1 Introduction

Novel load-reduction methodologies are imperative for deeply buried structures like high-filled cut-and-cover tunnels (HFCCTs), particularly regarding the enduring efficacy of their load-reduction mechanisms. Despite numerous proposed and investigated load-reduction methods [1,2,3,4], scant research has illuminated their long-term viability. The conventional soil-arching effect offers a solution for mitigating excessive earth pressures from high backfills. However, little attention has been given to how the time-dependent fluctuations of load-reduction techniques, coupled with soil creep, might precipitate structural failure in cut-and-cover tunnels (CCTs). The waning efficacy of load-reduction is often attributed to specific materials [4,5,6,7,8], including backfill soils, geogrids, and relatively low-compacted (RLC) soil. Prior explorations of potential load-reduction materials have involved expanded polystyrene and combined with concrete wedges [9], tire-derived aggregates [10, 11], and RLC soil [12]. Some studies have investigated how such materials, when utilized in highly compressible layers, induce the soil-arching effect, thereby reducing load factors, e.g., above pipelines [13]. Several research endeavors have demonstrated that geogrids, by furnishing tensile strength through interactions with soils, can bolster load reduction [7, 14, 15]. However, these methodologies primarily assume constant load reduction effects stemming from soil arching and tensile membranes. Conversely, research on long-term soil behavior and the time-dependent efficacy of load-reduction methods remains scarce. In contrast to expanded polystyrenes, RLC soil not only offers a locally sourced and environmentally friendly option but also exhibits enduring load-reduction capabilities [8].

Recent research indicates that the effectiveness of load-reduction measures is significantly compromised by soil creep, posing a threat to the long-term stability of CCT structures [16, 17], Guo et al., 2022; Guo et al., 2023). Moreover, apart from backfill soil, the creep behavior of materials employed for load reduction warrants attention [18,19,20]. However, the effects and mechanisms of various load-reduction materials remain unclear. In a previous study by [8], findings revealed that when loess backfill is utilized, the structural pressure of RLC soil load reduction increases over time, contradicting long-term load reduction objectives and compromising the stability of buried structures.

Li et al. further observed that soil creep markedly compromised the load-reduction capabilities of compressible soil layers [8]. Their study highlighted the pronounced creep observed in on-site loess from northwestern China when employed as backfill, following primary soil consolidation. Moreover, the utilization of RLC soil for load reduction exhibited uneven stress distribution atop an HFCCT.

In this study, our aim is to propose a novel geotechnical approach to enhance long-term load-reduction effects. Dry sandy gravel, known for its minimal creep deformation compared to loess, emerges as a promising backfill material for HFCCTs. Soil creep behaviors are inherently complex and become even more intricate when considering time-dependent factors. Leveraging available data on soil mechanical properties, including backfill soils and compressible layers for RLC soils, we conducted a series of parametric studies using the Fast Lagrangian Analysis of Continua (FLAC3D), a finite-difference program. In our numerical analysis, we account for both the creep characteristics of backfill and those of RLC soils to gain insights into how soil creep influences long-term load-reduction mechanisms. When interpreting the results, our focus lies on two primary factors: variations in internal forces and backfill deformation. Subsequently, we analyze and compare three effective load-reduction methods. Hereafter, we present the long-term load-reduction findings for HFCCTs utilizing borrowed soil and assess the efficacy of RLC soil, specifically dry sandy gravel. Additionally, we discuss the feasibility of implementing this novel method in HFCCT design.

2 Parametric Studies

Parametric studies were conducted based on the available indices and mechanical properties of soils. Notably, two soil properties were carefully considered: backfill soil and RLC soil.

2.1 Properties of Backfill Soil

For the backfill material in this study, dry sandy gravel with a median diameter (D50) of 0.6 mm and an average unit weight, γ = 16.28 kN/m3, was utilized. Table 1 presents the index properties of the soil [21] and the physical characteristics of the dry sandy gravel samples utilized in this study [8]. The dry sandy gravel consisted of 77% gravel and 23% sand.

Table 1 Parameters of Dry Sandy Gravel
Full size table

2.2 Properties of RLC Soil

The parameters of the studied RLC soil in Table 2 are sourced from [22]. The RLC soil comprises loess with a moisture content of 15.7% and a relative compaction of 85%. Given that the Burgers model was employed to capture the creep behavior of loess, determining the parameters of the loess based on initial stress is necessary. Obtaining the accurate initial stress of RLC soil entails iterative calculations. The specific steps to obtain this figure are outlined as follows:

Table 2 Parameters of RLC Soil
Full size table

Step 1: Calculate the initial stress, which equates to the self-weight of the RLC load-reduction layer, and assign the corresponding Burgers model parameters to the RLC load-reduction layer;

Step 2: Utilize FLAC3D to compute the vertical earth pressure upon completion of backfill;

Step 3: Define the vertical earth pressure at the backfill completion as the initial stress. Referencing the parameters in Table 3, obtain the creep parameters corresponding to the initial stress through interpolation and apply them to the RLC load-reduction layer. Recalculate the new vertical earth pressure using FLAC3D;

Table 3 Program for simulations
Full size table

Step 4: Compare the FLAC3D-calculated vertical earth pressure obtained in Step 2 with the corresponding pressure obtained in Step 3. If the difference in earth pressure is less than 5%, designate the vertical pressure in Step 3 as the initial stress. If the difference exceeds 5%, repeat Steps 2 and 3 until the earth-pressure difference falls below 5%.

2.3 Geogrid Properties

The viscoelastic property of the geogrid is described by Eq. (1), as presented in [23]. The viscoelastic parameters of the geogrid can be derived from Eq. (1). Embedding Eq. (1) into the creep model command stream using FISH (a programming language integrated into FLAC3D) allows for the simulation of geogrid creep.

$${\text{R}} = 108.18\varepsilon^{0.62966} \frac{1}{{1{ + }\left( {{\raise0.7ex\hbox{$t$} \!\mathord{\left/ {\vphantom {t {0.003}}}\right.\kern-0pt}\!\lower0.7ex\hbox{${0.003}$}}} \right)^{0.01} }} = 108.18\varepsilon^{0.62966} \frac{{1}}{{1{ + }2.2505t^{0.01} }}$$
(1)

In this formula, R = force per unit width, kN/m, ε = stress per unit width, t = time, h.

3 Numerical Simulation

A series of parametric studies were undertaken utilizing FLAC3D. The simulation of material creep behavior was achieved using FISH code across different loading stages. Consistent with prior research [8], this study examines backfill consisting of dry sandy gravel instead of loess. Additionally, various combinations of enhanced materials were compared to assess their effectiveness in load reduction.

3.1 Case Setup

Parametric studies were conducted for four cases, and their respective conditions are summarized in Table 3. These cases include: Case W1 (no load-reduction method applied), Case W2 (RLC soil only above the top of CCT), Case W3 (a geogrid layer based on Case W2), and Case W4 (concrete columns on both sides of CCT based on Case W3).

3.2 Model Establishment

Although the load-reduction methods vary, all models essentially consist of soil combined with a distinct load-reduction material. The materials in Case W4 exhibit greater complexity compared to those in the other three cases. Figure 1 illustrates the schematic diagram of Case W4. The numerical model dimensions for Case W4 are as follows: 101 m wide, 99 m high, and 10 m thick. The CCT measures 12.8 m in width, 11 m in height, and features side slopes inclined at 70 degrees. The backfill atop the CCT is segmented into 6 layers. In Case W2, RLC soil (highlighted in red) replaces a portion of the first layer of backfill. In Cases W3 and W4, geogrid (depicted in black) covers the top of the first layer. In Case W4, concrete columns (shown in light gray) are positioned atop the structure from Case W3.

Fig. 1
figure 1

Load-reduction model for a high-filled cut-and-cover tunnel (HFCCT) in Case W4 (unit: meters)

Full size image

In the numerical analysis conducted using FLAC3D, proper arrangement of measuring points is essential. Figure 2 illustrates the layouts of the measuring points. A total of 21 measuring points are positioned along the A-A section to monitor the vertical displacement distribution of the backfill, with a default spacing of 1 m between each measuring point. Similarly, another 21 measuring points are arranged along the B-B section to monitor the distribution of vertical earth pressure, with a spacing of 1 m between each point. The midpoint of the B-B section is utilized to track variations in vertical earth pressure over time. Additionally, 12 measuring points are situated adjacent to the lining structure to record changes in lateral earth pressure over time. Furthermore, this study investigates the analysis of the CCT's internal forces resulting from the load-reduction methodology and soil creep behavior. To facilitate this investigation, the bending moment, shear force, and axial force of the CCT are examined at specific locations. To calculate the internal forces of the CCT structure during creep, the measuring points forming black circles in Fig. 2 are utilized. These measuring points predominantly monitor locations such as the vault, spandrel, haunch, and skewback of the CCT.

Fig. 2
figure 2

Arrangement of measuring points for HFCCT

Full size image

3.3 Properties of Materials

Since the post-deformation of the HFCCT's foundation is relatively minor, creep deformation can be disregarded, leading to the adoption of the Mohr–Coulomb model. Furthermore, the creep of the CCT's structure and the surrounding slopes is negligible. Hence, the CCT's structure, concrete columns, and slopes on both sides are assumed to be linearly elastic. The parameters for these components are detailed in Table 4.

Table 4 Parameters of other materials
Full size table

In this paper, the backfill comprises dry sandy gravel and RLC soil predominantly composed of loess. Based on findings from [22], the Burgers model is employed to simulate the long-term behavior of loess. Additionally, insights from [23] suggest that the hyperbolic model can effectively describe the rheological characteristics of backfill. The CCT and slopes are simulated using an elastic model, while the foundation is modeled using the Mohr–Coulomb model. Equation (2) illustrates the creep behavior utilizing the Burgers model.

$$\varepsilon = \frac{\sigma_0 }{{E_{\text{M}} }} + \frac{\sigma_0 }{{\eta_{\text{M}} }}t + \frac{\sigma_0 }{{E_{\text{K}} }}\left[ { - \exp \left( { - \frac{{E_{\text{K}} }}{{\eta_{\text{K}} }}t} \right)} \right]$$
(2)

The constants in this equation are Kelvin Young’s modulus (EK), Maxwell Young’s modulus (EM), Kelvin viscosity (ηK), and Maxwell viscosity (ηM). Equations (3) through (5) represent the hyperbolic model. The constants in these equations are the deformation at the moment of loading (ɛi), creep strain over time (ε(t)), the reciprocal of the final deformation (b), and the reciprocal of the initial slope of the strain curve (a).

$$\varepsilon \left( {\text{t}} \right) = \varepsilon_{\text{i}} + \varepsilon_{\text{c}} \left( {\text{t}} \right) = \varepsilon_{\text{i}} + \frac{{\text{t}}}{{\text{a + bt}}}$$
(3)
$${\text{b}} = \frac{{1}}{{\varepsilon \left( {\text{t}} \right)\left| {_{{\text{t}} \to \infty } } \right.}}$$
(4)
$${\text{a}} = \frac{{1}}{{\left( {{{{\text{d}}\varepsilon } / {{\text{dt}}}}} \right)\left| {_{{\text{t}} \to \infty } } \right.}}$$
(5)

The corresponding parameters are assigned to the model to achieve ground stress balance after model establishment. Subsequently, upon completing the balancing task, the soil displacement generated is eliminated, and the filling calculation commences. For the filling process, six layers of filler are positioned above the CCT, and layered filling simulation is conducted by transitioning each layer of filler from a null model to a solid constitutive model. Single-time fillings can lead to calculation errors. Therefore, when the filler undergoes creep for 400 days in the model, the daily variation of horizontal displacement between the bottom of the CCT's sidewall and the settlement above the CCT remains below the specified threshold (0.2 mm/day, 0.15 mm/day). Hence, the model's creep calculation time is defined as 400 days.

4 Analysis Results

Following the model establishment process and the analysis steps depicted in Fig. 3, this section will proceed to analyze the research findings.

Fig. 3
figure 3

Analysis process

Full size image

4.1 Earth Pressure

Below, we present the earth pressure on the top of CCTs, encompassing earth-pressure distribution and time curves.

4.1.1 Earth-Pressure Distribution

Figure 4 depicts the distribution of earth pressure on top of the CCT. Figure 4(a) showcases the earth-pressure contours at the onset of time and on the 400th day following the stabilization of creep. Across all cases, a stress-concentration area is evident above the CCT, with notable concentration observed in Cases W2, W3, and W4. However, the vertical earth pressure is lower in these cases compared to Case W1. Following the stabilization of creep, the earth pressure remains unchanged in Case W1, whereas the stress concentration in Cases W2, W3, and W4 exhibits a declining trend.

Fig. 4
figure 4

Earth-pressure distribution of the CCT: (a) earth-pressure contours, (b) variation curves with distance

Full size image

Figure 4(b) illustrates the variation curves of earth pressure with distance at the onset of time and on the 400th day following the stabilization of creep. Across all cases, the earth-pressure curves exhibit M-shapes. Compared to Case W1, the earth pressures on the top of the CCT experience significant reductions of 52.78%, 52.78%, and 56.3% for cases W2, W3, and W4, respectively, with further reductions of 21.4%, 21.4%, and 18.77% after the stabilization of creep. As a result of the load-reduction measures implemented above the CCT, the earth pressure above the CCT is transferred to both sides.

The results demonstrate that RLC soil can effectively diminish the earth pressure on the top of the CCT, with the load reduction effect amplifying with creep over time. Conversely, on the foundation, inadequate tension in the geogrid hindered the complete mobilization of the tensile membrane effect, leading to the absence of a load-reduction effect. However, with the addition of concrete columns on both sides, full mobilization of the geogrid becomes achievable, potentially inducing a tension effect that generates an upward lifting force on the top of the CCT. Furthermore, the load-reduction effect augments with creep over time.

4.1.2 Time Curves

Figure 5 illustrates the time curve for the earth pressure above the CCT. Figure 5(a) presents the time curve for the vertical earth pressure. In Case W1, the vertical earth pressure above the CCT remains constant over time. Conversely, for Cases W2, W3, and W4, the vertical earth pressure undergoes a notable reduction. The most significant decrease in vertical earth pressure is observed in Case W4, with a corresponding reduction rate reaching 56.24%. When considering time-dependent creep, the vertical earth pressure continues to decrease over time. Ultimately, in Case W4, the vertical earth pressure experiences a reduction of 64.5%.

Fig. 5
figure 5

Time curves for the area above the CCT regarding (a) vertical earth pressure, (b) lateral earth pressure

Full size image

Figure 5(b) illustrates the time curve for the lateral earth pressure. Lateral earth pressure exhibits similar values across all cases but undergoes reductions in Cases W2, W3, and W4, with Case W4 exhibiting the most significant reduction rate. At the onset of time, the lateral earth pressure in Case W4 is 38% lower than in Case W1, with a further reduction of 40.62% observed after the stabilization of creep.

The results demonstrate that the load-reduction effect of vertical earth pressure is more pronounced for Cases W2, W3, and W4 compared to Case W1. Furthermore, the load-reduction effects of both vertical and lateral earth pressure are most significant for Case W4, and these effects amplify with creep before stabilizing after the creep stabilization period.

4.2 Bending Moment

The bending moment of the CCT is examined from two main perspectives: bending-moment distribution and time curves. Bending-moment contours primarily depict the internal force results of shell elements. Time curves illustrate the variations in bending moments across key components (the vault, arched shoulder, arched waist, and arched feet) over a period of 400 days. A positive bending moment indicates stretching along the inner contour of the lining, while a negative bending moment signifies compression.

4.2.1 Bending-Moment Distribution

Figure 6 illustrates the bending-moment contours at the onset of the four cases and on the 400th day. The bending moment for Case W1 is notably greater than that for the other three cases, with the maximum positive bending moment occurring at the vault. Progressing from top to bottom, the bending moment gradually decreases until the stress state of the CCT at the spandrel transitions from positive to negative. Subsequently, the bending moment gradually increases until the maximum negative bending moment occurs at the arch springing. In contrast, the bending-moment distributions for Cases W2, W3, and W4 are uniform, with Case W4 exhibiting the most uniform distribution. Here, the maximum positive bending moment occurs at the bottom of the arch, while the maximum negative bending moment occurs at the foot of the arch. The change in the CCT stress state still occurs near the spandrel.

Fig. 6
figure 6

Bending-moment contours

Full size image

4.2.2 Time Curves

Figure 7 presents the time curves of bending moments at various positions of the CCT. Bending moments at these positions are smaller in Cases W2, W3, and W4 compared to Case W1. Particularly, in Case W4, the bending moments at the vault, spandrel, and arch springing decrease more significantly compared to Case W1. Before creep, the bending moments decrease by 85.2%, 91.3%, and 45.3% at the vault, spandrel, and arch springing, respectively, in Cases W2, W3, and W4, with further reductions of 92.9%, 81.8%, and 43.1% after creep stabilization. The reduction of the bending moment for the haunch is 59.2% compared to Case W1 before creep and 64% after creep stabilization. These results indicate that the overall bending-moment state of the CCT is improved in Case W4 compared to all other cases, with creep causing a shift in bending moment from the vault and haunch to the spandrel and arch springing.

Fig. 7
figure 7

Time curves of the bending moments of (a) the vault, (b) the spandrel, (c) the haunch, (d) the arch springing

Full size image

4.3 Shear Force

The analysis of shear force for the CCT is conducted through cloud diagrams and time curves. In the figures, shear force is positive when it induces clockwise rotation of the lining at the back of the section, while the sign is opposite otherwise.

4.3.1 Shear-Force Distribution

Figure 8 depicts the shear contours for the four cases at the beginning of the time and on the 400th day. In all cases, the maximum shear occurs at the bottom of the arch. Compared to Cases W2, W3, and W4, Case W1 exhibits larger and more unevenly distributed shear force. Conversely, for Cases W2, W3, and W4, the shear-force distribution is relatively uniform, with Case W4 displaying the greatest uniformity. Moreover, the shear-force distribution in Cases W2, W3, and W4 becomes more uniform after the stabilization of creep compared to before.

Fig. 8
figure 8

Shear-force contours

Full size image

4.4 Time Curves

Figure 9 displays the shear-force time curves corresponding to four CCT positions. In Cases W2 and W3, the shear force at the vault, spandrel, and haunch decreases more compared to Case W1, while the shear force at the arch springing increases in Cases W2 and W3. Across Cases W2, W3, and W4, the shear-force distribution shows relative unevenness, exacerbated by creep. The shear force is lower in all four positions in Case W4 compared to Case W1. Before creep, the shear force from the vault to the arch springing declines by 98%, 71.5%, 60.6%, and 29% in Cases W1, W2, W3, and W4, respectively, and after stabilization, it decreases by 85.8%, 77.3%, 70.8%, and 23.6%, respectively. These results indicate that Case W4 exhibits the best overall shear state, and the shear-force distribution of the CCT becomes increasingly uniform over time.

Fig. 9
figure 9

Time curves of the shear force of (a) the vault, (b) the spandrel, (c) the haunch, (d) the arch springing

Full size image

4.5 Axial Force

The analysis of the axial force of the CCT includes cloud diagrams and time curves. Axial force is positive for tension and negative for compression.

4.5.1 Distribution of Axial Force

Figure 10 depicts the axial-force contours for the four cases at the beginning of the time and on the 400th day. In all cases, the axial force of the CCT consists of compressive forces. As the axial force transitions from the arch springing to the arch bottom, there is a sudden decrease near the arch springing followed by a gradual increase near the bottom of the arch. The axial-force distribution appears more uniform in Cases W2, W3, and W4 compared to Case W1, with Case W4 displaying the greatest uniformity. Additionally, it’s noteworthy that axial force is dispersed from the structure to the arch bottom.

Fig. 10
figure 10

Axial-force contours

Full size image

4.5.2 Time Curves

Figure 11 displays the time curves of the axial force at four positions on the CCT. Compared to Case W1, the decrease in axial force is more pronounced for Cases W2, W3, and W4, with Case W4 exhibiting the greatest decrease. Before creep, the axial force in Case W4 from the vault to the arch springing decreased by 40.8%, 47.1%, 35%, and 31.4%, respectively, and decreased by 46.9%, 52.5%, 39.4%, and 35.8%, respectively, after creep stabilization.

Fig. 11
figure 11

Time curves of the axial force of (a) the vault, (b) the spandrel, (c) the haunch, (d) the arch springing

Full size image

In Cases W2 and W3, the axial forces at each position vary greatly, whereas the axial forces in Case W4’s four positions are quite similar. These results indicate that the axial-force distribution for Cases W2 and W3 is more uneven than for Case W4. Overall, Case W4 yields better results in terms of axial force distribution compared to the other three cases. Moreover, it was observed that in all cases, the longer the creep time, the greater the decrease in axial force throughout the CCT.

4.6 Vertical Displacement of Backfill

The changes in vertical displacement of backfill for the four cases are analyzed and presented below.

For all four cases, Fig. 12 presents the cloud diagram and the distribution curve illustrating the vertical displacement of backfill along the A-A section above the CCT. In Fig. 12(a), unlike Case W1, Case W2 experienced an increase in the vertical displacement of backfill soil above the CCT. Similarly, Case W3 exhibited a similar increase in the vertical displacement of backfill soil above the CCT as Case W2. However, the vertical displacement of backfill soil above the CCT in Case W4 was smaller compared to Cases W2 and W3. In all four cases, the increase in backfill creep corresponded to a greater increase in vertical displacement.

Fig. 12
figure 12

The vertical displacement of backfill: (a) the cloud diagram, (b) the distribution curves

Full size image

Figure 12(b) illustrates changes in the vertical displacement of backfill soil for each of the four cases. The vertical-displacement distribution curve of backfill is U-shaped for each case. In Case W2, the displacement increases from the edge of the measuring point to the center of the CCT, where the maximum displacement occurs. pecifically, the vertical displacement in Case W2 increases by 0.0882 m. After the stabilization of creep, the vertical displacement of backfill soil increases when RLC soil is used, with an increase of 0.0133 m at the center. Similarly, Case W3 exhibits almost unchanged vertical displacement. In Case W4, the vertical displacement decreases by 0.0141 m before the stabilization of creep and 0.0144 m after the stabilization of creep.

The results indicate that RLC soil, characterized by its low pressure solidity, induced a displacement disparity between the external and internal soil columns. This led to the soil on the sides of the CCT exerting upward shear force on the soil in the middle, thereby transferring pressure from the top of the CCT to the external soil column [15].

The inclusion of concrete columns on both sides of the CCT stretched the geogrid fully, creating a tension effect that generated an upward lifting force on the CCT. In Case W2, the relative displacement contributed to a load-reduction effect. Conversely, in Case W3, similar to Case W2, there was a lack of tension in the geogrid.

Finally, in Case W4, the load reduction stemmed from both the displacement of backfill and the lifting force produced by the geogrid. Creep further reinforced these load-reduction effects.

5 Discussion

In this study, we employed a finite difference method to explore the mechanical characteristics of dry sandy gravel backfill under various load-reduction measures. Our findings revealed that when loess backfill was utilized, soil creep resulted in a long-term reduction in the effectiveness of load reduction. However, when the backfill comprised low-pressure soil (av < 0.1), we observed a significant increase in the load-reduction rate rather than a decrease, and creep further enhanced the load-reduction effect. Therefore, if the creep properties of the backfill soil can be more effectively controlled, the long-term load reduction effect would be amplified. Based on these insights, engineers can consider utilizing alternative backfill materials to better ensure the long-term safety of deeply buried structures such as HFCCTs. We recommend sourcing soil from other sites and compacting it, as these measures can enhance the long-term stability of the backfill.

While our quantitative analysis indicates that creep could potentially enhance load reduction when backfill soils are sourced from alternative sites, several factors warrant further consideration. Fluctuations in soil moisture content and the presence of structures above the HFCCT are particularly noteworthy. The constrained availability of land for construction often results in extensive building activities above many HFCCTs, inevitably altering their mechanical properties. Therefore, it is imperative that future research addresses these factors comprehensively.

In our study, we demonstrated that the load-reduction methods applied in Cases W2, W3, and W4 effectively mitigate the stress-concentration phenomenon at the CCT's vault site. However, as depicted in Figures 6, 8, and 10, the load-reduction methods employed in Cases W2 and W3 still resulted in uneven distributions of overall internal force. Our findings indicate that incorporating concrete columns into CCTs could potentially promote a more uniform distribution of internal forces at the vault location compared to scenarios where such columns are not utilized.

When employing loess-rich RLC soil as backfill for load reduction, we observed a short-term load-reduction rate of 12.5% at the top of the HFCCT. However, substituting the RLC soil with dry sandy gravel resulted in a substantial improvement, with the short-term load-reduction rate jumping to 52.15%. In cases where the backfill primarily consisted of loess, the long-term load-reduction rate was measured at 1.67%. Conversely, when utilizing dry sandy gravel as backfill, the long-term load-reduction rate significantly increased to 62.76%. This represents an approximate 10% rebound from the short-term load reduction observed in previous studies using on-site loess as the backfill soil (Table 5).

Table 5 Load-Reduction Rate Comparison Across Different Conditions
Full size table

6 Conclusion

In this study, we utilized FLAC3D, a finite difference method, to examine the load-reduction efficacy of dry sandy gravel. Dry sandy gravel, characterized by low compressibility, offers potential in mitigating the diminishing effects of load reduction caused by soil creep. The primary challenge facing the enduring stability of HFCCTs lies in the failure of load-reduction techniques due to creep. However, by substituting a site's soil with dry sandy gravel or other suitable offsite soils, the long-term stability of HFCCTs appears promising. We investigated four cases, analyzing outcomes related to earth pressure, bending moment, shear force, axial force, and displacement of HFCCT. Our findings yield several conclusions:

  1. 1.

    Load reduction is less effective when loess is used as backfill soil compared to dry sandy gravel. The presence of loess can lead to a rebound in long-term earth pressure due to creep. However, substituting in-situ loess with dry sandy gravel, or even RLC soil, can notably enhance short-term load reduction.

  2. 2.

    In terms of long-term load reduction, the backfill soil's low creep characteristics significantly improve the effectiveness compared to using loess as backfill soil. Dry sandy gravel demonstrates superior performance in prolonging load reduction. These findings hold promise for enhancing the long-term stability of HFCCTs specifically and offer potential solutions for ensuring stability in a range of engineering projects.

  3. 3.

    The principle of load reduction involves inducing the soil arching effect on the top of a CCT using RLC soil. This method proves effective, and further enhancements are observed with the addition of a geogrid and concrete columns. Concrete columns not only alleviate the uneven distribution of internal force along the HFCCT but also enhance its long-term stability by minimizing the effects of creep behavior.

  4. 4.

    The long-term load reduction mechanism is complex and greatly influenced by factors such as backfill soil, RLC soil, and geogrids. Creep behavior resulting from the interaction of these factors plays a crucial role in determining the effectiveness of load-reduction methods over time. Therefore, it is essential for engineers to thoroughly investigate the creep behavior of various materials to ensure the effectiveness of load-reduction methods, not only for CCTs but also for other types of structures.