5 Replies Latest reply on Apr 26, 2010 9:16 AM by Bill McEachern

    Averaging Von Mises Stress

      A recent comparison test between two alternate shapes used in an assembly simulation has me wondering if averaging listed stress values within some percentage of peak values is a viable practice in determining the more favorable outcome.  I've heard of this practice before but I'm not clear on what elements in the overall test should be removed from the stress output used to calculate the average, i.e. how far do you go in removing the outlyers.

       

      Regards,

      Dave Trent

        • Re: Averaging Von Mises Stress
          Ameer Chilakala

          Hi Dave,

           

          I did not understand your problem, can  u provide more information.

            • Re: Averaging Von Mises Stress

              Hi Ameer,

               

              If you list the element or node Von Mises stress values on a completed linear static test you can import these values into XL and create graphic plots of stresses.  I've been told that an exceptable way to evaluate results (for comparison between two similar tests) is to average the values.  The purpose is to remove node or element peaks that occasionally occur on a very few elements.  Assuming this is true, I'm wondering if anyone has experience using this technique and, if so, how to decide which elements to include in the data set (you wouldn't want to use all of the elements since the potentially large number of elements that see little or no stress would skew the results).

               

              Regards,

              Dave Trent

            • Re: Averaging Von Mises Stress
              Bill McEachern

              Hi Dave,

              I would suggest you put a picture of the geometry. This whole subject has many facets which would affect what you are trying to sort out. Stress singularities would need to be sorted out, along with bad element geometries. What is the yield stress of the material - assuming it is a ductile metal here (at least some what and you are looking at  Von Mises stresses). If ti isn't a fatigue issue things are far easier to sort out. If it is a fatigue issue then things can be a lot more complicated in my view. The whole averaging idea needs to be supported by a comprehensive strategy and probably by some sort of empirical data as well. Alternatively a non-linear analysis maybe the way to go. It does look like design 2 produces lower stresses though it may not matter much if this is a fatigue application combined with a stress singularity.

                • Re: Averaging Von Mises Stress

                  Thanks Bill,

                   

                  I agree with your suggestions although I should add that the purpose of my question was to delve into the subject in general.  This might be counter to your point that there are too many facets to consider on the subject although, since I'm currently limited to linear static testing (always on tool steel), I consider dynamic or fatigue conditions based on the linear static results.  As you point out, the conditions beyond yield can be corrected using a dynamic test but in the case of linear static results, is it possible to use the extreme stress values to bias the linear static results?  Your comment on bad element geometries is particularly relevent here.  Although I expect mesh adaptive convergence to correct for some extreme conditions, It's still common to see the occasional singularity-like elements that I often choose to ignore depending on conditions and location.

                   

                  Regards,

                  Dave Trent

                    • Re: Averaging Von Mises Stress
                      Bill McEachern

                      In the real world you don't get singularities - you get non-linearities. The other option I use is to put in a continuous approximation (a fillet typically) and find out what sort of radius produces the limit I am looking for with a mesh independent solution. That should provide some insight. Then its all about the gradient and how fast the crack will propagate.