Abstract
In this paper, a multiscale modelling strategy to simulate salt crystallization-induced damage in porous materials is proposed. Salt crystallization pressure exerted on pore walls is explicitly modelled on a nonlinear representative volume element (RVE) at the microscale of the porous medium. A macroscopic damage measurement of the whole RVE can be then extracted for any combination of crystallization pressure and pore filling time histories. The efficient coupling of moisture transport and salt crystallization with micromechanical damage is achieved by adopting a state-of-the-art multiphase model for the transport/crystallization part and by originally formulating an efficient phenomenological damage model, trained on a dataset generated through micromechanics-based simulations on RVEs. The effectiveness of this numerical strategy is shown via the comparison with an experimental campaign on salt-aged traditional Dutch tiles. The proposed numerical strategy appeared able to track the evolution of macroscopic damage in real-time along with salt transport and crystallization within the porous medium. The potential for using the proposed framework with extended datasets and simulation-driven machine learning is also highlighted.