Experimentally-validated stochastic model for freeze-thaw microstructural degradation and damage of hardened cement paste

The sustainability of concrete structures is highly influenced by their durability with regard to environmental exposures, such as freeze-thaw (FT) attack. So far, no sufficiently accurate model is available in the literature, which correctly predicts and quantifies the durability, i.e. the concrete degradation due to cyclic freezing and thawing. The reasons for that are numerous: on the one hand, the loading, i.e. the FT exposure is a highly variable process and the population of FT events is extremely difficult to classify. On the other hand, freezing and thawing cause very localized changes in the materials properties (which eventually lead to spalling), which due to their very small spatial extent, are hard to detect and to quantify and which are highly variable in nature, depending very much on the moisture content in the concrete as well as the freezing conditions. We address both these aspects by aiming to introduce a stochastic, spatially resolved model for the FT degradation of concrete. We focus our investigations on hardened cement paste (hcp) as the key material and formulate a representative volume element (RVE) consisting of a mixture of larger capillary pores with diameters between 10 µm to 100 µm, which are embedded in a homogeneous, isotropic solid, with sub-micron size porosity and structural features. The mechanical and physical properties of this RVE are simulated numerically, calibrated using statistical nanoindentation and NMR testing (amongst other techniques) characterizing the properties of the solid and µCT and microindentation, characterizing the properties of the entire RVE. The evolution of the properties of this RVE due to FT action is closely studied. We perform FT testing on hcp samples, numerically and experimentally, with high spatial resolution and with various porosities, pore size distributions and minimum temperatures. This study will form the basis for a layered model of the structure. As for the current state of the art, (i) such measurements until now were limited to bulk (non spatially resolved) measurements only and (ii) did not measure the actual mechanical properties but rather used auxiliary parameters such as ultrasound runtime. Here we clearly expect pronounced progress in knowledge with the proposed setup. Also, the strong interlink between experimental work and numerical simulation will allow for a much more complete understanding of the underlying processes, building the basis of a stochastic extrapolation of the experimental findings. The work in the project shall be carried out in close collaboration between Prof. Michael Beer and Dr. Matteo Broggi, who will focus on the stochastic model and analysis, and Prof. Michael Haist, who will carry out the experimental work and the physical modelling. The work will be greatly supported by 4 international collaborators.

Key terms

hardened cement paste, degradation, frost, stochastic damage evolution, random material behavior