Component reliability and sensitivity analysis of anchor-pile-wall reinforcement design of a multilayer soil slope

Abstract

Component reliability and sensitivity analysis of soil-slope combined-reinforcement deterministic designs are studied using a hybrid numerical-probabilistic method. Major components are qualitatively identified with numerical simulations to avoid redundancy evaluations. Quadratic polynomials as response surfaces are stepwise regressed to predict component load effects. Major component resistances of supporting structures are calculated in states of soil ultimate equilibrium and maximum allowable deformations obtained according to building and foundation types. Component performance functions (PFs) are established based on component failure models and soil-mass deformation properties. Reinforced slope stability reliabilities are calculated with first-order second-moment and design point methods. The reinforcement probability safety indicated by component reliability indices is adopted to check whether the deterministic design needs improvement. This methodology is applied to a loess slope reinforced by combined supporting. The resultant reliability indices of the slope imply some improving treatments should be made. Sensitivity analysis shows some variable variation has a more significant impact on the PFs; inversely signed sensitivity coefficients indicate that the equilibrium point between PFs should be considered when modifying parameters. Stable monitoring results of modified design implementation indicate that probabilistic analysis of component stability is necessary for some slope-treatment designs with distinct probability distribution variables.