Abstract
This study presents the numerical simulation of a shake table experimental earthquake campaign of a building aggregate composed of two adjacent unreinforced rubble stone masonry buildings. The experimental testing was performed with the purpose of studying the interaction between a single-storey and a two-storey building connected with a dry joint consisting of a smooth mortared surface. Before performing the experimental testing, various research teams were sent the construction details to participate in a blind prediction competition using different prediction strategies. The approach reported herein to simulate the shake table tests is the discrete element method (DEM) with rigid elements and damage and deformation lumped in inter-block joints that represent the mortar interfaces governed by a non-linear Mohr–Coulomb constitutive laws. The material properties implemented in the model after calibrating using piers shear tests was observed unrealistically stiff. Hence, it was reduced based on the outcome of pushover and eigenvalue analyses. A sequence of earthquakes with incremental acceleration was input to the real and numerical models. Numerical overestimation of damage and displacement was observed probably due to underestimating the damping ratio. Unexpected sliding of the single-storey building occurred at early stages of the simulation. However, the overall behaviour in terms of base shear force, building displacement and damage progression was well captured in the DEM model. The in-plane flexural and rocking mechanism in the two-storey building was correctly simulated. Damage at the interface between the two buildings with separation and pounding was also reasonably well predicted.