MSC vs. NEi 1

[Murree21]

My company is looking into buying one of the Nastran's. From everything I have read I am leaning towards NEi for their support and capabilities. During a conversation with an engineer from a large aeronautical company I just asked in passing which program he preferred. His answer was, "Do not waste your money on NEi, spend the money on MSC. Noran works well for simple linear problems but falls apart on more complex nonlinear solutions." Is there any truth to this statement? Is NEi still catching up on its complex solutions?

 

[johnhors]

MSC bought MARC some time ago specifically for non-linear analysis because MSC/Nastran is not good for it! I think that you got a very biased point of view from this aero guy. MSC having been receiving a lot of criticism lately on this web site, and not for a lack of good reasons.

Have you also considered NX/Nastran and The MacNeal Group version as well (the original "M" of MSC)

Perhaps even more important than the solver decision is the pre and post processor you get, MSC will offer you Patran, whilst NE and NX use Femap and TMG use Venus. If you've never used Patran before, then your choice of first vendor to eliminate is a no-brainer !!

 

[Murree21]

Since I am new to all of the Nastrans, choosing one based on which pre/post processor I know isn't a factor. However I would like to choose the program with the quickest learning curve.

You say that MSC bought Marc for nonlinear analysis because MSC.Nastran is not good for it. Does this mean that all of the Nastran lines are essentially "not good for nonlinear analysis?"

 

[johnhors]

If your work would involve complex non-linear analysis, then you should seriously consider using either Ansys or Abaqus, as these are generally accepted as the best solvers for this work. For example, contact analysis in Nastran (of any flavour) was possible using the CGAP element, but this is really limited to a non-sliding contact. To overcome the limitations of the CGAP element, MSC have recently introduced a more general contact capability which actually calls up the Marc solver in the background! NX/Nastran does something very similar and calls up the Adina solver, whilst NE have truly incorporated a general contact capability into their version of Nastran.

Sorry, but I really have to disagree with you on your statement "choosing one based on which pre/post processor I know isn't a factor". It is a factor, a very big factor, especially these days, where a lot of FEA being done uses CAD embedded systems, which are very very easy to use (IMHO too darn easy!), and as a result people are running analyses, who simply shouldn't be! My point being, everyone expects a relatively easy to follow, easy to learn, intuitive windows interface. I need say no more.

 

[GBor]

johnhors makes a couple of excellent points including some buried in the subtleties. Ansys and Abaqus are very good at highly non-linear problems...the explicit solvers are fast and accurate. Perhaps accidentally buried in the post is that

whilst NE have truly incorporated a general contact capability into their version of Nastran

the others have purchased the capability. You definitely need to consider the pre- and post-processor because that is where you will spend the vast majority of your time, but you also need to consider that NE is obviously investing in improving their software...what are the others really doing? Do they even understand their own "black boxes" of processing?

Great post, John, and considering your own FEA programming prowess, your objectivity is always impressive. Of course, your "ease of use" soap box is well appreciated on my side of the Atlantic!

Garland E. Borowski, PE
Borowski Engineering & Analytical Services, Inc.
Lower Alabama SolidWorks Users Group

 

[JacobL]

as a side note, MARC does the job quite well.
The only problem with MARC is the price, which is quite higher then the one of nastran.
the good thing about ansthat it's all in one solution, but MSC has specific apps for different calculations - MARC, NASTRAN, and DYTRAN for really cool stuff.

also, if you need to do CFD, then get ANSYS. MSC has no CFD capabilities (excluding limited CFD of DYTRAN).

 

[ThomasH]

First, when you say "complex nonlinear", what does that mean?

Traditionally MSC.Nastran is not a top quality software for that, that's why they added Marc. And there are things you can do in NE that you can't do in MSC in the same manner. And so on.

There is also a tendency from MSC-people to look at NE as a little brother. So bevare of those comments.

I can to some extent agree that the choice of pre/post is not critical. Bulkdata for all Nastran's are very similar so it will probably work.

But on the other hand, your working environment is the pre/post. To buy a solver and see what pre/post comes with the package would in my opinion be a misstake. I would say, test both or "all" alternatives. If there is no demoversion available that might give you a hint regarding the vendors confidence. But don't spend money based on a nice demo from a salesperson.

Good Luck

Thomas

 

[jakeline6]

Murree

Does the engineer in this "large aeronautical company" actually use NEiNastran? The comment almost sounds like something another FEA vendor would say and this guy is just passing on.

We use NEiNastran's non-linear all the time and it was one of the main reasons we have switched from MSC. With MSC and NX you have to purchase expensive add-ons to get the advanced non-linear which should be and is in the NEiNastran basic non-linear package. NEi developed their own surface contact which is as good as that in ANSYS or ABAQUS (we also have ABAQUS). They just released automated surface contact which you will want to look at. Also check the noran’s multimedia library and the engine block model.

http://www.nenastran.com/newnoran/animations

I would consider this a large complex non-linear model.

If you are also considering MSC, then you are looking at Patran for the pre/post processor. There are many interesting discussions on this versus FEMAP in eng-tips I would look at. MSC Nastran is not a bad product. I just think NEi is better and particularly for non-linear which is why I am so surprised someone would make this comment.

 

[A6FEA]

I recently purchased NEiNastran and also was very impressed by the non-linear analysis capability and handling of large solid models with surface contact. MSC never had surface contact and had to purchase MARC to get it. Also, many of their element types are still not supported including CQUADR. NEi supports it. If you want the advanced nonlinear in MSC you will be using a different product and the purchase price goes way up. As for NEi they have caught up and passed MSC as far as I am concerned.

Murree21, I am interested in your comparison between NEiNastran and MSC. Also, who is this "large aeronautical company"? I assume they use MSC primarily.

 

[fkmeyers]

I also work for a large aerospace company. I do a fair amount of nonlinear and I have seen models which run fine in NE Nastran and then have trouble in MSC or NX. So it can go both ways here. If you run a model where the NLPARM settings are set for MSC (MSC defaults), it may have trouble in NE Nastran. I have seen this a few times and when I set the defaults for NE, it usually runs fine. Most models I build do not have any issues but some require tweaking the NLPARM settings. I had to do the same with MSC when I was using that before. I have yet to build a model I could not run in completely NE Nastran.

NE has set up their own nonlinear and have evolved it towards what other high-end nonlinear packages are offering. I have built some real complicated contact models which I would never have been able to do with MSC or NX and their gap and slide line contact. If you want to spend the extra money and time learning ABAQUS or MARC or ADINA, then I am sure these programs will also handle complicated nonlinear very well. What is your budget and what types of models will you be running?

 

[Murree21]

This large aerospace company uses MSC as their primary solver. As for my budget, the smaller the better (of course) and my models will be 90% linear solutions. Basically, I am looking for a good solid product with an easy to learn pre/post, but I do not want to come across a nonlinear solution 6 months down the road and not be able to solve it.

I will be demoing both products before I make my final decision, but if what I am hearing is true Noran just seems to fit my needs better than the rest of the field. Thanks for all the help. Getting feedback from actual users of the programs is priceless to me, keep up the good work.