Error with solving

dammerel

Hi all, I have a model which was exported as IGES from Trimble Sketchup
i have an error with solving, im i doing something wrong??
i have attached the files
Andrew

Victor

Hello Andrew

This is actually a shell mesh, even though it looks solid at first. That's the reason it complains about no thickness in the Aluminium material. Not having a thickness is then the reason it doesn't solve.

Mecway can only make shell meshes from IGES files. It's better to use STEP if you can get that from Trimble Sketchup.

If not, a workaround is to convert this shell mesh into solids using Mesh tools -> Automesh 3D. However, doing this will break the link to the IGES file so you'll have to repeat everything if you modify the geometry or regenerate the original mesh.

dammerel

Thanks Victor, Sketchup will only export a surface model but once the IGES or STP have been meshed, are they still needed in Mecway? I have attached two files which are the same part (iges & stp) but different results

dammerel

attached files, i guess the mesh refinement would make the differents

Victor

These two files are identical (wrong upload?). Mesh refinement will probably have a big effect on them though because they use tet elements which can need a lot of refinement for good results, especially when they're so distorted like that.

You don't need to keep the IGES or STEP file after meshing. Although it's helpful if you do because then you can adjust the meshing parameters or do local refinement and mesh them again. It doesn't store the geometry data in the LIML file and depends on the actual IGES/STEP file to remesh them.

dammerel

Thanks again, typically what would be a node and element count on a bracket that size?

Victor

A few hundred or thousand, depending on what features you're interested in. Maybe down to 100 using hex20 elements (hand-made mesh). I notice that in this case the stress away from the fixed support mostly changes by less than a couple of percent when you refine the mesh from 300 to 1600 nodes, indicating that 300 nodes may be good enough.

If you're interested in the stress near stress concentrations, you'll need a finer mesh and curved internal radii to get it right on the corners.