As far as a simplifed rep goes, that's a tough question. For every little cut you knock out you will be ignoring stress concentrations. I suggest you look up Roger Toogood's "Pro/MECHANICA Tutorial on Structure," by SDC publishing. It has an interesting discussion about the differences between a CAD model and an FEA model.
As far as your constraints go you should try to make it as real as possible. There will be a constraint at the hub, a frictional load around the outer rim, etc. Try to make the constraints what they would be in a real situation. Run a quick pass analysis first to check that the model will solve and then compare and contrast what you believe will be the constraints.
THEN TEST IT!!! FEA is great but can often be wrong because the user has mad assumptions. There is no substitute for a rosette strain gauge, or a dyno, or a load cell, etc.