weaire phelan using galapagos

Hi again.

i just switched to gh 07.

galapagos seems great!!

i'm working on a weaire phelan structure

http://en.wikipedia.org/wiki/Weaire%E2%80%93Phelan_structure

this structure uses two types of geometry

Tetrakaidecahedron and Dodecahedron, wich have different kind of faces, hexagons and pentagons (Tetrakaidecahedron) and only pentagons (Dodecahedron)

With galapagos is quite easy to test different types of unions, but i have just been able to connect Tetrakaidecahedron with Tetrakaidecahedron, or Dodecahedron with Dodecahedron.

When i try to use both geometries, galapagos tries to connect hexagons and pentagons, so some unions wont work at all.

I think i have to list out hexagons so they onley connect with hexagons, but i have not been able to do it using galapagos

here you can find what i have at the moment

any advice is appreciated.

best

mario

Attachments:

galapagostest1.3dm, 139 KB
galapagostest1.ghx, 152 KB

▶ Reply to This

Replies to This Discussion

Permalink Reply by sujaygeorge on May 18, 2012 at 1:07am

hey jon,

i have downloaded structdrawrhino and have a liscence.

i downloaded the zip file and tried running it but for some reason the poly hedras do not follow the curve they form a huge box with 100s of them in it... pls help!!!!!

▶ Reply

Permalink Reply by Jon Mirtschin on May 18, 2012 at 1:14am

Hi,

Thanks for reporting. I think I've broken my polypacking routine some time in the recent past.

I'll try and check it shortly and post a new build ASAP.

Apologies, Jon

▶ Reply

Permalink Reply by sujaygeorge on May 18, 2012 at 1:29am

that would be such a relief jon...

looking forward for the new build...

cheers,

sujay

▶ Reply

Permalink Reply by sujaygeorge on May 20, 2012 at 2:31am

hey jon any luck with the polypacking new build?

▶ Reply

Permalink Reply by Jon Mirtschin on May 21, 2012 at 2:51am

I'm looking at it some improvements. For the model with this post, it seems the arrangement input has "changed" to boundingbox rather than intersect (I might have changed the order of the enumeration). If you try the attached, see if it works better. If you have other models you want me to look at, either email to me or post and I'm glad to help.

Attachments:

100903 weaire phelan.gh, 2 KB

▶ Reply

Permalink Reply by sujaygeorge on May 21, 2012 at 3:28am

wow jon that's absolutely brilliant!!!!!!

thank a ton!!!!

is there a way that the size of the polyhedron can be resized and if i could change them into ones having all sides as hexagons...

thanks a million and really appreciate the quick response...

Permalink Reply by Jon Mirtschin on May 21, 2012 at 3:51am

There's some different shapes in the various tessellation types accessible from the T input. Size can be controlled by the S input. The size is uniform without the pack, but you could perhaps apply a transform as a post process to get some variation if desired.

Permalink Reply by sujaygeorge on May 20, 2012 at 7:18pm

hey jon ,

would it be possible if you could get the polypacking working by a day or 2..

sorry for the trouble but in a desperate situation here for the project to move further... we are using a truncated tetrahedron for our system but the one u have currently used should be fine as long as it works... once again sorry for pestering just desperate times...

thanks in advance

▶ Reply

Permalink Reply by MARIO VERGARA on September 3, 2010 at 7:36pm

here's the file

Attachments:

galapagostest2.ghx, 366 KB

▶ Reply

Permalink Reply by Daniel Piker on September 4, 2010 at 6:13am

To answer your questions about the definition I posted -
The box has an area of 100, so it is trying to make each of the 3 cells have area 100/3 , which is why b = 33.333
The abs here means the absolute value.
So abs(a-b) finds how much (always positive) each cell's area differs from its target area.
I then took the negative of that (-abs(a-b)) because Galapagos always tries to maximize the Fitness, and used Mass Addition so it minimizes the total difference from ideal areas.
I was first assuming that you were trying to tackle Kelvin's problem with Galapagos - and wondering if it would end up with the Weaire-Phelan structure. I think this would be a very interesting one to try (especially as it is not proven that W-P is optimal, so there is a chance it might find something better!), but a little tricky to set up. You'd need to simultaneously ensure equal volumes and minimize total surface area, and for all of space, not just some finite volume. If you do pursue this, then Ruggero Gabbrielli would probably be a good person to talk to.

But from the image you posted it looks like the regular Weaire-Phelan packing, just with some cells removed. I don't really see how a genetic algorithm would be very useful in this particular case.

▶ Reply

Permalink Reply by MARIO VERGARA on September 4, 2010 at 8:13am

Daniel.
Thanks for the clear explanation.
I will try to contact ruggero.
the image is just for reference, i'm not trying to replicate it
testing wp structural behavior is really what i'm trying to achive
i'll keep on working taking the concepts you described.
i''l post something asap
thanks again