Random Field Modelling of Soil Shear Strength for Reliability Analysis of Deep Excavation in Sand

Random field model of spatial variability data plays a vital role in the reliability analysis of geotechnical problems. In this study, random fields of soil shear strength were defined using the Cholesky decomposition technique and Monte Carlo simulation. The correlation coefficient between cohesion and effective friction angle was assumed to simulate independent and dependent structures of soil shear strength. The reliability analysis of deep excavation in sand utilising the random finite element method was conducted. The instability of deep excavation is identified when the predicted lateral wall movements are higher than the limited criteria of lateral wall movements. In this study, 1,000 realisations were used in the Monte Carlo simulation. The results indicated that different correlation coefficients cause significant cohesion and effective friction angle variations. Failure probability is defined by the number of instability events divided by the number of realisations. The lateral wall movements increased with an increase in excavation depth. The dependent structure type of random field soil shear strength can underestimate the failure probability of deep excavation in the sand. This study highlights that the random field soil shear strength is vital in the reliability analysis of deep excavation.