Multi-scale multi-physics modeling of alkali-silica reaction in concrete: experimental investigation and numerical prediction

Pathirage, Madura

doi:https://doi.org/10.21985/n2-03xz-0s05

Work

Multi-scale multi-physics modeling of alkali-silica reaction in concrete: experimental investigation and numerical prediction

Public

Download PDF

The effect of Alkali-Silica Reaction (ASR) on concrete mechanical behavior and the multi-physics considerations that come along are highly complex. Hydration and other chemical reactions occur at the micrometer scale. Hygro-thermal phenomena and concrete cracks due to ASR are typically studied at the meso-scale, i.e. at the aggregate level. At larger scales, the effect of concrete reinforcement becomes essential in the damage assessment of ASR-affected structural elements. For massive structures such as dams and bridges, it is also necessary to consider the confining pressures that affect drastically the evolution of concrete deterioration over the course of time. Modeling all these physical processes is challenging and their combination seems almost impossible. The time is finally ripe for considerations of this nature. This thesis proposes an overarching multi-scale multi-physics computational framework to tackle this problem. The chemical reactions involved during hydration and occurring at micro-scale were directly simulated from a realistic micro-structure development model. All relevant information was passed to a macroscopic semi-empirical hygro-thermo-chemical model. The bridging was successfully applied for the parameter-free prediction of self-desiccation in concrete and other cementitious materials. At the meso-scale, the mechanical behavior was modeled by simulating the interaction between aggregates using a discrete lattice approach. Specific models to describe ASR, aging, shrinkage, and thermal expansion were incorporated. The coupled models were first calibrated on experimental data relevant to accelerated mortar bar tests. The overall computational framework was then used to predict the short term evolution of strength caused by the intertwined effects of aging and ASR. In order to continue the ascent, larger scale experiments were performed on reinforced concrete beams tested under four point loading. A novel hybrid experimental/numerical method was used to simulate their failures and understand the effect of shear reinforcement. The non-negligible effects of creep and shrinkage were this time taken into account in the modeling. All the previously stated models were fully calibrated apart of this thesis, which made the corresponding simulation results pure predictions. Finally, the effect of confinement was analyzed on one set of experimental data published in the literature. Stresses generated by different confinement pressures and the effect of ASR on concrete expansion was simulated. Strength reduction in concrete cores was also simulated with reasonable accuracy. The numerical framework presented so far was mainly deterministic as only the chaotic aggregate distribution was accounted for. In the perspective of extending the modeling capabilities to reproduce the true random behavior of concrete, a stochastic discrete lattice framework based on random fields was finally implemented.

Creator