Bayesian model and parameter calibration for braced excavations in soft clays

Abstract

This article presents a Bayesian approach for simultaneously calibrating models and soil parameters using field data on wall deflections and ground surface settlement from braced excavations. This approach is developed based on the Metropolis–Hastings algorithm, the Gibbs algorithm, and hybrid Markov chain Monte Carlo simulation. The Formosa excavation case history is adopted to demonstrate this approach. The predictions are made by using both calibrated models and soil parameters, and the predicted excavation-induced responses from the models are shown to have greater accuracy and less variability than the responses obtained when updating the soil parameters only. This Bayesian approach is effective given the availability of reasonably estimated prior distributions and field observation data with good quality.