Engineering Geology最新文献

Loess, a Quaternary wind-blown deposit, is a problem soil that gives rise to frequent geohazards such as landslides and water-induced subsidence. The behaviour of loess is controlled by its microstructure, consisting of silt-sized skeleton particles and complex bonding structures formed by clay-sized particles. Achieving a deep understanding and precise modelling of loess behaviour necessitates comprehensive knowledge of the realistic 3D microstructure. In this paper, a correlative investigation of the 3D loess microstructure is performed using X-ray micro-computed tomography (μXCT) and focused ion beam scanning electron microscope (FIB-SEM). Details of clay structures in loess, such as clay coatings, clay bridges and clay buttresses, are visualized and characterized in 3D based on FIB-SEM images with a voxel size of 10 × 10 × 10 nm³. The clay structures exhibit a diverse degree of complexity and their impact on the mechanical properties of loess is highlighted. Statistical analysis of the skeleton particles, including size, shape and orientation, are derived from μXCT images with a voxel size of 0.7 × 0.7 × 0.7 μm³. The findings provide insights into the collapse mechanism and particle-scale modelling of loess. The combination of μXCT and FIB-SEM proves to be a powerful approach for characterizing the intricate micro-structures of loess, as well as other geomaterials.

As a typical mechanism of internal erosion, suffusion has led to geological disasters in engineering structures worldwide. A slight deviation in soil structures, also known as the spatial randomness of soil parameters, determines the significant differences in this erosion process. However, owing to the lack of absolute quantitative prediction models for suffusion, this issue has not been effectively evaluated. This paper introduces initial random fields of soil properties into a hydromechanical model to quantitatively predict the possibility of suffusion, considering the random deviations in soil gradation, porosity, and permeability. Through the prediction of 50 sets of random fields, certain trends and uncertain deviations of suffusion are discovered. This certainty and uncertainty constitute the possible range of suffusion, which surrounds the prediction of the homogeneous model and will be temporally widened to larger deviations, indicating the unpredictability of the later stage of suffusion. Statistical analysis revealed that soils with more compacted porosity, more movable particles and less permeability at the seepage outlet are prone to suffusion, and this advantage gradually increases to form the upper envelope of the possible range. This phenomenon is attributed to the larger additional forces acting on the movable particles and the abundant movable particles. The hydromechanics-based model of random soil structures can theoretically estimate the possible development of suffusion and effectively assess the uncertainty of internal erosion risk in hydraulic engineering.

Construction of large underground infrastructure facilities routinely leads to leakage of groundwater and reduction of pore water pressures, causing time-dependent deformation of overburden soft soil. Coupled hydro-geomechanical numerical models can provide estimates of subsidence, caused by the complex time-dependent processes of creep and consolidation, thereby increasing our understanding of when and where deformations will arise and at what magnitude. However, such hydro-mechanical models are computationally expensive and generally not feasible at larger scales, where decisions are made on design and mitigation. Therefore, a computationally efficient Machine Learning-based metamodel is implemented, which emulates 2D finite element scenario-based simulations of ground deformations with the advanced Creep-SCLAY-1S-model. The metamodel employs decision tree-based ensemble learners random forest (RF) and extreme gradient boosting (XGB), with spatially explicit hydrostratigraphic data as features. In a case study in Central Gothenburg, Sweden, the metamodel shows high predictive skill (Pearson's r of 0.9–0.98) on 25 % of unseen data and good agreement with the numerical model on unseen cross-sections. Through interpretable Machine Learning, Shapley analysis provides insights into the workings of the metamodel, which alignes with process understanding. The approach provides a novel tool for efficient, scenario-based decision support on large scales based on an advanced soil model emulated by a physically plausible metamodel.

Wildfires striking vegetated hillslopes appear to increase the hazard towards rainfall-induced landslides. One mechanism little investigated in the literature consists in the formation of Wooden Embers Cover (WEC) following the wildfire. This layer has very peculiar thermohydraulic properties and may affect the interaction between the atmosphere and the subsoil. The paper presents an experiment conducted in an outdoor lysimeter filled with pyroclastic silt (SILT) up to 75 cm covered with 5 cm of WEC. Water storage in the SILT layer, soil water content, suction, and temperature were recorded for several years, initially under bare (no-WEC) condition (4 years), then vegetated (no-WEC) condition (5 years) and, finally, with a WEC placed on the top of the SILT (SILT+WEC condition; 3 years). The hydrological effect of the WEC was assessed by comparing the response of the SILT+WEC with the SILT under bare or vegetated conditions. The WEC reduces water losses by evaporation, thus increasing the average water content in the underlying SILT, an effect that is detrimental to slope stability. To discriminate whether the barrier effect was associated with the lower thermal or hydraulic conductivity of the WEC, a numerical simulation was carried out by considering the case of a WEC with its real thermal and hydraulic properties and the case of a fictitious top layer placed on the top of the SILT having the same hydraulic properties of the WEC but the thermal properties of the SILT. It is concluded that the barrier effect of the WEC is mainly associated with its hydraulic properties, i.e. the WEC acts as a capillary barrier. To demonstrate the practical implications of this findings, a case study of rainfall-induced landslide has been reanalysed by simulating the presence of a WEC layer having the same thermohydraulic properties as the material characterised in this study. It is shown that a WEC can substantially reduce the severity of the triggering rainfall event, thus increasing the vulnerability of the slope to rainfall-induced failure.

The computation of boundary frictional interaction between debris-flow and rough channel beds is crucial for simulating debris-flow dynamic behavior, owing to its impact on the resulting flow velocity and deposition area. Until now, some boundary treatment methods have been proposed in the Smoothed-Particle-Hydrodynamics (SPH) method, such as the conventional Dynamic-Boundary-Conditions (DBC) and Boundary-Critical-Layer (BCL) methods, which are limited in the effective consideration of boundary friction over complex topography. In this paper, instead of the fixed and predefined boundary critical layers in conventional methods, a concept of particlized frictional influence domain is defined, and a novel centroid aggregation-based boundary detection algorithm (CA-BD) embedded in the 3D-SPH framework is proposed. The algorithm captures the diverse interaction forms and computes mutual penetration between debris-flow particles and rough boundary particles, so that the frictional forces exerting on the debris-flow particles can be determined. Additionally, to enhance the computational efficiency, a CPU-OpenMP parallel acceleration framework is implemented. To validate the proposed model, a well-documented dam-break flow experiment and a debris-flow flume experiment are simulated, wherein the proposed model better reproduces the flow behavior compared to the DBC and BCL methods as observed in the experiments. Comparison on the computational efficiency indicates that the proposed model attains a 2.9 times acceleration factor than the CPU serial solution. Sensitivity analysis also reveals that the predefined length of the frictional influence domain $l_f$ has a significant influence and the value equating to the particle smoothing length $h$ is suggested.

Landslides pose a severe risk to humans, but accurately quantifying human risk remains challenging due to the less-studied fleeing process of humans during landslides. This study introduces a flight failure rate to represent the capacity of humans to escape from a landslide. A novel probabilistic framework for flight failure rate assessment is proposed by integrating uncertainties in both landslide runout and human flight. This framework distinguishes the individual flight failure rates at different locations and the total flight failure rate of the population in a landslide-threatened area. To aid in applying this framework in real-world communities, a network-based human flight model, embedded with the Ant Colony Optimization algorithm, is developed to simulate the heterogeneous human flight behaviors subjected to landslides. A catastrophic landslide in a community of Shenzhen, China, which caused 77 deaths, 17 injuries, and 900 homeless, serves as a case study to perform human flight simulation and flight failure rate assessment. Results indicate that the approach provides reliable and logical evaluations of individual and total flight failure rates. Individual flight failure rate varies significantly in spatial distribution due to differences in landslide available time and running distances to escape the landslide, which differs from the total flight failure rate of the population. Advancing and narrowing the distribution of response time, reducing the delayed time, and implementing pre-planned flight paths can significantly reduce the total flight failure rate and mitigate high-risk areas. This probabilistic framework provides a promising and valuable reference for landslide risk assessment and human disaster mitigation.

Erosion and entrainment significantly increase the volume and destructive potential of high-speed long-runout landslides. Previous studies seldom quantitatively address these effects, and even fewer incorporate the extent of slope weathering into the analysis of landslide dynamics. This study addressed this gap by developing a framework for dynamic analysis, combining Finite Element Method-Smoothed Particle Hydrodynamics-Finite Discrete Element Method (FEM-SPH-FDEM), and applying it to the Shuicheng landslide. Simulation results closely matched field data, revealing substantial sliding mass deviation and velocity variations influenced by rocky ridges and valleys. According to the simulation, the weathering degree of rock slope significantly affects landslide dynamic processes. The interparticle friction coefficient is crucial for accurately modeling these processes using the SPH-FDEM method. Additionally, by incorporating landslide erosion behavior into the framework, the case study indicates that the volume of landslides in Shuicheng County increased by approximately 0.6 times. Three stages of evolution mechanisms of high-altitude landslide-induced erosion behavior are proposed in this paper, highlighting the effectiveness of this framework in understanding landslide mechanisms and providing information for prevention strategies in high-altitude, highly weathered areas.