Problem with tiles optimization in Refinery

Hi All,
I made a script that divides the surface based on tiles’ size. Then I created a 10x10 m floor and divided it in 1x1m tiles.
The goal was to find the best solution of tiles positioning for minimum cut-off tiles.



As you can see, the best solution in this case is a 0 translation in both axis like shown below.


image

Howeverr, on Refinery I can’t manage to find that 0 0 solution…


What’s wrong? I just can’t find the issue.
Any suggestions are welcome, thanks!!

The fact you have a consistent result makes me wonder if you’re pulling the evaluation metrics after a Revit API call - can you post your graph?

Sure, here it is.
teste generative design piso.dyn (121.2 KB)

Which Revit version are you in, and what are your project units?

I’m using Revit 2021 and my project units currently are centimeters for length and square meters for area.
image

1 Like

Everything is happening without a Revit interaction, so my hunch was off… It may be that the design space is too wide and the results for anything other than the ‘perfect’ solution are the same - that is you’re always discarding a full tile for each row/column unless it’s the 'full tile offset… I can’t confirm this without some more data though.

Towards that end, can you provide the results logs and the HOF results from your Generative Design’s appdata folder? Navigate to %appdata%\GenerativeDesign and get the both the hof_hist-results-****.csv file and the sln_set_hist-results****.csv which have modification dates that align to the date your last attempt finished, as well as your Refinery-Server-Log.txt file and I’ll confirm there.

Strange, those files are missing from my folder.
I only found the server-log


refinery-server-log.txt (997 Bytes)

Try Generative Design with the space - there was a name change which can lead to some inconsistency. You may have to go to just %appdata% and find the relevant folder.

Sorry for the late response @JacobSmall, I couldn’t upload csv files here, so I uploaded on wetransfer:
https://we.tl/t-ug2TfBBgIH