FEA - Cosmos or Not

Sorry, I know this one has been done to death but I have read the
previous postings and can't find the answers I need.

I have a project to evaluate FEA software for use in determining the
long term performance of plastic products. Products contain complex
surfaces; a typical example would be a highly styled fuel tank. The
loads would be uniform forces and pressures at slightly elevated
general temperature with some localised non-steady state temperatures.

The material is known for its poor creep performance (and we don't
have an alternative we can use in our manufacturing process) so I
cannot make a linear behaviour assumption and yield results that mean
anything.

The obvious choice for a SolidWorks user would be CosmosWorks or
DesignStar because of the integration with Solidworks models and
assemblies, and the ease of use because of the familiar interface, but
I doubt either is capable of the analysis we require. Does anyone have
experience of this type of problem and if so what software do they
use, does it yield really useable results, and is it fairly easy to
get to grips with?

The other issue is cost, would I be able to get results from something
in the $10,000 - 15,000 range or do I need to spend $50,000 +. If the
latter then we would outsource but who to?

Too many questions!

 

In FEA, "nonlinear mechanics" usually means "stress/temperature dependent
material properties". Solvers like http://www.cosmosm.com/nonlinear.htm do
it, and support simple viscoelasticity models too (see
http://www.ansys.com/industry/nonlinear/examples.htm for examples)
I guess those tools should stay within your 10K-15K range .

Now, if I understood you correctly, your "long term creep" problem is a much
harder problem as it involves "time-dependent properties", or materials that
"remember" what happened to them in the past. In other words, if a small
portion of your material doesn't have the same properties after a cycle that
brings it back to the same stress/temperature state, you're in trouble!
If so, you'll certainly need a more specialized & much more expensive tool.

Anyway, I'd say you shouldn't consider the price of the software in the
outsource/in-house decision.
In my (small) experience, it rather depends on how often you'll use the FEA
software. Buy only if you'll use it more
than once per month, and have trained/trainable engineer(s) in house.
If not, outsource to someone who knows about your problem, in your case
plastics as it looks the problem will mainly be to find/develop/validate a
model of the material. I found
http://www.rapra.net/technicalservices/PlasticandRubberFEA.htm while
searching the web for you (I have no experience in plastics, I don't know
them...)

Good luck!

 

As mentioned in a reply to another post (which fem?) we've been evaluating
FEA software recently and found that the CosmosWorks mesher had problems
with complex surfaces. NE/Nastran did better and Ansys did best of all. (In
fairness to the FEA programs, SW has trouble making these parts, so it may
not be the fault of the meshers.)

I would guess that trying to analyze this with a non-linear material and
non-steady state temperatures would be incredibly slow. You'll probably end
up running analyses and physical tests at a constant temperature that
approximates the measured effects of the non-steady state temperatures.

Can you get good material characteristics for the creep behavior? It won't
matter how good the software is if you can't properly characterize the
material. As we get better and better tools for analyzing, it seems like
there is less and less data available on the materials. The folks Philippe
suggested may be able to test the material, but I bet it isn't cheap.

CosmosWorks doesn't do non-linear materials. DesignStar does either
non-linear materials or large-deflection analyses, but not both at the same
time. GeoStar will do both.

We don't have any real experience with what you want to do, as we were
running much simpler analyses to try out the tools, but CosmosWorks is
pretty easy to use, DesignStar is a little tougher, and GeoStar is not
designer-friendly at all. NE/Nastran is pretty easy to use, but doesn't have
a direct SW link yet. Ansys has an easy to use interface and link with SW,
but to get the full capability, you end up using the much less friendly
"classic" interface.

If you don't need large deflection analysis, then you could go with
DesignStar, which is in the range you want. We didn't find out how much
GeoStar was, but I'm sure it's above $15K. NE/Nastran is in your range and
has the capability you need. Ansys Structural is more like $20K.

I agree with Philippe's reply. It's a real rock and a hard place decision on
consulting and in-house.


Jerry Steiger
Tripod Data Systems

 

If you are going to do this with Cosmos products and it involves creep then
you are looking at Cosmos/M. This would likely be done with shell elements.
DesignStar or CosmosWorks can be used to generate a mesh and the mesh can
be exported to Cosmos/M for the creep study.

Typically meshing problems with Cosmos products are due to bad geometry such
as slivers and short edges. Judicious use of split lines can be used on
complex geometry to control and assist in meshing difficult geometry.

I would also be surprised if there wasn't some dynamic element to your
problem. Cosmos/M can deal with random vibration problems.

How do you propose to generate creep data for the plastic? Without creep
data no FEA package is going to be able to give meaningful answers.

You might also want to post this question on .
.