I cannot get my results to agree with reality

Hi, this is my first time posting so please bear with me.

I’m an architect and have only recently begun analyzing daylight through Ladybug, however I find that the results does not align with my real life perception of the space (upon visiting).

My results show an sDA of 15%, but the room appears bright in real life. See attached photo.

I initially exported my model from SketchUp, but have remodeled it entirely in Rhino to eliminate any import issues. The Grasshopper file has been shared with me through a course in daylight analysis and I have made very few adjustments.

Is there a way I can share my files in order for them to be reviewed by one of you experts?

Kind regards

1 Like

I can now attach multiple photos, so here are my results.

Hoping one of you can help me.

DA is calculated for the total occupied hours. You can not extrapolate from one image (hour) to the whole calculated period.
You can try to simulate the space for the specific date you took the picture, but to rely better on the simulations i recommend you to calibrate results with measurements on site.

-A.

Hi Abraham,

Thank you for your reply.

I realise that DA is an annual calculation and that it includes sun direction aswell, contrary to Daylight Factor for instance. However the photo was taken on a cloudy day so in my mind it would be considered an “average setting”, could be that is where I am mistaken?

Maybe the results are indeed accurate and I am just taken aback by the results. The room is large with a good height of 3.6m and the windows are large (height of 2.1m//width of 1.4m). They are spaced out with roughly 3.3m between them.

But anyway, thanks for your reply once again.

A space can still feel daylit without achieving a high DA, and I would expect that to be the case here. DA is looking for an ample amount of light (300 lux) on a horizontal work surface, meaning light coming from above plays a large role. This is a wide space (compared to the window height) with a dark ceiling, meaning it is mostly side-lit with very little diffusion of light from above, so low DA levels are not unexpected. If you run the model with a white ceiling, does that change the sDA significantly?

DA is often equated with whether a space will “feel” daylit, but that is not the case. Daylight autonomy is whether there is so much light that occupants can work without additional lighting, meaning artificial lighting can be reduced and energy saved. A space with a high DA is likely to also feel daylit, but low DA does not imply a lack of daylight access at an aesthetic/experiential level, just that there is not enough for all occupants to work without some supplemental artificial lighting. Your photo shows a very good example of where experiential and working-level daylight access can be quite different.

2 Likes

Agreed - here if you changed the threshold for DA from 300 lux to say 200 lux you might find the space performs much better and there’s just a significant period of time in the 200-300 lux range instead of above 300

I appreciate your thorough reply. I had not considered that an sDA below 300 lux could still feel well lit experientially, but it makes a lot of sense.

Thank you.

1 Like

Thank you for your reply.

I tried to reduce the threshold to 200 lux, but it only increased the sDA by approximately 10% (to an sDA of 26%) so I am still quite far away from the goal of an sDA of minimum 50%. I guess that it just might not be achievable with the current geometry/apertures.

1 Like

Interesting study, your account might be getting trusted enough to share files, you can see how that is determined here

To help understand the results more you could look at specific hours and see what the lux level is. You can visualise a single hour using this component

An example script here for you to plug your model into.
HourlyDaylightResult.gh (39.6 KB)

And then even do a point in time image based analysis to for a test hour to see how it compares to your image photo :slight_smile:

That would be interesting to try out. Thank you.

Of course there are natural limitations to how ‘correct’ the simulation can run in terms of accurately representing reality. Things like the detail level of my model etc will naturally impact that. But it would be interesting to try and simulate the sDA at exactly the point in time of my photo.

Hello everyone,

This is a very interesting topic, glad to see so many people jumping in. The theory has already been discussed in great detail in the posts before mine, it’s great input, but I have trouble connecting those explanations with the results I see in your grid. I wouldn’t be surprised to learn that there is a problem with your calculation model, I think the drop is way too dramatic from the 100% really next to the window to lower than 50% less than 1 meter away, a gradient is missing there.
It’s difficult to point for a problem without seeing your model, but I’ll give it a shot from what I learned from my past mistakes:

  • Make sure your detail level is set to a minimum of Medium
  • Aren’t you, by any chance, using a double glazing window, and having two layers of glazing?
  • Make sure your model units is correct. Are you using milimeters in rhino and in grasshopper?

Let us know your findings.

edit: I also have trouble believing such a space only reaches a sDA of 12%, that’s really low.

1 Like

For comparison, here is what I find for similar geometry (using ClimateStudio, sorry!). I do not know the location or orientation, so I took a south-facing facade in Seattle. I can approximately reproduce your results if I am including dynamic shading – are you including that? From top to bottom:

  • A dark ceiling (10% reflective), without operable shades. An sDA of 41%.
  • A dark ceiling, with operable shades. An sDA of 8%.
  • A light ceiling (70% reflective), without operable shades. An sDA of 73%.
  • A light ceiling, with operable shades. An sDA of 9%.




1 Like

Nice input, savage, great stuff! I find your results way more believable - an sDA somewhere between 41% to 73% depending on the reflectance factors.

It’s rather difficult to model dynamic shading in HBT (last time I checked, it was actually impossible), so I’d be really surprised that OP, as a beginner, managed to take that into account :slight_smile:

In my opinion, this confirms that it’s a technical issue with your model @PTR

Thanks for chiming in.

I must admit that I too am still not convinced that the results are not caused by some flaw in my model or my definition in Grasshopper.

I recently simulated sDA on another project with the same GH definition though and the results are much more credible.

• I’ve only been simulating at max detail level (2).
• I’ve modelled the windows inside Rhino as a singular closed polysurface.
• My model units is in meters.

Thanks for spending time on this. I really appreciate it.

My location is Denmark and the facade is facing north-east - which of course also contributes to the low score.

I am not including any form of shading apart from whatever the building geometry provides.

I am now able to upload my files here. I would really appreciate if any of you seasoned daylight experts could take a look.

You’ll have to look past some of the panel notes, those are for my own use and are in danish. I hope the definition makes sense to you. Please do not hesitate to ask questions if not.

Uploading: 3D model_Troubleshoot.3dm…
GH Definition_Troubleshoot.gh (544.4 KB)

@PTR, it seems that your model didn’t upload correctly…

Yeah, unfortunately it exceeds the limit of 1.5 mb.

I’ve uploaded both files to WeTransfer, hope that’s ok as far as filesharing goes.

Hey @PTR,
This is a very interesting discussion and I was eager to jump in as well!

I checked your model and it looks fine to me. The only things that I would suggest you to change are the reflectance of the ceiling, floor and wall.
I see that now you have 0.7 for the ceiling while from the picture is visible that is made of dark wood so I might consider switching the reflectance to something lower (0.2/0.3).
Similarly, you currently have a 0.2 value for the floor that I think can be increased to 0.5 while the white walls can be boost up to 0.7 from the current 0.5!
I think this might help to get results closer to reality (if this is your current objective and you are not trying to simulate a future refurbishment scenario instead).

I’ve also tried to run a quick Point-in-Time view-based analysis to see if the results agree with reality. I’ve used the 21st of Sept. at 12.00 with the Copenhagen EPW (is an hour with high sky coverage so it should reflect the condition under which your picture was taken). Here’s a picture of it, the light environment doesn’t look to different from the one you have after all and that should confirm that there is nothing wrong with your model!

The sDA results, despite low, might be right for me, especially considering the location, orientation and small glazed area you are simulating.

1 Like

Hi @PTR,

I had a look at your model, and I found out why the results are so low. There are some important details that are only visible opening the model and looking around, mainly the little amount of glazing present in the facade. In your model, there are 12 openings of 3.08 m² each in a wall of 58 x 4.1 meters. This is a WWR ratio of 16%, which is a rather small amount. However, it gets even smaller in reality, as from this 3.08 m² of opening, only 1.8 m² is glass (40% is frame). In total, the ratio glazing to wall is 9%, that’s a really small ratio, I would have never guessed that looking at your indoor picture!

I made a quick comparison of your window VS a frameless window to see if my gut feeling needs some calibrating. I replaced half of the windows, and left the second half as original:


The sDA is now 31%. If I’d have replaced the entire facade then it’d be close to 50%, it seems that everything is correct after all!

1 Like