Addressing uncertainty in U-value estimation with infrared thermography: A Kriging-based model calibration approach

The accurate estimation of U-values is crucial for assessing energy efficiency and planning retrofit strategies for the building stock. Conventional techniques, such as those based on infrared thermography (IRT), provide both qualitative and quantitative insights into surface temperatures. However, these methods often suffer from inaccuracies related to transient thermal conditions and the inaccuracy of infrared cameras (state of the art ±2 K). This paper aims to develop a method to obtain reasonably valid U-value estimates at urban scale under transient conditions. Therefore, a novel approach combining IRT with a model calibration and a Kriging-based interpolation method is developed. Kriging, a geostatistical interpolation method, is employed to model the temporal variations in thermalbehavior. The Kriging generated temperature interpolation is then used to calibrate a lumped resistor–capacitor (RC) model. The results of this simplified thermal model of the wall are used to estimate the U-value based on the dynamic thermal behavior. The methodology is evaluated using real-world NTC contact sensors and thermographic data collected from 3 different wall samples, allowing for a direct comparison with traditional IRT-based U-value estimation method (QIRT und IRT). The results demonstrate that the proposed Kriging-based model calibration approach enhances the reliability of U-value estimation to an accuracy of ±0.08 [W/(m 2 K)]. Hence, this study offers a robust and quantitative improvement over existing methods, with the potential to improve large scale, contactless U-value assessments in building energy diagnostics and contribute to more effective energy conservation strategies.