Geotechnics

This study explores the effect of cation adsorption on the shear strength and mineralogical characteristics of smectite-rich landslide clay collected from the Nishinotani landslide in Ehime Prefecture, Japan. Laboratory experiments were conducted using aqueous solutions of calcium, magnesium, and potassium chlorides at concentrations of 1000, 6000, and 12,000 mg/L. Ion chromatography, X-ray diffraction (XRD), and ring shear tests were conducted to evaluate the interaction between ion uptake and its influence on the change in shear strength. The results showed that calcium and potassium ion adsorption increased with both concentration and time, leading to enhanced residual shear strength and crystallinity, primarily due to stronger Coulombic interactions and favorable ionic size compatibility with smectite. Conversely, magnesium ions exhibited adverse effects, including reduced strength and mineral ordering, attributed to calcium leaching and weaker interparticle bonding. The findings indicate that selective cation exchange can be an effective, sustainable alternative to conventional landslide stabilization methods, especially in fine-grained, expansive clay systems. This work contributes to the development of geochemically engineered landslide mitigation strategies based on microstructural and mineralogical reinforcement.

This study evaluates the swelling potential in clayey soils of the Paleogene Brebi, Mera, and Moigrad formations in the Transylvanian Basin (Romania) by integrating direct free-swelling tests (FS; STAS 1913/12-88) with indirect index-property diagrams and semi-quantitative X-ray diffraction (XRD; RIR method). The indirect analysis combines three swelling-susceptibility classification charts—Seed et al. (AI–clay), Van der Merwe (PI–clay), and Dakshanamurthy and Raman (LL–PI)—with mineralogical trends from the Casagrande plasticity chart, complemented by Holtz and Kovacs’s clay-mineral reference fields and Skempton’s activity concept (AI = PI/% < 2 µm). The geotechnical dataset comprises 88 Brebi, 46 Mera, and 263 Moigrad specimens (with parameter counts varying by test), an XRD was performed on a representative subset. The free swell (FS) results indicate that Brebi soils range from low to active behavior (50–135%) without reaching the very active class; most Brebi specimens fall in the medium-activity range. Moigrad spans the full FS spectrum (20–190%) but is predominantly in the medium-to-active range. In contrast, Mera soils exhibit predominantly active behavior, covering the full range of activity classes (30–170%). The empirical classification charts diverge systematically: clay-sensitive schemes tend to assign higher swell susceptibility than the LL–PI approach, especially in carbonate-influenced soils. XRD results corroborate these patterns: Brebi is calcite-rich (mean ≈ 53.5 wt% CaCO₃) with minor expandable minerals (mean ≈ 3.1 wt%); Mera is feldspathic (orthoclase mean ≈ 55.3 wt%) with variable expandable phases; and Moigrad has a higher clay-mineral content (mean ≈ 38.8 wt%). Overall, swelling is controlled by the combined effects of clay-fraction reactivity, clay volume continuity, and carbonate-related microstructural constraints.

Factor of Safety (FOS) is a significant index to measure the stability condition of a rock slope in mining or civil engineering. In this paper, we evaluate and compare four different machine learning models, Gaussian Process Regressor (GPR), Support Vector Regressor (SVR), Random Forest (RF), and a hybrid genetic algorithm–multi-layer perceptron (GA-MLP), using two separate real-world datasets. The two separate datasets used in this study are from a previously conducted study on highway excavation with rock cutting in China, and another one in a mining site in Peru, with five geotechnical properties used as inputs, including slope height, slope angle, unit weight, cohesion, and friction angle. The two separate datasets were separated into training, validation, and testing datasets. The testing dataset of the models is unseen data used to assess model performance in an unbiased manner. The result shows that the SVR had the highest prediction accuracy, followed by GPR for the mining dataset, and GPR had the highest performance among all the models for the highway excavation dataset. From the boxplot, we can see that SVR, while having the highest predictive accuracy, has a larger variance in prediction compared to GPR for the mining dataset.