Not sure if this is the source of the error but you do not truly have cyclic symmetry. This post has an excellent explanation.
You can try 2D axisymmetric if you're trying to keep model size down. However I always model the full component in these cases. Also, it appears that everything is modeled as solids. The convolutions should be modeled as shells since solid elements will introduce artificial stiffness for such a thin material.
Thanks for the reply, problem solved!
You were right about it not being circular symmetry - modelling it as a quarter of the bellow and using the normal symmetry constraint on the orthogonal faces sorted things out.
Thanks very much!