Uplift capacity analysis of inclined strip anchors considering spatial variability of undrained shear strength: RAFELA and ANN

Abstract

This study investigates the uplift resistance of inclined strip anchors embedded in clays where the undrained shear strength varies spatially. Four input parameters are considered, namely the inclination angle (α), the cover-depth ratio (H/B), the coefficient of variation (COV), and the dimensionless correlation length (Θ) of the undrained shear strength. The random field is modeled using the Random Adaptive Finite Element Limit Analysis (RAFELA) and Monte Carlo simulation in OPTUM G2 under plane strain conditions. The results indicate that COV and Θ significantly affect the shape of probability density function (PDF) and cumulative distribution function (CDF) charts. The probability of failure (PoF) depends evidently on COV and Θ, while H/B and α have lower impacts. The correlation between four inputs and the mean of stability factor (μNc) is depicted through parametric studies. Additionally, Artificial Neural Network (ANN) is implemented to propose a regression model to predict the mean and standard deviation of the stability factor (μNc and σNc). Based on the optimal ANN structure, Permutation Feature Importance (PFI) is applied for the sensitivity analysis, showing that H/B is the most important feature, followed by COV, Θ, and α. The results from this study significantly contribute to the research on the pull-out behavior of strip anchors, particularly in clays with spatial variations in undrained shear strength.