Thanks for the suggestion! The gravity load is a good idea for keeping the hubs in contact! What I've tried is to set the reference temperature of the hubs to 200deg and the thermal condition at t=0 to 200deg, and add the radial thermal expansion to the tolerance in the geometry. That way, all the parts have the right temperature at the start and the hubs don't expand into each other due to thermal strain. My only concern is whether I should change the reference temperature in Mechanical or the zero-thermal-strain reference temperature in Engineering Data. Are these essentially the same?