Image-Based Modeling-to-Simulation of Masonry Walls

Abstract

This paper presents a streamlined image-based modeling-to-simulation framework to better assess the hazard vulnerability of masonry structures. In this framework, a masonry wall image is used as an input for 3D discrete element modeling and analysis. Therefore, individual bricks are directly detected from the image and the interactions of the bricks are explicitly considered in the numerical simulation of the masonry wall. The bricks in the masonry wall image are segmented first, from which the geometric information of the bricks is obtained using a bespoke algorithm. Segmented bricks are then approximated into n-sided polygons using the splitting and Douglas–Peucker methods to develop simplified brick geometries. This shape simplification is performed to lower the computational cost for the discrete element analysis while reasonably approximating the overall brick geometry and maintaining the simulation fidelity. The simplified polygons are exported to a free and open-source physics engine as scalable vector graphics, from which a 3D masonry model is developed, and rigid body discrete element simulation is performed using the impulse-based dynamics. This paper demonstrates the image-based modeling-to-simulation framework can estimate collapse scenarios from an input masonry wall image and proposes a prototype that transforms visual images into critical domain knowledge that would be useful for disaster preparedness. With enhanced predictive capabilities, this framework can contribute to innovations in the hazard vulnerability assessment of masonry structures.