Comparison of different approaches to evaluate the equivalent thermal diffusivity of building walls under dynamic conditions

Abstract

In the present work, three different methodologies to determine the equivalent thermal diffusivity of an opaque wall, part of a building envelope, were compared. The thermal behaviour of a typical wall of a recently built building has been recorded; the internal and external temperatures were measured for one month, during day and night, with one acquisition every 10 min. The experimental data were processed by means of three analytical models, getting as a result the effective thermal diffusivity of the wall. The first model used is the semi-infinite region, the second a symmetric slab with periodic surface temperatures and the third the ISO 13786:2007 rule. All models produce different values if thermal diffusivity is evaluated from the ratio of the external and internal amplitudes or from their phase shifts. For the first model the α values obtained are respectively (1.512·± 0.016)·10−6 m2 s−1 and (2.813 ± 0.004)·10−6 m2 s−1; for the second (1.741 ± 0.150)·10−6 m2 s−1 and (2.603 ± 0.007)·10−6 m2 s−1, and (1.880 ± 0.017PCNM)·10−6 m2 s−1 and (2.168 ± 0.003)·10−6 m2 s−1 for the third. This difference is likely due to the lack of correspondence of the wall models to its real structure. When a non-linear regression is applied to data, a unique value is obtained, intermediate between the two, i.e. (2.544 ± 0.021)·10−6 m2 s−1 (first model), (1.854 ± 0.010)·10−6 m2 s−1 (second model) and (1.988 ± 0.012)·10−6 m2 s−1 (third model). Both the difference between the two values of thermal diffusivity and the root mean square obtained from the nonlinear-least square regression asses that the best results come from the application of the ISO 13786 rule.