Form finding optimization with Octopus

Hi everyone,

I am working on a project to find the form of a facade for multiple objectives.I did a run of simulation combining the objectives at the end(see image) for total radiation for cooling and heating period.The variable was the glazing percentage of an external wall of one room.

I got a huge number for objectives and the simulation run almost 9 hours through the night and at the end(when I stopped it)there where no results to export.Except of the screenshots I made.I read the octopus manual and tried to understand from the forum discussions on how and what to select but still it seems to have lots of errors.

Any insight will be very helpful.

Attachments:

WWR_Octopus.gh, 1.4 MB

Replies to This Discussion

Permalink Reply by Robert Vier on May 12, 2015 at 5:55

hi nite,

your file gives me a very strange definition upon opening. also, all the objective values are 'null'.

could you please simplify and double check the file?

thanks

rob

Permalink Reply by Nitë on May 12, 2015 at 6:00

Hi Robert,

Thanks for the fast reply.those are components form Ladybug and Honeybee.Will simplify the file right-away.

Permalink Reply by Robert Vier on May 12, 2015 at 6:04

i see those components when i zoom in. just updated ladybug and honeybee yesterday. still they disappear when zooming out, and the objectives are zero.

Permalink Reply by Nitë on May 12, 2015 at 6:12

I have no idea why they appear to be null. :(

Permalink Reply by Nitë on May 12, 2015 at 6:06

Hi Robert,

here is the new file and .epw for Berlin.

thanks,

Nita

Attachments:

DEU_Berlin.103840_IWEC.zip, 239 KB
Octopus_WWR_new.gh, 885 KB

Permalink Reply by Robert Vier on May 12, 2015 at 6:29

Dear Nite,

you have connected only one parameter to be altered by octopus. Where is the point of the heuristic optimization? you could just use a newton scheme to quickly find the best setting for the single slider, or try manually.

this kind of optimization just makes sense if the problem is more-dimensional.

best

Permalink Reply by Nitë on May 12, 2015 at 6:38

Hi Robert,

What I am trying to do is a workflow similar to this video : https://www.youtube.com/watch?v=fYw7KexxThM .To simplify, I want to explore how and if form finding of a façade through optimization influences the energy and daylight performance of the building.

The point at this first step is to optimize WWR and than the "free form" of the facade.

Objectives: total radiation, daylighting, sunpath etc..In this file it is only Window to wall ratio.

Further when I put the whole facade as genomes(havent figured it out yet, if it should be a mesh or sliders), the problem will get more dimensional, I suppose..

Thank you,

Permalink Reply by JPE on May 12, 2015 at 20:04

Nitë –

Although this thread is primarily to address the use of GH+Octopus, I am compelled to point out that the objective function as it is now defined will produce a 2D solution (ratio). The experiment presented in the GH file will generate a simple curve as seen in your first image, which is nothing you could not calculate using a few E+ simulations with random WWR values and Excel to plot the slope and solve for the r value (or even calculate by hand if you feel so inclined). As you are trying to minimize the objective function in the Berlin climate, I suspect that you will have an “optimized” result at or around 0% window to 100% wall if you are solving for energy use and the opposite ratio if you are testing for maximizing available daylight.

If you change the objective criteria to minimize energy use (heating, cooling, lighting) AND maximize average daylight levels through manipulation of the WWR, your results will still be a 2D curve but less of a line as is seen.

Cheers – James

RSS