Uncertainties regarding structural safety of reinforced concrete structures may warrant a need for a detailed assessment. A detailed assessment using nonlinear finite element analysis is one of the alternatives which could help in decision-making about maintaining, upgrading or even demolishing and rebuilding of the structure. A reliable assessment should account for the existing damage in the structure, which may cause re-distribution of stresses within the structure, giving rise to unexpected failure modes. The existing approaches in nonlinear finite element analysis which account for the effect of already undergone damage in concrete, pose a number of limitations which either makes structural analysis ambitious or the uncertainty of concrete damage is not effectively accounted for. An alternative approach is adopted in this thesis, which is phenomenological and probabilistic in nature. Existing damage in concrete is conceived as a statistical field, which can be input into a finite element model, such that the existence of damage is taken as the starting point of the structural analysis. Focusing on indications of damage on the surface of concrete, i.e. crack patterns, an exploratory methodology based on image analysis is developed, to account for information obtained from crack pattern observation into nonlinear finite element analysis of reinforced concrete structures. The methodology is implemented on MATLAB and validated on damaged experimental specimens. Results of the computational analyses indicate good efficiency in predicting residual load carrying capacities and failure modes, alongside insightful numerical crack patterns. Through a critical examination of the obtained results and reflection upon the assumptions and simplification made in the methodology, recommendations for future research are provided.

One of the most severe deteriorations in reinforced concrete structures is associated with reinforcement corrosion, where the corrosion distribution shows considerable spatial variability. In the existing investigations of spatial variability of corrosion on the reliability of reinforced concrete structures, symbolic expressions are used for the structural performance. However, structural behaviors like stress redistribution and plasticity spread are not captured. In this research, a computational framework is designed to couple the probabilistic analysis with the nonlinear finite element analysis. Random fields is used with the computational framework to represent the spatial variability of corrosion.A 60 m girder of a reinforced concrete bridge is taken as the case study to explore the effect of spatial variability of corrosion on reliability of static indeterminate reinforced concrete structures. The reliability analysis is target on ultimate capacity of the beam to carry the traffic load. The girder is modeled as a 3 span continuous beam. The modelling of corrosion damages on reinforcement (area of cross section and physical properties) is based on assumption of pure pitting corrosion. In the axial direction of the beam, spatial variability is modeled with random field. In order to study the effect of different level of spatial variability, the correlation length of the random field is set from infinite large to 125 mm. Between adjacent bars, extreme cases of fully spatial correlated and totally spatial independent are studied.The case study shows:i) when corrosion develops, pitting corrosion can severely reduce resistance of the structure and localized damage may lead to a brittle structural response.ii) spatial variability of pitting corrosion in the axial direction of the beam leads to higher failure probability of the structure.iii) spatial variability of pitting corrosion between adjacent bars leads to lower failure probability of the structure.iv) the effect of spatial variability of corrosion on reliability of static indeterminate reinforced concrete structure is a collective effect of probabilistic and physical characteristic.In order to facilitate the reliability assessment of corroded reinforced concrete structures, additional efforts are also contributed to: quantification of uncertainty for the bond model of corroded reinforcement; comparison of various reliability methods with the probabilistic nonlinear finite element analysis framework.The thesis fills the knowledge gap of probabilistic nonlinear finite element analysis of reinforced concrete structures with spatial varied corrosion and quantifies the effects of spatial variability of corrosion on the reliability of a static indeterminate reinforced concrete structure. The analysis framework designed in this thesis is also applicable for other reliability assessment of corroded reinforced concrete structures.