Abstract

Mine waste dump material has no economic value to the industry. Therefore, the mine waste is dumped, forming slopes. Mine waste dump slopes obtain 30% to 50% of the mining area. To reduce the land occupancy of these slopes, they are created with high altitudes. Hence, they are susceptible to failure. Slope stability analysis is a major aspect of geotechnical engineering. Slope stability analyses are mostly done with assumptions on the geometry. This is avoided in this paper with the usage of UAVs. The 3D model is created from UAV imagery of a coal mine in Raniganj coalfield, India. The model is fine-tuned with the DGPS survey. Geotechnical data were collected and tested in the laboratory for various numerical analyses. An active mine waste dump slope is analyzed for slope failure. Earthquakes and rainfall cannot be controlled, and their effects on the stability of the mine waste dump slope were examined. The study extracted various factor of safety (FOS) analyses on static, seismic, and rainfall conditions. The seismic condition simulates a condition of the slope to be failed with low (0.948) FOS. However, rainfall condition predicts the slope to be more stable. The deformation pattern and magnitude of the slope failure are also discussed.

1. Introduction

Open-pit mining operation excavates the top portion of the soil to extract minerals or coals. During this process, the excavated soil had no economic value to the mining industry. These materials are known as waste [1, 2]. The limitation of the mining area makes it tedious to manage this waste. Therefore, the mine waste is dumped, forming a slope [3, 4]. These dump slopes are enormous. Mine waste dump is made very rapidly, therefore less time for consolidation of waste materials. This hasty approach is making the slopes susceptible to form disasters. From 1921, there were 23 accidents due to mine overburden slope failures only in India. These accidents cost 143 causalities [5, 6]. In 2016, a catastrophic slope failure in Rajmahal opencast mine caused the death of 23 people. Talcher, Bokaro, and Bharatpur recently faced similar accidents [79]. Mine dump slope stability analysis is a major subject in geotechnical engineering. The safety study is based on these slopes with SSR (Shear Strength Reduction). In this technique, soil material’s strength characteristics are reduced till the failure is obtained in the geometry [10]. This method provides a numeric value that represents the factor of safety [11]. If the value is greater than one, then the stability of the slope is high and vice-versa [1215]. For traditional numerical simulation, an assumption on geometry is considered [16]. With the usage of UAV, two significant problems in numerical simulations can be solved, the first reduction in geometry assumption, and the second, the GIS tools can be used for mapping, monitoring, and targeting the risk-prone areas for numerical simulation resulting in accurate analysis and less computational time. This approach is possible with UAV’s RGB (red, green, and blue) imagery. Overlapped images of UAVs can create a 3D point cloud of an area [1719]. Furthermore, DTM for slope targeting for numerical simulation and geometry for numerical simulation can be developed. For testing this approach, Raniganj coalfield, India, has been selected (Figure 1 represents the study area). Raniganj is situated in the eastern part of India. This coalfield region is about 1130 km2 from 23°25′N to 23°50′N latitude and 86°38′E to 87°20′E longitude [20].

Raniganj coal mines have most of the open cast projects of Eastern Coalfield Ltd., India. In Raniganj coalfield, the biggest open cast mine is situated in Sonepure Bazari [2]. The production of the coalfield is 33.90 million tonnes; thus, high mining activity occurs consistently. The mine we selected has an in-pit waste dumping system, i.e., waste is dumped within the mining area due to the limited range of available land. In-pit waste dump of the study area is shown in Figure 2.

An active mine dumping area was taken for testing for stability analysis, as the chances of failure will be high due to less time for waste to consolidate [12]. Hence, traditional numerical simulation is insufficient in this scenario as rainfall plays a vital role in slope failures and earthquakes [21]. As Raniganj is a subtropical place, heavy rain occurs, and precipitation of 200 to 400 mm every monsoon has been reported yearly. Also, regular cyclones have been impeding [22]. Therefore, an increase in rain directly affects the safety factor of the overburden slope [23].

2. Materials and Methods

This study focuses on the stability analysis of the active in-pit mine waste dump. Therefore, in data acquisition, UAV imagery, DGPS survey, and geotechnical samples were collected from the study area. With the help of UAV imagery,

3D modeling and proper geometry are extracted from the active waste dump slope. The area was cross-checked with GIS to ensure the proneness to failure. Finally, three geometry of the section was extracted, and numerical simulations were done with the help of geotechnical data. In Figure 3, the steps can be followed. There are four steps to achieving the desired goal; data collection, geotechnical analysis, image processing, and numerical simulation. The numerical simulation focuses on rainfall effects on the stability of the slope and the simulation of seismological impact on the slope. These parts are discussed in this section.

2.1. UAV Data Acquisition

In the field, DJI Phantom 4 Pro UAV has been used (Figure 4 shows the UAV). The UAV can fly for up to 19 minutes on a single charge and capture 20-megapixel resolution worth of imagery. Acquisition of images is made with overlapping of 75% in horizontal and 65% in vertical. DGPS (differential global positioning system) was also used for the accuracy of the generated model [2426].

The average GSD of the constructed orthoimagery was 5.022 cm per pixel. A total of 561 images were captured with two flights. And the flight plan was grid in nature. The area captured by the UAV is 4716900 m2. DGPS can extract coordinates of an unknown area without reference. Prominent “+” symbols were drawn in the field with whitewash for this study, as shown in Figure 5. Seven GCPs were used to process the images in Pix4DMapper. As per the software Pix4DMapper with seven GCPs, the RMS should be less than 0.10, making the model highly accurate.

The overlapping helps in recovering lost 3D information of the camera with the same feature extraction in multiple images, and triangulation is done to retrieve the position of the camera in x, y, and z coordinates by reversing the approach after extracting camera location, x, y, and z coordinates of features are extracted. This approach is made by structure from motion pipeline for generating 3D point cloud and other forms of data [27, 28].

2.2. Geotechnical Analysis

Accurate shear strength parameters, i.e., cohesion and internal frictional angle of the waste materials, are necessary for numerical simulation accuracy [29, 30], along with those compaction characteristics that are also needed. For shear strength parameters, triaxial tests were performed [31], and compaction tests were carried out according to Indian Standard (IS:2720) part-6 (1965). Hence, a geotechnical investigation was necessary. The materials were initially tested with grain size distribution. According to Indian Standard Soil Classification System (ISSCS) with various tests, the material showed coarse materials between 80% and 98%, and sand between 2 mm and 0.425 mm was ranging from 22% to 92% and, sand and clay ranging from 3.8% to 46% (particles are less than 0.75 mm) thus making the material poorly graded sand or SP. The samples were collected during UAV data acquisition and have been processed in the institute’s laboratory facilities, and the extracted geotechnical parameters can be seen in Table 1.

2.3. 3D Model of UAV Imagery

3D modeling from UAV imagery is based on epipolar geometry of photogrammetry [28, 32]. Therefore, feature extraction and matching of multiple images are done. Matched features of image pairs are used to create pose estimation (3D location of the camera of the same feature found in various images) with triangulation hence, spatial 3D point cloud data [33, 34]. To minimize errors, bundle adjustment is applied. Bundle adjustment is used on the accumulated 3D model to refine and prevent noise in triangulation [35]. This model is created from a UAV. This 3D model, along with an orthomosaicked image, was used to create a digital elevation model (DEM) for inspection of AOI.

Furthermore, to be accurate, the geometry DGPS surveyed data were integrated before processing the 3D point cloud as ground control points in Pix4DMapper. Figure 6 shows the 3D model of the study area, which is generated from UAV imagery.

2.4. Geometry Extraction for Numerical Analysis

The surveyed area from the UAV mainly covered the mining dump, but most of the area was not visible from the ground survey or observation. So, removing fewer risk-prone zones from the surveyed area helps reduce computational time. The authors targeted an active waste dump slope to be examined. A newly made dump has less time to settle down, and having loose materials causes low shear strength. Therefore, failure can occur.

In Figure 7, the active dumping zone is shown. This area initially needs to be examined with GIS. The GIS technology can cross-check whether the active dumping slope is prone to failure. This is done by developing a slope map of the study area from UAV data. Slope angles represent the relative angle displacements. In Figure 8, the slope map is shown.

From Figure 8, it can be seen that the angular value of the targeted zone exceeds 36 degrees. Thus, this site makes it ideal for examining numerical simulation for slope stability.

Three sections from the active dumping zone have been extracted. These geometries can be seen in Figures 9(b)9(d). The geometries from the same dumping slope have different angles and heights. Examining only one will derive less in-depth analysis. These sections have a length of 207, 210, and 198 meters horizontally and 66, 83, and 101 meters in altitude displacement (the projection in the figure is in 3D; hence, the perspective of displacement on different axis differs).

3. Results and Discussion

Flac comes with computing packages for numerical simulation applications. This is widely accepted and recognized. In Flac3D, numerical analysis is done with the shear strength reduction technique or SSR. Constraint boundaries restrict developed geometries. This approach improves accuracy, efficiency, and modeling time, ensuring accuracy and legitimacy [36, 37]. Three geometries were extracted from the targeted waste dump slope. These geometries range between 66 meters and 101 meters in height and 197 meters to 210 meters horizontally. Mohr-Coulomb’s failure criteria were adopted in the first iteration of stability analysis. The failure occurs with the critical combination of the shear strength of loose mine waste dump materials with normal stress. The loose, broken waste dump materials contain low shear strength [38, 39]. The dump slope was numerically analyzed with finite element method codes, and the FOS was calculated with displacement values, as shown in Table 2.

The slope seems to be stable with low displacement values in all the scenarios [38]. Furthermore, the sliding portion of the geometries can be seen in Figure 10. All the failures are circular. The FOS of the sections ranged from 1.26 to 1.53 of the slope under static conditions.

As discussed earlier, static slope stability analysis is insufficient as two significant reasons slope failures occur due to heavy rainfall and seismological effects. These two factors cannot be dealt with human intervention. Therefore, examining these two factors will ensure a more nuanced analysis on analysis.

3.1. Pseudostatic Condition for Seismic Effect

India is divided into four zones for an earthquake. The study area of this paper has zone III in terms of earthquake hazards. According to Figure 11, the study area lies in the moderate damage risk zone, and the factor is 0.16.

A seismic load is given in the numerical simulation of the seismological effect, i.e., 0.16. Therefore, the integrity of the analysis of the limit equilibrium method [41] with seismic conditions on the geometries is ensured. The extracted FOS from the simulation validated a lower stability factor than the static condition. In Table 3 no. 3 FOS of three sections can be seen.

The seismological effect also extracted similar failure patterns as a static condition. The LEM method provides only a stability factor. Therefore, stress and strain information is not achievable, resulting in nondeformation analysis in pseudostatic conditions. The sections after numerical simulation can be seen in Figure 12.

This section validates the analysis methodology as to its capability to calculate the slope failure surface with seismic acceleration. The simulated result is documented with zone III seismic acceleration (Figure 11), which has been analyzed through widely recognized conventional Bishop’s limit equilibrium methods [42]. The sections were further analyzed with a 2, 4, 6, and 8 seconds dynamic seismic load.

The el Centro dataset is adopted for input load. The given input data can be seen in Figure 13. This data propagates in the N-S direction. The pick of the acceleration is 0.3 g recorded around 2 seconds, and after 5 seconds, the acceleration gradually decreases with a low spike at 8 seconds and in around 25 seconds with 0.138 g. The displacement of the sections with dynamic load can be seen in Figure 14. In section 3 (Figure 13(e)), the displacement initially started with negative and then moved towards positive, which strongly proposes crack generation in the upper part of section 3 [43].

The simulation shows that the FOS of section 3 has a value of 0.953. This makes the section prone to failure. As for sections 1 and 2, they are shown a dip in FOS compared to numerical simulation in static conditions with 1.219 and 1.048.

3.2. Dump Slope Stability Analysis with Average Rainfall

The study area is subtropical in climatic nature. Hence, the rainfall occurs heavily after summer, i.e., from June to September [4446]. Mine waste dump failures occur primarily in this period due to heavy precipitation. The precipitation in the study area was highest in August. In August, the minimum rainfall was 207.3 millimeters recorded in 2017, and the highest was in 2016 with 601.3 millimeters. In other monsoon months, precipitation varies from 28 millimeters to 515 millimeters. The average rainfall in the monsoon of the study area has been recorded as 264.266 mm in the last ten years. The numerical analysis of the study area with precipitation data was simulated for seven months. Extracted FOS can be seen in Table 4. The FOS in this scenario improved [47].

The improvement in FOS is explainable by two factors. First, seven months for settlement of the slope which is after the monsoon. The second cause is related to the waste dump material characteristic. The material contains clay that has high adhesiveness when mixed with water, which vastly improves the cohesion that benefits the bonding and stability [48]. The saturation profile of the simulation can be seen in Figure 15. This phenomenon was derived from the theory of consolidation, in which precipitation decreases the pour spaces of sand and clay-type materials and improves the consolidation with time.

Deformation of the sections stayed almost the same with low movements of the slope materials, which maxed at 1.098 meters at section 3. The displacement profile of the sections can be seen in Figure 16. The displacement on the slopes with rainfall causes less than 1 meter of negative deformation. In sections 1 and 2, the displacement goes up to 0.05 meters and 0.16 meters, although insection 3 geometry, the deformation goes up to 1 meter.

As a future study, rigorous analyses can be done with various FORM algorithms, and these algorithms were derived from HLRF algorithms. HLRF and FORM algorithms are known for their reliability [49, 50]. Therefore, a basic comparison of the stability of waste dump slopes can be compared with various material properties [51].

4. Conclusions

The availability of the UAV and improvement on the image sensor is useful for creating 3D maps and models of the mining area. Proper scale and dimensions are extracted with the help of UAV imagery, DGPS, and photogrammetry. The geometries extracted from the UAV’s 3D model represent an active in-pit mine dump slope, which is examined based on less time for the slope to settle waste materials. A geotechnical investigation was carried out on the dump waste materials. The geotechnical properties are instrumental in numerical analysis reliability. Numerical simulations on these geometries are carried out in three different approaches, making the study cohesive. The environments for the numerical simulation consisted of static, seismic, and rainfall conditions. These simulations show the FOS value initially at a stable state, and rainfall with seven months of simulation time increased the stability. However, the seismic condition in Bishop’s limit equilibrium method shows significantly lower FOS of the sections. The least FOS was 0.948 achieved in section 3 geometry after seismic load. This outcome illustrates the failure in section 3.

Nevertheless, regular precipitation and time (or settlement of the loose materials) suggest a higher FOS, so the waste dump slope will be stable if no significant seismic acceleration occurs. The displacement analysis also indicates a maximum of 1-meter deformation. Although with seismic activity, cracks will be formed in section 3.

Data Availability

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

Conflicts of Interest

The authors declare no conflicts of interest.

Acknowledgments

The research is funded by the Science and Engineering Research Board (SERB), New Delhi (Grant No - CRG/2020/005919).