Confinement and alkali-silica reaction in concrete: Review and numerical investigation

The effect of confinement and Alkali-Silica Reaction (ASR) in concrete is of paramount importance when assessing damage in existing structures. It appears that this effect has been only seldom investigated with respect to the vast literature on free-expanding ASR, especially in its modeling aspect. This paper proposes to address this topical issue. For this purpose, an exhaustive review on the effect of confinement and ASR was first provided. Key features that must be taken into account were discussed and a discrete model equipped with the relevant multi-physics models was proposed. The computational framework which includes models for moisture diffusion, heat transfer, cement hydration and aging, thermal expansion, creep, and shrinkage was validated through a detailed comparison with a recent large experimental campaign for which all model parameters have already been calibrated. After accounting for possible alkali-leaching in the experiments and adjusting three model parameters, the numerical framework was used to simulate expansion curves, crack distribution, and damage evolution in unconfined samples, and drilled cores. Results show that overall, the model is able to well reproduce the transfer of expansion in the transversal unrestrained direction, cracks having a preferential direction in the restrained direction, expansion behaviors in the three spatial directions, and the evolution of mechanical properties in time for different confinement configurations. The important effect of creep and shrinkage was also emphasized. Crack distributions numerically generated were found consistent with Damage Rating Index analysis. One major finding of this study is that the true strength inside concrete subjected to multi-axial confining loads was found unaltered due to ASR in the direction of confinement. This striking result confirms many practitioners’ field experience with core testing and raises the fact that as sophisticated modeling tools become more and more reliable, existing methods of assessment must be improved by incorporating results from numerical modeling.