/sci/ - Science & Math » Thread #12375748

276KiB, 1324x769, Finite-element-analysis[1].jpg

View Same Google iqdb SauceNAO

Recommended FEA Software Packages

Anonymous Mon 23 Nov 03:06:03 2020 No.12375748 View Reply Original Report

Quoted By: >>12375798 >>12375859 >>12375995 >>12376005 >>12376298 >>12376648 >>12376784 >>12377558 >>12377700

So I have no idea if this is the right forum for this, but maybe some anons can help me out. I'd like to do some FEA Modeling, but don't have $6000 to pony up for a copy of LS-DYNA; and have no idea where I can "find" a working copy of MATLAB. Any suggestions on alternative packages I could acquire, or options for someone to do this? I'm no longer in college so getting a student license isn't realistic ATM. I'd also like to do this myself, if possible- Hiring someone else doesn't help me learn the software or let me play with some of the ideas i'd like.

Anonymous

Anonymous Mon 23 Nov 2020 03:26:36 No.12375798 Report

Quoted By: >>12375830 >>12375890 >>12375995 >>12376242 >>12376651

>>12375748
this is one of the areas of computing where FOSS hasn't caught up yet. you either learn FEM methods and program them yourself, or you pony up and pay for a license.

programming your own FEM can be interesting (see Galerkin method), but there are a lot of nuances to handling different PDEs and BCs optimally, and several non-trivial tasks that have to be done (like meshing). i was personally fortunate enough to be taught advanced and computational EM by a guy that runs an EM modeling company, and want to start putting together open source FEM on day, but that's after graduation

also, never EVER use matlab unless forced to. it wouldn't help with FEM anyway, but that's just sage advice.

Anonymous

Anonymous Mon 23 Nov 2020 03:36:20 No.12375830 Report

Quoted By: >>12375870 >>12375995

>>12375798
>also, never EVER use matlab unless forced to.
-t. asshurt CSvirgin who has an autistic meltdown when people don't reinvent the wheel every time they want to write a hello world

Anonymous

Anonymous Mon 23 Nov 2020 03:43:05 No.12375859 Report

Quoted By:

>>12375748
use FEniCS:
https://fenicsproject.org/

Anonymous

Anonymous Mon 23 Nov 2020 03:48:49 No.12375870 Report

Quoted By: >>12375874 >>12376242

>>12375830
wrong, Cleve. 7+ year matlab expert that can make that piece of shit sing louder than your mom last night

Anonymous

View Same Google iqdb SauceNAO 1591477354710.jpg, 54KiB, 600x750

Anonymous Mon 23 Nov 2020 03:50:11 No.12375874 Report

Quoted By:

>>12375870
>piece of shit
Oh shit you've done it now

Anonymous

Anonymous Mon 23 Nov 2020 03:55:15 No.12375890 Report

Quoted By: >>12375971

>>12375798
really autistic question but is it possible to learn this power without actually learning pdes first? like learn the numerical methods concurrently alongside pdes?
t. have not touched a math class beyond linear algebra in 4 years

Anonymous

Anonymous Mon 23 Nov 2020 04:23:04 No.12375971 Report

Quoted By:

>>12375890
there's enough extra consideration for implementing numerical methods that it really pays to have a good understanding of the math in question before setting out. the main benefit is that you have a better chance at deciding if your results are reasonable or not. moreover, people generally get almost nothing out of scientific computing if they aren't somewhat familiar with the underlying topic

for PDEs, this practically means being familiar with separation of variables, knowing the solutions to a handful of ODEs that keep popping up, and understanding that the separated solutions form bases upon which general solutions can be constructed.

linear algebra is absolutely a prerequisite, though. most important math i know by far

Anonymous

Anonymous Mon 23 Nov 2020 04:31:56 No.12375995 Report

Quoted By: >>12376242

>>12375748
I tried many, and they all sucked. Never tried LS-Dyna though.

I'd suggest using fenics, which is free and open source.

>>12375798
fenics and feapPV are foss.

>>12375830
He's right. Matlab isn't only shit, students are raped by it with mandatory classes to make them work with it later in life for heavily overpriced licenses.

Anonymous

Anonymous Mon 23 Nov 2020 04:35:29 No.12376005 Report

Quoted By: >>12376014 >>12376242

>>12375748
Unity, Python (plenty of libraries), opengl, javascript, blender, any programming language.
I highly recommend learning python.

Anonymous

Anonymous Mon 23 Nov 2020 04:38:04 No.12376014 Report

Quoted By:

>>12376005
https://www.researchgate.net/post/What-is-the-best-finite-element-library-for-Python-programming-language

Anonymous

Anonymous Mon 23 Nov 2020 06:17:29 No.12376242 Report

Quoted By: >>12376323

>>12375798
OP Here. Yeah, I wanted to model ballistic impacts on steel / composite plates; Just trying to verify something that was written in a book 100 years ago, and there isn't any info on web search. I very much know the pains of proprietary software;
>>12375870
If I could find a booty copy of matlab, i'd be using it already lol. Last copy I had with working toolboxes was in 2017. The copies on the net that claim to be unlocked don't work, admittedly I haven't dug too hard yet.
>>12375995
>Fenics and FeapPV
I'll look into these, thanks!
>>12376005
I was trying to avoid building from scratch, if only because this is just seeing if an idea may or may not have any merit. I run into serious opportunity cost problems these days and never have enough time. I can do this though if need be.

Anonymous

Anonymous Mon 23 Nov 2020 07:01:07 No.12376298 Report

Quoted By:

>>12375748
Foss engineering linux distro: caelinux.org
Thid includes salome-meca which is based on code-aster

https://www.caelinux.com
https://code-aster.org

Anonymous

Anonymous Mon 23 Nov 2020 07:12:38 No.12376323 Report

Quoted By: >>12376649

>>12376242
If you check the link you will see that you don't need to build it from scratch.
Probably still need to make your own modules but I think the learning experience would be worth it.
It's experience you could use to get a dev job if you're ever in a position where you can't find work

Anonymous

Anonymous Mon 23 Nov 2020 10:05:46 No.12376648 Report

Quoted By:

>>12375748
1) this is not a forum
2) it is the right place, welcome fren

Anonymous

Anonymous Mon 23 Nov 2020 10:06:13 No.12376649 Report

Quoted By:

>>12376323
I'll definitely look into it. I actually am looking for work (Brainlet recent grad EE here who pretends to know FPGA and RF stuff; I got C's in my higher level maths); but it's nice to know my little side project might have some career value.

Anonymous

Anonymous Mon 23 Nov 2020 10:06:49 No.12376651 Report

Quoted By:

>>12375798
>program them yourself
any books / resources you would recommend?

Anonymous

Anonymous Mon 23 Nov 2020 10:55:37 No.12376784 Report

Quoted By:

>>12375748
You can try openfoam

Anonymous

View Same Google iqdb SauceNAO 1430496792850.png, 283KiB, 625x941

Anonymous Mon 23 Nov 2020 17:29:10 No.12377558 Report

Quoted By:

>>12375748
>All those fucking triangles for absolutely no reason

Anonymous

View Same Google iqdb SauceNAO 41T3kV8CktL._SX332_BO1,204,203,2 (...).jpg, 21KiB, 334x499

Anonymous Mon 23 Nov 2020 17:41:40 No.12377610 Report

Quoted By:

if this is the numerical thread does anyone have any strong opinions on pic related?

Anonymous

View Same Google iqdb SauceNAO i happen to be an expert.jpg, 123KiB, 1920x1080

Anonymous Mon 23 Nov 2020 18:01:07 No.12377700 Report

Quoted By:

>>12375748
Consider the Mystran solver + pyNastran pre/post-processor, especially if you plan on sticking to linear static FEM, and want to learn about using Nastran (Mystran inputs/outputs are exactly like Nastran)

https://mystran.com
https://github.com/SteveDoyle2/pyNastran

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches