We ran across a HR system that experienced such sag. Turns out the original analysis had two "oversights" that allowed the creep sag - 1) the boiler connections were modeled as rigid restraints (with imposed displacement where necessary) & 2) wall thickness was not updated for bent pipe. These two errors predicted lower loads down at the bottom and the constant effort supports down there were undersized. In the beginning the pipe (and turbine?) carried the load but creep eventually allowed the supports to bottom out. Number 1) made the system too stiff at the top so the top of the model carried too much deadweight load. Number 2) ignored the added weight of the heavier pipe.