Meshing issue

Rino_aic

Hi, I'm a newbie. I was trying to repeat the Ansys example

to compare the results in Mecway, but unfortunately the test stops at the meshing stage. The attached file is a step generated in Rhino and is a valid solid with no duplicate edges. For me this is a big problem because the analysis of my models starts from solids in step format generated tutti in Rhinoceros V7.
Thank you

Victor

It meshes OK if you set Maximum element size to 10 mm. This seems to be necessary with thin-walled parts sometimes.

Rino_aic

Thanks Victor, but now a question is needed. That parameter that I had intended as the maximum value of the size of the single mesh element, based on what considerations should it be attributed? If in my model there are elements with a thickness of 3 mm I am inclined to think that that value should be less than 3.
Thank you again

Victor

I think it's the longest dimension of the element, or something close to that. I usually pick a value around 1-10 times the minimum wall thickness to prevent extremely flat elements which seem to cause meshing failures. You're right that for the best quality elements, it should be less than the minimum thickness.

disla

Be aware that your stp seems it is made of shells and solids. The main beam is meshed as solid section too if you apply those meshing parameters.

mesh

Rino_aic

Yes, in my second version I modified the model to have a tubular section and to better verify the behavior of the intersections of the parts of the solid generated in Rhinoceros. So I left all solid hollow elements. Unfortunately, even in this second version with the parameter set equal to 1mm the meshing is not performed, with 10 yes, but errors can be read during the generation as you can try.
Thank you

disla

Hi Rino_aic,

If you want to check numbers with the example problem the plate is solid as shown in the video and the piece of pipe probably too but not sure.
I think the values that are making you fail the mesh are the ones related with the minimum per curvature and elements per age.
Try 0.001 each

Rino_aic

Thanks, but with 1 mm maximum element size and 0.001 (M.N. o E.) I have the error. . With 10 mm and 0.001 in MIN. number of elements per curve I have loss of details in the mesh!

disla

Hi,
I would start with big uniform quadratic elements to check that the model runs ok and results. If they are well shaped the results will be ok.
1mm you are going too far. That is more than 2-3M nodes and the problem is probably that you run out of memory.
Remember the video says geometry is the less relevant of the error sources of the model.

Victor numbers look very good to start. I don't see you loose detail. If it works fine you could refine x2.