Historical Analysis of Hydraulic Bridge Collapses in the Continental United States

摘要

Predictions of the risk to built infrastructure posed by climate and land-use change have suggested that bridge collapses may increase due to more frequent or intense flooding. Assessments of the United States often assume that bridges may collapse when the 100-year flood (i.e.,a flood with 1% annual frequency of exceedance) occurs, but this assumption has not been fully tested because of a lack of comprehensive collapse records. Thirty-five bridges for which a stream gauge on or near the bridge recorded the flow during total or partial collapse were identified and used to test this assumption. Flood frequency analyses, other statistical analyses, and structural reliability methods were used to quantify the return periods of collapse-inducing flows, identify trends linked to event and site characteristics, and evaluate the potential importance of collapse return period variability in assessing the impact of climate and land-use change on hydraulic collapse risk. The results indicate that the collapse-inducing flow return periods varied considerably (range: 1 to >1,000years) and were frequently lower than values considered in many climate impact assessments: 23 of the 35 bridges were estimated to have collapsed during flows with return periods of lower than 100years. Annual failure probabilities computed using the full distribution of return periods of the collapse-inducing flows, as opposed to central values (e.g.,means), were more sensitive to an assumed increase or decrease in the underlying frequency of flooding. These results suggest that linking bridge collapse to only the 100-year flow does not capture significant variability associated with collapse return periods, potentially reducing sensitivity to flood frequency changes and reducing the robustness of assessments of the impact of climate, land-use, and streamflow-regulation change on hydraulic bridge collapse risk.