Assessing the Settlement and Deformation of Pile-Supported Embankments Undergoing Groundwater-Level Fluctuations: An Experimental and Simulation Study

Abstract

The intensification of extreme weather phenomena, ranging from torrential downpours to protracted dry spells, which trigger fluctuations at the groundwater level, poses a grave threat to the stability of embankments, giving rise to an array of concerns including cracking and differential settlement. Consequently, it is crucial to embark on research targeted at uncovering the settlement and deformation behaviors of pile-supported embankments amidst changes in water levels. In tackling this dilemma, a series of direct shear tests were carried out across a range of wet–dry cyclic conditions. The results confirmed that the occurrence of wet–dry cycles significantly impacted the resilience of silty clay. Additionally, it was observed that the erosion of cohesion and the angle of internal friction initially diminished sharply, subsequently leveling off, with the first wet–dry cycle exerting the most substantial influence on soil strength. Employing a holistic pile-supported embankment model, simulations revealed that variations in the groundwater level, fluctuations therein, varying descent rates, and periodic shifts in the groundwater level could all prompt alterations in soil settlement between embankment piles and could augment the peak tensile stress applied to geogrids. In summary, the orthogonal experimental method was utilized, indicating that, in terms of impacting embankment settlement under periodic water-level changes, the factors ranked in descending order were the following: pile spacing, pile length, embankment height, and the height of the groundwater table.

1. Introduction

Embankments serve as a vital component of transportation infrastructure, and their stability and integrity are paramount to engineering design and construction. Nonetheless, the inherent low strength and high compressibility of soils frequently constrain embankment construction, potentially leading to catastrophic events such as embankment collapse and landslides. The geogrid-reinforced and pile-supported (GRPS) embankment system has been employed extensively, due to its abbreviated construction duration and economies of scale, encompassing piles, sand-gravel cushions, reinforcements, and embankment fill materials [1,2,3]. Within this system, a portion of the soil load between piles is transferred to adjacent piles via shear resistance within the soil matrix, a phenomenon known as the soil arching effect [4,5,6]. Moreover, the tensile force provided by the reinforcement offers lateral support, constraining the embankment’s lateral spreading and facilitating the transfer of higher loads to the pile cap, a mechanism referred to as the membrane effect [7,8]. Both the soil arching effect and the membrane effect enhance the distribution of soil loads from between piles to the pile caps, thereby reducing differential settlement across the embankment [9,10,11].

In recent years, the frequency of extreme precipitation events has risen, causing a subsequent increase in groundwater levels [12,13]. The elevated groundwater levels not only diminish the shear strength of the soil [14,15,16], but also induce soil creep, leading to soil deformation [17,18,19], which in turn compromises embankment stability and triggers landslides. Additionally, the soil arching effect is mitigated as the soil’s strength (cohesion, friction, and Young’s modulus) wanes due to increased groundwater levels [20,21,22,23], reducing the load transfer between the subsoil and piles. Consequently, both total and differential settlements are exacerbated, posing threats to the stability and safety of road embankments. There is also an interplay between the underlying soil and the geosynthetic reinforcement; the higher compressibility of the soil, influenced by its saturation level, restricts the movement of the embankment soil, thereby reducing the displacement of the geosynthetic reinforcement, which can impact the integrity of the geosynthetic reinforcement.

To examine the impact of soil saturation levels on soil behavior, indoor laboratory experiments have been employed [24]. These studies have highlighted a correlation between soil strength and its saturation degree [25]. It is observed that soil strength diminishes with increasing saturation, which is influenced by the rise in groundwater levels. Additionally, subjected to multiple wet–dry cycles, soil strength is further compromised, although the rate of decay slows with the increment in wet–dry cycles, ultimately trending to stability [26,27]. Moreover, matric suction within unsaturated soil augments the effective internal friction angle and cohesive forces, thereby enhancing shear strength and embankment stability [17,28,29,30].

To delve deeper into the effects of groundwater-level fluctuations on embankments, full-section embankment models [31,32] and trap door models [33] have been utilized in simulation tests. The findings indicate that changes in the groundwater level, whether an increase or decrease, alter soil characteristics and modulate the soil arching effect, thereby affecting total and differential settlements within the embankment [27]. Particularly, under the combined influence of groundwater level fluctuations and cyclic loads, additional settlement of the embankment is incremented, and in severe cases, phenomena such as slurry boiling and mud leakage may arise [34,35].

Previous research has predominantly focused on the effects of isolated drops or rises in water level on embankment settlement through numerical simulation and model experiments, without fully considering the practical implications of periodic changes in groundwater level on embankment settlement in engineering practice. Consequently, there is a pressing need to investigate the deformation characteristics of pile-supported embankments subjected to periodic fluctuations in water level, building upon the foundation of existing research. The objectives of this study were to scrutinize the settlement and deformation dynamics of pile-supported embankments under varying groundwater levels. Direct shear tests were conducted to elucidate the impact of wet–dry cycles on the strength of silty clay. Subsequently, a simulation model was developed to examine the repercussions of various factors, including the depth of the groundwater level and the magnitude and rate of change of the groundwater level, on the overall and differential settlement within the embankment. Lastly, the study aimed to elucidate the settlement variation patterns of pile-supported embankments under cyclic groundwater-level variations, thereby informing the design and maintenance of roadbed engineering practices.

2. Influence of Wet–Dry Cycles on the Shear Strength of Silty Clay

2.1. Experimental Protocol and Soil Material

During the construction and long-term operation of pile-supported embankments, the subgrade soil, as the bearing layer of the embankment, will be exposed to a cycle of drying and wetting due to the periodic fluctuation of the groundwater level. Its strength will gradually change with the variation in soil saturation, leading to excessive settlement of the embankment, which affects the safety of embankment construction and operation. Furthermore, with the increase in the number of dry–wet cycles, the soil strength will also gradually decline, ultimately reaching a stable value. To investigate the impact of dry–wet cycles on soil strength, it is necessary to understand the intrinsic relationship between the shear strength of fine-grained clay and the number of dry–wet cycles, the lower-limit moisture content, and the amplitude of dry–wet cycling. Laboratory experiments were conducted to simulate the dry–wet cycling conditions of soil samples and to analyze the evolution of shear strength through direct shear tests.

The soil used in this study was sourced from Keqiao District, Shaoxing City, Zhejiang Province, China. The soil was extracted from a depth of 3 m to 3.5 m and exhibited a gray–yellow color. Through indoor natural moisture content tests, consistency limit tests, and compaction tests, the natural moisture content, plasticity index, maximum dry density, and optimum moisture content of the soil were determined to be 33.45%, 12.97%, 1.68 g/cm³, and 21.39%, respectively.

During the fluctuation in groundwater levels, the range of moisture content variation at different depths of the fine-grained clay differs significantly from the lowest moisture content, resulting in different dry-wet cycling paths experienced by the soil. Based on the results of natural-moisture-content tests, consistency limit tests, and compaction tests, this experiment selected soil with the same lower-limit moisture content (20%) and employed two sets of dry–wet cycling amplitudes (10% and 20%) to conduct dry–wet cycling tests. The number of dry–wet cycles for the soil was set at 5, as depicted in Figure 1 and

Table 1. Dry–Wet Cycle Test Protocol.

Test No.	Cycle Path	Compactness	Number of Dry–Wet Cycles
A	20% → 30% → 20%	0.93	5
B	20% → 40% → 20%

.

The dry–wet cycling test is primarily divided into two stages: wetting and drying. Wetting is achieved by the migration of a water film, where a predetermined amount of water is evenly dripped onto the surface of the sample (corresponding to the maximum moisture content in the dry–wet cycle). The sample is then sealed and maintained for 24 h under a moisture-retaining film to ensure uniform distribution of water both inside and outside the sample, thus simulating the wet phase during dry–wet cycling. Drying is conducted using a low-temperature drying method, where the sample is placed in an oven set at 45 °C to air-dry. The moisture content is controlled to reach the target moisture content (corresponding to the minimum moisture content in the dry–wet cycle) by periodic weighing, and the sample is sealed and maintained for an additional 24 h to achieve uniform moisture distribution, simulating the dry phase in the dry–wet cycling process. The duration of the drying process is determined through repeated measurements of the moisture content changes. By repeating the wetting–drying process, various numbers of dry–wet cycles can be simulated.

Figure 2 illustrates the process of curing the soil samples. During this stage, the soil samples are restored to their initial state under certain moisture conditions in preparation for the next round of dry–wet cycling tests. The curing process allows the moisture content of the samples to reach equilibrium and for their mechanical properties to be gradually restored. This stage is crucial for ensuring the accuracy and repeatability of the test results.

Finally, direct shear tests were conducted to measure the shear strength of the soil samples. During the tests, the vertical pressure was applied at 100 kPa, 200 kPa, 300 kPa, and 400 kPa. The shear rate within the shear box was set at 0.8 mm/min, and the shear displacement was fixed at 6 mm. This test protocol ensures consistent conditions for all samples and provides a standard frame of reference for comparing the shear strength properties of the soil under different dry–wet cycle conditions.

2.2. Discussion of the Experimental Results

As depicted in Figure 3, the cohesion of silty clay exhibits a decaying trend under various drying–wetting cycles. The “attenuation rate” is defined as the ratio of the reduction in soil cohesion following each wet–dry cycle to the soil’s initial cohesion. This measure provides a quantitative assessment of the impact of cyclic moisture changes on soil stability. In the initial phase of drying–wetting cycling, the cohesion of the soil diminishes rapidly, with the cohesion attenuation rates of paths A and B witnessing a decay percentage of 49.68% and 60.79% of the total attenuation rate, respectively, in the first cycle. Continuing with the drying–wetting cycles, the negative impact on soil cohesion gradually lessens, with the cohesion attenuation rate falling below 2% by the fifth cycle. This suggests that the soil’s capacity to withstand stress through cohesion stabilizes after repeated cycles, indicating an adaptation or saturation point in the soil’s response to moisture fluctuations.

After enduring five drying–wetting cycles, paths A and B experience a reduction in cohesion of 7.75 kPa and 7.65 kPa, respectively. In summary, the number of drying–wetting cycles exerts a significant influence on soil cohesion, with the decaying effect tapering off and eventually stabilizing as the cycle count increases. In the design and construction of embankments, it is imperative to consider the decrement in soil cohesion due to drying–wetting cycles, to mitigate settlement. Notably, due to the initial particle gradation of the soil specimen and variations in the soil collection location, discrepancies in initial soil cohesion exist along different paths, albeit these do not alter the regularity of soil strength decrement under drying–wetting cycles.

As illustrated in Figure 4, the drying–wetting cycles also exhibits a pronounced decaying effect on the internal friction angle of the soil, with decrements of 2.36 and 2.18 for paths A and B, corresponding to attenuation rates of 11.16% and 11.21%. Here, the term “attenuation rate” refers to the proportion by which the internal friction angle of the soil declines after each wet–dry cycle, in relation to its original value. Similar to the cohesion decay curve, the internal friction angle of the silty clay also gradually diminishes with an increasing number of drying–wetting cycles, manifesting an initial rapid decay, a subsequent slowing-down, and an eventual stabilization trend. For paths A and B, the internal-friction-angle decay values at the first drying–wetting cycle account for 36.02% and 56.30% of the total decay, respectively. Additionally, an examination of Figure 3 and Figure 4 discloses that, following five cycles of drying and wetting, the cumulative attenuation rates for soil cohesion along the distinct pathways are 28.18% and 31.88%, respectively. In contrast, the cumulative attenuation rates for the internal friction angle are significantly lower, at merely 11.16% and 11.21%, respectively. These results corroborate the stronger decaying effect of drying–wetting cycles on soil cohesion.

However, there is a discrepancy in the decay cycle of soil cohesion and internal friction angle under drying–wetting cycles. After the soil undergoes three drying–wetting cycles, the cohesion tends to stabilize at a certain value, with minimal or even negligible decay values with further cycles. In contrast, the internal friction angle continues to decay after three drying–wetting cycles and only stabilizes after four cycles. Consequently, in the design, construction, and settlement prediction of embankments, it is crucial to consider the differing stable cycles of soil cohesion and internal friction angle.

3. Numerical Simulations

3.1. Finite Element Model and Boundary Conditions

This study constructs a numerical model based on a pile-supported embankment project in the northern suburbs of Shanghai [36], to investigate the settlement and deformation characteristics of the embankment under conditions of groundwater-level fluctuations, as depicted in Figure 5. The foundation soil layer consists of a 25 m thick soft-soil foundation. The piles are represented by pipe piles with a diameter of 1.0 m and a wall thickness of 0.12 m, extending to a length of 16 m with a pile spacing of 3 m. The embankment height is 5.6 m, with a top width of 17.5 m and a 1:1.5 slope. At the bottom of the embankment, a 0.5 m thick crushed-stone cushion is laid, which includes one layer of isotropic geotextile to form a reinforced crushed-stone cushion.

Due to the symmetry of the cross section and the uniformity of the pile spacing along the longitudinal direction, only half of the embankment width is simulated. To eliminate the influence of boundary effects, the total model width is set to three times the length of the embankment bottom (77.7 m). Furthermore, to better validate the rationality of the numerical model, a double-row pile-supported embankment is modeled, which involves a 2-time pile spacing along the embankment longitudinal direction, as shown in Figure 6.

The embankment and coarse-grained fill are modeled using the Mohr–Coulomb model, while the base soil layer is represented by a soft-soil model, as shown in

Table 2. Soil Material Parameters.

Material	γ (kN/m3)	E (MPa)	λ	κ	v	c (kPa)	φ (°)	kw × 10−4 (m/day)
Embankment	18.5	20	-	-	0.3	10	30	-
Gravel	18.5	20	-	-	0.3	10	40	-
Coarse-grained fill	18.5	7	-	-	0.3	15	28	-
Silty clay	20	-	0.06	0.012	0.35	18	21.7	8.64
Soft silty clay	17	-	0.15	0.03	0.4	20	15.34	4.32
Medium silty clay	20.5	-	0.05	0.01	0.35	28	18.9	4.32
Sandy silt	220	-	0.03	0.005	0.35	40	22.75	43.2

. The geotextile is simulated using a linear elastic model, and the pipe piles are modeled using an embedded pile element. The geotextile is made of an isotropic linear-elastic material, which withstands tension but not compression, with a tensile stiffness of 1180 kN/m.

In PLAXIS 3D CONNECT Edition V22 Update 1 launched by Bentley Systems in USA, interface elements are commonly utilized to simulate the mutual contact between soil and structures. The parameter R_inter, which defines the degree of interaction between the soil and structure, is multiplied by the soil strength to determine the strength parameters of the soil–structure interface. When the deformation between the structure and the contacting soil is identical, i.e., there is no relative sliding between the two, R_inter is taken to be 1.0.

In this paper, the strength reduction factor R_inter for the gravel cushion layer is 0.8, for the coarse-grained fill and sandy silt it is 0.7, and for the silty clay, soft silty clay, and medium silty clay it is 0.6. The interaction between the pile and the soil is represented by the pile side friction, which is defined in relation to the soil layer when employing embedded pile simulations to model the pile body.

Upon completion of the geometric model and assignment of material parameters to the corresponding soil and structural elements, the model is meshed. For this study, a fine-density meshing approach is selected, with the structural elements and embankment appropriately refined. The model consists of a total of 17,067 units and 31,002 nodes, as illustrated in Figure 7.

For displacement boundary conditions, the bottom of the model (Z_min) is fully fixed, while the top (Z_max) is free. The lateral boundaries (X_min, X_max, Y_min, Y_max) are set to normal fixed. Regarding the drainage boundary conditions, it is assumed that the groundwater level is located 1.5 m below the foundation surface. The surface of the foundation soil (Z_max) is set to allow free drainage, while the remaining faces (Z_min, X_min, X_max, Y_min, Y_max) are impermeable. By adjusting the depth of the groundwater level, the periodic changes in water level within the subgrade soil can be simulated.

3.2. Construction Step Simulation

In the Phased Construction module of PLAXIS 3D, multiple construction stages are added to simulate the process of embankment filling. Initially, only the foundation soil is selected, with the calculation type set to the K₀ process, which directly creates the initial effective stress, pore water pressure, and state parameters. Subsequently, the pile driving stage is added, where all the embedded pile elements are activated and assigned a plastic calculation type. Then, the embankment construction stage is introduced, activating the embankment layers one by one according to Figure 8, selecting a consolidation calculation type for all the embankment layers. During the construction of the cushion, the geotextile elements and their interface units must also be activated. Finally, a consolidation stage is added, with a duration of 125 days, and the calculation type set to consolidation.

3.3. Model Validation

As depicted in Figure 9, the settlement rate during the embankment filling process is relatively high, slowing down gradually by the end of 55 days of construction, and stabilizing at around 150 days. At the conclusion of the filling period, the vertical settlement at the pile top and between the piles is 10.80 mm and 58.38 mm, with differences of 21.79% and 12.9% compared to the measured values. By the end of the consolidation period, the vertical settlement at the pile top and between the piles reaches 20.95 mm and 76.21 mm, with measurement differences of 13.22% and 8.62%. Throughout the entire process of embankment construction and consolidation, both numerical-simulation and engineering-field tests reveal a consistent trend in soil behavior, thus indicating that the numerical model in this study can effectively simulate the settlement variation trends during embankment construction and consolidation.

To further validate the accuracy of the numerical model in simulating the stress changes in the embankment, a comparison between the numerical simulation results and the field measurement results is conducted, as shown in Figure 10. Upon the conclusion of soil consolidation, the stress at the pile top and between the piles is 673.9 kPa and 58.69 kPa, with errors of 18.64% and 12.27% compared to the measurements. Among these, influenced by the soil arching effect, as construction progresses, the stress of the soil between the piles is transmitted to the pile top, with the stress between the piles being less than the embankment load, and a stress concentration phenomenon occurs at the pile top, with the pile-top stress being significantly greater than the embankment load. Throughout the entire process of embankment construction and consolidation, both numerical simulation and engineering field tests exhibit the same changing trend. In conclusion, there is consistency between the numerical model results and the field measurements, which validates the reasonableness of the model.

4. Influence of Groundwater Level on Pile-Supported Embankments

4.1. Influence of Groundwater Level on Soil Arching Effect

Subject to varying groundwater levels, the distribution of soil moisture content is altered, which in turn modifies the soil’s strength characteristics, potentially triggering deformation in embankments. To uncover the impact of groundwater level on the settlement and deformation of pile-supported embankments, simulations of the embankment-filling and -consolidation processes were conducted for three scenarios with different groundwater table heights: h = 0 m, −1.5 m, and −3 m. As shown in Figure 11, the groundwater level has a significant influence on the soil between piles, while its effect on end-bearing piles is relatively minor. This is because end-bearing piles primarily carry the load through the resistance at their ends, with settlement being related only to the mechanically robust strata and less affected by the soft-soil layers and groundwater level. For the soil between piles, however, a higher groundwater level increases soil saturation, leading to a reduction in soil strength (cohesion, internal friction angle, and compressibility modulus), and potentially causing creep in the soil, thereby increasing deformation. Additionally, during embankment filling, the saturated soil beneath the groundwater level generates excess pore water pressure, which, upon completion of embankment filling, dissipates, causing some degree of settlement. However, the above process does not occur for soil above the groundwater level. Therefore, the settlement of the soil between piles and the embankment soil in pile-supported embankments will gradually decrease as the groundwater level descends. The reduction in settlement, however, will exhibit a trend of gradual attenuation with the lowering of the normal groundwater level, and when the groundwater level reaches a certain height, further changes in the groundwater level will no longer affect the settlement of pile-supported embankments.

Furthermore, the height of the normal groundwater table will have a significant impact on the equal settlement surface of pile-supported embankments. As shown in Figure 11, as the height of the normal groundwater table decreases, the height of the equal settlement surface will gradually decrease. This is because as the height of the normal groundwater table decreases, the soil arch effect strengthens progressively, leading to an increase in the load transferred from the soil between piles to the pile tops, thereby reducing settlement between the piles.

As depicted in Figure 12, the tensile force of the geotextile increases with the increase in the normal groundwater table. This is because as the groundwater level rises, the soil arching effect in the pile-supported embankment weakens, resulting in greater non-uniform settlement between the pile tops and the soil between piles, and consequently causing the maximum tensile force of the soil-arching geotextile to increase. In engineering design, the variation in the maximum tensile force of the geotextile under different groundwater levels should be considered, to prevent embankment collapse accidents due to geotextile failure.

4.2. Influence of Water Level Changes on Pile-Supported Embankments

Upon the completion of pile-supported embankment construction, changes in the groundwater table can alter the soil’s mechanical properties, thereby triggering settlement in the embankment [37]. To investigate the influence of groundwater-level fluctuations on embankment settlement, a full-section model of the pile-supported embankment was established to simulate the settlement process during variations in the groundwater level, with a specified decrease rate of 1 m per month. As illustrated in Figure 13, the settlement of the soil between the piles progressively increases with the lowering of the groundwater table, and as the groundwater level continues to decline, the average settlement rate of the soil between the piles accelerates. This is attributed to the fact that with the sustained decrease in the water level, the effective stress within the soil gradually increases, leading to continuous compression of the soil layers and subsequent settlement deformation of the ground surface. Nevertheless, once the groundwater level recedes to a certain depth, its influence on the soil settlement between piles diminishes or may become negligible. Consequently, as time progresses, the rate of increase in the average soil settlement is expected to decelerate or potentially reverse.

Figure 14 demonstrates that upon the consolidation completion of the pile-supported embankment, under the same decrease rate as that of the groundwater table, the maximum tensile force in the geotextile initially shows a decreasing trend with the increase in the depth of the groundwater table. However, with further increase in the depth of the groundwater table, the tensile value in the geotextile gradually rises. This is because on one hand, the decrease in the groundwater level reduces the soil saturation, enhances the soil strength and the soil arching effect, mitigating the non-uniform settlement between the piles and the soil, thereby decreasing the maximum tensile force in the geotextile. On the other hand, the decrease in the groundwater level increases the effective stress within the soil, leading to continuous compression of the soil layers and increasing the non-uniform settlement between the piles and the soil, consequently increasing the maximum tensile force in the geotextile. In the initial stage of the groundwater-level decrease, the change in non-uniform settlement between the piles and the soil is significantly influenced by the soil arching effect, hence reducing the maximum tensile force in the geotextile. With further decrease in the groundwater level, the change in non-uniform settlement between the piles and the soil is primarily affected by the increased effective stress within the soil, leading to an increase in the maximum tensile force in the geotextile. Furthermore, the tensile-force–attenuation ratio is defined as the proportion by which the maximum tensile force diminishes in response to groundwater-level fluctuations, relative to its initial maximum value. Figure 14 reveals that at any given depth of the groundwater table, the attenuation rate of the maximum tensile stress in the geotextile is consistently below 0.5%, indicating that although the groundwater level does have an impact on the geotextile, this impact is relatively minor.

Figure 15 illustrates that under identical conditions of groundwater-level reduction, the settlement of soil between piles is 4.7 mm, 3.85 mm, and 3.7 mm for groundwater decrease rates of 1 m/month, 2 m/month, and 3 m/month, respectively. This phenomenon suggests that as the rate of groundwater-level decline increases, the settlement of the soil between piles correspondingly decreases. This may be due to the fact that during rapid decreases in the groundwater level, the soil matrix still contains a certain amount of moisture, which helps to share part of the load, thereby slowing down the settlement. Concurrently, as the rate of groundwater decrease increases, its mitigating effect on soil settlement diminishes, and when the rate of groundwater decrease reaches a certain value, the soil settlement no longer decreases with the further increase in the rate of groundwater decrease.

Figure 16 demonstrates that, with a constant reduction in groundwater level, the maximum tensile stress experienced by the geotextile lessens as the rate of groundwater-level decline accelerates. Specifically, for groundwater decrease rates of 1 m/month, 2 m/month, and 3 m/month, the respective attenuation ratios of tensile force are 0.4%, 0.88%, and 1.46%. This indicates a direct relationship between the rate of groundwater withdrawal and the reduction in tensile stress within the geotextile. Furthermore, as the rate of groundwater decrease increases, its weakening effect on the tensile stress in the geotextile becomes more pronounced. Upon comparing the attenuation rates of the maximum tensile stress in the geotextile as depicted in Figure 14 and Figure 16, it is observed that the attenuation rate in Figure 14 remains consistently below 0.5%, whereas in Figure 16 it exceeds 1.4%. This contrast highlights the fact that the rate at which the groundwater level decreases has a significantly more pronounced impact on the tensile stress within the geotextile than the mere reduction in groundwater level itself. Therefore, during the long-term operation of the pile-supported embankment, it is crucial to enhance the monitoring of the impact of the rate of groundwater-level decrease on the tensile stress in the geotextile.

4.3. Influence of Periodic Water-Level Changes on Pile-Supported Embankments

The Maintenance Period for Road Engineering specified in the “Methods for Completion and Acceptance of Highway Engineering” is generally 5 years. Therefore, this study investigates the impact of dry–wet cycles on settlement of pile-supported embankments after consolidation completion, based on the periodic changes in groundwater levels over a period of 5 years. The research utilizes data from the Monthly Report on Groundwater Dynamics (from January to December 2022) released by the Ministry of Water Resources of the People’s Republic of China to create a variation curve of the groundwater level in the Zhedong coastal plain region, as shown in Figure 17.

As illustrated in Figure 18, the settlement during the first year of water-level variation accounts for 73.95% of the total settlement after 5 years. Additionally, with the increase in the number of water-level cyclical variations, settlement continues to accumulate, although its development rate gradually becomes more stable. This suggests that after the first water-level change, the embankment will experience significant irreversible settlement (i.e., residual settlement) with an instance-dependent nature, meaning that as the number of water-level variations increases, so does the residual settlement. Combined with the results of indoor dry-wet cycle tests, it can be deduced that the variations in water levels result in a decrease in soil strength, leading to settlement of pile-supported embankments, and the settlement pattern is akin to the decaying trend of soil strength.

As shown in Figure 19, along with the periodic changes in water levels, the tensile force in the geotextile gradually increases. Simultaneously, with the increase in the number of water-level cyclical variations, its enhancing effect on the tensile stress in the geotextile becomes more pronounced. This is because the strength of the soil diminishes after dry–wet cycles, causing increased settlement, while the impact of water-level cyclical variations on the piles is relatively minor, leading to progressive increases in settlement between the piles and the soil, further manifested as an increase in the tensile-force value in the geotextile. Moreover, with the extension of the water-level variation cycle, the increase ratio of the maximum tensile stress in the geotextile becomes significantly greater. Therefore, in the design and monitoring of pile-supported embankments, the impact of water-level cyclical variations on the maximum tensile force in the geotextile should be considered to prevent embankment collapse due to geotextile failure.

5. Parametric Study on Settlement and Deformation of Pile-Supported Embankments under Water-Level Cyclic Variations

5.1. Orthogonal Test Method

Orthogonal experimental design is a widely utilized method in scientific research, particularly for the planning of multi-variable experimental designs. This approach involves the strategic selection and combination of representative factors within the experiment, utilizing orthogonal tables for the arrangement of trials, and employing statistical methods for comprehensive data analysis to assess the impact of each factor. For instance, in the context of four factors with three levels each, a traditional comparative experimental approach would necessitate 3⁴ = 81 test combinations. However, by employing an L9 orthogonal array, it is possible to achieve similar results with only nine test combinations, thereby significantly reducing the number of trials and enhancing efficiency. In this study, the orthogonal experimental design considers four factors at three levels each, and the experiment is arranged according to the L9 (3⁴) orthogonal array.

The method of range analysis is a commonly employed and efficient technique for analyzing the results of orthogonal experiments, offering a straightforward and convenient means to interpret the computed outcomes and accurately establish the relative importance of various factors affecting the results. In the application of this method, Kij is defined as the result for the j-th level of the i-th factor, while the average value $\overline{K_{ij}}$ serves as a basis for determining the optimal level and combination for the i-th factor. R_i, the range of the i-th factor, reflects the magnitude of the evaluation indicator’s variation when the levels of the i-th factor change. By comparing the Ri values of different factors, one can clearly assess the extent of their influence on the results; a higher R_i value indicates a greater impact. The calculation formula for R_i is as follows:

$$R_i=\max [K_{i1},K_{i2},\cdot \cdot \cdot ]-\min [Ki1,Ki2,\cdot \cdot \cdot ]$$

(1)

While the method of range analysis exhibits advantages such as intuitiveness, rapidity, and simplicity in orthogonal experiments, it does have limitations in distinguishing between the effects of factor-level changes and experimental errors on the variability of test results. To address these limitations, the method of analysis of variance (ANOVA) is an indispensable supplement to orthogonal experimental analysis, due to its powerful analytical capabilities. Variance is a key metric for assessing the level of data scatter and is fundamental to determining the stability of experimental conditions. When conducting orthogonal design experiments, variance analysis is performed based on the additivity principle of Fisher’s sum of squares, utilizing the F-test to delve into the effectiveness of various influencing factors and their interactions. This method not only identifies and calculates the variances from different sources, but also reveals the interrelationships between these variances and the underlying statistical principles, thereby providing a solid foundation for in-depth analysis of experimental results.

ANOVA is commonly used to test for interactions between factors, and the calculation formula for the F-value is as shown in Equation (2):

$$F_A={\displaystyle \frac{S_A/f_A}{S_e/f_e}}$$

(2)

where F_A represents the F ratio for factor A, S_A is the sum of squares for factor A, S_e is the sum of squares for the error of factor A, f_A is the degrees of freedom for factor A, and f_e is the degrees of freedom for the error of factor A. The calculation formula for the sum of squares for the i-th factor, S_i, is as shown in Equation (3):

$${\displaystyle S_i=\frac{a}{b}{\displaystyle \sum _{k=1}^b{{\displaystyle ({\overline{K}}_{ij}-\overline{K})}}^2}}$$

(3)

where a represents the number of trials, b is the number of levels for each factor, ${\overline{K}}_{ij}$ is the average result for the j-th level of the i-th factor, and $\overline{K}$ is the overall average of the results.

5.2. Orthogonal Experimental Design Scheme

To investigate the influence of different factors on the settlement and deformation of pile-supported embankments under the condition of water-level cyclic changes, orthogonal experimental design was employed, focusing on four critical factors: pile length, pile spacing, embankment height, and the height of the permanent groundwater level. The relevant factors and their corresponding different levels are presented in

Table 3. Experimental Factors and Their Levels.

Level	A (Pile Length)	B (Pile Spacing)	C (Embankment Height)	D (Groundwater Level)
1	16	3	4	0
2	20	4	5.6	−1.5
3	25	5	7	−3

. By filling the levels of each influencing factor in a random sequence into the orthogonal table, the orthogonal experimental scheme, as shown in

Table 4. Numerical Orthogonal Experimental Scheme.

Test No.	A (Pile Length)	B (Pile Spacing)	C (Embankment Height)	D (Groundwater Level)
1	16	3	5.6	−1.5
2	16	4	7	0
3	16	5	5.6	0
4	20	4	4	−1.5
5	20	3	7	−3
6	20	4	5.6	−3
7	25	5	7	−1.5
8	25	5	4	−3
9	25	3	4	0

, was constructed.

5.3. Analysis of Orthogonal Results

As illustrated in Figure 20, the settlement of the embankment varies with the cyclic changes in groundwater level, and the settlement caused by the first year’s water level changes accounts for more than 50% of the total settlement. This is attributed to the significant strength degradation of the soil after the first dry–wet cycle, as observed from indoor dry–wet cycling test results.

Through the analysis of nine orthogonal experiments, the increase in embankment settlement after 5 years of cyclic groundwater-level changes is obtained and listed in the orthogonal table for range analysis.

Table 5. Orthogonal Analysis Table for Settlement Increase Caused by Periodic Groundwater Level Changes.

Test No.	A (Pile Length/m)	B (Pile Spacing/m)	C (Embankment Height/m)	D (Groundwater Level/m)	Settlement Increment (mm)
1	1 (16)	1 (3)	2 (5.6)	2 (−1.5)	9.60
2	1 (16)	2 (4)	3 (7)	1 (0)	20.30
3	1 (16)	3 (5)	2 (5.6)	1 (0)	23.69
4	2 (20)	2 (4)	1 (4)	2 (−1.5)	3.80
5	2 (20)	1 (3)	3 (7)	3 (−3)	6.76
6	2 (20)	2 (4)	2 (5.6)	3 (−3)	8.08
7	3 (25)	3 (5)	3 (7)	2 (−1.5)	16.75
8	3 (25)	3 (5)	1 (4)	3 (−3)	13.82
9	3 (25)	1 (3)	1 (4)	1 (0)	2.69
Kij	53.59	19.05	20.31	46.68
	18.64	32.18	41.37	30.15
	33.26	54.26	43.81	28.66
$\overline{K_{ij}}$	17.863	6.350	6.770	15.560
	6.213	10.727	13.790	10.050
	11.087	18.087	14.603	9.553
Ri	11.650	11.737	7.833	6.007

reveals that as the pile spacing increases, the corresponding value of ${\overline{K}}_{ij}$ also increases, leading to greater embankment settlement under water-level changes. When the pile spacing is 3 m, 4 m, and 5 m, the values of ${\overline{K}}_{ij}$ are 6.350, 10.727, and 10.087, respectively, with range (R_i) values of 11.737. This indicates that in practical engineering, reducing the pile spacing can mitigate embankment settlement. Conversely, the lower the constant groundwater level, the smaller the value of ${\overline{K}}_{ij}$ and the lesser the embankment settlement under water-level changes. For example, when the constant groundwater level is 0 m, −1.5 m, and −3 m, the ${\overline{K}}_{ij}$ values are 15.560, 10.050, and 9.553, respectively, with a range of 6.007. It is apparent that in regions with lower groundwater levels, such as in the north, the embankment settlement caused by cyclic water-level changes is significantly less than that in southern regions. Upon examining Table 5 and

Table 6. Variance Analysis Table for Settlement Increment Caused by Periodic Groundwater Level Changes.

Factors	Sum of Squares of Deviations	Sum of Squares of Errors	F Value
A (pile length)	205.393	227.826	0.902
B (pile spacing)	211.085	222.147	0.950
C (embankment height)	111.297	321.918	0.346
D (groundwater level)	66.691	366.534	0.182

, it is evident that the factors influencing the settlement of pile-supported embankments under conditions of water-level change are ranked, in descending order of influence, as pile spacing, pile length, embankment height, and groundwater-level height.

Additionally, through the analysis of nine orthogonal experiments, the increase in tensile force of geotextiles after 5 years of cyclic groundwater level changes is obtained and analyzed using range analysis. As shown in

Table 7. Orthogonal Analysis Table for Tensile Force Increment Caused by Periodic Groundwater Level Changes.

Test No.	A (Pile Length/m)	B (Pile Spacing/m)	C (Embankment Height/m)	D (Groundwater Level/m)	Tensile Force Increment (kN/m)
1	1 (16)	1 (3)	2 (5.6)	2 (−1.5)	0.388
2	1 (16)	2 (4)	3 (7)	1 (0)	0.840
3	1 (16)	3 (5)	2 (5.6)	1 (0)	1.360
4	2 (20)	2 (4)	1 (4)	2 (−1.5)	0.930
5	2 (20)	1 (3)	3 (7)	3 (−3)	0.747
6	2 (20)	2 (4)	2 (5.6)	3 (−3)	0.936
7	3 (25)	3 (5)	3 (7)	2 (−1.5)	1.370
8	3 (25)	3 (5)	1 (4)	3 (−3)	1.421
9	3 (25)	1 (3)	1 (4)	1 (0)	0.355
Kij	2.588	1.490	2.706	2.555
	2.613	2.706	2.684	2.688
	3.146	4.151	2.957	3.104
${\overline{K}}_{ij}$	0.863	0.497	0.902	0.852
	0.871	0.902	0.895	0.896
	1.049	1.384	0.986	1.035
Ri	0.186	0.887	0.091	0.183

, the greater the pile spacing, the higher the corresponding value of ${\overline{K}}_{ij}$, indicating increased tensile force in the geotextiles and greater differential settlement between piles and soil. When the pile spacing is 3 m, 4 m, and 5 m, the ${\overline{K}}_{ij}$ values are 0.497, 0.902, and 1.384, respectively, with range (R_i) values of 0.887. Hence, in regions with significant groundwater-level changes, the impact of pile spacing on differential settlement should be thoroughly considered in practical engineering. Table 7, along with

Table 8. Variance Analysis Table for Tensile Force Increment Caused by Periodic Groundwater Level Changes.

Factors	Sum of Squares of Deviations	Sum of Squares of Errors	F Value
A (pile length)	0.066	1.219	0.054
B (pile spacing)	1.183	0.595	1.987
C (embankment height)	0.015	1.272	0.012
D (groundwater level)	0.055	1.231	0.044

, reveals that the factors influencing the tensile force of geotextiles in pile-supported embankments under conditions of water-level change are ranked, in descending order of influence, as pile spacing, pile length, groundwater-level height, and embankment height.

6. Summary and Conclusions

The present study delves into the attenuation characteristics of silty clay under alternating dry–wet cycles through direct shear tests. A comprehensive numerical simulation model of a pile-supported embankment in full section under variable water levels has been developed, and its validity has been corroborated with engineering field data. Utilizing this model, the study elucidates the regulatory patterns of groundwater-level fluctuations on pile-supported embankments, analyzing the impact of constant groundwater-level height, rate of groundwater-level variation, and periodic groundwater-level changes on embankment settlement. Finally, based on orthogonal tests, the relative influence of various factors on embankment settlement deformation under cyclic water-level changes has been determined.

Direct shear tests conducted under different dry–wet cycles indicate that both the cohesion and internal friction angle of the silty clay decrease incrementally with increasing cycles of dry–wet alternation. The strength of the soil mass exhibits a rapid decline in the initial stages, which gradually slows and stabilizes, with the first dry–wet cycle exerting the most significant impact.

Results from the full-section numerical simulation model of the pile-supported embankment reveal that a higher constant groundwater head results in greater settlement of the soil between piles, though this effect on pile settlement is negligible. Concurrently, as the groundwater head decreases, the soil arching effect intensifies, transferring more load from the soil to the pile heads, which in turn reduces soil settlement between piles, lowers the uniform settlement surface of the embankment, and correspondingly diminishes the maximum tensile stress on the geogrids. The simulation further suggests that as the water level continues to drop, the effective stress within the soil gradually increases, leading to continuous compaction of the soil layers and settlement deformation of the ground surface. However, the maximum tensile stress in the geogrids follows a pattern of initial decrease followed by increase. This is due to the fact that in the initial stages of declining groundwater levels, the uneven settlement of the pile-soil system is significantly affected by the soil arching effect, thereby reducing the peak tensile force in the geogrids. As the groundwater level falls further, the uneven settlement of the pile-soil system is more greatly influenced by the increased effective stress in the soil, leading to an increase in the maximum tensile stress in the geogrids. Furthermore, the settlement values of the soil between piles exhibit a trend of initially rapid reduction followed by stability under conditions of increased rate of groundwater-level decline and longer cyclic periods of water-level changes. Moreover, the maximum tensile stress in the geogrids continues to increase with the intensification of the rate of groundwater-level decline and the cyclicity of water-level changes. These findings confirm that both the rate of groundwater-level decline and the cyclic variation in water levels significantly influence soil settlement and the maximum tensile stress in geogrids. Such factors should be taken into account in engineering design, construction, and monitoring to enhance the safety and long-term stability of pile-supported embankments during construction and throughout their service life.

Finally, through orthogonal tests of pile-supported embankments under cyclic groundwater-level changes, it has been identified that the influencing factors on the central settlement of the embankment, from significant to minor, under cyclic water-level changes are pile spacing, pile length, embankment height, and constant groundwater-level height, in that order. The factors affecting the tensile stress in the geogrids are ranked as pile spacing, pile length, constant groundwater-level height, and embankment height. Future research will further examine the impact of traffic loads on pile-supported embankments and elucidate the settlement deformation patterns of pile-supported embankments under the coupled action of groundwater levels and traffic loads.