10.Discussion

10.1. State-of-the-art methods and novel method

The state-of-the-art analysis, part of objective 1 and covered in Chapter 3, showed that the currently applied methods have various limitations. The methods proposed by the current standards are based on highly simplified models that can lead to significant inaccuracies. The main issues of the models used in the standards are their restriction to near-normal incidence and inadequate modelling of diffuse and direct radiation components (see sections 3.4 and 3.5). The methods were originally designed for performance benchmarking of building components under lab conditions. They should not be applied for real-world irradiance profiles, as required for building performance simulation. The application case presented in this work (see Figure 153) demonstrates that a constant total solar energy transmittance coefficient (“g-value”) approach is not applicable, as g-values determined by inversion vary in the range of several tens of per cent depending on the solar angles and irradiance profile.
More advanced methods that consider the angular profile of the incidence light mostly rely on a combined approach. The properties of the glazings are determined based on empirical data and advanced models by applying the methods implemented in LBNL Window Software (Curcija et al., 2018). The combination of the glazing with the complex geometry of shading elements or surrounding parts is modelled with raytracing tools, particularly Radiance (Radiance Community, 2022). The resulting bidirectional scattering distribution data (BSDF) described in the form of Klems matrices (Klems, 1994a) is then further processed to directional solar heat gain coefficients (DSHGC) or directly evaluated in building performance simulation tools. The DSHGC method requires the assumption of thermal properties and boundary conditions that have to be anticipated for the application they are later applied to, which leads to inaccuracies, see e.g. (Wermke, 2021). Hence, a strict separation of the optical model and thermal domain should be followed (see section 2.5).
However, beyond this issue, the BSDF/Klems approach shows some systematic limitations that arise from its definition. The limitations are related to simplifications regarding the spectral, spatial, angular and polarisation properties of the transmitted or reflected light beam. The impact of these limitations becomes more significant if a system involving two or more BSDF layers is modelled. This is a typical case, as the commonly used raytracing tools do not allow accurate modelling of glazings, while LBNL Window does not allow the modelling of detailed 3D shading layers or other relevant three-dimensional geometry. Hence, both methods are commonly applied to form a combined system of two or more BSDF layers.
The RadiCal method proposed here allows accurate, microscopic modelling of coated and uncoated transparent layers as part of a complex, detailed three-dimensional structure. This allows a unified approach as all relevant optical processes can be considered based on a single, comprehensive threedimensional model. The optical modelling of the glazings is potentially more accurate than the LBNL Window method, as the angularly and spectrally resolved model considers the entire global radiation spectrum and interference or polarisation effects on material boundaries are intrinsic parts of the model. This determines the angular and spectral properties of the scattering on the modelled surface, as well as the outcome of any further scattering events. However, the achievable accuracy relies on the quality of the available optical data for the modelled surfaces.
The RadiCal method proposes a strict separation of the thermal and optical domains. The newly introduced solar incidence operator (SIOP) provides an interface to pass the optical data to thermal Discussion RadiCal, D. Rüdisser 224 simulations. The definition of the SIOP allows for an accurate modelling of transmitted and absorbed powers or energies. As the output of the SIOP is a scalar value, it cannot readily be used for lighting applications. For this purpose, the RadiCal raytracer can still be used to generate BSDF information based on a single comprehensive and detailed three-dimensional optical model.