Reliability Assessment of Structures With Bayesian Model Updating Accelerated Via Polynomial-Chaos Metamodeling

Abstract

Finite element models are often preferred in numerical modeling of structures, but model assumptions lead to inaccuracies and uncertainties. Measuring these is necessary to determine the reliability and accuracy of the numerical model. This has led to the development of FE model update methods that aim to calibrate the numerical model based on data obtained by structural health monitoring (SHM). However, a general framework that provides a realistic life cycle performance assessment of structures by efficiently incorporating monitored data into structural identification has not yet been impeccably presented. Bayesian modeling can characterize uncertain structural parameters as random variables and provide a systematic methodology for integrating a probabilistic SHM framework into model updating. Unfortunately, these lead to complex and time-consuming, causing limitations in their application. Metamodeling techniques which are effective stochastic predictors can be used to decrease the computational burden in the model updating. This study aims at adapting Polynomial-Chaos-Kriging metamodeling technique integrate to Bayesian model updating process to overcome the computational difficulties and reduce different source of uncertainty with using SHM, then, make more accurate reliability assessment. Therefore, an experimental study is used to assess reliability of structure that is exposed to different types of corrosion effects.