Bayesian Updating of Soil Parameters for Braced Excavations Using Field Observations

Abstract

A Bayesian framework using field observations for back-analysis and updating of soil parameters in a multistage braced excavation is presented. Because of the uncertainties originating from the poorly known soil parameters, the imperfect analysis model, and other factors such as construction variability, the soil parameters can only be inferred as probability distributions. In this paper, these posterior distributions are derived using the Markov chain Monte Carlo (MCMC) sampling method implemented with the Metropolis-Hastings algorithm. In the proposed framework, Bayesian updating is first realized with one type of response observation (maximum wall deflection or maximum settlement), and then this Bayesian framework is extended to allow for simultaneous use of two types of response observations in the updating. The proposed framework is illustrated with a quality excavation case and shown effective regardless of the prior knowledge of soil parameters and type of response observations adopted.