Mechanism-based continuum modelling of masonry

When dealing with the analysis of masonry structures, numerical models are generally adopted. Different modelling strategies have been proposed throughout the years, including detailed block-based models and homogeneous continuum approaches. While ensuring accurate predictions thanks to explicitly modelling masonry constituents, block-based approaches tend to be computationally expensive. Therefore, continuum models, representing masonry as an equivalent homogeneous material, are usually preferred. This approach allows a net saving in terms of computational cost, although it requires defining a proper constitutive framework. In this sense, accounting for different failure mechanisms appears essential. In this thesis, a novel framework for the mechanism-based continuum modelling of masonry is proposed. Emphasis is given to the possibility of representing different failure mechanisms within a homogeneous continuum approach. A selection of the most prominent ones for regular periodic masonry is taken as a starting point to derive a novel multi-failure strength domain. A single-surface formulation is also proposed, relying on the RealSoftMax function. Its employment in the framework of non-associated elasto-plasticity is investigated through displacement-based Finite Elements (FEs), showing promising results. Higher computational efficiency is then achieved by means of a novel hybrid stress FE. In both cases, the possibility of separately tracking the evolution of a plastic strain variable for each failure mechanism is highlighted. Finally, the post-peak quasi-brittle response of masonry is investigated within a mechanism-based elasto-damaging approach. A damage variable is conceived for each failure mechanism and suitably related with the damage variables for an orthotropic continuum model. The approach allows to track the evolution of damage and the mechanisms activated. Mixed stabilized displacement-strain FEs, recently proposed to deal with constitutive models involving softening, are also employed. Throughout this thesis, numerical examples are presented with the aim of highlighting the models capabilities and potentialities. Well-known benchmarks from the literature are also considered for comparison.