Hello Haiquan,

Thanks for your inquiry.

- No, you can not. Principal stress does not follow the normal distribution which is the basis of the PSD method.
- By default, the stresses in random will be in the solution coordinate system. You will see in the details it says solution coordinate system.

For solids, this defaults to the global, but for shells, and is on an element-per-element basis (x and y have to be inplane of the shell), so you should use an element orientation (right-click on geometry) branch to specify the orientation. You can then insert a results object under Results >Coordinate System > Element Triads to visualize the directions. There is a good write up in the help on this. Solution Coordinate System (ansys.com)

3. For layered sections in Mechanical the results are also in the solution coordinate system which is the layered direction (fiber/matrix), but in your case, you have short fiber. Are you using the short fiber workflow? Can you confirm back. If so, I will have to double-check.

Thank you

Sean