Effect of uncertainties in prediction models of shallow and deep foundations

Abstract

The primary objective in foundation design is to ensure that the foundation exhibits satisfactory behavior under regulatory loading conditions while addressing critical safety concerns. This study aims to investigate the impact of parameter uncertainties in prediction models for bearing capacity and settlement of shallow and deep foundations on resulting failure probabilities using various approximation methods. Recognizing the complexity of reliability studies, this research explores the possibility of not probabilizing all variables and simplifying calculations for future applications. Additionally, it aims to assess the effectiveness of commonly used approximation methods in geotechnical engineering for estimating failure probabilities. The study's findings would enhance foundation reliability indices and decrease the probability of failure. The parametric study has enabled us to identify the variables that are most sensitive to changes in their coefficient of variation. The influence of model selection on foundation reliability is emphasized, and a critical evaluation of model bias is crucial for reliability analysis. It is evident that conservative models may shift subsequently calculated failure probabilities by inaccurately assigning situations from the safe domain to the failure domain due to inherent conservatism.

Keywords

• reliability
• failure probability
• sensitivity
• foundation
• settlement
• bearing capacity