Neutral Axis-Based Anomaly Detection for Local Damage to Girders Affected by Nonlinear Temperature Gradients

Abstract

Damage to girders threaten a bridge’s safety. They can be effectively detected by the neutral axis (NA) position of a girder. However, the NA position is inevitably affected by the nonlinear temperature gradient of girders, which could mask the damage-induced abnormality. Aiming at the practicality of the NA position, an improved thermal-varying NA model was scientifically derived by accommodating vertically distributed measurements of a nonlinear temperature gradient, followed by several simplified models. By eliminating the effect of the nonlinear temperature gradient on the NA position to the maximum, an abnormality alarming framework was then established for damage detection of girders. Finally, a girder model imposed by an actual nonlinear temperature gradient was developed to validate the advantages of the improved model and the effectiveness of the alarming framework. The proposed models are an improvement over the existing model in terms of prediction accuracy and stability. Additionally, the NA-based alarming method can effectively detect minor damage to the girder subjected to varying nonlinear temperature gradients. The optimal strain measurement is recommended at the lowest turning point of the temperature difference characteristic above the NA position, and the optimal temperature measurement should effectively represent the statistical outline feature of the temperature difference.