ESTIMATION OF THE RELIABILITY OF THE THERMAL RECEPTIVITY COEFFICIENT OF THE SURFACES OF TIMBER FRAME WALL MATERIALS

A number of differences between the experimental results and the calculated thermal behaviour results of the timber frame walls have been obtained during the investigation carried out at summer operation conditions. The aim of the recent investigation is to determine at what stage in the calculation procedure the most significant errors in the estimation of the values of physical parameters might be made. The impact of errors on the further calculation reliability will be discussed as well. The methods of mathematical-statistics have been applied for establishing the parameters to characterize the sophisticated thermo physical processes.
The timber frame wall was used as the object of the experiment under unstable summer operation conditions. The calculations were performed according to the measured climatic input data (outside air temperature, solar radiation, wind speed, relative humidity, etc.) and the temperature distribution inside the tested wall.
The methodology for the calculation of the materials surface thermal receptivity coefficient Y is the object considered in the paper. Such methodology allows the estimation of the interaction between the physical parameters of different materials in the wall structure.
The investigation has resulted in the estimation of the parameter Y reliability. The modified Monte Carlo simulation, including analysis by the methods of the Paired t-Test, Tolerance intervals and Prediction intervals has been applied for the determination of the discussed reliability. Unstable processes of heat transmittance and interaction among the wall structural layers then have been described through parameter Y with adequate precision. The interaction was determined for the parameters included into the Y calculation equations, such as thermal conductivity ? W/mK, thermal resistance R m2K/W, the rate of thermal inertia D and materials thermal receptivity coefficient s W/m2K. The relative errors between the mean Ym and design Yds,0.9 values have been evaluated and the main dependences discussed.