Basic FEM

I'd be hesitant to use any unvalidated software commercially but given this is just a beta then this is surely part of the process that some of these questions are answered and the software is correctly compared.

Presumably no-one could hope to, or want to, sell software commercially for critical-applications without having checked their numbers, as it were.

I think the most interesting feature here, and reading the articles from back in '08 about the software, its likely original purpose is in the analysis of scan-data. High-density point clouds are always a pig to mesh even without running any further FEA on it, being able to handle this efficiently and without the meshing stage certainly has applications - though potentially fewer in the marine field where scanned data typically comes from proven designs.

 

Dear all

Thank you for your warnings and concerns regarding the general use of FEM like software. Scan and Solve displays following warning:

CAUTION
Design decisions require experimental data and substantial experience; they should never be made based solely on a software simulation. Simulation is not intended to replace physical testing of prototypes, which is required to validate any design.

This is not limited to this software but in my opinion to all engineering practices and tools. On top of that please consider this is Beta software, so use it only for experimental purposes.

I would appreciate it if we can leave the warnings by this and limit this discussion from now on to subjects regarding this particular software

Thank you in advance!

 

thanks for info,actually i dont know this details

 

Unfortunatley that often not the case. Just visit the forum for any software and search for the phrase fea bug or just bug and you will see that CAD software is often released with critical errors.

"I would appreciate it if we can leave the warnings by this and limit this discussion from now on to subjects regarding this particular software"

Instead of investing in proper development it looks like they are releasing it as a beta so that the users can be their free testing service. I wouldnt waste my time with it.

 

Software like Linux, Rhino and Freeship/Delftship started of like this and now offer packages that can compete or ar even better than software previously costing 1000's of Euro's more. For small and startup businesses this is a great thing. Not everybody can afford or is prepared to pay 5000 Euro plus for each piece of software, but they may be prepared to spare some time on testing beta software, knowing this is limited. A lot of them are around on this forum, and I started this tread mainly for them. If you want to tell how good your 5000+ euro software is please start your own tread, we know by now.

 

Oh, quite right, but there is a difference in a total failure in method and bugs. I wouldn't see a free-beta being released as a sign of cheapness. Have a free-beta is testing but that's fairly standard in my experience – in house testing would have real job testing a system across hundreds/thousands of systems and applications. You'd also have to be fairly mad to put something out to beta that hasn't been checked beforehand.

That said, there are plenty of people willing to ignore the chance of errors and any warnings just to get a free piece of software, but then that's the fault of the user rather than developer. I'm happy to watch how this software develops but I won't be using it for anything serious until I see it validated by independent reviewers.

---

Just out of interest, do you know much about the developer at all Windvang. There seems to be quite an active rhino Rhino community in the Netherlands at the moment, are there any FEM specialists who have already looked at Scan&Solve?

 

In your original post you said "I have very little FEM experience, so I can not commend on the results." so I dont understand why you are so upset to hear from someone who does.

I never said anything about expensive software and havent made any recommendations. I just dont understand why anyone would be a free beta tester when you can buy proven FEA for as little as $285.

Your anger makes me wonder if you have a financial interest in this plug in.

 

What low cost software do you recommend starting at $285?

 

I dont recommend it as I havent tested it yet (I have downloaded it though and will test it soon)

http://www.autofea.com/Product-Line.htm

 

@ ACuttle

I don't know anything about the developers other than their profile on the site. www.intact-solutions.com/company.php We do indeed a lot in Rhino here in Holland, but always have to export te dedicated FEM software, not always without problems.

@ Alidesigner

No financial interest. Just ask politely to restrict comments to this particular software, good or bad, like DCockey in post #10. Post like "i think solidworks is the best."
I, and most forum readers can do without.

 

I downloaded it and all I can say is that Scan and solve is going to be frustrating to use for anyone who really uses FEA. It has a very limited application at this stage.

If you wanted a supported and capable package with all the elements you would need to learn FEA in detail you can't better than Cadre Pro. Which I reckon will be the package Alidesigner is playing with.
http://www.cadreanalytic.com/Cadrepro.htm

 

I've downloaded the test version of CadrePro and it's generally impressive. One potentially significant limitation for me it that it appears to be limited to isotropic materials. Fine for aluminum or steel but limiting for wood or composites. I may buy a copy unless something else turns up.

 

This is one of the biggest misunderstandings of using FEA for composites.

Just because the FEM allows you to model layers and different layers with different properties, doesn’t mean it is correct. No FEA will give you correct results using anisotropic or orthotropic using “composites” in the sense of ‘fibre reinforced’.

There is no way of knowing if the results are correct and the variables, owing to the laminator, the temperature, the humidity, the amount of resin, fibre volume etc etc all play a major role in the material properties or simply its “E”, in each direction.

You would need to strain gauge each layer, ie lay down strain gauges in all the layers in all directions to verify the outputs to validate the model. This is impossible. Since the placement of the strain gauges in the composite affects the results. To my knowledge, no one has ever done this too. My old Prof doesn't recommend this too!

If you wish to model “fibre reinforced structures” in FEA, you make real coupon samples and mechanically test them to obtain the E, then use this value, not the “theoretical” value. If you don’t make coupons and test them to obtain E, your results are worthless.

 

All analysis is "wrong", if the standard of being "right" is a perfect match to "reality".

What is important is whether the results are useful, which requires an understanding of both what they represent and how they may/do deviate from reality.

A reasonable approximation of non-isotropic material properties should be better than pretending they are isotropic and can be represented with a single value of "E". Experimental results can be use to confirm and improve the approximations. A properly interpreted analysis using a reasonable representation is better than no analysis. Of course this assumes care and diligence in all steps of the analysis, and particularly in interpreting the results.

 

And that is the crux of the whole problem. Without an understanding of what is doing, the results are not worth much.

As for reasonable approximations, that is all and well good if one is using "known" or "similar" material properties to that being proposed. But many use FEA without knowing such; they use the FEA to give them this! Hence, how real is the E being used? The FE should not be used to provide the values of E, given that composite mechanical properties vary so widely, to guess what the E may be requires a large database of existing known structures and layups. Otherwise the guesses can vary so widely, they border on laughable.

Two different yards can lay up the same laminate and their E’s shall be different, how different?.. that depends upon their QA and skill, but can be significant, not minor. Not appreciating this simple but often over looked fact of using composites, leads to erroneous results of pretty looking colour plots from an FEM based upon assumptions that are flawed and do not reflect reality being proposed.

Why do you think Classification societies require coupon tests by the actual laminators building the boat using their own QA in the yards enviornment of build…it isn’t to have spare coasters for the office!