Two-dimensional (2D) mesoscopic simulations of accelerated corrosion tests are frequently performed to avoid the much higher computational costs of three-dimensional (3D) mesoscopic simulations. 2D models assume geometric invariance of the mesostructure in the direction perpendicular to the analyzed plane. However,…
Two-dimensional (2D) mesoscopic simulations of accelerated corrosion tests are frequently performed to avoid the much higher computational costs of three-dimensional (3D) mesoscopic simulations. 2D models assume geometric invariance of the mesostructure in the direction perpendicular to the analyzed plane. However, this is different from the real concrete mesostructure, consisting of polyhedral (rather than cylindrical) aggregate particles embedded in a mortar matrix. This provides the motivation for a comparative analysis of 2D and 3D mesoscopic simulations of the accelerated rebar corrosion test by Andrade et al. [Mat. Struct., 1993, 453-464]. The 3D model resolves the concrete mesostructure in the vicinity of the rebar into polyhedral aggregates, embedded in the mortar matrix. The 3D simulation accounts for non-uniform corrosion penetration into the rebar, non-uniform rust deposition on the rebar surface, and crack propagation through the mesostructure of concrete. Four 2D models are generated from four different cross-sections through the 3D model, perpendicular to the axis of the rebar. The 2D simulations are based on the assumption of either a plane strain state or a plane stress state. The comparison of the results of 2D and 3D simulations indicates that the 2D simplification does not necessarily result in the realistic simulation of the interac- tion of propagating cracks with the aggregates. This may lead to wrong predictions of crack propagation paths and crack opening widths.