a perplexing contact problem

[jisb007]

I am simulating a very simple problem using Abaqus: suppose a spherical solid of gel is enclosed within a rigid sphere of the same radius. The sphere was modeled as a membrane element but with thickness of zero. The interface between the spherical solid and the sphere is a frictional contact boundary condition. Then I apply a rotation to the rigid sphere about Z axis. In the physical model, because initially there is no overclosure between the solid and the sphere, the solid should remain stationary while the sphere revolves around Z. However, when I use Abaqus to run the model, the spherical solid actually revolves around Z as a result of the contact defined.

I carefully examined the *.odb file, and I think this is probably because of the finite discretization that the spherical arc is represented by a series of flat surface patches. So my question is, how can I correctly simulate the problem then? (i.e., should let the solid spherical gel intact while the outer spherical shell revolves.)

Thanks.