A numerical and experimental meso-mechanical approach for the prediction of creep of concrete

Abstract

A two-step meso-scale approach is proposed to predict the creep of concrete. The approach requires the creep of cement paste as input, combined with meso-scale analysis of digital images of mortar and concrete that are obtained from X-ray microCT scanning. For the generation of input data and validation of the model, a comprehensive series of creep tests are carried out on cement paste, mortar and concrete cylinders for a period of 6 months. The model uses a robust and automatic quadtree decomposition algorithm to convert digital images into meshes, which are then solved using the scaled boundary finite element method based on a continuum approach. The proposed model is separately validated on the mortar and the concrete scale respectively. In the first step, knowing the creep response of cement paste from the tests, the creep of mortar is simulated by considering it as a two-phase composite comprising of viscoelastic cement paste and elastic fine aggregates. Validation of the predicted response is obtained through comparison with the measured creep response of mortar. A similar procedure is then applied for concrete which is treated as a two-phase composite of viscoelastic mortar and elastic coarse aggregates. Finally, by accounting for damage of the mortar phase, it is demonstrated that the nonlinear creep of concrete observed at high stress levels can be attributed to material damage developed at the meso-scale.