A discrete transport-mechanical approach for modelling the durability of concrete

Reinforced concrete is one of the most commonly used structural materials in the world and is used for buildings and many different types of civil engineering structures, such as bridges, tunnels and airports. The majority of these structures are required to remain in service for at least 50 years while some are expected to last well over 100 years. Fracture of these structures leads to an increase in the permeability of the concrete which in turn can result in increased ingress of water and other aggressive agents, such as chlorides or carbon dioxide, that accelerate the deterioration of these structures. Likewise, the mechanical properties of the concrete can be affected by the transport of moisture into the structure. The increase in moisture can lead to a reduction of the strength and the stiffness of the material. The costs arising from structural failure are extremely high and in practice repair work tends to be implemented even when it is not entirely necessary. Therefore reliable approaches that can describe the interaction between the transport and mechanical properties of concrete and predict resulting structural degradation are of great benefit for practising engineers. Numerical models, such as the one proposed in this work, could be used for predicting when a repair is really necessary. In this work, a transport-mechanical lattice approach to modelling the durability of concrete is proposed. The discretisation of the specimen domain is based on a dual Delaunay and Voronoi tessellation in which the edges of the Delaunay triangles form the mechanical elements and the transport elements are placed along the edges of the Voronoi polygons. The mechanical response of the concrete is described using an isotropic damage constitutive law, while the transport of moisture through the specimen is described using constitutive laws developed for mass transport through porous materials. Both the mechanical and the transport models are assessed individually before the coupling ii between the two models is implemented. The accuracy of the proposed coupled approach is validated through the analysis of an elastic thick-walled cylinder, in which the numerical results are compared with an analytical solution derived as part of this work. The proposed coupled approach is then applied to the case of corrosion-induced cracking of reinforced concrete structures. In this approach, the corrosion products are assumed to behave as a fluid and therefore values of fluid properties are required. A value of viscosity is determined based on the analysis of a concrete specimen containing a single reinforcement bar. Finally, the proposed approach is applied to a concrete specimen containing four reinforcement bars to assess the approach as a predictive model. As expected with the concrete specimen containing a single reinforcement bar, very good agreement between numerical and experimental results is obtained. In the case of a specimen containing four reinforcement bars, it is observed that for small attack penetration depths the proposed approach is in very good agreement with experimental results. As the analysis continued, however, the numerical approach under-estimated the crack width when compared to experimental results