3D Connection Design by FEA and by hand

[tngv752]

I believe that it gonna take more time than doing connection design by FEA than by hand. My questions are:

- All type of complex connection can designed by hand so no need to use FEA ? That's what the old engineer did in the old day. I guess that we can but maybe conservative.

- 3D models of complex connections have to be created using 3D AutoCad and import them into the FE software analysis ? So You must learn how to draw 3D in AutoCad. Is there any special package to just create 3D connection model nowadays ?

 

[GBor]

The list is long and glorious:

AutoCAD just purchased Algor.
SolidWorks is closely tied to COSMOS and works well with NEiNastran and, most recently, AMPS
Catia has Ansys

Most CAD packages export some format that most FEA packages will read.

Most FEA packages have modeling capability:

All of those mentioned above, many of which also work with FEMAP

Roshaz which has internal analysis capabilities and works as a front-end for Calculix

Etc. Etc.

NEiNastran

Garland E. Borowski, PE
Engineering Manager
Star Aviation

 

[GBor]

OH! Forgot to respond to the original question:

No, FEA is not appropriate for ALL design and isn't necessarily faster in all cases. If the connection gets complicated enough, FEA becomes useful.

I believe RISA may be something to look at for your situation.

 

[rb1957]

has "back-of-the-envelope" calcs are by their very nature conservative (well, at least they should be). if the piece can be simplified (simple loading simple reactions, simple section) then a hand calc can work. if you can handle the conservativism (implied) then why not (ask your self what's the impact of having an extra 1lb of Al ?)

but these days FE tools are very adept at producing answers without much work ... if you start with a 3D CAD model, most CAD packages allow you to apply load here, react it there, and "push a button". the whole catch is "is the model valid?"; if my sentence appealled to you with it's simplicity then you probably should avoid FE 'till you know some of the traps.

good luck !

 

[corus]

If you're designing your structure to satisfy design codes then you'll find that the design codes have been formulated on the assumption that you've used hand calculations, and hence tend to be conservative to allow for the assumptions you've made, and those you've forgotten about. In typical structural FE models, the results you get probably give a better idea of the assumptions you should be making in your hand calculations. In most cases though FE probably isn't needed unless the structural detail isn't covered or your hand calcs put you over recognised limits. Of course with a FE model you can say that the results are from a computer, and hence must be right, surely?

corus

 

[GBor]

Depends on how pretty the colors are...

 

[rb1957]

"beautiful plummage ..."

 

[GregLocock]

Somebody (it might even have been corus) posted a question the other day that is either dead easy in FEA, or impossible.

He wanted to know the stress due to a shrink fit between two toruses, one inside the other.

I think in general shrink fits, and general 'cone and spike' problems, are a field where hand calcs are easier than FEA.



Cheers

Greg Locock

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

 

[GBor]

I can get a pretty picture of this Only program I would trust to attempt that type of problem would be AMPS. Cool analysis...I may try it just for fun!

 

[GregLocock]

thread404-238043

I almost gave him a star for exploding my brain.

Cheers

Greg Locock

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

 

[corus]

It was too easy in FEA, but these days I prefer to see a hand calculation in MathCad or an excel spreadsheet to confirm these electrickery-made answers. Must be the Luddite in me coming out.

corus