Seismic fragility assessment framework for highway bridges based on an improved uniform design-response surface model methodology

This paper proposes an improved uniform design (UD)-response surface model (RSM) method to reduce the impacts of sample correlation among small sample data caused by UD method and variable substitution process in the traditional UD-RSM method. The proposed B-spline-PLS-UD-RSM method, which is a systematic approach that combines nonlinear B-spline functions, the conventional UD-RSM method and the partial least-squares (PLS) regression technique, assesses the reliability of structures with high nonlinearity and multi-dimensionality. First, two reliability examples from the literature are investigated to demonstrate the efficiency and accuracy of the proposed RSM method. Subsequently, to investigate the applicability of the proposed method in seismic fragility analyses considering various parameter uncertainties and variable correlation, an alternative seismic fragility assessment framework is developed for highway bridges. Furthermore, this paper uses a typical multi-span reinforced concrete continuous girder (MSRCCG) bridge as a case study and performs a comparative analysis of the corresponding seismic fragility evaluations at both bridge component and system levels by using traditional and improved RSM methods. The results indicate the following conclusions: (1) The proposed B-spline-PLS-UD-RSM method can properly reduce the effect of sample correlation and achieve satisfactory reliability results for structures with highly nonlinear and multi-dimensional features through a much smaller sample size; (2) The proposed seismic fragility assessment framework is a good candidate for evaluating the seismic vulnerability of highway bridges incorporating variable correlation of structural parameters; (3) Seismic fragilities of bridge structures tend to be overestimated by using traditional UD-RSM methods and by ignoring the variable correlation of structural random parameters.