Crack-initiation from notch

Hi All

The fracture mechanics of multiaxial fatigue loading In steel plates have been studied comprehensively since the late 70s.

simplified the case with an inclined crack in uniaxial load has been investigated in metal by many researchers, as well.

in this forum, few researchers present their work that focuses on crack initiation from notched in metals.

So what I am studying is a combination of the all component above together, Yes the notch exhibits plasticity, under the load applied.

the tested specimens are a centrally inclined U-notch steel plates under axial load, Yes, It is fatigue load. different orientations of the notch were investigated included the case of falt U-notch (pure tensile)

We identified that the total fatigue life of any specimen with inclined U-notch is the summation of the life in mixed mode phase (crack initiation) + the fatigue life of pure mode-I (the life the of the same specimen but with falt U-notch/tensile stress only)

we also assumed the number of cycles in mixed-mode is the crack initiaion life. As we deducted the total fatigue life of plates with inclined U-notches from that of the plate with flat U-notch. This  difference represented the effect of combined load on the fatigue life...But can we consider it as the crack initation life? N-in mixed-mode, intiation= (N-total-notch angel=45)-(N-total-flat-notch), N=number of cycles

when I did simulation 

The output of the crack initiation life was almost always far less than that of test data (total life of specimen with inclined u-notch-the total fatigue life of specimen with flat-U-notch-pure mode-I/tensile stresses)

Yes, in the modelling, we got crack initiation life even for the plates with a flat notch, so it was deducted from the value of other plates with inclined U-notches, to get the crack initiation life for each specimen.

From literature, the crack initiation when the multiaxial load at the crack tip, the crack kinked (totally in mode-I) after being initiated. is this still correct when we have a notch?

Do you think, the difference between simulation (was less) and the test related to any factor has not been considered?

I am thinking if the difference comes from a missing stage in the simulation. The actual case may be three stages instead of 2. it is the crack initiates first, then undergo translation stage from mixed-mode to mode I, and then dominated by tensile stress only. If this is correct how to model the translation stage. 

noted that the specimens were notched using EDM method which was neglected in the simulation. However, as the predicted life was always less than the test data should not the EDM machining is the problem as we know that machining with EDM brings tensile stresses only at the notch that reduce the life. The case we have is the fatigue-life from the test is higher than the simulation.

Hopefully, will see your comments, it would be appreciated 

Thanks