Seismic Evaluation of Retrofitted RC Frames Using Neuro-Fuzzy Algorithms

Recent severe ground motions all over the world have indicated that seismic performance of existing Reinforced Concrete (RC) frames which had been designed based on old building codes were unacceptable, since such structures have low resistance to seismic loads which result in large deformations during earthquakes. The seismic retrofit procedure for bending frames must consider the strengthening of columns, beams and beam-column joints in order to prevent brittle failure modes. The Fiber Reinforced Polymer (FRP) strengthening procedure would cause minimal increase in size and weight of structural members, ease of installation and efficient corrosion resistance. In this paper, seismic behavior of RC frames retrofitted with FRPs under time histories of significant earthquakes have been investigated using neuro-fuzzy modeling. Neuro-fuzzy (as one of the fields of artificial intelligence), refers to combinations of artificial neural networks and fuzzy logic. Considering the available experimental data, the main parameters influencing on global behavior of retrofitted RC frames including the dimensions of RC members and characteristics of concrete and retrofitting system were assumed as the input parameters for generalizing the networks. The failure load of the frame was assumed as the target node. After training the networks and selecting the optimized one, the influence of important parameters on the behavior of the frame was investigated. The results showed that the thickness of FRP and ultimate tensile strength of FRP system are most effective parameters affecting on the global behavior of a non-ductile RC frame and will increase the load carrying capacity of RC frame.