Dynamic identification and model updating of full-scale concrete box girders based on the experimental torsional response

The authors examined the dynamic behaviour of seven concrete box-girders of the A24 Italian motorway. The bridge spans derive from the same design and are nominally identical. However, they manifest a varied dynamic response, imputable to several factors. Some discrepancies may originate from the construction time, due to geometrical irregularities and unlike concrete curing. Others may descend from the time-degradation, soil-substructures interactions and other modifications in the boundary restraints, like the substitution of the original rack and roller bearings with multi-directional pots. The first bending mode shape is negligibly affected by the specific bearing, and the corresponding natural frequency has a marked dependence on the values of the Elastic Modulus, obtained from static load tests. The Least-Squares fit between the Elastic Modulus from static load tests and the first natural frequency matches with the corresponding expression of a simply-supported beam model. Conversely, the second mode shape, which is mainly torsional, is likely to be more sensitive to the bearings. The authors endeavored to estimate the values of two equivalent torsional springs, representatives of the actual restraint conditions by maximizing the Modal Assurance Criterion (MAC) between analytical and experimental mode shapes. The analytical mode shape is the solution of the equation of motion for free torsional vibration, with warping stiffness and warping inertia both neglected. The obtained values reported as the ratio between torsional stiffness of the boundary spring and the torsional stiffness of the beam cross-section yield the estimate of the vertical restraint offered by the two different typologies of bearings: the rack and roller and the multi-directional pot. The study encompasses the analysis of the outcomes of the dynamic identification and an elementary model updating.