Seismic behavior & risk assessment of an existing bridge considering soil-structure interaction using artificial neural networks

Abstract

The Peak Ground Acceleration (PGA) is extensively used in earthquake engineering practice to describe the ground motion characteristics for establishing the seismic vulnerability curves. However, a single parameter is not enough to describe the seismic excitation and does not allow expressing the complex relationship between the structural damage and the ground movement. Motivated to overcome these shortcomings, several ANN-based models were proposed to predict the seismic structural damage using other parameters. Unfortunately, not all include soil structure interaction. This paper aims to explore the predictive power of an ANN-based approach to reproduce the nonlinear dynamic behavior taking into account the various ground motion intensities, the variability of soil, and SSI. The basic strategy is to train a neural network by a numerical database obtained from a FEM model. This numerical model is further validated by experimental results. An optimum prediction for a nonlinear dynamic response is achieved using Artificial Neural Networks. Finally, fragility curves were established considering SSI for three different soil classes. Results revealed the importance of considering SSI effects on the evaluation of seismic structural damage and risk assessment analysis.