FEA meshing - why not 2d mesh -> 3d mesh ??? 2

[Ziggizag]

Well,

I have just realized why the people do not use FEA (finite elements analysis) as often as they should do: FEA preprocessing phase is a disaster!!! Simply speaking - there is no simple, efficient meshing software available on the market. Existing software like Hyper Mesh is just an interface catastrophe. Everything you hate in a computer software interface is existent there!

OK - being serious. The problem seems to start at the beginning. Engineering design applications are NURBS driven. They describe surface in a topology unaware manner. Then the project goes into FEA preprocessor where appropriate mesh topology is being created for further analysis.

Well - while contemporary FEA preprocessors are powerful in terms of handling various mesh types, they are extremely poor modelling tools compared to professional modelling tools like 3DS Max or Maya.

Simply speaking - any 3D artists is able to deliver a model with nice surface mesh topology: definitely this model will be better than result of automatic meshing by FEA preprocessors or even better than human-optimized meshes delivered by inexperienced FEM engineer! The only problem is... how to proceed from surface (2d) mesh to volume (3d) mesh keeping surface topology of the input geometry?

Or rather - the real issue is that I have never heard about a FEA preprocessor accepting surface mesh data as input geometry. Why??? Aren't efficient algorithms generating volumetric 3d mesh from 2d surface mesh (volume tessellation)? Sure thay exist but FEA preprocessrs do not implement these algorythms. Why???

Would you be so kind to explain why streamline like below is non-existent in practice?:

1) 2D surface mesh geometry to be constructed in 3DS Max or similar application as triangular, tetra- or even hexa-mesh (with special modelling techniques hexa- is also is possible). Basic topology optimization is achieved already during modelling phase! 3D artist can easily recreate complete mesh geometry based on data imported from industrial design application. Dimension are not recreated perfectly but sure they are recreated in a near perfect manner. 3D artist doing in Max or Maya can easily skip unnecessary features and he/she can construct mesh of various density following appropriate specification.

2) Conversion of the surface mesh model into volume mesh model preserving imported surface mesh topology. A special subroutine is necessary for this within a FEA preprocessor.

3) Generated 3D volumetric mesh is used within FEA preprocessor as usual: materials, loads, constraints are applied.

4) Final result being sent to appropriate FEA solver.

The reason for the question is simple: an hour of a 3D generalist (a 3D modeler)) work is by far much cheaper than an hour of FEA engineer while 60% of FEA engineer workload is... meshing !!!

Would you be so pleased, to comment???

 

[jhardy1]

The work-flow you describe is most certainly possible already, but I think most FE analysts would rather deal with the actual underlying geometry, rather than a faceted approximation of the true geometry. Once you have meshed the model (using 3DS Max or whatever), you have already "dumbed down" the underlying content, and made it impractical (or difficult) to "recover" the lost data.

For example, let's say the designer / artist exports a surface mesh with a characteristic facet edge of 1 mm say, believing this is ample to define the geometry they want. But suppose there are some very fine features which are important for a particular analysis e.g. stress concentration around some 3 mm diameter screw holes - how are these going to be represented in a 1 mm mesh?

Or suppose there are edge fillets or other fine features that are captured OK in the generated mesh, but the analyst wants to "defeature" these because they are not relevant to the particular analysis required, because they add excessive computational burden without affecting the significant results. How do you removed a fillet if you do not have access to the original edge definition used to create the fillet (or other feature).

Or suppose the supplied mesh is far too fine for efficient analysis in a reasonable time scale - how do you go about increasing the mesh size to an "appropriate" extent (and only the FE specialist can say what that means in any given situation) if you are now working with an approximation of the defined geometry rather than the actual geometry itself.

I'm not saying that it is impossible to work effectively with a surface mesh rather than the actual geometry definition; just that it is typically preferable for the analyst to have access to the true geometry rather than a approximation of it. Give the FEA specialist direct "hands-on" access to the full geometry, and let _THEM_ decide what features to remove, how fine / coarse the mesh should be in each location, whether they need a surface or a solid mesh, etc.

http://julianh72.blogspot.com

 

[Ziggizag]

I can fully agree with you but (an this is a big BUT) - you are talking about completly different approach than myself:

1) I already said - 3D artist is capable to produce mesh of various density following guidelines. Usually the constructor has some ideas (by experience) where stress tends to concentrate and he/she can as for more dense mesh in these areas.

2) 3D artists are used to prepare meshes of various level of detail (LOD) - so it is pretty easy to prepare such set of meshes for coarse and fine FE analysis.

3) Sure - FEA specialist is better suited for model defeaturing but... it is a matter of cost/demand. If you look for an impact test simulation... you do not bother with fine details from the very beginning! Having passed rough estimation after few design modification final model can become a subject for further full-featured FEA but your company is left with a few bucks more in a pocket!

4) Last but not least - so often simple FEA problems could be solved by a constructor him/herslef but simply speaking he/she cannot pass through disastrously unfriendly meshing software interface filled with dozens of enigmatic options! They look more for some simple access to FEA. They want to select a beam and tell the FEA software: "this is an I-beam - please, use appropriate mesh and do not ask me what kind of mesh should it be for a side impact test! And this is a hole - please mesh it around appropriately too". But in a contrary - FEA software producers do not want to allow them for such a simplicity and ask them to call a FEA specialist for a simplest task!

 

[jhardy1]

Yes, I agree we are talking about a completely different approach!

If I understand you correctly, you are talking about a "3D Artist" who is "capable to produce mesh of various density" [sic] and who already "has some ideas ... where stress tends to concentrate", and what you want is "simple" FEA so that they can tell the software to "please, use appropriate mesh and do not ask me what kind of mesh should it be" - is that right?

Shudder!!!!!

http://julianh72.blogspot.com

 

[Ziggizag]

Nope - 3D artist is only following guidelines of a constructor who gives ideas where stress may concentrate.

I see nothing funny in expectation of "simplified" FEA for simple tasks. It is regular way of software development - initially highly specialized task completed by dedicated operators turn into semi-automatic process controlled by sophisticated software. I do not believe FEA is a special case where each and every simulation actually requires a PhD specialist. This is only because the software is in early stage you have all these nasty interfaces and you need a complete theory to operate basic analysis.

This will sure change. Time will tell. The same was with image processing some time ago. And the same was with static analysis what is now already implemented in each CAD package as a simple feature suitable for simple analysis.

 

[jhardy1]

"And the same was with static analysis what is now already implemented in each CAD package as a simple feature suitable for simple analysis."

And many of the analyses that I see undertaken by CAD Designers (as opposed to Professional Engineers / FEA Specialists) using integrated CAD / Frame Analysis / Pipe Stress Analysis / FEA tools are completely meaningless. Technicolour stress plots and groovy animations might look cool, but if the analytical model violates theory, the results are worse than useless - they can give a false sense of security, as people may actually BELIEVE that the design has been checked and is safe.

A case in point: most integrated CAD / FEA systems will automatically mesh a thin-walled structure (such as a pressure vessel) with a single layer of very thin (high-aspect ratio) 4-node tetrahedral elements. Most of the lower cost CAD / FEA systems have no means of meshing such structures with plate / shell elements. Some of them may generate a warning; many do not; in any case, most CAD Operators have no idea what the warning means, let alone what to do about it. They will then run the analysis, and it will generate very convincing stress plots and deformed geometry plots - and no-one realises that the whole analysis is fundamentally incapable of capturing plate bending behaviour, which can often be a critical design mode for connections of features like nozzles to the shell. That is, the analysis may look very impressive, but is literally absolutely useless for the purposes of design certification.

I am not trying to say that FEA must remain a tool only for the privileged elitist FEA specialists. My concern is that putting sophisticated analysis capabilities in the hands of users with no training or understanding of the theory very rarely yields meaningful results. Provided the CAD Operators / 3D Artists are working under the supervision of trained FEA users, you have a chance of getting good results, but if you are relying on untrained people to use FEA without any understanding of the underlying theory or limitations, well ...

Shudder!!!

http://julianh72.blogspot.com

 

[Ziggizag]

I would say - it's all about limits and experience. You usually check the part for machinability before sending design to the workshop and they also check it in the workshop before actually mounting block of metal in the vices for drilling.

FEA analysis is additional support - it is to reduce the cost of actual tests but it doesn't replaces it. It is often yet another tool of designer - nothing less but nothing more. It does not constitute safety certificate alone.

So - we need to bring some FEA to the masses - some will do worse and some better, but we do not need a PhD guy only to count a drop test. And basic warning system must be implemented anyway.

Cheers!

 

[rb1957]

"we need to bring some FEA to the masses" ... respectively, we don't ... FEA is a loaded gun, it produces lovely looking pictures that look great but can easily be utter garbage. personally, i had little trouble meshing volumes; the tricky stuff is in making a mesh of 2D elements.

meshing a volume from a 3D drafting package is easy. refining the mesh is high stress gradiants is also easy.

 

[johnhors]

Backing up what Julian and rb1957 have already said, FEA is just a tool, a very powerful tool that should only be used by competent stress analysts. By competent I mean someone who is capable of producing a hand calculation to accompany their FEA has a sanity check. The stress analyst will produce a free body diagram sketch and check that forces and moments sum to zero before proceeding with FEA. Failure to adhere to these simple steps will almost certainly lead to inappropriate boundary conditions being applied in the FEA, and as rb1957 states, utter garbage is produced.

http://www.Roshaz.com

quality, cost effective FEA solutions

 

[GregLocock]

We need to bring differential calculus to the masses. We need to bring open heart surgery to the masses.

guess what, in real life there are things you can't dumb down.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[imagitec]

When did Ziggizag write his post? 1983?

Those responding to his posts are taking up the usual argument of CAD-integrated, "simple" FEA versus standalone, full-blown FEA and the related danger of assuming a mesh and pretty pictures = valid results.

Ziggizag isn't even dimly aware that there is such a software divide.

Rob Campbell, PE