Bayesian Update of Load Capacity for a Large Steel Piling in a Stratified Soil Profile

Abstract

This paper applies Bayesian updating of the load capacity of a large steel piling foundation for the high load structure of the Alcântara Wastewater Treatment Plant (WWTP), located near the city of Rio de Janeiro in Brazil. Uncertainty is modeled by a priori and a posteriori distributions of the piling capacity. The a posteriori distribution is determined by updating the a priori distribution using a likelihood function, which incorporates records obtained during pile driving. The Bayesian update was applied to a dataset consisting of 645 steel driven piles. Two pile capacity design models and two different likelihood functions were used to verify their influence on the updated capacity estimates. Static and dynamic test results were compared to the updated estimates. The results demonstrate the ability of the Bayesian update technique to significantly improve the reliability of the entire piling.