What engine does the lighting analysis use for daylight calculations? I saw that the plug in has been validated against radiance, Is that validation data available?
Thanks!
Thanks for asking.
We are working on a validation whit paper now and ope to have that available soon.
The engine is the one that is behind the Autodesk Rendering (Render In Cloud in Revit) cloud service, which is a proprietary engine based on the Multi-Dimensional LIghtcuts method initially developed at Cornell University. Here is a link that that paper:
Keeping on the topic of validation, I recently compared the results from the Revit Lighting Analysis plug-in to Radiance.
Room details: West Facing room (4.5m tall, 4.75m exterior wall, 4.25m room depth)
Two windows (70% T-vis, #1 0.8m x 1.845m @ 0.33m AFF; #2 1.4m x 1.35m @ 0.75m AFF)
75% ceiling reflectance, 55% wall reflectance, 25% floor reflectance
Simulation 1: 9am on Sept 21 under CIE clear sky condition
Radiance Results 1: a maximum of 790 Lux close to the window on the working surface. the 110 Lux threshold is 2.5m from the window.
Revit Lighting Analysis Results 1: a maximum of 730 Lux close to the window on the working surface. the 110 Lux threshold is 3.6m from the window.
While the maximum illuminance values agree, the daylight penetration varies by over a meter.
Simulation 2: 3pm on Sept 21 under CIE clear sky condition
Radiance Results 2: a maximum of 2350 Lux close to the window on the working surface. the 110 Lux threshold is achieved throughout the room.
Revit Lighting Analysis Results 2: a maximum of 5200 Lux close to the window on the working surface. 110 Lux is achieved throughout the room.
In the case of direct daylight penetration through the window, the two model results vary greatly.
I followed the help guidelines for the revit lighting analysis and set up the radiance model to match. I am fairly comfortable with Radiance modeling and was hoping for some guidance/advice/thoughts on why the results vary so greatly particularly for the 3pm/direct daylight case. Are there any tricks or tips using the lighting analysis plugin?
thank you
Sam
Sam.
Thanks for doing the comparison. All of these comparisons done by different modelers really help us refine the workflow so we get consistent and valid results for everyone. We are very confident in the engine used by LAR, so variations in the results are due to a couple things:
1. User settings - Radiance is a bit of an art of simplification, so we see different styles of modeling creating varying results even if all the results would be considered valid. The LAR algorithm by definition takes care of most simplification settings, so the results are pretty consistent.
2. Material settings in LAR are incorrect - The materials editor in LAR is still a bit convoluted, so it's easy to make mistakes.
3. Running material types that are not supported or are supported differntly by LAR - Currently we don't model diffusing translucent materials precisely, so Radiance will be more accurate if you use a BSDF. On the other hand, LAR engine can handle detailed geometry like blinds directly, so the results for that will be more physically accurate at the detail level than a Radiance BSDF.
4. Window simplification in Radiance - Radiance uses a mono-planar element to model a virtual glass plane, while LAR engine models the solid geometry directly, resulting in slightly different response for higher incident angles and more diffusion at the edges.
The best way to go about it is probably for me to look at your Revit model to see if there are any settings I think would make a difference. Can you share the model confidentially with us? You can send the model or a link to download to LAR.Feedback@autodesk.com.
David
Thanks for the thorough reply David. I emailed you at the LAR helpdesk and will do a follow up post after we chat off line.
Regards,
Sam
Hi Sam,
Good and interesting post!
Did you already get any feedback.
I did some first test but I'm not sure if I did everything good?
I'm hoping that Autodesk has a document about how to make a realistic daylighting analysis. The help document is good, but it's very theoritical.
I'm looking for a good example.
Here is a little more info that might help you.
First of all, workflows and proper modeling.