Form finding with Galapogas

http://www.youtube.com/watch?v=kvySEV09cr0&feature=plcp

Using Galapagos to produce a schematic massing model. Galapagos tries to position programs based on a linear relationship while maintaining square footage for each program.

Used this process for my class project. Below

SITE

RELATIONSHIP DIAGRAM

ITERATIONS

FINAL RENDER

Tags: Galapagos, design, evolutionary, finding, form

Attachments:

foam finding.gh, 38 KB
site.3dm, 327 KB

▶ Reply to This

Replies to This Discussion

Permalink Reply by Edion86 on June 13, 2012 at 8:37am

That looks great, I'll give it a look. Seems a good point to start using galapagos.

Anyway... in which way is more efficient the final result than the others?

▶ Reply

Permalink Reply by George Faber on June 14, 2012 at 10:34am

Its still got some loopholes but fun to mess around with to see what you can get by changing up the relationship diagram (clustered, center,etc).

▶ Reply

Permalink Reply by Roberto Garcia Rosales on December 12, 2012 at 8:04pm

Hi George, we are currently working on a project for school where we started to develop program relationships in order to populate certain site.

We are planning to use Galapagos for some iterations of the program and we came across your example, however we were not able to run your script and we've been trying to figure out what happen. We are using Rhino 5.0 and grasshopper 9 and there are some function components who appear as "old" which (maybe)are limiting the script to run and watching the video we noticed that even the initial volumes doesn't appear. Would you let us some help? Thanks for your time.

▶ Reply

Permalink Reply by George Faber on December 22, 2012 at 6:39pm

Hey sorry for the slow reply.

I don't think the old components effect the script, but I could be wrong. Also, make sure you have the volumes turned on. I think they are hidden in the file I have above.

if you tell me more details about what you are trying to do, or send me a script I can take a look at it and try to help you out. I found through my project that there are still a lot of problems with my work flow and it takes a lot of tweaking to get a script and fitness to produce good results.

▶ Reply

Permalink Reply by djordje on December 28, 2012 at 6:21pm

Hi George,

This is very interesting project.

Can you tell me what do values like "ABS(6500-x)" and "X/6500" used in the Function component represent for each particular room?

Thank you.

▶ Reply

Permalink Reply by George Faber on December 29, 2012 at 10:47am

So, the two things I am having Galapagos solve for is to make each program be a certain square footage and be as close as it can to its adjacent program (that is the relationship diagram I have above).

In order to do that I need to give a numerical number to each of those properties and tell Galapagos to solve for either the minimum or maximum (in my case minimum).

To balance the different programs out I adjusted everything to be on a 0-1 scale. 0 being that it meets the requirement perfectly, 1 being the worst solution.

The "ABS(6500-x)" is the area. So I want my room to be 6500 sq feet. Lets say Galapagos produces a room that has a square footage of 4000 sq ft.

So, ABS(6500-4000) = 2500

the next part adjusts it to a 0-1 scale. 2500/6500= .38

now if you look at if the resultant sq ft was 6400. ABS(6500-6400)/6500= .01

the closer the area is to the desired then the smaller the resultant.

The next function components deal with the distance to the adjacent program. Once again adjusted to a 0-1 scale, I used x/702 because I think 702 is the furthest two points can be from each other on the site.

Finally, I add all these values for all the programs and have Galapagos choose the smallest number (i.e. the most "optimal" solution). So in theory the a result of 0 would be impossible due to the distance component would always be greater than 0.

I don't think I ever got Galapagos to solve for anything smaller than 4 or 5. There are still a lot of flaws, for example because everything is weighted the same (on a 0-1 scale) a result that produced all the programs to have the correct sq ft. and incorrect relationships would be the same as a result that gave the all the correct relationships and none of the correct sq ft.

▶ Reply

Permalink Reply by djordje on December 29, 2012 at 11:24am

Thank you for the reply namesake,

Can I quote you on some parts?

"I think 702 is the furthest two points can be from each other on the site."

How did you succeed in calculating this?

"a result that produced all the programs to have the correct sq ft. and incorrect relationships would be the same as a result that gave the all the correct relationships and none of the correct sq ft."

Above posted previews of "Iterations" represent that?

▶ Reply

Permalink Reply by Omar Helmy on December 30, 2012 at 4:29am

Very interesting study here, i have tried similar applications using Galapagos and got stuck at almost the same point (Weighing the parameters within the fitness)

Always wanted to try it out using the Kangroo geometry packing for the relationships and Galapagos for the Areas...

▶ Reply

Permalink Reply by andreaquartara on December 31, 2012 at 6:11am

Omar, can you post your files?

I'm trying to "populate" a lot with different volumes (the urban plan provides university, residences and businesses in presence) evaluating with Galapagos the best position on the lot (sun radiation, wind exposition, view...)

thanks