Again, I'm a newbie to FEA and Cosmos. Please be patient.

Moving on with my actual design project separate from my previous thread, I am trying to determine the minimal wall thickness I can obtain in a spline shaft while applying a predefined torque and reasonable FOS. I would assume the first failure mode to be when the part moves out of the true elastic limit. Now when modeling for shear, what would be the best material property to set as the target? I was looking at the tensile yield strength for AISI 1018, hot rolled, 19-32 mm round, but I'm not sure that's a direct correlation to shear.

In modeling the part, I have performed hand calculations as seen in the attached pics. I tried two different end conditions using the design torque and then the failure torque as seen with sample test pieces. With the first end condition, the ends are engaged up to a depth of 0.900" from the end faces. So I created 0.001" extrusions on the spline teeth to simulate an area of contact between the spline and yoke. With that I had to divide the torque total among the 20 teeth. This type of loading/restraint seemed to create the greatest stresses. The second end condition I used was to apply a torque to one end face and the restraint to the other end's face. This seemed to generate results more in line with the hand calculations. Now, I just want to replicate my real world results as reasonably as possible. Any suggestions on applying the load and restraints and then correlating the results back to an actual failure (yield) number that I can design with?

Sample 1 is solid. Sample 2 has a 0.5" inner diameter and was tested in the same manner as mentioned above.

Thanks.