Numerical modeling of time to corrosion induced cover cracking in reinforced concrete using soft-computing based methods

Abstract

Reinforced concrete (RC) is one of the most commonly used composite materials in construction industry. Corrosion of reinforcing steel embedded in concrete is a crucial issue leading deterioration of RC structure. In this study, mathematical formulations that predict the time from corrosion initiation to corrosion cracking in RC elements subjected to accelerated corrosion test are presented. For this, the time to corrosion cracking t cr in RC elements is evaluated and developed using soft-computing techniques, namely, genetic algorithms and artificial neural networks. In the models, nine critical estimation parameters were considered. The experimental inputs are mix design properties of the concretes, curing conditions, testing age, mechanical properties of concrete, and cover thickness of RC elements. The dataset was formed by collecting 126 experimental data samples reported in the technical literature. The dataset was randomly separated into three parts for training, testing, and validating the models. Through the numerical study, the influences of various experimental factors on the duration and extent of RC corrosion-induced cracking were shown. It was found that the soft-computing based models gave reasonable predictions of t cr in RC elements. However, the neural network model performed better and the highest correlation coefficient (R) between predicted and experimental t cr values was computed as 0.998.