A Modern Bayesian Approach to Model Updating of Bridges Considering Measurement Uncertainty

Abstract

Existing bridges were designed using contemporaneous standards, but increased loading due to increased freight demand must be considered in their ongoing safety management. Finite-Element (FE) modelling provides the structural responses under known loads. However, such models typically include many assumptions about structural behaviour. On the other hand, structural health monitoring (SHM) provides valuable data about the actual structural behaviour and potentially the structural condition. Typically, there are significant differences between the FE model prediction and the SHM measurements. Due to the significant prior engineering knowledge about structural behaviour and performance, an effective way to combine new observations with existing models is using Bayesian updating strategies. This paper reviews the current state of the art of Bayesian updating, which has undergone tremendous developments in the last decade or so. We discuss how engineers can benefit from these developments, specifically for updating structural models based on data through the fusion of prior engineering information and structural health monitoring data. The modern Bayesian workflow is applied to a simple case study of a moving load on a beam. The work illustrates the potential benefits of the approach for updating the performance prediction of structures based on data.