Radiance daylighting calculations are controlled using a set of options on the Daylighting calculation options dialog which is displayed before the calculations are started. It is important to make a selection here that is appropriate to the building you are analysing.
Select the report type (if any) that you will require after the calculations have finished. Options are:
If you do not require any particular report then select either 1-Map or 2-Grid.
Select a detail template from the drop list to load a set of ambient parameters to the dialog as shown in the table below.
You do not necessarily need to have an in-depth understanding of the meaning of each of these parameters to carry out an accurate assessment in DesignBuilder. However knowing a little bit about the way the calculations work and how the options affect tje calculations will help you to understand how to get the best trade-off between accuracy and simulation times.
It is important to realise from the outset that Radiance works using a statistical Monte Carlo approach and this means that you may (will) not get exactly the same results if you repeat a calculation using exactly the same model and options. You can reduce this effect by using higher detail settings but you cannot completely eliminate it. This is discussed in more detail under Advanced options below.
| Detail template | |||||
|---|---|---|---|---|---|
| Radiance code | -ab | -aa | -ar | -ad | -as |
|
1-Fast Typically used for initial calculations |
1 | 0.3 | 128 | 256 | 128 |
|
2-Standard |
2 | 0.25 | 256 | 512 | 256 |
|
3-Good (no interpolation) |
2 | 0.00 | 512 | 1024 | 512 |
|
4-Good |
2 | 0.22 | 512 | 1024 | 512 |
|
5-Accurate |
2 | 0.20 | 512 | 2048 | 1024 |
|
6-High quality Typically used for final results |
3 | 0.18 | 1024 | 4096 | 2048 |
Enter the height of the working plane above floor level for each zone in the daylight simulation (in m or ft). The default value will depend on the Report type selected. A typical value will be in the range 0.7-0.8m.
The working plane height is normally set at the average height of the top surface of the desks above the floor.
Enter the margin (in m or ft) around the zone boundaries where illuminance data is not to be calculated or included in summary results. This option can be used to help avoid inclusion of potentially misleading illuminance data close to walls and windows. A typical margin recommended by CIBSE is 0.5m.
Select the sky model to be used for the daylighting calculations. Select from:
Sky luminance profiles for 3 sky models (Source Rendering with Radiance: A Practical Tool for Global Illumination)
The main grid size that can be used to divide up the working plane when calculating illuminance. Larger values will speed up calculations but give lower resolution in daylighting outputs.
You should consider the size and number of zones in the building when selecting the grid size. A very fine grid will of course cause Radiance calculations to take a lot longer to complete. On the other hand a very coarse grid in a small zone may not give a good enough distribution.
You should use a maximum grid size no greater than 5ft or 1.5m for LEED EQ8.1 calculations.
This is the smallest grid size that can be used to fill in gaps in the working plane in cases where the maximum grid size is too large, especially at the edges of the plane. Use smaller values (e.g. 0.01m) to fill in most gaps between the main grid cells and the borders of the working plane for a more continuous output. It is common to use a Minimum grid size that is a factor of between 1 and 0.2 times the Maximum grid size.
To maintain a uniform grid across the working plane set the Minimum grid size the same value as the Maximum grid size.
The number of ambient bounces is the maximum number of diffuse bounces computed by the indirect calculation. The number of ambient bounces that Radiance should apply varies depending on the type of building and daylighting system you are analysing. It can be set based on the number of reflections typically required by the light to reach the task plus one or two extra for inter reflection within the space.
When the number of ambient bounces is set to 0 the ambient lighting calculations are switched off, so only direct sun/sky light patches are considered.
Tip: It is possible to calculate the fraction of the working plane that has a view of the sky by setting the Ambient bounces calculation option to 0 and the Illuminance threshold display option to 0, then viewing the Floor Area above Threshold data in the Grid output. This output is required by a number of codes, including "Code for Sustainable Homes".
If the number of ambient bounces is set to 1, light is considered to reach the interior surfaces from the sun's direct rays, from the diffuse sky, and from first bounce reflections of direct sunlight rays from both interior and exterior surfaces. Reflections of sky light off interior or exterior surfaces to other interior surfaces will not be considered. Additional bounces can be added to consider additional flux paths.
Note that a value of 1 for the number of ambient bounces is often enough for daylight factor calculations. However, if the daylight factor plot is "lumpy", setting a higher value alone will not fix it.
Doubling ambient bounces can double rendering time
Ambient accuracy (-aa) is the maximum error (expressed as a fraction) permitted in the indirect irradiance interpolation.
You should normally use a value between 1 and 0.1, with lower values giving the best accuracy. A value of zero gives no interpolation.
Halving Ambient accuracy approximately quadruples rendering time.
The Ambient resolution sets the distance between ambient calculations by determining the maximum density of ambient values used in interpolation. Factors that influence the scale over which interpolation may occur are:
The minimum possible spacing between hemispherical indirect irradiance sampling points is the maximum scene dimension multiplied by the ambient accuracy divided by the ambient resolution. In other words, for distances less than this minimum, the calculation will always resort to interpolation, rather than initiate more sampling, regardless of the error estimate associated with that interpolation. This prevents the calculation from spending unnecessary time resolving irradiance gradients over negligible scales. This distance gives the scale at which the interpolation accuracy begins to deteriorate from the ambient accuracy setting. The minimum separation for calculated irradiances, Smin, is:
Smin= Dmax × Ambient accuracy / Ambient resolution
Where the scene dimension, Dmax, is a measure of the maximum dimension of the zone being treated.
The effect of increasing Ambient resolution depends on the scene but it can quadruple calculation times for double the value.
Ambient divisions sets the number of initial sampling rays sent from each ambient point into the hemisphere to determine the indirect incident light. The error in the Monte Carlo calculation of indirect illuminance will be inversely proportional to the square root of this number. A value of zero implies no indirect calculation.
The Ambient divisions and super-samples parameters can be used to help reduce "noise" in a calculation. By setting these options higher more rays will be tested when calculating an ambient value for a point.
High values of this parameter (4096 is perhaps the highest value that would be used under normal circumstances) will minimise "patchiness" of daylighting outputs but slow calculations. Doubling Ambient divisions will approximately quadruple calculation time.
The number of extra rays that will be used to sample areas in the divided hemisphere that appear to have high variance. Ambient super sampling should usually be set to about one half or one quarter of the Ambient divisions parameter. Super-samples are applied only to the ambient divisions which show a significant change.
The effect of increasing Ambient super-samples is to reduce "patchiness" in regions where indirect illuminance is changing rapidly, but adding to the Ambient divisions and calculation times.
Hemispherical sampling at the first level will always be initiated from the first point supplied to rtrace. From these hemispherical sampling rays, the ambient calculation will predict the way the indirect irradiance is changing about that point (the indirect irradiance gradient). The calculation also evaluates an estimation of error associated with the prediction for the irradiance gradient. These quantities, together with the Ambient accuracy parameter, are used to determine a “radius of validity” for the gradient estimate. If the next point supplied to rtrace is within this radius, the indirect irradiance is evaluated from the gradient estimate and not from further hemispherical sampling. In other words, the value is obtained by a form of interpolation rather than by actual sampling.
Irradiance interpolation can occur across the points supplied to rtrace, so hemispherical sampling at the first level will not necessarily be initiated from every point in the working plane supplied to rtrace.
It is important to appreciate the element of chance at work whenever hemispherical sampling is used. If the number of initial sampling rays were set too small, the calculation might, for example, miss a bright patch even though it was visible from the point at which the rays were spawned. Likewise, an unrepresentative chance “hit” of a small bright patch by one of the sampling rays can produce a gross overestimate for indirect irradiance. In a rendering, the artefacts associated with ambient undersampling can cause bright and dark blotches.
To minimise blotches we need to set a sufficiently high value for the number of initial sampling rays, Ambient divisions. Hemispherical sampling is generally too expensive to initiate at every surface visible from the eye point. The calculation needs good indirect irradiance estimates from sampling at a limited number of locations. We then rely on the irradiance interpolation algorithm to estimate missing values. To generate a fairly smooth rendering for a daylight calculation accounting for the first level of inter-reflection, we would need to set moderately high resolution values for the ambient parameters.