FEA program opinions? 19

[MechElement]

Hello, I'm in the market for a license of SWx, but I would also like a good FEA program to accompany SWx. I've heard about NENastran & COSMOS running with SolidWorks, but it appears everyone has a different opinion about them. I'm a mechanical designer and working in the Middle East. I'm returning to the States for good in August to go back to school and finish my BSME.

Which FEA program works best with SWx?

 

[rtmpxr]

My dear GBor,

First of all, I spoke frankly, as you maybe observed, and I didn't attack anyone. If you beleive that, then you are in a wrong position.

PS About hiring, think twice, you'll never know...

 

[GBor]

There were no blatent attacks, but I do not believe anyone answered the original question with "sublime ignorance", nor do I believe the original statement regarding Ansys warranted your "excessive hope" for support. You made it clear that you were issuing your opinion and that's why this forum is here.

As for hiring, my #1 client is my former employer...not only do I already know, I wouldn't have it any other way.

fkmeyers,

Thanks for the support.

Garland E. Borowski, PE

 

[GBor]

rtmpxr,

Incidentally, I didn't see your support for the following:

"In general, Ansys appeals to the lower end of the market."
I disagree with such consideration.

Why exactly do you disagree...I haven't used Ansys, so I'm curious.

Garland E. Borowski, PE

 

[rtmpxr]

Evidently, you missed completly the point in my statement. There were two aspects.
First. Yes, I disagree with the way corus gave the sentence without any explanation. This was a valuable help ? In my opimion, not at all. And yes, I cheched before who corus is and I saw that is a person with high activity. Therefore my expectation was clearly motivated.
Second. The "sublime ignorance" made reference to THAT TYPE OF QUESTION (the original was FEA programs opinions), and NOT to the answers of the members in the forum. Search the forums and you''ll find a lot of "these questions" in disguise. And I stated "that type of question" didn't receive the right answer. What I did was to gave a such answer. Please, read again my post GBor.
Finally, I didn't intent to prove any "superiority". Others did. Ask them.

 

[corus]

As GBor states, people make opinions here. Some people have the opinion that Abaqus is a very powerful program for the 'harder' type of problem. Presumably then it has greater capabilites and appeals to the higher end of the market, ie. it has less mass market appeal. Other programs such as Nastran probably fall into this bracket too. At the lower end of the market, ie. those programs that appeal to the mass market, you probably have the freebies and then the likes of Pro-Mechanica etc., which are designed to be easier to use but in general have less capabilities. It seems to be the common opinion that the likes of Abaqus and Nastran have greater capabilities, and are at the top end of the market, though in my sublime ignorance I haven't listed these differences. In general, generally speaking in a general manner, I'd think that ANSYS has greater mass market appeal, but fewer capabilities than those programs that are generally at the top end of the market which can, generally, tackle those harder probelms, if generally required to do so, generally. To that end, if ANSYS isn't at the top end of the market then, logically, it must generally appeal to the lower end of the market, in my opinion, generally.

corus

 

[rtmpxr]

Ok corus, we cannot agree, generally speaking
I like Ansys because it helps me a lot in a field I will define a "harder" one. But I not refuse to use another tool which will give me better results. And I have a such example with ... yes, ABAQUS. And sometimes, I found Timoshenko's books to be my best friend. But this is another discussion and I will stop here.

 

[trainguy]

NE Nastran all the way.

The FEMAP pre-processor has literally enabled us to DO FEA jobs we would not otherwise have gotten (converting to and from ANSYS, for example).

On top of that, their support is second to none. They literally hold your hand all the way through, and they almost become a part of your team...

By the way, I do not work for them...

 

[fkmeyers]

I have had the same experience. There are many products out there that will do the job but the overall package...product and support from Noran has been the best for me and many of my associates that also have NEiNastran.

 

[iRocket]

Hello People,

I am given the task of evaluating and purchasing an FEA package for our company. Searching google, I came across your discussion group. I was impressed enough to join Engineering-tips. I would like to thank you all for providing so much useful information. I spent 3 hours in FEMLAB sale/educational course today. I had some questions that seemed not to have straight forward answers. We are making towing hooks, which are welded to the pulling vehicles. My objective is to find an FEA package that can handle "welds", non-homogenious material, and non-linear deformation. Of course it would be nice to look beyond my curent project and find something that can handle incompressible fluid flow.

After reading the above inputs, I am inclined to call Noran Engineering in the morning. All comments are welcome. Thanks in advance.

 

[fkmeyers]

I would call them. I have used NEiNastran for a welded tank application. We make tanks for aerospace vehicles (unmanned) and had an issue with some very bad welds of fittings on the domes of the tanks (LOX tank). These fittings were replaced with ones that were mechanically attached and welded as apposed to just welded. We used NEiNastran to analyze the welds and bolted connections of the fittings. We modeled a sector of the dome and the entire fitting. In this application we used all solid elements to model the welds and fittings. It was few years back and I am not sure if there were dissimilar materials but that is easily supported. We had to test the actual hardware and compare FEA results to test ones and they were very close. A rupture of the weld would have been catastrophic with the old design but with the new one it would have just leaked which would be bad as well. We instrumented the weld and the test results agreed well with our Nastran model (within 7%).

Another feature we have used for modeling welds in NEiNastran is their surface contact weld elements. These are not as good as continuous meshed parts for stress around the weld but are very easy to set up and modify. For example, we need to locate the fittings on the dome and did not want to build a bunch of models so all we did is build one model and used surface contact to connect the fittings to the dome. Then we could just move them around in FEMAP without re-meshing anything and run different configurations to assess the optimal location of each fitting. This was a huge time saver. Once the fittings were located we created the continuous meshed models for test support and flight loads verification.

 

[Drej]

Whilst I agree with a lot of the above comments (this is an excellent thread btw), I couldn't help but pick up on what corus + others state about ANSYS.

> In general, generally speaking in a general manner, I'd think that ANSYS has greater mass market appeal, but fewer capabilities than those programs that are generally at the top end of the market which can, generally, tackle those harder probelms, if generally required to do so, generally.
> To that end, if ANSYS isn't at the top end of the market then, logically, it must generally appeal to the lower end of the market, in my opinion, generally.

I'd have to disagree with that very generalising statement regarding ANSYS' position in the market. Yes, ANSYS does have great mass market appeal, but in my view there are a couple of fundamental reasons for this: (1) its marketing strategy (2) its capability as a code/engineering tool. The marketing by ANSYS is very agressive, and they sell licences by the truck load, especially Workbench. Why does it sell so well? Because simply it is such an excellent engineering analysis tool that people enjoy and trust. Being a staunch ABAQUS man for many years (there are a few here too), I was "forced" to use ANSYS about 5 years ago, and admit I hated using it at first; it was tempramental, buggy and just didn't feel like you would ever get the hang of it. However, for me now there is no other tool I would choose to use. I've done so much linear and non-linear work with this code: seismic/shock (response spectrum/t-h), dynamic transient, modal, time-history, contact, plasticity: every time I use it, it's right on the money in terms of asking it to do what I want. It has everything: programming flexibility (APDL is amazing), functionality, high-end capabilities (user materials/elements, state of the art contact algorithms and element formulations...) the lot: just as ABAQUS does. Which brings me on to the second point. I don't believe that today's ANSYS code can be considered a low-end tool, not with the Multi-Physics/Mechanical environment. Absolutely not. A few years ago yes, I would've held my hands up to agree with you, and would've said ABAQUS beats it for high-end capability no problem. Things have changed now. ANSYS' non-linear capabilities are now up there with ABAQUS - I'm not saying they're better, because that's far too difficult to quantify, but in terms of putting them side-by-side and saying: "So what can each of these codes do?" there's not much between them nowadays. And, of course, something doesn't get mass market appeal in our industry because it's a low-end code, it becomes popular (in this context) because it works, and engineers/analysts/customers trust it. Engineers are such a critical bunch (!), in a dod-eat-dog business, and they don't use if they don't trust it.

There is a bit of disclaimer here though, and that is ANSYS Workbench. This needs some work (in fact, it needs lots of work), and probably falls into the category described by corus above. Fair comment.

I'd be interested to know whether the people who've given negative remarks about ANSYS have ever used ANSYS previously for any length of time?

Cheers,

-- drej --

btw: Congratulations to the boys at Liverpool FC: European Champions!! Well done.


------------
See faq569-1083 for details on how to make best use of Eng-Tips.com

 

[corus]

In answer to the last question, yes, but admittedly a long time ago.

One thing you did bring up is the number of bugs that you find in a program. Every program I've used has them, and will no doubt will continue to do so. It's whether or not you can get around the bugs or, if they're that bad, do you have to wait for the next version for that bug to be fixed, and then only to find new ones. Unfortunately salesmen don't show you the bug list when you buy a program.

Liverpool had lots of bugs by the way, but their customer support was excellent, generally.

corus

 

[GBor]

I would echo fkmeyers that you should call NENastran. Most general FEA packages have a way of handling welds, but NENastran has some great market appeal, particularly if this is your first FEA package purchase. NENastran is very versatile.

As for the incompressible fluid flow, make sure any package that you look at can handle the Reynolds numbers you expect to see. Most packages offer some level of Fluids analysis and they will sometimes call it "incompressible fluid flow", but they are generally very limited in their Reynolds numbers until you start talking to Fluid specialty softwares.

Garland E. Borowski, PE

 

[paulliu]

Among the structural analysis codes, I like ANSYS and
ABAQUS. MSC/Nastran has some special capabilities
to deal with aerospace structures.

For metal forming, and crash-type analysis, I like
to use LS-Dyna3D.

I used COSMOS, ADINA et. al., they have very
limited analysis capabilities, and I doubt their
claims for their code. If they are good,
why do they have so few users?

 

[mrmartins]

At our company we are looking for a FEA software able to deal with to main areas

- Aerodynamics
- High speed non linear structural events (like impacts)

the software needs to give reasonable results.
We have reduced our choice to ALGOR, ANSYS and MSC software. What is your opinion.

There is also the possibility of other softwares, please give me your advice.

There are also good solutions, like abaqus but they do not have fluid dynamics and buying to different packages forces to learn working in two different ambients.

Thanks

 

[dmacx]

I recommend NE/Nastran over MSC for nonlinear. If you are doing impact then you will need decent contact and MSC does not have this unless you are going with the MARC add-on. Doing that will violate your requirement for a common interface.

I am curious as to how you reduced your list of contenders to the 3 mentioned. Have you evaluated any of these packages? I was not aware that ALGOR or ANSYS had aerodynamics. I assume you mean CFD which is different as it pertains to Nastran.

With FEMAP (the NE/Nastran modeler) you would have a common interface for both the CFD with TMG and Nastran for nonlinear impact. Most FEA products have different codes for CFD and structural eventhough they are operated under the same interface.

I do not think high speed impact is one of Algor's strengths, neither is CFD so I would not recommend Algor here.

I would look at ANSYS and NE/Nastran if I were you.

Don

 

[mrmartins]

Hi, dmacx,

thanks for your reply. I did not understand your comment about CFD and aerodynamics. Is it important for the selection?...

Do you think NE/Nastran (from Noran Engineering) is a good software for my application?

And do you have other software in mind?

Regards

 

[GBor]

mrmartins,

I wouldn't so quickly dismiss Algor, but I would definitely compare it to NE/Nastran to see which is the better solution for your needs. I also see you posting this in another forum and discussing MSC.Marc. I am a proponent of NE/Nastran when compared with MSC products. NE/Nastran and Algor are bitter rivals in the FEA market, so ask some very pointed, deep questions of each of them.

As for the importance of CFD vs. aerodynamics, you can package different products in different ways and still have the same capability. This is why the questions need to be deep and pointed -- ask about HOW things are calculated...figure out the physics that the software is using. What generally comes along with the repackaging is ease of use for a particular application and an increase in price. For instance, most packages these days have some form of contact, but some packages separate it into extensible elements, contractable elements, springs, etc. All of these elements are generally based on the same basic element of a truss to which you can apply a stress-strain curve, but they make it easier and generally converge better unless you know what you are doing with the curve. If your application is specifically and forever will be aerodynamics, NE/Nastran has a specific set-up to address this area and it may be worth the pricetag.

Back to Algor, I've used it for many years in impact situations. I have dropped things, thrown things, and hit things, but generally under what I would consider low-speed. I would be leary of simulating a bullet penetrating a target, but not of an automobile impacting a crash barrier at 60 mph. I've also used Algor for 2-D flow over a wing...I'm certainly no expert, but someone that is seemed satisfied with the results. I've seen the 3-D side and it looks extremely powerful.

If these software companies want your business (and in the current market, I think they do), devise a simple test in line with what you are trying to do and ask each of the packages that you are interested in to run them. If one of them refuses, you have to ask yourself what kind of support you will get. If they run them, ask what platform they ran them on (computer capability) and then ask if they could couple the analyses (could they take the fluids results and couple them with the structural analysis). Make sure the simulations are something for which you can perform a hand calculation to compare the results, and, finally, ask for a demo...see if you can get your hands on the software and figure anything out.

You're not asking for a small package. To do CFD and non-linear impact, you will have a pretty nicely priced package of $20,000 - $25,000 US or more. These software vendors should spend some time with you.

My 2 cents are probably worth about 1-1/2.

 

[Jordonlaw]

I've read this thread with interest and can add something to the comments about MSC MARC. I've used this now for 5 years to model non-linear geometries, non-linear materials and complex contact in both 2 and 3d. In my new job I started with a list of 50 codes that nominally would do the simulations I wanted if you took the sales pitch. In the end I asked a limited number of companies to benchmark a real problem and only one came out with a technical solution; that was MARC. MARC has the best contact algorithms of those I have encoutered and I'm sorry to disagree with what was said earlier as ANSYS simply said they could solve the problem I had.

I expected integration with CAD packages to be important but all the codes I reviewed did this with ease, accepting that 2d geometry translation is easier than 3d.

Price is irrelevent if it can't solve the problem and yes, I have managers grin like a Cheshire Cat (English phrase) by cute colours and animations on cheap packages. In the end via some deft negotiation a significant cost reduction was obtained. My experience with MARC support in the UK is simply that it is excellent.

In summary, when you want a FEA package check out the software with analysts not the salesmen. Do your homework and benchmark the codes at the software vendors (they should be better than you so if they can't do it, don't bother to try yourself). Get a evaluation copy of the code and have a play; I did this with ANSYS and COSMOS to compare with my experience of other codes. Negotiate a hard price; vendors want to make a sale rather than walk away - end of quarters are good as the bonuses ae due ! But never ever take the advice of a saleman or another engineer unless you've tried it yourself.

If this has helped anyone it was worth typing it.

CHE

 

[dmacx]

I work for the government (AFRL) and as a former MSC Nastran power user and loyal customer we now use and like very much NE/Nastran. We have replaced all of our MSC seats with NE/Nastran and have never once regretted it. There are many reasons why and I can tell you that if the choice is simply MSC or NE Nastran it is a simple one... NE! Unless you are a Nastran expert and an FEA Pro I would recommend Noran because they help their customers and seem to really care. Others say they care, Noran shows it.

As for costs our package was about 1/3 the cost of MSC for a new license and we justified the switch based on the maintenance we were paying MSC which was close to the NE Nastran list price for a new seat.

I agree, evaluate each product but cost should never be the only consideration. Never buy as a results of a sales persons late minute deal. Those deals are always bad deals because you are forced into a decision without the proper time to evaluate. That sounds like it is happening here. Becareful.