You know, I have seen this problem of multi-thickness shells/slabs over and over again. It is really a pain to model such things with the FEM due to the centerline modeling strategies. I see no other way to model such shells other than through the boundary element method.
It would be a better idea to use a boundary element method software to analyze column reactions than an FEM software. Internal supports in BEM are formulated to include the exact dimensions and locations of columns, walls and cores.
Hope this helps.
