Open source FEA for spaceframe analysis

[TimWrightTW]

Hi,

Does anyone know of a simple beam FEA solver that would be suitable for a spaceframe modelled with beam elements? This is not a uni or commercial project, this is just for my own project car.

Ideally I'm looking for something which will import a node and element description of the model and perform a static analysis based on a loadcase.

Specifications would be;
Solving for tensile, shear and bending forces/moments
Ability to add pin joints would be a plus
Frequency and modal analysis would be a plus
Bonus points if it has a post processor

I want to use it as a way to quickly try different configurations of the spaceframe to optimise the beam placements.

Matlab code will also be considered.

My intended work flow is;
1. Model frame in CAD program
2. Export wireframe beam representation to .igs or .stp file
3. Convert wireframe data to FEA input data in Matlab
4. Perform torsional and bending analysis in open source FE solver
5. Post process n Matlab

solid/surfacing elements are not required.

So far I have seen a few possibilities, so if anyone have experience in any of these, please pipe up!
Calculix
Code_aster
Elmer
FEAPpv

Cheers

Tim

 

[WolfHR]

There is a small (I like those the best) program for analysis of spaceframes- Analysis 2.04, but I don't know how suitable it is for automotive applications (I think it's primary use is civil engineering). What I like about it is that it is 'wireframe' based- simple, fast, ideal when designing the frame- doesn't require CAD modelling, and what's more important takes into consideration buckling of members.

It can be downloaded here:

http://www.cuylaerts.net/

(and it's only 500kB).

P.S. Sorry if I sound overly enthusiastic about this proram before I've used it in anger- I've been looking for something like this for a long time, and stumbled upon it only recently.

 

[TimWrightTW]

Cheers Mr Wolf,

This looks quite promising after a quick look. An automotive spaceframe is just a big 3D truss, so this software seems quite appropriate.

Tim

 

[GregLocock]

OpenFEM for matlab will do exactly what you ask, I suspect. Leastways it has beam elements, and is free.

Calculix is a great FEA program but last time I checked it didn't have beam elements.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[flash3780]

CalculiX lists beam elements in their overview:

http://www.dhondt.de/ov_calcu.htm

If you decide on CalculiX, I'd encourage you to join their users group on Yahoo. It also shares syntax with ABAQUS, so you should be able to leverage the ABAQUS documentation a bit. I'm not a user, but it's on the top of my list for programs to become well-acquainted with. From what I can tell, it's the best open-sourced FEM software that exists today.

//signed//
Christopher K. Hubley
Mechanical Engineer
Sunpower Incorporated
Athens, Ohio
--

http://engineeringliberty.wordpress.com

 

[GregLocock]

Well that is great news.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[ESPcomposites]

CalculiX does have beam elements, but they are not standard beam elements. Of course the shells are not standard shells either. Everything is a solid element. The beams are expanded into solids. It is quite clever, but does have some limitations. I have found it difficult to post-process sometimes because of the node reordering.

Nonetheless, it is one the better (if not the best) open source FEM solvers.

Brian

http://www.espcomposites.com

 

[GregLocock]

Oh, do you mean they are the original formualtion in calculix? I did a comparison between those and real beam elements on the calculix discussion forum, short answer is that they aren't good enough for my work.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[moa88]

Hi! I am working on something quite similar to what Tim has done. My aim is to script Calculix from Matlab; I mean I want to insert input data using MLab and, using a GUI, by pressing a SIMULATE button I want to call automatically Calculix and get the results for postprocessing in MLab. In this process the GUI's user has not to deal with Calculix at all.
Have you got any advise on how to do this?

Cheers

Gioacchino

 

[GregLocock]

It seems to me the only detail of interest is how to run calculix from a shell command, which is surely in the help file.

>ccx inputfilename

would appear to be enough

Everything else is the stuff you think you now how to do (building the model and post processing it) or boring file stuff.





Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?