Acta Geotechnica最新文献

This paper investigates the microstructural mechanism that triggers diffuse instability in two-dimensional granular materials, utilizing directional statistics and Discrete Element Method simulations. Hill’s second-order work criterion is employed to identify the onset of diffuse instability at macroscale. A second-order work expression, formulated in terms of micro-variables, is derived based on a Stress-rate Force Fabric function and a Strain-rate Displacement Fabric function proposed for two-dimensional granular systems. The second-order work is theoretically originated from three independent aspects: contact density variation due to contact loss/gain and volume change, microstructure anisotropy evolution caused by rearrangement of the contact network and the contact force networks, and alteration of directional average normal contact force. The microstructure anisotropy and the directional average of the normal contact force are connected with the fabric anisotropy in the strong and weak sub-networks. Two dual mechanisms underlying diffuse instability are identified, including: (a) degradation of the microstructure characterized by a diminished peak in microstructure anisotropy combined with a volumetric dilating regime; (b) degradation of directional average normal contact force or buckling of the weak sub-network, which ultimately leads to the collapse of the entire microstructure. Moreover, these processes are associated with the contact density and coordination number approaching their critical thresholds.

Accurate characterization of soil shear modulus and damping ratio at small-to-medium strain levels is essential for reliable prediction of ground response under dynamic loads and for the safe design of geotechnical structures. Despite the development of numerous empirical models, their predictive accuracy remains a critical concern owing to the diversity of soil types and the complexity of field conditions. Recently, data-driven deep neural networks (DNNs) have emerged as a promising approach for modeling complex systems. However, their generalization capability is often limited by the scarcity of high-quality experimental datasets. To address this issue, this study proposes a multifidelity neural network (MFNN) to leverage the accuracy of the high-fidelity experimental datasets and the abundance of the low-fidelity datasets synthesized using the simple empirical model. The MFNN model can automatically learn the correlation between the low-fidelity and high-fidelity datasets. The model has been successfully applied to predict the small-to-medium strain shear modulus and damping ratio of soils. The results demonstrate significantly improved prediction capabilities of MFNN compared to purely data-driven DNNs. The proposed MFNN framework provides a new solution for modeling soil dynamic properties.

Experience from the oil and gas industry offers valuable insights that can be applied to renewable energy projects, particularly in offshore environments. This study explores the use of offshore drilling data to support the development of fixed offshore wind turbine foundations by improving predictions of the rate of penetration (ROP). Drilling data from seven wells in the X oilfield in the North Sea—focusing on the large-diameter 17-1/2″ interval—were analyzed using artificial neural network (ANN) regression. The model incorporated eight drilling parameters (depth, weight on bit, standpipe pressure, torque, maximum torque, surface round per minute, bit round per minute, and mud flow rate) across depths of 0–125 m below seafloor. An initial ANN model trained on a single well achieved 0.946 of the coefficients of determination, outperforming previous methods such as the Bourgoyne and Young model approach. When data from all seven wells were combined, the model maintained a high coefficient of determination of 0.883, despite variations in lithology. Depth, standpipe pressure, surface round per minute, and torque were identified as key factors influencing ROP. This study highlights the potential of machine learning in offshore drilling applications.

Monitoring data were collected for four years at a microbially induced desaturation (MID) test site in Portland, Oregon. MID is an emerging ground improvement technique for mitigation of liquefaction triggering. In MID, a treatment solution is injected into targeted soils to stimulate native denitrifying microbes that produce nitrogen gas, desaturating the soil. In field trials in the summer of 2019, MID successfully desaturated fine-grained, granular soils to a degree of saturation (S_r) below 98.5% as determined by crosshole pressure wave velocity (V_p) measurements, and to less than 96.6% based on in situ moisture sensor measurements and physical sampling. Monitoring data collected in the four years since the field trial included V_p measurements from a seismic crosshole array, seismic cone penetration test (SCPTu) profiles with V_p measurements, in situ measurements of water content and electrical conductivity, piezocone dissipation tests, groundwater levels, and bulk density measurements on Shelby tube samples. In situ moisture and electrical conductivity sensors, V_p measurements, and soil sampling were all consistent with respect to changes in S_r. Overall, the monitoring data show that a reduction in S_r due to MID treatment was sustained throughout the four-year monitoring period.

The stress distributions vis-à-vis failure characteristics of rectangular pillars are different from those of square pillars. Most of the studies explicate the characteristics of square pillars by analysing the effects of two geometric parameters, i.e. width and height. However, the effects of a pillar’s length are scanty in the literature. The increase in the length of the rectangular pillar leads to an increase in the strength of the rectangular pillar. Generally, the stability of rectangular pillars is evaluated by incorporating the concept of effective width in the existing strength formulae of square pillars. Nevertheless, existing effective-width formulae tend to overestimate rectangular pillar’s strength, potentially leading to unexpected failures. In this manuscript, a comprehensive study has been carried out for analysis of failure characteristics, 3D stress distribution and evaluation of strength by the concept of effective width of rectangular and long coal pillars. The characteristics of rectangular/long pillars are evaluated and suitable formulae of effective width are developed by the analytical and numerical simulation approaches. Parametric study through numerical modelling reveals that the rectangular pillar’s strength increases with the increase in its length and becomes asymptotic in nature. The results of the numerical simulation are used to derive the formula for effective widths of rectangular/long pillars by regression analysis. The developed formulae are validated by an adequate number of failed and stable rectangular and long pillar conditions. It is found that the existing effective-width formulae do not predict all the cases correctly, while the formulae developed in this study are well validated with the actual conditions of rectangular and long pillars. Thus, this study will help to improve safety through the proper design of rectangular and long pillars by understanding failure characteristics and stress distribution.

A precise understanding of strength–dilatancy response in marine coral sand containing fines is crucial for ensuring the stability of offshore infrastructure. This study systematically investigates the impact of fines on friction angle and maximum dilatancy angle, as well as their relationship, through meticulously controlled geotechnical tests, complemented by in-depth discussion and interpretation. The findings reveal that as fines content rises, the friction angles at both peak (({varphi }_{text{ps}})) and critical states (({varphi }_{text{cs}})), as well as the excess friction angle (({varphi }_{text{ex}})) and maximum dilatancy angle (({psi }_{text{max}})), show a notable decline. Conversely, a higher relative density leads to an increase in ({varphi }_{text{ps}}), ({varphi }_{text{cs}}), ({varphi }_{text{ex}}), and ({psi }_{text{max}}). Additionally, higher stress levels lead to a reduction in ({varphi }_{text{ps}}) and ({psi }_{text{max}}), while their impact on ({varphi }_{text{cs}}) and ({varphi }_{text{ex}}) remains inconclusive, possibly due to minimal particle breakage under elevated stress conditions. Furthermore, this study defines the upper and lower bounds of variation in ({varphi }_{text{cs}}) and ({varphi }_{text{ex}}) relative to ({varphi }_{text{ps}}) across different stress levels and density states. The transition region where the mixture shifts from sand-controlled to fines-dominated behavior was examined. A significant observation is the relationship between ({varphi }_{text{ex}}) and ({psi }_{text{max}}), showing that Bolton’s strength–dilatancy theory, initially developed for clean sand, still applies to marine coral sand provided the fines content stays below a specific limit. This insight highlights the necessity of accounting for fines content when applying established dilatancy model to marine coral sand.

Unsteady flows of granular materials immersed in viscous fluids are commonly encountered in natural and industrial processes, exhibiting pronounced sensitivity to intrinsic properties of both particles and fluids, such as the particle–fluid density ratio. Understanding their complex dynamics and the underlying mechanisms, especially the intricate particle–fluid interactions at the pore scale, is challenging but crucial. In this study, a high-performance framework coupling the lattice Boltzmann method and discrete element method was employed for immersed granular flows. Experimental measurements of buoyant granular column collapses were taken to validate the accuracy and efficiency of the coupling algorithm. The influence of particle–fluid density ratios on the collapse dynamics of immersed granular columns was explored through macro- and microscopic analysis. Results show that collapses exhibit larger front positions and more efficient energy conversions as particle–fluid density contrasts increase, with floating particles achieving longer runout distances and higher front velocities compared to settling cases. Micromechanical analysis reveals that in settling scenarios, an enhanced contact force network can be formed to inhibit particle sliding during the initial stage, leading to longer collapse durations and slower dynamics, whereas floating columns tend to experience catastrophic failures. Furthermore, the particle Stokes number (St) serves as a governing parameter for immersed granular flows, exhibiting a strong power-law correlation with the normalized runout distance across various length scales.

The complexity of engineering problems generally makes it challenging to obtain explicit physical formulations and sufficient monitoring data. For the issue of shallow tunnelling-induced deformation, it is difficult to obtain an explicit analytical solution due to the anisotropy of in situ stress, asymmetric free surface boundary, and complicated elastic–plastic response. Physics-informed neural networks (PINNs) have demonstrated excellent capability in resolving boundary value problems. In this context, a data-driven and physics-informed neural network is developed to predict tunnelling-induced deformation. The underlying elastic–plastic governing equations and boundary constraints of a shallow-buried tunnel are encoded into a deep neural network framework, which is divided into two independent regions according to the Mohr–Coulomb yield criterion, ensuring that constitutive relations are executed in elastic and plastic regions separately. To mitigate nonlinearity and stress concentration effects, a global adaptive sampling strategy guided by probability distributions is introduced, which ensures a more efficient approach to enforcing physics laws. Comparisons between the analytical and PINN-based solutions of deformation induced by deep tunnel excavation demonstrate the robust performance of the proposed model, especially with the implementation of a global adaptive sampling strategy. For the solution of shallow tunnelling-induced ground deformation, the adaptive PINN model can accurately reproduce the elastic–plastic ground settlement fields with sparse labelled data. The data mining and physical information exchange character of the proposed adaptive PINN model demonstrates the promising potential of the scientific machine learning method in addressing engineering issues.

The durability of materials treated with the microbiologically induced calcium carbonate precipitation (MICP) technique has garnered increasing attention in the field of geotechnical engineering. A series of studies have been reported on the durability of biotreated soils, but relatively little attention has been focused on the durability of MICP-treated rock fractures, especially on the microscopic degradation characteristics. In this study, surface characterization, mass measurement, Brazilian splitting tests, and nitrogen adsorption tests were conducted on the specimens subjected to dry–wet cycles to investigate the multiscale degradation characteristics of sandstone with biotreated fractures. The surface of the specimen displayed loss of fines and dissolution of calcium carbonate with increasing number of dry–wet cycles, which resulted in decreasing mass, reducing tensile strength, and increasing specific surface area. The different trends in the variations of the surface fractal dimension and pore-structure fractal dimension indicate that the increase in micropores and complexity of pore structure are the main characteristics in the early stages of dry–wet cycling, followed by a significant increase in surface roughness. This study suggests the benefits of early repair for the sustainability of MICP-treated rock fractures in dry–wet cycle environments.