Probabilistic prediction with Bayesian updating for strength degradation of RC bridge beams

Abstract

A probabilistic prediction framework of corrosion-induced strength degradation for flexural beams is proposed in this paper. The proposed framework considers both ductile and brittle failure modes of reinforcements. The area loss of steel bars is established considering the likelihood of corrosion types. Statistical data analysis is used to quantify the uncertainties of capacity variation of corroded reinforcing bars based on the experimental investigation of tensile tests of 452 corroded reinforcements from different members. Following this, the static tests on 48 beams are conducted, and the finite element method (FEM) is used to evaluate the effects of corrosion on carrying capacity. A probabilistic model to include the effect of inaccurate modeling of corrosion on the beam bearing capacity is developed. Area loss and strength degradation of corroded reinforcing bar, possible ductile and brittle failure of reinforcement and model uncertainty are incorporated into analysis of time-dependent strength degradation. Finally, a Bayesian updating methodology is proposed to update the prior belief of the uncertainties and the updated posterior distributions are used for probabilistic prediction using field inspection results. Three beams demolished from a 36-year old concrete bridge are used to demonstrate and to validate the overall procedure. The prediction combined with Bayesian updating provides a satisfactory result by comparing model predictions with realistic field inspection.