On Mesoscale Modelling of Coupled Chloride-Moisture Transport in Concrete
The service life behaviour of concrete structures is highly influenced by the ingress of chloride ions. At sufficiently high concentrations, chloride ions can initiate corrosion of embedded reinforcement bars, which may substantially reduce the structure's bearing capacity and ultimately lead to structural collapse. Thus, a simulation tool for chloride ingress is desirable for predicting the time span during which reinforcement corrosion may occur.
Concrete is a strongly heterogeneous material used extensively in civil engineering applications. On the mesoscale, its heterogeneities can be characterized mainly by the presence of embedded ballast particles in the cement paste. On the microscale, pores in the cement paste constitute spatial variances in material properties. Taken together, these heterogeneities govern the overall transport behaviour of chlorides. However, this fact is often not considered by material models as they usually are based on the assumption of homogeneity.
This thesis concerns modelling of coupled moisture-chloride diffusion in concrete, considering the material heterogeneities by means of a multiscale model, including computational homogenization. The model involves two scales: the macro- and mesoscales. On the macroscale, concrete is considered homogeneous whereas the mesoscale comprises the material's heterogeneities. In the model, the mesoscale is said to be contained within a representative volume element (RVE), which, computationally, is introduced in the Gauss points on the macroscale. The RVE, in turn, contains the mesoscale constituents of concrete in terms of cement paste, ballast and the interfacial transition zone (ITZ).
Computational results obtained by the model establish how the diffusion of moisture and chloride ions on the macroscale is influenced by the concrete composition on the mesoscale. This is obtained for both stationary and transient conditions. The effects of ballast content, ITZ and coupling parameters, in particular, are studied.