Figure 8 : Comparison of illuminances monitored in the test module and in scale model 2 a. 2.2 m., b. 4.2 m., c. 6.2 m. from window side c. . Photometric Sensors

The comparison of the cosine response of the two different types of photometers used in this study shows large discrepancies (cf. Figure 10). The BEHA luxmeter, employed in the test room, contributes to the underestimation of illuminances (lower response to internal reflected component at grazing angles), while the LMT luxmeter used in the scale model, has tends to overestimate illuminances (larger response to internal reflected component at grazing angle). The average relative divergence between both sensors reaches up to 20 % point difference.

a. b.

Figure 10 : Comparison of cosine response of the photometric sensors a. LMT luxmeter used in the scale model, b. BEHA luxmeter used in the test module.

Degree

t— BEHA luxmeter

Degree — cos

According to this result, most of the remaining overestimation of daylight factors and illuminances observed in the scale model compared to the test module can be explained by the different features of the photometers use in this case (cosine-response).

CONCLUSION

This study is an attempt to identify the main sources of experimental errors occurring in the assessment of building daylighting perfoarmance by the way of scale models. Beside the impact of the mocking-up of the geometrical dimension of the test module (a 1 : 1 simple office room), indoor surfaces reflectance of the scale model, as well as the photometers cosine-response, remained the principal sources of experimental errors, leading to an overestimation of daylight factors and illuminance in scale models compared to the test module.

Large relative divergence were found when comparing the impact of slight differences in surface reflectance in the model, a 6% point difference of surface reflectance leading up to 84% divergence of daylight factors in the deeper part of the test module. Scale model location in this case appears to be non significant, the divergence remaining constant for two different locations close to the test module. The different cosine-response of the photometers used in the scale model and the test module are responsible for a 20% relative divergence between the monitored daylighting performance. Other experimental factors, such as photometers placement and levelling, can explain the remaining discrepancies, which appear difficult to reduce underneath 20 to 30% relative divergence.

Care should be brought, as a consequence, to the construction of scale models used to predict the daylighting performances of buildings, if a reasonable accuracy on daylight factors and work plane illuminance is expected to be reached. Photometers should be carefully calibrated and placed within the model. All these measures should contribute to reduce the overestimation tendency of the scale models in daylighting performance assessment.

Further studies are required however to investigate the other sources of experimental errors, still occurring today even with the new generations of sky simulator.

ACKNOWLEDGMENTS

The author would like to acknowledge the financial support of the Federal Commission for scholarship for foreign Students (FCS) of the Swiss Confederation. P. Loesch assisted in the scale model construction and A. Machacek in the model material selection.