Efficient image-based homogenized limit analysis for irregular masonry with lumped mortar joints representation

Mesomodels for masonry and historical structures typically focus on regular bond masonry, for which the geometrical description and the discretization are straightforward. The present contribution leverages image segmentation tools on acquired images of real historic masonry assemblies. The geometrical information of the mortar joints is lumped on null-thickness interface descriptions by exploiting the notion of medial axis, a concept from the field of image processing. A kinematic limit analysis problem is solved by formulating the problem statement as a linear programming problem, with the assumption of infinitely strong and rigid blocks. This technique is applied to analyse the homogenized failure behaviour of a representative volume element (RVE-) window, which corresponds to a representative portion of the historical masonry arrangement used to derive the overall mechanical properties and behaviour of the entire assembly. Based on successful image acquisition and segmentation, this approach computes homogenized failure envelopes in the stress space with the corresponding failure mechanisms with unparallelled computational efficiency, providing a streamlined pathway from image-based input to homogenization. Results in terms of homogenized failure envelopes in the stress space and the corresponding failure mechanisms are critically commented on. [Display omitted] •A methodology allowing image-based limit analysis homogenization is developed.•Homogenized strength is evaluated on real masonry images with unparallelled speed.•Lumping joints and assuming infinite block strength reduces the computational cost.•Simplified geometries yield similar results with enhanced computational efficiency.•The framework efficiency depends on the robustness of the image segmentation tool.