Statistical Analysis of the Factors influencing the In Situ U-Value of Walls

Abstract

Building thermal performance testing requires in situ measurement techniques that are well supported and validated by simulation with statistics to improve the accuracy of the results. Local on-site performance of building components is different from the theoretical one, influenced by factors affecting the building's thermal conditions. The current paper reviews the factors influencing the measured U-value results in the heat flux method based on quantitative findings of other studies through regression and correlation statistics. The findings regarding the current status of knowledge are limited to in situ methods without detailed insights of response time, sensitivity analysis, and thermal boundary conditions in the local context. Regression analysis between wall characteristics, time duration, temperature difference, and the measured U-value shows a very strong and statistically significant impact of these variables on the accuracy of the measured U-value of low transmittance walls. The R2 value indicates that three variables can collectively explain 91% of the variance in the measured U-value. There is a linear correlation between the wall characteristics and the measured U-value and a non-linear correlation between the time duration, temperature difference, and the measured U-value. Future work will focus on developing a measurement framework that considers time-dependent variables, dynamic weather, and uncertainty with high accuracy for different boundary conditions.