1 Introduction

Ore resource excavation developed multiple types of artificial slopes, such as the step-shape mine slope, waste dump slope and tailing pond (Dawson et al. 1998; Ulusay et al. 2001; Agurto-Detzel et al. 2016). The progressive exposure of geological structures, including folds (Cui et al. 2018), faults (Zhu et al. 2020), and discontinuities (Tang et al. 2021) and weak interlayers (Xue et al 2018; Tang et al. 2020; Chen et al. 2021), influences the mine slope stability after deep excavation (Benko and Stead 1998; Stead and Wolter 2015; Donati et al. 2021). The landslide occurrence by excavation is related to multiple factors (Stead and Wolter 2015), including the inherent lithology and structure (Stead and Eberhardt 2013), basal sliding surfaces structure (Zheng et al. 2018) and unloading disturbance by mining (Yu et al. 2020; Zhang et al. 2021), as also as the build-up of pore water pressure associated with rainfall (Cogan et al. 2018). But the mechanism of slope instability is usually contributed to one of the weak rock structure, which promotes slope deformation under external anthropogenic activities (Sabino 2014; Igwe 2015; Li et al. 2017; Ma et al. 2018). For example, utilization the fault as the major slip surface (Hart et al. 2012), influenced the 1963 Vaiont slide (Semenza and Ghirotti 2000). Identifying the controlling structural factors is important for evaluating slope stability.

From the perspective of slope structure, it is believed that anti-dipping rock mass has better stability (Xu et al. 2016). The alternated interlayers of mudstone and sandstone form the red lays in Sichuan Basin (Xu et al. 2016), which is well-known by the widespread bedding-controlled landslides (Bromhead 2013; Hu et al. 2018). At a mudstone mine used for cement additives, its initial design suggested the anti-dipping mudstone slope is generally stable. Since mineral resources are mostly found in complex geological bodies, periodic excavation will continue to reveal neglected geological structures in the depths (Ulusay et al. 2001; Agurto-Detzel et al. 2016). The resulting potential geological disasters not only threaten the production safety of the mining platform, but also restrict the sustainability of mining excavation. With further exposure of geological structure in mining process, it is of great significance to understand the control of different inducements on slope failure mechanisms for the prevention and treatment of mine slope instability.

In this case, the geological characteristics controlled by fault, the presence of weak interlayers and the occurrence of rainfall in a mudstone mine slope are presented on the slope instability analysis. A combination of field investigation, unmanned aerial vehicle, monitoring data interpretation, and numerical simulations is used in this study to investigate a mining-induced rockslide in the Sichuan Basin, China. This study provides a valuable case study of a mine slope whose stability is controlled by the exposure of fault.

2 Engineering and geological situation

2.1 Overview of Baiyian mudstone mine development

The Baiyian mudstone mine (E 103°38′33″, N 31°02′15″) is located in Jinfeng village, Dujiangyan city, Sichuan Province, China, as shown in Fig. 1. The Baiyian mudstone mine was constructed in 2001 and has been excavated since 2002. The ore is shoveled by excavators through stepwise mining from top-to-bottom in horizontal working platforms, without the need for blasting. The designed mining parameters are: (1) Production stage height: 5 m; (2) Working step slope angle: 70°; (3) Mining elevation: 890–722 m; (4) Safety platform: 5 m; (5) End slope angle: 24°–41°. Between 2005 and 2018, the mining slope underwent several local deformations during the mining process, ultimately leading to the formation of two large-scale landslides during the rainy season of 2018, namely the Landslide #I and Landslide #II landslides (Fig. 1c). By November 2018, when the landslide #I occurred, the mudstone mine had been mined to an elevation of 787 m, with a maximum slope height of 130 m.

Fig. 1
figure 1

(a) and (b) Location and regional tectonics of the open mine; tectonic and topography map (Qh and Qp: Quaternary deposit; E1r: Relu formation; K2g: Guankou fomation; J1z: Ziliujing formation; J3p: Penglaizhen formation; J2-3s: Shaximiao formation; T3xj2: Xujiahe formation; P2m: Maokou formation; Tγ, P3β, C1c, Dph, Pt1v, Pt2γ, Pt2δ: Igneous rock); (c) and (d) Spatial correlation of the two landslides and their 3D view

Full size image

Although multiple geological hazard surveys and analyses have been conducted in the past 12 years, the deformation mechanism of the Baiyian mine slope has never been explored. The governance measures carried out in the early stage only played a certain role in the short term, but failed to effectively prevent the occurrence of two landslides. Therefore, this article relies on detailed on-site investigation, including field investigations, drilling exploration and unmanned aerial vehicle surveys, to further identify the inherent mechanism of slope instability in mines, from the aspect of geological evolution process.

2.2 Geological setting of the open mine

The study slope, tending southwest, is located at the junction of the basin and mountain area densely distributed tectonic structures (Figs. 1, 2 and 3). Regionally, Baiyian Mine is located in the Guankou-Anxian fault zone, only 50 km away from the epicenter of the 2008 Wenchuan earthquake. In terms of topography, the mine is located in the steep drop zone between the Qinghai-Tibet Plateau and the Sichuan Basin. Therefore, the south-eastern side of the mine faces a stable and extremely thick piedmont clastic accumulation, backed by a structural and geomorphological activity area composed of sedimentary rocks, igneous rocks and fault. The strata exposed in the slope field (Fig. 1 b) are the middle section of the Triassic Xujiahe formation (T3xj2), Quaternary artificial fill (Q4ml) and Quaternary slope deposit (Q4el+dl). According to field surveys, the exposed bedrock in the local area is mainly carbonaceous mudstone, with interlayer of silty mudstone.

Fig. 2
figure 2

Rainfall history before the occurrence of landslide #I

Full size image
Fig. 3
figure 3

(a) Overview of Landslide #I and Landslide #II; (b) Tree skew caused by landslide. (c) Tension cracks at the trailing edge of the landslide; (d) Rock bending due to the fault; (e) Tension scarp at the trailing edge of the landslide; (f) Right leading edge of the landslide shear outlet; (g) Deposits on the right leading of the landslide; (h) A large boulder in the accumulation area

Full size image

A fault with a dip direction of 319° ~ 343° and a dip of 36° ~ 47° runs across the toe of the stepped slope, resulting in a difference in slope structure, upper wall dipping inward, and the footwall dipping outward (Fig. 1d). The study area has a subtropical monsoon climate, with an average annual rainfall of 1200 mm, mainly occurring from June to August. The rainfall data for the three months prior to Landslide #I were obtained from a local rainwater monitoring station (Fig. 2). Groundwater in the landslide area mainly exists as pore water in the loose rock mass, including landslide rock residual fracture and bedrock fracture. The simple water pump tests of three boreholes indicated that the permeability coefficient of the bedrock was in the range of 0.102 to 0.217 m/day, indicating that a low permeability for the rock mass.

2.3 Rock mass structure of mine slope

Previous surveys mentioned the existence of the F1 fault and found differences in stratigraphic occurrence, but did not reveal the location of fault fracture zone. During our field investigation, the mining depth has reached 787 m which allows us to identify the characteristics of F1 fault by comparing the occurrence of the upper and lower walls and tracing the extension of the surface fracture zone. The F1 fault extends from the 820 m platform on the west side of the current slope to 805 m on the east side, extending over 800 m in the occurrence of 319º ~ 343º∠36º ~ 47º. The fault zone is composed of mudstone debris, with a small amount of coal dust. And obvious folds and bends can be seen on both sides of the fault.

The slope was comprised of relatively fractured rock masses, principally the carbonaceous mudstone intercalated with silty mudstone. The footwall strata of the fault dip gently cataclinal to the slope, with a trend of 170°–190° and a dip angle of 12°-18°. The hanging wall strata of the fault dip toward the interior of the slope, with a trend of 260°–280° and a dip angle of 50°–55°. In the footwall, several weak interlayers are exposed at the leading edge and interlaid in the bedded strata according to the drilling core of ZK-04 and ZK-05 (Fig. 6). The gray-black weak interlayer contains a small amount of carbonaceous mudstone fragments with a thick of 5 cm ~ 30 cm, showing a liquid limit 44.6% and plastic limit 14.2%.

3 Landslide formation in the mining process

In March 2005, tension cracks first appeared on the mining slope. And more cracks emerged on the 915 m platform influenced by the Wenchuan earthquake. Then, slope cutting, load alleviation and a drainage system were constructed above the 867 m platform, effectively controlling the deformation of the cracks. As the increasing mining depth and free face, tension cracks were firstly observed in the central slope surface after August 2015. Since the rainy season of 2017, deep excavation and rainfall precipitation have led to a renewed reduction in slope stability, accompanied by a continuous increase in cracks that traverse the entire slope. Following a prolonged period of creep deformation, a substantial rockslide (landslide #II) moved below the 860 m platform in the central slope. Subsequently, the adjacent landslide #I slid to the 790 m platform in November 2018, with plenty of loose deposits. The two landslides threaten the safety of subsequent mining.

3.1 Characteristics of landslide #I

Landslide #I occurred in November 2018 and exhibited a continuous creeping movement, covering a distance exceeding 100 m over the course of a week. The affected area was delineated into two distinct sections: the source area and the accumulation area (Fig. 4). The morphological characteristic parameters and corresponding values are detailed in Table 1. Landslide #I manifests a prolonged tongue shape, indicative of a broad and gentle accumulation body at the leading edge, juxtaposed with a narrow and steep source area. The latter comprises multiple scarps and relatively gentle steps, featuring a sidewall approximately 5–10 m in height and presenting a lower topography than the original steps on both flanks. Notably, the height of the accumulation area surpasses that of the original steps, giving rise to an inclined scarp along the landslide boundary (Fig. 3a).

Fig. 4
figure 4

Geological profile (A1-A2) of Landslide #I. ZK is the borehole after landslide occurrence, and SK is the ore exploration borehole before mining

Full size image
Table 1 Morphological characteristic parameters and values of Landslide #I
Full size table

Based on the findings obtained through drilling exploration and field investigation, it has been ascertained that the slip source area comprises Quaternary deposits and carbonaceous mudstone. The Quaternary deposits consist of yellow silty clays containing crushed rock fragments, characterized by a substantial moisture content. The crushed coarse-grained rock fragments predominantly represent highly weathered yellow sandstone, with particle diameters ranging from 3 to 20 cm, constituting approximately 30% to 60% of the overall deposit. On the shear outlet platform (Fig. 3i), a significant quantity of sandstone fragments, with diameters ranging from 0.3 to 5 m (Fig. 3h), is present, with the largest fragment measuring 10 m in length. Furthermore, the accumulation body of Landslide #I extends over Landslide #II (as illustrated in Fig. 3i). The surface material of Landslide #II primarily consists of strongly weathered carbonaceous mudstone, interspersed with crushed rocks ranging in size from 0.5 to 5 cm. Notably, the accumulation body forms a scarp that inclines outward from the slope, being 5–10 m higher than the original mining step.

After the initiation of Landslide #I, the trailing edge and sidewalls became exposed, revealing a smooth sliding surface with a dip direction of 14° and a dip angle of 70°. The sliding wall is primarily composed of silty clay, interspersed with sections of strongly weathered sandstone. Individual blocks within the sliding wall measure approximately 50 cm in length. Notably, as one descends deeper into the landslide body, the proportion of rock blocks increases. On the east side, the sliding surface reaches a height of approximately 3–7 m, with the lower portion consisting of fully weathered black carbonaceous mudstone and silty clay. Conversely, the sliding surface on the west side exhibits a dip direction of 70°, a dip angle of 75°, and a height ranging from 2 to 8 m. The material composition of the west sidewall aligns with that of the east sidewall. Landslide #I is categorized as a bedrock landslide, and the drilling results provide clear insights into the condition of the rock mass near the sliding zone. At the ZK-01 site, the exposed sliding zone extends to a depth of approximately 9.5 m and is predominantly composed of strongly weathered sandstone gravel. At the ZK-02 site, the exposed sliding zone extends to a depth of around 10 m and consists of strongly weathered carbonaceous mudstone. The rock mass in proximity to this depth exhibits significant fragmentation, with particle sizes ranging from approximately 2 to 8 cm, and it locally contains argillaceous components. Throughout the drilling process, substantial water leakage was observed, while the rock mass above the sliding zone remained relatively intact.

Based on the available drilling data and survey information, the upper portion of the landslide's sliding bed comprises moderately to strongly weathered layered sandstone and carbonaceous mudstone. In contrast, the lower portion consists of moderately to strongly weathered thinly layered sandstone and carbonaceous mudstone. These interbedded layers exhibit a dip direction ranging from 290° to 330° and a dip angle ranging from 20° to 30°, with a downward inclination toward the slope. The carbonaceous mudstone, characterized by a gray-black coloration, possesses an argillaceous structure. The moderately weathered rock mass maintains a relatively intact structure, with an exposed thickness exceeding 5 m and including a localized coal seam. The carbonaceous mudstone dominates the slope's lithology, exhibiting low water permeability and acting as a barrier to water flow. In contrast, the yellow and coarse-grained sandstone, exposed to a thickness of approximately 7 m, has undergone strong weathering, resulting in significant fragmentation and reduced water permeability. Considering the underlying bedrock, Landslide #I can be characterized as a layered slope structure, with the rock layers dipping towards the slope.

The field investigation has uncovered multiple steps and staggered scarps on the slope's surface. These scarps range in height from 0.5 to 2 m, following a trend of 60° and a dip angle of 55° (Fig. 3f). The spacing between these scarps measures approximately 10 m. In addition, numerous cracks have emerged in the mining steps and the trailing edge of Landslide #I, measuring approximately 5 to 10 cm in length and 2 to 15 cm in width. These cracks align with the orientation of the scarps and have continued to widen during the field survey. In the rear portion of Landslide #I, several tensile cracks have developed, partially filled with yellow–brown clay and plant roots. Figure 3g depicts one of these cracks, situated roughly 10 m from the landslide's trailing edge, with an average width of 1 cm and a length of 5 m. The local cracks exhibit a staggered pattern, and the maximum displacement height reaches approximately 20 cm (Fig. 5).

Fig. 5
figure 5

Displacement and velocity monitoring data from YJ07 and YJ08 from October 2016 to March 2018(the interval of data is 2 or 3 months)

Full size image

To monitor the situation, two monitoring points, YJ07 and YJ08, were established on Landslide #I, as indicated in Fig. 6a. The monitoring data, collected during 11 monitoring periods from October 18, 2016, to August 3, 2018, along with other relevant information, can be found in Table 2 and Fig. 8 for YJ07 and YJ08, respectively. Evidently, the landslide has undergone significant deformation, with settlement reaching approximately 4 m. By November 2018, the landslide transitioned into a large-scale sliding event, ultimately giving rise to the present condition of Landslide #I.

Fig. 6
figure 6

Geological profile (B1-B2) of Landslide #II. ZK is the borehole after landslide occurrence, and SK is the ore exploration borehole before mining

Full size image
Table 2 Displacement and velocity monitoring data of Landslide #I
Full size table

3.2 Characteristics of landslide #II

Landslide #II's shear outlet is located on the mining platform, and the process of deformation and failure has persisted for four years. Notably, the landslide has exhibited a creeping behavior, lacking the distinct characteristics of short-term rapid sliding. The morphological characteristic parameters and corresponding values for Landslide #II can be found in Table 2.

Landslide #II features steep trailing edges and sides, with the trailing edge aligned parallel to the direction of the slope steps. The west sidewall trends at 201°, while the east sidewall trends between 90° and 126°. The predominant material composition of the landslide consists of gray-black carbonaceous mudstone and silty mudstone, displaying yellow rust staining on the surface. The slip zone, as revealed by ZK-04, is characterized by a weak intercalation of detritus intercalated clay. This weak component forms due to the softening of carbonaceous mudstone upon contact with water. The crushed rock fragments within the landslide primarily consist of moderately to strongly weathered carbonaceous mudstone, with particle sizes ranging from 1 to 3 cm. Based on the analysis of landslide deformation, failure mechanisms, and material composition, Landslide #II can be divided into two distinct areas: Zone A, representing the bedding slip zone in the footwall of the fault, and Zone B, representing the toppling deformation zone in the hanging wall of the fault, as depicted in Figs. 3 and 6.

Based on the drilling data and field investigation results, Zone A primarily consists of moderately weathered carbonaceous mudstone. Within this zone, the rock mass undergoes sliding along the weak interlayer. Core samples obtained from the boreholes exhibit relative completeness and thinness, occasionally with localized coal seams interspersed. In contrast, Zone B is primarily characterized by black, strongly to fully weathered carbonaceous mudstone fragments, partially intermixed with muddy components. These fragments, ranging in particle size from 1 to 7 cm, account for approximately 75% of the deposited material. Due to the folding and fracturing of the rock formation (Fig. 3d), the rock mass within Zone B exhibits severe fragmentation. The lithology of the landslide bed primarily consists of moderately weathered carbonaceous mudstone, displaying a trend ranging from 170° to 235° and a dip angle ranging from 14° to 33°. The carbonaceous mudstone exhibits a gray-black coloration and possesses an argillaceous structure. Overall, the structure of the carbonaceous mudstone is relatively intact, with poor water permeability, acting as a barrier to water flow. The sliding bed aligns closely with the depth of the weak interlayer, indicating that the bedded rock mass slides along this weak interlayer.

The bedding slip zone, referred to as Zone A, displays an approximately rectangular shape with a slope inclination of around 23°. The boundaries of this zone intersect with the slope direction at a small angle. Within Zone A, vegetation has thrived, and the original mining steps are still discernible. Significant swelling deformation is observed in this area. Moreover, local tension cracks have formed, extending for several meters and exhibiting an opening width ranging from 0 to 30 cm. It should be noted that the bedrock below the accumulation body of Landslide #I has also experienced damage as a result of its influence.

The toppling deformation zone, designated as Zone B, exhibits a slope angle ranging from 26° to 30°. The vegetation in this area is poorly developed. Within Zone B, a significant portion of the original mining steps in Zone A has been destroyed, leading to extensive folding and fracturing of the rock mass. Consequently, a large amount of bedrock is exposed. Tension scarps are widely distributed in the trailing edge and side edges of Zone B, characterized by folded and fractured rock masses. During the engineering geological survey, a total of 30 scarps were identified. These scarps range in height from 0.5 to 4 m, with spacing between them ranging from 5 to 10 m. Their lengths vary from 10 to 150 m, with the longest scarp measuring approximately 150 m. This scarp forms the left and rear boundaries of Landslide #II.

From August 6, 2015, to August 3, 2018, monitoring data were collected over 21 periods to assess the behavior of Landslide #II. The sliding direction of the landslide is determined to be 152°, closely aligning with the inclination of the mine slope. The positions of the monitoring points, YJ05 and YJ06, on the landslide are depicted in Fig. 7. The displacement and velocity monitoring curves, along with corresponding data for YJ05 and YJ06, are illustrated in Fig. 5 and Table 2, respectively. Analysis of the data and curves reveals that the landslide is undergoing long-term creep deformation.

Fig. 7
figure 7

Displacement and velocity monitoring data from YJ05 and YJ06 from August 2015 to March 2018 (the interval of data is 2 or 3 months)

Full size image

3.3 Formation sequence and correlation of the two landslides

The formation and evolution of the two landslides are closely related to the rock mass conditions revealed by mining excavation. As shown in profile xx, the landslide mass of landslide is almost above the strong weathering boundary. The mudstone developed on the top of the mining area has been continuously deteriorated by long-term rainfall, weathering and seismic vibrations. In particular, the excavation and erosion between 2002 and 2008 intensified the deterioration progress. As the excavation depth increased from 2008 to 2015, the deep mudstone began to undergo bending and toppling deformation due to weathering and unloading. However, the anti-dipping rock mass in the Landslide #II is mostly between the strong weathering boundary and the moderate weathering boundary. And its high strength results in a slight toppling deformation. At this stage, slope cutting and load reduction can effectively slow down the slope deformation. Before the complete instability of Landslide #I, the ballast of the upper slide body and the traction of the lower slide body is mutual influence in the potential intersection area of the two landslides. Before the fault was exposed, the two landslides had a similar evolution model that is the rock mass fragmentation process under the action of excavation, unloading-rainfall weathering and seismic deterioration. After the fault was exposed, the landslide #II was controlled by front edge creeping and trailing edge toppling deformation.

4 Mine slope instability mechanisms controlled by fault

4.1 Failure mode of Landslide #II

The investigation conducted reveals a clear relationship between the failure mode of Landslide #II and the structure of the rock mass. The presence of fault significantly influenced the slope failure mode, transitioning it from creeping-tension to creeping-tension-toppling. Such a distinct failure mode commonly occurs in slopes with thin bedding and significant variability in rock characteristics, often resulting from tectonic forces. Within the slope, there exists a potential slip zone that has the potential to develop into a failure surface. It is observed that thinner rock layers exhibit lower friction coefficients between the layers, making them more susceptible to this type of deformation. Based on the case analysis, the deformation evolution of the landslide can be divided into four stages. The results of numerical simulations support the occurrence of these four stages.

(1) Unloading recovery stage of slope front edge:

During the process of artificial excavation, the weak layer within the bedded rock mass at the front edge of the slope initiates the formation of a sliding surface. As excavation progresses, the rock mass situated above the weak layer begins to experience downward creep deformation. The deformation process is primarily governed by the properties and characteristics of the weak layer, which acts as the sliding surface for the movement. The occurrence and specific attributes of the weak layer play a significant role in determining the extent and nature of the deformation observed.

(2) The sliding surface extends from the bottom to top of the slope:

As the deformation progresses, the sliding surface extends towards the junction of different rock deformations, particularly where fault is present. The development of deformation also leads to a loosening of the rock mass structure within the slope. This loosening creates additional space for the deformation and failure of the anti-dip strata, particularly in the middle and rear portions of the slope. It is evident that fault plays a crucial role in facilitating this type of deformation evolution. The interaction and displacement along fault lines contribute to the overall deformation and failure mechanisms observed within the slope.

(3) The bending stage of anti-dip formation in the middle and rear of the slope:

As the front edge of the slope experiences creep, the fractured surface within the middle and rear portions of the slope continues to extend further backwards. This stage of deformation is often accompanied by localized caving of the slope, particularly when the angle of the anti-dipping rock layer is steep. The combination of ongoing creep and the presence of fractures can lead to the collapse or cave-in of specific areas within the slope, further contributing to the overall deformation and failure process.

(4) The fracturing and crushing stage of anti-dip strata in the middle and rear of the slope:

Due to the bending of the anti-dip strata, the moment acting on the strata also increases; therefore, once this kind of bending failure occurs, it usually shows progressive failure characteristics. This failure mode also reveals the inherent mechanical properties related to the slope development and changes, which largely determine the possible deformation characteristics and final failure of the slope rock mass.

Based on the occurrence and distribution of fault in the current mining area, it is inferred that the elevation of fault F1 exposure on the original terrain is about 816 m ~ 824 m. Between 2015 and 2018, the fault F1 was gradually exposed in the slope surface when mining ore in the bedding rock mass. According to the drill core of ZK-04 and ZK-05, the weak interlayer can be exposed on the mining platform at an elevation of 800 m ~ 810 m. Because the drainage on the slope surface almost failed during multiple deformations, rainfall can easily flow through the cracks in the loose rock mass to the mining working platform. The strength of the weak interlayer decreases rapidly after argillization. And the sliding of the bedding rock mass causes increasing in the free surface and decreasing in the supporting force of the anti-dipping rock mass in the hanging wall. Therefore, the planar sliding of the bedding rock mass in the footwall is transmitted to the anti-dipping rock mass in the hanging wall, causing toppling failure. The footwall rock mass slides greater in distance could bring the greater toppling degree and deformation depth of the anti-dipping rock mass. The failure mode of front part sliding and trailing part toppling is difficult to reinforce the slope through simple slope cutting and load reduction.

4.2 Configuration of landslide simulation

The physical and mechanical parameters used in the numerical simulation of landslides are comprehensively obtained from three aspects: mechanical tests, landslide inversion through limit equilibrium analysis and related research reports. Through the uniaxial test, direct shear test, and tensile test of the original sample, the range of physical and mechanical parameters of the slope rock and soil mass was obtained. By analyzing the stability coefficient of the pre-sliding slope is 1, the parameters under rainfall conditions were inverted (Table 3). The normal stiffness coefficient and tangential stiffness coefficient of the joints are converted by the elastic modulus and shear modulus between the rock mass and rock (Itasca Consulting Group 2014). Since there are too many parameter values involved, we will not expand further here.

Table 3 Rock mass properties used in the numerical model
Full size table

Except for major joints such as fault and bedding layers, there are no other major controlling joints of other tendencies developed in the mudstone rock mass. Therefore, two-dimensional numerical simulation can be used for analysis. The unique deformation mechanism of Landslide #II was replicated using the discrete element method (UDEC 6.0) through the establishment of an instability numerical model. The geometry of the Landslide #II numerical model is presented in Fig. 8, with dimensions of 195 m in length and 94 m in height. The left, right, and bottom boundaries of the model are fixed boundaries. In the numerical simulation, the thickness of the layered carbonaceous mudstone was set to 1 m, while the fault was assigned a thickness of 1.4 m. The weak interlayer, located at the foot of the slope, was given a thickness of 20 cm. To simplify the model and account for the negligible role of silt in the landslide instability, carbonaceous mudstone was used instead of silty mudstone in the simulation. The upper part of the model above the weathering boundary represents strongly weathered carbonaceous mudstone, while the remaining portion represents moderately weathered carbonaceous mudstone. The strength of the exposed strongly weathered carbonaceous mudstone was reduced due to rainfall effects, as determined by field investigations and laboratory tests. In the numerical model, the strongly weathered rock mass was assigned the strain softening model of Mohr–Coulomb in the discrete element software, where the intensity attenuation of the weak layer with the fitting equation is realized by the fish function in UDEC. While the other zones were assigned the traditional Mohr–Coulomb model. The required parameters for the rock mass and joints used in the calculation can be found in Tables 3 and 4, respectively. To enhance the computational efficiency of the numerical model, the weak interlayer zone and the fault zone were divided into blocks with an average size of 0.5 m, while the remaining zones were divided into blocks with an average size of 1.5 m. Additionally, five horizontal displacement monitoring points (M1, M2, M3, M4, and M5) were set within the slope, as depicted in Fig. 8, to track the displacements during the simulation. Since the mudstone is a weakly permeable layer, rainwater penetrates into the fissure rock mass in the surface weathering zone and then quickly drains to the foot of the slope, making it difficult to form a stable groundwater level. Therefore, the numerical simulation in this paper considers rainfall as a deterioration of mechanical parameters.

Fig. 8
figure 8

Configuration of the Landslide #II model using discrete element software

Full size image
Table 4 Joint properties used in the numerical model
Full size table

4.3 Deformation characteristics and instability mechanism of Landslide #II

The failure mechanism and kinematics of Landslide #II were investigated using the discrete element method. In the numerical simulation, the initial calculation focused on the slope without excavation, where the weak interlayer was not yet exposed. After approximately 5000 cycles, the numerical model reached an initial static equilibrium state, as shown in Fig. 8. When the slope was not excavated, no significant signs of deformation were observed, and the overall slope exhibited stability. The maximum horizontal displacement, measuring approximately 7 cm, was observed only at the top of the slope.

Taking into account the influence of excavation, the initial static equilibrium numerical model was excavated to expose the weak interlayer. Due to the presence of the open face of the excavation, the rock mass in Zone A (footwall of the fault) progressively slips along the weak surface under the influence of long-term gravitational forces. As the calculation progresses, the displacement zone induced by excavation propagates upwards, leading to the toppling and deformation of the rock mass in Zone B (hanging wall of the fault). After an additional 4000 cycles, the evolution of displacement characteristics is depicted in Fig. 9. Approximately 25,000 cycles later, the displacement of the landslide tends to reach a balanced state (Fig. 10). The final displacement characteristics of the numerical model are illustrated in Fig. 11. Within the bedding slip zone (Zone A) of Landslide #II, the maximum horizontal displacement reaches 1.5 m. In the topping deformation zone (Zone B), the maximum horizontal displacement measures 4.5 m, occurring in the middle section of the landslide. The horizontal displacement curves of the five displacement monitoring points (M1, M2, M3, M4, and M5) during the numerical simulation are displayed in Fig. 11. After the slope excavation, the displacement increment of Zone A is observed to be greater than that of Zone B in the later stages. Based on the results of the numerical simulation, the entire failure process of Landslide #II is explained. The exposure of the weak interlayer triggers the slip deformation of Zone A due to mine excavation, which subsequently leads to the toppling deformation of Zone B.

Fig. 9
figure 9

Evolution of horizontal displacement characteristics after excavation finishes: (a)–(d) Slip failure occurs at the leading edge of the slope after excavation, and deformation propagates upward, leading to toppling and breaking failures

Full size image
Fig. 10
figure 10

Final deformation characteristics of the Landslide #II numerical model

Full size image
Fig. 11
figure 11

Horizontal displacement at the monitoring points of the Landslide #II numerical model

Full size image

Numerical simulation results show that the overall deformation of the slope is mainly controlled by the sliding of the bedding rock mass along the weak interlayer caused by excavation. There are significant differences in the deformation of the five monitoring points at three stages (Fig. 11). In the initial stage before excavation of the footwall, the anti-dipping rock mass undergoes weak deformation due to the unloading effect of excavation, and the deformation amount gradually decreases as the slope elevation decreases. After the excavation of the footwall rock mass, the bedding rock mass immediately slipped, which was manifested as a sharp increase in the deformation of M4 and M5. The leading edge slip pulls the anti-dipping rock mass in the trailing edge to undergo coordinated deformation, and the amount of toppling deformation increases rapidly and overtakes the footwall. Comparing the monitoring points YJ05 and YJ06, it can be seen that the deformation of the footwall before the landslide occurred was greater than that of the hanging wall. It is more consistent with the second stage of the deformation curve of the numerical simulation. Due to the failure of the monitoring points when landslide occurred, no valid data could be obtained for the subsequent excess deformation of the hanging wall. The final numerical simulation indicates that the slope deformation following the sequence of bedding slippage of footwall rock mass, the fault zone shifted downward, bending and toppling of hanging wall rock mass (Fig. 12). In terms of the deformation characteristics of the landslide, the numerical simulation results align closely with the findings of the field survey.

Fig. 12
figure 12

The failure mode of creeping-cracking-toppling

Full size image

5 Discussion

Based on the geological survey after landslide failure, the landslide instability evolution was replayed from the perspective of DEM analysis. However, comparing the on-site deformation monitoring and numerical results (M3 and YJ06, M1 and YJ05), it can be seen that the maximum deformation amount obtained by numerical calculation is much smaller than the actual value. Field investigation found that significant bending deformation is still visible on the rear edge of the landslide #II (Figs. 3b, 6). At present, we believe that the anti-dipping rock mass above the weathering boundary exhibits toppling bending deformation. And its formation is related to the long-term deformation of the original slope under the action of gravity. The rock mass above the strong weathering boundary is mostly scattered. In the actual landslide evolution, the rock mass on the upper part of the fault F1 was gradually fractured to form debris accumulation during the instability process (Fig. 3e). In order to reflect the entire process of landslide evolution, the initial DEM model must treat the upper and lower rock masses as continuous structures (Igwe 2015; Li et al. 2017). Therefore, the model in Fig. 8 gives priority to the similarity of the initial geological structure, and does not yet consider the disintegration behavior of the sliding body (Fig. 10). But it is significant that the landslide evolution process shown by our DEM results is consistent with the instability mode given by geological survey and monitoring data analysis, effectively revealing and verifying the landslide deformation #II and instability mechanism. In addition, it is still unclear whether the bending and toppling deformation phenomenon is influenced by mining (Sabino 2014; Yu et al. 2020), and we still need to carry out long-term observations on site to this mine slope.

6 Conclusions

This study aimed to comprehend the impact of excavation and fault on the development of a rock slope. Through field investigations, monitoring, and numerical simulations, the deformation characteristics and failure mechanism of Landslide #II were thoroughly analyzed. The insights gained from this study shed light on how excavation and fault contribute to slope instability, which is crucial for assessing slope stability and implementing effective mitigation measures. The main conclusions are as follows:

(1) Landslide #II exhibits a complex geological structure, with a fault traversing the foot of the slope, resulting in distinct rock structures. The footwall comprises a bedded rock mass with weak interlayers, while the hanging wall consists of an anti-dipping rock mass. Excavation and rainfall emerged as key factors contributing to the occurrence of Landslide #II.

(2) Employing the discrete element method, we simulated the slope failure process of Landslide #II. Excavation exposed the weak interlayer in the bedded rock, causing slip and forming the bedding slip zone (Zone A). Displacement from excavation propagated upwards through the fault, leading to toppling failure in Zone B. In essence, the deformation and failure of Landslide #II were influenced by excavation, fault, stratigraphic structures, and the strength of carbonaceous mudstone affected by rainfall.