Efficiency and accuracy of a multiscale domain activation approach for modeling masonry failure

Abstract

Masonry is a composite consisting of two very different materials, which results in complex structural behavior. There exist accurate models in which both constituents are modeled explicitly, such as microscale and mesoscale models, but they require a great amount of computational resources. A faster, alternative, strategy to model masonry structures is the use of macroscale models, where homogenized elements are used which condense constituent behavior into one composite material. This approach also removes the upper bound on finite element sizes, which can lead to a reduction in element numbers. However, using one single material type makes it hard to capture the inherent nonlinear behavior when simulating masonry failure. In order to find a compromise between accuracy and computational efficiency one can use a scale embedding multiscale model where both macro- and microscale elements come into play, combining the advantages both have to offer. In this work a finite element based framework that formulates an adaptive scale embedding multiscale technique is presented, with the goal of both accurately and efficiently simulating large masonry structures. This theory is tested and compared to show its accuracy to rival a fully microscale model, while at the same time comparatively having a higher computational efficiency. In this work, the developed multiscale model is compared to its underlying microscale model using a couple example structures, ranging from small to large scale 2D unreinforced masonry walls with openings, including an application related to failure due to soil settlement, showing a potential for the application of this type of modeling.