Study on Mechanical Behavior of Excavation Supported by Rock-Socketless End-Suspended Piles in Soil–Rock Composite Strata Pit in Jinan

Abstract

Excavation in soil–rock composite strata poses significant challenges in regard to deformation control due to stiffness contrast and interface discontinuity. Based on the optimization requirements of a foundation pit project in Jinan Metro Line 7, we evaluated an end-suspended pile support system without rock-socket support through physical model tests and numerical simulations. The results indicate that ground settlement exhibits a typical “trough-shaped” distribution with an influence range of approximately 20 m. The pattern of retaining wall displacement evolves from being “inverted-triangular” into a “vase-shaped” during staged excavation, with maximum displacement remaining within code limits. Bending-moment peaks can be observed near strut levels and approximately 1 m above the soil–rock interface, reflecting stress redistribution and differential constraint effects. Parametric analysis demonstrated that increased rock weathering reduces formation stiffness and amplifies deformation and strut forces, whereas moderately weathered rock provides more effective restraint. A steeper interface dip angle induces asymmetric deformation due to stiffness contrast, increasing overall structural demand. An increase in rock-socketed depth, particularly within 4.0–4.5 m, significantly enhances anchorage performance and deformation control. These findings provide quantitative support for optimizing suspended pile systems in soil–rock composite strata.

1. Introduction

With rapid urbanization, development of high-rise buildings and underground spaces frequently encounters complex geological conditions, such as soil–rock composite strata. In regions like Shandong, excavation depths for foundation pits in such strata have been continuously increasing. However, due to economic and construction constraints, the embedded depth of supporting piles into the rock layer is often limited, preventing them from reaching the excavation base. This has led to the formation of a unique support system utilizing end-suspended piles. Consequently, the end-suspended pile support system has been adopted in numerous engineering projects [1,2,3,4,5,6,7,8,9].

The mechanical behavior of rock layers in soil–rock composite strata is highly heterogeneous due to factors such as weathering, jointing, and lithological variations. Understanding these rock-related mechanical properties is crucial for designing efficient and safe support systems in composite strata, particularly for end-suspended piles that rely on rock integrity for toe restraint. Li et al. [10] demonstrated that ignoring the spatial variability of mechanical parameters in layered rocks can lead to erroneous stability assessments. Similarly, Liu et al. [11] revealed that the anisotropic behavior of layered rocks controls the failure mode under compressive loads. Zhu [12] indicated that material heterogeneity directly induces localized stress concentrations. Chai et al. [13] proposed a refined heterogeneous modeling method based on borehole data, explicitly pointing out that traditional homogeneous models cannot reflect the true mechanical response of weathered rock layers. Finally, Zhang et al. [14] analyzed the nonlinear influence of rock content on the overall strength of severely fractured soil–rock mixture zones.

Numerous scholars have extensively researched and employed foundation pits in composite strata and their supporting structures. Liu [4] employed both finite element analysis and field monitoring to verify the applicability of the end-suspended pile support system in cover-excavation station pits within soil–rock composite strata. The results indicated good applicability. Lei et al. [15] systematically investigated the deformation and mechanical characteristics of a combined support system using ground anchors and end-suspended piles in soil–rock composite strata through field monitoring and numerical simulation, confirming the effectiveness of this integrated method in controlling deformation and reducing construction costs. Sun et al. [16] proposed three potential failure modes for end-suspended pile excavations in rock masses with outward-dipping structural planes or fractured rock masses and provided methods for calculating the stability safety factor under different failure modes using the limit equilibrium method. Wu [17] investigated the deformation patterns of the retaining structure and surrounding soil in deep excavations with end-suspended piles in a typical “soil–rock dual-structure stratum”, summarizing the deformation characteristics of the support structure and ground surface settlement. Huang et al. [18], based on field monitoring data from a soil–rock composite pit with end-suspended pile support, concluded that this support system demonstrates good adaptability in such engineering contexts. Diao et al. [19] conducted numerical simulations using both separate and integrated modeling approaches for the support structure in soil–rock composite pits, innovatively proposing a new calculation concept for the additional internal forces generated by deformation coordination in a modified rigid-flexible system transition zone. Chi et al. [20] pointed out significant deformation differences between the two sides of end-suspended piles in soil–rock pits under asymmetric stratum conditions and ground surcharge and analyzed the influence of ground surcharge on the displacement patterns at the pile head and toe. Wu et al. [21] analyzed the influence of the soil–rock elastic modulus ratio on the bending moment and lateral displacement of end-suspended piles, providing recommended values for the depth at which piles should be embed into rock and rock shoulder width. Bai et al. [22] studied the dynamic responses of end-suspended pile displacement under moving loads, revealing the displacement patterns at the pile head, along the pile shaft, and within the rock-socketed segment. Zhao [23] argued that the soil and rock layers should be calculated separately when designing retaining structures for soil–rock pits and proposed a calculation system suitable for an end-suspended pile support system. Based on the position of the pile toe relative to the potential slip surface on a rock slope, Chen et al. [24] proposed methods for calculating the pre-applied axial force in locking anchor rods. Yu et al. [25] developed wavelet neural network and conventional neural network models to predict deformation in composite stratum pits with end-suspended pile support, demonstrating that the wavelet neural network offers greater stability and accuracy in predicting deep excavation settlement. Han et al. [26] analyzed field construction monitoring data, finding that the deep horizontal displacement of end-suspended piles exhibits an “inverted trapezoidal” distribution. Liu et al. [27] conducted simulation analyses by varying the reserved rock shoulder width and pile embedment depth for end-suspended piles, discussing the resulting changes in pile displacement and bending moment to propose reasonable design parameters. Li et al. [28] applied the finite element strength reduction method, considering the interaction between the support structure and the soil–rock mass, to analyze the lateral earth pressure distribution on an end-suspended pile structure, along with the failure modes of the soil–rock mass and the corresponding safety factors. Wu et al. [29] combined uniform design with sensitivity analysis, concluding that the sensitivity of embedment depth, rock shoulder width, and prestress in locking anchor rods to the deformation of end-suspended piles decreases in that order. Baca et al. [30] revealed significant differences in how the pile base and shaft mobilize their capacity under unidirectional compression, uplift, and bidirectional loading, based on model tests and numerical simulations.

However, traditional end-suspended pile support systems are hindered by certain limitations; they often struggle to adequately meet the requirements of or adapt to diverse geological conditions, necessitating further research and the development of more adaptable end-suspended pile support systems. Zhao et al. [31] established numerical models for scenarios both with and without a rock shoulder for comparative analysis, summarizing the deformation characteristics of the foundation pit and its retaining structure under these two conditions. Xu et al. [32] investigated the stress and deformation patterns during excavation supported by a rock-shoulderless suspended wall through field monitoring, concluding that a support structure combining a rock-shoulderless suspended wall with robust internal struts can ensure the stability of soil–rock composite foundation pits.

This study is based on a soil–rock composite foundation pit project carried out at a subway station on Jinan Urban Rail Transit Line 7 (Phase I). Corresponding model tests were conducted, and a local three-dimensional numerical model was established using simulation software to simulate the entire construction process. This analysis focuses on the characteristics of ground surface settlement around the pit, lateral displacement of the end-suspended piles, and axial forces in the internal struts under the rock-socketless end-suspended pile (RSEP) support system.

2. Project Overview

2.1. Stratigraphic and Geological Characteristics of Jinan

The Jinan area is characterized by a typical “upper soil, lower rock” dual geological structure, a pattern governed by the region’s geological evolutionary history. The northern section is part of the Yellow River Alluvial Plain, which is covered by thick Quaternary loose sediments; the southern part is a bedrock mountainous area, predominantly composed of Paleozoic limestone and Mesozoic gabbro [33]. Notably, the gabbro massif is distinguished by its dense, hard structure and its crust that is undergoing spheroidal weathering. Residual soil layers primarily consisting of silty clay and weathered debris are widely distributed across the transition zone between the mountainous area and the plain [34]. The physical and mechanical properties of these soils exhibit a gradual transition from bottom to top. Within approximately the top 20 m below the surface in this region, the “upper soil, lower rock” dual-layer system is universally present [35]. Of particular importance is the frequent occurrence of transition zones between the bedrock and the overlying soil layer, characterized by mixed soil–rock materials and gradational properties. Together, these features constitute the complex yet orderly engineering geology setting of the area [36,37].

2.2. Project Background

The study presented in this paper is based on a foundation pit project for a metro station on Jinan Urban Rail Transit Line 7. The effective platform center is at chainage CK12 + 798.000. The pit will be constructed using the open-cut method. The pit depths are as follows: 20.6 m at the small-chainage end, with its base at an elevation of 12.92 m; 19.4 m at the center chainage, with its base at an elevation of 14.65 m; and 22.1 m at the large-chainage end, with its base at an elevation of 12.14 m.

2.3. Engineering and Hydrogeological Conditions

The site for this foundation pit project is within a piedmont alluvial–pluvial-plain landform, characterized by a relatively gentle topography. The stratigraphic sequence from top to bottom consists of miscellaneous fill, plain fill, loess-like silty clay, silty clay, and completely weathered, highly weathered, and moderately weathered gabbro. The main physical parameters of the strata are presented in

Table 1. Main physical parameters of strata.

Strata	Layer Thickness (m)	Gravity, γ (kN/m3)	Elastic Modulus, E (MPa)	Dynamic Poisson’s Ratio, μd	Cohesion, c (kPa)	Friction Angle, φ (°)
Miscellaneous Fill	1.5	17.66	40.42	0.402	8.0	12.0
Plain Fill	1.2	17.17	42.19	0.401	10.0	12.0
Loess-like Silty Clay	2.5	19.33	90.24	0.389	24.0	9.0
Silty Clay	4.2	19.13	206.06	0.389	30.0	11.5
Completely Weathered Gabbro	3.1	20.60	767.89	0.366	25.0	30.0
Highly Weathered Gabbro	7.4	23.05	1779.23	0.361	30.0	40.0
⑳3Moderately Weathered Gabbro	-	25.41	7263.86	0.318	150.0	40.0
⑳31Moderately Weathered Gabbro	-	27.66	19,495.09	0.299	200.0	40.0

.

The groundwater at the station site comprises Quaternary porous water and weathering-fracture water within magmatic rocks. As the Quaternary clayey soils in the Jinan area typically exhibit a fissured structure with relatively good permeability, there is no distinct aquitard separating these two types of groundwater, resulting in a close hydraulic connection. The stable groundwater level is buried at a depth of 3.50 to 3.80 m, corresponding to an elevation of 29.64 to 30.37 m. According to the collected data, the annual fluctuation range of the water level in the Quaternary porous aquifer is approximately 2 to 3 m.

3. Design of the Support Structure for the RSEP Foundation Pit

3.1. Characteristics of RSEP

Although end-suspended piles, as a special form of support, are widely used in projects involving soil–rock dual-strata, traditional end-suspended piles typically require a reserved rock shoulder at their toe to provide anchorage and reaction force. However, this traditional support method has several drawbacks: the treatment of the rock shoulder increases construction costs and complexity, reduces the utilization efficiency of the excavation space, and makes bedrock excavation more difficult. The core characteristic of the RSEP support system lies in its composite load-bearing mechanism consisting of “support and anchorage in the upper soil, and self-stabilization in the lower rock mass.” The upper part balances earth pressure through a pile–anchor (or pile–strut) system, while the lower part relies on the inherent strength of the hard rock mass for stability. In this approach, natural geological conditions are fully exploited by integrating the engineering structure with the strength of the rock mass. It avoids the high costs, low efficiency, and spatial constraints associated with rock shoulder excavation and treatment, significantly improving construction economy, operational efficiency, and space utilization within the pit. Furthermore, it offers flexibility to adapt to complex stratum conditions, achieving an optimal balance between economic benefit, construction efficiency, and spatial utility while ensuring project safety.

3.2. Foundation Pit Support Scheme

The underlying stratum at the pit site consists of highly weathered gabbro with good integrity and high hardness. Conventional support methods are hindered by challenges such as the difficulty of drilling into hard rock. Such issues can affect pile formation quality and project timelines. Therefore, the RSEP support system was selected. This approach avoids the difficulties associated with drilling into hard rock while balancing support safety and construction feasibility, as shown in Figure 1. The pit is planned to be constructed using the open-cut method. The standard section width of the pit is 36.0 m, with an excavation depth ranging from 20.6 m to 22.1 m. The support scheme employs end-suspended piles (Φ1000@1400, with an unsupported length of 3.5 m), along with a cut-off wall and three levels of internal struts. The first level consists of a 1000 × 1000 mm concrete strut (including a 600 × 600 mm concrete tie beam) at −1.0 m. The second (−7.75 m) and third (−15.0 m) levels are both Φ800 × 20 mm steel struts, equipped with 56C double I-beam waling beams and 40C channel steel tie beams. Vertically, the struts are supported by 490 × 490 mm lattice columns (composed of 4 L200 × 16 pieces of angle steels). The excavation is divided into four stages. Each stage involves excavation to 500 mm below the bottom of the corresponding strut.

4. Model Test for RSEP

This model test was static; therefore, we only needed to satisfy geometric and boundary condition similarity. To simplify the experiment, the following assumptions were made: To highlight the stress-redistribution effect induced by excavation, the self-weight of the surrounding rock mass was neglected in the test. In the design of the model, the length and spacing of the internal struts were strictly determined according to the geometric ratio, while their strength was allowed to deviate within a certain tolerance. When selecting simulation materials, a certain deviation in mechanical parameters such as compressive strength, elastic modulus, and Poisson’s ratio is permissible. Furthermore, due to the limitations of model testing, the focus was solely on the main research objectives. Issues of lesser importance for this study were appropriately simplified. Accordingly, the originally complex stratigraphic structure was simplified into two representative layers: the overlying soil layer and the underlying rock layer.

The actual depth of the foundation pit is 22,000 mm, and it is 36,000 mm wide. The model box dimensions are 1200 × 620 × 1000 mm. Fully considering the experimental assumptions, equipment, and difficulty, we selected the geometric dimension (L) and material elastic modulus (E) as the fundamental physical quantities for the model test. Based on the actual dimensions of the foundation pit and the model test dimensions, the scale factors were set to C_L = 50, C_ρ = 1, and C_E = 50. The similarity ratio for dimensionless quantities is 1— e.g., C_μ = 1, C_φ = 1, and C_ε = 1 [38,39,40]. Dimensional analysis was employed to derive the similarity coefficients for the main physical quantities, as shown in

Table 2. Similarity relationships of key physical quantities in the scaled model test.

Key Physical Quantity	Dimension	Similarity Relationship	Similarity Coefficient
Density, (ρ)	ML−3	Cρ	1
Geometric Length, (L)	L	CL	50
Elastic Modulus, (E)	ML−1T−2	CE	50
Poisson’s Ratio, (μ)	-	Cμ	1
Strain, (ε)	-	Cε	1
Stress, (σ)	ML−1T−2	CECε	50
Internal Friction Angle, (φ)	-	Cφ	1
Cohesion, (c)	ML−1T−2	CECε	50
Force, (F)	MLT−2	CECL2	50

.

4.1. Configuration of Geotechnical Materials

The physical parameters of the overlying soil layer are shown in Table 2. As this study focuses on the mechanical behavior of the RSEP support system, the cohesion of the soil was selected as the primary control parameter. Therefore, dried river sand was used as the simulative material during the experiment, with a density of 1.72 g/cm³, an internal friction angle of 32°, and zero cohesion [41,42,43,44].

The focus of this test is on the support structure and deformation characteristics of the end-suspended piles. The parameters of moderately weathered gabbro and highly weathered gabbro were selected as the target parameters for the experiment. To better simulate actual engineering conditions, the elastic modulus (E) of the simulative material was employed as the primary target parameter. Based on the similarity theorem, the target values for the required physical and mechanical parameters of the simulative materials were calculated and are summarized in

Table 3. Original parameters of the rock strata and target parameters for the scaled model test.

Category		Elastic Modulus, E (MPa)	Cohesion, C (kPa)	Friction Angle, φ (°)
Original Parameters	Highly Weathered Gabbro	1779.23	30.0	40
	⑳3 Moderately Weathered Gabbro	7263.86	150.0	40
Similitude Coefficient		50	50	1
Target Parameters		35.58~145.28	0.5~3	40

.

Based on an orthogonal experimental mix design, specimens with dimensions of Φ39.1 × 80 mm were prepared and subjected to unconfined compressive strength tests. This mixture yielded an elastic modulus of 40.3 MPa, a cohesion of 1.1 kPa, and an internal friction angle of 36.2°.

4.2. Support Structure Type Selection

4.2.1. Similitude Calculation for End-Suspended Piles

In the actual foundation pit project, the support structure consists of Φ1000@1400 bored piles, a 700 mm thick CSM (Cement Soil Mixing) cut-off wall, and internal struts. The 700 mm thick CSM cut-off wall in the support system can be substituted for a diaphragm wall in the design to satisfy key characteristics such as load-bearing capacity, deformation, and seepage prevention. Therefore, to simplify the indoor model test, the bored piles and the CSM cut-off wall are equivalently represented as a diaphragm wall. The similitude design is based on the equivalence of flexural rigidity. The simulation material selected was Nylon PA66GF30%. The dimensions of the similitude material can be calculated according to the similitude ratios, as shown in

Table 4. Prototype and target parameters of the support structure for the scaled model test.

Category	Support Type	Pile Diameter/Wall Thickness (mm)	Elastic Modulus, E (GPa)	Length (mm)	Width (mm)
Original Parameters	Bored Pile + CSM Cut-off Wall	Φ1000 + 700	31.5	16,000	596
Model Parameter	Nylon PA66GF30% Plate	6.4 + 6.0	8.0	320	596

.

4.2.2. Similitude Calculation for Internal Struts

In the test, the dimensions of the struts had to be determined based on controlling the axial compressive stiffness (EA). The cross-sectional dimensions of the struts were converted according to the principle of equivalent axial compressive stiffness. The internal struts are also made of Nylon PA66GF30%. The calculations showed that the first-level concrete strut from the prototype could be simulated using a PA66GF30% rod with a diameter of 6 mm and a cross-sectional area of 28.27 mm². The second and third-level Q235B steel struts were simulated using PA66GF30% rods with a diameter of 4 mm and a cross-sectional area of 12.57 mm², meeting the test requirements. To simplify the test, a symmetrical section of the actual foundation pit was used in the model experiment. To simulate the connection between the internal struts and the suspended wall, threads were machined on both ends of the simulated struts. Corresponding holes were drilled in the simulated suspended wall and the steel plate of the model tank. The end near the model tank wall was connected via a bolt fixed to the tank wall and an elongated nut, ensuring the support structure acts as an integrated load-bearing system, as shown in Figure 2a.

4.2.3. Model Test Setup

To intuitively observe the deformation characteristics, back-wall earth pressure variation and top settlement of the cantilever retaining wall during model test excavation, 12 strain gauges were arranged on both sides of the wall; three earth pressure cells were installed at −40 mm, −160 mm, and −280 mm along the wall’s central vertical line to monitor back-wall earth pressure variation. A dial indicator was mounted on the top of the wall to monitor its vertical settlement, as shown in Figure 2b.

4.3. Excavation Procedure for the Model Test

To systematically investigate the mechanical behavior of the foundation pit excavation supported by RSEP, we simulated the sequential excavation process and the installation of internal struts. The testing procedure strictly adhered to similitude theory, ensuring temporal similarity between the model and the prototype in terms of construction stages. Based on the construction sequence for the prototype foundation pit, the model test was designed such that there were four main excavation steps. Each step corresponds to the removal of one soil/rock layer and the installation of the corresponding level of support. With the top of the pit as the design reference elevation, the specific excavation steps are outlined in

Table 5. Excavation steps for the model test.

Test Step	Procedure	Excavation Depth
Step 1	Excavate from the top of the pit to a level 10 mm above the first strut position; then, install the first strut.	−30 mm
Step 2	Continue excavating to a level 10 mm below the second strut position; then, install the second strut.	−165 mm
Step 3	Continue excavating to a level 10 mm below the third strut position; then, install the third strut.	−310 mm
Step 4	Excavate to the final formation level (base).	−440 mm

below.

4.4. Analysis of Excavation Data from the Model Test

4.4.1. Settlement Pattern at the Top of the Equivalent Suspended Wall

Figure 3a shows that the degree of settlement at the top of the equivalent suspended wall in the model test changed significantly during the excavation process. From Step 1 to Step 3, settlement increased slowly, remaining within 1 mm, indicating that the internal struts provided effective counterforce, thereby suppressing settlement. Between Steps 3 and 4, i.e., during the excavation of the deep rock mass to the final base, the degree of settlement at the top of the wall increased rapidly, reaching a peak value of 1.84 mm. This is because a substantial portion of the soil–rock mass near the toe of the wall remained unexcavated prior to Step 4, so it continued to provide support. In Step 4, this supporting rock mass was completely removed, leading to the loss of its supportive function. Consequently, the combined effect of the earth pressure behind the wall and the self-weight reduced the effectiveness of the internal struts, resulting in peak settlement. Although the settlement was greatest in Step 4, the absolute value remained small, indicating that the RSEP support system maintained overall stability throughout the excavation process.

4.4.2. Pattern of Variation in Axial Forces in Internal Struts

Figure 3b shows the variation in internal strut axial forces during the model excavation steps. The axial force acting on the first strut increased from −0.45 N to −19.50 N; thus, it, bore the largest historical cumulative load. During the third excavation step, the axial force of the second strut rose rapidly to –14.50 N, marking this strut’s assumption of the role of the dominant load-bearing element in deeper excavation. The force acting on the third strut’s force increased from −0.25 N to −15.10 N. Overall, this resulted in a gradually decreasing axial-force distribution from the top strut to the bottom one. The data indicate that each newly installed strut initially exhibited near-zero axial force—a “preparatory” state—with force increasing rapidly only during subsequent excavation. This finding demonstrates that the struts’ primary function is to provide restraint against ensuing retaining-wall deformation. As excavation progressed downward, increased wall deformation was countered by strut reaction forces, leading to the observed rapid rise in axial forces from their initial values.

4.4.3. Pattern of Variation in Earth Pressure

Figure 3c shows the variation in earth pressure behind the wall during excavation. The pressure generally increases with depth. Simultaneously, at the monitored depths of −400 mm and −1600 mm, the pressure decreases with excavation steps—from 0.45 kPa to 0.37 kPa (−17.78%) and from 2.32 kPa to 1.83 kPa (−21.12%), respectively. This reduction is due to inward wall displacement, which shifts the soil from an at-rest state to an active stress state, releasing stress. The greater decrease at the deeper level indicates more pronounced unloading where initial stress is higher. The rate of pressure increase is higher between −400 mm and −1600 mm than between −1600 mm and −2800 mm. The slower rise in the lower segment in a result of the constraint imposed by the stiffer simulated rock layer below the soil–rock interface.

4.4.4. Pattern of Variation in the Bending Moment in the Equivalent Suspended Wall

Figure 3d shows the variation in the bending moment in the equivalent suspended wall during excavation (a positive moment indicates outward bulging). Initially (in Steps 1 and 2), the maximum positive moment occurs in the upper-middle section of the wall. As excavation deepens, the location of the maximum moment shifts downward, particularly after Step 2. A sharp, nonlinear increase in moment near the soil–rock interface reflects stress concentration due to the stiffness contrast between the upper sand and the lower rock. After entering the rock layer (from Step 2 onward), the moment decays rapidly, indicating that the rigid rock provides effective vertical fixity, dispersing the moment from the upper wall and stabilizing the support system at its base.

5. Numerical Simulation Analysis of the RSEP Support System

5.1. Establishment of the Numerical Model

We developed a local three-dimensional (3D) numerical model based on a foundation pit for Jinan Metro Line 7 to investigate the deformation behavior of a support system using RSEP with internal struts in soil–rock composite strata. The model dimensions (170 m × 30 m × 50 m) were designed such that the horizontal extent was 3–5 times the excavation depth to minimize boundary effects. The Mohr-Coulomb model was adopted for the soil–rock strata, and an elastic-plastic model was used for the support structure. The stratigraphy was simplified into two representative layers: an overlying loess-like silty clay layer (12.5 m thick) and an underlying highly weathered gabbro layer. The bored piles have a standard length of 16.0 m, with 3.5 m embedded in rock. The model configuration is shown in Figure 4.

5.2. Selection of the Parameters of the Numerical Model

To simplify calculation in the numerical model, an equivalent conversion for the diaphragm wall was performed. The pile-row system, which bears vertical loads in a manner similar to a wall-type diaphragm wall, is generally equated to a wall-type diaphragm wall of a certain thickness in design based on the principle of equivalent flexural rigidity [45].

$$E{\displaystyle \frac{\pi D^4}{64}}=E{\displaystyle \frac{\left(D+t\right)h^3}{12}}$$

(1)

$$h=\sqrt[3]{{\displaystyle \frac{3\pi D^4}{16\left(D+t\right)}}}$$

(2)

where E is the elastic modulus of the support structure, D is the diameter of the bored pile, t is the clear spacing between two adjacent piles, and h is the equivalent thickness of the diaphragm wall.

In the model presented in this paper, the equivalent suspended wall was simulated using 2D plate elements; lattice columns, uplift piles, capping beams, concrete tie beams, steel tie beams, and internal struts were simulated using 1D beam elements. Steel waling beams and steel tie beams, which have irregular cross-sections, were modeled by pre-defining their corresponding specially shaped sections and then applying 1D beam elements for simulation. The soil and rock strata were simulated using solid elements. The parameters of the support structure are listed in

Table 6. Main physical parameters of the support structure.

Support Type	Gravity, γ (kN/m3)	Elastic Modulus, E (MPa)	Poisson’s Ratio, ν
Suspended Wall	23.5	31.5	0.25
Concrete Strut	23.5	31.5	0.25
Concrete Tie Beam	23.5	31.5	0.25
Capping Beam	23.5	31.5	0.25
Steel Strut	78.0	206.0	0.28
Steel Waling Beam	78.0	206.0	0.28
Steel Tie Beam	78.0	206.0	0.28
Lattice Column	78.0	206.0	0.28
Tension Pile	23.5	31.5	0.25

.

5.3. Numerical Simulation Procedure

This simulation aligns with the actual construction process in the project, adhering to the following principle: “excavate in layers and support accordingly”. Supports were installed after excavation for each layer. The simulation proceeded step-by-step based on the characteristics of the retaining structure and the position of the struts. The excavation steps for the simulation are shown in detail in

Table 7. Numerical simulation steps.

Simulation Step	Procedure
1	Generate an initial geostatic stress field and reset stratum displacement.
2	Construct the equivalent suspended wall, lattice columns, and tension piles.
3	Excavate soil down to an elevation of −2 m.
4	Construct the first-level concrete strut, capping beam, and tie beams at −1 m.
5	Excavate soil down to an elevation of −8.7 m.
6	Install the second-level steel strut, steel waling beams, and steel tie beams at −7.75 m.
7	Excavate soil down to an elevation of −15.95 m.
8	Install the third-level steel strut, steel waling beams, and steel tie beams at −15 m.
9	Excavate to the final formation level (base) at −22 m.

.

5.4. Analysis of the Deformation Characteristics of the Foundation Pit and Suspended Wall

5.4.1. Characteristics of Ground-Surface Settlement Outside the Pit

Throughout this section, the terms “excavation step” and “support step” are hereafter abbreviated as E-Step and S-Step, respectively, to ensure conciseness. The full terminology is provided upon first mention, and any abbreviations are used consistently thereafter. As indicated by the data in Figure 5, the ground surface settlement curves on both the left and right sides of the pit exhibit a symmetrical distribution, with both presenting “trough-shaped” profiles. The primary influence zone extends to within 20 m of the pit. As excavation progresses, ground surface settlement outside the pit gradually increases, with the maximum degree of settlement reaching 2.58 mm, at approximately 6 m from the edge of the pit. From E-Step 2 to S-Step 3, slight heaving can be observed at the ground surface near the edge of the pit. This phenomenon occurs because, in the initial phase of excavation, stress release takes place in the soil mass. Subsequently, with the installation of supports, stress redistribution occurs. The support structure restricts the deformation of the retaining system, promoting a redistribution of stresses: part of the stress is transferred toward the bottom of the pit, reducing its effective stress and causing rebound and heaving at the base of the pit; another portion of the stress is transmitted through the retaining structure to the surrounding soil, exerting an upward force on it, leading to surface heaving around the pit, as shown in Figure 6.

5.4.2. Lateral Displacement Characteristics of the Equivalent Suspended Wall

Figure 7 shows the lateral displacement curves of the suspended wall during the excavation process. From E-Step 2 to E-Step 4, the lateral displacement curves of the equivalent suspended walls on both sides remain largely consistent. The lateral displacement profile of the wall exhibits a “vase-shaped” pattern, characterized by smaller displacements at both ends and larger displacements in the middle. The displacement value gradually increases with excavation, reaching 8.07 mm by E-Step 4, which is approximately 0.368‰ of the excavation depth. During E-Step 1 and S-Step 1, as a major portion of the wall remains embedded within the soil–rock stratum, restricting lateral movement at its lower part, the displacement profile shows an “inverted triangular” distribution (larger at the top, smaller at the bottom). As excavation depth increases, the location of the maximum lateral displacement gradually moves downward. In E-Steps 2, 3, and 4, the maximum displacement points are located at −6.3 m, −9.9 m, and −11.2 m, respectively. Furthermore, the point of maximum lateral displacement consistently remains below the previously installed support level.

5.4.3. Pattern of Variation in Shear Force in the Equivalent Suspended Wall

Figure 8 reveals that the shear force distribution along the wall is essentially symmetrical. The maximum shear force, which is less than 400 kN, occurs at the top and is primarily induced by the earth pressure resisted by the first-level strut. During E-Step 1 and S-Step 1, the direction of the shear force at −1 m reverses. This reversal is caused by the inward deformation of the support structure due to excavation, followed by the counteracting reaction force provided by the first strut installed at that location. In the initial excavation stage, shear forces are small and exhibit a wavy pattern, reflecting the constraining effect of soil–rock embedment. As excavation deepens, the shear force in the exposed wall segment displays characteristics akin to a simply supported beam. After strut installation, the wall gradually transitions to a load-bearing mode resembling that of a multi-span continuous beam. Specifically, after E-Step 3, the lower part of the wall is in a cantilever state, and the shear force continues to develop. However, as the struts, once installed, are not fully engaged until further wall deformation occurs, the curve shows a gentle change. By E-Step 4, distinct abrupt changes in shear force become apparent at the locations of the three struts. The wall overall exhibits the load-bearing characteristics of a continuous beam, with the influence of the 0.9 m cantilever segment at the bottom being limited.

5.4.4. Variation Characteristics of the Bending Moment in the Equivalent Suspended Wall

As shown in Figure 9, the bending moment distribution on the side walls are nearly identical. Taking the left suspended wall as an example for analysis, we can see that the bending moment distribution along the wall exhibits regular changes during the excavation process. The portion of the wall within the excavated zone is almost entirely under tension, as the wall deforms inward under the action of earth pressure from outside the pit. The unexcavated side, constrained by embedment in the underlying soil–rock mass, experiences a significant negative bending moment at its toe. When the equivalent diaphragm wall undergoes horizontal deformation due to earth pressure, the capping beam imposes a constraint on the top deformation of the wall. This constraint induces a relatively large bending moment at the top of the wall. In the bending moment curve for E-Step 4, abrupt changes occur precisely at the locations of the three internal struts. The internal struts apply constraint and reaction forces to the wall, limiting its lateral displacement. Due to the difference in stiffness between the soil and rock masses, stress concentrates near the soil–rock interface, resulting in a relatively large bending moment. The maximum bending moment (505.56 kN·m) occurs approximately 1 m above the interface.

5.4.5. Pattern of Variation in Axial Forces in Internal Struts

Figure 10 illustrates the pattern of variation in internal strut axial forces with the progression of excavation (in steps). To eliminate boundary effects, the average axial force of the central struts was taken for analysis. As excavation deepens, the axial forces in the struts generally show an increasing trend. Notably, distinct inflection points can be observed on the axial force curves for the first and second struts, corresponding to the installation times of the second and third struts, respectively. This finding indicates that newly installed struts can effectively share the lateral earth pressure. Upon completion of excavation, the maximum axial force occurs in the first strut, with a magnitude of 1508.61 kN. This is primarily because the first strut is located in the upper part of the foundation pit, where it is subjected to greater earth pressure, and the strong restraining effect of the rock mass underneath is absent.

5.4.6. Stress–Strain Behavior of the Foundation Pit

Figure 11a shows that compressive stress is dominant around the pit, with localized tensile stress (max: 0.096 MPa) at the near-surface, pit corners, base, and wall toe. Surface tension stems from soil displacement behind the wall, while base tension results from unloading rebound. Tensile stress near tension piles is relatively low due to their constraining effect. Stress concentrates at pit corners due to geometric discontinuity. The wall toe, lacking a rock shoulder, acts as a constraint-free transition zone, leading to tensile stress concentration.

Figure 11b indicates that the minimum principal stress is mainly compressive, concentrating at sidewalls, in the lateral rear area, and at tension-pile interfaces (max: 1.361 MPa in the upper sidewall corners). This results from stress-path alteration after excavation, which redirects surrounding stress toward these zones, aided by the restraint provided by tension piles against base rebound.

Figure 11c shows that after the excavation of the foundation pit, the peak maximum shear stress occurs at the lower rock wall, amounting to approximately 0.550 MPa. The shear stress concentration zone is primarily distributed from the toe of the suspended wall to the corner of the bottom of the pit. This section of the rock wall is in an unsupported, free-face condition without the restraint of a retaining structure and thus must bear the earth pressure transferred from above and the reaction forces from the struts, exhibiting a certain lateral bulging tendency towards the pit interior. Simultaneously, the rock surface at the base is subjected to vertical reaction forces. The superposition of these two effects leads to a significant concentration of shear stress in this area. Although the current level of shear stress does not yet pose a threat of failure, this phenomenon reveals the potential shear failure mode of the unsupported rock wall. Its stability requires close attention, especially when the rock layer is weak.

6. Analysis of Factors Influencing the Stability of Foundation Pit Excavation

6.1. The Degree of Weathering of the Underlying Rock Layer

The degree of weathering of the underlying rock layer is a fundamental geological factor determining the stability of soil–rock composite foundation pits. Weathering significantly deteriorates the engineering properties of a rock mass, causing its integrity, strength, and elastic modulus to decrease sharply with an increase in weathering intensity. This situation not only compromises the bearing capacity of the rock layer itself but also directly influences the overall stability of the composite pit, along with the deformation and internal force response characteristics of the support system. Completely and highly weathered rock masses exhibit soil-like behavior, and this state can easily lead to a sidewall collapse and excessive deformation in the pit. This section will analyze the deformation patterns of the pit’s retaining structure and the characteristics of support internal forces under different weathering conditions of the underlying rock layer.

6.1.1. Ground-Surface Settlement Outside the Pit

Figure 12a illustrates the variation in ground surface settlement outside the pit under different weathering conditions with respect to the underlying rock layer. The settlement profile maintains a trough shape regardless of weathering degree, with a similar primary influence zone. Maximum settlement decreases as weathering decreases: it is 1.86 mm for moderately weathered gabbro, having decreased by 28% from the highly weathered case (2.58 mm) and by 50.4% from the completely weathered case (3.74 mm). Outside the main influence zone, the settlement differences are about 0.2 mm between highly and moderately weathered rock, and about 0.83 mm between highly and completely weathered rock. The lower settlement in moderately weathered rock results from its relatively intact structure, higher strength, and greater stiffness, which provide more effective vertical and lateral support, thereby limiting soil deformation around the pit. In contrast, the looser and weaker completely weathered rock leads to greater deformation.

6.1.2. Lateral Displacement of the Equivalent Suspended Wall

Figure 12b indicates that the degree of weathering of the underlying rock layer significantly affects the lateral displacement of the equivalent suspended wall. As weathering decreases (from completely to moderately weathered), the maximum displacement decreases from 11.34 mm to 5.81 mm, and its location shifts upward from −12.5 m to −7.75 m. The wall toe displacement is about 1.95 times larger under completely weathered conditions than under highly weathered conditions, but it is only 19.5% as large under moderately weathered conditions. In the upper 3.5 m (within the soil layer), displacement is mainly controlled by active earth pressure, and it is barely affected by rock weathering. Below −12 m, displacement differences remain stable. The rock mass constraint dominates: the stiff moderately weathered rock provides strong base fixity, reducing overall wall displacement, whereas the weak completely weathered rock offers little restraint, resulting in larger deformation.

6.1.3. Bending Moment of the Equivalent Suspended Wall

Figure 12c presents the bending moment curves under different degrees of weathering of the underlying rock layer. Overall, the maximum bending moment increases with the degree of weathering. The completely weathered condition yields the largest moment (−575.2 kN·m), followed by the highly weathered condition, while the moderately weathered condition results in the smallest moment (−312.98 kN·m), representing a 38.2% reduction relative to the highly weathered case. The location of the maximum moment remains consistent across conditions. In the upper wall section (above −8 m), the moment for the moderately weathered condition is slightly higher due to the stiffer underlying rock restraining lower-wall displacement, shifting the moment upward. Below −8 m, the moment in the highly weathered case surpasses that of the moderately weathered case due to the weaker rock constraint. The abrupt moment changes at the locations of the struts confirm their constraining effect. Despite these differences, the fundamental load-bearing mechanism remains similar across all weathering conditions.

6.1.4. Internal-Strut Axial Forces

As shown in Figure 12d, the axial forces in the three struts are highest for the completely weathered condition, followed by the highly weathered condition, and are lowest for the moderately weathered condition. Furthermore, the difference in axial force between adjacent struts progressively increases from top to bottom. The fundamental reason for this lies in the alteration of the deformation mode and load transfer path of the support structure caused by the stiffness difference in the underlying rock layer. The weak rock masses under the completely and highly weathered conditions induce larger lateral displacements, particularly at the pile toe. Consequently, all struts, especially the third strut closest to the weak rock layer, must provide greater constraining reactions to counteract the deformation. In contrast, the stiff moderately weathered rock mass results in a support system with greater overall stiffness and less deformation. The earth pressure is more effectively borne by the inherent stiffness of the pile itself and the rock mass, leading to a significant reduction in the load transferred to the struts. As a result, the strut axial forces exhibit a characteristic distribution of decreasing magnitude from the topmost to the bottommost strut.

6.1.5. Mechanical Equilibrium Analysis of the Pile-Rock Interface

To further elucidate the shear slip behavior, a simplified mechanical equilibrium model for the stability of the equivalent suspended wall toe was established based on limit equilibrium theory. Considering the support structure per unit width, the wall toe embedded in the rock mass is subjected to the driving shear force (T_driving) generated by the active earth pressure, along with the resisting shear force (T_resisting), which consists of the passive resistance of the rock mass and the frictional shear resistance at the base interface.

According to the limit equilibrium theory, the stability condition against ‘kick-out’ failure can be expressed as follows:

$$T_{driving}\le T_{resisting}=E_p+s\cdot \left(c+{\sigma }_n\tan \phi \right)$$

(3)

where T_driving is the driving shear force per unit width, induced by the active earth pressure from the overlying soil, pushing the wall toe towards the pit; T_resisting is the total resisting shear force per unit width provided by the rock socket; E_p is the passive resistance of the rock mass in front of the wall toe; s is the effective contact length per unit width, c is the interface cohesion between the concrete wall and the rock mass; σ_n is the normal stress acting on the interface, primarily derived from the vertical weight of the support structure; and φ is the interface friction angle.

For moderately weathered rock, the high elastic modulus and strength parameters result in a substantial T_resisting. The interface remains in an elastic fixed state, effectively restraining the lateral displacement. For completely/highly weathered rock, the degradation of rock properties significantly reduces E_p and the interface friction parameters (c, φ). The driving force tends to exceed the resisting limit (T_driving > T_resisting), leading to plastic shear slip. This theoretical mechanism aligns perfectly with the excessive lateral displacement observed in the simulation results (Figure 12b).

6.2. Dip Angle of the Soil–Rock Interface

The dip angle of the soil–rock interface governs the stability, potential slip surfaces, and mechanical behavior of foundation pit excavations, and significantly alters the earth pressure distribution on the support structure. To systematically investigate the response of the end-suspended pile support system under gently varying dip angles, we compared three typical angles—0° (horizontal), 5° (gentle dip), and 10° (steep dip)—while keeping the unsupported pile length constant. This study examines the failure mechanism, overall stability, and distribution of internal forces and lateral displacements in the support structure under different interface angles. For the 10° case, where rock is nearly exposed on the right side of the pit, a minimum overburden thickness of 5 m was maintained in the model to better reflect realistic engineering conditions, as shown in Figure 13.

6.2.1. Ground-Surface Settlement Outside the Pit

Figure 14 shows that increasing the dip angle of the soil–rock interface causes asymmetric ground settlement. On the left side, maximum settlement grows with dip angle (0°, −2.57 mm; 5°, −4.19 mm; and 10°, −6.14 mm), and the zone of influence expands leftward. On the right side, settlement decreases (0°, −2.58 mm; 5°, −1.80 mm; and 10°, −0.90 mm) and the influence zone contracts. Both sides retain trough-shaped profiles. The settlement profiles on both sides maintain trough-shaped curves. The underlying mechanism here is that the increased dip angle results in a thicker overlying soil layer and consequently greater earth pressure on the left side. Furthermore, the inclined interface provides a diagonal slip path for the soil mass, pulling more distant soil towards the left. On the right, the soil layer thins, reducing the load. The resulting unbalanced pressure also tilts the support structure slightly rightward, further increasing deformation asymmetry.

6.2.2. Lateral Displacement of the Equivalent Suspended Wall

Figure 15 shows the lateral displacement curves of the equivalent suspended walls on both sides under different dip angles of the soil–rock interface. As the interface dips leftward from 0° to 10°, the lateral displacement of the wall on the right side decreases significantly, especially below –10 m, where convergence is most pronounced. In contrast, the displacement of the wall on the left side (the hanging-wall side) increases markedly, with the greatest increase occurring below –12 m—a trend opposite to that occurring on the right. This opposing behavior results from the different geometric and mechanical conditions created by the inclined rock layer. On the right, the rising rock surface restrains soil deformation, suppressing wall displacement. On the left, the thickening overburden increases soil weight and active earth pressure, while the inclined interface weakens rock support and promotes the sliding of soil toward the pit. Together, these factors drive the left wall inward.

6.2.3. Bending Moment of the Equivalent Suspended Wall

Figure 16 shows that the dip angle of the soil–rock interface significantly affects the wall’s bending moment distribution. For dip angles of 0° and 5°, the moment distribution above the second strut (−7.75 m) are similar due to the comparable soil layer thicknesses on both sides. Above the third strut (−15 m), its stiffness minimizes deformation differences, and the relatively uniform stress distribution, combined with a fixed unsupported length on the left, results in similar moment patterns on that side. At a 10° dip, conditions differ markedly: on the right, uplifted rock leaves minimal soil cover, making rock constraint dominant; on the left, the wall toe makes contact with the excavation base and thus maintain the specified unsupported length, enhancing basal constraint. This leads to greater fluctuation in the moment curve. Although the moments on both sides are smaller than in the 0° and 5° cases, a residual moment persists in the lower-left section.

6.2.4. Internal-Strut Axial Forces

As shown in Figure 17, for dip angles of 5° and 10°, the third strut is subjected to the largest axial force, while the second strut is subjected to the smallest. This is because the third strut, located deeper and closer to the soil–rock interface and the rock-embedded section, concentrates the sliding thrust from the inclined interface and experiences greater lateral earth pressure. Compared to the horizontal interface case (0°), strut forces are significantly higher for both 5° and 10° dips. Notably, the axial force in the third strut under the 10° dip is lower than that under 5°. This occurs because at 10°, the left suspended wall makes contact withs the excavation base, transferring part of the deep-seated deformation forces to the base and thus reducing the load on the strut. In the 5° case, without this base contact, the deep deformation is entirely resisted by the third strut.

6.2.5. Quantitative Analysis of Additional Eccentric Bending Moment and Instability Mechanism

Due to the inclined soil–rock interface, the active earth pressure on the “deep-soil side” (where the soil layer is thicker) is significantly higher than on the “shallow-soil side”. To quantify the impact of the asymmetric earth pressure generated by the inclined soil–rock interface, an additional eccentric bending moment (ΔM_ecc) is introduced. It is defined as the difference between the peak positive bending moments on the deep-soil side (M_deep) and the shallow-soil side (M_shallow):

$$\Delta M_{ecc}=M_{deep}-M_{shallow}$$

(4)

Based on the simulation results shown in Figure 16, ΔM_ecc increases non-linearly with the dip angle. Specifically, at a dip angle of 5°, ΔM_ecc is approximately 201.69 kN·m. At a dip angle of 10°, ΔM_ecc rises to 216.39 kN·m.

Comprehensive analysis indicates that the inclined soil–rock interface fundamentally alters the instability path and safety boundaries of the support system, which can be delimited by three progressive critical states: (1) Lateral Constraint Failure, manifesting as a non-linear surge in toe displacement on the deep-soil side due to exhausted passive resistance; (2) Interface Shear Failure, corresponding to global shear slip along the wall–rock contact, particularly in highly weathered strata; and (3) Flexural Failure, occurring when the significant eccentric moment (ΔM_ecc) induced by steep dip angles exceeds the wall’s bending capacity. Consequently, engineers must identify the dominant instability mode based on rock weathering and the interface dip angle, prioritizing the verification of interface shear safety in weathered rock and the wall’s flexural capacity under large-dip-angle conditions.

6.3. Rock-Contact Depth for the Suspended Wall

The rock-contact depth is a key parameter controlling the safety and economy of the support system in soil–rock composite foundation pits. It directly influences the anchoring effect, flexural stiffness, and deformation compatibility of the structure. Insufficient depth may lead to inadequate embedment, while excessive depth is both uneconomical and increases construction risks in complex strata. This section systematically compares three typical depths—3.5 m, 4.0 m, and 4.5 m—to analyze their effects on the deformation modes, lateral displacement, and internal force distribution of the support system.

6.3.1. Ground-Surface Settlement Outside the Pit

Figure 18a shows that as rock-contact length increases, the maximum surface settlement gradually decreases, measuring −2.58 mm, −2.32 mm, and −2.22 mm for lengths of 3.5 m, 4.0 m, and 4.5 m, respectively. These values represent reductions of 10.08% and 13.95% compared to the 3.5 m case. While the settlement trend beyond the influence zone remains similar for lengths of 3.5 m and 4.0 m, a length of 4.5 m results in a slightly reduced influence zone. The difference in maximum settlement between the 4.0 m and 4.5 m cases is marginal. Given that the distance from the third strut (at −15.0 m) to the wall toe increases from 1.0 m to 2.0 m across the three cases, it can be concluded that increasing the unsupported length has a limited restraining effect on both the magnitude and extent of ground surface settlement in soil–rock composite pits.

6.3.2. Lateral Displacement of the Equivalent Suspended Wall

Figure 18b shows that the maximum lateral displacement of the wall decreases with an increase in rock-contact length. When the length increases from 4.0 m to 4.5 m, the displacement decreases from 7.81 mm to 6.48 mm. However, the reduction between 3.5 m and 4.0 m is only 3.34%, indicating minimal improvement. This suggests that a critical effective embedment depth exists. Beyond this depth, the extended pile segment provides sufficient lateral reaction, redirecting the load to deeper stable rock and significantly reducing deformation. Thus, an optimal unsupported length range is identified for the rock-socketless support system.

6.3.3. Bending Moment of the Equivalent Suspended Wall

Figure 18c reveals that the bending moment trends are similar for 3.5 m and 4.0 m rock-contact lengths in the upper wall section, while the moment is lower for 4.5 m. The maximum moments for the 4.0 m and 4.5 m cases occur in the lower section, measuring −416.68 kN·m and −387.06 kN·m, respectively—both lower than the 3.5 m case. This indicates that sufficient pile embedment enables the lower section to provide a counteracting moment, optimizing internal load distribution and enhancing transfer to deeper rock. When combined with the lateral displacement data, this finding confirms that a critical effective embedment depth range exists for the rock-socketless system.

6.3.4. Internal-Strut Axial Forces

As shown in Figure 18d, a clear relationship exists between strut axial forces and unsupported length. At 3.5 m, the first strut is subjected to the largest force, followed by the second, with that for the third being the smallest. At 4.0 m, the forces rank as follows: first > second > third. However, at 4.5 m, this order changes significantly: the third strut is subjected to the largest force, followed by the second, and the first is subjected to the smallest amount of force. This shift occurs because the length of 4.5 m enables effective embedment of the pile toe, changing the system’s load-transfer mechanism from strut-dominated to one where loads are transferred through the wall into deep rock. Consequently, the third strut (closest to the embedment) resists higher deep earth pressure, while reduced overall deformation lowers forces in the upper struts. This indicates that the interval between 4.0 m and 4.5 m likely represents a critical range for effective embedment.

7. Discussion

The combined physical model tests and numerical simulations reveal that the RSEP system in soil–rock composite strata operates through a coordinated stiffness-transfer mechanism during staged excavation. As excavation proceeds, the pattern of lateral wall displacement evolves from being inverted-triangular to a vase-shaped, indicating progressive stress redistribution and stiffness mobilization. Upper soil pressures are primarily resisted by the pile–strut system, while the embedded rock section gradually activates toe restraint as excavation depth increases.

Within this process, the internal struts exhibit staged axial force development, rapidly mobilizing compressive resistance after installation. The uppermost strut generally sustains the highest amount of axial force due to greater earth pressure demand and relatively weaker lower embedment constraint during early excavation stages. This sequential activation enhances the global stiffness of the retaining structure and limits lateral deformation. Simultaneously, bending moments concentrate near the soil–rock interface, where the stiffness contrast is most pronounced.

Parametric analyses further clarified the dominant factors controlling the structural responses. The degree to which the underlying rock has been weathered directly influences toe restraint stiffness. Increased weathering weakens embedment constraints, resulting in amplified wall deformation and elevated strut axial forces, whereas moderately weathered rock provides a more favorable balance between stiffness and constructability. The inclination of the soil–rock interface introduces asymmetric boundary conditions and an uneven stiffness distribution, leading to differential deformation and a greater internal force demand on the side with thicker overburden. For foundation pits constructed in inclined composite strata, asymmetric structural design with enhanced stiffness on the thicker soil side is required. Monitoring efforts should likewise focus on lateral displacement and strut axial forces on this critical side. In addition, the identified effective embedment depth range of 4.0–4.5 m indicates that increasing embedment enhances anchorage performance only within a limited interval; beyond this range, structural benefits diminish as the stiffness contribution approaches saturation.

To clarify the fundamental mechanical mechanism of the RSEP in a soil–rock binary system under controlled conditions, the stratigraphy was simplified into soil and rock layers. By comparing this simplified model with a multi-layer stratigraphic model (Figure 19), we found that the two models exhibit similar deformation and internal force responses, indicating that this simplification represents a conservative analytical approach. Since the stratigraphic parameters, structural configuration, and excavation sequence were derived from an actual engineering project for design optimization purposes, the analysis reflects realistic boundary and loading conditions, enhancing practical relevance. Nevertheless, certain idealizations remain. In the physical model tests, sand was adopted as the equivalent material for the overlying soil by neglecting minor cohesion according to similarity principles, and this approach may have led to a slight overestimation of the influence range of surface settlement. Moreover, the homogenization of the rock mass idealizes mechanical continuity and may somewhat overstate the anchoring effect at the retaining structure toe. These potential systematic deviations should be considered when interpreting the results.

The findings are consistent with existing research on soil–rock composite foundation pits and further deepen our understanding of rock-socketless support mechanisms. The quantified relationships between rock weathering degree, interface inclination, embedment depth, wall deformation, and strut axial forces provide practical references for parameter selection in similar geological settings. Based on the identified deformation and internal force evolution patterns, in similar projects, we recommend prioritizing systematic monitoring of ground surface settlement, lateral wall displacement, and strut axial forces during construction and early operation stages. These parameters directly reflect stiffness redistribution and structural demand evolution, enabling dynamic stability assessment and timely risk warning.

Despite these findings, several limitations should be acknowledged. The conclusions are derived from specific geological conditions and excavation dimensions, and their applicability to other projects requires adjustment of mechanical parameters and geometric scales. This study focuses on static excavation stages and does not consider dynamic loading effects. Future work will focus on the mechanical response and stability control of this support system in deeper and larger foundation pits, the system’s interaction and failure mechanisms in fractured or jointed rock masses, its impact on adjacent existing structures, and its performance under dynamic loading, and validate and calibrate the model through additional field-monitoring data from similar projects [46].

8. Conclusions

Based on a soil–rock composite foundation pit project carried out on Jinan Metro Line 7, a numerical model of an RSEP with internal struts was established to simulate the excavation process. Analyses of ground settlement, wall displacement, bending moments, and strut forces led to the following conclusions.

• The support system demonstrates good stability, effectively controlling deformation. The numerical results reveal a “trough-shaped” settlement profile within 20 m of the pit, with the pattern of wall displacement evolving from being “inverted triangular” to “vase-shaped,” with a maximum excavation depth of 0.368‰. Model tests recorded a peak wall top settlement of 1.84 mm.
• Internal forces follow clear patterns. The bending moment peaks about 1 m above the soil–rock interface, with abrupt changes at strut locations. Strut axial forces increase with excavation, showing inflection points upon installation of new struts, confirming their effectiveness. In practical engineering, special attention should be paid to the soil–rock interface, and local reinforcement measures should be adopted, such as enhancing the support stiffness in the vicinity or strengthening the connection points of the nearby support structures.
• Key factors significantly influence performance. Less weathered rock (moderately weathered) provides optimal constraint, minimizing deformation and strut forces. A steeper soil–rock interface dip (up to 10°) increases loads and deformation on the thicker-soil side but reduces them on the opposite side, resulting in greater overall strut forces than a horizontal interface. Increasing the unsupported length from 3.5 m to 4.5 m reduces deformation, with 4.0–4.5 m being a critical effective embedment interval where load transfer shifts from struts to deep rock.
• The effectiveness of the rock-socketless system was validated for soil–rock composite strata, overcoming the need for a rock shoulder and balancing safety with economy. The findings provide practical guidance for similar projects in Shandong and analogous regions.