Invalid Geometry - Help Needed

[Roxana]

Hello,

I appreciate anyone who can explain to me how to find what is wrong with a structural model of about 20 member I built. The model looks fine, but when I try to run a stress analysis I get a message saying that there is invalid geometry, etc... I also try just processing the mesh, but same thing happens.

How can I find where the trouble is without rebuilding the entire model, and then hope again it won't fail...? How do I avoid this sort of things from happening? Can assemblies be stress analysed or just parts?

I appreciate all the help you can give me.

Thanks a lot.

[Roxana]

By the way, I did convert the assembly into a part using derived components...
Thanks again.

[Anonymous]

Roxana, Gary Tang made a reply a short time back about using derived assemblies for Stress Analysis: "FEA in AIP is intended for parts only. Although technically you can perform FEA on an assembly by converting it into a part using AIS derived functionality, in most cases, the results are incorrect or even misleading for the original assembly because the constraints between components of an assembly are not represented in the derived part. You can get reliable results for assembly analysis using a full-blown ANSYS product. Regards, Gary Tang Autodesk Inventor FEA QA" Does this answer your question well enough? If not, please elaborate. Best regards, -- Hugh Henderson (Inventor Workflow QA) "Roxana" wrote in message news:13463738.1101976220966.JavaMail.jive@jiveforum1.autodesk.com... > By the way, I did convert the assembly into a part using derived components... > Thanks again.

[Roxana]

Hi, thanks for your prompt reply.

I used another structure similar but smaller than the problematic one and did not have any problems with the mesh or the stress analysis, and I used the exact same method: assembly converted to a part using derived functionality. But I guess the method is just not reliable...?

Could you suggest what ANSYS product to use?

The exact stress analysis feedback error I get is:
"An error occurred while creating the finite element mesh. The cause may be invalid geometry or features that are extremely small in comparison to the overall dimensions of the parts."

Thanks for any further clues.

[Roxana]

If I were to use another product for FEA, could I import the same structure I built in inventor and would it work fine with ANSYS? Or would I have to build the models with ANSYS because the geometry built with Inventor would not be reliable for FEA...?

That would be a bummer...

[Roxana]

Another message I get is as follows:

"An error occurred inside the MASHER module: The mesh contained poor quality elements, or the process ran out of memory."

I have about 6G of hard space and 4G RAM. Could I ran out of memory with fairly simple structure? And if I have poor quality elements, how do I detect them without re-building the entire thing...?

Thanks in advance for your feedback.

[Anonymous]

Roxana, Doing an FEA analysis on a Derived assembly will work, as long as the "components" are actually touching that have boundary conditions on them. i.e. if you have a cylinder and a block that aren't touching and you but a BC on each of them, it is not possibe to analyze this type of case. Could it be that your second model has this situation? Would you consider posting the derived assembly or preferably the original assembly to the autodesk.inventor.customer-files forum? Thanks in advance, -- Hugh Henderson (Inventor Workflow QA) "Roxana" wrote in message news:25410763.1102012052915.JavaMail.jive@jiveforum2.autodesk.com... > Hi, thanks for your prompt reply. > > I used another structure similar but smaller than the problematic one and did not have any problems with the mesh or the stress analysis, and I used the exact same method: assembly converted to a part using derived functionality. But I guess the method is just not reliable...? > > Could you suggest what ANSYS product to use? > > The exact stress analysis feedback error I get is: > "An error occurred while creating the finite element mesh. The cause may be invalid geometry or features that are extremely small in comparison to the overall dimensions of the parts." > > Thanks for any further clues.

[Anonymous]

Roxana, You can import Inventor models into ANSYS Workbench, so you won't have to re-create your model. Workbench shouldn't have any trouble importing Inventor geometry. AIP Stress Analysis is a subset of Workbench functionality. Workbench can do analysis on multi-part, multi-material data. However, I'm not certain which "flavor" of Workbench has this type of functionality. You can request product information at: http://www.ansys.com/products/workbench.asp Best regards, -- Hugh Henderson (Inventor Workflow QA) "Roxana" wrote in message news:15911043.1102012365237.JavaMail.jive@jiveforum2.autodesk.com... > If I were to use another product for FEA, could I import the same structure I built in inventor and would it work fine with ANSYS? Or would I have to build the models with ANSYS because the geometry built with Inventor would not be reliable for FEA...? > > That would be a bummer...

[Anonymous]

Roxana, This can happen if you have a part with a very large aspect ratio with thin walls. Shell elements would work better in these types of cases. AIP Stress uses solid elements exclusively. ANSYS Workbench has multiple types of elements that can be used depending upon what type of geometry is being analyized. Investigating why the memory ran out is difficult at best without having the model or an image to inspect. Thank you for your inquiries, and keep the questions coming. Best regards, -- Hugh Henderson (Inventor Workflow QA) "Roxana" wrote in message news:10396280.1102012859827.JavaMail.jive@jiveforum1.autodesk.com... > Another message I get is as follows: > > "An error occurred inside the MASHER module: The mesh contained poor quality elements, or the process ran out of memory." > > I have about 6G of hard space and 4G RAM. Could I ran out of memory with fairly simple structure? And if I have poor quality elements, how do I detect them without re-building the entire thing...? > > Thanks in advance for your feedback.

[Roxana]

Thanks for all your hints and information. I am digesting your answers and will surely post more questions and probably the model itself.

I am wondering if I export the problematic model in some format such as IGES or SAT and the import again... Shouldn't that take care and heal any problems with the geometry? Maybe that is nonsense, but I wonder...

Thanks.

[Anonymous]

Roxana, It may be the case that there is a tiny "sliver" face that is tripping up the mesher. In that case, I don't think a translation will heal it, but I could be wrong. Give it a shot and if it still doesn't work I think we're to the point where we will need to see the model in order to determine the cause of the failure. Best regards, -- Hugh Henderson (Inventor Workflow QA) "Roxana" wrote in message news:20739633.1102031795217.JavaMail.jive@jiveforum1.autodesk.com... > Thanks for all your hints and information. I am digesting your answers and will surely post more questions and probably the model itself. > > I am wondering if I export the problematic model in some format such as IGES or SAT and the import again... Shouldn't that take care and heal any problems with the geometry? Maybe that is nonsense, but I wonder... > > Thanks.

[Roxana]

Hi:

After careful attention to your comments and analyzing my model. I seem to recognize that the problem relates to the aspect ration of one of the structure members. There is one pipe 60 ft long and .322 inches thickness. This indeed seems to be the problem.

However, how can I get around this issue with Inventor? I tried to use 12 sections 5 feet long each instead of one section 60 foot long, but this did not work either, even though a mesh can be generated for one 5 ft section alone...

How should I build the structural model given the fact that I will need to use a pipe .322 inches thick?

Thanks for your help.

[Anonymous]

> ... structural model of about 20 member I built ... > ............... > How should I build the structural model given > the fact that I will need to use a pipe .322 inches thick? > Solid FEA is suitable of machined fittings, castings, etc. Analysis of the type you want is done using elements that mimick the members; for your model solving for fewer than 40 (if there are no plates; two per bar / beam element minus common) nodes vs. hundreds of thousands or millions. Possibly if after analysis using suitable methods you have a weld joint or something of the sort that's of interest the solid FEA might be useful. If you hunt around a bit (search Google) you can probably find freeware to analyse your structure as well as a lot of useful info ... http://www.dermotmonaghan.com/ http://analyst.gsfc.nasa.gov/FEMCI/femcibook.html http://www.eng-tips.com/threadminder.cfm?pid=727 Seems I have a fuzzy memory that MSC's eval lasts indefinitely, but it limited to 300 or so nodes so looking into various vendors trial versions might be worthwhile for educational (you don't want something that's going to timeout in 30 days) purposes. Oh, and if you do use a preprocessor that's capable of geometry import you might want to use Acad to create a wireframe and export. IV's no help for that, either. Good luck with it. ===================================

[Roxana]

Thanks for your feedback.

Would you happen to know which MSC product would be adequate? I see they have many products, like Visual Nastran 4D, etc...

[Anonymous]

> Would you happen to know which MSC > product would be adequate? I see they > have many products, like Visual Nastran 4D, etc... Actually, I'm not up on the current products. I'm running NASTRAN for Windows, which I think has evolved into VisualNastran for Windows. I was thinking that the 4D stuff was a subset of the FEA with parts of Adams dynamic analysis (kinematics and rigid body dyanamics) added, but I don't know. I could be going on some bad assumptions and my idea of what you want may be out in left field (I do aviation related structures which are mostly shell / plate models vs. bar and beam structures and don't really claim to know a lot about any of it), but I think you'll want to make sure whatever you get has a full featured pre / post processor vs something that uses a CAD system interface unless you want to go with a fully intergrated system like Pro/E / Mechanica (? no idea really, just guessing. Maybe someone will correct me if I'm wrong). I also don't know what context to put your needs in; student, business related, what sort of structures, etc. You might do well to ask on an analysis forum and describe you situation and requirements. (MSC might not be the most cost effective choice, either, if purchase is the goal. I only mentioned them for what I thought I remembered about the trail version.) ============================

[Anonymous]

Roxana, As you've found out, AIP Stress Analysis has a difficult time meshing a thin-walled tubes that are very long since we use solid elements exclusively. ANSYS Workbench utilizes different types of elements, including Shell elements, which are perfect for this type of analysis. On this type of geometry, a good mesh might take hundreds of thousands of solid elements (and eventually we might run out of memory). However, a good mesh on this type of geometry can be made with perhaps just a few hundred Shell elements, and probably get better results than the hundreds of thousands of Solid elements. Cheers, -- Hugh Henderson (Inventor Workflow QA) "Roxana" wrote in message news:3993784.1102237436556.JavaMail.jive@jiveforum1.autodesk.com... > Hi: > > After careful attention to your comments and analyzing my model. I seem to recognize that the problem relates to the aspect ration of one of the structure members. There is one pipe 60 ft long and .322 inches thickness. This indeed seems to be the problem. > > However, how can I get around this issue with Inventor? I tried to use 12 sections 5 feet long each instead of one section 60 foot long, but this did not work either, even though a mesh can be generated for one 5 ft section alone... > > How should I build the structural model given the fact that I will need to use a pipe .322 inches thick? > > Thanks for your help.

[cfresh4u]

I ran into the same problem as you but I found a work-around for it... Follow my steps below...

1. I exported the orignal part to an IGES file.
2. Opened a NEW part drawing.
3. Re-introduced the IGES file using the import function.
4. Added back all the material properties.
5. Go to Stress Analysis.
6. Click on Stress Analysis Settings.
7. Click on Preview Mesh.

If it gives you a solid part with Mesh around it you are now ready to add your Loads & Constraints. If this does not work you will have to make sure to MAX out your Virtual Memory Page. If your Virtual Memory Page is a low number by Windows default then you will have to clean up your hard drive by deleting ALL temp and un-wanted files. Doing this will increase your default Virtual Memory Page file size.

8. Add your Loads & Constraints.
9. Click on Stress Analysis Update.

And whala! Success! You have your part analysis!

I hope this helps!
Cesar

[xavierl]

This type of problem is best done using beam elements. The answer will be just as good with only a few elements.
The full ansys module can also change a thin walled pipe into mid level shell elements (ie not solid elements)
Mostly using fea you steer clear of brick solid elements due to the huge processing demands.As mentioned above this should be reserved for complicated castings,etc.
regards
Frans X Liebenberg