A damage tolerance analysis contains a number of elements; such as expected usage, structural and fatigue-related material properties, crack size, geometry, stress intensity factor, retardation model and constants, and others. Understanding of the relative significance of these elements is integral to performing a successful damage tolerance analysis and determining future research efforts to improve the accuracy of damage tolerance analysis. These elements have associated with them inherent uncertainty (as in material properties) or statistical uncertainty (as in loads, initial crack sizes, and geometry). Also, the assumptions contained in the analysis methods may result in modeling errors, that is, the use of simplified models to represent complex behavior. In this research, the relative significance of the elements of a damage tolerance analysis were calculated and ranked from a probabilistic sensitivity standpoint. The expected variation of damage tolerance analysis inputs, for example, initial crack size, fracture toughness, hole size, etc., were modeled with probability distributions determined from experimental test: and aircraft or component teardowns. Discrete parameters such as the retardation model were varied through discrete changes to the model and reanalysis. The probabilistic sensitivities, the derivative of the probability of failure with respect to statistical moments (mean, standard deviation), were determined using Monte Carlo sampling and used as the metric to rank the importance of the damage tolerance analysis elements. The methodology was applied through a numerical analysis to a fatigue critical location of a T-38 wing considering three different usages ranging from severe to mild.
