The durability and serviceability of concrete structures is influenced by both the early-age behavior of concrete as well as its long-term response in terms of shrinkage and creep. Hygro-thermo-chemo-mechanical models, as they are used in the present publication, offer the possibility to consistently model the behavior of concrete from the first hours to several years. However, shortcomings of the formulation based on effective stress, which is usually employed in such multiphase models, were identified. As a remedy, two alternative formulations with a different coupling of shrinkage and creep are proposed in the present publication. Both assume viscous flow creep to be driven by total stress instead of effective stress, while viscoelastic creep is driven either by total or effective stress. Therefore, in contrast to the formulation based on effective stress, they predict a limit value for shrinkage as observed in long-term drying shrinkage tests. Shrinkage parameters for the new formulations are calibrated based on drying shrinkage data obtained from thin slices. The calibration process is straightforward for the new formulations since they decouple shrinkage and viscous flow creep. The different formulations are compared using results from shrinkage tests on sealed and unsealed cylindrical specimens. Shrinkage strain predictions are significantly improved by the new formulations.
Keywords: creep; hardening concrete; hygro-thermo-chemo-mechanical modelling; shrinkage.