PSI - Issue 78

Michele Mattiacci et al. / Procedia Structural Integrity 78 (2026) 1159–1166

1160

1. Introduction

Historic masonry structures constitute a significant portion of the global architectural heritage and are widely acknowledged as invaluable assets due to their unique historical, cultural, and architectural relevance. As such, their preservation has become a strategic priority for governments, international organizations, and heritage protection agencies worldwide. Ensuring the structural safety and long-term conservation of these buildings, particularly in the face of natural and anthropogenic threats, has led to the growing adoption of structural health monitoring (SHM) strategies as a tool to support informed decision-making and risk mitigation Adam et al. (2022). However, the implementation of e ff ective SHM systems for masonry structures remains a significant scientific and technological challenge, particularly in the case of heritage assets. The inherent mechanical complexity of masonry, characterized by its orthotropic and heterogeneous nature, stems from the interaction between its constituent units and joints as well as their arrangement. Typical mechanical features of masonry include a high specific mass, low tensile and shear strength, and limited ductility, especially under out-of-plane loading, which results in quasi-brittle behavior. Moreover, these structures are particularly vulnerable to progressive deterioration that, in severe cases, might potentially lead to partial or total collapse. Such degradation is frequently driven by a combination of long-term phe nomena, such as material aging Saviano et al. (2022) and insu ffi cient maintenance, as well as sudden events, including earthquakes Zizi et al. (2021) or other extreme environmental loads. Additional structural vulnerabilities commonly observed in masonry buildings involve di ff erential settlements or failures of key restraining elements (e.g., tie rods). Environmental stressors, such as pollution and the impacts of climate change, can further accelerate deterioration mechanisms Phillipson et al. (2016), increasing the complexity of reliable monitoring. A major challenge in applying SHM techniques to masonry structures lies in the sensitivity of monitoring features to environmental and operational variability, including fluctuations in temperature, relative humidity, solar radiation, and so forth Sivori et al. (2024). While this issue has been widely acknowledged, most literature contributions ad dress it through simplified compensation strategies, often neglecting key variables such as humidity or solar expo sure—especially when dealing with actual SHM applications. On the other hand, more detailed models that attempt to rigorously capture the hygro-thermal e ff ects on the static response of masonry have proven e ff ective only in controlled settings Ramirez Alvarez de Lara et al. (2023), typically involving small-scale wall specimens. These physics-based approaches are not suitable for real-time deployment on actual structures, which are often characterized by complex geometries, heterogeneous materials, and pre-existing damage, all of which limit the accuracy and generalizability of such models. One of the novel contributions of the present work lies in the introduction of a data-driven compensa tion strategy capable of accounting for multiple environmental influences and their potential nonlinear relationships with strain measurements. Unlike previous methods, the proposed approach can model each sensor’s response indi vidually, enhancing its adaptability and precision in handling real-world variability. Notably, none of the previously proposed SHM methodologies for masonry buildings have been validated on full-scale structures subjected to realistic environmental fluctuations and actual damage scenarios. The present study addresses this gap, providing a robust and scalable framework for EOV e ff ect compensation under complex, real-life conditions. Additionally, while data-driven approaches are particularly attractive for masonry SHM due to their non-intrusive nature, scalability, and automation potential Gentile et al. (2019); Masciotta and Pellegrini (2022), their ability to achieve higher levels of damage iden tification remains limited and is currently the subject of active investigation. Nonetheless, such capabilities become critically important following disruptive events, where the timely identification of damage location and severity can guide emergency response and prioritize structural interventions. To tackle these challenges, this work presents a comparative study of two innovative static SHM strategies tailored for historic masonry structures. The first is a fully data-driven approach, combining time-dependent multilayer per ceptron regressors with nonlinear cointegration theory to extract residuals that are robust to environmental e ff ects yet sensitive to structural changes. These residuals enable both global and sensor-level control charts, supporting damage detection, localization, and quantification via change-point analysis. The second is a model-based strategy grounded in model class selection theory. It employs finite element and surrogate models representing various damage scenar ios, framing damage identification as an inverse calibration problem. This enables the selection of the most probable damage scenario, along with estimates of its severity and location.