Simplified mechanical models for the seismic collapse performance prediction of unreinforced masonry structures

Unreinforced masonry, including non-confined unreinforced masonry (NCURM) and confined unreinforced masonry (CURM), constitutes a significant portion of low-rise buildings and is prone to collapse due to high seismic vulnerability. Seismic collapse performance prediction is an effective method to reduce structural collapse, and it requires the analytical model that can correctly simulate structural mechanical behaviors, especially for the deterioration behaviors from accumulation damage. Macro-scale structural models constructed based on hysteretic models of walls have small computational efforts compared to micro-scale finite element models, and are preferred for structural seismic collapse performance analysis. However, existing hysteretic models of NCURM wall and CURM wall lack accuracy in considering the deterioration behaviors due to incorrect model forms or lack of available model parameter values. In this study, an experimental database consisting of cyclic tests on 47 NCURM walls and 50 CURM walls is assembled based on the existing literature. The hysteretic models of NCURM wall and CURM wall are developed by the combined use of the assembled database and the Ibarra-Krawinkler (IK) model from Ibarra et al. The IK model has versatile consideration of the in-cycle and cyclic deterioration behaviors and is used for the modeling of walls. A calibration procedure that clearly distinguishes the in-cycle and cyclic deterioration behaviors is established, and the model parameters of test walls in the database are calibrated. By regression analyses of calibration results, a set of empirical equations are created for determining model parameters. To satisfy the seismic collapse performance analysis of regional masonry buildings, the multi-degree-of-freedom (MDOF) shear model of masonry structure is further given based on the wall hysteretic model, which is constructed by the structural inter-story mechanical behaviors and has very small computational efforts. Good accuracy of both wall hysteretic model and structural MDOF shear model is verified by the comparisons with experimental and numerical results.