Quantification of rust properties and distribution of non-uniform corrosion in concrete

Abstract

Reinforcement corrosion is one of the most significant deterioration of reinforced concrete structures. As a fundamental step to predict the service life for protection against corrosion and evaluation of repair and maintenance schedules, it is essential to accurately simulate the crack process of concrete structures subjected to corrosion of reinforcement. This is however challenged by the quantification of many key parameters, including the deformation modulus of rust, the distribution of rust and the amount of diffused rust. This thesis is herein devoted to develop new methods to quantify the above mentioned parameters for prediction of corrosion induced cracking of concrete structures. To evaluate the deformation modulus of rust in concrete, a new displacement-based inverse analysis method is proposed by combination of theoretical and experimental analysis. The crack propagation of concrete is modelled by a smeared crack model, in which the anisotropy and strain softening effect in cracked concrete are considered. New empirical relationship of the deformation modulus of rust is presented and validated by both of the previous laboratory experiments and the field tests in literature. By investigating the deposition coefficient, defined as the ratio between the thickness of the deposited rust and the corrosion depth, the relationships among corrosion depth, expansion and deposition of rust are quantified through accelerated corrosion test. By testing parameters including the geometry, mix of concrete and corrosion rate, the value of deposition coefficient was evaluated to estimate the diffusion of rust. The non-uniform distribution of rust can always cause stress concentration on interface of steel and concrete. In the present work, a new method to predict the non-uniform distribution of rust in concrete is developed by comparing the Wasserstein distance (WD) between the target distribution and a known distribution. The non-uniformity and the total area of rust are employed as the parameters for prediction, which are functions of the degree of corrosion. By comparing the predicted distribution with experimental data, it is proved that the WD-based analogous method can accurately predict the non-uniform distribution modelled by one-peak Gaussian function. The crack pattern for multi-peak distribution of rust is investigated by accelerating corrosion test. The observed cracks are thereafter prescribed as the positions of peaks of the Gaussian function-modelled distribution of rust. Numerical investigation validates that the multi-peak distribution of rust can be predicted through a new method by using accelerating corrosion test. The methods presented in this thesis are non-destructive by monitoring the displacement field on concrete surface with digital image correlation (DIC) or utilizing the relationships between peaks of rust and cracks of concrete, hence they can be easily applied to test different parameters. A framework was proposed to optimize the two dimensional (2D) DIC measurement. By employing the accelerating corrosion test, the effect of rust diffusion can be neglected or incorporated into the deformation of rust. All the work presented in this thesis has filled a gap of accurate quantification of the parameters for prediction of corrosion induced cracking of concrete structures.