Influential Mechanisms of Roughness on the Cyclic Shearing Behavior of the Interfaces Between Crushed Mudstone and Steel-Cased Rock-Socketed Piles

Abstract

In the waterway construction projects of the upper reaches of the Yangtze River, crushed mudstone particles are widely used to backfill the foundations of rock-socketed concrete-filled steel tube (RSCFST) piles, a structure widely adopted in port constructions. In these projects, the steel–mudstone interfaces experience complex loading conditions, and the surface profile tends to vary within certain ranges during construction and operation. The changes in boundary conditions and material profile significantly impact the bearing performance of these piles when subjected to cyclic loads, such as ship impacts, water level fluctuations, and wave-induced loads. Therefore, it is necessary to investigate the shear characteristics of the RSCFST pile–soil interface under cyclic vertical loading, particularly in relation to varying deformation levels in the steel casing’s outer profile. In this study, a series of cyclic direct shear tests are carried out to investigate the influential mechanisms of roughness on the cyclic behavior of RSCFST pile–soil interfaces. The impacts of roughness on shear stress, shear stiffness, damping ratio, normal stress, and particle breakage ratio are discussed separately and can be summarized as follows: (1) During the initial phase of cyclic shearing, increased roughness correlates with higher interfacial shear strength and anisotropy, but also exacerbates interfacial particle breakage. Consequently, the sample undergoes more significant shear contraction, leading to reduced interfacial shear strength and anisotropy in the later stages. (2) The damping ratio of the rough interface exhibits an initial increase followed by a decrease, while the smooth interface demonstrates the exact opposite trend. The variation in damping ratio characteristics corresponds to the transition from soil–structure to soil–soil interfacial shearing. (3) Shear contraction is more pronounced in rough interface samples compared to the smooth interface, indicating that particle breakage has a greater impact on soil shear contraction compared to densification.

1. Introduction

The completion of the Three Gorges Dam has significantly advanced the development of inland waterway construction along the upstream Yangtze River. Rock-socketed concrete-filled steel tube (RSCFST) piles, known for their unique structures, are widely used [1]. The base of these piles is composed of reinforced concrete embedded in the rock strata, while the upper section consists of a steel tube in contact with backfilled crushed mudstone, forming the soil–steel interface, as illustrated in Figure 1. RSCFST piles constitute an inherently end-bearing friction pile system, where the vertical load-bearing capacity is distributed between the tip and shaft, leading to a complex load distribution. Shear displacements induced by cyclic loads such as ship impacts, water level fluctuations, and wave-induced loads will gradually weaken the shear strength between the steel tube and backfill-crushed mudstone particles. This deterioration can compromise the bond strength at the pile–soil interface and the durability of the structure, potentially resulting in increased maintenance costs and shortened service life. Over time, the decline in dock-bearing capacity and reduced service life may significantly limit the Yangtze River channel’s cargo throughput, hindering economic development in the upper Yangtze region. Therefore, investigating the factors influencing the bearing capacity of RSCFST piles is crucial. Previous studies on RSCFST piles primarily apply numerical models and laboratory-scale models to investigate the effects of various factors on the pile’s bearing capacity [2,3,4,5,6,7,8]. For example, He et al. [3] explore the dynamic impedance and response of large-diameter rock-socketed single piles under harmonic loading using a finite–infinite element combined model, considering variables such as embedment depth, rock weathering degree, and excitation frequency. Xu et al. [8] investigate the skin friction behavior and ultimate skin friction of the pile–mudstone interface for completely, highly, and moderately weathered mudstone using in situ self-balanced pile tests. Some researchers have also investigated the degradation of pile foundation-bearing capacity caused by the fatigue deformation of steel tubes under horizontal cyclic loading [2]. These studies typically considered the entire pile as a uniform entity, assuming a consistent outer profile along the pile shaft. In practical engineering, the roughness of the steel casing can vary at different locations, and roughness is a critical factor influencing the shear resistance and strain characteristics of soil–structure interfaces. However, research on the effects of roughness on the shear behavior of the RSCFST pile–soil interface under vertical cyclic loading has rarely been reported. Therefore, this study adopts a microscale perspective to investigate the influential mechanisms of roughness on the shear behavior of the RSCFST pile–soil interface under cyclic loading.

The soil–structure interface has long been a primary concern in geotechnical engineering, commonly observed in shallow foundations, pile foundations, diaphragm walls, pipelines, tunnels, and other structures that rely on surface friction for loadbearing [9]. The direct shear test is a widely utilized method for investigating soil–structure shearing mechanisms. Scholars often use this test to summarize and deduce the relationship between shear strength and various influencing factors. For example, Clough and Duncan [10] reveal that the relationship between contact shear stress and relative shear displacement is hyperbolic. Uesugi and Kishida [11] determine that the relationship between surface roughness and the coefficient of friction at yield is independent of the mean particle size. Fakharian and Evgin [12] find that both the reduction in normal stress and the increase in interface shear amplitude lead to the attenuation of maximum shear stress. Some scholars focus on studying the strain characteristics of the shear interface. For example, Zhang [13] finds that the volumetric strain due to the stress–dilatancy behavior of sand is found to be composed of a reversible dilatancy component and an irreversible dilatancy component. Zhang and Zhang [14] observe that shear deformation is composed of slippage at the contact surface and deformation of the soil constrained by the geotextile. In most engineering projects, buildings are subjected to cyclic loads. Consequently, cyclic shearing is a common boundary condition considered by scholars [15,16,17,18,19,20,21,22]. In these studies, factors such as the particle size, particle angularity, shear amplitude, shear rate, and surface roughness are investigated.

Among soil–structure interactions, the roughness of the pile–soil interface critically influences load transfer and deformation between the pile and the soil [18,23]. Therefore, numerous scholars have extensively studied the impacts of surface roughness. Al-Douri and Poulos [24] investigated the static and cyclic behavior of various carbonate sediments and silica sand, finding roughness to be a significant factor influencing static shear behaviors. Later, scholars employed various methods to quantify the surface roughness of structures. For instance, Mortara et al. [25] defined steel plate surface roughness using a normalized roughness value and conducted a series of cyclic shear tests, finding that due to the absence of dilation on smoother interfaces, no recovery of shear stress was observed in the post-cyclic phase. A similar study was also conducted by Wang et al. [26], where they adopted the average roughness value to quantify the surface profile. To investigate the shear behavior of foundations and frozen soils in permafrost regions, Zhao et al. [27] conducted monotonic and cyclic shear tests on artificially frozen silt–structure interfaces under varying temperatures and roughness, where the roughness was defined as the maximum peak-to-valley height. Chen et al. [28] utilized a large direct shear apparatus to investigate the effect of roughness on the shear strength of the red clay–concrete interface, defining concrete surface roughness using a modified silica powder clamping method. Yang et al. [29] conducted a series of shear creep tests on clay–concrete interfaces with varying roughness and normal stresses to investigate the effect of the joint roughness coefficient (JRC) on long-term cohesion and friction angle. Jitsangiam et al. [30] conducted a series of cyclic direct shear tests on a granular soil–rough interface under constant normal load (CNL) conditions to investigate the effects of cyclic shear amplitudes and loading sequences, and monotonic interface direct shear tests were performed to examine the post-cyclic behavior of the soil–rough interface immediately after the cyclic shear tests were completed.

In a prior study, we introduce an enhanced large-scale direct shear apparatus to examine the cyclic shear behavior of the RSCFST pile–soil interface [31]. The apparatus’s stability and efficacy in replicating the actual conditions of the steel–soil interface are confirmed through two sets of parallel tests. Subsequently, under specified initial normal stress (INS = 400 kPa), number of cycles (200), and roughness (R = 1), the cyclic shear behavior of the RSCFST pile–soil interface was analyzed under CNS and CNL boundary conditions. Given the significant influence of loading conditions and surface roughness on the mechanical properties of the pile–soil interface, and the limited research on the shear behavior of the RSCFST pile–soil interface, it is essential to extend previous studies to explore the impact of surface roughness. In fact, the cyclic behavior of pile–soil interfaces is better investigated using CNS direct shear tests, as any volume change in the pile–soil interface zone is constrained by the soil beyond this zone, and the normal stress on the interface may decrease or increase when the interface zone contracts or dilates [32], as illustrated in Figure 2. Therefore, the cyclic interface pile–soil interface’s behavior for pile foundation problems is usually investigated by employing the CNS direct shear apparatus [21,33,34,35,36]. Based on this, a series of cyclic direct shear tests are conducted under CNS boundary conditions under four initial normal loads to investigate the cyclic shear characteristics of the RSCFST pile–soil interface with four different roughness levels. Initially, the transformation of the shear stress–shear strain curve with the number of cycles is plotted, and the anisotropy of shear stress is presented by comparing the maximum shear stress in different shear directions within a single cycle. Subsequently, two parameters, shear stiffness and damping ratio, are employed to characterize the dynamic behavior of the interface. Then, the shear contraction and dilation of the soils are characterized by the relationships between normal stress and normal strain and between normal stress and the number of cycles. Finally, the gradient of soil particles within a 3 cm thickness from the shear surface after tests is statistically analyzed, and the particle breakage ratio index B_g is introduced to characterize the fragmentation of particles at the shear interface. The structure of this paper is as follows: an introduction is provided in Section 1; a detailed description of the apparatus and materials used is given in Section 2; the results and phenomena observed in the tests are presented in Section 3; the influential mechanisms and characteristic patterns are discussed in Section 4; and the conclusions are listed in Section 5.

2. Test Methodology

2.1. Materials

2.1.1. Soil Specimen

In this study, the mixed soil specimen consisted of lightly weathered sandstone and mudstone blocks. The mudstone blocks were collected at the construction site of Guoyuan Port in Chongqing, where RSCFST piles are widely used. To measure the uniaxial compressive strength and shear strength of the sandstone and mudstone in both their natural and saturated states, uniaxial and triaxial tests were conducted using the RMT-150C rock mechanics testing system (Institute of Rock and Soil Mechanics, Chinese Academy of Sciences, Beijing, China). The average uniaxial compressive strength of the mudstone tested in the laboratory was 21.0 MPa (natural state) and 11.5 MPa (saturated state), while that of the sandstone was 66.1 MPa (natural state) and 63.6 MPa (saturated state). The dry density of the mudstone was 2.4 g/cm³. The Young’s modulus and Poisson’s ratio of the mudstone were 4.3 GPa and 0.35, respectively. The mass percentage of the mudstone particle in the soil specimen was 20.0%, which is close to the mudstone content in practice. The cohesive strength and angle of internal friction of the mudstone were 5.2 MPa and 40.0° (natural state) and 3.8 MPa and 37.8° (saturated state), respectively.

To unify the particle size distribution of the soil samples, the collected mudstone blocks were manually crushed into gravel with pieces smaller than 110 mm, followed by further crushing using an EP-3B jaw crusher (Hebi Minsheng Science and Technology Development Co., Ltd, Hebi, China) to achieve a particle size ranging from 3 to 40 mm. Subsequently, the particles were preliminarily sieved using a sand screening machine and then refined with a ZBSX-92 laboratory standard vibrating sieve shaker (Shaoxing Shangyu Yueda Instrument Manufacturing Co., Ltd, Shaoxing, China) employing sieve apertures of 20 mm, 10 mm, 5 mm, 2 mm, 1 mm, 0.5 mm, 0.25 mm, and 0.075 mm. The particle groups with sizes from 2 to 5 mm, 5 to 8 mm, and 8 to 10 mm were selected to prepare the specimens, as shown in Figure 3a. The proportions of these three groups followed the composition of an actual fill material in the field. The particle size distribution curve is shown in Figure 3b. The uneven coefficient, C_u, and the curvature coefficient, C_c, were 2.3 and 1.0, respectively. To ensure the consistency of the initial conditions, the moisture content of the specimen was in accordance with the air-dried moisture content, 0.8%, and the density in this condition was 1.7 g/cm³. The prepared mudstone particles were then sealed separately according to their particle sizes and stored in a hermetic environment.

2.1.2. Steel Plate with Saw Teeth

Customized steel plates with saw teeth were employed in our test to maintain consistency with the steel tube used in the RSCFST pile, as illustrated in Figure 3. The density, elasticity modulus, and Poisson’s ratio of the steel material were 7.8 g/cm³, 200.0 GPa, and 0.3, respectively.

Scholars have applied several approaches to define roughness from a surface profile in their tests, such as the mean roughness R_a [26,37,38], maximum peak-to-valley elevation R_t within a sampling length L_r, maximum vertical peak-to-valley elevation R_max over a reference length Lm [39,40], where Lm is normally taken as either an absolute length in millimeters, or the mean particle size D₅₀ [41], and the normalized roughness R_n [11,39,42]. In our study, the peak-to-valley distance of the V-shaped saw teeth on the steel plate surface, denoted as R and measured in millimeters (mm), represented the roughness of the steel plate surface, as illustrated in Figure 4.

2.2. Test Setup

2.2.1. Apparatus

The adopted apparatus is illustrated in Figure 5. Designed and improved by Liang et al. [31], the main components consisted of a shear box, a normal actuator, a horizontal actuator, and a supporting frame. The upper shear box comprised a detachable iron plate and a three-sided iron plate to form a hollow iron cuboid with an inner diameter of 20 cm × 20 cm × 10 cm, and it was firmly fixed to the left support base. The lower shear box was placed on the rollers to ensure that it could move freely in the horizontal direction, and at the top of lower shear box was a customized steel plate providing distinct roughness, with a size of 28 cm × 28 cm, equal to the outer diameter of the shear box. Then, a normal actuator and a horizontal actuator were installed at the top of the upper shear box and the right end of the lower shear box, respectively; these servo-controlled hydraulic actuators had a maximum push–pull capacity of 100.0 KN. Additionally, each actuator was equipped with stress and displacement sensors with resolutions of 10 N and 0.01 mm, respectively. A special pressure plate with four springs was placed between the upper shear box and the normal actuator to achieve the CNS boundary condition. Each spring had an outer diameter of 50 mm, a free length of 180 mm, and a maximum compressed length of 35 mm. The stiffness of each spring was calibrated using a universal testing machine, and this showed that the relationship between deformation and the applied force was linear. The stiffness of a single spring was 225.66 N/mm, and the total stiffness of the four springs is 905.24 N/mm. This configuration generated a maximum normal stress of 792.1 kPa at a compression length of 35 mm in the shear box, significantly exceeding the maximum normal stress of 500 kPa required for this study. Therefore, the selected springs met the test requirements.

2.2.2. Test Procedure

Prior to the shear test, mudstone particle groups with sizes from 2 to 5 mm, 5 to 8 mm, and 8 to 10 mm were air-dried first and then quantitatively weighed according to the grading curve. These particles were evenly mixed and divided into three portions. Water was added to each portion and stirred until the moisture content reached 0.8%. The specimens were then sealed in a hermetic environment for 12 h to ensure uniform soil moisture distribution. Before installing the specimen, white petroleum jelly was applied to the bottom of the upper shear box to reduce friction with the lower shear box. After ensuring close contact between the upper and lower shear boxes, the specimen was loaded into the upper shear box in three stages, with each stage compacted using a compaction instrument to control density. Once the specimen was compacted and flushed with the top of the upper shear box, a pressure plate capable of applying CNS boundary conditions was placed on top of the specimen. Then, the pressure plate was connected to the normal actuator, and the position was adjusted to bring the pressure plate into contact with the sample surface. The power and data acquisition control systems were subsequently activated, the single shear cycle time was set to 58 s, the maximum shear displacement was set to 15 mm, the number of shear cycles was set to 100, the data collection interval was set to 0.64 s, and the normal pressures were set to 300 kPa, 400 kPa, and 500 kPa according to prior settings. Upon reaching the designed normal pressure, the horizontal actuator immediately applied shear stress to the lower shear box to prevent specimen consolidation from affecting the test results. In each cycle, the initial motion direction was designated as the positive direction, and the shear displacement of the sample changed from positive to negative and then returned to the origin in a cycle, and the movement of the lower shear box is illustrated in Figure 6. After the test, the instrument was disassembled from top to bottom, and soil specimens located within 3 cm of the contact surface were collected for particle breakage analysis. All experimental conditions are summarized in

Table 1. Summary of general information in cyclic direct shear tests under CNS conditions.

Test Number	Initial Normal Stress (kPa)	Roughness (mm)	Total Spring Stiffness (N/mm)
T-1	300	0	905.2
T-2	300	1
T-3	300	2
T-4	300	5
T-5	400	0	905.2
T-6	400	1
T-7	400	2
T-8	400	5
T-9	500	0	905.2
T-10	500	1
T-11	500	2
T-12	500	5

.

3. Results and Analysis

3.1. Effects on Shear Stress

Figure 7a–l shows the relationship between shear stress and shear displacement under three normal loads and four different levels of roughness, where N = 1, 5, 30, and 100, respectively. It can be seen that maximum shear stress occurs at the first cycle and decreases successively as the test progresses, indicating that the steel–mudstone interface is characterized as experiencing cyclic softening with increasing umber of cycles. For a given normal stress, the shear stress–strain curve of the interface with no saw teeth retains a parallelogram shape before and after the test, and only slight attenuation in shear stress is observed in the initial cycle, while for the rough steel interface (R = 1, 2, 5), the shape gradually transforms from rhombus-like shape in the initial cycle to a parallelogram shape in the final cycle, and greater roughness leads to more evident attenuation in shear stress. Previous studies indicate that the degradation of shear resistance at soil–structure interfaces is primarily caused by particle rearrangement and breakage [15,32,43,44]. According to Figure 7a–l, the stress–strain curve for the smooth interface shows no significant reduction in shear stress throughout the test, whereas the rough interface exhibits a considerable decrease in shear stress. This confirms the occurrence of pronounced particle rearrangement and breakage within the rough interface (further supported by the particle breakage rate data in Section 3.4). By comparing the stress–strain curves, it can be inferred that the nonlinear phase of shear stress-shear displacement observed in the rough interface corresponds to particle rearrangement and breakage, as illustrated in Figure 7b–d. In contrast, the smooth interface experiences only minor particle rearrangement and breakage, resulting in a predominantly linear relationship between shear stress and shear displacement before yielding, as illustrated in Figure 7a. In addition, when N = 1, greater roughness has a larger area enclosed by the shear stress–strain curve, i.e., the average shear stress is greater. On the contrary, when N = 100, greater roughness has a smaller area, i.e., the average shear stress is smaller. For a given roughness, R = 0, 1, and 2, and the greater the normal stress, the larger the area enclosed by the curve. However, at R = 5, there is no significant difference in the area enclosed by the curve when N = 100. This may be due to the fact that, under CNS conditions, the decrease in soil specimen height is close to the total compression of the springs when the number of cycles reaches 100, resulting in no significant difference in normal stress, and hence the final shear stress–strain curve area is similar under the three INS levels.

Figure 8a–l shows the variation in maximum shear stress in the positive and negative directions with the number of cycles, where the values corresponding to the dashed lines in the figure represent the difference in shear stress in two directions. These results show that maximum shear stress in both directions decreases with the number of cycles at a decelerating rate when R takes 1, 2, and 5. However, when R takes 0, the maximum shear stress decreases with the number of cycles within the initial ten cycles and then increases with the number of cycles when normal stress increases. The decrease in shear stress corresponds to particle fragmentation and attenuation of normal stress, and the uprising behavior on the flat interface may contribute to the increment of friction between soils and steel plates caused by the densification of specimens. The maximum shear stress in the positive direction always exceeds that in the reverse direction, indicating the presence of anisotropy at the interface. Greater surface roughness results in more pronounced interface anisotropy during the initial stages of shear, which gradually decreases with cycles. This decline trend is more prominent for R = 2 and R = 5 compared to R = 0 and R = 1. After 100 cycles, rougher surfaces tend to exhibit lower shear anisotropy. Similar observations can also be found in Wang et al.’s [26] work, in which they conducted cyclic shear tests on concrete–clay interfaces under constant normal load (CNL) conditions with up to 1000 cycles. Their findings indicated that higher roughness leads to greater shear anisotropy. They attribute this to the complex surface topography of rougher surfaces, which leads to intense adjustment of soil particles during cyclic shear. However, according to their results, rougher surfaces exhibited greater shear anisotropy even after 1000 cycles, contrasting with our findings. We attribute this discrepancy to differences in boundary conditions. In our experiment, the CNS boundary condition was applied, causing the specimen’s height to gradually decrease as the test progressed, which led to a corresponding reduction in normal stress. In the later stages of shearing, the rate of normal stress attenuation decreased. Assuming that the internal friction angle and soil cohesion remain constant within a single cycle, according to the Mohr–Coulomb criterion, the attenuation ratio of the shear stress decreases correspondingly. This results in a smaller difference between the maximum positive and negative shear stresses within a single cycle, thereby reducing shear anisotropy. In contrast, Wang et al. employed the CNL boundary condition, where normal stress remained constant throughout the test. Under these conditions, shear stress primarily depends on the interface friction angle, which increases with surface roughness [5,28,45,46]. Consequently, roughness shows a positive correlation with interface anisotropy.

3.2. Effects on Shear Stiffness and Damping Ratio

To characterize the dynamic behavior of the pile–soil interface during cyclic shear processes, parameters, including equivalent secant shear stiffness and damping ratio, are introduced. Desai et al. [47] modified this approach and employed it in characterizing the dynamic response of sand–concrete and sand–geosynthetic interfaces during shear. Figure 9 presents a typical hysteresis loop used to calculate the modified secant shear stiffness and damping ratio using Equations (1) and (2).

$$K={\displaystyle \frac{K_1+K_2}{2}}={\displaystyle \frac{{\tau }_1+{\tau }_2}{2A_w}}$$

(1)

where K₁ and K₂ represent the secant shear stiffness in the first and second half-cycle, respectively, A_w is the shear semi-amplitude, and τ₁ and τ₂ correspond to the shear stress at the shear semi-amplitude in the positive and negative direction, respectively.

$$D={\displaystyle \frac{D_1+D_2}{2}}={\displaystyle \frac{1}{2}}\left({\displaystyle \frac{S}{4\pi S_1}}+{\displaystyle \frac{S}{4\pi S_2}}\right)={\displaystyle \frac{S}{4\pi A_w}}\left({\displaystyle \frac{1}{{\tau }_1}}+{\displaystyle \frac{1}{{\tau }_2}}\right)$$

(2)

where D₁ and D₂ represent the damping ratio in the first and second half-cycle, respectively, S is the hysteresis loop area, and S₁ and S₂ are shadow areas.

In Figure 10, the relationship between shear stiffness and the number of cycles is plotted. It can be seen that the variation in shear stiffness with the number of cycles is generally characterized by a gradual descent. A significant decline is observed within the initial several cycles, and then the shear stiffness continues to decrease at a decelerating rate from about 5 to 20 cycles, depending on the magnitude of roughness. Eventually, the decline stabilizes at a low rate. Interestingly, the shear stiffness on flat surfaces begins to show a gradual increasing trend after approximately 10~20 cycles, and this phenomenon is more evident when subjected to greater INS. For a given normal stress, greater roughness has larger initial shear stiffness but smaller final shear stiffness, as shown in Figure 10a, where the initial shear stiffness values at R = 0, 1, 2, and 5 are 11.07, 13.22, 15.20, and 17.34 kPa/mm, respectively, while the final shear stiffness values at R = 0, 1, 2, and 5 are 9.63, 5.09, 3.30, and 1.70 kPa/mm, respectively. The observed variations in shear stiffness at smooth and rough interfaces are consistent with the previously described trends in shear stress. This is because, in our tests, the shear semi-amplitude A_w remains constant, and, according to Equation (1), shear stiffness is positively correlated with the average absolute maximum shear stress in both the positive and negative directions within a single cycle, hence resulting in similar trends. By analyzing the intersection points of curves for different roughness levels, it becomes evident that the enhancement in shear stiffness due to roughness is short-lived. For R = 5, the shear stiffness falls below that of the smooth interface after only six cycles, while for R = 2 and R = 1, this occurs after three and two cycles, respectively. This indicates that, although greater roughness corresponds to higher initial shear strength at the interface, under cyclic loading, this enhancement is transient and accelerates interface degradation.

Figure 11a–c illustrates the evolution of the damping ratio in the soil–steel interface with the number of cycles under varying roughness and INS levels. In general, the damping ratio of the flat interfaces initially decreases, then increases, while it displays an opposite trend in rough interfaces, an initial increase followed by a decrease. Throughout the experiment, it was observed that the damping ratio exhibited extremum points (refer to the inflection points in Figure 11). For rough interfaces under the same INS, surfaces with higher roughness required more cycles to reach the inflection points. In contrast, for smooth interfaces, the number of cycles needed for the damping ratio to reach the inflection points fell between those required when R = 1 and R = 2. However, under certain experimental conditions, no extremum points were observed for the damping ratio, whereas under other conditions, two extremum points emerged. For instance, in Figure 11a, when R = 1, the damping ratio significantly increases after one cycle and then remains almost constant without decreasing. In Figure 11b, the damping ratio at R = 5 consistently exhibits a fluctuating upward trend. In Figure 11c, the damping ratio at R = 0 and R = 1 initially decreases, then slightly increases, and finally gradually decreases. Furthermore, it can be observed that during the first cycle, the greater the roughness, the smaller the damping ratio. When R = 0 and 1, the variation in the damping ratio is relatively limited with the increase in the number of cycles. It is only when R = 2 and 5 that a significant variation in the damping ratio can be observed. The damping ratio of the interface material is significantly influenced by its physical properties. When the roughness is minimal, the disturbance area on the soil is confined, maintaining the structural stability of soil and resulting in negligible damping ratio variations. With an increase in roughness, the self-adjustment mechanism fails to counteract this external disruption. Fine particles, resulting from the fragmentation of coarse particles, migrate downwards through the pores and settle in the grooves. Once the grooves are filled with particles, particle fragmentation and slippage primarily occur due to the compression and rolling between particles. The underlying steel plate is unable to directly participate in shearing, leading to a transition in the shear type at the shear interface from soil–steel shearing to soil–soil shearing. This alters the mode of energy dissipation at the interface; therefore, a noticeable change in the damping ratio with the number of cycles is observed at the rough interface. In addition, the densification of the specimens under different normal stresses varies, which also impacts the damping ratio of the interface.

3.3. Effects on Normal Stress

Figure 12a–l presents the variation in normal stress with shear displacement under varying roughness levels and number of cycles. In general, normal stress displays a gradual decrease as the shear test progresses, and thereafter the soil specimens are characterized by shear contraction behavior. However, within a single cycle, both shear dilatancy and contraction are observed in certain instances as the height of the soil specimen experiences periodic upward and downward movement. This behavior is more frequently observed in the initial few cycles, particularly when the surface is rougher. In the later stages of tests, abundant coarse particles crushed into finer particles, coupled with the enhancement of densification of the soil, render both shear dilatancy and contraction behavior less discernible. A similar phenomenon is observed in the study of Zhang and Zhang [48], who argue that the volumetric strain due to stress–dilatancy behavior at the contact surface comprises two components: a reversible dilatancy component and an irreversible dilatancy component. The reversible shear dilatancy is primarily associated with reversible changes in the average orientation rate of the sand particle assemblage induced by relative sliding and particle rotation during the shearing process, while the irreversible shear dilatancy, on the other hand, stems from the fragmentation of sand particles, a decrease in average porosity, and the disappearance of larger pores during the shearing process.

Comparing the normal stress–shear displacement path of smooth interfaces, it is evident that normal stress markedly decreases at different roughness levels. This suggests that particle breakage at the interface primarily contributes to irreversible volumetric strain, with densification being a secondary factor. Moreover, roughness amplifies the effects of particle ascent and descent over the structural surface ribs, thereby accentuating reversible volumetric strain as roughness increases.

Figure 13a–c presents the relationship between the normal stress (the last value within a single cycle) and the number of cycles. These results show that as the cyclic shear test progresses, the normal stress exhibits a decelerating decrease. This behavior can be segmented into three stages: initially, the normal stress decreases rapidly within the first 5 cycles, with 60%~70% of the reduction in normal stress; in the second stage, the decreasing rate begins to slow down before reaching about 15 cycles (R = 0, 1, 2) or 50 cycles (R = 5), with 10%~20% (R = 0, 1, 2) or 35% (R = 5) of the reduction in normal stress; and in the final stage, the decreasing rate stabilizes at a low level, with approximately 10% ~ 20% (R = 0, 1, 2) or 5% (R = 5) of the reduction in normal stress. In the first stage, the specimen particles fracture into finer particles under the combined effect of substantial normal stress and shear stress, with the fine particles filling the voids to compact the specimen, leading to a significant shear contraction strain. In the second stage, the normal stress decreases due to the shear contraction behavior, with particle fracturing primarily induced by shear stress, and thus the rate of shear contraction strain decreases, leading to a gradual reduction in the rate of decrease in normal stress. In the third stage, both the normal stress and shear stress are relatively low, with particle fracturing primarily resulting from mutual compression and rolling, and the shear contraction behavior is not pronounced, and hence the normal stress tends towards stability.

To better quantify the impact of surface roughness on normal stress, the attenuation ratio S_σN is introduced, calculated by Equation (3):

$$S_{\sigma N}={\displaystyle \frac{{\sigma }_0-{\sigma }_N}{{\sigma }_0}}\times 100\%$$

(3)

where S_σN is the cumulative ratio of attenuation up to the N-th cycle (in Figure 14, N takes at 100), σ₀ is the INS, and σ_N is the last measured value of normal stress in the N-th cycle.

According to Figure 14, when the roughness R = 0, the average value of S_σN is 35.59%. For roughness values of R = 1, 2, and 5, the average values of S_σN are 61.14%, 74.35%, and 97.35%, respectively. This indicates that rough surfaces significantly reduce the confining pressure at the pile–soil interface under prolonged vertical cyclic loading. In fact, under CNS boundary conditions, the attenuation in normal stress is influenced not only by surface roughness, but also by volumetric contraction of the specimen under compression, which leads to a continuous decrease in normal stress due to the constant normal stiffness.

3.4. Effects on Particle Breakage Ratio

In our study, particles ranging from the shearing interface to one-third the height of the specimen were collected after each test. The particle size distribution was then determined using sieving tests, as in Figure 15a–c. It is observed that the particle size distribution curves all exhibit conspicuous elevation after the test, with a more pronounced rise in elevation associated with greater roughness. For a given roughness, an augmentation in normal stress correspondingly escalates the particle size distribution curve, although the impact is less conspicuous compared to the effect induced by an amplification in roughness. The particle size distribution curves provide an intuitive representation of the significant degradation that roughness inflicts upon the structure of the soil specimen.

The particle breakage ratio is introduced to better quantify the relationship between roughness and the degree of particle fragmentation. Various scholars and institutions, including Marsal [49], Miura and Yamanouchi [50], and Hardin [51], have proposed formulations to quantitatively assess particle fragmentation. In this paper, we adopt the method introduced by Marsal, who proposed the index B_g to evaluate the degree of particle breakage. This index is calculated using Equation (4):

$$B_g=\Sigma \mathrm{\Delta }{w}_k$$

(4)

where w_k represents the positive difference in soil particle content for each particle size group before and after the test; therefore, in our study, B_g refers to the mass fraction of particles with a diameter below 2 mm.

Figure 16a–c delineates the statistical outcomes of B_g under different roughness and INS, as well as the mass fractions of particles with different sizes. It is observed that B_g increases with the augmentation of roughness, albeit at a gradually diminishing rate. We believe that this phenomenon may be attributed to the fact that, within a certain range, the augmentation in roughness significantly elevates the interlocking force between the soil particles and the steel plate saw teeth. Consequently, the resistance of the specimen to shear deformation increases, leading to a greater degree of particle breakage during the shearing process. However, when the roughness surpasses a certain degree relative to the D₅₀ of the specimen, a portion of the soil particles become fully embedded within the grooves. This leads to an expansion in the soil–soil contact area and a contraction in the soil–structure contact area, consequently diminishing the interlocking force offered by the saw teeth. Therefore, beyond this point, an increase in roughness does not result in a significant change in the particle breakage ratio. In addition, it is found that when R = 0, the mass fraction of particles with a diameter of 1–2 mm is the highest, followed by 0.075–0.25 mm, 0.5–1 mm, 0.25–0.5 mm, and <0.075 mm. Conversely, when R = 1, 2, or 5, the mass fraction of particles with a diameter of 0.075–0.25 mm is the highest, followed by 1–2 mm, 0.5–1 mm, 0.25–0.5 mm, and <0.075 mm. Within the flat interface, solely the surface layer particles in contact with the steel plate surface undergo fragmentation under the influence of normal and shear stress. The coarse particles located deeper within the specimen are safeguarded due to densification, resulting in a higher residual content of coarse particles. Within the rough interface, the saw teeth disrupt the soil particles beneath the surface layer, compelling more coarse particles to participate in fragmentation, resulting in a higher degree of particle breakage; therefore, the content of fine particles significantly increases.

4. Discussion

4.1. Roughness and Interface Strength

Under CNS conditions, the influence of roughness on interface strength manifests both as an enhancement and a diminution. In the initial stage, compared to a flat interface, the shear stress of a rough interface is much greater, but with the increase in the number of cycles, the shear stress of the rough interface is far less than that of the flat interface. On the one hand, roughness implies a higher degree of interlocking force between the soil and the steel plate, where the transverse ribs between the saw teeth provide a considerable part of the shear strength. Under CNS conditions, the influence of roughness on interface strength manifests both as an enhancement and a diminution. In the initial stage, compared to a flat interface, the shear stress of a rough interface is much greater, but with the increase in the number of cycles, the shear stress of the rough interface is far less than that of the flat interface, which may be attributed to variations in shearing types. According to Chen et al. [28], when the shear failure slip plane is located on the concrete surface, the interface friction angle approximates that of a smooth interface; when the slip plane occurs within the clay matrix (soil–soil shearing), the interface friction angle aligns with the red clay’s friction angle; and when the slip plane lies at the clay–concrete interface, the friction angle fluctuates between the red clay’s friction angle and the smooth interface’s friction angle. To validate the applicability of this theory to the present study, we calculated the interface friction angles for R = 1 and R = 0 under 1, 10, 20, 50, 75, and 100 cycles. The results, shown in Figure 17, reveal that under R = 1, the interface friction angle consistently lies between the values of the smooth interface and the intrinsic friction angle of weathered mudstone particles (40°, as referenced in Section 2). Furthermore, as the tests progress, the interface friction angle increasingly approaches the intrinsic friction angle of weathered mudstone. This suggests a gradual transition of the shear failure slip plane from the steel surface into the mudstone specimen, indicating a shift in interface shear type from soil–structure shearing to soil–soil shearing. In the early stages of the cyclic shear tests, the interface is characterized by soil–structure shearing, where roughness contributes to a higher degree of interlocking force between the soil and the steel plate. In the later stages, the grooves between asperities are filled with fragmented fine particles, and the ribs on the steel plate no longer directly participate in the shearing process. The shear failure slip plane then migrates into the specimen (soil–soil shearing), transitioning the interface shear type to soil–soil shearing.

Roughness primarily affects interface strength during the initial soil–structure shear stage, where the shear plane is located between the structure’s surface and the soil, demonstrating a strong correlation between roughness and interface shear strength. However, in the later stages of cyclic shear, the rough surface becomes covered with crushed fine particles, and the shear plane moves above the surface. At this stage, roughness no longer influences interface shear strength, which becomes predominantly determined by the particle size distribution of the interface soil and normal pressure, with no direct relation to roughness.

4.2. Three Shearing Stages

It can be seen that normal stress, shear stress, and shear stiffness all exhibit a nonlinear relationship with the number of cycles, which can be divided into three stages: the rapid decline stage, the transition stage, and the stable stage, as shown in Figure 18a. In the first stage, the original structure of the soil specimen exhibits a certain degree of resistance to shearing forces. However, once the shearing force surpasses the critical shear strength of the interface, it triggers extensive particle sliding and fragmentation immediately, leading to severe structural damage. This process endures as displaced and fragmented particles seek new voids to settle down into. In the second stage, an increasing number of “finer particles” generated by fragmentation gradually fill the pores and grooves within the interface. At this point, the arrangement of soil particles within the interface becomes increasingly compact, the interlocking force provided by the saw teeth weakens, and the shear strength provided by the internal forces between coarse particles decreases. Consequently, the rate of shear contraction in the specimen and the rate of attenuation of the interface strength slow down. Hence, a deceleration in the decline of normal stress and interface strength is observed. In the third stage, the space between the steel plate and the undisturbed soil of the upper specimen is mainly filled with fine particles, as shown in Figure 18b. At this point, the fine particles primarily serve a lubricating function, resulting in a significant decrease in particle breakage. The specimen exhibits no apparent shear contraction, and the shear stress remains nearly constant.

The observed shearing behavior of the soil–steel interface serves as a foundation for parameter determination in the evaluation of the load-bearing capacity of RSCFST pile foundations in practical engineering applications. Particularly in regions experiencing substantial water level variations, recurrent seismic activity, or considerable shifts in dock cargo volume, the lateral friction coefficient of piles should be categorized into distinct ranges, contingent on the operational lifespan of the dock. This stratification ensures a more accurate evaluation of the actual load-bearing capacity of pile foundations.

4.3. Suggestions for Future Research and Engineering

This study investigates the effect of roughness on the shear behavior of RSCFST pile–soil interfaces under CNS boundary conditions by conducting a series of cyclic direct shear tests. The results demonstrate that under prolonged vertical cyclic loading, rougher RSCFST pile–soil interfaces exhibit reduced shear performance. To ensure the long-term load-bearing capacity of RSCFST pile foundations, steel tubes with smooth and uniform surface profiles should be prioritized during the design phase. During penetration, measures should be implemented to prevent significant deformation of the steel tubes caused by surrounding rocks, and weld protrusions should be minimized. Nevertheless, this study may have several limitations. In practical engineering, pile foundations may undergo vertical shear cycles far exceeding the 100 cycles set in our test, which restricts the applicability and reliability of our results. Additionally, the absence of a scaling ratio in this study limits the practical applicability of our findings. Future research should adopt more realistic boundary conditions and appropriate scaling ratios for field conditions to improve the relevance of the results for engineering applications.

5. Conclusions

Cyclic direct shear tests on a steel–mudstone interface were carried out for four different roughness levels under CNS boundary conditions where INS took 300 kPa, 400 kPa, and 500 kPa, respectively. The variations in shear stress, shear stiffness, damping ratio, normal stress, and particle breakage ratio were investigated to study the influential mechanisms of roughness on the cyclic dynamic response of the RSCFST pile–soil interface, which is extensively employed in ports within the mountainous regions of the upper Yangtze River. The concluding re-marks of this study can be summarized as follows.

(1)

During the initial stages of cyclic shearing, surface roughness amplifies interface shear strength and anisotropy. Conversely, in the subsequent stages, greater roughness leads to lower shear strength and anisotropy. On the one hand, enhanced roughness in the initial cycles induces greater saw tooth-particle interlocking forces, bolstering the shear resistance of the interface, necessitating a more extensive particle rearrangement and fragmentation, thereby increasing both the interface strength and anisotropy. On the other hand, however, the intensified interlocking force disrupts the soil structure more comprehensively, producing a larger volume of fine particles with diminished angularity and dip direction during the shearing process, which results in a reduction in both interface strength and anisotropy.

(2)

The shearing type of the interface and the content of fine particles significantly impacts the variation characteristics of the damping ratio. In the cyclic shear process, the damping ratio of rough interfaces initially increases before decreasing with the number of cycles, whereas the flat interface demonstrates a converse trend with a less pronounced variation. The alteration pattern for the flat interface is predominantly associated with the densification of soil. As for the rough interface, the shear type gradually transitions from soil–steel shearing to soil–soil shearing with the increment in the number of cycles. Consequently, the amplified content of fine particles and the diminished contribution of the interlocking force between the saw teeth and specimen particles to the shear strength induce a substantial change in the variation in the damping ratio.

(3)

Under CNS boundary conditions, particle breakage is the primary cause of specimen shear contraction. Under the influence of cyclic shearing with CNS boundary conditions, both shear contraction and dilation were observed during the experiment, and the general trend of volumetric strain demonstrates shear contraction, while the phenomenon of shear dilatancy occurs within individual cycles. The attenuation of normal stress and the particle breakage rate at smooth interfaces is significantly lower than at rough interfaces, indicating that particle breakage, rather than densification, is the primary cause of specimen shear contraction.

(4)

Our test results indicate that, under CNS boundary conditions, the enhancement of cyclic shear strength at the RSCFST pile–mudstone interface due to roughness is significant but short-lived. Future research should explore strategies to balance the advantages and disadvantages of rough surfaces to optimize pile foundation engineering by enhancing the ultimate bearing capacity without compromising the service life of piles.