In highway construction across the southeastern coastal regions of China, granite residual soil is widely used as subgrade fill material in pavement engineering. Its mechanical behaviour under dynamic loads warrants in-depth investigation. Dynamic events such as vehicular traffic and earthquakes are complex, involving multidirectional loads. The dynamic behaviour of soil under bidirectional cyclic loading differs significantly from that under cyclic loading in one direction. A large-scale bidirectional cyclic direct shear apparatus was utilised to carry on a series of horizontal cyclic direct shear tests on granite residual soil with water contents of 14% and 24% at different normal stress amplitudes (sigma a) (0, 100, 200 kPa). Based on these tests, discrete element method (DEM) models were developed to simulate the laboratory tests. The test results revealed that cyclic normal stress increases dynamic shear strength during forward shear but reduces it during reverse shear. The energy dissipation capacity increases with rising sigma a. The dynamic behaviour of granite residual soil is more significantly affected by cyclic normal stress when the water content is higher. The DEM simulation results indicated that as cyclic shearing progresses, the location of the maximum principal stress (sigma 1) shifts from the top of the specimen toward the shear interface. The distribution of the angle between sigma 1 and the x-axis, as well as sigma 1 and the z-axis, transitions from 'M' distribution to 'Arch' distribution. With increasing sigma a, during forward shear, the magnitude of the maximum principal stress increases, and the orientation of sigma 1 rotates toward the normal direction. Conversely, during reverse shear, the magnitude of the maximum principal stress decreases, and its orientation shifts toward the horizontal shear direction. The material fabric anisotropy coefficient decreases with increasing sigma a, while the anisotropy orientation increases with increasing normal stress.

Internal erosion induces alterations in the initial microstructure of soils, simultaneously affecting physical, hydraulic, and mechanical properties. The initial soil composition plays a crucial role in governing the initiation and progression of seepage-induced suffusion. This study employs the controlled variable method to develop granular soil models with varying particle size ratios, initial fine particle contents, and coarse particle shapes. Seepage suffusion simulations coupled with microstructural analyses are conducted using the CFD-DEM approach. Results demonstrate that particle size ratio, fine particle content, and coarse particle shape exert distinct influences on cumulative erosion mass, fine particle distribution, contact fabric, and mechanical redundancy at both macroscopic and microscopic scales. This numerical investigation advances the fundamental understanding of internal erosion mechanisms and informs the development of micro-mechanical constitutive models. Furthermore, for binary granular media composed of coarse and fine particles, careful control of the particle size ratio and fine content is recommended when utilizing gap-graded soils in embankment and dam construction to improve structural resilience and resistance to internal erosion.

Buried pipelines are essential for the safe and efficient transportation of energy products such as oil, gas, and various chemical fluids. However, these pipelines are highly vulnerable to ground movements caused by geohazards such as seismic faults, landslide, liquefaction-induced lateral spreading, and soil creep, which can result in potential pipeline failures such as leaks or explosions. Response prediction of buried pipelines under such movements is critical for ensuring structural integrity, mitigating environmental risks, and avoiding costly disruptions. As such, this study adopts a Physics-Informed Neural Networks (PINNs) approach, integrated with a transfer learning technique, to predict structural response (e.g., strain) of both unreinforced and reinforced steel pipes subjected to Permanent Ground Displacement (PGD). The PINN method offers a meshless, simulation-free alternative to traditional numerical methods such as Finite Element Method (FEM) and Finite Difference Method (FDM), while eliminating the need for training data, unlike conventional machine learning approaches. The analyses can provide useful information for in-service pipe integrity assessment and reinforcement, if needed. The accuracy of the predicted results is verified against Finite Element (FE) and Finite Difference (FD) methods, showcasing the capability of PINNs in accurately predicting displacement and strain fields in pipelines under geohazard-induced ground movement.

Liquefaction resistance and post-liquefaction shear deformation are key aspects of the liquefaction behavior for granular soil. In this study, 3D discrete element method (DEM) is used to conduct undrained cyclic triaxial numerical tests on specimens with diverse initial fabrics and loading history to associate liquefaction resistance and post-liquefaction shear deformation with the fabric of granular material. The influence of several fabric features on liquefaction resistance is first analyzed, including the void ratio, particle orientation fabric anisotropy, contact normal fabric anisotropy, coordination number, and redundancy index. The results indicate that although the void ratio and anisotropy strongly influence liquefaction resistance, the initial coordination number or redundancy index can uniquely determine liquefaction resistance. Regarding post-liquefaction shear deformation, the above quantities do not dictate the shear strain induced after initial liquefaction. Instead, the mean neighboring particle distance (MNPD), a fabric measure previously introduced in 2D and extended to 3D in this study, is the governing factor for post-liquefaction shear. Most importantly, a unique relationship between the initial MNPD and ultimate saturated post-liquefaction shear strain is identified, providing a measurable state parameter for predicting the post-liquefaction shear of sand.

This study presents a hierarchical multiscale approach that combines the finite-element method (FEM) and the discrete-element method (DEM) to investigate tunneling-induced ground responses in coarse-grained soils. The approach considers both particle-scale physical characteristics and engineering-scale boundary value problems (BVPs) simultaneously, accurately reproducing typical tunneling-induced mechanical responses in coarsegrained soils, including soil arching and ground movement characteristics observed in laboratory tests and engineering practice. The study also unveils particle-scale mechanisms responsible for the evolution of soil arching through the underlying DEM-based RVEs. The results show that the rearrangement of microstructures and the deflection of strong contact force chains drive the rotation of macroscopic principal stress and the formation of soil arch. The microscopic fabric anisotropy direction can serve as a quantitative indicator for characterizing soil arching zones. Moreover, the effects of particle size distributions (PSD) and soil densities on ground deformation patterns are interpreted based on the stress-strain responses and contact network characteristics of DEM RVEs. These multiscale insights enrich the knowledge of tunneling-induced ground responses and the same approach can be applied to other geotechnical engineering analyses in coarse-grained soils.

In practical engineering, earthquake-induced liquefaction can occur more than once in sandy soils. The existence of low-permeable soil layers, such as clay and silty layers in situ, may hinder the dissipation of excess pore pressure within sand (or reconsolidation) after the occurrence of liquefaction due to the mainshock and therefore weaken the reliquefaction resistance of sand under an aftershock. To gain more mesomechanical insights into the reduced reliquefaction resistance of the reconsolidated sand under aftershock, a series of discrete element simulations of undrained cyclic simple shear tests were carried out on granular specimens with different degrees of reconsolidation. During both the first (mainshock) and second (aftershock) cyclic shearing processes, the evolution of the load-bearing structure of the granular specimens was quantified through a contact-normal-based fabric tensor. The interplay between mesoscopic structure evolutions and external loadings can well explain the decrease in reliquefaction resistance during an aftershock.

The stress state and density of soil have been considered as the key factors to determine the liquefaction resistance. However, the results of seismic liquefaction case histories, laboratory tests and centrifuge model tests show that the fabric characteristics also influence liquefaction resistance, even more significantly than the contributions of stress state and density. In this study, anisotropic specimens with different consolidation histories were prepared using the 3D Discrete Element Method (DEM) to investigate the influence of fabric characteristics on the mechanical behavior of granular materials and the underlying mechanisms. The simulations revealed that under monotonic shear conditions, horizontally anisotropic specimens exhibited strain hardening and dilatancy characteristics, as well as higher peak strength. Under cyclic shear condition, the normalized liquefaction resistance of the specimens showed a strong linear relationship with the degree of anisotropy, independent of confining pressures and density. Microscopic results indicate that the fabric arrangement aligned with the loading direction leads to the evolution of the mechanical coordination number and average contact force in a manner favorable to resisting loads, which is the underlying mechanism influencing macroscopic mechanical properties. Additionally, the evolution patterns of contact normal magnitude and angle in anisotropic granular materials under cyclic loading conditions were also analyzed. The results of this study provided a new perspective on the macroscopic mechanical properties and the evolution of the microstructure of granular soils under anisotropic conditions.

A tensor-type capillary stress, instead of a scalar suction, has been proposed to serve as a stress-like state variable to capture the effects of capillarity in the mechanics of unsaturated granular soils. However, the influence of water content on the evolution of capillary stress in such soils remains insufficiently understood. This study performs numerical simulations of unsaturated granular soils in the pendular regime using the Discrete Element Method (DEM) involving a volume-controlled capillary bridge model. In these simulations, water content is maintained constant by redistributing the water from ruptured capillary bridges to adjacent ones. The evolution of capillary stress with varying water contents during triaxial and biaxial loading conditions is systematically examined. The DEM simulation results show that, under both loading conditions, the mean component of the capillary stress generally decreases, while its deviatoricity gradually develops. These changes are observed to become less significant as the initial degree of saturation increases. At low saturation levels, capillary bridges between non-contacting particle pairs rupture due to soil deformations, and the water from these ruptured bridges redistributes to existing contacts. This redistribution leads to an anisotropic distribution of pore water aligned with the contact network. At higher saturation levels, non-contacting capillary bridges persist due to their ability to sustain large relative displacements between particles, allowing the spatial distribution of pore fluids to remain less constrained by the solid contact network. Additionally, at higher water contents, relative sliding and particle rearrangement are the primary factors influencing the directional distribution of capillary bridges.

Laboratory experiments have shown that the proportional shearing of granular materials along arbitrary strain path directions will lead to stress states that converge asymptotically to proportional stress paths with constant stress ratios. The macro- and microscopic characteristics of this asymptotic behaviour, as well as the existence of asymptotic states exhibiting a constant stress ratio and a steady strain-rate direction, have been studied using the discrete element method (DEM). Proportional shearing along a wide range of strain-rate directions and from various initial stress/density states has been conducted. The simulation results suggest that general contractive asymptotic states (except for isotropic states) do exist but may be practically unattainable. Dilative strain path simulations, on the other hand, result in continuously changing stress ratios until static liquefaction occurs, indicating the absence of dilative asymptotic states. Despite this difference, a unique relationship between the stress increments and the current stress ratio gradually emerges from all strain path simulations, regardless of strain path direction and initial stress/density conditions. At the particle scale, the granular assembly sheared along proportional strain paths exhibits a constant partition ratio between strong and weak contacts. Although general proportional strain paths are associated with changing geometric and mechanical anisotropies, the rates of change in these anisotropies for contractive strain paths are synchronised to maintain a constant ratio of their contributions to the mobilised shear strength of the material, with a higher proportion being contributed by geometric anisotropy for more dilative strain paths.

受青藏高原暖湿化的影响，多年冻土呈现快速退化状态，并由此诱发大量的冻土滑坡灾害。为深入探讨多年冻土区滑坡失稳机制，本研究基于多年冻土滑坡区活动层（融土）粉土、黏土及相应土-冰界面直剪试验结果，通过离散元分析软件MatDEM对应修正了直剪试验模型，并将数值模拟结果与直剪试验结果进行对比分析。结果表明：修正后的直剪模型可有效地对粉土和黏土进行数值模拟计算；模拟与试验结果的剪切特性曲线及抗剪强度拟合曲线变化趋势基本保持一致，且黏土抗剪强度明显低于粉土，表明土体抗剪强度随土颗粒粒度的减小而降低；在土-冰界面处，黏土-冰的抗剪强度最低，土体稳定性最弱。模拟结果中位移场图、单元连接状态图均表明在剪切过程中形成了明显的剪切带，且融土界面处及非界面处颗粒的平均配位数变化规律显示剪切带内的颗粒在试样变形中起到主要作用。此外，热量变化曲线表明，剪切过程中的热量主要来源于剪切带中上下剪切盒产生的摩擦热。本研究可为高原冻土滑坡区土体抗剪强度数值模拟研究提供有效参考模型。