1 Reply Latest reply on Jul 22, 2011 10:32 AM by Tim Goddard

    Mesh failure for simple shape.

    Chris Hagen

      I have created a simplified airframe structure consisting of a circular tube, stringers (10 simple flat sheets that run the length of the tube on its inside), and 15 transverse disks repeated down the length of the tube on its inside.  The stringers were made with the circular pattern tool and the disks were made with the linear pattern tool.  They do overlap if that makes any difference. I deleted all patterns and extrusions that it told me caused the mesh to fail one by one and eventually I was down to the original tube and it said that the simple tube section failed to mesh.  This is rediculous, please help.

        • Re: Mesh failure for simple shape.
          Tim Goddard

          For most FEA programs there is an accuracy and computation cost associated with trying to mesh very thin structures with solid elements.  The cheaper computational element provides totally inaccurate results, while more accurate results are computationally prohibitive.  Validating any analysis results is an important step to any engineering decisions.  Improper use of the variables involved can lead to bad decisions.  That being said, SW simulation makes it a point to obscure the technical backbone to make it easier and less likely to be used improperly. 

           

          Why am I saying all this? 

           

          Your problem sounds like a great application of shell elements for more reasons than a successful mesh.  I would suggest you create your model using surfaces in lieu of solids.  In another post it was suggested you could keep both solid geometry and surface bodies in the same part file by using surface->offset and then removing the solid bodies from the analysis.  Alternatively you could start with surfaces and then use the thicken command.  I suggest the latter as separate surface bodies may prove hard to connect the mesh boundaries.

           

          Shell elements require thickness and offset definition.  Right clicking a surface body in simulation allows for 'Edit Definition..'  Offset is the position of the shell with respect to the 'real' solid.  So if the surface is dimensioned as the outer skin, inner skin, or mid-plane that is where you define what it is. 

           

          Meshing may require using  'curvature based mesh'

           

          Validity:

          Always construe some analysis model that you can verify with other means.  Roark's formulas for stress and strain is great.  I believe there are a multitude of examples to pick from. 

           

          As with all stiffened panel calculations buckling is a concern so do not forget to consider it.  Let us know how you make out.