Damage assessment of bridges with jacketed RC columns using vibration test

This study develops a method of detecting the damage of jacketed RC columns measured in terms of stiffness degradation by horizontally exciting the damaged bridge and by identifying the change in its vibration characteristics including frequencies and mode shapes from those of the undamaged bridge. The method combines neural network techniques with finite element analysis to establish a reliable relationship, a well trained neural network, between the vibration characteristics and the stiffness degradation. The unknown stiffness degradation is then identified by inputting the vibration characteristics obtained from the vibration tests into the neural network. This method was verified in an earlier study by the present authors' on the basis of experiments performed on two half-scale bridge columns wrapped respectively with carbon and glass fiber jackets. The present study validates the usefulness of this method for the assessment of column damage on the basis of the vibration test involving an entire bridge. Specifically, the validation is carried out by comparing damage scenario with the stiffness degradation computed from the vibration characteristics of a three-span bridge model with jacketed columns. It is demonstrated that the knowledge of the first mode of the bridge induced by the horizontal vibration test is sufficient for the identification of the stiffness degradation at jacketed columns. This fact indicates that the proposed damage assessment method of combining horizontal vibration tests with neural network-based identification is practical and effective because of the ease with which the first mode vibration can be induced in the field.