What causes Inverted Mesh's

[ryandias]

I am new to FEA. I am having some trouble with a simulation stopping due to an "inverted mesh."

What sort of geometry would cause this? What sort of changes should I preform to rerun the simulation?

Thanks.

 

[btrueblood]

Your mesh is likely generated using a left-handed instead of right-handed coordinate system. The result is that the Jacobian of the "stiffness matrix" is negative, causing havoc with the number crunching. Did you generate the mesh youself, or use an auto-mesh utility? Did you combine multiple meshes, possibly inverting (mirroring) one mesh?

Last, are you doing a fluid dynamics simulation or stress analysis? This is a fluid dynamics forum, not a FEA (which is more classically for stress analysis) forum...

 

[ryandias]

btrueblood

Yes my simulation is actually a mechanical stress simulation.

My previous experience is only with CFD simulations. I was not familiar with a mechanical stress simulation folder/section. I was hoping the mesh generation was similar.

I specified 4 regions mesh size, and then used an automatic mesh (in the program - comsol).

 

[btrueblood]

Hmmm. Yes, meshing techniques for FEA can be more stringent than for CFD. Sounds like the automesher is not robust enough, though. You may want to repost this question in the Structural Engineering/FEA forum; if you do so, post a note here redirecting people to your new thread. Some questions you could answer in that post or here include: What software package(s) are you using (I assume SolidWorks and the Cosmos FEA solver). Does the part have multiple holes, and/or is it an assembly of multiple parts; if an assembly, have you tried exporting the assembly as a part file first? Is the automesher working from a "native" solid part file, or have you tried exporting the solid using STEP or similar, and meshing that?

 

[ryandias]

The program I am using is called COMSOL. IT is ultra NOT userfriendly. Its appeal to my company initially is that it does multi physics, heat transfer, Stress, columbic heating etc.

It is a part file, and there are 4 holes in the part.

I "reworked" some of the geometry to get rid of fillets and other complicated features.

Thanks for your input!

 

[btrueblood]

Oh, COMSOL, sorry, missed it in your earlier post.

Hm. I wouldn't think a high-end product like that would have auto-meshing issues. Have you talked to the vendor, and what did they say? Can you try meshing without specifying the mesh density at first, and let it get a coarse solution, then increase the mesh density after getting the initial coarse mesh.