8 Replies Latest reply on Aug 9, 2013 3:44 PM by Jared Conway

    Coarser mesh control not possible?

    Justin Strempke

      So is this something with my setup that I cannot apply a mesh control that is more coarse than the global?  I checked the KB, and saw SPR 318867 that related to this.  There wasn't any info on when/if it'll be addressed, wondering if anyone has thoughts.  I couldn't vote on it either (haven't been able to on ANY SPR's, need to check into that...).  It would be helpful many times where 80% of my model needs smaller elements, but larger bodies can use larger elements and it's more tedious setting a large global mesh, then refining all other parts.

        • Re: Coarser mesh control not possible?
          Chris Michalski

          If you use the adaptive meshing you can select "mesh coarsening" which will dumb down areas that don't need a fine mesh.  But that requires you to let it run the iterations to optimize the mesh. 

          I always assumed that if you apply a local mesh control and set it to coarse it would dumb down the global (but I've never tested that assumption).

            • Re: Coarser mesh control not possible?
              Justin Strempke

              From what I've seen the mesh applied is the smallest value in any setting, which is why a smaller global would override a local coarse mesh.  I'd like to set the coarse mesh to lower sim times, so the adaptive mesh is a bit counterproductive.  I use that occasionally, but not often.

                • Re: Coarser mesh control not possible?
                  Chris Michalski

                  Huh, I just assumed that if the adaptive could do a coarser mesh then you could use the mesh control to do the same thing.  Learn something new everyday (unfortunately in this case).

                  I tried it on a simple model I've been working with and it seems you're right, setting a super coarse mesh control on some parts actually increases my number of nodes and elements.

                  • Re: Coarser mesh control not possible?
                    Roland Schwarz

                    An inaccurate analysis costs a lot more time than one that runs long.

                      • Re: Coarser mesh control not possible?
                        Justin Strempke

                        So very true in the long run (no pun intended)!  There's just times where a large body that has finer mesh at connections to other parts can afford larger elements, but the other 100 parts in the assembly need smaller meshes geometrically.

                         

                        I know I should just get in the habit of using a large global mesh with adaptive, let the software dictate the specific parts of faces that might need refinement instread of splitting and applying mesh controls manually.  In those cases yes, even with adaptive it may be quicker solve times and overall!

                          • Re: Coarser mesh control not possible?
                            Roland Schwarz

                            I got to the point where I would just start coarse and run refinement (including coarsening) on almost all my analyses.  It just worked beter that way for me.  Added benefit of the refinement seeking convergence in regions of high stress gradient (eliminating "lowball" stresses).

                              • Re: Coarser mesh control not possible?
                                Justin Strempke

                                You know the single one thing I do hate about adaptive studies is that it locks the software while solving.  Non-adaptive only are locked during meshing, but when solving you can still multitask and work on other models, run other studies, etc.  Huge pain in my opinion.

                                  • Re: Coarser mesh control not possible?
                                    Jared Conway

                                    i think this came up awhile ago, the software is built to refine the mesh from global to control, so doing the opposite isn't recommended. but certainly something you could follow-up with your reseller about.

                                     

                                    we follow the same process as roland for all our consulting projects. an additional benefit is if you setup the problem wrong, you get there quickly so that you can fix without waiting.

                                     

                                    also on adaptive, mesh coarsening is the best of both worlds. you get more accurate results which require more elements but then balance the solve time by coarsening.