3-phase/5-phase method: is a high -ad value for vmtx_par or dmtx_par?

Hi there :slight_smile:

I have been working on comparing the 5-phase method with the 3-phase method by using a shoebox room with a south window. After I studied the Three-phase and Five-phase tutorials by Andy McNeil, one thing has made me confused.

[1] The Three-Phase method for simulating complex Fenestration with Radiance. A McNeil, LBNL. Last revision: 2014.
[2] The Five-Phase Method for Simulating Complex Fenestration with Radiance. A McNeil, LBNL. 2013


According to a previous discussion with @Sarith, -ad needs to be set a high value. So previously I set -ad 22400 to dmtx_opt_par_. The calculation for this case took 3-5 minutes.
However, in the tutorials, it is addressed that the view matrix usually has high -ad values, which are 50000 and 65536 in both papers. Meanwhile, for daylight matrix, a typical -ad value is 1024 according to [2].

So I am confused about:

  • Is dmtx_opt_par meant to be Daylight Matrix parameter and vmtx_opt_par meant to be View Matrix parameter in Honeybee Plus?
    If so, should we set a high value -ad for dmtx_opt_par and a low value for vmtx_opt_par in each run?

Actually I tested it, and the output of vmtx_par is null. Moreover, it took 25 minutes to finish the calculation with the same configuration and location. The outputs (average UDI and DA) don’t have a big difference with the previous calculation.

I’d like to learn more about how to set the acceptable values for vmtx and dmtx in HBPlus, any suggestions will be appreciated :smiley:

I leave the hard parts for @sarith - :slight_smile:

Did you change the recipe type? I feel we should break down this component in at least two separate components. One for multi-phase and another one for the normal raytracing studies.

You may enjoy taking a look at this page. Room two is probably more helpful to understand the effect of the parameters. The change in the x axis is -ad and the y axis is -ab. You can change the -c parameter from the drop down.

1 Like

Thanks @mostapha

Did you change the recipe type?

Thanks for pointing it out :grinning: I forgot to change the recipe type to 3, and that’s why vmtx output was null :sweat_smile:
It gets fixed and shows as below:
I will remember to keep a number list to remind myself lol

Thanks for sharing!

In the previous case, it has:
vmdx: -ab 5 –ad 22400 –lw 0.00005
dmtx: -ab 2 –ad 1024 –c 1000
which took 15 mintues to run,

Then I set:
vmtx: default (-ab 3 -ad 1000)
dmtx_opt_par: -ab 5 –ad 22400 –lw 0.00005
it took 10 minutes to run.

Since -ad and -ab are involved in both view matrix and daylight matrix calculation, it seems that increasing -ad and -ab values for view matrix can greatly increase the runtime?
The test room looks more like room 1, without furniture and context. I will try to display renderings.

Hi @catsquito,

The whole thing with regards to ambient parameters really has to do with how complicated your scene is. If your external scene is complicated, then you’d require high values of ab and ad for dmtx too. In the same way, if your room geometry isn’t too complicated, then setting high values for those parameters in vmtx is unnecessary.

As far as the parameters being set for the HB[+] simulation are concerned, I’d rather have a look at the .bat file in the project folder than the outputs from gh canvas.


1 Like

Hi @sarith ,
Thank you for the clarification!
So for the less complicated room geometry, daylight matrix plays the main role in the calculation, so we only need to set a high -ad value for dmtx when running HB+. In the other way, for a more complicated room geometry, view matrix would participate more than it does in a simple scene, so we need to adjust -ad values in both vmtx and dmtx. Did I understand it correctly?
I am also curious about parameters and the annual results in 5-phase method. Theoretically, does the -ad value connecting to vmtx or dmtx differently have an impact on annual results (average UDI, DA, sDA)?
I am asking this because for the same geometry configuration and material, sDA and DA results from Annual Daylight Recipe are much smaller than the results from 5-phase recipe, UDI results are quite similar though, while ASE is always zero.
I guess radiance parameters don’t have too much to do with the annual results in different recipes, and it is more likely due to materials, but I am not very sure…

The calculation, and impact on accuracy of, vmtx and dmtx are independent of each other. So, you need to increase/decrease ambient parameters for them based on the complexity of their geometry. I don’t think it is appropriate to assume that one matrix is more important than the other based on complexity of geometry in a specific context.

Yes to all metrics. DA and sDA are measures of daylight sufficiency. UDI also accounts for excess daylight, but this is supposed to be on account of direct sun than diffused daylight. So if the ambient values in the calculation are lower than required, there is a likelihood that those metrics will be affected.

As I mentioned in my previous post, I’d rather look at the actual commands in the .bat file, than what the recipes are doing. ASE is likely zero because in the direct-sun part of your calculation, the entire scene might be getting blacked out.

The radiance programs used for the annual calculation in HB[+] don’t employ ambient caching algorithms and rely on Pure Monte Carlo calculations instead. So, several paraters like -aa, -ar, -as etc. have no relevance.


1 Like