Damage detection is the process of identifying the presence, location, and severity of damage in a structure using measurements of its response or properties. It is the primary objective of structural health monitoring and is applied across a wide range of civil, mechanical, and aerospace systems.
Damage detection is the identification of changes in a structure's material or geometric properties that adversely affect its performance. In structural health monitoring (SHM), damage detection is typically the first level in a four-level hierarchy (detection, localization, quantification, prognosis). The fundamental principle is that damage β whether it is cracking, corrosion, delamination, yielding, or loss of connection β alters the stiffness, mass, or damping of the structure, and these alterations manifest as changes in the measured response. Detection methods vary widely depending on the type of damage expected, the accessibility of the structure, and the required level of confidence.
Damage detection approaches are commonly classified into local and global methods. Local methods examine a specific region of the structure for damage using targeted measurements β such as visual inspection, ultrasonic testing, acoustic emission, or strain measurements. These methods are highly sensitive but require access to the damage location and are labor-intensive for large structures. Global methods examine the entire structure for evidence of damage using measurements from a distributed sensor network β typically vibration-based. These methods can detect damage without prior knowledge of its location, but they are generally less sensitive to small or localized damage. Vibration-based damage detection relies on the fact that damage reduces structural stiffness, which changes natural frequencies, mode shapes, and damping. The change in these modal properties is used as a damage indicator, though the relationship between damage and modal change is often complex and affected by environmental and operational variability.
The central challenge in damage detection is distinguishing damage-induced changes from those caused by environmental and operational variability β temperature, humidity, wind, traffic, and occupancy. These sources of variability can produce changes in modal properties that are comparable to or larger than those caused by early-stage damage, leading to false positives and missed detections. Modern damage detection methods increasingly use statistical approaches β such as outlier analysis, novelty detection, and machine learning classifiers β to address this challenge. Methods are typically trained on baseline data from the healthy structure, and deviations from this baseline are flagged as potential damage. Advanced methods include guided wave techniques, electromechanical impedance, fiber optic sensing, and computer vision-based crack detection. In earthquake engineering, damage detection is particularly valuable for post-earthquake assessment, where rapid identification of damaged structures enables efficient allocation of inspection resources and timely decisions about occupancy, repair, and demolition. In Iran, damage detection for post-earthquake assessment is an emerging practice, with growing interest in deploying monitoring systems on bridges, critical buildings, and lifeline infrastructure. The integration of damage detection with earthquake early warning systems is an active area of research and development.