PSI - Issue 66

Domentico Ammendolea et al. / Procedia Structural Integrity 66 (2024) 350–361 Author name / Structural Integrity Procedia 00 (2025) 000–000

351

2

1 Introduction Masonry buildings constitute a substantial portion of the global construction landscape, and many of them hold significant cultural and architectural value. Preserving and rehabilitating these structures, particularly residential ones, is crucial to reducing land consumption and promoting sustainable urban development (Gaetano et al. (2022b), Greco et al. (2022)). During the last decades, numerical simulations have become a widespread tool for analyzing the structural behavior of masonry structures. In such a context, different numerical methods have been developed to reproduce the structural behavior of the masonry. These methods are frequently classified into micro- and macro-modeling approaches (Lourenço (2002), D’Altri et al. (2019)). Micro-modeling approaches delve into the intricate details of masonry at the scale of its material constituents, i.e. , brick units and mortar joints (Pulatsu et al. (2016), Pepe et al. (2019), Gaetano et al. (2022a), Greco et al. (2024)). This level of detail allows for an accurate simulation of the mechanical behavior of masonry, particularly in replicating failure mechanisms. However, this precision comes at a cost-computational efficiency. Micro-modeling approaches are generally computationally expensive, limiting their use to the analysis of small-scale masonry components. Macro-modeling approaches simplify masonry as a fictitious anisotropic homogeneous material with phenomenological mechanical properties, offering good computational efficiency (Lourénço et al. (1997), Berto et al. (2002)). However, these methods sacrifice accuracy in predicting damage mechanisms and strain localization. Recent research has increasingly focused on using multiscale models to address the key limitations of both micro- and macro-modeling approaches (Luciano and Sacco (1997), Trovalusci et al. (2009)). Indeed, multiscale models integrate detailed local behavior from micro-modeling approaches with the computational efficiency of macro modeling ones, enabling the analysis of complex structures such as composite materials, biological tissues, and large-scale infrastructure across various scales while achieving a balance between accuracy and efficiency. Most multiscale models have been developed according to the concept of " separation of scales ," which consists of analyzing the material at distinct levels where each level addresses a specific range of scales. The separation of scales typically involves solving detailed problems at smaller scales to inform the behavior at larger scales, thus allowing for a comprehensive understanding of the material or structure while managing computational complexity effectively. In such a context, homogenization techniques are commonly employed at small scales to derive effective material properties that represent the averaged behavior of the heterogeneous microstructure (Trovalusci et al. (2015)). These properties are then used at the large scale, bridging the gap between different levels of analysis and ensuring that the intricate details of the microstructure of the masonry are captured without the need to model them explicitly at larger scales. Multiscale models that adopt the concept of separation of scales are generally classified into hierarchical and semi-concurrent. The former transfers information in a one-way direction from the micro- to macroscale, allowing efficient computations but risking the loss of accuracy for problems involving damage. The latter enables two-way information exchange between micro and macro scales, better representing the damage but at a higher computational cost. More recently, the promising class of concurrent multiscale models has been frequently proposed for analyzing the structural behavior of masonry structures (Greco et al. (2016), Driesen et al. (2021)). Based on the concept of 'scale embedding ,' these models can potentially revolutionize the field. They involve the simultaneous coupling of different representation scales for the material, allowing interactions between fine-scale and coarse-scale models throughout the simulation. Concurrent multiscale approaches integrate detailed microscopic and macroscopic models, enabling a more accurate representation of material behavior and structural response. By solving sub-models at various scales concurrently, these models capture complex phenomena like strain localization and failure mechanisms while maintaining computational efficiency, instilling optimism and hope for the future of structural analysis. This work proposes a novel adaptive concurrent multiscale model for analyzing the failure behavior of periodic masonry structures under general in-plane loadings. The proposed multiscale model adopts a domain decomposition strategy, which consists of assigning different resolutions for the masonry within the computational domain depending on the evolution of stress and damage. In particular, the proposed concurrent model uses a coarse-