Contribuição à confiabilidade de estacas tubulares metálicas na Baixada Fluminense com aplicação da teoria bayesiana

Abstract

In foundation design, the geotechnical engineer faces numerous uncertainties, mainly those due to the spatial variability of the geotechnical parameters, the limitation of the bearing capacity estimation models and those attributed to the reduced geotechnical investigation. With the main objective of contributing to the reliability of steel tubular piles, this research intends to analyze a set of driven steel tubular piles executed in the Baixada Fluminense, using a probabilistic approach. Bayesian theory was adopted, with the uncertainties involved in bearing capacity estimation modeled by a priori and a posteriori distribution. The posterior distribution of the bearing capacity is obtained from the updating of the a priori bearing capacity by means of the likelihood functions. For a priori distribution, two models load capacity estimation were used, one theoretical, named as β by Fellenius (2016), and another semi-empirical one, Aoki-Velloso (1975). In relation to the likelihood functions, the results of the dynamic formulas of Sorensen and Hansen (1957), known as the Danish formula, and also that known by Aoki (1986) were adopted. The results of the posterior distribution presented less uncertainty when the Danish formula was considered for likelihood function. An estimation of set-up was also made, both on the basis of the interpretation of dynamic tests and also with interpretation through static tests.