Vibration based performance assessment of concrete-concrete composite bridges

Includes bibliographical references (leaves 142-148).Concrete composites consisting of precast pre-stressed standardized beams and a cast in-situ deck slab have been used for the construction of short to medium span bridges for the past four decades in South Africa and worldwide. The pre-cast beams and cast in-situ slab are commonly connected using shear connectors. Failure of these connectors would compromise the composite action of the structure, thus reducing the load carrying capacity and hence its efficiency. This study seeks to assess the integrity of such shear connectors using dynamic testing and Finite Element (FE) analysis. The main objective of the work is to assess the practicality of vibration-based techniques to detect damaged shear connectors using experimental and analytical modal data. A scaled bridge model was constructed and 10 mm bolts connected the beams and slab to simulate shear connectors in the prototype bridge. Different damage scenarios were introduced by loosening some of the connectors and vibration testing was done to detect the artificial damage. An FE model of the system was also developed. The shear connectors were modelled as non-linear spring elements capable of simulating the composite action between the slab and beams. Damage of shear connectors was simulated by reducing the spring stiffness. The updating of the FE model was done manually by adjusting appropriate spring stiffnesses. The experimental and analytical results show that the natural frequencies are sensitive to this damage. The frequencies dropped from undamaged to severe damaged structure. Very little information was deduced from the damping ratios, modal assurance criteria (MAC) and coordinate modal assurance criteria (COMAC) values. The experimental and analytical first bending, torsion and transverse modes were sensitive to the damaged shear connectors. 65% of damaged connectors were located using these modes. Using experimental modal data, the mode curvatures and flexibility changes were able to locate the damaged region when more than 35% of shear connectors were loosened. However, using numerical data, the mode curvatures and flexibility changes were able to localize the damaged region for 6% damage introduced. 75% of the loosened connectors were identified. The stiffness change technique could only identify less than 10% of damaged shear connectors using experimental modal data. The same technique was applied on analytical data and over 75% of damaged shear connectors were located. The FE modelling of shear connectors used in this work was applied on an existing bridge. Van der Kloof bridge (South Africa) was constructed using precast pre-stressed beams and a cast in-situ slab. Extended beam web stirrups were used as shear connectors. The main aim was to develop a robust FE model for this bridge that could be used in future to investigate the condition of shear connectors. Using 6-0 non-linear spring elements to model the shear connectors, a maximum difference of 0.98% was observed between the measured and theoretical frequencies after manual updating. This is quite a small difference. This model could therefore be used as a true representative of the physical structure for future investigations