My understanding is the the sub modeling with take the displacement field of the system level model at the specified cut planes and apply that to the sub model. With that the difference between rigid and sub modeling should be obvious.
If, in fact, the displacements were concentrated in the area of interest, then I suppose sub-modeling would not be very useful and you could easily approximate the rest of the structure with rigid elements. Folks do that all of the time, when we simplify our model of a cantivlever spring plastic part, for example, as being fixed at one end. Sub-modeling is good for the many cases where you need to look closely at the stresses and displacements in a local region, but the larger structure still has a substantial impact on that local region.
Thanks Bill and jerry. That was very helpful