3 Replies Latest reply on Aug 9, 2011 11:10 AM by Loren Sackett

    Large shell model - Buckling Analysis problem

    Ben Towell

      I have been asked to do a buckling analysis on a model of part of a large ship.  The model has been generated using surfaces with the intent of doing a shell mesh.

       

      The model meshes as shells and runs as a static analysis.

       

      However I am not able to run the model as a buckling analysis.

       

      I keep getting a range of errors as I have tried different options. (Status code 2:Preconditioner fails to factor, Stop 5 Singular maxtrix, degenerated surface: see output file...)

       

      The model has 3000 surfaces. I am using SW2010 with the latest service pack. With a 250mm element size it takes 2.15hrs to mesh. With a 200mm element size it takes 45mins to mesh.

       

      Does anyone have any suggestions as to why it will run as a static analysis, but not as a buckling one?

       

      I can't find any information from the failed analyses, to identify what has gone wrong.

       

      Thanks for any help.

        • Re: Large shell model - Buckling Analysis problem
          Loren Sackett

          I have a few thoughts:

           

          1 - Double check your boundary conditions to make sure that you do not have unconstrained degrees of freedom.  This seems to give lots of different errors including "Singular Matrix".  Does the buckling analysis have the same restraints as the static analysis?

           

          2 - If I remember correctly, a fine mesh increases stress accuracy, but has little affect on deflections.  I believe that a course mesh would be fine/preferable for a buckling analysis.

            • Re: Large shell model - Buckling Analysis problem
              Ben Towell

              The buckling analysis has the same constraints as the static analysis that ran.

               

              I have tried the analysis with a finer and courser mesh and that did not help.

               

              I am going through the model to try and see if I can identify a particular surface that is the problem. I have tried to simplify the model as much as possible to establish what is causing the error.

               

              Are there any limits on model size for shell buckling? A limit on number of elements/nodes?

                • Re: Large shell model - Buckling Analysis problem
                  Loren Sackett

                  I have never tried to do a buckling analysis with shell elements, but your idea of checking the surfaces sounds like a good idea.  Along similar lines, you might try making an experimental model with the same element type, and boundary conditions, but with grossly oversimplified geometry.  If it works, then you know the problem is geometry related, but if it still doesn't work, then you will know that you need look somewhere else.

                   

                  Maybe someone else will be more helpful.