Development of Interpretable Regression Models for Slope Stability of Nano-Silica Stabilised Soils in Mountainous Terrain

Mountainous regions, particularly the Lesser Himalayas, face persistent slope stability challenges due to complex geological formations and extreme climatic variability. This study investigates the potential of Nano-silica (NS) as an eco-efficient soil stabilizer for enhancing the mechanical behaviour of fine-grained soils in these terrains. Consolidated undrained (CU) triaxial tests were performed on clay of intermediate plasticity (CI), silt of intermediate plasticity (MI), and low-plasticity clay-silt (CL-ML) soil types treated with different NS contents and subjected to various curing durations. To develop reliable predictive insights, both linear and non-linear regression models were constructed using essential geotechnical parameters, including cohesion, internal friction angle, and the pore-water pressure ratio. A novel model simplification process was employed to derive explicit closed-form equations for the Factor of Safety (FoS), retaining only the most influential terms. The non-linear models demonstrated high predictive accuracy, with R2 values exceeding 0.97, outperforming their linear models for all the soil types stabilized with NS. These interpretable regression models offer a practical tool for slope stability assessments in NS-stabilized soils. The integration of material innovation with simplified computational models contributes to resilient and sustainable infrastructure design in data-scarce highland regions.