A concept for quantifying equivalent initial flaw size distribution in fracture mechanics based life prediction models

A stochastic regression analysis technique for the prediction of equivalent initial flaw size (EIFS) distributions based on crack size against time data is presented. This technique correctly accounts for the variability of multiple crack growth drivers to enable a data set independent EIFS distribution. A numerical example illustrates the accuracy of the predicted EIFS distribution in the presence of noise in the crack growth rate data, and shows the applicability of the developed stochastic regression analysis technique to small data sets with large scatter.