Abstract:
A new approach to the description of short crack growth - the P - a curve - has been tested using extensive data on submillimetre cracks in a grey cast iron. It is shown that this approach, in which the probability of growth is plotted as a function of crack length, is valid in that it produces a plot that is consistent from specimen to specimen and is a function of both crack length and stress level. The growth probability, Pg, can be defined systematically in terms of the average amount of crack growth, οcmean within a given interval of cycles, οN; this leads to the concept that Pg characterizes growth at a particular rate: οcmean/οN. Predictions were made of the endurance, Nf, using a numerical model that simulates the growth of a large number of cracks. Crack coalescence was found to be a critical feature, both of the model and of the experimental findings. The great majority of crack growth was found to occur by coalescence; a simple model of coalescence, based on fracture mechanics, was successfully incorporated into the predictions.
