PSI - Issue 84

Antonio S. López-Cuervo et al. / Procedia Structural Integrity 84 (2026) 223–230

224

1. Introduction Structural Health Monitoring (SHM) has gained increasing relevance in both the scientific community and industry, with a growing number of bridges being equipped with dense monitoring systems. Operational Modal Analysis (OMA) based on acceleration data is one of the most widely implemented techniques, due to its non-intrusiveness and relatively easy automation for extracting modal parameters (i.e., resonant frequencies, mode shapes, and damping ratios) as damage-sensitive features (Rainieri and Fabbrocino, 2014; García-Macías and Ubertini, 2022). However, despite its wide use, the low sensitivity of modal characteristics extracted by acceleration-based OMA to local defects has been extensively reported in the literature (Limongelli, 2019; Anastasopoulos and Reynders, 2025). Conversely, strain-based OMA has attracted growing interest because strain mode shapes are inherently more sensitive to local stiffness changes than displacement mode shapes (Reynders et al. , 2007; Kranjc et al. , 2016). Notable full-scale deployments have been reported by Anastasopoulos and co-authors, who instrumented multiple railway bridges in Belgium with dense long-gauge FBG networks, enabling the identification of strain modes under in-service excitation. For instance, the KW51 steel bowstring bridge in Leuven was instrumented with an 80-FBG array, enabling continuous tracking of ten modes over a one-year campaign (Anastasopoulos et al., 2021). More recently, on the 110 year-old Nieuwebrugstraat railway bridge in Ronse, one span was monitored with 80 FBG sensors, consistently identifying four strain mode shapes over long-term measurements (Anastasopoulos and Reynders, 2023). Despite these successful applications of FBG-based sensing systems, the high-cost interrogator and the installation complexity of the fibers, among others, have prevented widespread adoption of these systems in bridge monitoring. As an alternative to fiber-optic sensing, conventional resistive SGs can be employed for dynamic strain monitoring and strain-based modal identification. In practice, however, their use has been largely confined to Experimental Modal Analysis (EMA) of machinery and small-scale structures, as ambient strain amplitudes in civil structures are often too low to provide sufficient signal-to-noise ratio. The objective of this study is to demonstrate the feasibility of strain based OMA on a real-world civil structure using a low-cost SG-based acquisition system. To this end, a low-cost DAQ is implemented, and two AVTs are presented from two optimized SGs layouts. The identified modal properties from these tests are benchmarked against conventional acceleration-based OMA and FEM predictions. The linear independence of the identified mode shapes is also assessed. The case study is a steel pedestrian footbridge located in Granada (Spain), spanning the Bobadilla–Granada railway line (Fig. 1(a)). The main span consists of a Warren steel truss, comprising two vertical planes of tubular diagonals and a transverse bracing system. Access to the truss is provided by ramps and stairways located on both sides. The superstructure is made of A42 steel with welded connections, whereas reinforced concrete (C25/30) is used for the deck pavement and foundations. An FEM was developed in SAP2000 (Fig. 1(c)), based on the original construction project (Socran Ingenieros S.L., 2002). The superstructure was modeled using beam elements for the diagonals, main girders, and bracing members, and plate elements for the deck. Mass contributions from non-modeled components, such as steel gusset plates, handrails, and connection details, were included. Boundary conditions were idealized as fixed supports at the connections to the foundations. The FEM-based modal analysis provided the input for the EfI optimal sensor placement algorithm and served as a baseline for comparison with the experimentally identified strain mode shapes. 2.2. Experimental setup and data acquisition Four AVTs were conducted: two accelerometer-based campaigns (AVT-Acc-1 on 18 November 2024 and AVT Acc-2 on 10 December 2024) and two strain-gauge-based campaigns (AVT-SG-1 on 18 December 2024 and AVT SG-2 on 7 October 2025). The study was restricted to the main Warren truss. For the accelerometer-based AVTs, a commercial National Instruments acquisition system was used, consisting of a cDAQ-9184 chassis and three NI9230 modules (3 channels, 24-bit). PCB KB12VD accelerometers with a sensitivity of 10 V/g were employed, as shown in 2. Methodology 2.1. Case study