Combined Cracking in Concrete : A Study About the Combination of Restrained Shrinkage and Flexural Cracking

This study focuses on the cracking behaviour of reinforced concrete (RC) slabs subjected to restrained shrinkage and bending. Cracking due to shrinkage is a common concern in concrete structures and often occurs before external loads are applied. In practice, design methods typically treat shrinkage and bending cracks separately, calculating their effects independently and then combining them linearly. However, this approach may result in overly conservative estimates of crack widths, which can lead to inefficient designs. The primary aim of this study is to evaluate the interaction between restrained shrinkage and bending to determine how these affect the development of cracks in RC slabs, both individually and in combination.The study was carried out in collaboration with ELU Konsult AB and the Division of Concrete Structures at KTH the Royal Institute of Technology in Stockholm. A non-linear numerical model was developed in the finite element software Abaqus, where concrete was modelled using the Concrete Damage Plasticity (CDP) model. The interaction between concrete and reinforcement was represented using a connector-based bond-slip approach validated to match realistic slip behaviour. The model was verified against experimental data to reflect cracking under shrinkage and bending.A parametric study was performed in 2D to explore the influence of several key variables, including slab thickness, concrete strength class, reinforcement diameter, spacing, and concrete cover depth. Due to the limitations of 2D modelling – such as the inability to include out-of-plane effects or fully represent reinforcement arrangement – reinforcement was idealized using an equivalent diameter. The findings demonstrated that the interaction between shrinkage and bending is non-linear, and that performing two separate analyses and combining the results linearly leads to overestimation. In numerical simulations, applying shrinkage first and then bending as an integrated model, resulted in crack widths less than half as large as those predicted by linearly combined numerical approaches.Analytical calculations were performed using Eurocode 2 (both current and next generation) and an evolved Björn Engström method for shrinkage. The results showed that when using current Eurocode 2 for both shrinkage and bending, predicted crack widths were up to 4 times larger when linearly combined, than those from the integrated numerical model. When using the evolved Björn Engström method for shrinkage and the new Eurocode 2 version for bending, the overestimation was around 2.6 times. These comparisons suggest that linear addition provides a conservative estimate, but can significantly overpredict crack widths, particularly when using the current Eurocode 2.