[Radiance-general] CFS with Radiance

Thu Jul 9 11:56:11 PDT 2015

I must say that I agree with Rob, however, a short answer that might help.

For evaluating illuminance levels in a space with different CFS you,
strictly speaking, DO NOT HAVE TO do anything. You may choose between
different methods that have some pros and cons.

*ray-tracing*

   - Requires ray-tracing for each time-step and each system, which is slow
   (*four systems simulated annualy in an hourly basis --> 4*8760 = 35,040
   simulations*.)
   - As accurate as it gets, if options are defined correctly.

*2 phase method*

   - Do not need the BSDF
   - Requires ray-tracing for each CFS (*4 systems --> 4 ray-tracing
   simulations*)
   - After ray-tracing, annual simulation is fast.

*3-phase method:*

   - Requires BSDF data in KLEMS basis, which may be slow to compute,
   unless it can be exported from WINDOW, for example, or such data has
   already been calculated (the idea is to make a database, I think).
   - Does not really work well for specular systems (Klems patches are too
   big)
   - Requires 2 ray-tracing runs, always (*4 systems --> 2 ray-tracing
   simulations*)
   - After ray-tracing and BSDF calculation, annual simulation is fast

*5-phase method:*

   - Requires BSDF data in KLEMS  basis AND/OR Tensor tree format, which
   may be slow to compute, unless it can be exported from WINDOW, for example,
   or such data has already been calculated (the idea is to make a database, I
   think).
   - Works well for specular systems
   - Requires 4 ray-tracing runs + 1 for each system, always (*4 systems
   --> 4+1 = 5 ray-tracing simulations*)
   - Hard to code...?
   - After ray-tracing and BSDF calculation, annual simulation is fast

I would not say there is a recipe. If you have to choose between 1 or 2
CFSs, maybe it is faster to just draw them and use the 2 phase method (or
ray-tracing if an annual simulaton is not required). On the contrary, if
you are going to test 10 different CFSs, 3 and 5 phases may make sense,
since you reduce the expensive ray-tracing calculations... However, this
will also depend on weather you have the BSDF data AND/OR if it makes sense
to calculate it and store it AND/OR if you intend to simulate a dynamically
controlled CFS.

Best!

2015-07-09 14:29 GMT-03:00 Guglielmetti, Robert <
Robert.Guglielmetti at nrel.gov>:

> Some very quick replies and clarifications within:
>
> On 7/9/15, 10:26 AM, "Ikrima Amaireh" <ezxia at nottingham.ac.uk> wrote:
>
> >Hi G. Larrain,
> >
> >Thanks for your detailed clarification (below).
> >
> >I wonder if you kindly can help me toward better understanding and
> >answering the following:
> >
> >We agreed that for daylight modelling of spaces with CFS, Radiance is a
> >suitable tool as it applies ray-tracing method. And for annual and/or
> >climate-based daylight calculations, as calculation time becomes a
> >critical factor, using BSDF data for CFS is highly recommended to tackle
> >the task with radiance (using phases methods). However:
>
>
> BSDF data in an annual simulation context is generally limited to Klems
> basis BSDF data, which may not be high enough resolution for some CFS. And
> the 5-phase method, which can circumvent this, is not necessarily "quick".
> This is all still very much a quandary and the newest daylight metrics
> have added confusion to all of this, IMO.
>
>
> >
> >- if someone needs to perform daylight calculations for given space with
> >different CFS (different cases for comparison purposes) to get
> >illumination levels for horizontal grid points (and not pictures nor
> >scenes), is Radiance still the most convenient option?
>
>
> Absolutely. The lack of an image-as-output requirement does not change the
> fact that ray tracing is a good/convenient algorithm option for daylight
> simulation problems, especially when dealing with diffusing media and CFS
> in general. BSDFs allow you to do lots of "what-ifs", relatively quickly,
> but are beholden to the limitations of the resolution of the BSDF.
>
>
> >- is it still needed to use any of the phase methods (2, 3 or 5)?
>
>
> Is what still needed? BSDF data? Ray tracing? I don't know of other
> lighting simulation tools that can employ the multiphase methods. I would
> say BSDF data is optional for 2-phase, required for 3-phase, and optional
> for 5-phase.
>
> >- most importantly, is it still needed to get BSDF data for CFS or just
> >can model the detailed CFS in Radiance (and, if yes, how?)
>
>
> Again I'm confused as to what "it" is. A couple of approaches are
> available to you in general. If you have a geometric model of the CFS you
> can use Radiance (genBSDF) to make a BSDF of the CFS and use that in a 3-
> or 5-phase context. However in an annual/climate-based simulation context,
> you will be stuck with a Klems basis BSDF and that will not be very good
> resolution for a lot of CFS; here you may want to use the 5-phase method
> and stick the actual CFS geometry in the building model.
>
> Some things to consider here are the photon map, now a part of Radiance
> proper, or using the 2-phase method where appropriate. By 2-phase method I
> mean generating a daylight matrix for your calculation points (or view),
> and modeling the window material as-is. If the "CFS" is a shade cloth, you
> can approximate that with a Radiance "trans". Same for translucent panels.
> With this single daylight matrix, you can throw a vector of sky matrices
> at it and get an annual climate-based result very quickly. Problem is, we
> want to do stuff to the windows, sometimes at the time step level, so:
>
> Blinds and compact daylight redirection devices (e.g. Lightlouver) are
> best represented as BSDF, and if most of the redirected flux is headed up
> and away from the points of interest, a Klems basis BSDF is good enough
> IMO. In these cases you could use the 3-phase method. Problem is when you
> have a BSDF for blinds, and you also want to simulate the blinds-up
> condition (i.e. clear, specularly transmitting glass). Using a Klems basis
> BSDF for this is sub-optimal. You end up needing to do two annual
> simulations, one as 2-phase for the clear glass scenario, and again as a
> 3-phase (with a blinds BSDF for the transmission matrix) for the blinds
> down condition. If you have a lot of different window groups, your
> simulation space can get large in a hurry. But it's still do-able and you
> can get results that tell a story you simply couldn't tell 5 years ago,
> informing the newest daylight metrics as well.
>
>
>
>
> _______________________________________________
> Radiance-general mailing list
> Radiance-general at radiance-online.org
> http://www.radiance-online.org/mailman/listinfo/radiance-general
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.radiance-online.org/pipermail/radiance-general/attachments/20150709/cfbdd4da/attachment-0001.html>