Best vibration+noise FEA 1

[plbm]

what is the best vibration FEA solver, for problems related to sound and noise?

I mean modal and forced vibrational analysis with an eye on acoustics.

An integration between vibration and sound would be great! What software can does this nowadays?

Material description should be as accurate as possible (I am thinking of complex materials like resin+fibers, foams, composites, etc.)

 

[bjpil]

For what it’s worth, I use MSC/Nastran for the vibration analysis and then import the results into LMS Sysnoise for the noise analysis. Never have had an opportunity to correlate the results and I’ve only used it for metals. The two software work well together.

BJP

 

[Drej]

What do you mean by "best"?


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

 

[plbm]

by "best" I mean that it gives reliable results, even for complex materials.

I have the possibility to measure acoustically, so that would be the final validation of FEA efforts. Therefore a feature I would consider is the possibility to integrate the vibrational analysis into an acoustical analysis.

GUI, features for speeding up the analysis, and price are important, but I would give a lower importance to them now.

 

[Drej]

> by "best" I mean that it gives reliable results, even for complex materials.

Most, if not all, commercial codes will give you "reliable" results. This is a topic that has been discussed previously (the definition of "best" results or "better" software), and is difficult to quantify. In the end, most codes will do the job for you, but it's down to the experience of the user whether or not the answers are garbage or useful.

For me, I've been using ANSYS and ABAQUS for many years, as well as a few other lesser known codes (NASTRAN, PAFEC, DIANA, NISA...). For a budget package which will do a job for you, I would consider looking at NISA, although this is quite an old code and has been lacking in development somewhat. If you have complex geometry, NISA has a limited geometry integration capability, so this is something to think about. It does however, have a brilliant eigenvalue solver, probably the best I've used along with NASTRAN. ANSYS is also excellent. With this you have very good functionality (forced vibration, acoustics, emag, statics, transient, coupled-field, &c.), an excellent GUI and excellent CAD integration. The flexibilty of using ANSYS programming language as well (APDL) is also very useful. I'm a big ANSYS fan, and I've been using it many years, so there maybe a *bit* of bias in there as you can tell. Drawback of ANSYS, NASTRAN, ABAQUS is the licence cost. If you want a purely dynamics solver for your work, it may be worth paying the extra for NASTRAN. Otherwise, have a look into ANSYS. You can check ANSYS capabilitites here

http://www.ansys.com/assets/brochures/capabilities-90.pdf.

The above is of course non-exhaustive, and only a fragment of the info available. A good idea might be to demo a couple on your shortlist for a month each - this is very popular nowadays and is encouraged by the suppliers.

Best of luck,

-- drej --


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

 

[plbm]

BJP,
thanks for your comment, LMS looks good...
do you know if it also imports from NEiNastran?

 

[plbm]

> Most, if not all, commercial codes will give you "reliable" results. This is a topic that has been discussed previously

drej,
do you know if describing mechanical properties varying with frequency leads to more accurate results for some applications?
I was showed MSC Nastran, where there is the such possibility of material properties description. I thought that if they keep this description possibility open, it should be relevant for some applications. Unfortunately they are not giving me a demo.
For example Cosmosm does not have this material properties description feature. Only has creep. But they give out demos!

 

[Drej]

I've never had to change mechanical properties as a function of frequency before, and to be honest I wouldn't know why (maybe it's different in the acoustic world, but here I'm talking from the purely mechanical perspective). Certainly the big three ANSYS/ABAQUS/NASTRAN you can change *some* material properties dependant on the frequency, but the properties that can be changed are very limited (and for good reasons as well). However, in ANSYS there is an undocumented procedure for changing the other mechanical props (mu, E, etc.) in terms of frequency; it's a cheat, and you get a warning message, but it certainly works and I know lots of people that have used it (in fact, search the archives here and there will be a dsicussion thread on this).

You can easily set the damping characteristics as a function of frequency "perfectly legally" (as a material property) within ANSYS and ABAQUS (don't know about COSMOS) and probably NASTRAN. These would be the Rayleigh damping constants (alpha/beta damping) and the material damping, but that's about it.

Of course, the above three codes also have all of the common material models (elasto/visco/plastic/etc.) such as creep you mention as well as the capability of user materials/compound (hybrid) models and composites. ANSYS especially covers the use of composites very well in the current V9, but again not sure about the other two.

Cheers,

-- drej --


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

 

[GregLocock]

For our noise and vibration models we use Nastran to generate the linear bits, and then glue the results from Nastran together with a systems model that specifically allows for frequency (and other) dependent properties. I think it is in-house code, leastways I cannot find a web reference to it. This then allows us to drive the system with measured force or vibration spectra, and generate sound spectra.

It's not rocket science, but it is certainly more complex than using one package for everything.







Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[plbm]

Greg,
if I understood right from your post, you are assuming that it is relevant to use frequency dependant properties. I would like to know for what applications!

And you are saying that you are using a software tool that "learns" that frequency dependance (and other dependancies) from measured data? That would be exactly what I'm looking for!

Cheers

 

[bjpil]

PLBM,

Not sure how well NEiNastan works with Sysnoise but I assume it is the same as MSCNastran. All you need is to export out a punch file or an op2. The question I would ask is how good is NEiNastran's dynamic solver? I know MSCNastran is very good up to a certain frequency. I have more trouble after about 2 Khz where the modes become more complex and is harder to correlate. The other thing is can NEiNastran use acceleration test data directly?

BJP

 

[fkmeyers]

We use NEiNastran for complex dynamic models (aerospace) and it works as good or better than MSC and other FEA products we also have in-house such as ABAQUS. I think one of the major strengths of Nastran in general is dynamics. NEiNastran supports residual vectors which I have been using recently to help with the higher frequency modes. There should also be no problem using acceleration test data.

 

[GregLocock]

We use frequency dependent modulii for rubber bushes, displacement dependent moduli for rubber bushes, and all sorts of strange things for hydraulic engine mounts and the like.



Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[fkmeyers]

Nastran (MSC, NX, NEi) has support for this with their CBUSH element.

 

[rb1957]

pet peave ...

i wish people would stop asking for the "best" software (for just about anything). there is no universal "best". even "good" software depends on so many variables.

in my humble opinion, a more sensible question would be to ask people to recommend a FEA (or whatever) package. better would be something like "i'm thinking (there's a good clue) about buying FEA package "XYZ" ... how robust is this ? what sort of problems might i find ?" why do i think this is better ... it shows you've taken the time to research the question a little (i mean, how difficult is it to yahoo or google ?)

thank you, and good night

 

[rtmpxr]

Strongly agree.

 

[Drej]

rb1957/trmpxr

Tis also one of my pet hates, so you can imagine when this came up thread727-123750 I was like a coiled spring!

Cheers,

-- drej --


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

 

[glsv]

If you are working on dynamics problems in NEiNastran with an eye toward acoustics, you may want to cosider VNoise.
It is a fully capable BEM code that has a direct interface to NEiNastran and FEMAP. It is a fraction of the cost of Sysnoise.

 

[ramshree]

My Suggestions:
Vibration: ANSYS/Nastran
Acoustics: LMS Sysnoise (This has been proven and tested for several years and is the most widely used acoustic software in the world).

Expensive: Yes!

But my philosophy is you get what you pay for!

 

[gwolf]

ABAQUS also has a full set of acoustic elements which integrate seamlessly with the more conventional structural elements. People use them to analyse the noise in car interiors for example.

I have used these elements and it's very easy but I am not an acoustics expert so I don't know what their real limitations are. You would have to take a look at the abaqus.com web site to get a datasheet.