Progression from failure to collapse of two-way spanning URM walls studied using DEM parametric dynamic analysis

The out-of-plane seismic collapse mechanisms of two-way spanning unreinforced masonry (URM) walls vary depending on boundary conditions, overburden loads, wall length, diaphragm stiffness, masonry properties, and the presence of openings in the wall. Interpreting the collapse mechanism from the final collapsed state can be misleading. This study investigates the full collapse progression of two-way spanning URM walls under seismic loading, from initial crack formation to final collapse, focusing on the role of boundary conditions, masonry properties, wall length, and diaphragm interaction. The findings aim to bridge the gap between assumptions that original boundary conditions and bond properties can be inferred from the collapsed state and provide guidance on details to observe in post-earthquake scenarios for an informed assessment. The discrete element method (DEM) was used for quasi-static simulations of experimental testing to calibrate the model, followed by a parametric dynamic analysis to explore collapse mechanisms. As a consequence of the parametric study dynamic resurrection and the collapsing sequence that led to a similar pattern of remaining brickwork after collapse of two-way spanning walls with varying boundary conditions and bond strengths are reported for the first time. Simulated walls with flexible diaphragms showed similar performance to those without diaphragms, while heavy diaphragms caused significant damage due to hammering of the heavy joists. Plates and anchors connecting the wall to the diaphragm enhanced collapse resistance. Comparisons of the DEM results with real post-earthquake observations revealed that the collapsed state often could not be accurately correlated to boundary conditions or bond strength of the as-built wall but was consistent with wall length. • DEM simulations showed that original boundary conditions and bond strength cannot be reliably predicted from collapsed state. • Dynamic resurrection observed in two-way spanning URM walls for the first time. • Analytical models compared to simulations revealed key limitations and highlighted the need for refinement. • Longer walls experienced lower collapse acceleration and fewer remaining bricks after collapse due to arching effects. • Simulated collapse patterns matched real post-earthquake observations.