8 Replies Latest reply on Dec 18, 2013 11:42 PM by Jared Conway

    Shared Edges, Different Stress Readings

    Nate Sieger

      I have a shell model where two perpendicular edges intersect at a shared edge with nodal bonding. When I deselect "average results across boundary for parts," I see that I can get two different sets of stress readings at the shared edge depending on which edge I select to measure.

       

      I'm just curious how to interpret this. If I'm measuring nodal stress results, shouldn't there only be one set of stress results along a shared edge?

       

      The attached images show the same location being measured, but I selected the two edges. You will see that the same node numbers are listed, but displaying different stress readings. Again, this has "average results across boundary for parts" unchecked.

       

      Any insight is appreciated, thank you.

        • Re: Shared Edges, Different Stress Readings
          Jared Conway

          are you saying the answer is different when you enable the average results across boundary?

           

          just to make things more clear, can you make an exploded view and then pull the results. this will just make sure everything is separated

           

          and also, is this a bonded contact set or automatic bonding happening here?

           

          overall what is your concern about these results and what do the displacements say?

            • Re: Shared Edges, Different Stress Readings
              Nate Sieger

              When I enable averaging, the nodal stress is the same at this location whether I measure edge 1 or edge 2; the opposite is true when I disable averaging, which is displayed in the first set of pictures I attached.

               

              This is a joined edge at a T intersection of two shell bodies in a multibody part. I have a screenshot of another view of this model, but the attach picture function is not working for me (though the attach video function does). Picture a simple T intersection with shared nodes, I am able to select two edges at the single colinear intersection. The highlighted seam in the first two pictures is the one I am measuring, which allows me to measure two edges at the same location. I think this is because the program is allowing me to either select the split line or the edge of the end plate; they are co-linear but still separately identified in the program.

               

              This is a single part with split lines where edges of bodies intersect. I have shared nodes at 100% of the seams in this model. This is why I am confused because I would think that the program is retrieving the stress data at the nodes, which I only have one set of at this seam (and others as well). However, in my picture you can see the same node numbers having two different stress readings (with averaging disabled) depending on which edge I select at the same location.

               

              Thanks for looking at this and responding Jared, I appreciate it.

                • Re: Shared Edges, Different Stress Readings
                  Jared Conway

                  displacements calculated at the nodes

                  stress calculated at the elements

                  and i'm pretty sure that nodal stress is taken from the elements to the nodes, so i'm pretty sure what you're seeing is expected. that is why i'm trying to understand what your concern is regarding the results as i think you're convinced that they should be different.

                    • Re: Shared Edges, Different Stress Readings
                      Nate Sieger

                      I'm still confused here. What I am viewing in those first two screenshots is the nodal stress. The VM stress is provided at each node, which is independently identified with its own node number. Stress is calculated in the element and extrapolated out to the node. So if I turn off nodal averaging, how is it then that I can have different stress results at the same nodes? Also, I have the exact same displacement at these locations since the nodes are shared.

                       

                      My concern is essentially that I do not understand why I am able to select two different edges at a colinear seam and where the mesh has produced nodal bonding. I only have one set of nodes along this edge but I can read two different stress values at the same nodes. If I were to measure stress at this joint, which would I select and what would the justification be?

                        • Re: Shared Edges, Different Stress Readings
                          Bill McEachern

                          If you have averaging on it uses all connected elements to compute the nodal stresses. If you have averaging off across part boundaries (bodies really) then it willnot use the element from the other body in computing the average sp they should be different regarles of whether the edge is shared or not - it has to do with which elements are in the average and which ones are not. I may not have this fully understood but I think this is what you are asking.

                           

                          Wow that was a bunch of typeo's - sorry about that. One more thing - the stresses are computed at teh guass points and averaged to get the element stress. The element stresses are average based on what is connected to a given node to get the nodal stresses. Nodal stresses are are a derived quanity. If in doubt look at the element stresses and their will be no ambiguity. If the materials are teh same between teh bodyies and say they are welded then across part boundaries is probably the way to go.