Fragility assessment of RC buildings in southern spain based on neural network predictions

Abstract

The computational burden needed to perform a fragility analysis of structures can be excessive and beyond the capability of regular computing systems. In this work, a Neural Network (NN) implementation is presented to make fragility analyses attainable. Neural Networks allow finding solutions to complex problems at a fraction of the computational time required by conventional analyses. The fragility assessment has been developed for low- and mid-rise 3D buildings located in southern Spain, a moderate earthquake prone area. Nonlinear static analyses are carried out to determine the capacity curves of reinforced concrete buildings, avoiding their specific modelling. The curves are predicted with minimal error, requiring only basic geometric and material parameters of the structures to be specified. Four levels of performance-based seismic design have been considered to assess the seismic performance. Fragility curves have been developed for the structural models with different types of structural configurations and heights. Finally, it should be noted that fragility curves have not been obtained to date for the reinforced concrete buildings of the area.