Numerical Simulation of Geophysical Models to Detect Mining Tailings’ Leachates within Tailing Storage Facilities

By inergency On Mar 3, 2024

1. Introduction

Mining tailings are the residual materials remaining following the extraction of valuable minerals from ores. These tailings usually contain toxic chemicals such as arsenic, lead, and mercury, which pose a significant environmental threat if not managed properly [1]. Mining tailings can contaminate soil, water, and air, leading to health hazards for humans and wildlife. The management and disposal of mining tailings are essential to prevent their negative impact on the environment [2,3]. Some of the strategies for managing mining tailings include the construction of tailing dams, the backfilling of mined-out areas, and reprocessing to extract any remaining valuable minerals [4,5]. Despite these efforts, mining tailings remain a significant challenge for the mining industry, as the tailings produced can accumulate for decades or even centuries, with the potential to cause harm to the environment and communities surrounding the mine sites [6].

Geophysical techniques are extensively employed for the monitoring of groundwater contamination, which can arise from diverse sources, including hydrocarbon contamination [7,8], landfills [9,10], saline water intrusions [11,12], and mine tailings [13,14]. These techniques are cost-effective, repeatable, rapid, and non-destructive during data acquisition. Moreover, they provide an informative image of the subsurface petrophysical conditions. These geophysical surveying methods involve the use of the physical and petrophysical properties of the earth to identify the presence and movement of fluids within the ground [15]. The presence of hydrocarbon contaminants stimulates biological and chemical activity, resulting in bio physicochemical changes induced in the subsurface during the reaction of contaminants [16]. The increase in reactions in the surfaces of the substrate and between pore openings in rocks and sediments can induce physical changes in the porous material itself [17]. These reactions result in the production of acids which are responsible for the leaching of the surrounding soil matrix causing an increase in the ionic strength of the pore water. Hence, the dissolution of minerals results in the etching of the grains of the soil matrix [18]. These changes may cause an increase in the bulk electrical conductivity of contaminated soil that is detectable by geophysical methods such as Electrical Resistivity Tomography (ERT) and Ground-Penetrating Radar (GPR) [16,19,20]. These techniques play a crucial role in assessing and understanding the extent and nature of groundwater contamination, enabling effective mitigation strategies and resource management [21,22]. However, choosing the optimal geophysical method that meets the purpose of a project sometimes is challenging and may dictate the success or failure of the subsurface imaging [23,24].

The ERT method has been widely applied in groundwater, mining, environmental, and civil engineering investigations [25,26]. Mineral grains in the subsurface constitute the primary component of nonconductive soil and rock fragments [27,28]. However, in contaminated areas like mining sites, the presence of conductive metallic ores leads to variations in conductivity, creating anomalous zones [29]. As a result, the resistivity of soil and rock fragments can be readily differentiated from conductive zones. Other factors that may affect the resistivity include water in the pore spaces, water salinity, permeability, and porosity [30]. Resistivity measurements can detect anomalies in the subsurface and hence provide an image of the subsurface conditions [31]. Recently, there have been significant advances in automated ERT data acquisition approaches as well as 2D and 3D inversion software [32,33,34]. Thus, ERT has become a more attractive exploration method and a proven non-invasive and cost-effective technique. Many authors have demonstrated the conceivability to obtain accurate resistivity data of the subsurface structure using 2D/3D ERT inversion models [35,36,37,38,39,40,41].

As ERT is very sensitive to electrode array types [42], choosing the right array for each case study can have a substantial impact on the produced image [43,44]. Various numerical, experimental, and field studies have been conducted to compare different ERT electrode arrays for near surface geophysical settings and structures [45,46,47,48,49,50,51,52,53,54]. To our knowledge, there has not been a comprehensive comparison of ERT electrode arrays for mining tailings’ leachates (MTLs). Most of the previous work on MTLs using ERT employed a single electrode array without justification for why this array was employed. Therefore, this study aims to determine the appropriate electrode array that could be used to detect and monitor MTL. In this study, “leachate” pertains to the fluid that accumulates within tailing storage facilities (TSFs) subsequent to leaching through the tailing materials. To achieve this goal, a review of the published case studies in the literature that applied ERT to characterize mining tailings was performed to identify the most commonly used electrode array for such applications. Subsequently, a numerical investigation was conducted to evaluate the imaging capability of the commonly used electrode configurations. This evaluation was performed on a synthetic mining tailing backfilling model. A robust inversion approach was utilized to compare the configurations for 2D ERT, because it is a common inversion scheme for such projects. The following factors were evaluated for each electrode array: the imaging resolution, the depth of investigation (DOI), the data density, and the sensitivity to noise. Additionally, field surveys were conducted at the El Mochito mine tailings site to detect zones saturated with leachate within the tailing storage facilities (TSFs). This methodology is expected to help in selecting the most appropriate electrode array for mapping the MTL.

2. Literature Review

The use of ERT in characterizing mining tailings has become increasingly popular due to their accuracy and cost-effectiveness. In the section below, a review of published case studies that employed ERT to characterize mining tailings with emphasis on identifying the most commonly used electrode array for detecting MTLs within TSFs was conducted. Several criteria were considered while reviewing the published literature, including the following:

(1): Research focus: choosing the published literature that is closely aligned with the research focus and objectives of this study;
(2): Relevance: selecting articles that directly address similar or related research questions allows the authors to build upon existing knowledge and establish a coherent framework for their study;
(3): Methodological compatibility: limiting the selected literature that uses the same electrode arrays and employs a similar or complementary methodologies
(4): Limitations and scope: Because of limitations on space and the scope of the study, it is not feasible to include all the articles available on a particular topic. Thus, the authors had to make strategic choices to include representative studies that adequately cover the range of relevant perspectives and findings.

The Wenner-Schlumberger (WSC) array was successfully used during a 2D-ERT survey to characterize the environmental hazard of lead and zinc leaching of mine tailings at Frongoch Mine, Ceredigion, UK [55]. Another study was conducted in the Sierra Minera region, Spain, and employed the WSC array combined with soil chemical analyses. Their study helped in mapping the mine-tailing ponds [38]. Similarly, the WSC array was used to identify acid waste drainage from Zn-Pb post-flotation tailing ponds in Olkusz, Poland [14]. The WSC array was also used in the characterization of tailing dams [56], in the evaluation of tailing ponds in Linares, Southern Spain [57]. Meanwhile, the dipole–dipole array was used to assess waste materials in a mining tailing pond in El Gorguel, Spain [58]. ERT was employed to diagnose alteration channel penetration in uranium waste in Osamu Utsumi, Brazil, using the Schlumberger array [59]. The Schlumberger array was used for investigations and evaluation of environmental pollution in a mine tailings area in Komsomolsk, Russia [60]. Moreover, the most suitable spots for mine waste disposal were identified using the Wenner array [35]. Also, the Wenner array was utilized for the assessment of mine wastes in Regoufe, Portugal [61]. Copper mine tailings in Rio Tinto, Spain, were characterized utilizing the Wenner array [62], while the same array was used for undrained oil sands’ tailing ponds in Alberta, Canada [30]. For the characterization of water contamination due to metal mine waste in EsgairMwyn, Ceredigion, a Wenner-α electrode array was used [63]. WSC was employed to characterize abandoned sulfide mine-waste ponds in Iberian, SW Spain [64], and the same array was used for the assessment of the environmental threat of reclaimed mining tailing ponds [65]. The Schlumberger array was utilized for the delineation of subsurface structures in U-tailings in Jaduguda, India [66]. Table 1 summarizes some of the case studies where ERT was utilized to characterize mining tailings.

5. Discussion

The results of the modeling investigations indicate that the Wenner Alpha (W-α) array provides a relatively decent data resolution (data density) by measuring the apparent resistivity at various electrode spacings (i.e., a moderate data density compared to other electrode arrays). The top layer, representing a dry tailing region, has a thickness that is almost the same as the synthetic model (2.5 m) with a small difference in resistivity value. Similarly, the second layer represents a percolation/leaching (semi-saturated tailings) region of which the boundaries are well resolved in the inverse resistivity model. However, the third layer, which represents the saturated tailings region, has a relatively low data resolution due to the wide electrode spacing. The maximum model depth of this array has a moderate depth of investigation because the depth to which the Wenner-alpha array can investigate the subsurface is much smaller than the maximum AB separation (Figure 7, where the maximum AB separation of ~400 m corresponds to a maximum depth of ~56 m. Thus, for deeper investigations, a wider electrode spacing is required, which results in data resolution attenuation. Additionally, the data density of the W-α array is relatively low as it requires a relatively large number of electrodes. For sensitivity, the W-α array showed a good sensitivity to resistivity changes at shallower depths, which decreased at greater depths due to current dispersion as it travels deeper into the ground.

The Wenner Beta (W-β) array is a modification of the W-α with a shorter current electrode spacing. Despite the shorter electrode spacing, the W-β array produces a data resolution for the first layer nearly like the W-α array. However, for the other layers, the resolution is lower than that of the W-α array since the DOI of the W-β array is relatively shallow. The W-β array has the same data density as the Wenner array since both use four electrodes. Nevertheless, the W-β array offers a higher sensitivity to subsurface resistivity changes than the W-α array. This increased sensitivity is due to the short electrode spacing, which allows for a greater portion of the subsurface to be investigated, resulting in higher sensitivity to subsurface resistivity changes, especially in shallow applications.

The Wenner Gamma (W-γ) array is another modification of the W-α, where the current and potential electrodes are placed in an alternative order. This means that the electrode spacing is twice that of the W-α electrode spacing. The W-γ array provides a similar data resolution to the W-α, with the electrode spacing being the primary factor affecting data resolution. Smaller electrode spacings result in higher data resolution. The DOI of the W-γ array is better than the W-α, as the current is focused between the outermost electrodes, resulting in deeper penetration. While the data density is the same as the W-α, the sensitivity of the W-γ array is like the Wenner array. However, the higher DOI may result in lower sensitivity at shallower depths.

The dipole–dipole (DD) array is a commonly used configuration for ERT surveys. This array provides good data resolution, as the spacing between the electrodes can be adjusted to optimize data resolution for a particular target depth. The array provides a relatively high data density compared to Wenner arrays, as multiple measurements must be taken at each electrode location. However, this can increase the time and cost of the survey. The DD array is sensitive to both lateral and vertical changes in resistivity, making it a good choice for investigating complex subsurface structures.

The Schlumberger (Sch) array is a commonly used electrode configuration in ERT surveys for characterizing mine tailing dumps. It provides a relatively high data resolution compared to Wenner and DD arrays due to its sensitivity, which is important for detecting small-scale features in the subsurface. The Sch array gives the DOI better than Wenner arrays and can be adjusted by changing the electrode spacing, making it suitable for studying shallow subsurface layers. Additionally, the data density is higher also than that of the Wenner and DD arrays and can be controlled by adjusting the number of electrode measurements, allowing for higher-resolution data in areas of interest. The Sch array is also sensitive to changes in subsurface resistivity, making it useful for detecting subtle variations in the MTL subsurface. The Schlumberger array produced high-quality data with relatively low RMS errors.

The Wenner-Schlumberger (WSC) array is a modification of the Wenner array in which the current electrodes are kept at the same spacing, but the potential electrodes are spaced closer together to increase the sensitivity of the method. The WSC array provides a higher data resolution compared to Wenner and DD arrays due to the increased sensitivity achieved by spacing the potential electrodes closer together. This allows for a better delineation of near-surface features. Moreover, the DOI of the WSC array is higher compared to other arrays such as Wenner, and the data density of the WSC array is higher also than that of the Wenner and DD arrays. Additionally, the increased sensitivity of the WSC array results in a higher-resolution dataset due to the closer spacing of the potential electrodes. This allows for a better delineation of near-surface features and can help to identify subtle changes in subsurface MTLs. Therefore, the WSC array is useful in situations where a high resolution is required at shallow depths, making it a popular choice for environmental and engineering applications. The WSC array, while effective, does come with certain drawbacks. It requires a larger surface area for cable deployment due to the substantial distance between current electrodes. Consequently, surveying expansive areas for MTL plumes can be cumbersome, particularly in access-limited regions. Moreover, the slower acquisition of the WSC array can be time-consuming, which may pose challenges for the dense data collection often required for precise MTL plume delineation.

In addition to the results and analyses obtained from synthetic modeling MTL data, we conducted a field survey of an old mining-tailing pond in Honduras to validate these findings. The inversion results of the field survey are depicted in Figure 7. It is noteworthy that the survey utilized the W-α array for practical reasons, such as the ease and speed of use, as well as calculations. Moreover, this array is widely employed in previous studies for similar applications, as indicated in Table 2. Additionally, the choice aligns with the principle that obtaining satisfactory results with a less efficient array suggests better outcomes with a more efficient one.

Upon the analysis and interpretation of these resistivity profiles in Figure 7, a consistent and distinguishable layering pattern became apparent across all surveyed profiles which is practically consistent with the results of the synthetic modeling for the MTL. The profiles reveal the presence of three geoelectric layers, each with varying thickness and depth. This observation is summarized in Table 4. The uppermost layer (A) is characterized by relatively high resistivity values ranging from 60 to 100 Ωm and is located approximately 2.5 m below the surface. Layer A is interpreted as a dry tailing cover, primarily composed of solid tailing/waste materials. It is worth noting that the bottom boundary of this layer exhibited uniformity in most profiles.

The second geo-layer (B) is sited beneath the dry covering layer for tailings, typically at depths of 3–10 m below the surface. It demonstrates the resistivity values of relatively moderate magnitudes, 30 to 60 Ωm, and exhibits a thickness that varies between approximately 10 and 20 m. This layer is understood to be a partially saturated zone for the tailings. The third geo-layer (C) represents the lowest geoelectric layer in all the resistivity profiles. It displays resistivity values that are very low, measuring less than 30 Ωm. The transition from B to C is noteworthy, as this transition exhibits an irregular boundary.

The utilization of the Wenner array in our field survey of an old mining tailing pond in Honduras presented both advantages and disadvantages. The practical advantages of choosing the W-α array were evident in its ease and speed of use, making it a convenient choice for efficient data collection in challenging field conditions. This array’s widespread application in similar studies, as reflected in Table 2, added credibility to our selection. However, it is important to acknowledge the trade-offs associated with the Wenner array. While it provides practical benefits, its efficiency may be compromised when compared to other arrays, such as WSC or DD. The principle guiding our choice suggests that satisfactory results obtained with the Wenner array indicate the potential for even better outcomes with a more efficient array. The resistivity profiles resulting from the survey, depicted in Figure 7 and summarized in Table 4, aligned well with the synthetic modeling results for MTL. The consistent layering pattern observed in the field matched the synthetic profiles, revealing three geoelectric layers with varying resistivity values and thicknesses. This alignment strengthens our confidence in the validity of our field survey results and the applicability of the chosen Wenner array in characterizing the subsurface structure of the tailings pond.

To contextualize the findings within the existing literature, it is evident that each electrode array in our study exhibits distinct advantages and limitations during experimental and fieldwork endeavors. Our discussion focuses on the data resolution capabilities of the W-α array across various tailings layers, acknowledging its diminished resolution at greater depths attributable to wider electrode spacing. Notably, prior investigations [51,59], including those by Olayinka and Yaramanci [45], have corroborated similar trends regarding data resolution concerning electrode spacing and DOI. While Olayinka and Yaramanci [45] observed lesser noise contamination in the W-α array when compared to other configurations in different geological settings, they noted that the W-β and W-γ arrays exhibited lower noise contamination during imaging surveys, albeit with inconsistent anomaly effects and signal-to-noise ratios, as reported by the same authors. The DD array, characterized by relatively high anomaly effects, is susceptible to noise contamination, resulting in lower signal-to-noise ratios compared to Wenner arrays, a trend also observed in prior studies [18,45]. Both the DD and W-β arrays, boasting symmetric electrode configurations for normal and reciprocal measurements, facilitate robust data quality control to yield well-resolved images. Despite the DD array’s superior imaging resolution, particularly for vertical and dipping structures, its depth resolution may not be optimal [45].

In comparison, the Sch array, while demonstrating imaging abilities akin to the W-α array due to similarities in electric field and measurements, excels in DOI relative to other arrays, as highlighted in previous research [59,66]. However, challenges persist, including higher noise contamination and fewer signal-to-noise ratios compared to the W-α array [45,60], rendering the Sch array less suitable for multichannel applications without a reciprocal array. Research conducted by various authors [14,65] aligns with our study’s findings, indicating that the WSC array achieves superior data resolution, particularly for near-surface features, while also enabling deeper plume detection compared to traditional Wenner arrays [56,57,64], attributed to its heightened sensitivity arising from closer potential electrode spacing.

6. Conclusions

The study utilizes and compares the ERT numerical simulation for six different electrode arrays to find the optimal array for detecting MTL. The comparison considered various aspects including data resolution, depth of investigation (DOI), data density, and sensitivity. The 2D resistivity inversion results indicated that each electrode configuration has its advantages and disadvantages. It also highlighted that the choice of the optimal array is dictated by the desired goal/s such as the target depth, required data resolution, and data density.

The results of the modeling investigations indicate that the Wenner Alpha array provides relatively decent data resolution (data density) by measuring the apparent resistivity at various electrode spacings (i.e., a moderate data density compared to other electrode arrays). The Wenner Beta array offers higher data resolution for the first layer than Wenner Alpha array due to the shorter electrode spacing, but its DOI is relatively shallow. The Wenner Gamma array provides similar data resolution to the Wenner array, but with better DOI. The dipole–dipole array is commonly used and provides good data resolution and a relatively high data density, but it has a shallower DOI compared to the Wenner arrays. The Sch array provides a better data resolution than the Wenner arrays, with a relatively deep DOI and higher data density; however, it is very time-consuming and very costly to use. The WSC array provides a higher data resolution compared to the Wenner and DD arrays due to the increased sensitivity.

Finally, the choice of electrode configuration depends on the specific goals and conditions of the survey. The Wenner Alpha array may be suitable for shallow investigations, while the dipole–dipole array may be preferred for complex subsurface structures. The WSC array may be useful for high-resolution data. Consequently, the WSC is highly recommended for MTL detection using ERT surveys. The MTL models developed in this research were tested and validated through real field surveys. In the field surveys, the identification of three distinct geoelectric layers, a dry tailing cover, semi-saturated zone, and saturated layer, provides a nuanced understanding of the subsurface composition at the El Mochito mine-tailing pond. Ultimately, this study significantly contributes to the detection of MTL and its subsurface characteristics through a combined approach of synthetic modelling and field surveying.