Is free (or very cheap) FEA software really available and worth it?

[Mech_Engineer]

Last night I tried answering a question I've had for a while- can you really do a proper FEA analysis using free (or at least very cheap) software? I've found myself recommending a couple of times to find a free FEA package for running an analysis on simple geometries, but I don't actually have first-hand knowledge towards that end... :uhh:

So at home I did a little looking around online and found LISA (http://www.lisa-fet.com/), an FEA package that is able to solve a nice array of structural FEA problems (and some CFD it claims). It's "free" up to 1300 nodes, and $80 unlocks it completely. So I downloaded LISA and messed around with it a little bit; the geometry meshing interface isn't the best (no surprise, I'm used to operating ANSYS which is around $35,000 per seat) but it does have the capability of making a simple FEA model for a block or other geometries. There are also some good tutorials available on YouTube and elsewhere to allow you to follow along in some basic structural analyses. Overall it seems capable but it will take me a little work to get comfortable with the interface and starting a mesh from scratch (it's hard to go old-school carbureted beater when you're used to flying in corporate jets )

http://www.lisa-fet.com/images/anim_usability.gif​

LISA is also able to import several different solid models including STEP, so my next question was can I find a simple CAD package (hopefully parametric) which is better in 3-D modeling and can export STEP for LISA. This basically mimics my at-work practice of modeling a design in Pro-E or SolidWorks and then import to ANSYS using parasolid or STEP. I ended up finding what I consider to be a nice little package on Sourceforge called FreeCAD (http://sourceforge.net/apps/mediawiki/free-cad/index.php?title=Main_Page ) This was a big find IMO because its geometry modeling capabilities are well ahead of LISA and it enables you to model a geometry and then have LISA import and mesh it.

Picture of FreeCAD interface:

http://sourceforge.net/apps/mediawiki/free-cad/nfs/project/f/fr/free-cad/thumb/3/33/Rim_bling.png/800px-Rim_bling.png​

So that's about as far as I got last night, but I think I've got the workings of a nice structural FEA solution for about $80. I haven't bought LISA yet but I'm going to try running an analysis tonight on one or two simple geometries and see how they turn out with only 1300 nodes.

 

[Mech_Engineer]

Student versions are good for students (but can get a bit pricey and typically have limitations), I'm more interested in software that is free or less than $100 and doesn't require you to be a student. Basically FEA for the engineer tinkering on home projects or a professional that doesn't have FEA access through his/her employer.

 

[Studiot]

This is a very good thread to start.

Thanks.

I've been trying to work up to shelling out (pun intended) for an FE program that I can't justify on economic grounds.

I will watch developments with interest.

PS I have been very disappointed with MathType and declined the last upgrade version because it still doesn't include either of the two things I want for internet forums, viz the ability to include tables and pictures.

 

[huntoon]

As for 3d parametric modeling, have you heard of Alibre? It's in the neighborhood of $200.

I had not heard of LISA, but will check it out.
Thanks.

 

[Vadar2012]

What about the Strand7 trial version. It's not too bad. I used to use it at home to avoid going to uni to use the full version. I could never find a decent free CFD program.

 

[AlephZero]

I could never find a decent free CFD program.

Give it time. At work we reckoned the level of development of CFD software lagged about 20 years behind mechanical FEA. One reason was the heavier computational demands of CFD meant it couldn't become "mass market" software product like linear stress and dynamics analysis. But modern PC hardware has enough oomph to run useful CFD models, so that only (!) leaves the problem of teaching the average Joe Engineer how to use CFD without getting nonsense resolts.

 

[Vadar2012]

so that only (!) leaves the problem of teaching the average Joe Engineer how to use CFD without getting nonsense resolts.

So true. It takes years to learn how to construct a proper mesh and to recognise when there's errors in it. For example, why does a mesh that is too fine start to yield worse results than a coarser mesh? You'd think finer the better.

I think the main problem is, most models require like 64 cpus to be run over a couple days, which can only be done on a supercomputer. For which no one is going to pay just to use at home. It also requires one to learn linux.

 

[berkeman]

Thread closed for Moderation (not of Mech_Eng)...

 

[berkeman]

Thread re-opened after some deletions and infractions. Student versions of software are for verified students.

 

[pantaz]

CalculiX is open-source FEA. I've never used it, so I can't comment on its usefulness.
http://www.calculix.de/
http://en.wikipedia.org/wiki/Calculix

 

[Vadar2012]

I was reading about FreeCAD on that website. It can do a whole heap of stuff. Quite impressive. Surely it has some restrictions, being free. One question though, what was it like to compile? There's always problems with optimisation when performing this stuff. Also nice to see something written in Python.

 

[Mech_Engineer]

I did some messing around and was actually happy with the results, especially for the price paid

First, downloaded a simple CAD model in STEP format from McMaster-Carr (a snap ring). Here it is displayed in FreeCAD:

So, the question is what does it look like when LISA auto-meshes it... I was pretty happy with the result with basic default settings, and the node count was just under the limit:

Next, I applied a basic set of fixed conditions to some nodes, applied some deflections to some nodes, applied some basic "steel" peoperties, and solved it. It was pretty easy, and the result was what I would consider an expected result.

I need to gain some better understanding of the units system as I think I'm off in there somewhere, but I thought the post-processing was good, had some nice-to-have display options, and makes overall output of the results easy to visualize (even has a video animation mode).

So far, I'm liking what I'm seeing considering it's all free up to this point I don't currently have an imminent need for home FEA, but if I did I think I could recommend LISA for basic structural at least. The node limit for the free version is probably better suited for 2-D models, but $80 really is a steal and that would allow you to solve some simple problems that I could imagine coming up.

I was happy with LISA's auto-mesh algorithm as well, importing the c-clip was as easy as opening it and setting a max element size. It automatically refined on curves, and put together overall a nice starting mesh (although it would need some refinement to get better stress results in the corners). This could be solved by meshing just the surface of the c-clip and doing a 2-D model, which I'll see if I can get done next.

FreeCAD is also a big win in this I think, it's an effective tool that has most of the "need-to-have" features in a 3-D parametric CAD modeling tool, and exports to STEP files for LISA. Its graphics are nice, the layout seems pretty intuitive, what else can I ask from a completely free piece of software?

 

[Mech_Engineer]

On the topic of free or cheap engineering resources I like to take advantage of as well:

Material properties:

• Matweb- nice website with many (unverified for the most part) material properties.
• MIL-HDBK-5J - Great resource for detailed material properties of engineering metals including Steel, Aluminum, Magnesium, and Titanium. Spec is "cancelled" but still very relevant, and easily found as a legal free download (link provided at assistdocs.com). Which leads me to...
• www.assistdocs.com - A great resource for MIL specs provided by the DOD. All MIL and FED specs are available as free PDF downloads (I can't believe there are sites out there that try to charge for them) with searchable OCR-scanned documents (or in some cases native PDF files). Very good resource for alternative specifications if you can't afford (or don't want to buy) specs from ASTM, ASME, SAE, ISO etc. All "cancelled" specs in the database also reference the spec that supercedes it (and you can still download the latest version before cancellation).
• 3DContentCentral.com - A crowdsourced CAD model database hosted by the makers of Solidworks (but parts are also available in neutral CAD formats), a nice resource for a variety of CAD models with searchable tags.
• McMaster-Carr - What's not to like about McMaster-Carr? They have a search engine that's best-in-the-business IMO, and they provide CAD models for a TON of stuff, mainly engineering component related (fasteners, components, etc.).
• NIST Thermophysical Properties of Fluid Systems - A great free resource provided by NIST with amazingly detailed thermodynamic properties of engineering fluids. A great resource for CFD properties!

 

[Mech_Engineer]

By the way I would like to revise my statement that LISA : ANSYS :: Carbureted Beater : Corporate Jet. I'm thinking LISA is at least as good as a mid-90's fuel injected vehicle, assuming you utilize external CAD capability for modeling something from scratch :rofl:

 

[ThermalModeler]

I've used LISA 8.0 for heat transfer simulations, including phase change and thermal stress, and it is quite capable for 3D transient simulations.

 

[Ryoko]

This looks interesting (but their website is lacking):

http://www.z88.uni-bayreuth.de/z88aurora/wasistz88_e.htm

 

[bigfooted]

Or this one: a linux distribution with a bunch of FEA/CFD software
http://www.caelinux.com/CMS/

 

[tioker]

I was on Alibre forum and someone there mentioned there is a new product called Mecway that is basically a more advanced version of Lisa.

http://forum.alibre.com/viewtopic.php?f=7&t=17263

Here is a list of features shown on their website, but I haven't had time to play with it yet.

 

[tioker]

I eventually got to looking into Mecway (http://mecway.com) and I feel I should say something about it here. As a Lisa user, I have to say it's very impressive! It's clearly based on Lisa but absolutely full of enhancements even though it's the same price (both are now $100). Things just feel more comfortable. Here are some differences I noticed:

Using STEP files is easier. In Lisa, you had to scroll through a list of surfaces to find the one you want. On a complex model with hundreds of surfaces, that was sometimes very tedious checking them one by one. In Mecway you simply click on it and apply a load or whatever right there.

Everything has units of measurement now which is a real breath of fresh air. I was restricted to using pure SI in Lisa because it was "agnostic" and didn't know anything about units. But with Mecway you can make dimensions in mm and mix that with force in pounds then show the stress in psi or MPa.

Mecway doesn't seem to use named selections anymore. This is a bit tidier because when you create a force it only adds a force, not an extra named selection along with it. There are also some new loads - an interesting one is frictionless which let's a surface slide in whatever direction it's at. With Lisa you could do this in theory using rotated coordinates on every node but it would be too impractical in most cases. Here's an image of it used on a screw head. It looks like it's working as expected with the head being pulled into the cone.

I've done a bit of sheet metal work before and shell elements are essential. Mecway extends these to include buckling and dynamic analysis which would have been a lot easier than building everything from bricks like I did with Lisa. Here's an image of local buckling in a channel steel.

Other things it claims to have are a nonlinear solver and a higher limit of a million nodes which I havn't been able to test using the demo but will be fantastic since I was alway running into the brick wall of Lisa's much smaller limit (60,000?). I didn't try the nonlinear solver either.

On the downside, the meshing is still a bit rough. Sometimes it created "invalid topology" elements although instead of waiting till you solve it to discover the problem, it displays a red X on the invalid elements immediately.

 

[Arek]

Hi Guys I have just seen this post and as I used LISA a little bit and I acquired a license for MECWAY I thought I will post few things.
Main thing is the question of the topic: Is cheap FEA soft worth anything. If you look at Calculix, Elmer, OpenFOAM, Code Aster and Saturn you will realize that YES it is and it is used by big companies, even nuclear firms are developing one and using it for civil, structural and CFD analysis.
@tioker - in LISA you can use any type of units but you have be consistent as just like doing analytical calculations, if you use wrong units in formula you will receive magnitude higher or lower results or just a scrap results. According to license LISA and Mecway are the same 100$ but just LISA allow you to use it commercially, in Mecway it's just for educational purposes or private without any commercial use, if you want one you have to spend 350$. In LISA you are able to do most of the things that on Mecway you will do , there are couple enhancements as geometrical nonlinearity, compression constraint, elastic support, stress linearization graph and few more which would be hard to perform/apply in LISA or not possible at all. I am going to do some Creo Simulate comparison with LISA/MECWAY and other software to help in learning them and FEA itself and I have already done some comparison. Simulate has got some tutorials which were done professionally so I take results as correct. Please have a look at results and compare if 100$ is worth it's price.
Comparison shows how close LISA and MECWAY are to Creo Simulate in this simple case with regards to stress and displacement. Creo is much more powerful and easy to apply boundary conditions but it doesn't do Flow Analysis for example and you can do it in LISA/MECWAY. Results of LISA/MECWAY were indentical, the mesh was generated identically using the same setting of course as it uses Netgen for meshing. Mecway was created by LISA's ex-programer so this is why they are identical. I like in LISA adding faces, volumes, nodes to one section and couple other things but I have to admit that Mecway wen't forward so much in ease of use and that it looks very promising. Please have a look at my website, where I used LISA for some analysis structural, fluid and thermal transient for real and theoretical applications. I intend to use it more and will try to compare its functionality with Creo Simulate as I just started to go thorough it's training. I will post in more detail comparison of LISA and Mecway to Creo and hopefully for some tough subjects which L/M are not capable to deal with eg. contact analysis I wil try to go for Calculix/Elmer or eOOS. My web page is www.arek.pajak.info.pl and of you want to have a look at some more free software please go here: http://feacompare.com/

 

[berkeman]

Thanks Arek! Welcome to the PF.

 

[Eng.112358]

You can try Heron Rotordynamics (Link to heronrd.com). The student license is around u$s10/month and a complete license near u$s100.

 

[A B H]

Alongside courses that are available for MOOC platforms these days, you can also get a hands-on experience through SimScale. It is a cloud-based tool where simulations can be set up on a browser. It does not need the hassle of licenses or installations. It allows structural, fluid and particle simulations.

SimScale offers a free version where one can get up to 3000 hours for computing. For the free version, the projects created are publicly available. However, if one would like to keep your project private, a professional version is also available. There is also a two-week free trial for the professional version of the platform.

The SimScale tutorials and materials provide easy to understand materials to help with usage of SimScale platform. In addition, you can explore the public projects database, where there are loads of projects already available.

 

[goeasyon]

WELSIM, a 3D FEM-based simulation software, may be useful for your projects.