Vibrational monitoring of a full-scale highway bridge undergoing a destructive test

This research examines the vibrational response of a full-scale highway bridge before and after the introduction of a simulated crack in one of the three supporting steel girders. The simulated cracking of the girder was carried out in five stages beginning with the cutting of the entire bottom flange at the mid-span of one fascia girder. Subsequent stages increased the severity of the crack by continuing the cutting of the girder through the web in six-inch increments.^ The overall goal of this work was to determine what processing techniques and response components could be used to form the basis of a continuous monitoring system for in-service highway bridges. The system would detect a change in the stiffness of the structure that could present a severe hazard or even collapse.^ A large portion of the study was experimental, and involved the field testing of a portion of a highway bridge that was being replaced with another structure at the same location. A minimum of 10 passes were made with a full-size pick-up truck during each stage of the test to simulate traffic passing over the bridge.^ The vibrational response of the bridge was measured at eight locations on the structure for each vehicle pass and recorded for analysis. A Frequency Response Spectrum (FRS) was produced from the results of each vehicle pass. An average FRS was also produced. The location of peaks within the average and individual FRS and their amplitudes were compared between the different stages. The values of the Signature Assurance Criterion, which utilize the response of the structure at specific frequencies, and the Cross Signature Assurance Criterion, which utilize the response at specific locations, clearly indicate a change in stiffness.^ A finite element model was developed and calibrated using the experimental results from the unaltered structure. Reduction in the stiffness of the primary beam elements resulted in good correlation with the experimental results to a certain degree. Changes in the torsional frequency range were not duplicated with the model.^ Recommendations for further research are discussed.