Abstract:
This thesis primarily addresses the variation in identified modal parameters of bridges from traffic-induced vibration measurements. Effects of traffic loading on this variation are aimed to be analyzed. A typical mid-sized post-tensioned highway bridge, the Jamboree Road Overcrossing (JRO) that is located in Irvine, CA, US was used as a case study. The JRO was monitored over a five-year period and 1707 traffic-induced vibration data sets were collected through 13 acceleration sensors. First of all, modal parameters and structural properties are identified by using the Frequency Domain Decomposition (FDD) method and radial basis neural networks, respectively. Variations up to 10% in modal frequencies and 6% in structural properties are observed under the same environmental conditions over the same period. Then, in order to investigate the reasons of the variations, three different finite element analyses are carried out. While static mass and moving load analyses are conducted by the initial finite element model of the JRO, correlated white noise analysis is conducted by a single span finite element beam model. As a conclusion, it is observed from the results of static mass analysis that additional masses have significant contributions in modal mass of the system. Increase in the modal mass, then, results in decrease in natural frequencies. Numbers, locations and quantities of the additional masses are the determinative factors of this decrease. Furthermore, responses of the moving load and correlated white noise analyses are obtained. Afterwards, modal parameters of the models are identified from these responses by using FDD method. As a result, speed of the moving load causes significant variations in the identified natural frequencies. It is also concluded that correlated white noise results in suppressed modes, pseudo peaks and shifts in the identified natural frequencies.
