The model is fully consistent with the sample. The hardening model was taken from the library data - this is stock steel without processing, previously it always coincided in different calculations of the samples (even if there is a slight difference).

In the calculation, the model is torn in a different place, because the real sample broke on the thread. I think this is the case, but why then did the loads coincide?

If the actual sample failed at the threads at the end, then you can adjust the model to include them.

- I don't understand this, can you explain in detail?