ABSTRACT
Most long-span bridges are now incorporating a Structural Health Monitoring System (SHMS) within the technical specifications of their elicitation process.
However, there are limited opportunities to test the effectiveness of a data processing scheme on a real damage scenario. There are some study cases of instrumented full scale bridges where several damage scenarios are inflicted (Farrar et al. 1994, Maeck & Roeck 2003). In these studies, the damage scenarios are controlled and a shaker is used as an excitation source. The data sets from these cases allows testing the effectiveness of different data processing schemes for damage characterization. The excitation source and the ambient condition differ from the most realistic case, where a bridge suffers progressive damage over a long period of time, subjected to normal operational loads and ambient variations.
In this research, vibration data of a real bridge that suffered from severe structural damage is analyzed, and a data processing scheme for locating damage is proposed and considerations about the design of the instrumentation system for encouraging subsequent diagnosis is discussed.
Damage diagnosis in a SHMS is defined by increasing levels of specificity, considering: damage detection, localization, severity, damage type, and eventually damage forecast. This paper presents a methodology for analyzing accelerometers data installed on a full-scale suspension bridge. Applying dynamic identification techniques (Frequency Domain Decomposition technique), fundamental frequencies and mode shapes of the structure can be obtained and used as features for further physics based statistical analysis. Complementing sequential damage detection and damage localization algorithms one can achieve a high detection and localization accuracy. In this work, 4 different techniques based on modal shapes were used to locate damage. These techniques are: Analysis of the curvature modal shape changes, change in the flexibility of the structure (CIF; Pandey & Biswas 1994), modal strain energy also known as modified damage index (MDI; Stubbs et al. 1995; Choi et al. 2008); and hybrid of MDI and CIF was performed (HMC; Choi et al. 2008). Each technique gave results of the locations more susceptible to concentrate damage.
The damage localization methods based on vibrations are very useful when used together with dynamic parameters obtained empirically. The curvature modal shapes where outperformed as damage locators by CIF, MDI and HMC. The later was the best damage location indicator.
Lastly, this case shows that sensors configuration in a structure must be designed foreseeing the most likely failure scenarios to be detected and located. In this case, having accelerometers near the pillars supporting the main deck would have improved the damage detection and location of the SHMS.
