PSI - Issue 78

Domenico Cefalì et al. / Procedia Structural Integrity 78 (2026) 1358–1365

1360

2. Bridge Health Index: a review of methodologies and implementations The Bridge Health Index serves as a vital tool for assessing bridge functionality and structural integrity, guiding crucial decisions in maintenance, rehabilitation, and replacement activities. It provides a comprehensive measure of a bridge's overall health based on the condition of its structural elements and the services it renders, with the escalating number of aging bridges worldwide emphasizing the need for robust BHI methodologies to ensure infrastructure safety and longevity. Methodologies for BHI development span from traditional inspection-based approaches to advanced data-driven models incorporating Artificial Intelligence (AI) and sensor technologies. Traditional condition rating, which relies on qualitative visual inspections, often introduces subjectivity; to counter this, techniques like fuzzy logic and non-destructive evaluation are explored (Faris et al., 2025). The aggregation of detected defects leads to an overall BCI or BHI, while deterioration models track changes in condition over time for efficient planning. Hybrid deterioration models, integrating AI with stochastic and physics-based methods, show promise for enhanced predictions (Faris et al., 2025). A significant challenge in BHI implementation is the scarcity of objective, high-quality data, often relying on expert opinions (Jeon et al., 2024a, 2024b). Novel health indices for quantitative evaluation use deterioration models from historical data and Long Short-Term Memory (LSTM) neural networks to model time deterioration relationships, improving efficiency, accuracy, and long-term prediction (Jeon et al., 2024a, 2024b). For steel bridge decks, a BHI can be determined by considering deterioration mechanisms based on statistical analysis of inspection data (Otake et al., 2015, 2012). A Weighted Priority Index (PI) approach specifically assesses maintenance prioritization, crucial for managing aging infrastructure where collapses highlight urgent maintenance needs (Lee et al., 2025). The Analytical Hierarchy Process (AHP) prioritizes concrete bridges using eight indices (structural, hydrology/climate, safety, load impact, geotechnical/seismicity, strategic importance, facilities, and traffic/pavement), with expert-assigned scores and AHP-determined weights identifying bridges most in need (Darban et al., 2021, 2020). A reliability-based methodology for aging bridge health monitoring proposes critical failure criteria, defines safety margins, assumes normally distributed random variables for failure quantities, and estimates elementary reliability indices and failure probabilities, leading to a system reliability index (Dissanayake and Karunananda, 2008). The Pontis bridge management system utilizes BHI for project and network level evaluation, investigating linear/non-linear scales for condition state weights, amplification weights, and element weights based on replacement costs, long-term costs, and hazard vulnerability (Inkoom et al., 2017; Inkoom and Sobanjo, 2018). However, the current Pontis BHI, as applied in networks like Denver, can be subjective due to reliance on imprecise cost data and strong correlation between cost and condition, suggesting modifications (Jiang and Rens, 2010a). An alternative, the Denver BHI, was developed to reflect element damage's effect on bridge health and function rather than just value (Jiang and Rens, 2010b). For highway bridges, a Global Flexibility Index (GFI), the spectral norm of the modal flexibility matrix from dynamic responses, can infer health deterioration; a sharp increase indicates further investigation is needed (Patjawit and Kanok-Nukulchai, 2005). Effective BHI implementation relies on advanced data collection, signal processing, and analytical techniques. Nanosensors are emerging tools for SHM, measuring physical attributes and environmental variables at the nanoscale, promising improved maintenance despite data interpretation, power, and signal processing challenges (Adarsan et al., 2025). Comprehensive measurement techniques for highway bridges include PTS-C10 crack width and DJCS-05 crack depth meters, and strain tests to analyze strain-load relationships (Zhou et al., 2020). Vibration responses, typically accelerometers, offer insights, necessitating advanced processing like Empirical Mode Decomposition (EMD) and Wavelet Packet Decomposition (Ramadhan et al., 2025). Finite Element Analysis (FEA) simulates and predicts vibration responses, especially given large structures and limited sensors; integrating FEA with signal processing provides reliable tools (Ramadhan et al., 2025). A SHM system for railway bridge substructures can use a soundness diagnosis index correlated with natural frequency (Abe et al., 2016). A novel damage index for railway bridges, based on synchronous strain and displacement data during train passages, identifies a transformation operator for converting strains into displacements without prior knowledge. The displacement prediction error serves as a robust damage index, insensitive to vehicle loads and temperature (Quqa et al., 2024). AI algorithms form the basis of modern Predictive Maintenance (PdM), crucial for accurate SHM. SHM leverages a limited number of sensors on critical components, with AI performing complex statistical analysis using Machine Learning and Deep Learning to model behavior, detect trends, and predict failures, enhancing reliability (Bianchi et al., 2025). Another implementation is an HI-based bridge evaluation framework for prestressed concrete bridges, integrating Building Information Modeling (BIM) for inspection data and LSTM for deterioration prediction (Jeon et al., 2024a). This approach boosts evaluation