Which to choose? Cosmos, Nastran, Catia, Ansys Abaqus... 1

[arranot]

Hello all,
Os story that is history. I work in the company are considering implementing a computer program using finite elements. And I am told that I report on which is best suited for our needs. The company manufactures machines for the steel sector as boxes rolling, straightening, shearing ...
Usually what you use to perform linear static calculations. And perhaps some of fatigue.
What program do you think is best for us? Cosmos, Femap, Nastran, Catia, Ansys, Abaqus ..

thanks

 

[flambter]

It always depends on your own preferences and the type of problems you have to sovle. They all have their pros and cons. Knowing mostly Catia V5 and Nastran/Patran, Catia V5 is nice to make simple 3D solid models if you design with Catia V5. Otherwise, I would stay away from it unless you get the package of AFC (Abaqus for Catia V5) which now could be good. But you are stuck with Catia V5 mesher which is terrible.

MSC Nastran/Patran, pretty powerfull for beams and 2D. Mesher is so-so unless you have a lot of knwoledge on how to make it look good but then it can become very complicated. Not user friendly. Can do non-linear analysis.

Your solution can be a mix of some softwares. Nastran solver, Hyperwork mesher (nicest mesh I have ever seen!)

You have to test them for your own needs and make your choice. FEM company are usually happy to come to your company and make demonstration and participate in pilot projects. So use them!

 

[arranot]

You say that the program MSC Nastran/Patran can become very complicated. Is that true with FEMAP-NX NASTRAN?
thanks again

 

[rstupplebeen]

Their are many other threads about this same topic. Check them out. In the end it's cost vs. features vs. usability. I hope this helps.

Rob Stupplebeen

 

[flambter]

Don't know about FEMAP. MSC Nastran/Patran becomes complicated to me, but ask an expert and he will tell you otherwise. That was only my own little opinion.

 

[BlasMolero]

Dear Arranot,
I run since years the suite "FEMAP & NX NASTRAN", this is the best price/performance you have in the market, very good professional people behind the software, with more than 25 years of developping FEMAP. Also very good VARs all around the world. The basic package include linear static, modal frequency, buckling, heat transfer, and nonlinear analysis.
Best regards,
Blas.

 

[flash3780]

I can speak to Pro/Mechanica, Ansys, and ABAQUS as those are the programs that I've worked with.

Pro/Mechanica is probably the simplest to use, although it (in my opinion) leaves a bit too much of the analysis up to the computer, and can come out with some screwy results if the automesh doesn't work right. They do, however, use an innovative approach to FEA in general. They use a P-method solver, so rather than increasing the number of elements in areas of higher stress, they instead use smarter elements. I expect that future solvers will combine this approach with the standard H-method.

Ansys is probably the most popular finite element solver, and their new Workbench product puts a great front-end on the solver. Classic Ansys has plenty of functionality to deal with more complex problems as well. Unfortunately the Ansys Classic GUI can be a bit tricky to pick up. Still, Ansys is a great option for an implicit FEA solver. The one thing that I wish that they had in Workbench is a command line.

Abaqus/CAE is my favorite of the three, and has a very intuitive GUI interface (which is worlds better than what they had 5 years ago) and is also quite powerful. Abaqus does a good job of handling contact, plasticity, and other non-linear problems. From the sound of it, you'd probably be most interested in their Abaqus/Standard solver, but they also offer an explicit solver for looking at dynamic problems. Abaqus also has the ability to incorporate Python scripting into a model, which gives the user a lot of flexibility in their analysis.

 

[loki3000]

ansys workbench would be my best bet, if you're only doing LSA. it also has a fatigue module.
for LSA just about anything will do from the numerical point of view, even Solidworks simulation (and it too has fatigue)

 

[4Pipes]

I use Autodesk Algor Simulation. I purchased it when it was just Algor. I am more than happy with the flexibility.

I would think carefully before looking at options that have "built in" solid modelling. Their "forecourt test drives", bi-directional modelling and visual appearance can be extremely impressive and very tempting. If it does what you want then great but you have got to make sure it will do what you want. It might seem obvious but you don't need FEA for easy shapes - Will it work the difficult shapes? I know folk who have become unstuck on this.

I like Algor for that purpose. I can either do my own solid modelling with Alibre - Yes it does work well for most things. If it does not I still have the flexibility to get models drawn up with anything I like. The only caveat would be Autodesk's very long term intentions. At the moment, and at least for the next 2011 version being beta tested, it is still standalone and unchanged from the original Algor multi-source interface. I have no reason to believe this will change but I would ask the question.

 

[gwolf2]

Decide on your budget and then pick one of these in descending order of preference:

(CATIA IF IT COMES WITH ABAQUS)
ABAQUS CAE
PATRAN/NASTRAN
FEMAP/NASTRAN
ANSYS WORKBENCH
COSMOS, ALGOR, NISA etc.

 

[arnie333]

Hi

Is the "CATIA WITH ABAQUS" the Simulia range ?

Thanks

 

[cooperjer]

Hello Arranot,

There are several factors to consider when purchasing an FEA package. The factors I used were as follows: Price, Ease of use with the UI, Customer support times available, Online tutorials or support, and speed of processing a complex geometry model with simple supports and loads.

I've used Algor and ANSYS in the past. Both were user friendly and had a lot of customer support.

Thank you,

Cooperjer
Mechanical Engineer

 

[GregLocock]

You might also consider that it is a good idea to use the same package as your main customer or supplier. Admittedly that is less of a consideration with FEA than with other CAE products.

Cheers

Greg Locock


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

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

 

[gwolf2]

> Is the "CATIA WITH ABAQUS" the Simulia range ?

Yes and no. I use the "CATIA WITH ABAQUS" term to suggest CATIA for the modelling environment and get the ABAQUS enhanced FE option for CATIA for the analysis - an extremely powerful modelling environment and great FE but possibly hugely expensive.

ABAQUS/CAE is more commonly known as Simulia these days but the modelling environment is not CATIA (although it has many user interface similarities and possibly shares some code with CATIA underneath because of the historical link between Dassault and HKS.

> I've used Algor and ANSYS in the past. Both were user friendly and had a lot of customer support.

Wow! You should get out more!

For those familiar with my past stated references note how PATRAN has dropped down my list these days. The Heresy!

Gwolf

 

[johnhors]

Gwolf,

I totally agree on your comment about Abaqus enhanced FE in catia.

However I do not believe that Abaqus/CAE actually shares much code with Catia. The obvious similarity is that CAE adopted the Catia axes triad, but beyond that what is there that is common? Any linking of the two softwares would only have commenced when Dassault bought out HKS. CAE is still developed in Rhode island and Catia is developed in Paris. CAE is based on the ACIS solid modeller geometry kernel which has nothing in common with the Catia solid modeller. Indeed import of Catia geometry into CAE can still be problematic (especially legacy Catia V4 stuff that has previously been imported into V5). CAE and Catia also use very different meshing technology.

As for Patran, with venture capitalists that now own MSC, its future has got to be uncertain.

http://www.Roshaz.com

 

[gwolf2]

On PATRAN I recently met one of the clever buffoons (now unemployed) who buggered it up. Sadly he had not been responsible for the contour plotting re-write so I had no immediate excuse to punch him.

I thought of lots of other reasons later on when we had parted :-(

 

[kellnerp]

If you are using SolidWorks then you might look into NE/Nastran as they have an integrated FE package. The ease of use of CosmosWorks with the robustness of Nastran.

TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

"Node news is good news."

 

[SAITAETGrad]

For what you are doing, I have to ask why you think you need FEA at all.

 

[johnhors]

I have to ask why you think you need FEA at all.

To get the meaningless pretty colour (color) pictures that ignorant management love to see of course!

http://www.Roshaz.com

 

[gwolf2]

Of course. Don't forget to switch on all the stress averaging you can, switch off the mesh outlines so they can't see what the mesh is like and use a graduated contour scale. Work of art mate.