Acta Geotechnica最新文献

Crack segmentation in soils poses significant challenges in geotechnical applications, particularly due to the progressive evolution of microcracks into dense and complex networks during desiccation. These temporal variations in crack morphology, significant for understanding soil behavior under environmental and mechanical stresses, have been overlooked in previous studies. To address these challenges, a novel cross-temporal soil crack instance segmentation pipeline is proposed in this research. Specifically, a local aggregation transformer is designed to effectively capture diverse crack patterns through adaptive token aggregation. Furthermore, a crack spatial feature pyramid is developed to model the hierarchical spatial characteristics of crack patterns at different stages of desiccation. To accommodate the varying densities of cracks during desiccation, an adjustable segmentation non-maximum suppression mechanism is proposed, which dynamically adapts its suppression strategy based on the density of bounding boxes. The proposed framework is validated through end-to-end training and demonstrates breakthrough performance on both speckle and non-speckle datasets. These findings enhance the understanding of soil mechanical properties and provide a basis for improving geotechnical stability assessments and foundation design methodologies.

This study proposes a novel three-fidelity physics-informed neural network (TF-PINN) to solve forward and inverse problems in tunneling applications. The proposed model incorporates: (i) low-fidelity data generated by numerical simulations, (ii) mid-fidelity physics-based analytical patterns for closed-form solution, (iii) high-fidelity field data from a tunnel case study, leveraging the underlying physical information to simulate the soil–tunnel interactions within a complete simulation environment. The model hierarchically trains three-fidelity learning with uncertainty quantification and transfer learning to explicitly enhance training efficiency and preserve critical physical knowledge. TF-PINN is applied to analyze the interaction mechanism between the shield and the surrounding strata. Results reveal that the proposed model exhibits inherent advantages in handling physical constraints and data-driven approaches over existing physics-informed methods. TF-PINN provided the highest correlation coefficient (92.5%) and lowest root mean square error (0.052), demonstrating the precision of the testing model in forecasting the ground deformation. The proposed method can be utilized in the early design process to give accurate estimations for soil–tunnel deformations when data are limited.

The theory of large-strain consolidation of high compressibility soft soil under time-dependent drainage at the boundaries involves complex material and geometric nonlinearities. Existing analytical solutions are applicable to specific working conditions only and are not suitable for general evolving compression and permeability laws. A Physics Informed Neural Network (PINN) is here introduced to solve the Partial Differential Equation (PDE) governing the nonlinear large-strain consolidation induced by a loading applied on the top of the soft soil layer, under exponentially time-growing drainage boundaries. To describe the nonlinearities linked to the mechanical behavior and the permeability of the soil, double-logarithmic models are employed. To avoid issues related to the length and timescales governing the problem, a scaling transformation of the entire solution is also exploited. The effectiveness of the proposed PINN-based approach is reported via a comparison with available analytical solutions. After this preliminary assessment, the effects on the consolidation process of drainage, initial stress, nonlinear mechanical and permeability properties of the soil are analyzed. The PINN-based approach is reported to significantly reduce the difficulty in solving the nonlinear consolidation problem characterized by large strains and overcome the deficiency of existing methods to provide exact solutions only for some specific cases.

In geotechnical site investigation, Robertson’s soil behavior type (SBT) chart is widely used for soil classification based on two quantities measured during a cone penetration test (CPT), the normalized cone resistance Q_t and the normalized friction ratio F_R. Q_t and F_R are negatively correlated and provide complementary information for soil classification. However, this cross-correlation between Q_t and F_R has not been explicitly modelled in previous studies of subsurface soil classification and stratification using an often-limited number of CPT soundings from a specific site. This study aims to leverage such cross-correlation for improving CPT-based stratification and zonation by a joint sparse representation of Q_t and F_R in a vertical cross-section, as well as quantifying their uncertainty under a Bayesian framework. In addition, direct application of the SBT chart to a vertical cross-section often leads to noisy results (e.g., SBTs fluctuate rapidly and unrealistically within short distances). The noises are subsequently removed mainly by subjective engineering judgment in current practices. In this study, a randomization of input measurements is proposed to filter out the noise and improve computational efficiency simultaneously. Both simulated and real data examples are used to illustrate the proposed method. The results indicate that the proposed method significantly improves accuracy of the soil classification and stratification and automatically removes the noise.

This study investigates the impact of clay shale reconstitution on its hydro-mechanical behaviour, using grain size distribution as a key parameter to assess the degree of sample preparation. This topic is of practical relevance, as the characterisation of fine-grained soils often relies on their intrinsic properties, typically measured after reconstitution. While fine-grained soils that can be homogenised with minimal mechanical effort follow a fairly uniform preparation process, this is not the case for materials requiring initial pulverisation. To address this, three batches of remoulded Opalinus Clay samples (fine, medium, and coarse) were prepared. Their hydration curves, swelling pressure, and compressibility behaviour were analysed. Additional classification tests, pore size distribution measurements, and X-ray diffraction (XRD) analyses were performed to support physical characterisation. XRD revealed no preferential breakage of specific clay minerals or segregation of grain sizes. Test results indicated that preserved bonding in coarser grains acts as an additional attractive force, limiting hydration potential and reducing water adsorption. In contrast, finer materials showed higher hydration and swelling pressure due to increased surface area. While compressibility behaviour remained similar across samples at high stress, coarse-grained samples exhibited overconsolidated behaviour at low stresses, attributable to the structural integrity of larger grains.

Interfacial shear characteristics of geocell-reinforced backfill are critical to stability in reinforced retaining wall structures. The influence of size-related factors (i.e. geocell equivalent diameter d, height h and aspect ratio h/d) on the shear behavior of geocell-sand interface was explored using a three-dimensional discrete element method (DEM). DEM models of geocell-sand interfaces with different specifications were validated against experimental data. Results show that strains at welded junctions of geocells are generally greater than those at the strip sections, and larger equivalent diameters of geocells correspond to lower strains. Within the range of variables in this study, changes in geocell diameters and heights influence the magnitude of mean contact forces by 5.33% and 3.14%, respectively, while their impacts on the principal direction of contact force are measured at 4% and 2.7%. Additionally, aspect ratio (h/d) significantly affects the location of the maximum normal contact force between geocell and sand. For the geocells with d ≥ 170 mm, their sidewalls exert a relatively weak confining effect on the internal soil particles within a range of 1/3d from the cell center. Finally, the geocell-sand interface shear strength was predicted by two machine learning models.

Biomineralization has been used for the treatment of calcareous sand to improve its properties. Although many studies have been performed on the biomineralized calcareous sand under freshwater conditions, few studies were focused on the behaviors of calcareous sand in seawater. As the freshwater is scarce in island areas, the freshwater-based biomineralization technology may be unsuitable for the treatment. In this study, seawater-based bacterial enzyme induced carbonate precipitation (BEICP) was proposed to treat calcareous sand. A series of tests were conducted to verify the feasibility and efficiency of this treatment method through investigating the effects of seawater on the biomineralization and the properties of biomineralized calcareous sand in comparison with microbially induced carbonate precipitation (MICP). Test results reveal that seawater leads to the decrease of urease activity of bacterial cells and urease. NaCl, MgCl₂, Na₂SO₄, and CaCl₂ are the main inhibitory components in seawater, of which MgCl₂ and CaCl₂ have a strong influence on the urease activity of bacterial cells and urease, respectively. Compared to MICP treatment, BEICP-treated calcareous sand exhibits higher unconfined compressive strength and better biomineralization effects. The findings of this study can contribute to the application of biomineralization technology in island areas.

Earthen buildings have been used for thousands of years, but they could face threats by fire during the service life. The evolution of thermal properties of compacted earth after high-temperature exposure is of great significance. In this study, thermal property tests were conducted on compacted earth with different non-hydraulic air lime contents after high-temperature exposure up to 900 °C. Predictive models for thermal expansion coefficient, thermal diffusivity coefficient and thermal conductivity of compacted earth with air lime content were established. Scanning electron microscopy (SEM), thermogravimetric-differential scanning calorimetry analysis (TGA–DSC) and mercury intrusion porosimetry (MIP) tests were performed to reveal the material changes, pore parameters and pore structure distribution characteristics. The temperature of 500–600 °C was identified as the threshold for sudden changes in specific heat capacity, thermal conductivity and thermal expansion coefficient. The major weight loss for unstabilized earth was 5.07% at 400–600 °C, while for stabilized earth with 30% air lime, it was 11.97% at 600–800 °C. The addition of air lime optimized the pore structure of compacted earth and reduced the combined percentage of macropores and mesopores by 63% compared to unstabilized earth. Furthermore, air lime significantly reduced the most probable pore diameter, when the temperature increased from 25 to 400 °C.

Electroosmosis is currently limited by high energy consumption and electrode corrosion. In this study, electroosmosis was carried out for gold tailings under intermittent current conditions using electrokinetic geosynthetic (EKG) electrodes. The influences of the on–off cycle and on–off ratio on water discharge, energy consumption, and resistance were investigated. The results show that water discharge increases and then decreases with the on–off cycle. The average total resistance increases and then decreases with the on–off cycle, while it shows the opposite trend with the on–off ratio. Within the first 30 min, the current displays a peaked "Λ" shape that is mainly affected by the anode contact resistance. After 30 min, the current tends to increase and mainly depends on the cathode contact resistance. Within a cycle after 30 min, the current follows an "L" shape. Upon turning on the current, H⁺ is enriched near the anode to rapidly elevate the anode contact resistance and suppress the current. When H⁺ in this region becomes saturated, the anode contact resistance stabilizes, the cathode contact resistance decreases, and the current increases slightly. The optimal on–off cycle was 12.5 min with an on–off ratio of four, at which point the total energy consumption was reduced to 93.9% and the energy consumption ratio to 84.2%. These results facilitate further studies on the action of intermittent current during electroosmosis and reducing the energy consumption in electroosmotic technology.

This paper provides insight into the inner soil behaviour of stiffened caisson penetration in stiff-over-soft clay that is a common soil profile for the application of caissons. Large deformation finite element (LDFE) analyses were conducted to simulate the installation process of stiffened caisson from the seabed surface with jacking process. Soil-flow mechanism and inner soil heaving were analysed. After the validation of LDFE method, an extensive parametric study was performed, investigating the effects of caisson geometry and layered soil strength profile (including the normalized strength of the top soil layer, soil strength ratio, and the thickness of the top layer). The results showed that three soil-flow mechanisms could occur during stiffened caisson installation in stiff-over-soft clays: Mechanism I—surcharge mechanisms; Mechanism II—trapped mechanism; and Mechanism III—plugging mechanism. Based on the three mechanisms identified, a design framework for assessing inner soil heaves was established to provide guidance for caisson foundations.