NASTRAN versus COSMOSWORKS 1

[dshaffer1001]

Howdy folks, new guy here. I hope this is the right place to post this, if not, please accept my apologies.

I don't know much about COSMOSWORKS. Can anybody clue me in on how it compares with NX NASTRAN ? Is there any particular reason to favor one over the other ?

I am an aerospace engineer with several years experience using MSC/NASTRAN, now NX NASTRAN, on both civilian and military aircraft. My company has been asked to analyze a crashworthy crew seat installation on a military transport. The seat faces aft and is attached to the aircraft side wall, not the floor. It is required to fold against the side wall when not in use, so its support structure incorporates lug fittings, plus several removable struts that are held in place using lock pins and stowed when not in use. Various intercostals and brackets will be added to the aircraft side wall. The structure is to be analyzed for static emergency landing and flight loads only.

I will have to analyze the support structure, struts, lugs, attaching intercostals and brackets, fasteners, the whole nine yards.

We have been given Solidworks models of the entire installation. My fellow stress guy has a copy of COSMOSWORKS, and says he can get the model built and running in a lot less time than I can using NASTRAN. I believe him, as the NASTRAN model will be a considerable amount of work, even using the Solidworks geometry to help me set it up.

Are you folks aware of any reasons NOT to use COSMOSWORKS ?

 

[ZillionM]

This is what a created COSMOSWORKS report says in a note:
'Do not base your design decisions solely on the data presented in this report. Use this information in conjunction with experimental data and practical experience. Field testing is mandatory to validate your final design. COSMOSWorks helps you reduce your time-to-market by reducing but not eliminating field tests.'

I dont know how NASTRAN works, but I guess you have to some how connect the parts and limit their freedom of movement/direction?
With SolidWorks & COSMOSWorks you first have to assemble the parts using mates. Then afterwards, in COSMOS, you have to restrain the parts using connectors without moving any part.

ZillionM

 

[MadMango]

You've been using Nastran for so long, I would think that a few roadblocks may be ahead of you. You've probably got a fine collection of material properties that won't be available in COSMOS, not a huge problem, but it will take time to recreate your data. You are familiar with the work flow in Nastran, you'll have to learn how COSMOS behaves (again more time). Little things like these, you'll have to judge exactly how much time you will actually save. I have very little experience with Nastran, so on an "accuracy" level, I can't say if there is any technical problem with using COSMOSWorks.

"Art without engineering is dreaming; Engineering without art is calculating."

Have you read faq731-376 to make the best use of Eng-Tips Forums?

 

[dshaffer1001]

Thanks for the responses.

My coworker will be doing most of the FEA work. He's pretty well qualified on Solidworks and COSMOS so no worries on that score. However, he has no experience on NASTRAN, so we don't really know how they compare.

I'm more interested in how the FEA actually works compared to NASTRAN, i.e. how accurate is it, how hard is it to validate the results, are there any traps or pitfalls I need to be aware of, that sort of thing. I've spent some time web searching without much success.

And yes, since my background is mostly NASTRAN and Autocad, my learning Solidworks looks like it will be a bit of a hair puller.

 

[gfbotha]

I believe that user-friendly pre-processing does not necessarily mean bad results, but I do admit that such FEA software is more susceptible to misuse by inexperienced/casual users.

Before we selected Cosmos DesignStar (basically the same as CosmosWorks) a coupe of years ago, I asked the same kind of questions as you do. Depending on what effects you want to include in your analysis, and given that the problem is indeed modelled properly and representatively using the tools available in CosmosWorks, it is my opinion that your results should compare favourably with that obtained using say, Nastran. To have confidence, I suggest you figure out and understand the underlying methods employed when utilizing some of the semi-automated modelling or constraint tools available in CosmosWorks. A problem is that the underlying theory might not always be documented that well. What is good to know is that many times the actual solvers used in these designer packages are the same as those employed in the conventional/more comprehensive software (CosmosM in this case).

We quite regularly also perform non-linear analyses and verification through the years generally pointed to good results. What I do not like is some element types not being available (e.g. 2D types). For what it is worth, a couple of years ago, two quite detailed studies performed by so-called independent bodies, were published comparing CosmosWorks and a couple of other programs (performing different types of analyses and also referring to Nastran): the one was by IMPACT Engineering Solutions and the other by ARA Engineering. When pondering about acceptable error, one should keep in mind assumptions made about loading, material, etc.

Regards

 

[Brodie79]

I currently use CosmosWorks and FEMap with NX Nastran. I've carried out a few comparison checks of static, contact and eigenvalue analysis. I've generally found the results to be within a few significant figures.

You can set up the model in CosmosWorks and export it to Nastran and run it there for certain analysis types, which give you the speed of the Cosmos preprocessor and the power of your Nastran post processor.

FEMap with Nastran obviously gives me greater control over the model and mesh and is better when I need to extract specific nodal data but generally I've found the results in CosmosWorks to be acceptable.

 

[kellnerp]

We have been given Solidworks models of the entire installation. My fellow stress guy has a copy of COSMOSWORKS, and says he can get the model built and running in a lot less time than I can using NASTRAN. I believe him, as the NASTRAN model will be a considerable amount of work, even using the Solidworks geometry to help me set it up.

I have run into a couple of problems with CosmosWorks.

1. The FEA model results may not reflect the geometry in SW if the SW model is modified after the FEA model is solved.
2. Moving a CW model from one computer to another once caused a major problem with the model reporting incorrect results because of a confusion of temp files. I don't know if this has ever been fixed.
3. CW is not certified as far as I know. There are Nastran versions that are.
4. Documentation of the model may be difficult because the CW software does not expose nodes and elements.
5. If you have to use shells and solids or just plain shells I have not seen CW to be totally reliable in stitching adjoining faces properly (this has to do with sharing nodes on edges.)
6. Setting up shell models is a major pain in CW.
7. You have to modify the geometry in SW (splitting faces and solids) to get the mesh the way you want it.

TOP

TOP

http://www.engtran.com