Research On Fatigue Reliability Evaluation Of Highway Steel Bridges Based On Fracture Mechanics

Fatigue life prediction is of great importance to the fatigue design and evaluation ofhighway steel bridges. From the point of view of probabilistic fracture mechanics (PFM), thefatigue life prediction and fatigue reliability evaluation methods for highway steel bridges aresystematically studied in this dissertation by combined applications of fracture mechanicstheory, crack growth theory, cumulative fatigue damage theory and structural reliability theory.Both the variation characteristics of fatigue life and the fatigue failure probability within thedesign period of the structure are obtained by the present approach, indicating the accuracyand efficiency of the proposed method in solving the fatigue and fracture problems ofengineering structures.The major work in this dissertation is described as follows:(1) The research history of fatigue problems and the fatigue design methods arepresented first, followed by an introduction to the two major types of approaches for fatiguelife prediction based respectively on S-N curves and fracture mechanics. The tradictionalmethods of PFM are summarized, and three major models of fatigue reliability analysis arealso introduced.(2) The response surface-spline fictitious boundary element method (SFBEM) and theimpotant sampling Monte Carlo-SFBEM are proposed to assess the fracture probability of alinear-elastic cracked structure. In the above approches, the SFBEM based on the Erdoganfundamental solutions is adopted to perform deterministic fracture analyses for numericaltests.(3) A fatigue life prediction approach based on SFBEM is proposed for the case ofconstant-amplitude (CA) loading, in which the Paris law for crack growth is used inconjunction with the SFBEM for fracture mechanics. To account for the inherentuncertainties in fatigue problems, a response surface method based on the above approach isfurther applied to the evaluation of the fatigue failure probability of a structure under CAloading.(4) In consideration of the loading and structural characteristics of highway steel bridges,a fatigue life prediction approach based on SFBEM is proposed for the case ofvariable-amplitude (VA) loading, in which the Miner rule for linear cumulative damage isused in combination with the fatigue life prediction approach for CA loading. To account forthe inherent uncertainties in fatigue problems, a response surface-Monte Carlo method basedon the above approach is further applied to the evaluation of the fatigue failure probability of a structure under VA loading.(5) Using the proposed fatigue life reliability analysis approach based on SFBEM for VAloading, the statistical characteristics of the fatigue life and the fatigue failure probabilitywithin the design period are obtained for the main steel girder and the cross steel girder of theHong Kong Ting Kau Bridge.It has been shown that high accuracy and efficiency can be obtained by the present PFMapproach, since the complex formulations for the derivatives of crack-driving forces arecompletely avoided and a much lesser number of numerical tests for deterministic fractureanalyses is required in the proposed method. Good performance can also be observed whenusing the present fatigue life prediction and reliability analysis approaches for CA and VAloading, since repetitive calculation of stress intensity factors during the crack growth isconducted efficiently by SFBEM based on the Erdogan fundamental solutions. Theeffectiveness of the proposed approach is well validated by the successful application of themethod to the fatigue reliability analysis of the Ting Kau Bridge.