![]() |
Simulation of Daylight Redirection Systems | ||
Raydirect Plug-In for Rayfront | |||
Caveats and PitfallsVisibility Checks Only from Center of PaneRaydirect scans the outer hemisphere for determining the irradiance caused by diffuse reflections and the sky, and it also checks how much light actually arrives from the sun, as filtered by window panes and other obstruction. However, those scans and checks are only performed for a single point on each window, its centerpoint. This means that there will be an error in the results when a window is partially obstructed. Either the center is unobstructed, then the result will be as if the window was fully exposed, or the center is obstructed, then the result will simulate a fully obstructed window. If this is a problem for a specific simulation, then you'll have to subdivide the affected polygon into several pieces that then get processed independently. A future version of Raydirect will include an automatic control mechanism to determine how much of the surface of each window receives how much light, then averaging the results. Unfortunately, doing this check correctly for all situations is non-trivial and relatively expensive in terms of the computational effort required.
Possible irregular results for measuring fieldsThis is not really a problem of Raydirect, but a general issue about measuring fields. It can be tempting to run numerical simulations with lower quality simulation settings than one would do for a visualization. The problem with this approach is in the uneven results it will generate. In a visualization, uneven results give the observer an immediate and unmistakeable clue about the low reliability of the picture. With numerical output, the problems may not be as easily recognizable.The situation is further aggravated by the fact that the sequence of the points calculated may strongly influence the light distribution in a low quality simulation. This means that the results may depend on the orientation of a measuring field. The solution is of course to run the simulation with higher quality settings. One of the best ways to verify your settings is by generating a visualization as well. If the picture doesn't show any disturbing artefacts, then the numerical simulations with the same settings will be ok too. Sampling errors measuring direct irradianceIf any sensor positions are located in a place where the system sends visible amounts of light redirected from the direct sun, then the jittered sampling of the sources will cause erratic results, which may be a lot higher than expected. While the speckles on room surfaces in a visualization can be reasonably interpolated by supersampling and filtering the picture, there is no such mechanism available for light sensors. Therefore Raydirect can't be used to measure irradiance/illuminance values that are caused by the solar component arriving at a specific point directly from the redirection system. The software is only capable to determine the distribution of light in a complete space, and for measuring irradiance/illuminance values that are reflected off of other surfaces such as the ceiling or the walls. | ||||
|
|
|
|
| |||
|
http://www.schorsch.com/raydirect/manual/caveats.html Copyright © 1998-2003 Georg Mischler. All rights reserved. |