The model provides an adaptation to Paris-Erdogan in order to take into account the variation of the growth rates caused by the first grain boundary.
The reliability predictions were consistent with what was predicted by the model for the 140kN and 160kN data. The specimen at 140kN did not fail. The specimen at 160kN failed at 486900 cycles.
The results for the 200kN weren't so promising. Failure actually occurred between 50000 and 60000 cycles. The 200kN data exhibited coalescence to a greater degree than the other data, and on closer examination it was clear that this phenomenon was the cause of failure at such an early time.
In order to generalise the model further the question of coalescence had to be reexamined.