5 Replies Latest reply on Jun 18, 2012 5:58 PM by Kevin Corr

    Relief cuts in FEA

    Charles McRaven

      relief cut question.png

      I'm trying to determine internal stresses produced by a series of press fits onto a shaft.  Please refer to the image so you can follow along.

       

      Part C is a shaft, the centerline, would be the bottom of the image.  All parts can be considered rings.  Please excuse the crude graphic.  IP regs prohibit my posting of a screen shot of the assembly.

       

      First part (A) is mounted to part (B).

      Contact set (I) is a stress fit (press fit, interference fit, whatever).

      This causes the inner diameter of (B) to collapse slightly.

      The inner diameter of the (A)/(B) assembly is then machined to compensate for for this collapse.

      The (A)/(B) assembly is mounted onto the shaft (C).

      Contact set (II) is a tight slip fit (clearance fit).

      Then part (D) is mounted onto the (A)/(B)/(C) assembly.

      Contact set (III) is a heavy enough interference fit sized so that the contact set (II) becomes a heavy fit itself.

       

      If I run an analysis on parts A, B, and C, it shows an interference fit at contact set II because SW does not acknowledge the machining step to remove the excess internal material.  This results in increased stresses once the entire assembly is analyzed.  Also, the assembly will spin at high speed(~20K RPM) so I need to know accurately if interference fits will be maintained.  A false interference fit invalidates the results.

       

      My question is this: Is there any way to have SW factor in that there has been material removed after the initial deflection?

       

      Simply constraining that face produces higher stresses then would exist in reality.

       

       

      Please do not suggest design changes as there are other factors requiring this construction.

       

      Thanks in advance

        • Re: Relief cuts in FEA
          Kevin Corr

          Run a shrink fit contact pair with just A and B first, excluding C and D? That would tell how much the machining step removes. In turn, make a cut extrude of that amount in the bore of B which would simulate the slip fit AB to C when the full four item analysis is run with the two shrink fit contact pairs ie. AB and DB.

            • Re: Relief cuts in FEA
              Charles McRaven

              Great minds think alike. That was my solution, too.  It's a pain because the part does not shrink uniformly so I have to output a csv of the displacement plot, reformat that, import that into SW, and use it to create a revolved cut.  Seemed like a lot of work and hoped there was a built in function that I just didn't know about.

                • Re: Relief cuts in FEA
                  Kevin Corr

                  Design study might be how to proceed if choice of variables(AB ID), constraints(max. stress over AB ID less than x) and goals is all that is needed to get AB's ID, which, as non constant in your experience, but would be acceptable as a constant.

                    • Re: Relief cuts in FEA
                      Charles McRaven

                      the non-constant ID of the A/B assembly is very problematic.  At 20,000 RPM, the interference fit must be maintained across the contact set or the results are catastrophic and very dangerous.  However, at rest, when the stresses from the fit are greatest, there cannot be yielding or the fits will not be maintained at speed.  So we have a very small window of fits that will work.  Currently, there is a variation of 0.002, radially, from one side of the (II) contact set to the other.  So the part must be machined to be parallel with the shaft surface.  The simulation must reflect that process or it's just a shot in the dark and we don't find out if it is accurate til the machine fails at great cost and danger.

                       

                      I don't believe a design study will work because the results still come from SW incorrect simulation of the deformation (that is, not taking the mid stage machining into account).

                        • Re: Relief cuts in FEA
                          Kevin Corr

                          The simulation can't help but reflect the maching process, in fact, it idealizes the machining too much by just changing the diameter in solidworks.  It seems like "centrifigal" load will identify the II boundary stress(eventually less than yield but greater than some minimum to avoid UFO's.) then a subsequent study to identify the II boundary stress at rest without the centrifigal load. these two stresses inform the II boundary diameter pair to be closer(stress too high) or farther apart(stress too low). I feel your pain getting that ID free of variation to avoid yield and fly away. It seems like I would try to imitate martin taper bushings if I could get away with it.