PSI - Issue 84

Enrico Pasquale Zitiello et al. / Procedia Structural Integrity 84 (2026) 360–367

361

of the infrastructure heritage, yet they are often characterized by an advanced state of deterioration due to their long service life, increasing operational stresses, and limited financial resources for maintenance. Most of these structures, built during the post-war reconstruction period, have now exceeded 50–60 years of service, approaching or reaching the end of the expected useful life of reinforced concrete structures. Adding to this scenario, in several national contexts, are structural deficiencies in existing infrastructure management and monitoring systems, which hinder a systematic assessment of the state of conservation of these structures. The lack of comprehensive and up-to-date knowledge of the infrastructure heritage, combined with the intrinsic limitations of traditional inspection methodologies, based primarily on visual inspections and periodic interventions, highlights the need for a paradigm shift towards more preventative, continuous, and data-driven approaches. In this context, the adoption of structural monitoring systems based on sensors and ICT technologies emerges as an effective solution to support advanced maintenance strategies. The digitalization of the built environment, through the integration of Building Information Modeling (BIM), the Internet of Things (IoT), and Machine Learning (ML) techniques, enables the development of dynamic digital models, known as Digital Twins, capable of representing the structural behavior of structures in near real time. The unification of static data from information models with dynamic data from sensors enables continuous monitoring of structural performance, early detection of anomalies, and the activation of alarm systems based on safety thresholds. The evolution of these systems towards predictive models, powered by machine learning algorithms, also opens new perspectives for optimizing maintenance policies, reducing reliance on widespread inspections and improving the efficiency of resource allocation. The evolution of these systems towards predictive models, powered by machine learning algorithms, opens new perspectives for optimizing maintenance policies, reducing reliance on widespread inspections and improving resource allocation efficiency. However, the extensive implementation of advanced monitoring systems across a country's entire infrastructure assets would currently be technically complex and costly. A more sustainable and effective approach involves the targeted selection of strategic or representative infrastructure assets for continuous monitoring to characterize their structural behavior throughout their entire life cycle. The information thus acquired allows for the identification of degradation patterns, recurring damage mechanisms, and significant performance indicators, which can then be extended, through generalization and classification processes, to infrastructures similar in terms of structural type, materials, construction period, and operating conditions. This approach allows the benefits of advanced structural monitoring to be scaled to the network level, supporting predictive and risk-based maintenance strategies even in the absence of widespread sensor systems across all assets. This work presents an integrated methodology for bridge monitoring and maintenance based on the use of structural sensors and Digital Twins, applied to two case studies of bridges in operation in Italy, currently undergoing continuous monitoring. 2. State of art To rigorously define the concept of a Digital Twin in engineering, it is necessary to move beyond the reductive view that equates it to simple three-dimensional modeling. The Digital Twin must be interpreted as a complex technological ecosystem, consisting of interconnected hardware and software components, capable of transforming raw data from the actual work into useful information for decision-making through a closed-loop system. From this perspective, the digital twin does not represent a static snapshot of the physical asset, but a dynamic and predictive platform, characterized by a continuous information flow between the real world and the virtual model. According to the framework proposed by Wimmer and Braml (2024) (1), the technical architecture of the Digital Twin is based on a multi-level system in which the physical asset, the data management infrastructure, and the virtual model operate in an integrated manner. At the core is the physical layer, consisting of the structure and a network of monitoring devices. In the case of bridge infrastructure, this includes structural, environmental, and dynamic sensors capable of detecting deformations, vibrations, thermal variations, and other significant parameters for assessing the structure's state of health(2). This component is complemented by high-precision geometric survey techniques, such as laser scanning and photogrammetry, which enable the generation of point clouds and three-dimensional models faithful to the actual geometry, ensuring dimensional consistency between the physical asset and its digital counterpart (3). The second