Simulation methods for practitioners

radiance-general
1

Hello all,

I wanted to ask you, people, a question that has something to do with simulation methods, specifically focused on practitioners. Several of my (Groundhog) users are students and people who just want results.

Context

As a few of you may know, I am developing EMP, which intends to become sort of a “higher level Radiance” that not only does simulation, but also pre and post-processing. The point is to allow a user to script on a higher level. For example, to say “hey, emp, calculate the UDI on this workplane” instead of saying “hey Radiance, grab this bunch of sensors (which I need to figure out), calculate the daylight matrices, multiply by a gendaymtx with this weather, etc.”

Question

In the previous context, I want to use a method that is not too sensitive to user errors. For example, multi-phase methods have the problem caused by averaging partially shaded windows, which is why I tend to not use them.

In this regard, I was reading THIS article by E. Brembilla, and noticed that for global illumination calculations, the two-phase method (i.e. without sharp suns) works well. For direct solar illuminance, on the contrary, we need better methods.

I wanted to know what you think about the following methods for different metrics

Proposition

  1. UDI, DA, etc.: 2-phase method with 145 Tregenza patches (seem to be fast and does not sacrifice too much accuracy for annual results)
  2. ASE: Direct sunlight calculation (i.e. black world with MF6 source suns, like the 5-phase method)
  3. Photosensor (i.e. not workplane) calculation: this intends to calculate sub-hourly illuminance. I want to use Standard Daylight Coefficients (i.e. Tregenza 2-phase method plus MF6 direct sharp suns)
  4. UDI, DA. etc. in models WITH BSDFs: 4-component method (this is not yet supported)

Just wanted to know what you think about this. Would this combination be “good enough”/convenient for practitioners and students. I want to encourage the use of PROPER daylighting analysis on design stages, so accuracy is a concern.

I am planning to interpolate weather to 20-minute intervals, by the way. Is that enough?

Your opinion would be really appreciated! Thanks very much.