Simple Torsion Stress in Tubes Part 3

Ok, I've come back around to this Simulation type with a different issue.  I've stripped away the extraneous parts to the most elemental study.  I am trying to understand why an input torque on the end face of a 12" tube is yielding less stress than an equivalent torque applied across a 0.5" wide perimter around the OD of the tube.  The other end is fixed in a similar fashion with respect to each load condition.  Hand calculations predict for a 0.594 ID and 0.794 OD tube, the shear should be 222ksi converting to a Von Misses of 384ksi.  The first pic (25Bar_Ends_Long) I applied the load and the fixture to the end faces and got 387.9.  The second pic (25Bar_Ends_Short) is of the load being applied at the end face with the 0.5" removed from the tube length to mimic the torque being applied on the 0.5" sections and transmitted through at the shear cross seciton and got 389.0.  The third pic (25Bar_Ends_Split) is the most representative of the system I'm tyring to model complete with component contacts.  The 0.5 width is split from the rest of the tube to simulate the levers that would be acting at those sections.  That yielded 493.6.  So that's nearly 100 ksi difference in results and this difference is even more so for the system I'm actually modeling.  Can anybody shed some light on the difference?

Thanks.

SolidworksSimulation