Estimating Prestressed Concrete Bridge Reliability and Rating Factors Using Bayesian Networks with an Application to a Bridge Made Continuous for Live Load

Abstract

The bridge inspection process has multiple steps. One obvious element is for inspectors to identify defects in the main components of the structural system and assign condition ratings. These condition ratings are somewhat subjective because they are influenced by the experience of the inspector. In the current work, processes were developed for making inferences on the reliability of prestressed concrete (PC) girders with defects at the girder component level. The Bayesian network (BN) tools constructed in this study use simple structural mechanics to model the capacity of girders. Expert opinion is used to link defects that can be observed during inspections to underlying deterioration mechanisms. By linking these deterioration mechanisms with changes in mechanical properties, inferences on the reliability of a bridge can be made based on visual observation of defects. The BN can then be used to directly determine the rating factor (RF) of individual structural elements. Examples are provided using BNs to evaluate an existing older PC bridge currently behaving as two simply supported spans. The bridge is modeled using two scenarios with the spans acting as simply supported, and then also with the link block (continuity joint) repaired so that the spans are continuous for live load. The spans are considered simply supported for all dead load.