Honeybee Intermezzo 6: Customized set of radiance parameters
Option A: Select values based on the Detail level
Given that the detail level that you select defines the radiance parameters for the simulation, you should choose it based on the complexity of your model. As a rule of thumb, we can say that:
0-Low level: Corresponds to a room with window without shading
1-Medium level: Can be used for the parametric analysis of the interior of a room with windows and simple shading.
2-High level: Can be used for the analysis of the interior of a deep room without shade or a normal room with complex shading.
In case you want to set an even higher level of detail (e.g. for a deep room with complex shading) or you want to further adjust the radiance parameters, you can hold a convergence test as described next.
Option B: Customize the values based on a convergence test
A ‘convergence test’ is always recommended to find the most suitable set of ambient parameters for a given scene (3D geometry and optical properties). The test consists of running multiple simulations and gradually increasing the ‘resolution’ of the ray-tracing process by changing one parameter at a time. The radiance ambient parameters that can be specified are:
-ab: ambient bounces (number of inter-reflections to take into account)
-aa: ambient accuracy (maximum error allowed in the ambient interpolation)
-ar: ambient resolution (density of sample points for detailed regions)
-ad: ambient divisions (number of rays in the sampling hemisphere)
-as: ambient super-samples (additional rays for highly varying regions)
In this example, we are going to hold a convergence test for the Daylight factor simulation. For the first iteration:
1. Create a Params » Input » Panel component and write the following values (pay attention to write it using exactly the format, along with the dashes in front!):
-ab 1 -aa 0.4 -ar 8 -ad 32 -as 16
2. Connect the panel to the ‘additional_par_’ input of the HB Radiance Parameter component.
3. Run the Daylight factor simulation.
4. Create a Math » Util » Average component and connect the ‘results’ output of the HB Daylight Factor component to the ‘input’ of the average.
5. Create a Params » Input » Panel component and connect it to the ‘Arithmetic mean’ output of the Average component.
6. Point down the value that is shown in the panel.
Now increase the number of ambient divisions (-ad) to the next value as indicated in the table. You can change it along with -ad, as the two parameters are related. Set -as to half the value of -ad, at most. Go through all the -ad and -as values while keeping the other parameters constant and take note of the average DF from each iteration. After a sufficient resolution, the outcome values are not changing significantly. Pick the lowest values that guarantee a stable result (±10%).
Now that you chose values for -ad and -as, keep them constant and do the same type of iterations by increasing ambient resolution (-ar), again until it stabilises. Again, take note of the resulting average DF values and pick the lowest resolution above which the DF values are not changing significantly.
Then decrease the ambient accuracy (-aa) in the same manner, until you find the parameter value that stabilizes simulation results.
Once you are satisfied with these parameters and the average DF value does not change much when launching multiple simulations, you can start increasing the number of ambient bounces (-ab) until you reach the convergence and your final set of ambient parameters, which guarantee a reliable simulation result.
Important to note:
a. Halving -aa or doubling -ar will result in computational times approximately 4x longer, while doubling -ad and -ab will result in computational times 2x longer.
b. By setting -ab = 0, you can simulate the contribution of direct sunlight alone, without any light from the sky or from inter-reflections.
