I saw that including the effects of pressure (Burdon pressure effect) increase the loads on the nozzles. If you didn't consider this effect in your analyze I think the loads will exceed the allowable with more then 200%.
I also think that including the friction in analyze is very important, but also the results must be interpreted to see if they could be real. In such cases (sensible equipment) I am using expansion joints and/or very flexible piping traces. I usually avoid the friction supports in that case because it is difficult to appreciate if the friction value is real, or it is to high. In any case, taking care about friction is conservative.

But, in my opinion, only these loads couldn't cause the failure "in a few seconds from start-up" (purely if these loads are not extraordinary high) . You have to check if the installation of the machine was correct (alignment, clamping by screwing up bolts and nuts, etc), if the operations indicated by the vendor was respected, etc.

In any case, it is hard to convince the vendor or anyone that the cause of failure is not the piping system, because of these omissions from the stress analyze.

I should like to know if you would discover the real causes of the failure.

Best regards,
Dragos