and just to finish my dissertation....
We have today better tools FEA and B31J to evaluate flexibility of components- that is really great.
We have today better tools (again FEA and B31J) to evaluate SIFs, fine!
But we have a lot of cases in which the number of cycles is somehow arbitrary highly set for services where there is no evidence such cycles will ever appear. For such cases a rational engineering approach would be to set (conservatively anyway!) the number of cycles as coincidentally with the new accurate values for SIFs to maintain the same level of safety (against fatigue) that was with the old values of Appendix D. I think this is a rational engineering approach because it appears there is no real need to increase the level of safety against fatigue for such regular applications.
Of course applications where such fatigue is a real concern must be evaluated with a higher number of cycles and accurate SIFs.
In the end, the evaluation of number of cycles would be the Owner responsibility, B31.3 has such tools in 302.3.5 (d) but not giving the freedom to set the number of cycles below a high limit (3000 or 7000 cycles). That limit can be lower for regular applications- that's the point.