8.2. Results

8.2.1. Introduction

In the following sections, only the two unshaded cases and the four cases with horizontal slats (0°) of the total 23 test cases are presented in the main text. These cases were selected because they have the highest number of evaluated measurements. Hence, the results provide the highest statistical relevance. However, the results of all test cases are provided in the same way in Appendix A. All evaluation results, as specified in section 8.1.12, are presented in tabular form for each test case. In addition to the evaluation results, relevant specifications of the measurement conditions are provided in a statistical form. The resulting irradiances and transmittances depend on the actual irradiance profile and therefore cover a specific range. Since these parameters’ mean or average values are of little relevance, the quantities minimum, first quantile, third quantile and maximum are provided only. Finally, two charts offer a graphical overview of the main results for each test case: a correlation chart shows the modelled irradiance values as a function of the measured values; a second chart depicts the determined transmittance values as a function of the solar azimuth angle. In both charts, colours are used to indicate the sky condition at the time of the measurement. The diffuse ratio that was used for this purpose relates the diffuse irradiance on the horizontal plane (𝐷𝐻𝐼) to the total irradiance (𝐺𝐻𝐼).
Additionally, the measured and modelled irradiance values are presented as time series for the two unshaded cases (see Figure 136 and Figure 139). Presenting the shaded cases as time series is less relevant, as the duration of a single measurement is longer and the temporal resolution is significantly more coarse.

8.2.2. Case #1 – UNSHADED_SOUTH

8.2.4. Case #3 – ANTHRA_wC_00

8.2.5. Case #6 – SILVER_wC_00

8.2.6. Case #8 – WHITE_wC_00

8.2.7. Case #10 – Z-SILVER_wC_00

8.2.8. Result overview and discussion

When assessing the comparison results, it must be considered that the virtual model measurements are not compared against an ideal standard but against a complex empirical measurement depending on many variables. While the modelled results can be considered noise-free, the empirical measurement contains significant noise. For this reason, the (signed) mean errors 𝑀𝐸𝜏 , indicating systematic deviation, are considered most relevant to assess the quality of the model, provided the statistical errors are in an acceptable range. The latter can be assessed by the standard deviation of the errors 𝑆𝐸𝜏 , as well as the regression analysis, provided that sufficient data points are available.

Figure 148 and Figure 149 provide an overview of the error measures in graphical form. Note that the mean absolute error 𝑀𝐴𝐸𝜏 covers both statistical errors as well as systematic deviations. Additionally, the values are provided in Table 24. Note that, based on their definition, the stated errors are absolute errors. Hence, like their base quantity transmittance, they are related to the incidence irradiance.

8.2.8.1. Unshaded cases

It has to be noted that the evaluations for the unshaded case are not trivial, as they deviate significantly from the simple on-plane irradiance. As can be seen in Figure 135 and Figure 138, the determined transmittance values for the two unshaded cases vary significantly in the range of approx. 20% to 45%. This is a result of the strong angular dependence that the transmittance of the tripledglazed window exhibits. Beyond that, the shading of the reveal limits the amount of diffuse irradiance reaching the measurement area (located at the centre of the glazing). If only diffuse radiation reaches the window, either because direct sunlight originates from the opposite directions or due to cloud cover, the modelled transmittance converges to a constant value of around 34%. This represents the Full system empirical validation RadiCal, D. Rüdisser 214 average value for all directions of the visible sky and ground segments. The convergence to a constant value for diffuse irradiance can also be observed in the following shadowed cases.
The two unshaded test cases generally showed the highest level of variance. This can partly be attributed to the generally higher level of transmittance but also seems to indicate that reflections significantly impair the measurement. If no shading is present, specular reflections in the environment can directly reach the vertically oriented sensor. Considering the related, relatively high standard deviations of errors (4% each), the mean errors of determined transmittances were low, with only 1.3% for the south-facing case and 0.3% for the west-facing case.

8.2.8.2. Shaded cases with closed window

The results of these practically most relevant cases show good agreement of the modelled and measured transmittances. Most errors range well below 2%, with standard deviations in the same range. Only one case (#9) indicates a systematic error, as the mean error exceeds the standard deviation. However, only 22 measurements were available for the related case. The determined transmittances again showed a strong variation depending on the slat angle, sun angle and sky condition. The resulting transmittances range from approximately 1% to 20%. The lowest values are achieved for the anthracite slats in a low-diffuse sky condition, whereas test cases with horizontal slats in diffuse conditions show the highest transmittance.

8.2.8.3. Shaded cases with an open window

The measurements taken with open windows generally exhibit the highest errors and variances. Approximately half of the measurements show systematic deviations in the sense that the mean error exceeds the related standard deviation. Mainly two reasons could be the potential cause for the errors. Firstly, opening the window negatively impacted the internal scattering situation, as the black curtain behind the PyroScanner had to be partially retracted. An additional blanket was then used to cover the window sash, which was opened at an angle slightly greater than 90 degrees. The now visible parts of the window frame additionally contained some metallic parts that were left uncovered. Beyond the scattering, the open window also allowed external air to reach the measurement equipment leading to thermal gradients. Additional and more significant thermal gradients arose as the internal pyranometer and electronic equipment were exposed to the strongest direct solar irradiance in these cases. Accordingly, the highest deviations were observed in case #20 with a negative slat angle. In this shading configuration, the internal sensor was exposed to up to 900 W/m² direct normal irradiance with a strong temporal (spatial) fluctuation behind the tilted slats.

8.2.8.4. Albedo and ground reflected irradiance

The evaluation of the measurements showed that the combination of the bright metallic roof on the neighbouring building and the relatively dark floor is not captured appropriately by the idealised constant albedo approach. Performing adaptations of the albedo values used for the evaluation significantly improved the results regarding correlation and absolute error (not shown here). Nonetheless, for the purpose of validation, three constant albedo values (see section 8.1.11) were used for all evaluations performed here. The observed parameter sensitivity regarding the ground-reflected irradiance demonstrates the need for more detailed modelling of the surrounding environment, as discussed in section 6.2.4. Implementing such a refined approach that considers the geometry and reflectance values in the environment is potentially the most efficient method to increase the general accuracy of calculations for solar-induced energy components.

8.2.8.5. Conclusion PyroScanner measurements

Though the empirical measurements proved to be challenging, as they are influenced by many external factors, the comparison of the empirical measurement results with the model measurement results is satisfactory. The mean transmittance values that were determined in the measurement for each case ranged from approximately 1% (ANTHRA_wC_45) to approximately 75% (WHITE_wO_m45) depending on the material, slat type, slat angle and presence of the window. In addition, variations of the transmittance of 10 to 20 % and beyond (in terms of “absolute” transmittance) were recorded for the individual test cases. These variations depend on the actual irradiance profiles, mainly determined by cloud cover and solar angles.
The RadiCal method is able to predict the transmittances values that vary significantly depending on the shading type and configuration, as well as on the irradiance profile, mostly with accuracies in the range of 1 % to 2% in terms of “absolute” transmittance (see Table 24). Nevertheless, some cases show systematic deviations that would require further investigation. Potential sources for the errors can either be linked to the empirical measurements, e.g. in the form of thermal gradients (as “windowopen” measurements generally exhibited higher errors). They could, however, also be caused by reflections in the environment or testing facility that were not covered by the model. The simplification implied by the constant albedo model generally restricts the achievable (see section 8.2.8.4). Further, it must be considered that the detailed spectral and angular information, as required to apply the RadiCal method to its full extent, were not available, neither for the glazing nor for the materials of the shading devices. The inversion method developed in the PhD (sections 4.12 and 4.13) were applied to derive the required parameters of available data. Hence, the PyroScanner validation measurements prove that the method can be applied with limited data to predict the realworld performance of shaded or unshaded glazings with reasonable accuracy. It has already been demonstrated that the implemented optical models are able to provide high accuracy if the detailed data required to model the surface is available (see section 4.9.3).