Checking the FEA quality 5

[feajob]

Hi every body,

I would like to know how do you check the quality of your FEA for static linear stress analysis?

I know that an error estimation method was proposed by Zienkiewichz on 1987 and many other publications came later for this purpose. I am not sure if there is any adaptive mesh refinement strategy available based on this norm (in any commercial FEA code)! I use MSC.Nastran, but I am not sure if this error estimation method is available in Nastran too.

Usually, there are many small geometrical features in my models, so I do some geometrical clean up and then I mesh my model with HyperMesh. The mesh quality is very good (about 1 million degrees of freedoms for my quadratic tet meshes). My boss recently asked me, how we can be sure about the stress quality in the stress concentration zones? I thought that the best way is checking with error estimation norms. But, I would like to ask you first, may be some one have a better idea or easier way for this purpose. By the way, I found this forum very helpful.

Regards,

Ali

http://www.geocities.com/fea_tek/asd/

 

[Breezer]

MLoew,

How does this capability in MECHANICA work?

I fully agree, I would not trust comparing results from models that had unknown or different convergences either. But we are still talking linear and convergence in respect to mesh discretization – regardless h- or p-elements and assuming a good mesh quality. Perfectly converged for me means 0% difference in results due to mesh discretization. It is the duty of every analyst to check convergence, but there is often a limit, since we unfortunately don’t always get paid for this.

Assuming a concept design: An analyst’s model setup is well done, he made good assumptions and did some convergence studies. He ends up with an error of 10% in the results due to convergence, but stresses just reach 10% of the yield strength. Fine.

Assuming the same, but final design and resulting stresses reaching much closer to the yield strength. Definitely get your result difference due to convergence as close to 0 as possible.

This again looks different for an analyst doing e.g. a material or geometry parameter study. The analyst has to decide and QUOTE for the convergence requirements.

Good links.

A

 

[mloew]

AnnHer,

Please take a look at the FAQ P-Elements (FAQ828-810) and a couple of the links above. Briefly, in MECHANICA, the polynomial level of the elements is increased until convergence of a predetermined level (typically 10% on strain energies) is reached for a multi-pass adaptive analysis (MPA). There is also a single pass adaptive (SPA) method that presumes the increase in p-order to achieve convergence.

Best regards,

Matthew Ian Loew
"I don't grow up. In me is the small child of my early days" -- M.C. Escher

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[jhardy1]

A star for MLoew - I will be circulating the links to these papers to a few colleagues who will benefit greatly!

 

[mloew]

JulianHardy,

Thanks for the

! I included these links in the FAQ Errors in FEM/FEA (faq727-980) along with some other links I found.


Best regards,

Matthew Ian Loew
"I don't grow up. In me is the small child of my early days" -- M.C. Escher

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[chris9]

When I check FEA models I mainly look at the following:

1) Are the loads realistic? Shock load, fatique etc.
2) Is the model over/under contrained? In other words does the displacement plots make sense.
3) Is the CAD geometry good? Check for cracks, overlapping surfaces.
4) Correct material property assumptions?
5) Correct failure mode assumptions? Brittle or ductile etc.
6) Benchmark against something similar that you know is OK.
7) Has the solution converged? Adequate mesh density.

If your model has sayisfied 1-6 then you should be in the right ball park. I hardly ever do mesh convergence studies unless I am sure I have satisfied all the above criteria first.

 

[mloew]

I recently updated the FAQ Errors in FEM/FEA (FAQ727-980) with the link How to Model and Interpret Results Part 1: An Illustration of Idealisation & Modelling Strategies available from NAFEMS. I especially like the input checklist.



Best regards,

Matthew Ian Loew
"I don't grow up. In me is the small child of my early days" -- M.C. Escher

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[JimNDT]

Dear feajob:

I am with a company that manufactures measuring systems for optical full-field strain and stress analysis. This is a 3-D non-contact measurement.
They enable fast acquisition of the strains and stress in the observed surface section with regard to magnitude and direction.

In order to present the acquired data of all different regions simultaneously, 3-D measuring results are mapped e.g. onto a 3-D CAD model of the tested component. Using known location point coordinates, both in the Field of Measurement as well as CAD model coordinate systems, the image re-sectioning process is accomplished. This 3-D data mapping capability provides a unique, direct comparison between the experimentally collected strain fields and the analytically calculated strain data by CAE.


You might have a look at

http://www.dantec-ettemeyer.com

and follow the link to the strain and stress product lines.

That might provide help with your callenging task.

Regards
Jim



Btw.: Here, this approach might be superior to the strain gauge because the latter can only determine the strain at the point where it is mounted, averaged over its length and in the specified measurement direction. So, easy full-field hot spot detection becomes reality.

 

[feajob]

Hi Jim,

Thank you for your reply. It is an interesting approach, I wlll send your link to our test engineer.

AAY