Utilizing the system instantaneous frequency for the Structural Health Monitoring of bridges
Neda Mostafa is a PhD student in the department Applied Mechanics & Data Analysis. (Co)Promotors are prof.dr.ir. T. Tinga, dr.ir. R. Loendersloot and dr. D. Di Maio from the faculty of Engineering Technology.
Bridges are continuously subjected to material aging, like corrosion of steel bars in reinforced concrete bridges or corrosion of steel structures and components. These processes combined with increasing traffic load result in overall deterioration and loss of structural integrity, which requires proper estimation of the maintenance needs to ensure safe operation of the asset. Vibration-based structural health monitoring is a non-destructive in-situ technology mainly based on performing three steps: 1) measure the dynamic response of the structure; 2) analyze the response; and 3) interpret the analyzed response for either system identification or damage detection. Structural Health Monitorin (SHM) then contributes to plan and perform maintenance or to provide early warnings of growing damage that can affect the safety and the bridge lifetime.
This thesis focuses on the time-frequency analysis of the vehicle-bridge dynamic interaction response to identify the time-dependent resonances of railway bridges which are incorporated into a damage detection approach, rather than only system identification. Most of the current system identification techniques applied to bridges are based on the free vibration response analysis. It is known that the bridge free vibration response is sensitive to environmental conditions such as temperature and it is not sufficiently sensitive to damage. Input-output modal analysis or output-only modal analysis are the other most used techniques for the bridge system identification. The train-bridge dynamic response, obtained during passage of the train is potentially more sensitive to damage, but also a more complex signal to analyze. First of all, it is a non-stationary signal that is not valid for modal analysis. In addition to the time-variant nature, the vehicle-bridge dynamic response can show closely-spaced spectral components response. These features disrupt the performance of the most advanced signal processing techniques. This thesis therefore applies a recently developed technique, Wavelet Synchrosqueezed Transform (WSST) to extract the Instantaneous Frequencies (IFs) of the Vehicle-Bridge Interaction (VBI) system response. A comparative study is performed on the various commonly used time-frequency analysis techniques. The obtained results were further validated using field measurements on a real bridge.
Subsequently, a concept for damage detection in (railway) bridges based on the instantaneous frequency analysis of the bridge's forced and free vibration responses is proposed. Within this concept, based on the bridge natural frequency extracted from the bridge free vibration, a healthy baseline is obtained of the bridge forced vibration response. The shape correlation and the magnitude variation are proposed to distinguish between the global characteristics of the bridge baseline induced by variable operational conditions and the local deviations caused by damage. If the source of the baseline deviation is damage, then the magnitude variation can be used as a damage index. The proposed damage index is a preliminary step toward damage quantification. Furthermore, the local deviation of the baseline instantaneous frequency around the damage location shows the potential of the proposed method for damage localization. However, damage localization is out of the scope of the current study.
The proposed Vehicle-Induced Delta Frequency (VIDF) quantifies the influence of the vehicle dynamics on the response of the intact bridge. The Damage-Induced Delta Frequency (DIDF), as a damage sensitive feature, quantifies the influence of the vehicle dynamics on damage detection. The final objective of this study was to investigate the effectiveness of the proposed damage sensitive feature using different train types, specifically freight trains and passenger trains. Therefore, two vehicle models were employed to calculate VIDF and DIDF. The results of the numerical studies show that trains with single suspension systems cause more pronounced changes in the bridge's frequency response, specifically the Vehicle-Induced Delta Frequency (VIDF) and Damage-Induced Delta Frequency (DIDF), than dual suspension trains. This characteristic indicates that single suspension trains are better suited for efficient bridge health monitoring and damage detection.
The work in this thesis shows a methodology to use the non-stationary dynamic response of a bridge passing event of a train for structural health monitoring purposes. The method is applicable for relatively low-speed train passages (no high speed lines) and identifies a number of key points to take into account when implementing such a monitoring system. The signal processing technique is of great importance, but also the type of train passing the bridge is an elementary part of successful damage identification. These insights form the base for guidelines to design a monitoring system using the dynamic response to a train passing a bridge.
