Towards automated image-based cohesive zone modeling of cracking in irregular masonry

The models developed for masonry and historical structures in the literature are usually classified into macromodels considering masonry as an equivalent continuum; and micromodels in which brick, blocks, or stones and mortar joints are represented explicitly. In this second category, many contributions dealt with regular bond masonry for which the geometrical description and the discretization are rather straightforward. For irregular masonry structures however, even though both finite element method (FEM) and discrete element method (DEM) approaches have been developed, obtaining a versatile geometry representation and its discretization remains much less straightforward. This becomes an important issue nowadays with the availability of image acquisition techniques, based on which computational models could be derived. The present contribution proposes an automated methodology to produce a line description of the mortar joints of an irregular masonry blocks/stones and mortar assembly, which can subsequently be used in modeling approaches. The starting point of the development is a generation technique based on inclusions packings which uses distance fields to the nearest neighboring inclusions to describe an assembly of blocks or stones geometrically. It is shown that such an assembly and the corresponding distance fields can be used to extract efficiently and in an automated way a line or lumped description of the corresponding mortar joint based on the concept of a medial axis. This line description can subsequently be used to define computational models. This is illustrated by the automated generation of FEM models that represent mortar joints by interface elements equipped with a cohesive law. Simulations on representative volume elements (RVEs) and on a wall are shown to illustrate the methodology that paves the way towards the image-based modeling of irregular masonry structures through the automated generation of cohesive zone-based models.