Stability Control Method and Field Testing of High Embankment with Red Bed Soft Rock and Soil Stone Mixture Fill Roadbed

Abstract

Post-construction settlement in embankments is a crucial quality indicator and a significant factor influencing the long-term stability of roadbeds. Especially for the mixed-fill materials, by considering the uncertainty of composition and mechanical properties, it is important to predict and take construction measures to control post-construction settlement. In this paper, taking the construction of high-fill embankments with red bed soft rock mixture as the background, the deformation characteristics of mixed-fill materials were revealed first. Then, a dynamic–static coupling method for roadbed filling was proposed, and corresponding control parameters were provided. Finally, by employing ABAQUS 2016 for long-term settlement numerical simulations and conducting load-bearing preloading tests, the deformation patterns of the high embankment with red bed soft rock mixture fill roadbed were revealed.

1. Introduction

Red bed formations are extensively employed in various engineering projects, such as highways, railways, water resources, and hydropower applications [1,2].

Different from typical filling materials such as rocky soil and industrial slag, which have high friction coefficients, are easily compressed, and exhibit good permeability, red bed soft rock undergoes significant variations in strength due to differences in mineral composition and cementing materials. Particularly in atmospheric conditions or under the influence of wet–dry cycles, rock blocks can disintegrate into soil or even become mud. Therefore, its use as a filling material for highways can easily lead to roadbed settlement.

Engineers and researchers have addressed specific challenges related to the unique physical properties of red layer formations, including efforts to improve the use of red layer soft rock as a filling material [2]. Liu et al. conducted numerical simulations of the stability of red layer soft rock slopes in the Zhetang channel area of the Xinjiang reservoir using the Mohr–Coulomb criterion. The results validated that the safety factor of red layer soft rock slopes met design requirements [3].

In previous research, the road performance of red bed soft rock as a filler has been extensively examined through field tests, involving the analysis of the chemical composition of typical red bed soft rock and its correlation with disintegration resistance and physical properties [4]. Zhao et al. proposed a transformative approach for red bed soft rock, involving a sequence of ‘pre-disintegration-raking-compaction,’ effectively eliminating the water sensitivity of red layer soft rock and creating a dense, impermeable embankment. This method ensures the water stability and strength of the roadbed, meeting the highway roadbed index requirements [5]. Liu et al. [6] simulated four kinds of red bed embankments with optimal water content and compaction degrees of 87%, 90%, 93% and 97%, respectively, through centrifugal model tests, obtained settlement, compaction degree and acceleration, and analyzed the relationship between construction period and post-operation settlement. Su et al. [7] simulated the natural disintegration process of red bed soft rock in the atmosphere by using the circulation methods of sprinkling and blowing, and conducted particle screening and quality analysis of the disintegration products, drawing the particle mass change curve under different particle size conditions. On this basis, the control index of construction technology is put forward, which can be used to guide the construction of a highway embankment project to control the quality of roadbed. Qing et al. [8] used geotechnical centrifugal model tests to analyze the settlement characteristics of red bed soft rock fillers under different compaction degrees, reasonably determined the compaction parameters of embankment filling, and verified the suitability of red bed soft rock fillers for embankment filling. Through a series of laboratory tests, researchers tested the expansion rate, triaxial shear strength and compression modulus of red bed soft rock under subgrade load, studied the change rule of CBR load ratio, compaction properties and unconfined strength of red layer soft rock, and verified the applicability of using red layer mudstone as the filling material of high-speed railway subgrade [9,10].

Numerous studies on red bed soft rock subgrade materials and their parameters have been conducted both in China and internationally [11,12,13]. Due to limitations imposed by on-site conditions during construction operations, there is a current need to propose a method for the quantitative assessment of the physical and mechanical characteristics of soil–rock mixtures when using red bed soft rock as a filling material. Additionally, specific control methods for ensuring the safety and stability of high-fill embankments tailored to the unique features of mountainous roads need to be developed.

In this paper, taking the 50 m high embankment construction on Chu Yao Expressway in Yunnan Province as the background, based on the strength and deformation characteristics testing of the soil–rock mixed materials, a dynamic–static coupling method for roadbed filling was proposed, and corresponding control parameters were provided. Meanwhile, the long-term settlement numerical simulations and load-bearing preloading tests were conducted, and the deformation change laws of high embankments with red bed soft rock mixture fill roadbeds were revealed.

2. Construction Process of High-Embankment Roadbed

High-embankment roadbeds, typically exceeding 20 m in height, demand stringent quality control to ensure that post-construction settlement adheres to standards and prevents issues like road surface cracking and water seepage that could compromise roadbed stability [14,15]. For the construction control of red bed fill materials, in accordance with China’s leading “Technical Specifications for Highway Subgrade Construction” (JTG 3610—2019) [16], the widely adopted construction process is based on the work of Zhao et al., who proposed a construction methodology centered around the core sequence of “pre-disintegration-raking-compaction” based on on-site experiment [5].

2.1. Key Embankment Construction Processes

The embankment construction discussed in this paper involves the use of a crushed or weathered red layer soft rock fill material sourced from fragments generated during the excavation of nearby roadbeds and tunnels. These fragments consist of abandoned pieces composed of the typical soft red bed rock found in the Yunnan region. Mechanical crushing and pre-breaking processes are applied to these fragments, forming a soil–rock mixture filling material, as illustrated in Figure 1.

Leveraging a self-developed testing system and advanced high-energy compaction equipment, control of the soil–rock mixed-fill compaction is conducted using parameters such as rebound modulus and deflection. The average stone content in this fill is approximately 45%, with stones up to 30 mm in size, and a loose layer thickness of 50 cm. High-energy impact compaction is applied every 4 m in height.

Following the acquisition of the fill material, the high embankment adopts a layered construction process, as shown in Figure 2. During the layered construction, thorough compaction is essential to ensure the strength, rigidity, and water stability of the embankment body.

2.2. Sprinkling and Compaction

Red bed soft rock exhibits characteristics of disintegration when exposed to water. In the construction of soil–rock mixed-fill roadbeds, to enhance compaction and improve the rolling effectiveness of the soil–rock mixture, it is necessary to increase the water sprinkling appropriately. For soil–rock mixtures with stone content below 50%, water usage can be controlled at 3% to 5% by weight, as shown in Figure 3.

2.3. Layered Compaction and Dynamic Compaction

In soil–rock mixed-fill roadbed construction, the compaction process is a critical step to ensure the effective compaction of the soil–rock material [17]. Typically, a construction approach involving layered compaction and quantitative dynamic compaction is employed. The specific steps include site clearance, maintaining site-level evenness, layered compaction, and quantitative dynamic compaction [18,19,20]. Quantitative dynamic compaction involves establishing compaction elevations, marking compaction point locations, measuring site elevations, positioning the compaction machine, measuring the pre-compaction hammer head elevation, performing compaction, and measuring the hammer head elevation.

Dynamic compaction is an indispensable and crucial step in soil–rock mixed-fill roadbed construction as it effectively enhances the density and stability of the soil–rock mixture. To determine the number of compaction cycles, on-site trial compaction is generally conducted by measuring settlement and surrounding uplift. Furthermore, it is essential to control the layer thickness to ensure the final compaction effect. Typically, in practical construction, the spacing between compaction points can be chosen within the range of 5 m to 10 m based on engineering experience. To establish the process control parameters during the construction, on-site construction process trials should be conducted to determine reasonable parameter values.

3. The Dynamic–Static Coupling Compaction Method and Control Parameters

To analyze the impact of construction control parameters on the final compaction effect of high-embankment roadbeds, prior to roadbed construction, dry density and gradation curves of loose soil–rock materials are obtained through particle classification tests. During the filling process, a rational compaction process is determined by employing three compaction techniques: high-frequency vibration, static pressure, and low-frequency vibration, along with adjustments in pavement thickness.

3.1. The Impact of Different Compaction Methods on Compaction Density

The comparative results of the compaction density variations using different compaction methods are shown in Figure 4 and

Table 1. Variations in subgrade compaction for different rolling methods.

Compaction Passes	Compaction Density/%
	High-Frequency Vibration	Static Pressure	Low-Frequency Vibration
2	83.38	90.78	78.17
4	93.87	90.34	88.81
6	100.94	93.19	95.72
8	100.28	93.37	99.14
10	101.23	95.3	102.53
12	97.3	97.44	-

. It can be observed that, with an increase in compaction passes, the compaction density gradually increases for various compaction methods and eventually stabilizes. Among them, static compaction exhibits more stable variations, resulting in the highest final compaction density.

Furthermore, in comparison to vibratory compaction, static compaction generates a higher initial compaction density during the initial compaction phase. The compaction of loose roadbed fill materials is a crucial aspect of roadbed engineering, effectively improving the bearing capacity and stability of the roadbed. When applied to roadbed fill materials, static compaction, in particular, holds significant advantages, producing a smoother road surface and effectively filling the gaps between coarse and fine particles.

After multiple compaction passes, vibratory compaction proves more effective in balancing the porosity of the soil–rock mixture. Therefore, in practical construction, a combination of compaction methods should be selected based on specific conditions. To achieve better compaction density, a combination of static and vibratory compaction, along with quantified high-energy impact compaction, can be employed. These methods ensure the uniformity of roadbed fill materials, achieve excellent compaction, enhance bearing capacity, and improve stability.

3.2. The Impact of Dynamic Compaction Layer Thickness

As previously mentioned, in roadbed construction, layer thickness is a crucial factor in roadbed construction, impacting both construction progress and compaction quality. Excessive thickness can hinder the effective transmission of compaction effects, while insufficient thickness can lead to effective compaction but may extend the construction period and escalate costs. Therefore, in the embankment construction experiment, maintaining the use of the static compaction method, layer thicknesses of 30 cm, 40 cm, and 50 cm were employed. The variations in deflection, compaction density, and porosity were then examined, as detailed in

Table 2. Comparison of data for different dynamic compaction thicknesses.

Layer Thicknesses/cm	Deflection/0.01 mm	Compaction Density/%	Porosity/%
30	118.34	97.71	2.71
40	124.34	93.86	4.06
50	131.66	91.09	6.08

.

Referring to Figure 5, it becomes apparent that an increase in layer thickness leads to higher deflection and porosity values. Specifically, deflection increases from 1.18 mm to 1.32 mm, and porosity rises from 2.71 to 6.08. However, compaction density decreases from 97.71% to 91.09%. This suggests that when the layer thickness is relatively small, applying the same external load to a thinner roadbed results in smaller inter-particle gaps, thereby achieving more efficient compaction.

Taking into account these three indicators and assuming the availability of high-energy impact compaction, it is recommended to choose a layer thickness of 30 cm to achieve optimal compaction. Specifically for soil–rock mixed-fill embankments, especially those incorporating red bed soft rock, employing a construction method involving high-energy impact compaction in passes of approximately 4 m in height is advisable. This approach allows for adjusting the layer thickness to 50 cm, effectively controlling post-construction settlement, improving construction efficiency and benefits, and ensuring compaction quality.

4. Long-Term Settlement Prediction Analysis

To explore the long-term settlement patterns of high-embankment roadbeds, this section focuses on the construction of a high-embankment roadbed in the context of the Chu Yao Expressway in Yunnan, specifically the section from K67+380 to K69+183.967.

As shown in Figure 6, a typical cross-section located at K68+948.243 is selected for analysis. Numerical simulations of long-term post-construction settlement are conducted using ABAQUS 2016 finite element analysis software.

ABAQUS is a powerful general-purpose finite element software with a rich set of built-in constitutive models suitable for geotechnical materials. These include models such as the Mohr–Coulomb model, extended Drucker–Prager model, and others. ABAQUS can automatically generate self-weight stress fields, perform seepage–stress coupled analysis, and conduct saturated and unsaturated flow analyses.

In this segment, the highest point of the embankment exceeds 50 m in height, with a total length of 1803.967 m. The roadbed fill material consists of crushed or weathered red bed soft rock, and the construction process follows the procedures described in Section 2, resulting in a soil–rock mixed fill of red bed soft rock after compaction. To ensure the quality of the roadbed construction, adjustments are made to the moisture content of the fill material during the construction process. Trials are conducted to determine the optimal moisture content to guarantee compaction quality, and strict control measures are in place to manage post-construction settlement.

The underlying red bed soft rock foundation has a relatively minor impact on roadbed settlement and is assumed to be uniform. Given that the Mohr–Coulomb nonlinear elastic model is widely used in geotechnical calculations, this model is an incremental elasto-plastic model that determines the failure envelope by employing shear yield functions and tensile yield functions, ensuring that the tensile stress flow rule is associated while not being correlated with shear flow. Based on the on-site testing by the construction unit and preliminary geological investigation by our research team, the Mohr–Coulomb model parameters are as shown in

Table 3. Rock physical parameters.

	Unit Weight kg/m3	Elasticity Modulus /MPa	Poisson Ratio	Cohesion /kPa	Internal Friction Angle /°
Rock	2600	800	0.28	300	40

.

In the ABAQUS finite element analysis software, considering that the creep time scale of the red bed soft rock is on the same order of magnitude as the loading rate, it is necessary to simulate the coupling of creep and plasticity. For this purpose, the ABAQUS 2016 software includes a built-in time hardening and Drucker–Prager coupled creep model.

Referring to the formula principles provided in the ABAQUS user manual [21], the introduction is as follow.

When the applied stress on a material remains essentially constant and considering the time-hardening characteristics, the equivalent creep strain rate can be calculated using Equation (1).

$$\stackrel{•}{{\overline{\epsilon }}^{cr}}=A{\left({\sigma }^{cr}\right)}^n{t}^m$$

(1)

where $\stackrel{•}{{\overline{\epsilon }}^{cr}}$ is the equivalent creep strain rate, ${\sigma }^{cr}$ is the equivalent shear creep stress, and A, n, and m are material parameters used to describe the material’s creep behavior.

The equation can be integrated over time to yield Equation (2).

$${\overline{\epsilon }}^{cr}=\frac{A}{m+1}{\left({\sigma }^{cr}\right)}^n{t}^{m+1}$$

(2)

According to the definition of time hardening, to ensure the stress invariance characteristic during the creep process, it is necessary to divide the time history into several stages and assume that the stress remains constant within each stage, then solve it segment by segment. Therefore, when considering the influence of previous creep on subsequent creep, $\stackrel{•}{{\overline{\epsilon }}^{cr}}$ can undergo sudden changes while t remains constant.

The plastic yield surface uses a linear Drucker–Prager yield surface with a meridian (sublinear) function as given in Equations (3) and (4).

$$F=t-p\mathit{tan}\beta -d=0$$

(3)

$$t=\frac{q}{2}\left[1+\frac{1}{k}-\left(1-\frac{1}{k}\right){\left(\frac{r}{q}\right)}^3\right]$$

(4)

where t is the deviatoric stress parameter, q is the Mises stress, r is the deviatoric stress invariant, β is the tilt angle of the yield surface in the p~t stress space and is related to the friction angle φ, and k is the ratio of the strength from triaxial tensile tests to the strength from triaxial compression tests, reflecting the influence of principal stresses on yield.

When k = 1 and t = q, the yield surface is the Von Mises circle in the π-plane. To ensure that the yield surface is convex, it is required that 0.778 ≤ k ≤ 1.0. d is the intercept of the yield surface on the t-axis in the p~t stress space. It represents another form of cohesion and is related to the input hardening parameters. It can be determined as follows:

$$d=\left\{\begin{array}{l}\left(1-\frac{1}{3}\mathit{tan}\beta \right){\sigma }_c\\ \left(\frac{1}{k}+\frac{1}{3}\mathit{tan}\beta \right){\sigma }_t\\ \frac{\sqrt{3}}{2}\left(1+\frac{1}{k}\right)\tau \end{array}\right.$$

In a linear model, the expression for the plastic flow potential $G$ is given by Equation (5).

$$G=t-p\mathit{tan}\psi$$

(5)

In this equation, ψ represents the dilatancy angle in the p~t stress space. When ψ = 0°, it indicates that the volume of the plastic deformable material remains unchanged, whereas when ψ > 0°, the material exhibits dilatancy effects. When using the associated flow rule, ψ equals β, and when the value of k is 1, the linear Drucker–Prager model degenerates into the classical Drucker–Prager model [22,23].

Therefore, we can combine the time-hardening process with the Drucker–Prager yield-failure criterion and its associated flow rule to jointly formulate the time-hardening and Drucker–Prager plastic-coupled creep model of the soil [24,25].

The values of the relevant parameters are shown in

Table 4. Time-hardening creep model parameters (different degrees of compaction).

Compaction Density	σ0/kPa	A	n	m	R2
85%	50	1 × 10−8	1.21398	−0.72969	0.98891
	100	1 × 10−8	1.26747	−0.82022	0.95937
	200	1 × 10−8	1.23795	−0.92080	0.88839
	400	1 × 10−8	1.22657	−0.92554	0.84951
90%	50	1 × 10−8	1.11915	−0.82803	0.92200
	100	1 × 10−8	1.15649	−0.75232	0.92676
	200	1 × 10−8	1.15450	−0.87983	0.93797
	400	1 × 10−8	1.13348	−0.91752	0.89378
95%	50	1 × 10−8	1.10226	−0.88783	0.81795
	100	1 × 10−8	1.13559	−0.74755	0.96733
	200	1 × 10−8	1.14587	−0.81434	0.95700
	400	1 × 10−8	1.11372	−0.90938	0.90810

and

Table 5. Linear Drucker–Prager model parameters.

	Unit Weight kg/m3	Elasticity Modulus /MPa	Unit Weight kg/m3	β (°)	ψ (°)	σc (kPa)	k
Fill material	2080	38	0.30	38	38	30.2	1

, where σ₀ represents the axial load from the compressive creep characteristic test, and A, n and m are material parameters for the time-hardening creep law. R² represents the coefficient of determination obtained from the fit.

Considering the Drucker–Prager model parameters for a plane strain problem with the assumption of k = 1 and using the associated flow rule where ψ = β, we obtain Equations (6) and (7).

$$\mathit{tan}\beta =\frac{\sqrt{3}\mathit{sin}\phi }{\sqrt{1+\frac{1}{3}{\mathit{sin}}^2\phi }},d=\frac{\sqrt{3}\mathit{cos}\phi }{\sqrt{1+\frac{1}{3}{\mathit{sin}}^2\phi }}c$$

(6)

$${\sigma }_c=\frac{1}{1-\frac{1}{3}\mathit{tan}\beta }d,{\sigma }_t=\frac{1}{1+\frac{1}{3}\mathit{tan}\beta }d$$

(7)

According to the above formula, β(ψ), σ_c(σ_t) can be obtained. Based on the parameters of the existing Mohr–Coulomb model, the parameters of the Drucker–Prager model were calculated, and the results are shown in Table 5.

The maximum long-term settlement (10 years) for the typical cross-section is shown in Figure 7. Based on numerical simulation results, the maximum pavement settlement-time curves for different degrees of compaction of the typical cross-section are illustrated in Figure 8.

From these two figures, we can observe that the relationship between the maximum roadbed settlement and the degree of compaction for the typical cross-section is nonlinear. Particularly, as the degree of compaction increases from 85% to 90%, the variation in roadbed deformation decreases. When the degree of compaction exceeds 90% and falls within the 90% to 95% range, the maximum pavement settlement remains relatively stable.

For red bed soft rock fill materials with different compaction densities, under the same load, lower compaction density results in a larger porosity and lower compaction degree. Macroscopically, this manifests as smaller stiffness, weaker resistance to deformation, larger ultimate creep strain, and, consequently, greater pavement settlement [26,27,28]. With an increase in compaction density, the material’s porosity decreases, compaction degree increases, macroscopically exhibiting higher stiffness, enhanced resistance to deformation, reduced ultimate creep strain, and, ultimately, smaller pavement settlement.

5. High-Embankment Settlement Preloading Test

To test the long-term settlement characteristics of the high-embankment roadbed, in this section, a preloading test was conducted for verification. Due to the considerable height of the high-embankment roadbed, a significant load was required for preloading. And the choice of loading method should be tailored to the specific environmental conditions. In this study, the water bag pre-compression method was employed, and the water source for the water bags was derived from a nearby reservoir. The height of the preloading was determined using pre-existing numerical simulation methods, which calculated the relationship between the embankment thickness and deformation of the high-embankment section. This process ultimately determined the load capacity and testing duration for the preloading test, as well as the number of water bags required.

5.1. Experimental Plan Simulation

As mentioned earlier, to determine the experimental parameters reasonably and consider the displacement monitoring requirements during the pre-compression, the ABAQUS analysis model described earlier was used to numerically simulate the pre-compression effect. The analysis included factors such as pre-compression time, settlement variation patterns, and monitoring range requirements.

The model used red bed soft rock fill material with a compaction ratio of 85% and σ₀ = 50 kPa. The typical duration for preloading tests is generally between 30 and 90 days. Therefore, the simulated pre-compression time was set to 90 days. Based on the numerical simulation results, graphs depicting the maximum settlement–time variation curve for different preloading heights and the relationship between settlement during preloading and preloading height were generated. The results are shown in Figure 9 and Figure 10.

As shown in Figure 9 and Figure 10, within the range of numerical simulation, there is an approximate linear relationship between the preloading height and the embankment settlement. To obtain an appropriate experimental settlement that fits this relationship, we can determine the preloading value by extrapolating along this linear relationship. At the same time, this helps avoid excessive settlement during preloading, ensuring that it complies with the embankment design requirements.

Additionally, based on the numerical simulation results for the typical cross-section, graphs were created to illustrate the maximum post-construction settlement of the road surface over time under different preloading heights and the relationship between them. The results are shown in Figure 11 and Figure 12. From these graphs, it can be observed that as the preloading height increases, the post-construction settlement generally decreases. When the preloading height is less than 1 m, there is a significant reduction in settlement after preloading. Hence, considering factors such as cost-effectiveness and site conditions, it is recommended to choose a preloading height between 1 and 4 m based on the embankment height and other relevant factors.

5.2. Preloading Design

• Load design

Based on the previous numerical simulation results, this plan proposes the use of water from a nearby reservoir for preloading the embankment. Calculations are made based on a maximum fill height of 20 m, and the compacted soil–rock mixed-fill material is assumed to have a maximum density of 2.0 g/cm³. If the preloading head is controlled at 5–10% of the total mass of the high embankment, then the preloading height using reservoir water would be approximately 2.0 to 4.0 m for a maximum fill height of 20 m. Similarly, if the maximum fill height is 10 m, and the material density is 2.0 g/cm³, the preloading height using reservoir water can be set at 1.0 to 2.0 m.

2.

Preloading design parameters

The thickness of the on-site water-stabilized layer is 38 cm, and the width is 11.25 m. For the preloading design, an 11 m width is used, and the density is calculated at 2.0 g/cm³, equivalent to an actual head of 0.76 m. After comparison, PVC water bags are chosen for the preloading. The dimensions of the water bags are 11 m in length, 5 m in width, and 1.3 m in height. The maximum water filling capacity is 66 tons, requiring a total of 10 water bags. Calculations for the preloading water head, based on an average embankment height of 20 m, set a lower limit of 2.0 m. In addition to the embankment’s bearing capacity, a water filling height of 1.2 m is also set, and it can be adjusted in three load levels: 0.8 m, 1.0 m, and 1.2 m.

The placement of water bags and drilling is shown in Figure 13. The preloading section covers milepost K68+480 to K68+530. After placing two water bags from the starting point, a deep borehole is positioned at K68+490, with its center 1 m away from the hard shoulder. Calculations based on the embankment depth reveal that the filling depth at K68+490 is 36 m, the drilling depth is 46 m, and the installation of settlement rings requires 41 rings. Similarly, after placing two more water bags, a second deep borehole is located at K68+511, with its center 1 m away from the hard shoulder. The filling depth at K68+511 is calculated to be 30 m, the drilling depth is 40 m, and 35 settlement rings are required for installation. The total drilling depth is 86 m, and a total of 76 settlement rings are installed.

Surface displacement monitoring piles are installed at two locations: K68+490 and K68+511, with the pile centers positioned 2 m away from the hard shoulder, as shown in the design style in Figure 14.

Each water bag measures 11 m × 5 m × 1.3 m and weighs 180 kg when empty. They can hold a maximum of 66 tons of water. In total, 660 tons of water are needed, which can be sourced from a nearby reservoir.

5.3. Implementation Process of Preloading at the Site

The on-site implementation process of preloading is as follows:

(1)

Preloading is carried out on the left side of the roadbed while the right side remains open to traffic. The preloading area is fenced off to prevent collisions or damage.

(2)

First, two deep boreholes are drilled at the specified locations as shown in Figure 13. The borehole size is 108 to 130 mm in diameter, and mud wall protection is applied. Deep borehole observation pipes are installed, filled with fine sand to fill the voids between the observation pipe and the borehole, and the first horizontal displacement and vertical settlement observations are conducted.

(3)

Two surface displacement monitoring piles (boundary piles) are installed at the specified locations as shown in Figure 14.

(4)

Clear any hard debris such as rock blocks that may cause damage to the water bags in the stabilized layer. Lay a layer of non-woven fabric on the preloading section to prevent stones from puncturing the water bags. Mark the direction of the water bags and spread them evenly from the center of the designated location towards the periphery. Inflate the water bags using a blower and then fill them with water. Water is injected to three different load levels: 0.8 m, 1.0 m, and 1.2 m. The injection ports are placed on one side of the central median.

(5)

Leave a gap of more than 50 cm on both sides of the deep borehole locations for pedestrian access and deep hole monitoring.

(6)

Lay non-woven fabric and water bags one by one and inflate and fill them with water using the blower.

(7)

Conduct a water injection test with a height of 0.8 m for the first 10 water bags while simultaneously monitoring deep borehole and slope displacements. Adjust the water head height to 1.0 m or 1.2 m based on the monitoring results as needed.

(8)

The preloading period is determined comprehensively based on different fill thicknesses and preload loads, combining theoretical calculations with monitoring results, and is expected to be between 30 and 120 days.

(9)

After preloading is completed, recover the water from the water bags and observe the rebound of the fill.

The real-time image of the preloading process is shown in Figure 14, and deformation monitoring of the roadbed is conducted during the loading process.

6. Analysis of Field Monitoring Results

The on-site monitoring work began after filling the water bags and lasted for 48 days. The first deep-hole monitoring was conducted to obtain baseline data, and the arrangement of monitoring points is shown in Figure 15. On the 14th day, the water bag filling reached a height of 0.8 m, and the second deep-hole monitoring was carried out. On the 30th day, the water bag filling reached a height of 1.2 m, and at the same time, 14 additional surface-monitoring points were set up. The sixth round of monitoring was conducted, as shown in Figure 16a. The left diagram displays the initial positions of various monitoring points in the vertical direction, while the right diagram shows the cumulative settlement displacement values of each point after reaching a filling height of 1.2 m, with a significant cumulative settlement displacement in the 20 m depth range. As illustrated in Figure 16b, the left diagram represents the initial positions of various monitoring points in the horizontal direction, while the right diagram presents the cumulative horizontal deformation values of each point after reaching a filling height of 1.2 m. On the 35th day, the seventh round of monitoring was conducted, as shown in Figure 17. On the 44th day, the eleventh round of monitoring was conducted, as shown in Figure 18. Finally, on the 48th day, the last round of monitoring was performed, as depicted in Figure 19.

By comparison with Figure 17 and Figure 18, it can be seen that the maximum variation of deep settlement in hole 1 is about 40 mm, the maximum deep settlement point is about 25 m, and other monitoring points are generally less than 25 mm. Most of the locations with large settlement deformation are located in the upper part of the roadbed with a hole depth above 15 m. The maximum depth settlement change in hole 2 is about 23 mm, the maximum deep settlement point is located at 4 m~7 m, and most of the locations with large settlement deformation are located in the upper subgrade with a hole depth of more than 15 m. The maximum change in deep horizontal displacement in hole 1 is about 3.5 mm, and the maximum deep horizontal displacement point is located at a hole depth of 22 m~25 m. The maximum depth horizontal displacement of hole 2 is about 3.5 mm, and the maximum depth horizontal displacement is located at about 26 m of the hole depth. Most other monitoring points with large horizontal displacement changes are located at the upper part of the roadbed with a hole depth of more than 20 m. The maximum surface settlement monitoring point is located near the monitoring point on the southernmost side of the slope, which is about 40 mm.

The on-site preloading test demonstrated that the settlement and deformation of the high-embankment fill fell within the range of 23 mm to 40 mm. These settlement values can be effectively managed through suitable compaction methods and settlement allowances, thereby alleviating potential post-construction settlement issues. The monitoring results closely correlated with the numerical simulation findings, affirming the successful control of post-construction settlement in this challenging ultra-high-fill scenario.

Long-term observations conducted two years after the road’s opening, along with the driving experience, were consistent with the earlier test predictions. This validates the accuracy and reliability of the applied methods in ensuring the stability of the high-embankment fill.

7. Conclusions

Considering the context of constructing a 50 m high embankment in Yunnan Province, China, and a thorough investigation into the deformation characteristics of soil–rock mixture fill materials, a comprehensive approach was adopted to manage long-term settlement. In this paper, taking the construction of high-fill embankments with red bed soft rock mixture as the background, the deformation characteristics of mixed-fill materials were revealed first. Then, a dynamic–static coupling method for roadbed filling was proposed, and corresponding control parameters were provided. Finally, the long-term settlement numerical simulations and load-bearing preloading tests were conducted, and the deformation change laws of high embankments with red bed soft rock mixture fill roadbeds were revealed. And the main conclusions are as follows:

(1)

The cohesion and internal friction angle of the fill material will affect compaction efficiency and the development of long-term settlement; especially for the cohesion values, it will lead to the high-embankment failure.

(2)

To ensure the quality of high-fill embankments with soil–rock mixture fill, the paving layer thickness should be less than 40 cm, and the rolling times using the coupling dynamic and static compaction method should be more than 6.

(3)

On-site preloading tests demonstrate that the settlement and deformation of high-fill embankments range from 23 mm to 40 mm. Proper measures, such as super-elevation to accommodate settlement, can balance the post-construction settlement. The monitoring results align closely with the numerical simulation results, confirming that controlled settlement is achieved for the super-high-fill embankment. Long-term observations two years after opening to traffic and driving experiences match the earlier test predictions.

While significant advancements have been achieved in the construction of a 50 m high embankment in Yunnan Province, China, it is important to acknowledge potential limitations stemming from geological and soil variations. Future research endeavors should aim for broader regional applicability, encompassing additional influencing factors, and incorporate advanced monitoring technologies to enhance this study’s practicality and generalizability.