Influence of soil parameter spatial variability on railway loess-filled embankment settlement

This study involved an embankment settlement analysis of a case study of the railway loess-filled embankment section of the Baozhong railway in Guyuan, China. The two-dimensional spatial variability of the soil parameters in the embankment fill layer was simulated using random field theory and the finite difference method. The covariance matrix decomposition method was used to discretize the soil parameters within different geological cells. Settlement predictions were evaluated through Monte Carlo simulations with varying horizontal scale of fluctuations (SOFs), and the embankment surface settlement values at the centerline were compared to the deterministic calculation model (DCM), random variable model (RVM), and random field model (RFM). The results indicated that the horizontal SOFs ranging from 40–60 m in the loess-filled embankment can effectively capture the spatial distribution pattern of the soil parameters. Spatial variability exerts a substantial influence on embankment settlements. Neglecting the spatial variability of soil properties underestimates the risk of surface settlement in embankments compared to the DCM. Therefore, RFM analysis that considers the spatial variability of soil parameters must be incorporated to ensure safe operation and design in subgrade engineering.