PSI - Issue 78

Marta Bertassi et al. / Procedia Structural Integrity 78 (2026) 1521–1528

1522

1. Introduction The seismic vulnerability of unreinforced masonry (URM) structures has been widely recognised and extensively documented in the literature (Dhizur et al ., 2011; Giuffrè, 1996; Ingham et al ., 2011; Penna et al ., 2014; Sorrentino et al ., 2014). Among the various failure mechanisms, the out-of-plane (OOP) collapse of masonry walls is of particular concern due to its inherently dynamic nature and its critical implications for vulnerability assessment. This response is commonly governed by the rocking motion of wall segments that rotate and impact against each other or the floors under seismic loading (Costa et al ., 2013). In current engineering practice, OOP mechanisms are typically assessed using simplified nonlinear kinematic analysis (NLKA) procedures prescribed by the Italian Building Code “ NTC2018 ” (MIT, 2018; MIT, 2019) and the final draft of the second- generation Eurocode 8 “ FprEN 1998 ” ( CEN 2025a,b), which assume rigid-body motion and rely on a code-like formulation of floor acceleration spectra. However, such approaches may neglect important aspects of the dynamic interaction between the wall and the rest of the building, potentially leading to inaccurate estimations of OOP displacement demands. This study aims to highlight this gap by performing a comprehensive parametric investigation of vertical spanning strip wall (VSSW) behaviourusing two alternative assessment strategies: (i) a code-based NLKA approach, where the seismic demand is defined by standard code spectra and the wall response is evaluated using a SDOF model, and (ii) a refined approach based on nonlinear dynamic analyses (NLTHA) with increasing scale factors applied to the input motions. The comparison between these approaches is carried out across a wide range of configurations, including variations in wall geometry, vertical overburden, and floor level, aiming at quantifying discrepancies in predicted displacement demands. The results contribute to the ongoing discussion on the adequacy of current code-based seismic assessment procedures for OOP mechanisms in URM structures, with particular reference to the NTC 2018 and FprEN 1998. NLTHAs were carried out using two different SDOF numerical models. The first one was used to simulate in a simplified way the global seismic response of masonry buildings, allowing for a proper estimation of acceleration amplification at different floor levels (Graziotti et al ., 2016). The second was employed to simulate the nonlinear OOP response of masonry walls (Tomassetti et al ., 2018). As case studies, three simplified URM buildings with one, two, and three floors were considered. Each building was modelled as a nonlinear SDOF system (Graziotti et al ., 2016), with fundamental periods ( T 1 ) of 0.1 s, 0.2 s, and 0.3 s, respectively. These configurations are referred to as 1F , 2F , and 3F (Fig. 1). In this way, the ground motion was filtered through the SDOF model, providing floor-level accelerograms representative of the expected seismic demand at the gutter level. For each of these cases, only the walls highlighted in grey were analysed (Fig. 1), and two different structural scenarios were considered. The first assumes a high lateral strength of the building (referred to as EL ), effectively resulting in an elastic response throughout the seismic excitation. In contrast, the second assumes limited lateral strength, with an equivalent SDOF acceleration capacity a * = 0.3 g, leading to significant nonlinear behaviour. This latter assumption reflects the realistic expectation that URM structures are unlikely to remain elastic under seismic excitations intense enough to trigger OOP failure mechanisms. These two scenarios were thus selected to bound the expected range of dynamic amplification at the floor levels under consideration. Additionally, a ground floor ( GF ) configuration was analysed, involving the application of ground acceleration directly to a wall, thereby representing the base-level input without structural filtering. The input motion at the ground floor level ( GF ), as well as those for the upper floors ( 1F , 2F , and 3F ), were directly applied to OOP SDOF models without averaging with respect to the floor below. Thus, the same signal was applied both at the base and at the top of the considered masonry wall. 2. Modelling assumptions and case study definition 2.1. Global building models and floor accelerograms