Life-Cycle Fatigue Performance of Coastal Slender Bridges Subject to Multi-Hazards

Serving as critical links in the transportation network for coastal regions, costal slender bridges could constantly experience complex dynamic interactions with strong winds and/or high waves during extreme weather conditions, in addition to moving vehicles, such as cars, trucks, or trains. Continuously repeated stress cycles as well as corrosive coastal environments could cause significant fatigue damage accumulations at complicated weldments of the orthotropic steel deck (OSD) during lifetime, which could be critical and might affect structural safety and reliability. Nevertheless, fatigue is a damage accumulation process that is subjected to various aleatory and epistemic uncertainties from ambient environment, model simplifications, measurement error, etc. Challenges, such as realistic load characterization, modeling and simulation of complex structures, model parameter identification and calibration as well as uncertainty quantification, exist when evaluating the dynamic performance and fatigue damage of structural details in the vehicle-bridge-wind-wave (VBWW) system. To address these challenges, this dissertation proposes a list of versatile and efficient numerical schemes to enable: (1) comprehensive dynamic performance analysis of coupled VBWW system; and (2) probabilistic assessment and prediction of fatigue damage of OSD accounting for various uncertainties. A general analytical VBWW platform is first established based on the finite element analysis (FEA) software ANSYS and programing software MATLAB. With the established VBWW platform: (1) global dynamic responses of the vehicle-bridge system subjected to various service and extreme wind and wave loads can be rationally predicted; (2) comprehensive vehicle driving safety and ride comfort evaluations are also carried out using current state-of-art evaluation criterion. As an extension of the VBWW platform, two probabilistic fatigue damage assessment schemes were developed based on machine learning algorithms. The first one is to integrate the multi-scale FEA and the support vector machine (SVM) for fatigue reliability evaluation considering life-cycle stochastic dynamic loads. The second one is to use the dynamic Bayesian network (DBN) for fatigue damage diagnosis and prognosis of an OSD through integrating the physics-based model with field inspections while accounting for the associated uncertainties. Through the two established numerical schemes, the fatigue damage of the coastal bridge in the context of VBWW system can be evaluated.