Help needed in choosing software 1

[phoenix221]

Hi,

I am currently looking into which modelling environment I should be using. I would like to model (2D/3D) an entire airframe structure with the objective of using the model for stress test and FEA.

The material used in construction would be largely glass fiber "sandwich" composite on a foam core.

I've been playing around in Autodesk Inventor for the past week or so. I really like the "assembly" concept in it, but I am getting the distinct feeling that it may not be the best modelling tool. I also looked at Advanced Aircraft Analysis (from DAR Corporation) but it seems to completely lack the modelling environment! Concepts 3D just popped on my radar screen, but I haven't had time to really look at it yet...

I was wondering what other people might be using for similar purposes?

In my choice of software, I was also hoping to keep the learning curve short if at all possible, since I am doing this in my free time,... and since I will most likely end up using another piece of software for the stress testing, etc. file/data exchange would be a priority as well.

Any suggestions are appreciated, Thanks...

 

[franck]

Hi Phoenix,

I am not sure if this is going to be helpful since those S/W are quite expensive but here are the software used in my company for automotive/aerospace/space analyses:

To create your geometry (2D/3D):

Catia
Pro Engineer
Solidworks (probably the best choice since the less expensive and it works really well too)

To create your meshing and visualize your processing results (stress, normal modes):

I-DEAS
PATRAN
FEMAP (affordable and really excellent)

To perform your calculation:

Nastran
Ansys

both of those can be run in Windows environment.

Good luck!

Franck

 

[phoenix221]

I was just reading up on Solidworks actually so your tip was great ... I am not sure how easy it would be to model "shell" structures such as layered fiberglass ???

I also noticed that the same company puts out "Cosmosworks" for FEA. If Cosmosworks can handle anisotropic materials, I would not need the other packages, well at least for FEA... ?

Would be nice to invest the time in learning a package that can also do aerodynamic analysis on the airframe as well as the mechanical/FEA analysis... any thoughts/tips?

 

[franck]

I will not recommend using Cosmos for your complex analysis. This software is really basic unless they have improved it a lot!

 

[phoenix221]

I am gravitating towards ANSA for mesh generation and LUSAS or ANSYS for FEA.

I know some products require that you have a Phd in FEA before you can get reasonably accurate results :-( ... from this standpoint, I suppose I would need tools more targeted towards designers rather than FEA analysts ???

Am I going down the wrong path with my preliminary choices?

 

[franck]

Phoenix,

I am not sure if a PHd is needed to run FEM analyses but the only advise I can give you is to always check if your results are making sense! Really, now most of the S/W especially Nastran have built-in parameters that will make your analysis go through even when you are doing important modelling mistakes. In that case the S/W will give some results but they might be really wrong. Those features are required by companies that want to have the analysis time reduced! The quality of their product may be questionable.

As a rule, try to always develop simple test case to check if your modelling is in agreement with the theory. This is especially valid for your composite structures for which there are several ways of modelling.

I cannot say for ANSA or LUSAS but ANSYS is a good S/W.

Franck

 

[phoenix221]

franck,

>the only advise I can give you is to always check if your
>results are making sense!

I was hoping to be able to rely on the sofware analysis within reasonable margin

>Really, now most of the S/W especially Nastran have built-in
>parameters that will make your analysis go through even when
>you are doing important modelling mistakes. In that case the
>S/W will give some results but they might be really wrong.

This is not a feature set I desire :-(. I would like to avoid this as much as possible!

>Those features are required by companies that want to have
>the analysis time reduced!

What is the purpose of analysis if it is compromized? Please do tell which companies these are I will avoid their products for sure !!!

>As a rule, try to always develop simple test case to check
>if your modelling is in agreement with the theory. This is
>especially valid for your composite structures for which
>there are several ways of modelling.

Are you referring to "modelling" for FEA purposes? I intend to use

http://www.thelaminator.net

to determine theoretical properties values for the composite. I do plan to do some simple testing of small pieces to verify that the sofware is not "making things up". Beyond that, I would like to definitely avoid making up very complex tests! I was hoping that by choosing the right FEA sofware, I can rely on it more :-(

 

[franck]

Phoenix,

Your approach is correct! Compare the results of simple test cases derived by testing and analysis. By the way, I used laminator before and it gave me good results. ESA is selling an excellent one as well but I guess a bit more expensive. Check out the following site:

http://esapub.esrin.esa.it/pff/pffv7n1/henv7n1.htm

Franck

ps: Yes I was referring to different ways of modelling composites in FEA (for instance in Nastran you can use PCOMP or PSHELL elements).

 

[phoenix221]

>Franck,
>ESA is selling an excellent one as well but I guess a bit more
>expensive. Check out the following site:
>

http://esapub.esrin.esa.it/pff/pffv7n1/henv7n1.htm

It seems more extensive!... Re. Laminator... did you find that the numbers were below or above the actual values based on tests?

 

[franck]

Phoenix,

As a rule for my mechanical parameters (Young modulus,...) I am always using the B allowables (99% of probability with 90% of confidence level). That way we are always conservative (for instance the test produces higher stiffness than predicted).

Franck

 

[phoenix221]

Well... I thought to post an update and seek more counsel ...

I have looked at MSC.Patran which has a bewildering array of options and settings. I found it strange that a product of this caliber is not better integrated with its sister product, Nastran. You have to export for an analysis, run Nastran manually, then import results for post processing. You'd think that the products come from two different companies :-(

I also looked at Femap which has fewer options, but shares the same attrocious interface philosophy with Patran, i.e. there is no global view of the project.

Next came Lusas which has a friendlier interface but (at least the version I was looking at) lacks the laminate materials definition and draping features of Patran. I believe they have "Lusas Composite" which I could not look at :-(

I also took a look at the Algor website, and the presentations seems to indicate that the product really has a nice interface with good support for laminates and good integration with Solidworks and Nastran... problem is I can not obtain a trial version to actually test it out... the company does not respond to contact.

So now for a shameless pleading :-o... If anyone has an ISO image of Lusas Composite or Algor (Laminate), please email me at michael221@hotmail.com.

Finally, (more to summarize my own thoughts...) the main issues seem to be:
1. Integration with the CAD software and solvers
2. Well designed user interface with a manageable set of features.
3. Support for layup definition and draping
4. analysis/visualization of analysis results.

I find the different views on "integration" to be quite funny actually. Some companies feel that their software integrates as long as you can read/write files of the other software... furthermore, some software doesn't even integrate with it's own complementary products...what companies get away with these days ...

The interface to the features is another issue altogether. FEA software by definition will be complex. To facilitate the workflow, an interface that does that is an absolute must, otherwise you need to keep track of what you have done in your head... a proposition that makes the whole process more difficult than it should be.

So this was a rather lengthy post with some ranting and raving thrown in for good measure... if you have additional recommendations or wish to correct my oversights/errors please let me know...

 

[franck]

Hi Phoenix,

Have you checked Nastran for Windows? It is using the FEMAP interface and you can directly excute/run your model from FEMAP. I am able to open Solidworks/ProE or Catia models even if it always takes some time to find what is the appropriate/compatible format when you perform that operation for the first time. However, similar to PATRAN, you must be very careful when looking at the results (especially the stress) since they are so many parameters (averaging, smoothing, corners, etc.) that the results may not reflect the analysis.

Franck

 

[phoenix221]

Nastran for Windows... no I have not looked at it yet...

In general I find the Patran/FEMAP style interface poorly designed. There are just too many parameters and the learning curve is very steep. I can't spend the next 2 years learning the software... I would like something that I can use for reasonable FEA with say a 2 mo time investment. From this standpoint the interface is key, the other aspect has to do with the sheer number of features that are user controlled when using Patran/FEMAP. If they provided defaults for some so that at least you don't have to worry about each!... but as I have concluded, they designed their products like this on purpose. The user is meant to closely control every aspect of the FEA. Their training offerings (which in some cases can total to more than the software itself) must be selling like hot-cakes

Obviously I am not an "expert" FE analyst. For this reason I liked the products such as Algor because the interface at the least is not getting in the way of the workflow. Interface wise, Cosmosworks is really a well thought out product. Its interface is easy to use, and they provide reasonable defaults in case you don't yet know what to put in :-0.

 

[phoenix221]

... well my post continues here since I hit "Submit" rather than "Preview"...

If I did not need composite/laminate support, Cosmosworks would probably do just fine, though its handling of one of the simpler assemblies was not quite good, so I had to test some of the parts separately. Cosmosworks's support for composites/laminates is not there yet either :-(, it only allows orthtropic definition of materials, and no support for ply layups as far as I could see.

Lusas is running strong. It's interface is also more user friendly. I was however unable to evaluate "Lusas Composite", so the composite support in Lusas is unknown to me :-(.

This leaves me with "Algor" with the "Composite material Extender" plug-in which looks really good from their web site ... have not been able to evaluate it though because the company has yet to get back to me :-(. I am wondering if there is another way to get an eval copy?

 

[mwj1]

phoenix221 said:

Nastran for Windows... no I have not looked at it yet...

In general I find the Patran/FEMAP style interface poorly designed. There are just too many parameters and the learning curve is very steep...

IMHO FEMAP rules and PATRAN is nothing like FEMAP. Check out:

http://www.ugs.com/products/nx/simulation/advanced/femap/index.shtml

I have used this product (FEMAP and/or MSC NASTRAN for Windows - two different vendors same pre/post) for over 10 years and choose it over any other FEA product I have been exposed to. Price is good, solver is completely imbedded in the package, translators work well and it has been created to work in the Windows environment.

 

[crackman]

Just to add my two cents... I have done quite a bit of FEA over the past 18 years in the aircraft industry and have tried, reviewed and evaluated many FEM packages and for the specific application of airframe modelling (either metals or composites), NASTRAN is the ONLY way to go. There are an enormous amount of capabilities within Nastran and output (grid point force balance, element oriented forces, etc.) which have made it THE tool for FEA in aircraft. The tools are not obvious to the new modeler but become highly valued to the experienced. The main thing to remember is that in the aircraft industry (due to size and complexity) vary rarely are solid FEMs built. The lion's share of stress analysis is performed with coarse grid models and then performing detailed stress analysis using validated stress methods (typically from OEM stress manuals). NASTRAN provides the unique capability (remember unlike new FEA packages, NASTRAN was specifically developed for the aircraft industry) of providing the necessary data to accomplish this type of analysis. I have found many FEA packages (for example ALGOR) for which getting something as simple as a Free Body Diagram is impossible. Too many of the current FEA codes have been developed purely with the idea of performing Solid Models only.

In the past (many many moons ago - figure it was Nastran ver 64 and Patran 2.5), I very effectively modeled a 3000+ element compound curvature composite laminated radome (18 feet long, 3 foot wide) by using NASTRAN CQUAD4 elements and PCOMP and MAT8 cards and using the then Patran PLAM module to do a Tsai Wu failure analysis for all elements - all plies. Worked very well in that the PLAM module translated all of the NASTRAN output and could perform the analysis of all plies very quickly (for the time of course - the radome was atop a section of boeing fuselage and I ran it on a cray - probably could run it on my lap top now .

Selecting an FEA package is a tough decision but if you are planning on staying in the aircraft business I too recommend NASTRAN and either Patran or Femap.

Good luck

JB