Radiation analysis horizontal grid

Hi everyone,

I tried to run some tests to calculate the solar radiation on a horizontal plane with different components (PIT, RAD and DC). I used San Francisco climate file. I found some weird results:

  1. As you can see in the figure attached for all the methods the radiation value is not uniform all over the plane.
  2. Using RAD component the value doesn’t “match” the global horizontal radiation value from the EPW.
  3. Different results between RAD e DC simulations. DC results are closer to the climate file.

Can someone help to figure out why these strange results occur?

Attached the GH file.


05_annual_radiation_gridbased.gh (423.8 KB)

solar rad

Hi @AndreaZani905,

Thank you for posting this. As you know getting the results with a slight difference from radiance is normal but in your case it seems more than a bit. I won’t be able to check this right now but I marked this to check next week.

In case of direct comparison between the values from weather file and the simulation keep in mind that the simulation values are for between the hours (e.g. 8:30, 9:30 and so on).

cc: @sarith

@mostapha I thought all the recipes were for illuminance only? If he is actually doing a full-fledged radiation study he might have flipped the -O flag for gendaymtx, right?

We have a recipe for radiation which generates the sky matrix with -O1:

gendaymtx -m 1 -r 0.0 -O1 -s -v sky\skymtx_sol_r1_2_725300_41.98_-87.92_0.0.wea

I wonder if the difference happens in sun matrix. I didn’t get a chance to address the -m in gendaylit.

Yes, it seems that happens with the calculation of the sun matrix.




I checked the file and I’m getting different results from yours.

  1. Radiance’s gendaylit overestimates the values.
  2. Radiation recipe is right on with the results.
  3. DC recipe for Radiation is noisy but the average results is pretty close to the value from the weather file.

Here is the file: radiation_study_comparison.gh (433.7 KB)

The difference between RAD recipe and DC recipes is that RAD is using a simpler approach to calculate the skies.

DC method generates the Tregenza total sky, Tregenza direct sky and sun matrix (analemma) and runs 3 calculations. The final result is total - direct + sun. In this approach we try to replace the inaccurate patched sky with correct sun matrix but at the same time the calculation under the patched sky is happening twice which adds the noise to the scene. Here is the results from total and direct sky side by side.

RAD recipe on the other hand generates Tregenza diffuse sky and sun matrix and adds them up. In this approach we run the analysis under patched sky once and so there is less noise in the calculation. If you’re looking to calculate annual radiation I suggest you to use the Radiation recipe component.

PS: Update both ladybug and honeybee from GitHub before testing the file.

Thanks Mostapha, with the updated components I get your same results.



probably this is an old topic but I find that is relevant to what I want to ask, so I just post here and revive it instead of creating a new one.

I was trying to understand a bit better the differences between the different analysis recipe in HB+ and when is better to use one instead of another. Mostapha explained well how the RAD and DC calculate the results and why in the DC there is more “noise” in the results (which are, however, very similar to each other if averaging them).

However, I still have doubts about when to chose between PIT, ANN, RAD, or DC. I get that ANN, RAD, and DC can provide annual results, but also hourly results can be easily retrieved using hoys input, potentially making the PIT useless.

I tested the different recipes by doing a sensitivity analysis of the radiation on a south-facing façade for one hour (16.00) on the 16th of August in an urban context and all the results are really similar to each other, which makes the choice for a specific one difficult to justify.

  • Can someone summarize for what is each of them more correct to be used?
  • And, which one can be considered the best to conduct radiation studies in urban contexts like the one I have done?

Best regards,