Hi, I’m having a strange problem where changing the VLT of the assigned glazing from 0.66 to 0.67 results in drastic changes in sDA results. The results increase from 11.5% with VLT of 0.66 to 89% with VLT of 0.67. An identical simulation in IES-VE yields a result of over 90% with a VLT of 0.66 using identical Radiance parameters and thresholds. I’ve uploaded a simplified script to Onedrive (I’m a new user so wasn’t able to upload here): https://1drv.ms/u/s!AnQyzy4_hhlYhsUCW1d2qHwDvJ37Uw
Hi @SH ,
This is why threshold-based metrics like sDA aren’t always ideal since you’ll sometimes experience a step function when you “cross the threshold.” Still, the big reason why you have such a big “step” in your case is that you are using really low-quality Radiance parameters that are causing the results to be unreliable.
I suggest just using the default radiance parameters for the Annual Daylight recipe and you will get much more expected results.
Hi Chris, thanks for taking the time to look through this and provide feedback. I’ll try running with higher quality Radiance parameters.
Is there a way to run an annual daylight simulation using a Perez sky? I believe IES-VE uses this which is giving much more favourable results for sDA analyses (this sky model is allowed under the certification scheme I am using)
I agree with Chris. But now that we talk about parameters for the Annual Daylight recipe, one thing that you should be careful with is the -lw parameter, which should be 1/ad or lower. This is taken care of if you use the default parameters - but remember to set the type to 2 = annual (rfluxmtx).
Hi @SH ,
The annual daylight recipe in the LBT plugin uses a Perez sky with Tregeza discretization (145 patches) in order to model the indirect contribution to the hourly illuminance. Importantly, the direct contribution is modeled using an “improved DC” method, in which rays are traced directly from the sensors to the disc of the sun. So, whereas older tools like DAYSIM struggle to model the direct contribution of sun on an hourly basis, the LBT recipes will be much closer to the values you get from a point-in-time simulation (the middle option below):
Thank you both for your input. To clarify, should I be using recipe type 2 rfluxmtx for all annual simulations (eg. sDA and ASE)?
If you mean the
recipe_type_ of the HB Radiance Parameter component, then yes, you should use it for all of the annual simulations with the LBT plugin.