Statistical Analysis of Coastal Bridge Vulnerability
NOTE: This report is a paper published in Structure and Infrastructure Engineering, an online journal. Print publication in the journal is pending.
Jamie E. Padgetta*, April Spillera, and Candase Arnolda
aDepartment of Civil and Environmental Engineering, Rice University, Houston, Texas, USA
This paper analyzes bridge damage along the US Gulf Coast from the 2005 Hurricane Katrina. Most bridges were low-lying water crossings having multiple-span concrete superstructures with limited connectivity to the substructure. The paper emphasizes statistical analysis of hazard-damage relationship for the Katrina exposed region. Multivariate logistic regression is conducted to identify which hazard intensities and bridge characteristics were strong predictors of damage level, considering such parameters as surge elevation, water speed, wind speed, relative surge, number of spans, and age of bridge. The results indicate that storm surge, as well as number of spans, correlates to increasing level of damage, and subsequent analyses evaluate this hazard intensity at the damaged and undamaged bridge locations. Empirical damage probability and damage state exceedance matrices are developed. Empirical fragility curves, estimating likelihood of being in or exceeding a damage state, are presented as a first step toward risk-based analysis of coastal bridges.
Keywords: bridge, coastal, storm surge, probabilistic analysis, fragility