Open source FEA softwares

[Silente]

Hi all,

lately i have been looking for some decent and user friendly open source FEA software to perform simulations of both space frame chassis and solid models (parts).

I found some interesting feature in a software called LISA, which moreover offers the possibility for a complete version at only 50 $.

But i would really appreciate to have your opinion about that. Have you ever used Open Source FEA softwares? Do anybody knows if there is some decent tutorial around on how to use them (i am not particularly inside FEA and i think i have alot to learn)?
Last, do anybody has ever checked the quality of CAD FEA modules results (like cosmos or INVENTOR simulation)?

Thank you very much.

[Silente]

Hi all,

lately i have been looking for some decent and user friendly open source FEA software to perform simulations of both space frame chassis and solid models (parts).

I found some interesting feature in a software called LISA, which moreover offers the possibility for a complete version at only 50 $.

But i would really appreciate to have your opinion about that. Have you ever used Open Source FEA softwares? Do anybody knows if there is some decent tutorial around on how to use them (i am not particularly inside FEA and i think i have alot to learn)?
Last, do anybody has ever checked the quality of CAD FEA modules results (like cosmos or INVENTOR simulation)?

Thank you very much.

[Spetsnazos]

Note that there is a significant difference between free and opensource.

Honestly I've looked when I first started FSAE and I didn't find anything. There would be little reason to produce opensource FEA software because the number of users would be so limited. A large majority of people using FEA software aren't programmers or even computer people, they are engineers and designers.

Free software? Well I have seen some limited versions of ProE but otherwise its usually free for 30 days.

check with your school to see what kind of licenses they can get you. That is probably the best legal route you can take

[Demon Of Speed]

A few CAD companies sponsor FSAE teams. I know for sure that autodesk and solidworks are sponsoring teams with at least free software, try contacting someone and see what you can do.

[Simon Dingle]

Ansys gave us free Workbench and Fluent licences a couple of years ago. Not sure if they're still doing it, but well worth asking.

[Silente]

thanks everybody for the replies.

Spetsnazos, i didn't talk about free software but about open source.

I have tried both autodesk inventor and solid works modules, but would like to understand if i can trust their results, guessing i am modeling the problem correctly (of course, this is experience depedent, but also, for example, conneceted to the possibility that the softwares give with constraints and forces).

Did anyone tried any open source program anyway? or know of good tutorials to follow for the first steps?

Thank you very much

[Spetsnazos]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Silente:
thanks everybody for the replies.

Spetsnazos, i didn't talk about free software but about open source.

I have tried both autodesk inventor and solid works modules, but would like to understand if i can trust their results, guessing i am modeling the problem correctly (of course, this is experience depedent, but also, for example, conneceted to the possibility that the softwares give with constraints and forces).

Did anyone tried any open source program anyway? or know of good tutorials to follow for the first steps?

Thank you very much </div></BLOCKQUOTE>

What OS are you running? Ive somewhat of a positive experience in Linux using Z88. The problem with Z88 is that it has a freeware version and an opensource version. The opensource version doesnt have a way to import from solidworks, the mesher is limited, no shell element support, and doesnt support multi-core processing as far as I could tell.

I couldn't get Calculix to work so I'm not even going to comment on it. I spent a solid 2 hours and then just gave up.

Trust me, Solidworks Simulation is sufficient for most applications and opensource software is generally for very simple geometries. Think about it, if you spent all that time developing a FEM software, wouldn't you try to sell it?

Anyways let me know if you want more info. I'm running Ubuntu 11.04 ifyc.

[Spetsnazos]

by the way, when looking for FEA/FEM software, make sure it is capable of pre and post processing.

[Silente]

Hi,

thanks for your reply.

I am using autodesk inventor at the moment (i am really missing solid works!) and it has a FEA module where i tried to perform some simulation. You have some serious limitation on force and contraints definition and sometimes how it works it's not completely clear and you don't find so much informations in help files.

But to be honest, the OS code i found seems to be far more complete, already in the modeling phase. For sure, it is also less easy to use and it seems you need to gain some experience to use it properly. It also has a 300 cells limit, but you can remove it with this 50$ license. So it's not a big deal.

i read that with OS the companies earn money by giving support, not by selling the software itself. This is a different way of doing business, since soon or later nearly everybody need to be trained to use this softwares to their maximum. On CFD, there are a lot of applicantions which are growing on open foam for example, which is an OS code you can download for free i believe.

I would like to understand if it is possible with any sort of FEA, since i have much more interest in learning it.

are you an expert user of FEA? what kind of software you are using?

[Spetsnazos]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Silente:
Hi,

thanks for your reply.

I am using autodesk inventor at the moment (i am really missing solid works!) and it has a FEA module where i tried to perform some simulation. You have some serious limitation on force and contraints definition and sometimes how it works it's not completely clear and you don't find so much informations in help files.

But to be honest, the OS code i found seems to be far more complete, already in the modeling phase. For sure, it is also less easy to use and it seems you need to gain some experience to use it properly. It also has a 300 cells limit, but you can remove it with this 50$ license. So it's not a big deal.

i read that with OS the companies earn money by giving support, not by selling the software itself. This is a different way of doing business, since soon or later nearly everybody need to be trained to use this softwares to their maximum. On CFD, there are a lot of applicantions which are growing on open foam for example, which is an OS code you can download for free i believe.

I would like to understand if it is possible with any sort of FEA, since i have much more interest in learning it.

are you an expert user of FEA? what kind of software you are using? </div></BLOCKQUOTE>
No prob man but I'm by no means an expert. I learned it through school and practice.

If the software your using is trully opensource then it should be fairly easy to remove the 300cell limit and there should not be a $50 fee. What software are you using and what prog lang is it written in? I could possibly help.

About the only company I know thats opensource and is able to make a profit by providing support is redhat...although I don't know many individual users that get it.

I agree with you that opensource is the future, but once someone learns a specific program, they are very resilient to changing to something else.

At the moment I've only played around with Z88 and Calculix. I can use Solidworks Simulation just fine(hardest part is setting up boundary conditions anyways).

If you're really into CFD check out OpenFOAM. I have no experience with CFD, but I read articles that show it to be somewhat decent. It supports pre and post processing and is capable of dynamic mesh.

Just remember that OpenFOAM has no GUI...

Remember, a big part of FEA/FEM is that you trust the calculations behind it. If I am able to tamper with the calculations to achieve the desired results, how credible is that?

The big reason behind why companies go with the trusted names like Solidworks and ProEngineer(Creo) is peace of mind knowing that there is lots and lots of money in the development of those packages. If the software was written poorly and lots of projects failed, the company would likely go out of business from lawsuits and bad press.

[Silente]

the software i was talking about is that:

http://www.lisa-fet.com/

the limit is not 300 but 1300 cells, i had a mistake.

The possibilities seems to be that of a real FEA software. I don't know if it is actually a real OS, looking better to the site they talk about a free software.

i was working with inventor FEA and to be honest the limtis you have on force and constraints make me think it is not so trustable. You van have an idea of how the stree and deformations are going to be, but if you can't model at least constraints in the proper way, i think it's impossible to trust the results.

Don't have real experience with Solid works FEA, i used Solid Works a lot and with great pleasure but have never used its FEA module. I just used cosmos works some times but don't know if they now developed something different.

And actually don't know how much time and money they spend to make these packages ok. their job is manly connected to 3d cad, not to simulation. Real FEA are Abaqus, MSC Marc, ansys. But they are too expensive and too time consuming to learn properly.

Have you ever tried to validate solid works results with some test evidence? or with some other FEA softares?

[wagemd]

I have found SW Simulation to be sufficient (even if not idea) for most things other than composite and some assembly FEA.

The most recent simulation seems to have a decent set of constraints... One thing to note, SolidWorks includes a Simulation and a Simulation Express. Express loads automatically with SolidWorks, but you have to manually enable the add-in for the full blown version. Express is crap as far as I can tell (extremely limited loads and constraints).

I have a feeling that SW has invested quite a bit into Simulation. Remember that SW is owned by Dassault who also own Abaqus and Catia. There are quite a few companies in industry that use Simulation as their primary FEA tool.

I would highly recommend validating FEA physically regardless of package. We have done a bit of validation in the past, but nothing I remember quantitatively at the moment.

[Pico]

I can say I personally checked a result from Ansys Workbench with some hand calculations and it was fine. I wouldn't dismiss the open source one but with 1300 nodes you're very limited in the type of model you're going to mesh. You'll get 1300 elements max and from experience to mesh a clevis is 5000 or more.


Checking if you FEM correlates with a test is very involved to the point where it takes hundreds of checks to see if your model is well defined and that you'll get a converging result. They're all based on the same solving techniques with tweaks between them.

All FEA software follow the same pre-processing procedure. I've used Ansys + WB, Patran, Cosmos, Marc. They're all time consuming to learn properly but it's worth it to learn at least one. I'd suggest Ansys WB as a starter.

I'm going to personally try LISA, let's see what she's made of http://fsae.com/groupee_common/emoticons/icon_smile.gif

[Silente]

I would really love to have your opinion about it.

It seems to have similar functionalities than "real" FEA softwares, like Marc.

But i am not expert at all, so maybe i just see a small part of the pack.

I wish i can have some time to learn how to use it in a decent way, it would be really useful.

[Joe_Little]

MSC has free students versions of Nastran and Mark, all you have to do is show you're a student by sending a scan of your school ID and you get a version for your laptop. There is some restriction, # of nodes I would assume, but it'd likely more than enought to do a beam model chassis.

http://www.mscsoftware.com/con...-Center/Default.aspx (http://www.mscsoftware.com/contents/Academia/Student-Center/Default.aspx)

Also, FSAE teams can register for the FEA and Motion bundles, getting full versions of Nastran, Patran, Marc, Adams, etc. This requires a little paper work and is "sponsership" meaning you put an MSC logo on your webpage and car.

Email university@mscsoftware.com

Obviously, this is applicable if you want "free", not "open source"...

This don't mean much in academia, but these bundles are nearly half a million worth of commercial software. To put it into perspective, solidworks premium simulation costs maybe $10k. Serious analysis isn't done with CAD programs, and their simplicity can be deceptive, so use them with caution.