6 Replies Latest reply on Jun 2, 2008 1:38 PM by Bill McEachern

    Interpreting results

    Koel Abell
      I've modeled a 1/2-20 Gr.5 bolt. My test ran with the threaded end of the bolt fixed, while loading the underside of the bolt head to the manufacturers specified bolt preload of 12,360 lbs.

      The results show stresses at the shank to bolt head fillet at around twice the ultimate strength of the bolt material.

      Obviously real bolts will not "fail" at this preload so what are the results telling me?

      When I run fatigue studies of my assembly using these bolts with the prescribed preload, I get cycle lifes of only around 100.

      Thanks for any and all help.
        • Interpreting results
          Michael Atlas
          Sounds like 2 different questions here.

          1) re. the "stress concentration" at the shank to bolt head: what is your mesh like here? I would expect that you will have to specify a finer mesh in this area or set the Mesher Option "Automatic transition" to ON and try again. Keep going finer and finer in this area till peak stresses appear to converge.

          2) Fatigue life of a bolt loaded to Preload: What are you inputting as the cyclic loading on the bolt? If you are cyclingthe bolt itself from 0-Preload then I wouldn't be surprised at all for it to fail within 100 cycles. The whole idea of a preloaded bolt is to prevent the load on the bolt from varying. This is a result of the interaction of the ratio of the relative stiffness of the parts being bolted to the stiffness of the bolt. Cycling the bolt by itself is not realistic.

            • Interpreting results
              Koel Abell

              Using a finer mesh on the fillet didn't have much affect on the stress concentration results in the pure preload test (the mesh was initially fairly fine as it was a simple part and I turned the mesh as high as it would go).

              As for the fatigue test, the bolt preload is held constant while a load perpendicular to the centerline of the bolt is cyclically applied to the center of the bolt shank, a lot like a pin and clevis with the pin being the preloaded bolt. The problem is, the bolt head is initally stressed past yeild and tensile strength.

              Thanks for your help.
                • Interpreting results
                  Michael Atlas
                  Are the stresses in the rest of the bolt correct? Should be somewhere around 65K PSI.

                  If you are just using the slidebar to set mesh size and you don't have Automatic Transition on you still may not be meshing anyway fine enough.

                  I would try a local condition that set the mesh about 1/2 the smallest radius of the fillet to start.

                  Otherwise, something else is going on.

                  Results picture?
                    • Interpreting results
                      Vince Adams
                      Do your average shank stresses conform to F/A?

                      Instead of guessing at an appropriate mesh size, try uisng h-adaptive and let COSMOSWorks size the mesh for you.

                      If you could post a snapshot of your results as Mike suggested, we might be able to see something.

                        • Interpreting results
                          Koel Abell
                          The stress on the 1/2" shank is about 63ksi.
                          I tried a much finer mesh on the fillet. There were 1.1 million DOF and for a 2 inch long 1/2" diameter bolt, that's a lot. With this finer mesh, the stress at the fillet increased.

                          Two pictures are attached. One shows the finer mesh on the fillet and the other is a stress fringe plot. Using ISO Clipping. over 9% of the part is stressed over the ultimate strength of 120ksi.

                          Thanks for all your help.
                            • Interpreting results
                              Bill McEachern
                              you need to due an NL analysis - the areas higher than yeild have just lost stiffness and the load is re-distributed which, if the analysis is linear is not accounted for. You can not predict failure with a linear analysis. You can barely get it done with an NL analysis. With a linear analysis you can figure out if anything is above yeild but that is about it. It doesn't tell a thing about anything higher other than it is higher - you have no clue how much.

                              On the faigue issue the pre-load is used to lower the stress ratio to ensure that the cyclic component is small and if done right pretty much zero.