5 Replies Latest reply on May 19, 2009 9:32 AM by Devi Prasad Samal


    Devi Prasad Samal
      Hi all,

      i have a basic but most important question. i have created a plate of rectangular section. i have restrain one face & apply some force on the other side.

      Now when i take the full model and analysis the part it gives some result. but when i take symmetry condition, it will give some deviation.

      so finally my question is during taking the symmetry it deviates from it results. so is it safe to take the symmetry for analysis?
        • Symmetry
          Prashanth Srinivasan

          Did you check if your mesh size is same for the full model and the half model?

          Also you may need to check if your symmetry (half-cutting) plane is not cutting across any elements in your original full model mesh, but rather it should cut at the nodes between each adjacent elements, so that you have equal mesh profiles in both half models.

          From my experience, the change in mesh size has some impact on the results - and it could be that you are facing the same issue
          • Symmetry
            Chad Schmidt
            Are you sure your halving your load? I'm guessing the load on one of the faces has been split up. Did you account for this?
              • Symmetry
                Devi Prasad Samal
                yes i have taken the same default mesh settings, & calculate with half loads as i divided the part in to 2 equal halves.

                And also i have keep in mind that it should reflect a symmetry.

                i am not saying that due to symmetry i am getting wrong results. but some deviation is there.

                i dont think it only happens in my case.

                just take a plate then restrain it from one side. then apply some froce on the other side. and clculate result. in my case i have got let 54652N/m2

                but while takeing the symmetry i get the result 54108 N/m2.

                so imean to ask is this deviation is ok in case of analysis.

                just try i am sure you also get some deviation. otherwise i am doing some mistake.

                thanks for the reply.

                i also try your suggestion.....
              • Symmetry
                Anthony Botting
                Hi Devi: that is apprx 500 parts in 50,000, or about 1% difference. It is not likely that a laboratory measurement can get that kind of close tolerance (or, difference) in strain gauge data during testing - so, you are good. I teach the software, and recommend very close examination of significant figures. I instruct students to generally not use more than three significant figures. If you conducted a convergence check on both meshes, theoretically they should give you the same answer to more precision as long as the mesh is perfectly symmetric on the "whole" model, and the same size and type elements are used for both models. Just keep in mind the FEA is only an approximation to real life. Good luck.