Bulletin of Engineering Geology and the Environment最新文献

The mechanical behaviors of rock mass containing discrete fracture networks (DFNs) are always a challenging topic in rock engineering. A methodology is presented that incorporates the generation of synthetic rock mass (SRM) and investigates its mechanical behavior using a GPU-based discrete element method (DEM) software, CoSim-DEM. This approach considers factors such as fracture intensity, variations in domain size, and other relevant parameters. The particle size ratio (PSR), which means the ratio of the particle size to the characteristic scale of the sample, is used to analyze the influence of the element/particle size on DEM numerical results. When the PSR is less than 0.1, it has little influence on the numerical results for both rock and rock mass samples. The numerical tests indicate that both fracture intensity and sample size significantly affect the failure and mechanical behaviors of rock mass samples. Based on a generated SRM, rock mass samples with different sizes are generated and used for numerical uniaxial tests. The results indicate that the rock mass exhibits an obvious scale-dependent characteristic, with the representative elementary volume (REV) being approximately 1.5 times the maximum size of the fractures. This study may provide a novel approach to the study of rock mass mechanics and the development of numerical test methods.

While-drilling identification technology is a crucial part of intelligent mining development. The results provide a scientific basis for real-time adjustment of support parameters, promotes safe and efficient mining. Noise and vibration generated during the rock drilling process vary significantly between different rock materials, offering a new method for identifying rock characteristics. This study used a digital drilling test system to conduct experiments on layered rocks with weak interlayers, monitoring drilling parameters and borehole sound pressure and vibration acceleration in real time. Time-domain and frequency-domain features of borehole sound pressure and vibration signals were analyzed to identify rock characteristics in weak interlayer areas. The results show that these methods can clearly identify the location of weak interlayers. The spectral characteristics of different rock materials are significantly different, indicating a strong correlation between borehole sound and vibration characteristics and the mechanical properties of the rock. Comparing sound and vibration identification results with drilling parameters at weak interlayers verified the accuracy of the identification, providing a basis for lithology identification using multiple parameters. This study shows the importance of drilling noise and vibration parameters in enhancing while-drilling parameters in the digital drilling process.

Spatiotemporal patterns of earth surface deformation are influenced by a combination of static and dynamic environmental characteristics specific to any landscape of interest. Nowadays, these patterns can be captured for larger areas using Interferometric Synthetic-Aperture Radar (InSAR) technologies and yet, their spatial prediction has been poorly investigated so far. Here, we initially compute the InSAR-derived line-of-sight hillslope velocities (V_LOS) and calculate their mean (ranging from 0 to ~ 30 mm/y) and maximum (ranging from 0 to ~ 60 mm/y) values per Slope Units (SUs). These separately constitute the response variables to be modelled through a series of deep learning routines: i) a basic neural network (Multi-Layer Perceptron), ii) a Graph Convolutional Network implemented to capture spatial dependence among neighbouring SUs, and iii) an Edge-Featured Graph Attention Network sensitive not only to the interdependence brought by the SU positions in space but also to reciprocal terrain characteristics. We assessed the model performance for both models via Mean Absolute Error (MAE), r-squared (R²), and Pearson Correlation Coefficient (PCC). The Edge-Featured Graph Attention Network model produced the best performance. The result for the first model targeting the mean V_LOS are 4.75 mm/y, 0.63, and 0.79 for MAE, R², and PCC, respectively. As for the second model, targeting the maximum V_LOS, these are 19.52 mm/y, 0.55 and 0.75. We also showcased interpretable multivariate models, where the contribution of each predictor to the InSAR velocities is summarized and interpreted. This represent a clear example where InSAR-derived hillslope velocities are accurately estimated at regional scales, thus setting up the scene for further advances towards space-time regional deformation modelling. 

Recent advancements in rainfall observation and forecasting have enabled the real-time identification of rainfall temporal patterns, thereby enhancing the prediction of landslides by accounting for their temporal dynamics. However, conventional landslide prediction methodologies rely on average rainfall characteristics, such as intensity (I) and duration (D), overlooking the influence of rainfall temporal details. To overcome this limitation, this study proposes an innovative approach to defining rainfall thresholds that integrate rainfall temporal patterns. These thresholds are designed to provide probabilistic estimations of landslide occurrences for rainfall events characterized by constant intensity and duration, while considering the temporal dynamics. Investigating the impact of rainfall temporal patterns on landslide occurrences reveals that the initiation of landslides is primarily influenced by the infiltrated rainfall rather than the total rainfall amount. Specifically, rainfall temporal patterns characterized by fewer high-intensity values tend to result in greater infiltration, thereby increasing the likelihood of landslide triggering. Furthermore, rainfall concentrated in the early and middle stages of an event is associated with a higher probability of landslide occurrence. This study's findings underscore the importance of incorporating rainfall temporal patterns into landslide early warning systems, thereby facilitating more effective risk mitigation strategies.

A multitude of subterranean engineering projects are established within granite bodies, where the mechanical properties of granite are predominantly influenced by the existing stress environment and inherent microfabric. As the depth of engineering projects increases, the characteristic stress differential becomes more pronounced, indicating a discernible distinction from the stress conditions encountered in shallow engineering applications. To elucidate the influence mechanism of microfabric and stress disparity on granite strength, a quantitative analysis is conducted on the microfabric of five types of granites. The results show that the strength of granite is primarily determined by initial damage, structural coefficient, biotite content, and quartz content. With an increase in stress differential, the impact of initial damage and biotite content on granite strength diminishes, while the influence of quartz content and structural coefficient on granite strength begins to intensify. A subjective and objective comprehensive evaluation model is established to quantify the weight ratio of microfabric on granite strength. The coupling mechanism of stress difference and microfabric on the granite strength is revealed from the rock fracture directional development induced by the increase of stress difference. The results can be used as a guide to understand the granite strength characteristics according to the lithofacies and true three-dimensional stress environment, and provide an effective theoretical basis for the safe construction of deep granite engineering.

In pipeline engineering in expansive soil areas, swelling pressure is frequently determined by laboratory tests as an essential design parameter. During the processes of filling and rolling, lateral residual stress exists in the soil, which significantly influences the swelling pressure. In this study, a series of expansion and residual stress tests is conducted on compacted expansive soil s with different initial dry densities and moisture contents. The variation rules of the swelling pressure for different specimen preparation methods are compared. The effects of the initial conditions on the lateral residual stress are analyzed. The range of initial conditions for the influence of lateral residual stress on swelling pressure is investigated. In addition, based on the Mercury Intrusion Porosimetry (MIP), the change rule of the soil microstructure under different initial conditions is obtained. A microcosmic mechanism for the lateral residual stress acting on the vertical swelling pressure is proposed. The results demonstrate that the lateral residual stress increases with increasing dry density and decreases with increasing moisture content. Based on the variation range of the initial specimen condition, the influence of the lateral residual stress on the swelling pressure can be distinguished using slash (({omega }_{0}text{=}112.5{rho }_{0}-157.75)). When the lateral residual stress exceeds the range of approximately 130–170 kPa, the swelling pressure is influenced, and the greater the lateral residual stress, the greater the effect.

A hybrid discrete-continuum numerical method was employed to study the shallow red-bed soil landslides that frequently occurred in Southwest China. This study used a clay slope with weak interlayer in Guangxi, China, as a geological model to simulate water infiltration driven by differences in water content in particles. The landslide’s kinematic behavior was performed by the discrete element software MatDEM. Results indicate that the spatial variability of the wetting front is caused by the thickness difference of the soil layer. The water content of the weak interlayer particles increases nonlinearly with time, and the bond between particles decays exponentially with the increase of its water content. In red-bed soil slopes, both higher initial water content and higher recharge boundary water content promote landslide initiation. After triggering the landslide, the soil particles at the foot of the slope and in the weak interlayer move first, and the maximum velocity and kinetic energy of the particles show a trend of first increase and then decrease. During the dynamic process, the sliding body may have an extrusion effect on the underlying soil layer, resulting in a new sliding surface. This study evaluates the effectiveness and potential of this discrete-continuum hybrid numerical approach for revealing the disaster-causing mechanism of rainfall-induced landslides in red-bed areas.

This study investigates the effect of excavation-induced phreatic line drawdown on seepage discharge in mountain tunnels situated in horizontal strata. By integrating the Dupuit hypotheses with the Boussinesq equation, theoretical solutions were developed to describe the equilibrium state phreatic line and quantify the corresponding seepage discharge. These solutions were derived using the integral method and were further analyzed to identify potential sources of error. To validate the theoretical framework, numerical simulations were conducted using ABAQUS, revealing strong correlations and offering insights into discrepancies. Key parameters influencing the equilibrium state phreatic line and seepage discharge were systematically examined, including hydraulic conductivity, the horizontal distance from the tunnel’s central axis to the lateral boundary of groundwater recharge, recharge height, the vertical distance from the tunnel’s central axis to an impervious layer, tunnel radius, and the thickness of horizontal strata. The applicability conditions of the theoretical solutions were also established through comparative analysis with numerical results. Despite the notable relative errors inherent in the theoretical solutions, they account for phreatic line drawdown caused by tunnel excavation, offering an analytical tool for preliminary design and seepage evaluation in tunnel engineering.

Landslide susceptibility prediction (LSP) is a complex task with unresolved uncertainties, such as errors in sample classification and intricate relationships among environmental factors and spatial grid units. Additionally, the absence of interpretable black box models restricts the credibility and effectiveness of prediction models. To tackle these problems, an innovative interpretable deep learning model based on self-filtering graph convolutional networks and long short-term memory (SGCN-LSTM) is proposed. In the SGCN-LSTM, a self-screening strategy is employed to remove landslide/non-landslide samples with substantial errors that fall outside a defined threshold interval. Furthermore, SGCN-LSTM extracts nonlinear connections between environmental factors and long-range dependencies among grid units through spatial nodes and information gates. The Anyuan County in south China, with 2,655,972 grid units, 16,594 labeled, served as the study area. The LSP models used numeric inputs from the Frequency Ratios of 10 environmental factors in these spatial grid units. Results show that the accuracy and area AUC of the SGCN-LSTM achieve 92.38% and 0.9782, which are higher than those of one deep learning model cascade-parallel long short-term memory and conditional random fields (by 5.88% and 0.0305), and four machine learning models (by 12.44-20.34% and 0.0532–0.1909). This article delves into SGCN-LSTM ‘s evaluation results using the SHAP method, providing insights into the landslide development patterns and spatial heterogeneity of associated environmental factors in Anyuan County, with a global interpretability perspective. In conclusion, the SGCN-LSTM automatically screens erroneous samples, effectively extracts nonlinear features and spatial relationships from various environmental factors and delivers superior prediction accuracy and robustness for LSP.

Polatlı, which is the largest and tenth out of twenty fifth most populated county of Ankara is well known for being one of the most productive agricultural districts in Türkiye in terms of its barley and wheat production. However, despite that Polatlı has a relatively dense and rapidly growing population, and bears environmental problems, it does not possess a proper municipal solid waste landfill. Since the county currently lacks a proper landfill, the municipal waste is deposited in an improper open dump site that is located to the south of the county. Concerns have been raised due to fire incidents reported and due to scattering of the waste material throughout the neighborhood of the open dump site and to the other parts of the city due to the lack of fencing at the open dump site. Another environmental problem is caused by biogas energy producing companies in the district that dump their processed animal wastes in the farm fields which endangers public health. In addition, extensive illegal waste dumping in the neighborhood of the open dump site exists. The objective of this study is to select the best alternative municipal landfill site location for the Polatlı County, Ankara. To fullfil the disposal needs of the county, landfill site selection has been performed in this study by considering criteria including, air traffic safety, geology, land use, distance to settlement, distance to roads, drainage, slope, erosion, distance to fault and distance to earthquake epicenters. These criteria have been ranked and evaluated in a GIS environment prior to selecting the best alternative landfill site through “The Technique for Order Preference by Similarity to Ideal Solution (TOPSIS)” method of Multi-Criteria Decision Analysis (MCDA). The results of the landfill site selection methodology indicated that amongst the three alternative landfill sites, the best locations to construct a landfill were chosen to be those two alternative sites that were situated north northeast (NNE) and north northwest (NNW) of Polatlı, respectively.