Using the Ordinal Regression in the Probabilistic Analysis of Seismic Vulnerability for a Model of Concrete Bridges

Abstract

     The increasing interest in seismic fragility curves and in developing them to be more effective and useful requires the use of statistical and probability techniques different from those used in the traditional case, so in this research we used an unfamiliar statistical method in this field, which is the ordinal regression.  We adopted the analytical method to create seismic fragility curves for a class of concrete bridges, where a sufficient number of samples of bridges belonging to this class were analyzed using csibridge v20 software, and damage data were collected in each of them for different seismic intensities, then the damage data was processed based on five different models of ordinal regression then the corresponding fragility curves were plotted.  The research showed that the ordinal regression with its five models used in this research is well suited to constructing seismic fragility curves for the studied class of bridges with some variation in the quality and fitness of the five models, but they are all acceptable.