Probabilistic crack growth behavior of aluminum 2024-T351 alloy using the 'unified' approach

In a previous paper, a deterministic experimental crack growth model for aluminum 2024-T351 alloy was formulated based on published experimental results, using the 'unified' approach. In the present paper, random equivalent initial flaw size (EIFS) values obtained from other experimental results are introduced to the model, and the crack growth probabilistic behavior is demonstrated. The EIFS values computed from published experimental results of 3-D cracked fastener holes are well-approximated by a Weibull distribution. This approximated EIFS distribution is used as a distribution of the initial cracks in a large number of computations for a 2-D case which is very similar to the 3-D one, and serves for demonstration purposes. A computation program was written to numerically solve the behavior of crack lengths vs. the number of load cycles. It is shown that part of the initial cracks do not propagate, as their initial value is smaller than the required threshold criteria set by the 'unified' approach. A dispersion in the number of load cycles required to obtain a given crack length is shown. Although the coefficient of variation (COV) of this dispersion is smaller than the COV of the initial random cracks, it may be of concern to designers, as the life period of the design may change by a factor of more than 3.