I think I didn’t read this whole thread carefully enough, so I’ll try to fix what I said. It seems I never tried to run a pretension analysis beyond separation or at least didn’t take much care about the results since usually separation is a failure condition, so I updated the boltpipe4 example, this time in prepomax with axial load in a second step: it is necessary to edit the calculix keywords as shown in the picture, do it after creating the pretension load so you know the node id (seems I always did this directly the .inp file).
The results for the bolt load are in the attached picture, same as yours (time 1.0 is full pretension, then at about time 1.5 separation happens, always load a bit bigger that pretension, from then all load is taken by the bolt).
Just a small remark, the “flat” portion of the curve is not completely flat but its slope depends on the relative stiffness bolt/grip as explained here, and the non-linear behaviour can be described as 2 lines with different slope before and after separation. This is the usual “hand calc” analysis.
I provide both the prepomax file and the inp file so anyone can check the results and confirm if appropriate.
Contrary to what I said before the pre-tension section model works well beyond the applied pretension load (if CCX keywords are used properly of course).
