I would model it thusly (for code compliance:

1. As installed with all supports. If it wasn't installed per code to begin with, you can't really expect to make things better while removing supports, and this is to defend yourself and your decision.

2. With the intended thermal displacements as though the shoes were incapable of falling off. Again, if it wasn't passing before this event, you're not making it any better, again to defend yourself and your decision.

3. With the "lift-off" supports completely removed. You never specified piping code, but B31.3 wants you to run any long term "lift off" in the SUS case. If it can't pass this, then it can't really pass this while under duress of permanent deflection.

While I have not performed any fitness for service analyses personally, I would be inclined to expect you can use values from 3 above within such analysis to determine length of life left with permanent deflection and lack of support.