FEA SOFTWARE

[civeng80]

I did my course in civil engineering about 30 years ago. Things were different back then FEA was around but quite distant. I now use microstran (structural analysis program both 2d and 3d). Structural engineering often involes analysing quite vexing models e.g. plates Im currently going thru this now. Can anyone recommned a good FEA software for structures and how would it be different to what I have?

Thanks

 

[GregLocock]

That depends on what you want. One way of looking at it is to separate the problem into three parts

building the model
solving the model
displaying the results

If you are lucky you will find a package that does all 3 in an integrated environment.

In my opinion the easiest pre and post processor, where all you have is sketches, is femap. I only do linear stuff so have no particular opinion about solvers.

Hypermesh is a powerful pre and post processor,especially for meshing solids, but it has a 'unique' user interface. It does have a built in solver, but is designed to integrate with other solvers just as well.

Call me a luddite, I still write decks by hand.

Cheers

Greg Locock

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

 

[rb1957]

ok greg ... "luddite"

 

[EdR]

Greg:

You are confusing the younger members with "decks"......"luddite"

Ed.R.

 

[civeng80]

Pardon the ignorance

But what are decks?

and while your at it what is a luddite?

 

[GBor]

Can't help you with "luddite"...Aussie-thing? or ludicrous? But shortly after chiseling computer programs out of rock, "decks" of cards were used to enter instructions into a computer.

 

[GregLocock]

If you look in your nastran bulk file I think you'll still find the various sections refered to as cards.

Cheers

Luddite

 

[civeng80]

Thanks Greg !

Card decks bring back good/bad memories lining up for computer runs. Also thanks for new word "Luddite" I may use it first opportunity I have.

Happy Australia Day !

Cheers !!

 

[prost]

Luddite:
one of a group of early 19th century English workmen destroying labor saving machinery as a protest;
broadly : one who is opposed to especially technological change

 

[Co]

ANSYS Multiphysics........

 

[hansm2]

For general purpose FEA I recommend NEiNastran from Noran. There are many posts on the subject on eng-tips and people give opinions of this and other products. I would check these out.

Hans

 

[GBor]

civeng80,

Could you answer Greg's original question? As hansm2 says, there are many posts on this subject, but the answer may vary depeding on your application. It appears you are civil-structural, are you working on concrete? Steel? Poured structures? Beam-reinforced wood? Houses? High-rises?

 

[civeng80]

I mainly work on steel structures (e.g portal frames) with tilt up panels. Already have microstran wich is good for linear structures, but interested in plates e.g. torsion and biaxial bending of tilt up panels.

Thanks !

 

[GregLocock]

So what level of detail and complexity do you use, or want?

For instance, I do a lot of work on spaceframes and suchlike, with beam elements, which could be up to 500 mm long. On the other hand I occasionally have to model the joints in laddder chassis, for which I use plate elements, these might be perhaps 5 or 10mm on a side. Alternatively I might be modelling a suspension arm in which case the tets might be only 2mm on a side.



Cheers

Greg Locock

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

 

[civeng80]

Interesting question Greg.

Microstran does not have any limitations for linear structures. So does a FEA program have limitations then?

Now 500 mm would be significant in structures whereby 2mm would not be significant. So about 100mm would be about the linear acuracy required. But is this important in FEA?

But with structures anything may come up for analysis really. So Im looking for something which is fairly general.

Hope this makes sense.

Thanks !

 

[GregLocock]

Sorry, what I'm getting at is that different people need different levels of mesh detail. For some purposes 5, or more typically 30, elements might be enough to be useful, whereas for other jobs half a million is none too many.

FEA programs don't really have a scale. That is set by your units, inside it is all just ones and zeroes.

Cheers

Greg Locock

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

 

[civeng80]

Thanks Greg

Being new to FEA (as I said did my course 30 years ago) I simply dont know how accurate the mesh should be for the structural applications Im looking for. Im using microstran now which is a matrix structural analysis software, so Im going in new territory, maybe I dont need it at all so what Im looking for maybe is more guidance from a structural engineer.

Thanks anyway.

 

[fmguru]

Depends on what you want to do. Some FE packages are tailored to deal with continum models and others specialize in skeletal structures. Not many handle both types efficiently.

Finele Consulting Engineers

 

[salmon2]

FEA is popular, but really much slower than just doing some tests and provides nothing unless the simulation matched experiments. A lot of big companies do not even do it.

Good FEA codes could require months, years practice before you feel comfortable with it. Simple FEA codes will allow you go much quicker at the beginning, but very limited down the road.

 

[GregLocock]

Really ? Tooling lead time is 18 months for some of our parts. It sure would be a shame to wait 18 months for a part, and then test it, and then have to redesign it.



Cheers

Greg Locock

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