Finite Element and Equivalent Frame modeling approaches for URM buildings: Implications of different assumptions in the seismic assessment

The work aims at discussing the influence of modeling assumptions on the seismic assessment of Unreinforced Masonry (URM) structures when the Finite Element (FE) and Equivalent Frame (EF) modeling approaches are adopted. This influence was evaluated with reference to a benchmark URM wall by performing pushover analysis. The geometry of this wall, with two stories and four vertical rows of windows, is representative of a facade of an existing building that was heavily damaged after the 2016-2017 Central Italy earthquakes. Typical issues faced by practitioners in the modeling process are widely discussed, contributing to the harmonization in the use of FE and EF based approaches (i.e., reduction of scatter in predictions). Several numerical models were assessed as a result of the consideration of different modeling assumptions, either individually or in combination. Pushover seismic analyses of the reference wall were performed after validating the FE approach at the panel scale against benchmark shear tests of representative piers. Then, the capacity curves of the reference wall were compared between the FE and EF models as well as the predicted damage patterns against the actual damage. Large differences in terms of secant stiffness were observed, higher in EF models (83%), while there were smaller differences in the maximum base shear force (35% in FE models). All models predicted a similar displacement capacity, except in one case, for which the displacement was much larger. For this reason, there was a wide range of predictions of acceleration capacity for the Near Collapse limit state (2.35-7.22 m/s2). Regarding the damage patterns, the predictions showed a higher concentration of damage at the ground story when compared with the damaged existing wall. The failure mode of base piers is however consistent with the one observed after the 2016-2017 Central Italy earthquakes, i.e., diagonal cracking. Both FE and EF models were unable to predict the severe damage state at the 1st story due to considering a mass proportional pushover. The underestimation of damage is also related to the fact that orthogonal walls are not considered in the 2D pushover analysis. Adopting consistent assumptions for the FE and EF based approaches results in good agreement of predictions at wall and panel scales.