Comparison of Measurements and Modelling

Modelling

For the modelling of the lamella and roller type blinds in combination with glazing two different tools were used. The first one is the well-known European Window Information System WIS (version 2.0b) which has been developed further within the European WINDAT project (see http://www. windat. org). The algorithm used for the blinds is the standard algorithm using a radiosity method, splitting up the individual blinds in 10 flat sections reflecting completely diffuse (5 on the upper part and 5 on the lower part). Mirror — type blinds cannot be modelled which such an algorithm). A second simplification within WIS is the treatment of the slats as completely flat and without extension. Thus lamellae like the ones in Figure 1 pose a problem for this algorithm when radiation is passing nearly parallel through the slats. The model underestimates the possibility to hit a lamella and overestimates the transmission. WIS does estimated the convective heat flow through
devices based on a plug-flow model using the temperatures of the layers. This model is described in the ISO/FDIS 15099 standard.

Because of the optical simplification a second simple radiosity model (using only one section of the slat — one facing upward and one downward) based on view factors has been programmed. The approach is similar to the one documented in prEN13363-2 [ 2], however with two important extensions: firstly the lamella might be curved with a radius given, and secondly a direct transmission part is taken into account. Using the solar transmittance and reflectance calculated with such an algorithm, the total solar energy transmittance of the shading device in combination with a glazing (either inside or outside) is calculated using a simple resistance model. Convective and radiative surface coefficients from the glazing surface to the blinds, and through the blinds to the environment are estimated based on the "openness” of the blinds. The convective part is always constant. This model is called the “ISE model” in this paper but should not be confused with another more refined inhouse model utilising raytracing. [ 5]

So two simplified models were used, each having deficiencies in some areas. The models are both quick and can be used also with spectral information on the optical components (glass, slats). They should give a clue how good certain approximations and simplifications are for the estimation of total g-values.