Qu？o confiáveis podem ser os modelos físicos em escala reduzida para avaliar a ilumina？？o natural em edifícios?

daylight is one of the individual strategies with greatest potential for reducing energy consumption in buildings. to realize this potential it is essential to characterize the environment precisely and quantitatively. for decades, reducedscale physical models has been employed for daylight evaluation. however, despite its benefits, this method has been the target of criticisms that have pointed out errors as intrinsic shortcomings. this study aims to evaluate two of the most cited sources of error: measurement under real sky conditions, and the effect of scale. the study was divided into two stages: (a) comparison of illuminances measured simultaneously in a real environment and a physical model in reduced scale, exposed to the real sky; and (b) comparison of illuminances measured in physical models built in three different scales subjected to an artificial "mirror box"-type sky. in the first stage, the results showed errors less than 5%, except in those situations where the reflected component was relevant. in the second step, the results were even better, showing that the scale effect was insignificant, with differences less than 4%. the results clearly indicate that this method is reliable, provided care is taken in producing the models and the measurements, especially regarding the optical properties of the surfaces, the conditions of exposure models (surroundings), dimensional accuracy, and photometric procedures.