Numerical analysis of ground surface settlement induced by shield tunnelling considering the soil parameter spatial variability in loess strata

Abstract

Ground surface settlement (GSS) induced by shield tunnelling in loess soils represents a substantial challenge for geotechnical and structural engineers. Despite numerous attempts to predict GSS caused by tunnelling, considering the spatial variability of the soil parameters remains open. Thus, this study analysed the Lanzhou Metro Line 1 in China. In particular, the spatial variability of Young's modulus in the loess layer was simulated using random field theory coupled with finite difference analysis. Random fields were generated using the covariance matrix decomposition method, which is consistent with a specified numerical mesh. Subsequently, Monte Carlo simulations were performed using both deterministic and random field difference methods to explore the effect of horizontal scale fluctuations due to shield tunnelling on the ground motion. The results indicated that Young's modulus of loess exhibited a normal distribution. The vertical scale of fluctuation of Young's modulus in the loess layer spanned from 1.32 m to 1.51 m. These findings demonstrate that random field analysis can comprehensively characterise the trend, magnitude and fluctuation range of the GSS. Moreover, the findings indicate that GSS is underestimated when neglecting spatial variability compared with traditional deterministic analysis.