Daylight results when models are close to / far from origin

grasshopper honeybee
1

Hi all,

I have encountered some differences (quite big) in results from HB Annual Daylight when using the same model but

    1. placed far from origin (532800, 182000, 0) which references the British National Grid coordinates
    1. brought close to origin in model space (0,0,0)

The first test suggests a quite a dark floor plate - less light through the southern facade than the western side (even though there is much less obstruction to the south) - I struggle to understand the results in this case:

image
image1482×597 89.3 KB

The second test produces more convincing results as light comes through the western and southern facades as expected:

image
image1636×603 121 KB

In both tests, a threshold of 300 lux is used to visualise DA results.

Does HB not work well if geometries are very far from origin? would be grateful if anyone could shed any light

2 Likes
3

Hi @wxcchuck,

This is more of a Radiance problem. Not only will the results be obviously wrong, but the runtime can also increase drastically. Here is another example. The solution is, as you have found out, to move the model close to the origin. In case you do not know, you can use HB Move to move the model if you do not want to move the original geometry.

4

Thanks @mikkel

I was trying to use HB Move but it failed to run the analysis.

image
image480×519 16.6 KB

I basically used HB Move to move HB Faces, Apertures and Sensor grid, am I missing something?

5

Hi @wxcchuck,

Did you move the Faces, Apertures, and SensorGrids individually or did you move the whole model?

Can you share the log file or your script so I can try to recreate your issue? The log file is found in your project folder, e.g:

C:\Users\[USERNAME]\simulation\sample_model_grid\annual_daylight\__logs__\logs.log
6

Hi @mikkel

I tried with both, but neither worked…

The log file below is from the scenario when HB objects are moved individually:

Thanks :slight_smile:

7

Thanks, @wxcchuck

The issue happens when creating the Radiance octree, likely because there are too many polygons in the model or too many polygons sharing the same vertex. Can you confirm if the recipe runs if you do not move any geometry? (although with wrong results as shown in the images in your first post)

8

@mikkel

Yes the recipe runs successfully before I used HB Move component. Also the same model has been used for all the tests…

9

Hi @wxcchuck,

Are you able to share the Rhino / Grasshopper file so I can recreate the issue? You can send a PM if you cannot share it in public.