Time-varying mechanical properties tracking and predicting of in-service cable-stayed bridges
With the increase of service time, most of in-service cable-stayed bridges are aging and experiencing performance deterioration. Periodic field measurements are then conducted on such bridges to provide necessary data to numerically assess their mechanical properties. But due to the complexity of structures, the numerical models of cable-stayed bridges often differ from actual structures in service. Besides, the cable-stayed bridge is high statically indeterminate system, and the internal and external time-dependent factors will alter structural mechanical properties in the service phase. With the development of detection technology and structural health monitoring (SHM), the external factors can be accurately measured and analyzed. Because of the non-detectability of the internal factors, it is laborious to consider effect of the internal factors. So the target of the project is to study effect of the internal time-dependent factors, track time-varying mechanical properties, and predict the long-term properties of in-service cable-stayed bridges.
The internal time-dependent factors include the concrete shrinkage and creep, strands relaxation, cable damage, while the external time-dependent factors consist of the uneven foundation settlement and ambient temperature.
With the coupling effect of inherent numerical analysis errors and time-dependent factors, numerical results of long-term mechanical properties are often inconsistent with periodic on-site measurements of in-service cable-stayed bridges. Accurately tracking and predicting the time-varying mechanical properties of in-service cable-stayed bridges still remains extremely challenging.
An analytical method is proposed for static and dynamic characteristics of damaged cables. Damage (corrosion or broken wires) is equalized to the axial stiffness reduction in certain region according to microscopic law of damage. Governing equations for static configuration and in-plane free vibration of damaged cable are derived. Parametric studies are performed to study effect of damage on the static and dynamic characteristics of cables.
A numerical method is presented to track time-varying mechanical properties of in-service cable-stayed bridges with numerical analysis errors and concrete effect. The multi-objective optimization method is adopted to update the original numerical model of the cable-stayed bridge. The step-by-step finite element method is then employed to analyze the mechanical properties of the cable-stayed bridge with concrete time-dependent effect. The presented numerical method is applied to track the time-varying mechanical properties of Haihe Bridge (old) serving from 10 to 12 years under dead load. The feasibility and accuracy of the presented numerical method are verified by periodic field measurements.
A dynamic prediction algorithm is planned to be proposed for the long-term properties of in-service cable-stayed bridges. The uncertainties of internal and external time-dependent factors during the service of cable-stayed bridges will be firstly analyzed. Artificial neural network (ANN) is proved suitable for accurate long-term prediction because it can better contain the long-term information of the system. Besides, artificial neural network has strong self-adaptability, which can automatically adjust the network structure parameters to adapt the uncertainties of internal and external time-dependent factors. The combination of artificial neural network and time series prediction theory will then provide a new way for the dynamic prediction of the long-term properties of in-service cable-stayed bridges. (In progress)
Cooperation publications in other project