Not an uncommon question. I wish that this question were addressed in the beginning of the manuals.
The majority of problems in the training manuals are run using a default mesh. The goal is to teach some aspect of the tool or process, not to refine each problem to convergence. From release to release (SP or major), changes can occur in the way that a part or assembly is meshed. These changes may be small, but the result is that nodes fall in different locations on the model. In high stress areas, the relative location of the nodes to the high stress area can result in different peak stresses being reported. If you refined the mesh to convergence for these problems, you would likely get yet another (correct) value. If you ran that same problem to convergence in different releases / service packs, you should get the same result each time. They don't go through the convergence process with each exercise in the manual because that would be time and resource intensive and is not necessary for each topic being covered. Therefore, this question comes up over and over again. Why aren't I getting the same answer that is in the book? It should be one of the first things discussed before starting exercises in the class and should be addressed in the manuals themselves, because it will come up.
I have heard that the exercise results should be within 5% of the manual. This is not always the case. Regardless, it should not be a concern or a reflection on the software. Its simply a coarse mesh and a general result that has the tendency to drift.
A better check is to compare displacements. Being much less sensitive to mesh refinement, displacements should be close to what is reported in the manual.
Thank's a lot my friend, by the, if you have any informations about the same problemme in SW Motion, because we haven't a mesh there?