6 Replies Latest reply on Jul 7, 2009 1:48 PM by Roland Schwarz

    is this a mathematic singularity?

      Dear Simulation experts:

      I am modeling two pieces screwed together using "bolt". At the edge of the bolt head, I got high stress and the stress increases when I refine the elements in that area. Is this due to numerical singularity and can be ignored? If this is real, how can it be designed out? Thanks!

      Pei
        • is this a mathematic singularity?
          Chad Schmidt
          Try putting a small fillet in the area of stress. See if this makes your stress go down with increasing refinement on the fillet.
            • is this a mathematic singularity?
              Hi, Chad,

              Thanks for the reply!

              I am using "bolt" BC to model the screws. So, the screws are not physically in the model. Plus, the screw heads act on a flat surface, so, I cannot add any fillet.

              I just ran the same model using ANSYS. In ANSYS, it can model bolts with pre-load and the screws are physically in the model. I did not get the same high stress in the same areas. However, in ANSYS, I got high stress (but, not as high as predicted in SolidWorks) at 4 spots on the inside edges of the screw holes.

              So, ANSYS and SolidWorks gave two different results near the screws areas. I am not sure which one to trust. I probably need to try a third FEA package?!

              Pei
            • is this a mathematic singularity?
              Roland Schwarz

              Pei-Ying Hsieh wrote:

               

              I got high stress and the stress increases when I refine the elements in that area. Is this due to numerical singularity and can be ignored?

              To the best of my knowledge, the opposite is true. You need to refine the mesh until results stop increasing significantly. Coarse mesh tends to produce results that are skewed low in areas of high stress gradient.

              If you find it is not possible to refine your mesh to the point where results converge (stop increasing w/ finer mesh), you need to reevaluate the construction of your model.
                • is this a mathematic singularity?
                  Hi, Rogier and Roland,

                  Thanks for the answers.

                  SolidWorks support engineer sent me an article indicated that bolt connector is not accurate near the bolt region. After physically modeled the bolts in SolidWorks (and ANSYS), I found that this is true (the high stress near the bolt head is gone). Unfortunately, it is critical in my case that the stress is accurate, so, I have to physically include the bolts in my model.

                  I am also refining my element size to make sure I get converged solution.

                  Pei
                    • is this a mathematic singularity?
                      Roland Schwarz
                      It is true that the bolt joint produces singularities that simply do not exist in real life. The area under the bolt head will deform in ways to prevent such singularities in an actual assembly. Even modelling the bolts might not completely eliminate the "hot spots".

                      Another thing to consider is the plastic deformation that occurs when a bolt exerts stress beyond yield.