Need advice on choosing FEA software 5

[semni]

My company is planning to purchase some analysis software and I need help in deciding which product (or combination of products) is best for our specific needs. We use Pro/E for our mechanical design work, and most of the parts we design are Aluminum. We are primarily interested in the stresses and deformations that our assemblies will face when subjected to shock and vibration conditions at temperatures ranging from -40 C to +70 C. We are also interested in heat transfer analysis.

We have virtually no prior experience using FEA tools, but we want to have some capability to do this in-house. We will be using the analysis software only occasionally, and cannot justify a dedicated FEA Engineer. Our budget for the software is fairly generous, but we don't want to invest a lot of time or money on training. Please offer me your advice on which product (or combination of products) makes the most sense for our circumstance.

 

[GregLocock]

The quickest learning curve in my opinion is MSC Nastran for Windows, and it has modules for transient and thermal analysis.

However, it probably makes more sense in the first place to get something that will integrate more tightly wth ProE, but I have no experience with that... and for what it is worth we do NOT generally use the FEA that is integrated with our solid modelling package.




Cheers

Greg Locock

 

[forkhandles]

I have MSC visualNastran for Windows and would agree with this choice, but would give serious consideration to nx Nastran. Its the same thing, but you might get it cheaper or with more features for the money.

 

[corus]

Greg is correct in saying that it is not common practice to use solid modelling CAD software with FEA. In my experience it's more trouble than its worth even though management have the view that you can do the CAD drawing and press a button and out pops the stresses without having spent any money on people or training. If, however, you want to go that route then PRO/Mechanica is probably scraping the bottom of the barrel in terms of capablities, but it integrates well with Pro/E.

 

[RdJ]

Like GregLocock and forkhandles said. The best available is most probable the MSC VisualNastran 4D, Which gives you an 3D drawing with color coded stresses. This can also be used with various different CAD programs Ex. Solid works. Although it is very user friendly, it is most probably much easier to do the stress analysis with intergrated CAD software.

Riaan de Jager
Mechanical Projects Engineer
British American Tobacco Manufacturers
South Africa

 

[pjhype]

I have used MSC Nastran 4W since its initial release, and would agree that it is easy to use. What makes it easy to use, however, is the pre/post processor, FEMAP. The ownership of FEMAP has changed hands a few times over the years, but the code has managed to keep up with the latest trends in geometry modeling. The basic finite element "engine" in Nastran hasn't changed much over the years, however.

I would recommend that you settle in on a pre/post processor (like FEMAP) and then look for codes that FEMAP supports (there are several). It appears that what you are trying to analyze is not too complicated and that all of the FE codes on the market should be capable of solving. If you want the FEMAP pre/post processor and a solver something like MSC Nastran but don't want to spend the money, you should look at CAEFEM, which offers a full range of capabilities at a fraction of the cost of MSC Nastran. CAEFEM is available at the following website:

http://www.caefem.com

pj

 

[JKEngineer]

I do not generally do mechanical/stress analysis, rather mostly heat transfer either stand alone or in conjunction with infrared thermography. My primary FEA tools are CosmosWorks and Cosmos/M. CosmosWorks is integrated with SolidWorks. They support the pretty pictures described above and give the types of results you are seeking. They can export to Cosmos/M for more rigorous analysis. I find them generally fairly easy to use and straightforward.

Jack M. Kleinfeld, P.E. Kleinfeld Technical Services, Inc.
Infrared Thermography, Finite Element Analysis, Process Engineering

http://www.KleinfeldTechnical.com

 

[corus]

If you're buying software then I'd advise that you buy a single package and not separate pre and post processors. These can be more trouble then they're worth as they attempt to convert the data into the solver package. In my experience you have to sort out the bugs in their packages as well as the bugs in the solver package, and all programs have them. It also increases the amount of time required to learn separate packages.
ANSYS and Abaqus have full pre and post processing capabilities together with the solver in a single dedicated package. These can also read in CAD drawings if that's what you want. Nastran I don't know, but that is also popular.
Personally though, if you don't have someone who knows what they're doing then, as with most computers, if you put garbage in you'll get garbage out. Training can be as worthwhile an investment as much as buying the right software.

corus

http://www.corusresearch.com

 

[vorwald]

I recommend purchasing NX Nastran for Femap from eds. See

http://www.eds.com/products/plm/nastran/femap.shtml.

Femap will import files from Pro E, but I don't know how well it works. I have meshed models based on CATIA models, and there tends to be some back and forth refinement required to get the CAD model compatible with FEMAP.

 

[zcong]

Hi, semni,

As far as I know, ls_dyna is good at shock analysis, but still there are some limitations for its application, and i think such limitations are due to the assumptions it bases, this is the case for any computing software.
And if you want to do some nonlinear analysis and want to get a farely accurate as well as reasonable result, I reconmmand ABAQUS, since v6.3 is a great solver than v6.2, also it can offer a complete solutions for contact problem. But i dont know how it is for heat transfer problem.
and I was heard from my friends that ANASYS is good at couple-field problem, maybe you can try to refer to the theory manual of ANSYS since it publishes independently.
Good luck!

 

[CALATO]

Hi Semni

I have experience working with Samcef V10 and with the new great interface Samcef Field (more user frinedly). I'm not sure, but i believe is cheaper than ansys , abaqus , MSC Nastran, etc.

http://www.samtech.fr

 

[Softedge]

I use Solidworks 2004 and Algor FEA with an InCad add-in that automatically launches the Algor software and creates the mesh. I am pretty sure there is a similar add-in for Pro/E. Talk to Ed Simmons at Algor, he will be able to give you more information. The software is easy to use and the conversion from the solid model occurs without problems. The mesh generation engine appears to yield reasonable results. The software allows you to modify the mesh.

 

[jeeleong]

Hi Semni!

I just joined the forum so my suggestion might be a little late. But anyway, every software vendor may tell you that their product is best and it can do lots of things. Usually this confuses us further as to which is best. In my opinion, you should pick a few software vendors (maybe one or two or three would be sufficient) and ask them to help you run some examples of the things that you do as an evaluation process. Compare the results with for example your experiments or any data or calculations. Then just get the software that can do the job properly. Of course, good technical or after sales support is also important. I hope this helps.

 

[DickButkus]

Of all the solvers, (and someone can correct me if I am wrong), I believe that only ABAQUS provides an explicit and an implicit solver. The explicit solver excels at shock and impact, while the implicit solver does the traditional non-linear (and linear) analysis including heat transfer. Both of these solvers have a single user interface wrapped around them. Results may be transported between solvers if necessary.

I believe that ANSYS has a license to resell LS-Dyna so that they can sell you an implicit and an explicit solver, but I don't believe that these models can be built from their preprocessor and I doubt you can send results between the two.

I also believe that MSC will sell you Dyna along with their solvers. I am guessing that Patran will write Dyna decks as well as decks for MSC products as well.

Besides these, I think it will be tough to find an integrated explicit/implicit/pre and post processing solution.

Best regards and good luck!
-KF9RI

 

[johnhors]

LUSAS has both explicit and implicit solvers within the same pre and post processor user interface, check out

http://www.lusas.com