Galapagos and Geometry Gym

Hi Jon and all.

Is there a to send a GH model to GSA for analysis, and get results back to GH for further action?

My intention is to try and use Galapagos to optimize a structure via GSA. Any thoughts on how to do it? Samples would be great.

Cheers,

Itai.

Replies to This Discussion

Permalink Reply by Robert Vier on March 10, 2011 at 3:25pm

Hey Itai,

nice to see you here, getting into structural mechanics ;)

I already asked myself the same question, since GSA has got an API (and samples doing an optimization from excel. included in the program's install folder) it is possible.

Either by scripting it in GH (hm not sure if this works)

or writing a custom GH-component (not that hard, just download the GH sdk-help in the 'help'-register and/or this ) maybe i will get into this some time this year.

for preliminary analysis you also could use GH 'karamba'-plugin (which recently was developed in vienna ;) ) working seamlessly with galapagos. me testing > http://shelf-rob.blogspot.com/

basically it is a linear FE-solver that makes it's beams out of lines, but until now just gives you 'the truth' = the displacements. pretty useful, innovative and fast, though.

best

robert

Permalink Reply by Jon Mirtschin on March 10, 2011 at 11:16pm

Hi Itai and Robert,

I have some working examples of this, I will be posting some examples to the blog soon.

There's still some issues I'm trying to resolve as best I can, if the model contains an error, it's not obvious how to extract it to report back to the user. Also, I need some time to create new components to generate strings for extracting results, I will prioritize those specifically needed.

If you have an example model or idea to test and explain this, on I'm happy to start letting this into the wider user base. Using a commercial solver such as GSA (and will follow hopefully with Sap, Robot, Strand7 etc) allows access to all sorts of advanced analysis options including non-linear analysis, rigid links, more advanced loading etc etc.

I'm also progressing quickly form finding, this would include membrane, tensegrity etc.

There's some performance issues with this, it's not exactly quick (although again there are some accelerations if you are only changing aspects such as properties and not entire topology). It's pretty exciting and powerful so look forward to seeing what others produce with this.

Cheers,

Jon

RSS