Scaled boundary finite element model-based Bayesian updating for subseabed shield tunnels utilizing distributed strain data

Abstract

The structural state of subseabed shield tunnels presents significant uncertainties throughout their service life. Bayesian model updating provides a more accurate parameter estimation by effectively accounting for these uncertainties. However, the Markov Chain Monte Carlo (MCMC) simulation faces challenges of computational intensity and convergence issues. To address these challenges, a novel methodology based on the scaled boundary finite element model (SBFEM) is proposed to efficiently update the structural parameters of subseabed shield tunnels within the framework of Bayesian inference. The circumferential strains of the tunnel are obtained using distributed fiber optic sensor as measurements. The SBFEM allows for efficient simulation of the structural responses and is incorporated into MCMC simulations to derive the posterior distributions of model parameters. Validation results indicate that the computational time for the SBFEM is only 2.7 s, reflecting an efficiency improvement of 90.7% compared to conventional finite element models. The average discrepancies between the updated and actual values of the external loads, tunnel, and stratum elastic modulus are 3.8%, 2.1%, and 3.5%, respectively. This approach ensures analytical accuracy while addressing the high computational expenses associated with MCMC sampling. These findings contribute to a more precise and efficient assessment for the structural states of underground tunnels.