Fatigue failure is a multi-stage process. It begins with the initiation of cracks, and with continued cyclic loading the cracks propagate, finally leading to the rupture of a component or specimen. The demarcation between the above stages is not well-defined. Depending upon the scale of interest, the variation may span three orders of magnitude. For example, to a material scientist an initiated crack may be of the order of a micron, whereas for an engineer it can be of the order of a millimetre. It is not surprising therefore to see that investigation of the fatigue process has followed different paths depending upon the scale of phenomenon under investigation. Interest in the study of fatigue failure increased with the advent of industrial ization. Because of the urgent need to design against fatigue failure, early investiga tors focused on prototype testing and proposed failure criteria similar to design formulae. Thus, a methodology developed whereby the fatigue theories were proposed based on experimental observations, albeit at times with limited scope. This type of phenomenological approach progressed rapidly during the past four decades as closed-loop testing machines became available.

This paper reviews the capabilities of a plasticity-induced crack-closure model and life-prediction code to predict fatigue crack growth and fatigue lives of metallic materials. Crack-tip constraint factors, to account for three-dimensional effects, were selected to correlate large-crack growth rate data as a function of the effective-stress-intensity factor range ([pie]Keff) under constant-amplitude loading. Some modifications to the [pie]Keff-rate relations were needed in the near-threshold regime to fit small-crack growth rate behavior and endurance limits. The model was then used to calculate small- and large-crack growth rates, and in some cases total fatigue lives, for several aluminum and titanium alloys under constant-amplitude, variable-amplitude, and spectrum loading. Fatigue lives were calculated using the crack-growth relations and microstructural features like those that initiated cracks. Results from the tests and analyses agreed well.

With the development of technology, damage tolerance design becomes compulsory and fatigue crack propagation life is a necessary design case, e.g. in aerospace industry. For low cycle fatigue problems, the failure process is generally ductile which cannot be described by the known Paris' law properly. Predicting elastoplastic fatigue crack growth life remains one of the most challenging problems in fracture mechanics. Cohesive zone modeling provides an alternative way to predict crack growth in ductile materials under elastoplastic loading conditions. The investigations of constraint effects have confirmed that cracking depends on the applied load intensity and the load configuration. Present dissertation concerns the constraint effect on the cohesive zone model and the application of the cohesive zone model for three-dimensional low cycle fatigue crack growth predictions. - A new stress-triaxiality-dependent cohesive zone model is proposed to describe 3D elastoplastic fracture process. - A new cyclic cohesive zone model is proposed to describe the fatigue crack growth with both low and high growth rates. - A new stress-triaxiality-dependent cyclic cohesive zone model is proposed and the stress-state affects both the cohesive law and the damage evolution equation.