Thanks for the answer. In hand calculations I assumed only forces in the ends of the shaft but for it to be fully fixed there should also be a moment in each end.
By allowing the computer model's surfaces to rotate around the vertical axis, representing a pinned support, I got the same result.