FEA Sofware Reccomendation 3

Savis,

If you're looking for a recognized leader, with coupled-field capabilities, and a sound element library, I believe ANSYS is you best choice. The pre-processor is not very good but you've got that in Solidworks. The solver is good.

BatMan2

Thanks BATman,

I am new to this particular field. The preprocessor capability in Solidworks refers to??

I will be getting training with the software purchase.

If you are looking at doing anything with contact between parts then you need to be looking at full featured packages such as ANSYS, Nastran, Marc, and ABAQUS. Of these companies, MSC is probably the largest followed by ANSYS. ABAQUS is privately held so you can't look at their books.

If you want to include the nonlinear material behavior, drop impact, or other more advanced applications you need to look at ABAQUS since they give you both a standard and explicit solver in one environment.

You will want a package that can read results from your mold filling simulation so that you can get the proper material orientations if you wish to do cooling and warp simulations.

Cosmos and Algor are fine starter packages for simple problems, but don't expect either one to have the depth of the mainstream packages.

Good luck with your decision.

Best regards,
KF9RI

Are you saying you can't do contact analysis with COSMOS ? See They certainly claim to be able to do this, and most other things too. COSMOS may not be as powerful as NASTRAN or ANSYS, or have the track record etc., but I don't think you can make such a blanket statement. If you look at their user list, its not too shabby. COSMOSWorks is very limited of course - perhaps you are thinking just of that ?

take a look of rockfield fea software at It might be smaller company but
it is based on swansea fe development team which is doing serious job.

I have been pretty happy with Algor after about a year. It has a good user interface and imports/meshes Solidworks pretty well.

Looked at Cosmos also, beware that for more advanced analysis (can't remember details but I think nonlinear materials and dynamic response) you have to go into a different older command line environment.

Algor claims to be good at contact analysis and multibody physics, couldn't comment on performance here versus the big $$ guys.

Personally, for contact analysis, I prefer MSC Nastran or Patran. And if you are needing a non-linear analysis, MSC MARC is also very good. MSC has a good pre-processor and modifying the mesh is much easier than most.

If you are looking for a decent Multiphysics software, Algor has a decent package. Also, MSC's 4D is pretty good.

Just be careful about using the Multiphysics packs. They still do a stress analysis in a linear mode and as everyone knows kinematic stresses are non-linear by nature and definition.

John Petty
Sr. Mech. Designer/FEA Analyst
Clarke-Alto

If you want one package to serve all possible FEA needs, ANSYS is very good to outstanding in every area of structural analysis. If you must have the best possible answers for contact or non-linear material problems, ABAQUS is the leader. NASTRAN is acceptable for everything and remains the leader in dynamic (vibration) problems. If you want easy to use Mechanica is suitable for complicated geometries that can't be meshed any other way and only for linear stress and vibration problems but Mechanica is not appropriate for computing contact stresses. I haven't tried it yet but I have been told that the ANSYS Workbench makes it as easy to use (and abuse) as Pro/Mechanica. I don't like Algor because their codes have never been state-of-the-art.

Doug

ANSYS, NASTRAN, COSMOS - I don't see a big difference in their capabilities- in terms of what a user can get out of them. They all are General purpose FEA packages. Linear analysis and run of the mill non-linear problems (including contact, small strain plasticity etc) COSMOS should do just fine. For the really advanced non-linear problems you might be better off going with a product like ABAQUS or LSDyna.

I strongly recommend to you "ANSYS".it has a complete element library and has different capability to solve linear and nonlinear analysis.you can use solidworks as geometry modeler and export your model to ansys by ACIS format.you can easily setup the solver(frontal,pcg or other methods) and change the amount of minimum requirment memory but in other software like NASTRAN sometimes you should have change the DAT file adding or deleting the diffrent lines by specific formats.you should consider that NASTRAN is only a solver and you should use PATRAN as pre and post processor but you have integrated envioremment in ANSYS.

REZA

I use IDEAS linear and nonlinear which is integrated preprocessor solver and postprocessor. I have not had problems or defficiencies with it . But it seems that either I am missing something or IDEAS is. Can some of you please tell me why there are no pros or cons about IDEAS in this thread?
thanks, fsi

If you have mastered the solid modeling portion of SDRC ideas, I take my hat off to you. It is the most quirky, non-intuitive and difficult to learn solid modeling software that I have yet come across. Because of this I never mastered the FEA portion before we dumped it (I wasn't the only one )!. Could this be the reason for its apparent lack of popularity ? (We can't all be geniuses!)

Hi Savis,

Sorry for the delayed post, but I just joined this site. I am replying because I fear that one point may not yet have been made: a big part of choosing your FEA package comes from how much you plan to use it. As with everyone else, I can only comment accurately on packages that I have used recently (in my case, Algor and ANSYS).

If you will use FEA only once in a while for specific tasks, then get one that is easy to use, cost effective, and from a company that has decent tech support. If you've got an unlimited budget, you can go with ANSYS and use Designspace (which I have not used, but is included with the full version now and is supposed to be easy to use), but at a factor of 2x or 3x $$$, that really seems like a waste of money to me. Thus, in this case, I would recommend Algor -- the auto-transfer process works well, the mesher is decent, and it has a very short learning curve for this type of software.

If, however, you are going to be using FEA often, then you would be well advised to consider ANSYS because once you get up the rather substantial learning curve I doubt that there is a more flexible and powerful PC-based package out there.

FWIW, I fit more into the first category -- I will sometimes go months without needing to do much FEA, followed by several weekes of almost full-time modeling -- so I use mostly Algor.

Regards,
Greg

First to answer fsi: In the 1980's we used to use IDEAS as a 2D mesh generator for nonlinear analyses in ANSYS. However, when they introduced the Master Series it became much more complicated to use and we hated it. When the ANSYS mesh generation capability caught up and surpassed IDEAS, we didn't need it anymore.

Now back to savis: Code selection depends on who will be using it and how often. The classes you will take will teach you how to use the code, but not how to do analysis. If the code will be used occasionally by designers with little or no analysis background, you should stick with a simple easy to use code. You should avoid the high end codes unless the user(s) has the background to validate and verify an analysis with hand calculations and simplified sample problems while also knowing how to properly interpret the results.

--kan

Desktop Engineering magazine < reviews different programs. For example, the August issue looked at NE/Nastran, <

The choice of software should be driven by what you need to do with it. Look at typical jobs that you will need to analyse and draw up a list of things that you may need to include (e.g. contact, thermal, stress, vibration (type), 2D or 3D?) and how you want to use it (how often, in CAD, expert user?). Keep a thought on your future needs too.

Use product literature to eliminate any packages that do not meet your requirements. Next interview sales reps of short listed packages, don't be taken in by flashy demos; give then a simple analysis based on your requirements to solve. Also consider what the after sales service may be like.

Buying an underspeced system may mean extra expense, delays or even not being able to do what you need to.

Buying an overspeced system will mean extra software and training costs and possible inefficiency.

In summary be driven by your needs and then use the above recommendations to help you make a decision. The cost of getting the choice wrong can be great.


TERRY

After reading all the posts, this may widen the tolerance band some.

I'd agree you need to:

know what the budget is
know the capabilities of the User
know the weightiness of the projects (FEA to be used 10% or 70% of the time)

From a thermal analysis perspective, I've had Algor for 2 years and have yet to get useful results from it. The meshing and overall learning curve have been very painful and disappointing, given the price tag. The software seems to solve slowly and, in meshing, seldom converges to a usable mesh.

This has sent us shopping for:

- $0-10,000 package
- something that can solve electronic package analyses in less than a day
- a reasonable learning curve for once-in-3-month simulations.

We're considering Flotherm, Cosmosworks, and Harvard Thermal. HT I don't care for much, since it doesn't have a direct 3d importing capability and has an interface like Algor's for building models.

Cosmos seems faster and more intuitive for use.

Am beginning the eval on Flotherm.

Can anybody comment on using FEA packages Strand7 vs ANSYS vs CosmosM. I'm new to FEA & would like to know your ideas on these packages.

check out DEFORM.


I have used it. Pretty impressive.