9 Replies Latest reply on Dec 2, 2011 10:44 AM by Bill McEachern

    Mass Participation

    Alec Chalmers

      The universally accepted level of mass participation for a linear dynamic study is 0.80, at least in the axis of interest. In several studies llately I am having a hard time getting to the 0.80 participation without including hundreds of frequencies which creates extremely long duration solutions or forces the engineer to shut the simulation down as it begins to page to virtual memory. Ultimately the frequencies included in an analysis will be in the 10kHz - 20kHz range, frequencies that are far removed from the frequencies in the dynamic profile being simulated and at levels that do not have significant mechanical effects. Does anyone have a guideline as to the acceptable mass participation that will generate accurate results?

        • Re: Mass Participation
          Paul Kellner

          I have to ask what it is you are analyzing?

            • Re: Mass Participation
              Alec Chalmers

              Lately, I am performing random vibration studies to support the design of vibration fixtures. In many simulations, I cannot get over a mass participation factor of 0.50 without hundreds of frequencies which include frequencies that are in the RF spectrum. I also have access to Abaqus which has the option to include a "residual mode" that captures the effects of the eigenmodes beyond a specified upper limit and assures that the mass participation is greater than 0.80. I have not investigated the mathematics supporting this feature, but it is very widely accepted. In general the correction that this residual mode makes to the results of a simulation are not dramatic. When using Soliworks Simulation I am often times constrained by the computer resources to include a set of frequncies that may only represent a 0.50 mass participation in the direction of interest. SW should look into establishing a method to include a "residual mode".

                • Re: Mass Participation
                  Bill McEachern

                  Hi Alec,

                   

                  Getting anything accurate with anybody's code over about 10k Hz is a bit of a stretch based on my experience. there are many cases where getting to the 80% mass participation is just not reasonable - but it is case specific. As you mentioned the inclusion of the high modes with the ABAQUS technique you mentioned "are not dramatic" then maybe they are worth ignoring altogether. If you have any flat shell structures in your model, you will get a lot of out of plane modes (like maybe thousands depending on the structure) to show up before you get an in plane mode to appear. hence you mass participation in directions in the plane of the shell might very well be totally acceptable down in the 30% range. It depends on the specific situation.

                    • Re: Mass Participation
                      Alec Chalmers

                      Bill,

                       

                      I don't believe that the effects of modes in the 10k Hz and higher are meaningful for mechanical structures, especially when the base excitation has an upper limit of 2k Hz. The problem arises when documenting a simulation in an engineering report; reviewers are going to be looking for the 0.80 mass participation. Many are not acquainted with the problems inherent in attaining that figure.

                        • Re: Mass Participation
                          Kevin Corr

                          Are there nonlinearities significant to the response, ie. material, geometric, contact? If so, wouldn't nonlinear dynamics solver be a match. It doesn't care about about mass participation since it doesn't do an integration of eigenvectors, right?

                          • Re: Mass Participation
                            Bill McEachern

                            If they are competent enough to review the report then they should be competent enough to understand why getting to 80% isn't a sensible thing to strive for don't you think?

                              • Re: Mass Participation
                                Alec Chalmers

                                Is there any general rule other than the often unobtainable 0.80 mass participation that can be referenced? We are leaning towards the use of a lower limit for mass participation in the axis of interest for inclusion of modal frequencies in a study.

                                  • Re: Mass Participation
                                    Bill McEachern

                                    Maybe you should take Kevin's suggestion and just do run with the direct method. Use the non linear solver without any non linearities enabled (no contact - other than bonded, leanr elastic materials, no large displacements). Pick an appropriate time step and pick some places to monitor so you get the maximums. You will have to pcik a sensible saving step. It will take a while to run and generate lots of data - last time I did this it was over 25GB on a some what simple model. then you don't have to worry about calculating modes. It will be a more accurate solution - some customers demand these solutions as opposed to using the modal method. Then you will have a reference case you can compare various mass participation factors to so you can demonstrate what modes are of use and what the errors are. Let us know how it works out if you do this.