PSI - Issue 78

Francesco Testa et al. / Procedia Structural Integrity 78 (2026) 905–912

906

1. Introduction Historic masonry structures form a significant part of the Italian built environment, with churches standing out due to their cultural importance and number. Past seismic events have shown their significant susceptibility to earthquake damage. In response, there is a growing need for reliable seismic vulnerability models that can be applied at a large territorial scale. Such models are essential for informed decision-making and efficient resource allocation, particularly in light of the need to preserve listed historic buildings. Conservation places a heavy burden on public authorities, especially when financial resources are limited and many buildings are exposed to both natural and human-made hazards. Territorial-scale approaches align with international preventive conservation recommendations (Lourenço et al., 2022) and with the multi-level risk assessment and planning strategies outlined in the national DPCM (2011) guidelines. In particular, empirical models, developed using observed damage data from post-earthquake surveys, have become widely adopted tools for predicting the seismic performance of masonry structures. These models also support the prioritisation of assets in greater need of conservation by enabling rapid condition screening. However, most existing models treat masonry churches as single structural entities. This approach tends to overlook the distinct seismic behaviour of individual components, or macroelements, despite their well-recognised influence on overall performance. Among these macroelements, historic masonry bell towers have exhibited particularly high vulnerability during seismic events (Canuti et al., 2023; Sisti et al., 2023; Testa et al., 2024a, 2024b). Their damage level is often markedly different from that observed in other parts of the church, owing to their largely independent structural behaviour. For this reason, bell towers require focused analysis within broader seismic vulnerability assessment frameworks. The effectiveness of risk analysis models, such as fragility functions, largely depends on the ability to identify homogeneous asset classes. Within these classes, it is critical to distinguish the specific attributes that influence the response to seismic hazard (Lagomarsino et al., 2021; Rossetto & Ioannou, 2018). This distinction is essential for reducing the wide variability often observed in the seismic performance of assets within the same class when subjected to identical intensity measures. While ordinary buildings may exhibit greater homogeneity, heritage monuments typically defy broad classification schemes. The literature highlights several key attributes that significantly affect their seismic susceptibility, including the level of maintenance, material quality, structural regularity, dimensions and aspect ratio, interaction with adjacent structures, and the presence of interventions (Lagomarsino, 2006, Masciotta & Lourenço, 2022, Cianchino et al., 2023). However, access to this detailed information is often limited. The quality and completeness of data collected during post-earthquake inspections are influenced by the structure of the survey forms used and the time available for conducting the assessments. To gain a deeper understanding of the seismic response of historic bell towers, this study investigates specific attributes and features recorded in a well-established inspection form used by the Italian Civil Protection. The aim is to identify subsets of towers exhibiting more homogeneous seismic behaviour. For this purpose, a selection of historic masonry bell towers affected by the 2016 – 2017 Central Italy seismic sequence was analysed. The initial dataset, drawn from the Database of Observed Damage (Da.D.O.), is described in detail in Section 2. The first stage of analysis involved grouping the towers by geographical location, at both regional and provincial levels – corresponding to the first and second tiers of administrative division in Italy, respectively. This was done to assess whether geographical classification could capture relevant factors such as traditional construction practices, seismic history, patterns of local response and mitigation strategies. This approach enabled the development of tailored fragility models, which are presented in Section 3. To further explain the differences observed in the seismic response of towers across different areas, three additional explanatory variables were considered. In particular, the assets were differentiated based on their conservation state and the presence of recent strengthening interventions, both prior to the seismic sequence, as well as the characteristics of the seismic scenario preceding the inspection, as discussed in Section 4. 2. Dataset The towers investigated in this study were affected by the Central Italy seismic sequence, which struck a broad area along the borders of the Lazio, Abruzzo, Marche, and Umbria regions. The swarm presented a first main shock on 24 August 2016. This event, with a magnitude (Mw) of 6.2, had its epicentre in the Monti della Laga area near