I’m running a daylight simulation for a notional residential building, and would like a second opinion as to the results being obtained.
The aim is to obtain an average daylight factor results for each space shown (side note: running an annual simulation for this model is far more time intensive than previously thought - to the point where Rhino crashes from lack of memory). I haven’t encountered any issues modelling geometry and setting up the simulation, although post-processing the results has left me second guessing myself somewhat.
I have attached the GH file for info - which used this feedback as a guide. Specifically, I’m looking for some input related to:
- The script shown to produce average daylight factor results and show on model. Are these values correct? How can I post-process the figures to show a % DF figure, rather than shown to decimal place?
- Comparing this model to the same design ran through a seperate software shows quite considerable variation in results reported- hence my reservations.
Thanks in advance,
GH Daylight Analysis _OKR_Tower INT.gh (889.6 KB)
- The default Radiance parameters can give you a high error. See this: DF variations in different runs
- If the values are less than expected and you are using the version from Food4Rhino and not the one from GitHub see this: Ground Light in Daylight factor
In any case, if you provide the expected results it will be easier to answer to this question.
@mostapha I’ve attached some screenshots of (1) HB results for each space, and (2) Comparative results in Sefaira.
Please let me know your thoughts.
Thanks for sharing the images. It would help if you share whether the legend scale are the same in both the tools. Also is the sky type same in both the studies?
From the image it seems like there’s not much of any shade to the windows…
- For a without shade window, I doubt the Sefaira results. The daylight factors drop too early as we go away from the windows.
- In the Sefaira image, at the windows on the right hand side of the building, I observe a directionality in the daylight factor results. A 45 degree pattern can be recognised. For a DF study, this should not be the case, as the overcast sky is supposed to be rotationally invariant.
The colours suggest a particular difference between the penetration of the light deep into the space. Perhaps you are not using the same number of ‘ambient bouces’ in the calculations. A lower number of ambient bouces with result in lower light levels particularly at the back. 6 ambient bounces would be the ballpark looking at your plans.
But first id try to standardise the colours and scaling of the results mesh - they might be implying a difference that isnt there. Better still, put the values on the points. Easy in GH, i dont know about seferia.
Agreed with the comments from @devang and @john.everist. It’s really hard to comment on the images with two different color scheme. Can you get the numbers out of Sefaira?
For the results from Honeybee we need to see the section so we know the relationship between the height of the window and the depth of the room. The pattern looks right to me.
This is how DF results should look. No directionality in the results.
@devang Yeah, My model is running using Overcast Sky. So there is no directionality .If @jwoodall running Sefaira using Sunny Sky with sun, the result will be directional.
Thanks @minggangyin @devang @mostapha @john.everist for your input and feedback, and apologies for the long delayed response.
Unfortunately I’m unable to edit the Sefaira output to normalise colour scale or show point values - this is one of the (many) limitations of the software (hence looking at HB as a viable solution). Sefaira don’t offer much verification as to simulation inputs either, however I do now note the directionality of DF contours on the East facade. This would surprise me if Sefaira were using a sunny sky, as the simulation is described as a Daylight Factor analysis.
Does anyone have a simple(ish) method to convert the average DF figures shown from decimal place to % values?
Finally - I’ve uploaded a section of the model in question showing window and room height/depth, following @mostapha’s request.
I havent seen the figures youre referring to but i expect you simply multiply the figures by 100.
If youre not familiar with daylight factor heres a useful link.
I’m trying to manage this same comparison. I’ve done both Honeybee and Sefaira daylighting analysis for floor plans, and have considered three variables:
- Graphical resolution
Of those variables, it appears that Honeybee is more specific in terms of input and perhaps accuracy, but at the expense of both time and graphical resolution. The graphics can be resolved in photoshop, but it just adds even more time. How much time did each of those take you? My professor said an entire floor plan would take 10 hours for a yearly analysis, compared to a 15 minute sefaira analysis of the same variables at a much higher resolution.
Is sefaira generally regarded as a cheap/informal software, laughed at by professionals? My professor insists on having everyone use Honeybee but so far nobody has produced the assignment because it takes 10 hours per level for a yearly average… What is the consensus on Sefaira?
The resolution (not optimal) that I see from the 1st image, is not dependent on HB, but depends on the Radiance parameters.
Find the right balance and I assure, that you will be able to have a good “smoothing” in the coloring.
You don’t need Photoshop.
Play around with the parameters, trying to keep good accuracy.
Right, but the question here is about a “yearly average” analysis for an entire floor plan, which my professor says would take 10 hours with a good computing power, compared to 15 minutes for Sefaira. If you have a life, you won’t increase the values as you suggest because it would take such a long amount of time.
Is Sefaira a good alternative or is it so inaccurate that it’s not worth using?
Below is the Sefaira version of my radiation, not including interior walls, that only took 15 minutes