Editorial: How AI and digitalisation transform bridge engineering, from modelling and monitoring to reliability and resilience Free

Bridge engineering is entering a new era shaped by the integration of machine learning (ML), advanced sensing and computational modelling. Traditionally reliant on simplified analytical models and periodic inspections, the field is now evolving toward data-driven, continuously updated and predictive frameworks. The seven papers in this issue collectively capture this transition. They span the full life cycle of bridge engineering – from form-finding and aerodynamic design, through digital measurement and intelligent monitoring, to load modelling and time-dependent reliability – demonstrating how digital technologies are reshaping both practice and theory, while recent studies in this journal also highlight the growing importance of resilience under flood and extreme wave hazards (Beaver and Mitoulis, 2026; Li et al., 2026), further reinforcing the need for integrated, data-informed approaches.

A fundamental shift in bridge engineering lies in how design is conceived – no longer as a static, code-driven exercise, but as a process increasingly informed by simulation and real-world data. Tang et al. (2026) present a refined rigid-connection-based approach to shape-finding in steel truss suspension bridges, improving force distribution and structural efficiency, enabling more rational configuration under construction constraints. In parallel, Duranovic et al. (2026) demonstrate how computational fluid dynamics can move beyond a specialist research tool to become a practical component of early-stage design, offering a cost-effective alternative to wind tunnel testing while capturing key aerodynamic coefficients and vortex-induced vibration behaviour. By showing that such analyses can be performed within standard engineering workflows, their work enables faster and more informed aerodynamic decision making. These contributions illustrate how computational methods are extending beyond analysis toward active design exploration, enabling engineers to interrogate both structural form and aerodynamic performance early in the design process.

As monitoring capabilities improve, there is a parallel need to reassess the assumptions underpinning structural loading and analysis. Hernandez-Martínez et al. (2026) address this challenge by comparing codified load models with weigh-in-motion traffic data. Through the introduction of exceedance rate spectra, the study exposes systematic discrepancies in predicted support reactions, questioning the continued adequacy of conventional load models. The findings point toward a necessary recalibration of design assumptions to reflect evolving traffic patterns and highlight the growing role of data in closing the loop between design and observed structural behaviour.

A critical enabler of digital transformation is the ability to capture accurate, high-resolution data on bridge geometry and behaviour. Wu et al. (2026) demonstrate how point cloud data can transform alignment assessment, marking a decisive shift away from sparse, discrete measurements toward dense, high-resolution structural capture. By processing laser-scanned data to derive precise geometric alignment, their approach not only improves efficiency but also enables continuous tracking of deformation and construction tolerances. This capability aligns closely with the emergence of digital twins, as explored by Xu et al. (2026), who evaluate sensor layout strategies through integrated finite-element modelling and monitoring frameworks for two cable-stayed bridges (Figure 1). Their work shows how optimised sensor placement can enhance damage detectability while reducing redundancy in instrumentation, offering a more strategic approach to structural health monitoring. Together, these contributions point toward a new paradigm in which bridges are no longer represented through isolated datasets, but as integrated digital systems, combining geometry, behaviour and sensing within a unified and continuously updated framework.

Inspection and monitoring are undergoing one of the most visible transformations through ML and remote sensing technologies. Do et al. (2026) demonstrate how the integration of unmanned aerial vehicles and deep learning can enable automated damage detection, with fully convolutional networks capable of simultaneously classifying bridge components and identifying defects with high accuracy. By leveraging fused datasets, their framework improves robustness and generalisation, making large-scale, repeatable inspection both feasible and practical, particularly in hard-to-access areas. In parallel, Xu et al. (2026) show how monitoring systems themselves can be optimised through digital twin-based approaches. By coupling finite-element models with sensing strategies, their work enables more effective sensor placement, enhancing damage detectability while avoiding unnecessary redundancy. These advances point toward a shift from episodic, labour-intensive inspection to continuous, data-driven condition assessment. While this transformation does not diminish the role of engineering judgement, it fundamentally redefines it – moving away from manual data collection toward interpretation, validation and informed decision making.

As monitoring capabilities expand, there is a parallel need to rethink how structural performance is assessed over time. Kim et al. (2026) introduce a durability design framework based on time-dependent reliability, integrating resistance degradation, traffic growth and environmental effects throughout the service life. By incorporating real inspection data and simplified reliability formulations, their study offers a practical pathway for implementing life-cycle-oriented design. This perspective aligns with the variability in loading conditions identified by Hernandez-Martínez et al. (2026), the increasing availability of monitoring data from digital systems (Xu et al., 2026) and recent work on flood resilience and hydrodynamic loading (Beaver and Mitoulis, 2026; Li et al., 2026). These contributions point toward a shift from static safety factors to dynamic, continuously updated reliability assessment, where structural performance is treated as an evolving condition rather than a fixed design outcome.

The developments presented in this issue are not isolated innovations, but components of a broader transformation toward an integrated engineering paradigm. Simulation is informing design, sensing is redefining observation, machine learning is automating interpretation and reliability frameworks are structuring these inputs into actionable insights. The challenge now lies not in developing new technologies, but in integrating them into coherent workflows and embedding them within engineering practice and standards. This requires a shift in mindset: from deterministic to probabilistic thinking, from periodic inspection to continuous monitoring, and from reactive intervention to predictive management. Bridge engineering has always balanced innovation with caution. As these technologies continue to mature, their effective adoption will depend on how well they are incorporated into design methodologies, assessment frameworks and professional practice. The tools to deliver smarter, safer and more resilient infrastructure are already available; the key question is how they can be implemented at scale to realise their full potential.

For further reading on related topics, readers are referred to recent papers in this journal on differential interferometric synthetic aperture radar (DInSAR)-based structural monitoring (Di Carlo et al., 2025), machine learning models for bridge response prediction (Ha et al., 2025), large-scale monitoring systems such as the Queensferry crossing (Cousins et al., 2025), the themed issue on intelligent asset management (Christodoulou, 2024) and data-driven condition assessment approaches (Moutsianos et al., 2026).