| Damage tolerant design is used in many industries to allow safe and cost effective operation of structures, vehicles, and machinery with cracks. The goal of damage tolerant design is to know when flaws will prevent their safe and proper function. Fracture mechanics is used to predict the effect of a crack on the body in question.; Most fracture mechanics problems involve crack propagation, and therefore a geometry that changes over time. A series of solutions is required to reflect the changing geometry of these problems. Although fully arbitrary, three-dimensional analyses are possible for many fracture mechanics problems, most commercial fracture mechanics codes limit the shape of all three-dimensional cracks to elliptical quadrants. This restriction allows the use of influence functions or previously obtained curve-fit solutions. These approaches are much faster than direct finite element or boundary element solution techniques, which are required for arbitrarily-shaped cracks. However, the increased solution speed might be obtained at a cost in accuracy in the predicted fatigue life.; The main questions addressed in this work are: Is the current generation of numerical fracture mechanics codes adequate to predict behavior in real structures? Can acceptable predictions be obtained quickly by making assumptions regarding crack shape? Can the methodology be improved to allow for more accurate predictions?; Fatigue crack propagation in three different configurations and stable tearing of corner cracks are investigated experimentally. The prediction methodology is evaluated through comparisons of numerical predictions and experimental observations. Three different fracture mechanics codes are investigated; two commercial codes that feature simplified geometric modeling and pre-calculated stress intensity factor solutions, and a research code that features arbitrary geometric modeling capabilities.; Conclusions are drawn regarding modeling capabilities. In the cases where predictive capability might be unacceptable, suggestions are made for improvement and for future research. Guidelines are suggested for when the simplified shape of the flaw is inadequate for engineering analyses. |
