Prediction models for bond strength of steel reinforcement with consideration of corrosion

Abstract

Corrosion phenomena is one of the main deterioration causes, which remarkably affects the behavior of structural reinforced concrete (RC) members in seismic regions. Researches on reducing rehabilitation cost, performance assessment, and accurate modelling of corrosion-affected RC structures are progressively becoming popular in recent years. Corrosion diminishes bond capacity between reinforcement and surrounding concrete, which induces reduction in strength and ductility of members. The aim of this investigation is to provide a prediction approach based on a large number of results from published researches related to corroded reinforcement in concrete members using artificial neural networks (ANN). The minimizing mean square error criterion and increasing regression value of predicted results are considered for evaluation of training performance of ANN models. The validity of proposed model is checked using collected experimental database. Results show that estimated model has acceptable agreement with experimented data.