The problem of daylight simulation

Hi, I find the daylight simulation component in my file cannot generate the annual files and show that the result for the study cannot be found? I compared the file and the example and I think the setting is right. So, what’s wrong? The attachment is the file. Thank you for your help.

the problem of daylight (199 KB)

Hi Han, The name of the zone had white space which was causing the issue. In general don’t use white space in naming in Honeybee.

Mostapha (204 KB)

Thank you Mostapha and sorry for my late response.

Could the honeybee perform simulation for the free-form building? The video showed that it could perform simulation when the building form was rotated. But when I do that, some of the window surfaces will be disappeared. However, when I turn the angle to a higher value, the surface will show again. Does honeybee could only work during some rotate angles?

Hi Han,

I need your file to check. Honeybee should generate the window for both cases. It can also be a tolerance issue. Sometimes setting up a lower tolerance helps but it can add up to the time of creating windows.


Hi Mostapha,

Thank you for your response. The attachment is the GH file.

Thanks a lot.

Yunsong HAN

The problem of the rotation (165 KB)

Thanks! I added this as an issue and will look into it soon. (


This is a bug with the Rhinocommon “OffsetCurveOnSurface” method. If the curvature becomes too great, the method fails. There’s nothing that we can do about it as far as I understand except report it to Guilio and hope that it is better in Rhino 6. I imagine that offsetting the window in Rhino manually would likely produce the same erroneous curve. Han, if you really need a window with that great of a curvature, you might try a few different manual method in Rhino (like trimming the original surface with planes) to get the geometry of the window you desire. Then you can use the surface-by-surface method to make your zones.

Mostapha, are you ok with me closing out your github issue?


I should say that it might not necessarily be large curvature that causes the method to fail. It’s something on McNeel’s end, though, and hopefully they will figure it out if we report it to them.