Abstract
Robustness evaluation of bridges within a risk-based framework requires estimation of the probability of occurrence of different hazards followed by an assessment of the vulnerability of the bridge with respect to those hazards, as well as quantification of the consequences of potential failure. The first part of the paper deals with a statistical analysis of past metallic bridge failures which can help in identifying the principal hazards affecting bridges and their associated vulnerability. The results show that natural hazards, design errors and limited knowledge are amongst the most commonly encountered causes of collapse in metallic bridges, followed by accidents and human error aspects other than in design. When analysed chronologically, the data demonstrates a decreasing trend for collapses attributed to limited knowledge and an increasing trend in failures resulting from accidents and natural hazards. The paper continues by presenting a categorisation procedure through which consequences arising from potential bridge failures can be estimated. Associated models for quantifying their magnitude considering both spatial and temporal domains are highlighted. Finally, the predictive capability of the models is outlined through a case study.