 11 meq/100 g and 136.4 m2/g, respectively. This confirms the presence of clay minerals (kaolinite) (Pejon, 1992). The pH measurements graph of the leach extraction tests, shown in Figure 3(a), indicates consistent results compared with tests by Machado and Schneider (2008) who studied sulfide coal tailings from a region in southern Brazil mixed with steel slag. The ES1 sample has maintained an alkaline pH between 10 and 12 during all experiments. Furthermore, this sample exhibited a white leachate, suggesting that the material is highly rich in calcium. The pH of the column that contained only AMP remained acidic at about 2.3. Values below pH 3.5 indicated the presence
Downloaded from wmr.sagepub.com at UNIV OF PITTSBURGH on November 14, 2015
357
Pereira de Almeida et al. Table 2. Results of total acid digestion chemical analysis for steel slag. Constituent
Al2O3
CaO
Fe2O3
MgO
MnO
P2O5
TiO2
SiO2
S
ES1 ME1
5.41 2.36
32.17 43.80
35.83 36.86
11.38 1.68
2.73 4.95
1.06 1.43
0.48 –
9.32 8.03
0.09 0.09
(%) (%)
of acidophilic bacteria (MEND Program, 2009). To this extent, Martins et al. (2004) studied samples from the Mina de Pirita and identified the T. ferrooxidans bacterium. These mixtures had the pH gradually increased throughout the test period. There was an exception for M33 that had fluctuated for the first 20 days and, afterwards stabilized at about pH 6.5. A pH in the range of 6–8 can be considered within a neutral “range” according to the Canadian MEND Program (2009). The M25 pH was neutral after 100 days at a proportion of 1:3 that was suggested by Machado and Schneider (2008) to neutralize sulfide coal tailings. The reduction potential (Eh) of AMP showed high values, above 500 mV, throughout the monitored period as demonstrated by Figure 3(b). These values are in agreement with Martins et al. (2004) and Machado and Schneider (2008). In addition, these data indicate a strong reduction potential and a presence of acidophilic bacteria. The M14 and M20 mixtures attenuated the oxidation process (Figure 3(c)). However, they remained in the range 400–500 mV. The M25 test reduced the values of Eh, and after 90 days this value was reduced to about 300 mV. M33 has reached similar values. The M33 mixture has showed Eh values between 200 and 300 mV since the beginning. These results demonstrate an immediate neutralization. In the first 10 days, the EC dropped in all columns containing AMP. Initially, these values were high, which showed elevated ion concentrations. After 10 days, the values were reduced and were subsequently stabilized in the range of 2000 mS/cm, indicating a polluted leachate. Other works showed stabilization around 2000 and 2500 mS/cm (Machado and Schneider, 2008; Moraes, 2010). The EC reduction may be explained by the absorption process proposed by Ekolu and Azene (2012) as a possible mechanism in the reduction of ion concentrations in the leachate. The highly alkaline mixture from the calcium oxides is negatively charged and the cations (mainly metals) in water are attracted and chemically absorbed onto the calcium oxides. The ES1 column maintained about 2000 mS cm-1 for almost all studied periods. This is due to calcium release. Finally, temperature measurements (Figure 3(d)) have remained close to the environmental temperature, even though the ARD reaction is exothermic.
Geotechnical results The geotechnical tests provided the results cited in Table 3. According to the Unified Soil Classification System (USCS) (American Society for Testing and Materials, 1985), the materials were classified as follows.
- ES1: well-graded sand (SW). - AMP: well-graded sand with clay (SW-SC). - M33: well-graded sand with clay (SW-SC). The soil–water characteristic curve is shown in Figure 4(a), in which the ES1 exhibits a distinct behavior compared with M33 and AMP. It is typically due to the difference in PSD. The ES1 is coarsegrained soil, while M33 and AMP present fines in their composition. Materials that present high moisture retention can be used as a capillary or oxygen inhibiting barrier (O’Kane et al., 2002). The graph of the unsaturated hydraulic conductivity versus matrix capillarity can also be seen in Figure 4(b). The ES1 shows high values of conductivity for small suction values and decreased values for higher suction values, this being typical for coarsegrained soils (O’Kane et al., 2002). In contrast, the AMP conductivity is smaller than ES1 for small values of matrix suction and increases for high levels of suction. This means that this material has clay in its PSD. The M33 unsaturated hydraulic conductivity has shown the smallest values in the beginning of the matrix suction. This may be explained by a clogging process caused by ferric hydroxide (“Yellow Boy”) precipitation.
Numerical modeling As M33 has efficiently neutralized ARD during the leach extraction tests, it was proposed to be the material for layer 2. Thus, to control its efficiency as a cover system, the water flow and oxygen ingress have been addressed at the bottom of this layer. The thickness of layer 2 and the climate boundary were varied for each interaction performed as observed in Figure 5(a). The climate boundary is based upon the precipitation related to the 1021 and 2200 mm m-2 per year (Oliveira, 2008). The results for other precipitation data in Ouro Preto are intermediate values. By analyzing the graph of maximum cumulative gas flux in Figure 5(a), the reduction in the gas flux (up to 1 m thick layer) is small; gas passes with little interference between the pores. In addition, the cumulative gas flux decreases significantly for the thickness between 1 and 3 m. For 5 m thick and more, the gas flux is very low. The maximum surface cumulative liquid flux is constant up to 1.60 m thick layer (Figure 5(b)). This is likely due to the two well-defined seasons. During the wet season, infiltration dominates; during the dry season, evaporation is predominant. This impacts the cumulative flux at the base of this layer. With thicknesses above 1.6 m, the cumulative liquid flux increases linearly. This layer contains more material; thus, it is impacted less by evaporation. During the wet season, this layer allows water infiltration, but, in the dry season, it maintains a high saturation degree due to moisture storage.
Downloaded from wmr.sagepub.com at UNIV OF PITTSBURGH on November 14, 2015
358
Waste Management & Research 33(4)
Figure 3. Leach extraction test results: (a) pH; (b) Eh (mV); (c) EC (electrical conductivity, µs/cm); (d) temperature (°C).
Downloaded from wmr.sagepub.com at UNIV OF PITTSBURGH on November 14, 2015
359
Pereira de Almeida et al. Table 3. Results obtained from geotechnical tests. Geotechnical results Material PSD Clay (φ< 0.002 mm): Silt (0.002
