Publications

Deep-sea mining is typically conducted on the seabed with water-depths exceeding several thousand meters, necessitating accurate understanding of soil mechanical behavior under ultra-high back pressure (bp) while preserving conventional effective stress states. However, current high-pressure triaxial systems fail to achieve sufficient measurement accuracy for effective stress, and investigations on undrained behavior of sand under ultra-high back pressure remain scarce. To overcome this fundamental constraint, we propose a testing method that incorporates an auxiliary high-precision differential pressure sensor into high-pressure devices, enabling direct measurement of effective stress and substantially improving measurement accuracy. Moreover, the principles of calibration procedures and additional techniques to further enhance accuracy are demonstrated. To isolate the role of bp, undrained monotonic and cyclic tests are performed on quartz sand under bp ranging from 300 kPa to 29 MPa utilizing this enhanced high-pressure triaxial apparatus. The results demonstrate that bp variations neither alter the slope of the critical state line in p′–q space nor significantly affect the evolution of effective stress ratios during shear, confirming the validity of the effective stress-based mechanical principle under ultra-high hydrostatic pressure. Furthermore, an energy-based analytical framework reveals a unique relationship between normalized energy dissipation and pore pressure evolution during cyclic loading regardless of bp.

Understanding the deformation and non-coaxiality behavior of granular materials under pure principal stress rotation (PSR) is a highly challenging topic in soil mechanics. However, there has been limited research investigating the influence of particle morphology on these behaviors. In this study, drained torsional shear tests with PSR stress path were performed on four glass beads with different particle morphologies, under various intermediate principal stress coefficients and stress levels. Microstructural analysis using X-ray CT scanning of dummy cylinder specimens was employed to support the interpretation of the experimental data. Results reveal that as particle shape becomes more irregular, the deformability and shear modulus of granular materials initially increase and then decrease, while non-coaxiality consistently increases. A moderate increase in the shape irregularity leads to an increase in the mean coordination number Z̄ in particle packing, while a highly irregular particle shape results in a decrease in Z̄ and a looser fabric, which is detrimental to shear modulus. The contact normal directions in non-spherical materials are more concentrated, exhibiting a stronger initial fabric anisotropy that potentially promotes non-coaxiality. Additionally, the proportion of deviatoric strain that is non-coaxial with deviatoric stress tensor significantly increases as particle shape becomes more irregular.

This study investigates the combined effects of particle morphology and the intermediate principal stress ratio b on the multiscale mechanical behavior of granular materials. Four distinct particle morphologies, ranging from nearly spherical to highly irregular, and with or without intrapores, were obtained via X-ray tomography (μCT) scanning and digitally reconstructed. A series of virtual true triaxial tests with varying b values was subsequently conducted on cubic particle assemblies using the level set-discrete element method (LS-DEM). Macroscopically, both the peak and critical stress ratios exhibit a monotonic dependence on b, characterized by an initial rapid decline, followed by a gradual reduction before stabilizing. In contrast, the influence of b on dilatancy becomes significant only at relatively high values. Microscopically, the mean coordination number correlates linearly with the friction angle across all morphologies and stress conditions. Analysis of shear-induced anisotropy and particle rotation characteristics reveals that more complex morphologies promote the development of more anisotropic internal structures and restrict the rolling motion of individual particles by locking them in place to enable higher bulk strength. In contrast, increasing b reduces the contact anisotropy and facilitates particle rotation, thereby decreasing the shear resistance. These findings contribute to an improved understanding and modeling of granular materials with realistic particle morphologies under complex stress paths.

Particle-scale tracking and breakage analysis are essential for understanding granular mechanics, but existing methods struggle with irregular/crushable particles and large interval tracking. To solve these challenges, this study presents a novel framework that formulates particle tracking as a graph optimization problem, solved through physics-informed spatiotemporal optimal transport (PSOT-Track). The proposed method uniquely integrates: (1) spatiotemporal graphs encoding particle neighborhood relationships, (2) physics-based costs (mass conservation and shape similarity) for transport optimization, and (3) breakage-aware matching via fragmented particle reassembly. Validation using X-ray microcomputed tomography (μCT) data sets (9,248 regular lentil particles, 4,765 irregular Fujian sand particles, and 1,727 irregular porous coral sand particles) demonstrates superior performance: 99.5%/93.1%/97.3% accuracy under small tracking interval (≤5% strain), maintaining over 70% accuracy at large tracking intervals (>5% strain), achieving a 50% improvement over conventional methods. Furthermore, PSOT-Track successfully identifies 97.1% of splitting and 85.7% of chipping breakage, highlighting its potential for breakage analysis. The proposed approach establishes a generic and robust paradigm for particle-scale tracking and breakage analysis, paving the way toward a deeper understanding of the complex granular micromechanics.

The micromechanical behaviors of coral sands remain poorly understood, primarily due to the inherent complexity of their highly irregular particle shapes, which pose significant difficulties for accurate three-dimensional (3D) particle reconstruction and tracking using X-ray tomography (µCT). To address these challenges, this study proposes a novel framework that integrates large vision models and discrete label optimization for efficient and accurate 3D particle reconstruction with optimal transport for robust particle tracking. This framework effectively resolves tracking both particle breakage and internal voids in coral sands, as validated by in situ mini-triaxial µCT tests. Compared with the state-of-the-art method, the proposed approach achieves comparable reconstruction accuracy (90%) while reducing computational time by 50%. For particle tracking, accuracy between adjacent µCT scans (corresponding to axial strain increments of 2.5%, 5%, and 10%) reached 95%, 86%, and 71%, respectively. Micromechanical analysis of coral sands further reveals that heterogeneous local shear deformation develops as axial strain increases, forming an X-shaped shear band. Significant fabric anisotropy emerges after peak stress, with preferred orientations aligning with the shear band. Moreover, particle breakage was observed to occur primarily during the strain-softening stage. Splitting induced by stress concentration was the predominant failure mode.

Granulated sand–rubber mixtures have been widely employed in ground improvement. Previous studies have extensively investigated the mechanical behavior of such geomaterials under conventional triaxial conditions, but little is known about the influence of varying principal stress directions (α_σ) on their strength and deformation characteristics. To address this gap, drained hollow cylinder torsional shear tests were conducted, including monotonic tests with different α_σ values and cyclic tests involving cyclic principal stress rotation (CPSR), on Fujian sand mixed with granulated rubber in six different mass ratios. The test results reveal that the inclusion of granulated rubber has a significant impact on the variation of peak strength with respect to α_σ. As the rubber content (R_c) increases, the strength anisotropy of the mixture reduces, and the strength failure surface in the (σ_z−σ_θ)/2–τ_zθ plane transitions from a circular to an elliptical shape with a noticeably reduced radius in the τ_zθ-axis direction. A formulation was proposed to characterize the strength envelope of sand–rubber mixtures. The inclusion of granulated rubber promotes the contractive behavior of the mixture. Under CPSR cyclic loading, higher rubber content leads to an increase in various strain components. The addition of rubber particles was observed to reduce the mixture’s noncoaxiality under CPSR, but this effect reduces at higher stress levels.

Particle morphology, as intrinsic properties of granular materials, significantly affects the mechanical response of granular materials, but little attention has been paid to cyclic response together with monotonic behavior. In this study, six cereals with similar grain sizes but different shapes were selected as testing materials due to their minimal variation in particle morphology. First, three-dimensional particle shapes of six cereals were obtained using X-ray microcomputed tomography (μCT). Then, a series of monotonic and cyclic triaxial tests were conducted on all cereals with and without staining. The staining could effectively eliminate the surface characteristic differences among six cereals, enabling them to be employed to better investigate particle shape effects. Meanwhile, since the surface characteristics of cereals, such as surface roughness and surface-adhered starch, underwent significant changes after staining, their effects could also be analyzed by comparing the results of cereals before and after staining. Repose angle tests were conducted to investigate changes in the surface roughness of six cereals after staining, which showed that staining had no significant effect on the surface roughness of mung but significantly increased it for all other cereals. Triaxial test results showed that the peak friction angle decreased with increasing particle shape regularity and decreasing surface roughness, while accumulated volumetric strain during cyclic tests decreased under both conditions. Furthermore, the cereals with surface-adhered starch exhibited quite different mechanical behavior than conventional granular materials. The starch bond under confining pressure significantly enhanced their shear strength and reduced their compressibility. However, the specimens experienced an abrupt strength reduction and volume contraction when the deviatoric stress reached the critical threshold of starch bond rupture, which required special attention during the storage and transportation of cereals.

Particle abrasion is a critical phenomenon in the study of granular materials, as it leads to a progressive decrease in the mechanical strength of the granular aggregates. In this study, we present a novel computational framework based on Firey’s law for predicting the abrasion-induced shape evolution of railway ballast. This approach provides a more efficient alternative to laboratory experiments for generating abraded particle shapes for use in numerical simulations. Using 3D scans of fresh ballast particles, we simulate their abrasion process at various levels of mass loss and statistically analyze the resulting shape changes. We then adopt the level-set discrete element method (LS-DEM) to investigate their mechanical response under triaxial compression. Macroscopically, both the stress ratio and void ratio decrease with increasing abrasion level, which can be attributed to abrasion-induced corner rounding and surface smoothing. Microscopically, the decrease in stress ratio is associated with the loss of highly stable contact subnetworks characterized by larger contact deviation angles, and the weakening of contact network anisotropy. Meanwhile, the decrease in void ratio is attributable to the transition of ballast particles toward more ellipsoid-like shapes, which promotes denser packings.

The influence of particle shape on the shear modulus at very small strain (G_max) of granular materials remains poorly understood and correlated. Using both micro-CT and bender element tests, this study aims to further systematically investigate this influence by comparing six granular materials with distinct particle shapes. The study included materials with angular and rounded particles, as well as relatively spherical and moderately angular particles, with particle morphological factors assessed using micro-CT. A series of bender element tests was conducted on these materials under various relative densities (D_r) and mean effective stresses (p′). Additionally, computed tomography (CT) technique was employed to interpret the role of particle shape on G_max from a microstructural perspective. The test results reveal that under the same relative density condition, as the irregularity of particle shape increases, the G_max of the materials first increases and then decreases. Angular materials exhibit the lowest G_max values, primarily due to their larger void ratio, while the mediumly angular materials display the highest G_max values compared to rounded and angular materials. Additionally, it was observed that overall regularity (OR) can be used to describe the significant transitional G_max response of granular material in relation to the variations in particle morphology. As OR decreases, the sensitivity of G_max to p′ initially decreases and then increases, which was found to be related to the shape-dependent particle mean coordination number (Z̄). Notably, in materials with an extremely low Z̄ value, G_max exhibits a significantly faster increase with p′. Consequently, based on test data from granular materials with a wide range of particle shapes and transitional G_max responses, practical equations for correlating the parameters of G_max prediction model with particle morphology were formulated and validated.

Three-dimensional (3D) particle reconstruction from X-ray micro-computed tomography (µCT) images is essential for digital twins and understanding the micromechanical behaviors of granular media. Despite large vision models (LVMs) having shown remarkable effectiveness across various domains, their application to accurate 3D particle reconstruction remains underexplored. This study proposes a systematic framework that enhances and leverages LVMs for the reconstruction of arbitrary 3D particles. The proposed framework includes three key steps: (1) enhancing LVMs with higher computational efficiency for two-dimensional (2D) label extraction, (2) mapping stacked 2D labels to 3D, and (3) extracting particle surfaces to generate 3D models. The enhanced approach is applied to reconstruct four distinct samples to validate feasibility. Six conventional and four lightweight LVMs are selected to explore the influence of model size and the number of prompts on reconstruction accuracy. The H-extreme watershed method is chosen as a benchmark for comparison. The results demonstrate that the enhanced framework can accurately reconstruct irregular and complex samples, such as carbonate sands, with a greater than 50% improvement in accuracy compared to the benchmark prediction. Additionally, the framework effectively reduces over- and under-segmentation errors, resulting in accurate reconstruction of microstructural characteristics. This versatile framework presents a promising alternative for investigating complex micromechanical mechanisms of granular media.

Reconstructing three-dimensional (3D) granular microstructures through X-ray micro-computed tomography (μCT) imaging is significant for elucidating micromechanical behaviors of granular media and optimizing geotechnical designs. However, due to the irregular morphology and dense packing of granular media, traditional image-processing techniques often lack the precision required for accurate reconstructions. This paper presents a novel framework for accurate 3D reconstruction of arbitrary granular media using vision foundation models (VFMs). Two-dimensional (2D) mask maps representing the granular media are extracted from μCT images along the x, y, and z-axes using VFMs and then processed by a two-step strategy to repair textures and remove noises. An optimal transport (OT)-based method is employed to reconstruct a complete 3D mask map based on 2D mask maps. The proposed method is applied to reconstruct two carbonate sand samples with irregular grain shapes and four lentil samples composed of nearly 10,000 grains captured in triaxial loading, utilizing various VFMs and prompt configurations. The framework demonstrated a 50% improvement in reconstruction accuracy over the state-of-the-art method for carbonate sands and achieved a 95% accuracy for lentil samples. This advancement offers a more effective alternative for investigating the micromechanics of granular media.

The major principal stress direction angle (α_σ) experienced by granular soils varies widely in engineering, causing different strengths. However, how particle morphology affects the strength anisotropy behavior under different α_σ remains unclear. To address this gap, this study performed drained hollow cylinder torsional shear tests under different α_σ on six granular materials with distinct morphologies. Results highlight the significant dependence of peak strengths of granular materials on both particle morphology and α_σ. Increasing particle shape irregularity and surface roughness leads to a considerable enhancement in peak strength, while this peak strength significantly degrades with increasing α_σ. Materials with more irregular shapes were found to have a more pronounced strength anisotropy. Furthermore, the initial fabric of particle packings, derived from three-dimensional X-ray microtomography, was used to interpret microscopic mechanisms behind the morphology-dependent strength anisotropy. Irregular-shaped materials display broader preferred particle orientations and higher initial fabric anisotropy compared to relatively regular-shaped materials. This higher morphology-induced fabric anisotropy contributes to strength anisotropy, and a correlation was established for describing this trend. Additionally, an anisotropic failure criterion incorporating fabric anisotropy was developed to characterize the strength envelope for granular materials with diverse shapes.

Accurate three-dimensional (3D) reconstruction of granular grains from X-ray micro-computed tomography (µCT) images is a long-standing challenge, particularly for dense soil samples. This study develops a machine learning (ML) enhanced approach to automatically reconstruct granular grains from µCT images. The novel academic contributions of this paper include (a) a hierarchical strategy based on parameter-independent polygonal approximation, area, and concavity analysis, for the first time, to identify and eliminate both intergranular and intragranular voids; (b) incorporation of a recursive segmentation scheme and ML-based grain classifier to avoid over-segmentation; (c) novel modifications on the determination of splitting paths to enhance segmentation accuracy; and (d) an effective approach of assigning initial level set functions for reconstructing granular grains automatically. The hybrid ML algorithm is applied to µCT images of dense Mojave Mars Simulant. The results indicate that the proposed method can accurately segment grain clumps with unclear boundaries. The new automatic reconstruction algorithm eliminates ineffective operations and achieves a three-fold increase in computational speed than previous methods documented in the literature. Ninety-one percent of grains with distinct boundaries can be reconstructed and the reconstruction ratio reaches 81% even for grains without distinct boundaries. The overall reconstruction ratio of grains increases by 20% compared with previous methods, achieving a step-change improvement for one-to-one mapping of real soil samples.

Reproducing realistic particle shape is important for discrete modeling of granular construction materials, such as sands and gravels, which are widely used foundation soils in civil engineering. The existing algorithms to acquire particle shapes from the raw images of construction materials are still very limited and can hardly deal with binary mixtures, such as sandy gravel. To address this issue, this study aims to develop a systematic framework for realistic simulation of gravel-sand mixture based on deep-learning-enhanced discrete element method (DEM). An efficient and convenient method is proposed to quickly identify particle contour and establish particle shape libraries based on the combination of YOLOv5 and U-Net algorithms. Furthermore, the DEM-based biaxial compression tests are conducted on two groups of gravel-sand mixture based on the acquired realistic coarse and fine particle shapes. The influences of coarse particle shapes and fine sand content on the macroscopic and microscopic behaviors of gravel-sand mixtures are quantitatively and comparatively studied. The proposed framework for deep-learning-enhanced particle shape acquisition and realistic DEM simulation will provide the researchers with more convincible physics-based insights into granular mechanics and has the potential to be extended into 3D to benefit the practical problem.

Fiber reinforcement techniques can effectively optimize the engineering properties of ground soils. This study aims to investigate the influence of fiber stiffness on the shear behavior of fiber-reinforced granular soil. First, the particle shapes are automatically extracted based on the improved Viola–Jones algorithm. Subsequently, the fiber stiffness is calibrated considering the results of laboratory tensile tests. Moreover, a series of numerical biaxial tests are conducted considering different fiber stiffnesses and contents and a flexible boundary. The influence of the fiber stiffness on the strength and volumetric strain is analyzed, and the stress–dilatancy relationship of the fiber-reinforced soil is discussed. In addition, the interaction between the fibers and soil particles is clarified according to the internal structure and stress network, including the coordination number, contribution of the contact, sliding of the particles and mobilization of the tensile force in the fibers. The results of this study can help clarify the reinforcement mechanism of fibers with typical stiffness values.

Fast and accurate three-dimensional (3D) particle reconstruction from X-ray micro-computed tomography (μCT) images is crucial for understanding granular micromechanics and developing digital twins. In this study, we propose a novel method that employs a large vision model (LVM) for efficient and precise 3D particle reconstruction. First, the point-prompted LVM extracts two-dimensional (2D) particles from μCT images along two arbitrary orthogonal axes. Then, these 2D segmentations are optimized through a discrete relabeling scheme, which operates on the principle that segmentations along one axis can inform and improve segmentations along the other axis. Following this optimization process, 2D segmentations are transformed into one 3D segmentation, where each particle is assigned a unique label value. The effectiveness of this method is validated on a reported dataset comprising 2,123 carbonate sand particles. Compared to the benchmark, our method achieves a 10% improvement in accuracy while reducing computational time by 43%. These results demonstrate the superiority of our method in 3D particle reconstruction.