Thermo-hygro-chemical model of concrete: from curing to high temperature behavior

Abstract

Concrete is a heterogeneous multiphase material composed of various solid phases that interact both physically and chemically with each other and with the water filling the pores. Among these solid phases, a crucial role is played by the calcium silicate hydrates (C-S-H), which are the primary products of cement hydration and are primarily responsible for the material's physical properties. When concrete is subjected to high temperatures, the chemically bound water in C-S-H is progressively released, leading to a degradation in the strength and durability properties of the concrete. Hence, understanding how the dynamics of C-S-H dehydration and the corresponding evolution of hygro-mechanical properties (e.g. strength, permeability, porosity) are related with the characteristic observed phenomenology of spalling is crucial to assess the resistance of a structure under high temperature. Within this context, multiphysics thermo-hygro-chemical (THC) numerical models now play a pivotal role in predicting and analyzing structures' performance under fire accidents. However, to enhance the reliability of numerical results, properly accounting for the initial hygro-chemical state of the structure just before the accident is of chief importance. This work presents a monolithic fully-coupled unified THC mathematical model enabling the simulation of the full service life of the material: from casting (early-age behavior and curing), through aging, until the eventual occurrence of an accident (high temperature, high pressure, ...). The model provides the evolution of the hydration reaction as a function of time, temperature, and relative humidity, as well as the eventual dehydration occurring at high temperature. The main contribution of this work lies in the proposition of general chemo-physical constitutive relationships that incorporate the influence of the hygro-thermal state of the material as well as that of C-S-H hydration/dehydration in a fully-coupled manner. The evolution of volume fraction of phases and porosity during hydration/dehydration follows Powers' stoechiometric model, while a novel adsorption-desorption model is proposed to properly account for the irreversibility of chemical damage in the porous microstructure. This enables an alternative, simpler approach requiring only a limited number of experiments for the model calibration. The model is firstly benchmarked by simulating the early-age behavior of a concrete sample, and it is then validated against experimental results of temperature, gas pressure and mass loss under heating. Our results highlight a non-negligible impact of the initial (which in real cases is usually heterogeneous) hygral state on the predicted behavior at high temperature and unravel new perspectives on understanding the physics underlying concrete spalling. The thermo-hydro-chamical code developed in this paper is made available in a GitHub repository.