1. Introduction
Slope problems have become one of the three most prevalent major global natural disasters due to population growth and overexploitation of the natural environment [
1]. Many countries around the world have experienced devastating landslides in recent decades [
2,
3]. Several alpine countries in Europe, such as Italy and Austria, have experienced serious landslide disasters. Also, landslides often occur in South America, North America, and Asia. In these areas, stratified rock slopes are one of the most common types of slopes. One example of a tragic landslide event occurred on 29th April 1903, in Alberta, Canada, where a slope instability event of 30 million m
3 buried all nearby villages, resulting in the loss of 70 lives [
1,
4,
5]. On 31st May 1970, a rock slide in Peru, triggered by an earthquake, claimed over 18,000 lives. In 1974, one of the largest landslides ever recorded occurred in Peru’s Mantaro Valley, with a volume of 1.6 billion cubic meters, resulting in 450 people either missing or dead [
4,
6,
7,
8].
A bedding rock slope is a dipping surface composed of a rock mass with an inclination close to or consistent with the inclination of the bedrock’s bedding [
9,
10]. Several researchers have extensively studied the fault factors and modes of bedding rock slopes due to their impact on engineering construction and human life. Qin Hui et al. (2023) [
11] proposed the use of an unloading stress analyzing method (USAM) to determine the sliding surface, safety factor (Fs), and reinforcement force (Fr) of bedding rock slopes. Zhenlin Chen et al. (2020) [
12] found that the inclination angle of weak interlayers is a crucial factor that affects the fault mode of rock slopes. Zhang Jipeng et al. (2021) [
13] categorized the fault modes of bedding rock slopes into three distinct modes, namely, integral slip mode, collapse fault mode, and slip tensile fracture mode. Zhang Zhuoyuan et al. (2009) [
14] summarized the evolution of the fault modes of rock slopes as follows: sliding fault along weak structural planes, collapse, and destruction. Zhu Hanya et al. (2004) [
15,
16,
17] compared the fault of bedding rock slopes with the buckling instability of slabs and beams. They provided a criterion to determine stability displacement. Thanks to years of extensive research and analysis, the procedure used to analyze and study slope stability has matured. The methodology and work flow of our study can be described as follows: recognition and characterization of the actual slope; development of mechanical and mathematical models; application of a calculation method; and elaboration of the concluding remarks. Our slope stability research is focused on the construction of an appropriate mechanical and mathematical model, as well as on the use of reasonable calculation methods for slopes [
18,
19,
20,
21,
22,
23,
24].
The stability of rock slopes is often determined by locking sections along potential sliding surfaces, which have a high bearing capacity in order to resist instability. Hongran Chen et al. (2018) [
25] discovered that the physical model of a multi-locking-section slope is solely dependent on the displacement of the first locking section’s volume expansion point and on the number of sections. Hongnan Qin et al. (2020) [
26] proposed that the time series curve of fault deformation in the multi-locking-section slope differs significantly from the typical three-stage fault deformation theory of slopes with creep characteristics. Qifeng Guo et al. (2020) [
18] considered the slope as a system and used the energy method of system stability to calculate the safety factor of rock slopes with locking segments. They also considered the weakening effect of joints and the locking effect of rock bridges.
With the rapid development of society, various types of large-scale engineering projects have been constructed. During and after their construction, it is essential to involve a great degree of slope engineering, including engineering many bedding-layer rock slopes. Especially in areas with complex geological conditions, construction inevitably involves manual excavation, which poses extremely high risks. At this point, the feasibility, safety, and economy of the entire process need to be judged by analyzing the stability of the bedding-layer rock slope. To some extent, it can also be said that the stability of the slope determines the success or failure of the project and will greatly affect the investment and benefits of the project.
Currently, research on bedding rock slope primarily focuses on stability and post-landslide treatment measures. However, there is a lack of scientific and rigorous research focused on the formation mechanism of slope instability, as well as on its prevention and on issuing timely warnings when it occurs. In this work, an experimental series is performed through the application of different loads to a bedding rock slope model in order to study the changes in its macro-characteristics and analyze its fault mechanism. These findings provide a significant reference for slope treatment and similar slope engineering.
3. Process of Slope Load Deformation Fault
DH3821 displacement sensors were fixed on the sliding surface of the slope model, and acoustic emission probes were fixed onto the side of the slope, allowing them to be connected to the corresponding acquisition system (
Figure 3). The model was subjected to loads according to a predetermined loading scheme until the end of the test (
Table 3), with the load being applied successively through the console of the YDM-D-type geotechnical structure model testing machine.
The DH3821 data acquisition system was used to collect the displacement response data of the model under each load condition, in conjunction with the displacement sensor and the acoustic emission acquisition system.
Figure 3 illustrates the data acquisition process.
The main technical indicators of DH3821 are shown in
Table 4:
To observe the test process, the deformation and fault characteristics of the slope model were analyzed using a 3D laser scanner. Based on the analysis of displacement and acoustic emission data collected during the test, the slope model revealed varying degrees of fault in response to different loads [
32,
33,
34,
35,
36].
The load was applied along successive increments ranging from 0.143 MPa to 1.786 MPa. Real-time displacement data were collected using the DH3821 data acquisition system at a frequency of 0.5 s, as
Table 5,
Table 6 and
Table 7,
Figure 4,
Figure 5,
Figure 6,
Figure 7,
Figure 8,
Figure 9,
Figure 10,
Figure 11,
Figure 12,
Figure 13,
Figure 14 and
Figure 15 below.
As
Figure 4,
Figure 5,
Figure 6,
Figure 7,
Figure 8,
Figure 9,
Figure 10,
Figure 11,
Figure 12,
Figure 13,
Figure 14 and
Figure 15, the red markings on the graph represent the external manifestations of deformation and failure in the slope model. During the loading test, cracks became visible at the front and back of the model. As the load increased, cracks developed on the middle side, right side, and slope foot, and they continued to grow as the load increased. The cracks inside the slope model eventually extended from top to bottom and connected with those at the foot of the slope. As a result, the slide surface of the slope slid significantly, forming a through-slip surface, causing complete destabilization and damage to the model.
4. Data Analysis
The DH3821 data acquisition system collects real-time displacement data at a frequency of 0.5 s.
Figure 16 displays the horizontal displacement of various parts of the slope for different loading intensities.
Figure 16 displays displacement sensors DH1, DH3, and DH5 located at the bottom, middle, and top of the slope sliding surface, respectively. The curve illustrates that an increase in the strength of the vertical load on the top of the model results in an increase in the lateral displacement on the side slope.
During the initial loading stage, shear deformation primarily occurred at the top and middle sections of the slope, while the foot of the slope experienced minimal to no deformation. Under increased load, the slope experienced a sudden small change in shear deformation due to its low shear strength. This deformation gradually increased. In the middle of the slope, deformation increased steadily, but it remained smaller than that of the upper part. At the foot of the slope, deformation was minimal or nonexistent.
When the load strength increased beyond a certain point, the slope’s sliding surface experienced significant shear deformation due to the absence of retaining on the free surface.
Figure 4 shows that as the load strength increased step by step, DH5 at the top of the slope underwent large-scale shear deformation and fault at a load strength of 0.786 MPa, followed by DH3 in the middle of the slope at 0.857 MPa. Finally, DH1, located at the locking section of the slope foot, experienced abrupt lateral displacement at a load strength of 0.929 MPa. As the load increased, the displacement at the sliding surface of the slope decreased until the slope completely destabilized.
Figure 17 and
Figure 18 show the analysis of signal strength and energy at S1–S7 during the test based on real-time photography and video recording, which was performed using a 3D laser scanner. The analysis is based on the increase in load and the passage of time during the test.
Figure 17 and
Figure 18 demonstrate that the slope exhibited varying signal strength and energy as time and vertical load increased at its peak and near the top at around 1300 s, specifically at measurement points S7 and S6. No significant changes were observed in other areas. At around 13,500 s, the maximum peak occurred at the S4 measuring point in the middle of the slope. Finally, the maximum value was observed at the S1 measuring point located at the bottom of the slope at around 15,000 s. It can be inferred that, as the load strength increased, the displacement changed from the top of the slope to the bottom.
It is evident that the shear strength was smaller near the top of the slope sliding surface, stronger in the middle, and largest at the bottom.
6. Numerical Simulation Analysis Based on FLAC 3D
6.1. Introduction to FLAC 3D
In the 1980s, British scholars Peter Cundall and Itasca developed finite difference software and applied it to the design and construction of large-scale projects such as water conservancy and civil engineering, where the software played an important role. Now, FLAC has been widely used around the world.
Based on the finite difference Joseph-Louis Lagrange algorithm, FLAC 3D can simulate the whole process of deformation and failure of the model, which can provide effective guidance for engineering design and research. Because FLAC 3D uses a hybrid discretization method that is more advanced than progressive iteration, it is more suitable for simulating problems in large deformation or torsion mechanics of rock masses; problems in simulating slope stability can be solved without obstacles. At the same time, FLAC 3D is solved by display without memory matrix, which can be used to solve large strain problems in a short time.
6.2. Numerical Model Setting
Compared to the model test, numerical simulation calculations are less restricted by particular conditions. In this paper, we used numerical simulation to analyze the fault mode of the model under different conditions and to simulate the macroscopic deformation, fault geometry, and degree of slope for different loads.
The first step in the FLAC 3D simulation calculation is the selection of the calculation model, which generalizes the actual slope model based on the research purpose and the size and complexity of the calculation model [
37]. For this test, the simulation analysis uses the numerical calculation model shown in
Figure 19, in which only the upper part and the sliding surface can be freely deformed, while all other surfaces are static.
The laboratory test set the slope model in place for more than 15 days after the completion of the backfill in order to maintain the slope model and, simultaneously, carry out the stress compensation and deformation adjustment of the slope model in advance. Therefore, the initial deformation of the model is not considered, and displacement equilibrium is performed after calculating the initial stress. This paper uses the total stress analysis method for numerical simulation.
6.3. Model Material Parameters
The setting of parameters has a direct impact on the results of numerical simulations. To ensure that the conclusions are well constrained and of a certain reference value, it is essential that we select the calculation parameters reasonably. The parameters used in this work are derived from the laboratory model tests, and the specific parameters are shown in
Table 8.
6.4. Numerical Model Results
6.4.1. Geometry of Fault Plane
The geometry of the fault plane can be obtained through the calculation and analysis of the numerical model of a bedding-locked rock slope under load. The model shows that when the slope model is in the final unstable state, it is affected by a complete continuous fault plane, resulting in the instability of the slope and the locking section. Its geometry represents a combination of a straight line and a circular arc. The shape of the sliding surface when the numerical model of the slope finally fails is shown in
Figure 20.
6.4.2. Displacement Characterization Studies
In this study, the numerical simulation slope model was calculated, and the law of the variation in its horizontal displacement with the curve of the load applied to the top of the model was analyzed (
Figure 21).
The figure illustrates that the displacement of each measuring point increased gradually along with the load on the top surface of the side slope. Additionally, the lateral displacement was larger closer to the top surface, with a pattern of DH5 > DH3 > DH1. Between 0.929 and 1.000 MPa, DH1’s displacement point mutated, resulting in a complete through crack, which caused slope instability. Maximum instability displacement was obtained at 3.37 mm. The law can also be derived from the horizontal displacement cloud map in cases where the slope numerical model is unstable.
Figure 22 displays a horizontal displacement cloud image when the model is unstable.