Modelling strategies for structural health monitoring of cable-stayed bridges

Abstract

Long-span cable-stayed bridges have gained increasing popularity due to their appealing aesthetics, increased stiffness compared to the suspension bridges, and relatively small size of bridge components. For cable-stayed bridges with steel girders in particular, fatigue has been one of the most important failure modes and the prediction of the remaining fatigue lives has been the primary focus in recent years. To enable the evaluation of fatigue performance, a reasonable baseline finite element model reflecting the real structural behaviour is indispensable. This thesis describes the establishment of baseline finite element models of cable-stayed bridges for fatigue analysis considering the mean stress effect. Ting Kau Bridge in Hong Kong is chosen as a real life example to demonstrate the practical application of the methods developed. Firstly, the initial cable forces of the finite element model of the cable-stayed bridge are calibrated to ensure that the initial geometry of the finite element model under permanent loading agrees with that specified on the as-built drawings within reasonable tolerance. The traditional cable force adjustment method is often a trial-and-error process which is empirical, time-consuming and occasionally difficult in convergence. An optimization method based on the Kriging surrogate model is therefore developed in order to establish the relation between the deck geometry and he initial cable forces of the cable-stayed bridge. The efficiency of the proposed approach is further verified using Ting Kau Bridge against the traditional method. Furthermore, the initial finite element model developed on the basis of engineering blueprints is updated by modifying the uncertain parameters so as to achieve a refined model using the Kriging predictor. An objective function can be formulated in terms of the discrepancy between the theoretical and measured responses captured by the Wind and Structural Health Monitoring System (WASHMS) installed on the bridge. A novel feature of the proposed method is that it enables the simultaneous use of static load testing data and the dynamic information, which is not feasible by using the conventional sensitivity-based approaches. Finally, the fatigue lives of selected critical components of Ting Kau Bridge are worked out with different methods including the deterministic and probabilistic approaches based on the baseline finite element model. In the deterministic approach, the equivalent annual standard fatigue vehicle spectrum is built based on the WASHMS measurements to estimate the damage accumulation and predict fatigue lives. In the probabilistic approach, a probabilistic loading model is proposed to simulate the vehicles running along the bridge by using the available information collected by the weigh-in-motion system. Combined with the traffic load model, a statistical approach is employed to obtain the stress time-histories, and the probability density curve of the fatigue life can be obtained. One significant advantage of the proposed method is that the diurnal variation of traffic flow within different time intervals can be accounted for properly. The results of these approaches are compared with one another. As the mean stress effect is significant in the estimation of fatigue lives, the baseline calibration of cable-stayed bridges is considered essential.