Construction materials used for masonry units and mortar are typically quasi-brittle. The current research presents a computational framework to simulate fracture in these materials using the discrete element method. In this framework, the surface geometry is first discretized into a fine unstructured triangular tessellation, which is then extruded to form triangular blocks that constitute the 3D volume. The fine discretization enables modelling of the initiation, propagation and coalescence of cracking within the simulated materials. The ability of the proposed modelling approach to reproduce realistic fracture patterns is first evaluated using experimental results from the literature. The comparisons demonstrate that the proposed framework can be used to simulate Mode-I fractures in direct tension as well as mixed mode fractures under compression. The influence of block size on the crack pattern and global load–displacement predictions is systematically explored. The framework is then utilized to investigate the influence of pre-existing morphological irregularities on the fracture response. For this purpose, a custom algorithm is developed to introduce joined element clusters (to simulate stiff inclusions within the material) and voids (to simulate defects within the material) within the geometry. Results show that defects within the material, particularly in masonry mortar joints, act as crack initiators and significantly reduce load-carrying capacity, underscoring the importance of mortar integrity in structural performance. The examples highlight the framework’s potential to enhance the performance evaluation of masonry materials and structures.

Unreinforced masonry (URM) walls are common load-bearing structural elements in most existing buildings, consisting of masonry units (bricks) and mortar joints. They indicate a highly nonlinear and complex behaviour when subjected to combined compression–shear loading influenced by different factors, such as pre-compression load and boundary conditions, among many others, which makes predicting their structural response challenging. To this end, the present study offers a discontinuum-based modelling strategy based on the discrete element method (DEM) to investigate the in-plane cyclic response of URM panels under different vertical pressures with and without a damp-proof course (DPC) membrane. The adopted modelling strategy represents URM walls as a group of discrete rigid block systems interacting along their boundaries through the contact points. A novel contact constitutive model addressing the elasto-softening stress–displacement behaviour of unit–mortar interfaces and the associated stiffness degradation in tension–compression regimes is adopted within the implemented discontinuum-based modelling framework. The proposed modelling strategy is validated by comparing a recent experimental campaign where the essential data regarding geometrical features, material properties and loading histories are obtained. The results show that while the proposed computational modelling strategy can accurately capture the hysteric response of URM walls without a DPC membrane, it may underestimate the load-carrying capacity of URM walls with a DPC membrane.

Post-earthquake investigations show that unreinforced masonry (URM) buildings may exhibit diverse failure mechanisms depending on the construction morphology and the connection detailing between their structural components. Advanced computational models are necessary to consider the influence of these aspects. However, realistically reproducing the post-collapse state of an existing URM building is challenging when limited data is available on the aforementioned features. To address this challenge, a framework for exploring the seismic behavior of URM buildings is presented. The current investigation presents two case study buildings located in Türkiye's Hatay province: the Mithatpaşa Primary School in Iskenderun and the Liwan Boutique Hotel in Antakya, both of which suffered partial collapses during the recent Kahramanmaraş Earthquakes in 2023. Discrete block models of the two case study buildings are generated based on geometrical information obtained from various pre- and post-collapse vision-based data sources. An automatic block generation algorithm is proposed to replicate periodic and nonperiodic masonry wall patterns. Next, the generated discrete block media are analyzed using discontinuum-based structural analysis to predict the seismic response of the structures. Comparisons between the preliminary pushover analysis results and collapse observations inform further analyses, and lead to an exploration of how construction morphology and connection detailing may have contributed to the partial collapse of the buildings. It is demonstrated that this iterative approach, supported by forensic site evidence and reverse engineering analysis, provides new insight into the influence of key factors that contribute to collapse. This information can help safeguard similar structures and inform the development of effective retrofitting solutions.