USE OF STOCHASTIC SUBSPACE IDENTIFICATION MEHODS FORPOST-DISASTER CONDITION ASSESSMENT OF HIGHWAY BRIDGES

摘要

After the occurrence of a major earthquake, reliable methods are required for post-disaster conditionassessment of highway bridge structures. Structural health monitoring techniques based on field vibrationmonitoring can be applied to post-earthquake investigations of structural integrity. The systemidentification techniques applied to structural testing in laboratory environment are typically based onimpact-load or other forced vibration excitation techniques to obtain information on the dynamicproperties of the structure. In this case, the characteristics of the applied forces are known, and thus theresponses of the test structure can be correlated with the input excitations. For large-scale structures, suchas major bridges, the use of a forced vibration testing technique in the field is very often impractical andexpensive to undertake. Ambient vibration responses due to transient dynamic load effects, such as, trafficloads, wind, and earthquakes, may be used instead in this case. Special numerical techniques of systemidentification are needed to analyze the ambient vibration data obtained by field monitoring systemsbecause of the lack of information on the input excitation. Previous studies have shown that the stochasticsubspace identification (SSI) methods are one of the most robust output-only identification techniques forcivil engineering applications. In this paper, the SSI method is presented in detail along with case studyresults using field-monitoring data from the Confederation Bridge in Canada. The results of the casestudies show that there is significant variability in the dynamic properties of the structure extracted fromdifferent datasets collected at different times under different loading scenarios and/or differentenvironmental conditions. The variability in the extracted structural parameters represents a challenge forstructural condition assessment algorithms based on detecting changes in the structural static or dynamicproperties. Observed changes in the structural properties may be due to one or a combination of thefollowing effects: (1) stiffness degradation due to deterioration or damage; (2) environmental effects suchas temperature variation; (3) differences in the loading scenarios; (4) computational inaccuracies andmodeling assumptions. In order to perform accurate and reliable post-disaster condition assessments athorough understanding of these effects is needed. Studies have been conducted to evaluate the influences