I  would like to share a  better way to model tie rods  in bellows. The traditional method with Tie/link parallel to the bellows limits movement in the as modeled direction of the tie, but a real tie rod will actually move in a arc and the restrict movement in the direction the tie rod  is pointing and any given time.

The pendulum option for inclined supports can provide a more accurate model.  Use this instead of the Tie/link.